J . .-llhlm LIBRARY Michigan State University This is to certify that the dissertation entitled Structure and dynamics in the colorado forest of the Luquillo Mountains of Puerto Rico presented by Peter L. Weaver has been accepted towards fulfillment of the requirements for Ph.D. _ Ecology degree In Qt Q. mm "'L Major professor ‘ \ Date 0A 6‘8 MS U is an Aflirnum‘ve Action/Equal Opportunity Institution 0-12771 MSU RETURNING MATERIALS: Place in book drop to LJBRAfiJES remove this checkout from .-,— your record. FINES VIII be charged if book is returned after the date stamped below. L Wat/73" ”313 we :4" STRUCTURE AND DYNAMICS IN THE COLORADO FOREST OF THE LUQUILLO MOUNTAINS OF PUERTO RICO by Peter L. weaver A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Botany and Plant Pathology 1987 ABSTRACT STRUCTURE AND DYNAMICS IN THE COLORADO FOREST OF THE LUQUILLO MOUNTAINS OF PUERTO RICO BY Peter L. Weaver The influence of past hurricanes and environmental gradients on forest structure. species composition. regeneration and productivity. was investigated in the colorado forest (subtropical lower montane wet and rain forests. sensu Holdridge) of the Luquillo Mountains of Puerto Rico. Forest recovery after the passage of hurricane San Cipriano in 1932 was assessed on 2.8 ha of permanent plots located in natural forest. During the 35-year period of measurement from 19u6 to 1981. secondary tree species declined in abundance whereas shade tolerant climax species. the size of residual trees. and the mean specific gravity of the overall wood biomass. increased. Species richness was greater on lower elevation valley plots than on higher elevation slope plots. and declined over time on the most diverse plots and increased on the least diverse plots. Aboveground woody biomass increased from 121 to 135 t/ha with an average accrual of 0.67 t/ha/yr. In comparison. the increase in stem numbers and biomass on a O.h ha thinned plot in the same forest was greater than the mean for the natural forest plots. The regeneration of mill; tnnnfllitlnna. the largest and oldest tree species within the forest. appeared related to past hurricanes. Forest structure and soil characteristics were studied on 75 temporary 500 m2 plots. stratified by elevation. life zone (wet and rain). and topography (ridge. slope. and valley). throughout the colorado forestu Stem density increased with elevation in the rain forest but not in the wet forest whereas tree height declined with elevation in both life zones. Species richness showed no relationShips with any of the environmental factors. Basal area. wood volume. and wood biomass varied with topography and elevation in the rain forest. but not in the wet forest. and bulk density. percent organic matter. and soil organic matter content. varied with elevation in the wet forest. but not in the rain forest. Reciprocal averaging and polar ordinations of species and plot data from the temporary plots provided evidence of species' preferences in a complex. montane habitat. including trends by topography. aspect and elevation. Of the 39 species used in ordinations. five were largely confined to valleys. eight were found in valleys from 30 to 65 percent of the time. and the remainder were tallied on ridges or slopes more than 75 percent of the time. The regeneration of 20 canopy tree species was studied using centrally located subplots on each of the temporary plots. Data on seedling regeneration. understory stems. seed size and wood density were used to classify canopy species as shade intolerant pioneers. gap or late secondary species. and shade tolerant species with different gap requirements for maturity. The net primary production of 7.83 t/ha/yr for two 0.h ha permanent colorado forest plots was determined by summing biomass accrual (0.82 t/ha/yr). litterfall (6.80 t/ha/yr). and herbivory (0.21 t/ha/yr). To elucidate spatial and temporal trends in hurricane disturbance and recovery. as well as primary production and related measures in the Luquillo Mountains. the results of this study were compared with intonation for the forests above and below the colorado forest. Overall. the structural and dynamic parameters for the colorado forest were intermediate between those of the lower and higher elevation forests in the Luquillo Mountains. This dissertation is dedicated to Alba Iris ACKNOWLEDGMENTS I am grateful to Dr. Peter Murphy. my major professor. and Drs. Donald Dickmann. Jay Herman and Fred Tschirley. dissertation committee members. for their assistance and guidance during the conduct of this research. The colorado forest permanent plots. which provided information on long-term changes. were established by Dr. Frank Wadsworth during the mid-1980's. Part I of the dissertation was reviewed by Ms. Barbara Cintron of the Commonwealth of Puerto Rico Department of Natural Resources and Dr. Frank Nadsworth of the Institute of Tropical Forestry; Part II was reviewed by Dr. Thomas Crow of the Forest Sciences Laboratory in Rhinelander. Hisconsin. and Mr. Lew Ladd of the Southern Forest Experiment Station Statistical Office in New Orleans; and Part III was reviewed by Dr. Carl Jordan of the Institute of Ecology. Athens. Georgia. and Dr. Ariel Luge of the Institute of Tropical Forestry. Ms. JoAnne Feheley. the Institute librarian. was extremely helpful in locating literature relating to the three sections of the dissertation. Mr. Martin Chudnoff of the Forest Products Lab in Madison. Wisconsin. provided specific gravities for many of the tree species. Assistance in field work was provided by many of the Institute's forestry technicians. Most notable for his contribution was Carlos Rivera who helped tally trees on many of the permanent and temporary plots. Others who assisted at one time or another were: Carlos Dominguez. Rector ii {and Lisette Esteban. Luis Totti. Mary Jeane Sanchez and Felipe Torres-Pollack. Study at Michigan State University was made possible through a one year GETA (Government Employee Training Act) program. iii TABLE OF CONTENTS LIST OF TABLBSOOOOOOOOOOOOOOO0.00......OOOOOOOOOOOOOOOOOOOOOOO LIST OF FIGURES.COOOCOOOCOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO GENERAL IMRODnfllonOOOOOOOOOCOOOCIO0.00.0.0...0.00.00.00.00. 0 Purpose and Organization of the Study................... The Study Site.......................................... Site Characteristics.................................... Geology and Soils..................................... Climate............................................... Vegetation.............................................. Past Research in the Luquillo Forest.................... LITERAflRE CIT-OOOOOOOOOOOOOOOO0......OOOOOOOOOOOOOOOOOOOOOOO PART I. CHANGES IN FOREST STRUCTURE AND GROWTH AFTER HURRICANE DISTURBANCE IMRmUCTIONOOOOOOOOOOOOOOO0.00.0.0.0...00.00.000.000...0.0... MODSCCOOOOOOOOOOOOOOOOOI...0.00.0.0.0...OOOOOOOOOOOOOOOOOOO Rmrs.OOOOOOOOOOOOOOOOOOOOO00....00......OOOOOOOOOOOOOOOOOOO Stand Changes........................................... Ingrowth. mortality and number of stems............... Diameter class distribution........................... Height class distribution............................. Crown class distribution.............................. Specific gravity distributions........................ Changes in Species Composition.......................... Survival of indicator species......................... Diameter class distributions of canopy species........ Species survival on thinned plot...................... Species richness...................................... Rare and endangered species........................... Similarity of the plots............................... Growth and Production................................... Diameter growth of select species by crown class...... Basal area changes.................................... Volume changes........................................ Biomass changes....................................... ‘8. Of mullmam.o..ooo..................... DISCUSSION.O.COOOOOOOOOCOOOCOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO. iv Page vii ix The N.tura1 Forest PlotaoOOOOOOOOOOOIOOOOO0.00.00.00.000 Hurricane-induced changes in forest structure and compositionOOOO00.0.0000...OOOOOOOOOOOOIOOOOOOOOO GmuthCOOOOOOOCOOOOOOIOOOOOOOOOOOOIOOOOCOOOOOOOOOOOOOO The Thinned PlotOOOOOOOIOOOOOOOOOOOOIOOOO0.00.00.00.00.0 Changes in forest structure and composition induced by thinninSOOlOOO0000......OOOOOOOOOOOOOOOOOOOOOO Gmuth.00......IOOOOOOOOOOOOOOOOOOOOOOOOOIOOOOOOOOOIOO Age and Dynamics of Cyrillahzacemiflgra................. Hurricane Damage and Recovery in the Luquillo Mountains. SWIG O O O O O I O O O O O O O I O O O O O O O O O O O O I O O O O O O O O O O O O O O O O O O O O O O O O O O 0 LITERATURE CITED. 0 C O O O O O O O O O O O O I O I O O O O O O O O O O O O O O O O I O O O O O O O O O O 0 APPENDIX TABLE. 0 O O O O O O O O O O O O O O O O I O O O O O O O O O O O O O O O O O O O O O O O O O O O O 0 APPENDIX FIGUREOOOOCOOOOOOOOOOOCOOOOOOOOOOOOOOOO0.0.0.0000...O 9:1 911 100 103 103 107 107 112 116 120 128 130 PART II. STRUCTURE AND COMPOSITION OF CLOSED CANOPY COLORADO FOREST RELATIVE TO ENVIRONMENTAL GRADIENTS IMRmUCTION.O0.0.0....O...I.OOOOOOOOOCOOOOOOOOCOOOO0.0.0.0... mmODSCOOOOOCOCOOOOOCOCOOOI0.00......OOOOOOOOOOOOOOOOOOOOOOOO REwLTSANDDISCUSSION.00.0.0.0...OOOOOOOOOOOOOOOO0.0.0.000... Forest Cover............................................ Species Richness........................................ Stem Density............................................ Tree Size Gradient in the Forest........................ Organic Matter of Forest Soils.......................... Ordination of Arborescent Vegetation.................... Ordination 1 - sample plots for the entire forest..... Ordination 2 - species distribution for the entire forest........................................... Ordination 3 - composite plots........................ Ordination h - leeward wet forest plots to the west... Other ordinations..................................... Species distributions................................. Regeneration of Canopy Species.......................... Seedling and understory success....................... Seed size of canopy species........................... Specific gravity of canopy species' woods............. Tentative Life History Classification of Canopy Trees... Shade intolerant pioneer species ..................... Gap or late secondary species ........................ Shade tolerant species ............................... SWOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO00.00.00.000... 132 136 1N3 1N3 133 153 156 166 173 17! 177 180 183 183 186 199 201 20! 205 206 206 209 210 215 LITERATURE CITED.............................................. APPENDIX TABLE................................................ APPENDIX FIGURES.............................................. PART III. FOREST STRUCTURE AND PRODUCTIVITY ImnonumIou000.0000000000....0000.00000000000.000000000000000 MODS.0.000000000000000000.00000000000000000000000000000000. REULTS000000000000.0.000.000.000000.000.000.000.0000000000000 Leaf Characteristics in the Forest...................... Leaf area index....................................... Leaf weights.......................................... Litterfall and Loose Litter............................. Standing Herbivory and Herbivory Rates.................. Bimaa800000000000000000000.0.000000000000000000.0000000 ".t Prim” PmuOtion (NPP). 0 00 00 00 000 00 00 000 0 0 00000 0 0 0 DIstSION0000000.0000000.0000000000000000.0000000000000...000 [Great StmOturG...0000000000000000000000000000000000000 Forest Bimaaa Eqmtion800 .00. 0000 0.0000000... 00000000 . 0 Forest Dynamics and Productivity........................ Th. Luquillo Forest continum.00000000000000.00000000000 smx000000000000000000000000.00000000000000000000000000000. LITEMNRB CITEDeeaeeeaaaaaaaaeeaaeeaaaeaaeaaeeaeeeaeaaaeaseaa ‘melx ngeeeeeeeeeaaeaeeeeeeeaeaaeeaeaeeeeaaeaaeeeeeeeaae vi 218 22“ 230 236 238 283 2&3 283 283 2N6 255 255 258 272 272 275 277 286 290 292 296 9. 10. LIST OF TABLES GENERAL INTRODUCTION Comparison of stand structure in four forests of the Luquillo Mountain30000000000000000000000.0000000000000000000000000.0 An approximation of the areal extent of colorado forest elsewhere in Latin America 00000000000000.0000...0..0000.00 PART I Original stem densities. survival. ingrowth. mortality. and final stem densities for the colorado forest plots from 19u6 to 1981000000....0000.000000000000000000000000.000000000000 Relative density. relative frequency and relative basal area and percent importance for major tree species on all natural forest plots in the colorado forest in 19fl6 and 1981. Species are arranged from highest to lowest percent importance based on the 19u6 values for the 25 most important species......................................... Shannonéwiener indices (Krebs 1972) in 19u6 and 1981 on the permanent colorado forest plots........................ Rare and endangered tree species of Puerto Rico found on the colorado forest permanent plots........................ Rare and endangered tree species (stuns/ha) found on the colorado forest temporary plots............................ Percent similarity indices for the colorado forest permanent plota 1n 19n6 and 1981000000000000000000000000.0000...00.00 Mean annual diameter increment (MAI) of 35-year survivors of select species by crown class on the natural forest permanent p10t8000000000000000000.00000000000000000000000000000000000 Comparison of mean annual diameter increment (MAI) of surviving dicot trees by crown class between 19fl6-51 and 1976-81 on four natural forest permanent plots............. A1. Biomass accumulation rates for secondary and mature natural 11. forests in the American tropics.0.0000000000.000.00.0000000 PART II Summary of studies using numerical analyses in tropical forests.000000000000.000000000000.000000000.000000000000000 vii Page 15 18 35 52 73 7h 75 77 79 80 128 133 12. Summary of percent cover in closed canopy stands in the eclorao for°8t0000000000000000000.0000000000000000.00000.0 1“” 13. Values of forest parameters by life zone and topography.... 1117 111. Summary of regeneration characteristics of canopy species in the calorado forest00000000000000000.000.000.00000000000 200 15. Ranking from most secondary to most primary (top to bottom) of tree species that reach the canopy in the colorado forest 207 AZ. Alphabetical listing of tree species abundance in the ommtion studyOOOOOOO000.000.000.0000000000000.00....0000. 22“ PART III 16. Leaf area indices by compartment in the colorado forest.... 2AA 17. Leaf characteristics for common species found throughout the calorado for°8t000000000000000000000000000.00000000000000000 zus 18. Standing herbivory sample for 12 tree species in the colorado foreat00000000000000000000000000000000....000000000000000000 256 19. Herbivory rate sample for nine tree species in the colorado forest over 90 daya00000000000.0000000000000000000000000000 257 20. Detailed calculation of net primary productivity (NPP) for the calorado forest.000.0000000000000000.0000...00000000000 269 21. Comparison of stand structure for four forests within the Luquillo mtaina00000000000000000000000000000000000000000 273 22. Comparison of stand dynamics for four forests within the Luquillo muntain80000000000000000.0000.00.000000000000000. 278 A3. Summary of biomass field data.............................. 296 viii Fienmre LIST OF FIGURES GENERAL INTRODUCTION The Luquillo Experimental Forest. The colorado forest is divided into wet forest (oblique lines) and rain forest (cross-hatched) life zones (Holdridge 1967; Ewel and Hhitmore 1973). Dark circles represent temporary plots used in species-site ordinations. Dark triangles represent the permanent plots used in growth and stand dynamics studies.... Climate diagrams (Halter 1975) for La Mina based on data collected from 1935 to 19A3 (Wadsworth 19u8) and for P100 del Este from 1969 to 1982 (0.8. Department of Commerce 1970-82). Striped areas indicate periods of water surplus and stippled areas indicate rainfall greater than 100mm/month Monthly occurrence of 102 hurricanes in Puerto Rico. 1508-1970 (Salivia 1972). Oblique lines indicate hurricanes for which only the month was recorded. These were arbitrarily assigned to the middle of the month................................... Hurricane record for Puerto Rico from 1700 to 1970 (Salivia 1972). Long vertical I bar - type A storm (vertex of hurricane passes directly over the island with hurricane force winds); intermediate vertical I bar - type B storm (vortex of hurricane may or may not have passed over the island. but only part of the island experienced hurricane force winds); short vertical I bar - type C storm (vortex passed at some distance from the island. which experienced high winds (not hurricane force) and heavy rains); double I bar - two storms in the same year with intensity of second storm indicated by position of second I bar within the first; horizontal line from 19A6 to 1981 - years of measurement of tree growth in the Luquillo Forest; black circle - type A storms that passed directly over the Luquillo Forest......... Path of type A hurricanes with trajectories directly over Puerto Rico from 1700 to 1970. Solid lines are known trajectories and dashed lines are assumed trajectories based on descriptions (adapted from Salivia 1972)................ PART I Diameter class distributions for seven natural forest permanent plots and one thinned permanent plot in the calorado forest 1n19n6 and1981000.0.00.00.00.0000000000000 Diameter class distributions by topography for natural forest permanent plots in the colorado forest in 19h6 and 1981..... ix Page 10 12 1A 37 no 1C). 11. 12. 13. 1A. 15. 16. 17. Height class distribution by topography for natural forest permanent plots in the colorado forest 1986 and 1981........ Crown class distribution for the natural forest and thinned ”fluentplots19u6md19810000000000.00000000000000000000 Specific gravity by groupings (0.25 = 0.2 to 0.29. etc.) for number of stems. basal area. and volume in 19u6 (shaded) and 1981 (unshaded) on the natural forest and thinned permanent colorado plots.............................................. Percent survival of indicator species in the colorado forest from 19h6 to 1981. Lines indicate survival of original trees. Dot is 1981 population of trees (survival of residual trees and ingrowth). Bars indicate basal area: shaded bars. 19h6 basal area; unshaded bars. 1981 basal area survival and growth on survivors; and obliquely-lined bars. 1981 total basal area. growth of residual basal area and ingrowth. Number of trees/ha are indicated for slepe (S). slape and valley (M). and valley (V) plots............................. Diameter class distributions for indicator species on 2.8 hectares of natural forest permanent plots in 19fl6 and 1981. Class A = ”.1 to 9.9 on; class B = 10.0 to 1h.9 cm; subsequent classes are in 5 cm increments through class I = 130.0 to 138.9 cm. The species indicated by asterisk are also canopy species............................. Diameter class distributions for canopy species on 2.8 hectares of natural forest permanent plots in 19fl6 and 1981. Class A = 1.1 to 9.9 cm; class B = 10.0 to 1u.9 cm; subsequent classes are in 5 cm increments through class 0 = 75.0-79.9 cm. Indicator species not included............ Change in number of stems/ha by species on the thinned colorado plot vs. all natural forest colorado plots. In each case. the thinned plot is the first pair of entries by Bpeci°800000000000000.000000000000000000000.0...00000.0000000 Change in number of species by plot and for the entire forest from 19166 to19810000000000.0000000..OOOOOOOOOOOOOOOOOOOOOOOO Basal area changes for all permanent plots from 19u6 to 1981. Mean is for all natural forest plots combined. Ingrowth includes recruitment and growth of residual stems. Mean annual basal area growth (in m /ha) is shown above the bars. The top value is for ingrowth alone. The bottan value sums both ingrowth and mortality as growth............. Volume changes for all permanent plots from 19h6 to 1981. Mean is for all natural forest plots combined. Ingrowth includes recruitment and g wth on residual stems. Mean annual volume growth (in m /ha) is shown above the bars. ”2 115 I17 50 55 63 68 71 83 18. 19. 20. 21. ‘-1e 22. 23. ‘The top value is for ingrowth alone. The bottom value sums both ingrowth ”d mortality “ growth.0.000000000000000000000 Aboveground woody biomass changes for all natural forest permanent plots from 19fl6 to 1981. Mean is for all natural forest plots combined. Ingrowth includes recruitment and growth on residual stems. Mean annual biomass growth (in t/ha) is shown above the bars. The tap value is for ingrowth alone. The bottom value sums both ingrowth and mortality as growth.......................................... Scatter of points for diameter growth by diameter classes for 9121113 racemiflora. Circled points were those used in the weighted least squares regression. Points with standard error bars represent the means and standard errors for the circled points within the respective diameter classes. The mean and standard error at 10 cm is for the ingrowth on the thinned plot. Solid dots indicate growth rates that were discarded in the calculations for tree age.................. Curve of age related to diameter for Cyrillaflraaemiflgza. The circles represent the upper endpoints of the diameter classes on the ordinate vs. the cunulative ages on the ab8°18“0000000000.000000000000000.00000000.000.00.0000000000 Net biomass growth rates (aboveground woody biomass) for secondary and mature natural forests in the American tropics: (1) tropical moist forest (Salas 1978; Folster. Sales and Khanna 1976); (2) tropical moist forest (Ewel 1971); (3) subtropical moist forest (Snedaker 1970); (A) tropical wet forest (Galley £1.31, 1976); (5) tropical moist forest (Uhl 1980); (6) subtropical wet forest (Crew 1980); (7) lower montane wet forest (Tanner 1980); (8) lower montane wet forest -- colorado forest (this study). Forest formations are Holdridge (1967) life zones ................... Diameter class distribution of Cyrilla,raeemiflgna_on the natural forest plots within the colorado forest in 19h6. The time scale indicates the years to which the diameter class corresponds as well as the years that type A storms passed directly over the Luquillo Mountains. according to the synthesized age curve. The shaded portion of smallest diameter class. that is. < A.1 cm in diameter. was not tallied in the field.................................................. PART II Tree species-area relationships for 25 valley. 25 slope. and 25 ridge plots 500 m in size in the colorado forest. All Stems “re 2 16.1 cm dbh...000000000000.000.000.00000000000 Distribution of data points in the covariance model I = T + LZ + TxLZ + E where I = the number of species per 500 m2. T : topography. L2 = life zone. and E = elevation. None of xi 85 88 91 93 105 130 1A6 2“. 225. 26. 27. 28. 29. 30. 31. the component terms of the model was significant............. Distribution of data points in the covariance model I = E where I = the number of stems/ha and E = elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in the rain forest. but not in the wet forest. Topography of plots is indicated............. Distribution of data points in the covariance model I = E where I = thg mean height of dominant and codominant trees on the 500 m plots and E = elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in both the wet and rain forest. Topography of plots is indicated..................... Distribution of data points in the covariance model 5 = T + E + TxE were I = the basal area (m lha) on the 500 m plots and T = topography. E = elevation. and TxE = inter- action. according to life zone. The topography and inter- action components of the model are significant at alpha = 0.05 in the rain forest. but none of the components are significant in the wet forest................................. Distribution of data points in the covariance model I = T + E + T E where I = aboveground woody volume (m lbs) per 500 m plot and T = topography. E = elevation. and TxE = interaction. according to life zone. All components of the model are significant at alpha = 0.05 in the rain forest. but none of the components are significant in the wet foreat0000000000000000000000000000000000000.000000000000000... Distribution of data points in the covariance model I = T + E + TxE where I = aboveground woody biomass (t/ha) per 500 m plot and T = topography. E = elevation. and TxE = interb action. according to life zone. All components of the model are significant at alpha = 0.05 in the rain forest. but none are significant in the wet forest........................ Distribution of data points in thescovariance model I = E where I : soil bulk density (g/cm ) and E : elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in the wet forest. but not in the rain forest............................................ Distribution of data points in the covariance model I = E where T = percent organic matter in soil and E = elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in the wet forest. but not in the rain forest............................................... Distribution of data points in the covariancg model I E where Y = soil organic matter content (kg/ m ) and E elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in the wet forest. xii 151 155 157 161 163 165 168 170 32. 33. 3h. 35. 36. 37. 38. but not in the rain forest0000000000000000000000000000000.0000 172 A reciprocal averaging sample plot ordination comprised of 75 plots based on 39 tree species for the entire colorado forest. Two configurations are separated by the dashed line. Group A contains mainly valley plots ranging from low to high eleva- tion. Group B contains largely slope and ridge plots. again ranging from low to high elevation. The 800 m contour. the midpoint elevation. is indicated.............................. 176 A reciprocal averaging species ordination comprised of 39 tree species on 75 plots for the entire colorado forest. Three configurations are separated by solid lines. Group A contains species predominantly found in valleys. from low to high elevation; group B. species with 30 to 65 percent occurrence in valleys. from low to high elevation; and group C. species with combined ridge of slope occurrence greater than 75 per— cent. from low to high elevation. See legend for more infor- mation on topographic occurrence of species................... 179 Reciprocal averaging sample plot ordination for 25 composite groupings of ridge. slope and valley topography at the same elevation. based on 39 tree species for the entire colorado forest. Groups 1 through A are partitioned on the basis of life zone (coincident with aspect) and elevation. as indicated. Exceptions are explained in text.................. 182 Polar sample plot ordination for 27 individual leeward wet forest plots based on 38 tree species. Solid lines separate different elevational groupings of plots and dashed lines separate different topographic groupings of plots............. 185 Distribution of the major tree species on 3.75 ha of colorado forest by topographic position (V = valley. $3: slope. R a ridge). The specific gravities given in g/cm are discussed later in text................................... 188 Elevational distribution of common colorado forest tree species to the windward and leeward of El Yunque peak. Each point represents the mean value for a ridge. slope. and valley composite plot at a given elevation........................... 190 Panel A: The seedling success of canopy tree species based on the seedling relative densities of canopy species divided by the relative density of canopy size plants of the same spe- cies; Panel_B; The understory success of canopy tree species based on the understory relative densities of canopy species divided by the relative density of canopy size plants of the same species; Panel 9: Seed size of canopy tree species. defined as the sum of half the length + the width of the disseminule; and Panel D: Wood specific gravities of canopy tree species. The species are arranged in the same order as they appear in Table 18.............................. 203 xiii A—2. Reciprocal averaging ordination: species ordination of 27 A-B m .A-l. 39. 40. “10 N2. 43. m1. “5. 46. #7. I8. l9. windward and 30 leeward plots and 30 species on ridge topwraphy0000000000000OI000000.00000000000.000000000000000... Reciprocal averaging ordination: species ordination of 27 ‘windward and 30 leeward plots and 26 species on slope topography000000000000000000000....000000.0000000000000000...O Reciprocal averaging ordination: species ordination of 27 windward and 30 leeward plots and 2H species on valley topxraphy000000000000000000.0000000000000000.0000000000000000 PART III Total litterfall for the colorado forest by month. from January 1’ 1981 thm‘lgh December 31’ 19820000000000...00000000 Leaf litterfall and woody litterfall for the colorado forest by month. from January 1. 1981 through December 31. 1982...... Flower/fruit litterfall and miscellaneous litterfall for the colorado forest by month. from January 1. 1981 through Decemmr 31’ 198?§ aaaaaaaaa O OOO00...00000000000000.0000.00000 Woody looselitter. miscellaneous looselitter and total looselitter for colorado forest by month. from January 1 thmuh Decmber 31’ 198200000.000.000.000.0.0000000000000000. Leaf weight expressed as an allometric function of DZH. Tabonuoo data were obtained from Ovington and Olson (1970) and the colorado data from this study......................... Branch weight expressed as an allometric function of DZH. Tabonuoo data were obtained from Ovington and Olson (1970) and the colorado data from this study......................... Trunk weight expressed as an allometric funtion of DZH. Tabonuco data were obtained from Ovington and Olson (1970) and the colorado data from this study......................... Total aboveground woody tree weight (b anches. trunks) ex- pressed as an allometric function of D H. Tabonuco data were obtained from Ovington and Olson (1970) and the colorado data from this study.......................................... Total aboveground tree weight (trunk. br ches. and leaves) expressed as an allometric function of D H. Tabonuco data were obtained from Ovington and Olson (1970) and the’ colorado data from this study................................. Mean annual organic matter (0. M.) budget for two plots of 0.A0 ha in size in the colorado forest........................ A - Estimates of loose litter storage in montane tropical xiv 230 232 23" 2&8 250 252 25” 260 262 26A 266 268 271 forests over an elevation gradient on the same mountain: dots - Khao (Mt.) Luang. peninsular Thailand (Yoda and Kira 1969); triangles - Luquillo Mountains. Puerto Rico (Table 22). Circled triangles. palm forest. B - Estimates of litterfall in montane tropical forests over an elevation gradient on the same mountain; dots - Gunung Mulu. Sarawak (Proctor et,al, 1983); triangles - Luquillo Mountains. Puerto Rico (Table 22). Circled triangles. palm forest........................................ 281 XV GENERAL INTRODUCTION Purposg_§nd Orgagigation of the Study Forest vegetation types develop mainly in response to climate and exhibit intrinsic variation in structure and species composition according to environmental conditions. On the islands of the Caribbean, the major spatial variables within montane forests are elevation, aspect and topography, all of which comprise gradients that influence microclimate. The major stochastic climatic event is the hurricane. The four major hurricanes that traversed the Luquillo Mountains during the past two centuries had an impact on the vegetation. For most tropical forests, historical and environmental information is extremely limited. This, in part, has contributed to the view that tropical forests are homeostatic ecosystems with a complex, randomly distributed, species component. Only recently has there emerged a more dynamic concept of tropical forests. The availability of structural data for the colorado forest a few years after the passage hurricane San Cipriano, as well as adequate taxonomic information on its component tree species, provided a rare opportunity to demonstrate the effects of both historical and environmental influences on forest structure and composition. Because forest structure, species composition, and dynamics are all related, this study of the colorado forest was divided into three parts. The purpose of the first part was to determine temporal changes in forest 2 structure and species composition wrought by hurricane San Cipriano of 1932. Eight permanent 0.40 ha plots established in the mid-1940’s and re-measured in 1981 were used to assess these changes. The plots were located between 650 and 900 m in elevation on windward and leeward aspects. In essence, this section provides a chronology of the colorado forest between 1946 and 1981, and characterizes typical changes in species composition and structure associated with the recovery of the colorado forest from a major hurricane. The second part of the study concerned the current condition of closed canopy forest, including regeneration, and elucidated spatial patterns in species composition relative to environmental gradients. Canopy vegetation was sampled in 1981 on 75 temporary, 500 m2 plots, stratified by elevation, topography and aspect, and regeneration was tallied on 20 subplots of 1 m2 located in the center of each plot. Soil organic matter in the top 50 cm was also determined. Species distributions were investigated using direct gradient analyses (summarizing distributions by elevation and topography), and indirect gradient analyses including reciprocal averaging and polar ordinations. Structural parameters were investigated using covariance analysis. The third part Of the study was to determine forest productivity and to assess the influence of elevation on forest structure and dynamics by comparing the results of the colorado forest study with those available for the forest types above and below it. The Study Sitg_ The Luquillo Experimental Forest is located in northeastern Puerto Rico at lflPN latitude and 660W longitude (Fig. l) and encompasses Figure 1. The Luquillo Experimental Forest. The colorado forest is divided into wet forest (oblique lines) and rain forest (cross-hatched) life zones (Holdridge 1967; Ewel and Hhitmore 1973). Dark circles represent temporary plots used in species-site ordinations. Dark triangles represent the permanent plots used in growth and stand dynamics studies. DUE ~ . Obmw2a1~ H shaman v 3:20 2: .. 33> .m - 09:20 05. . xsasu - ocoh .u - vacca> .w n oczz o4 - o.moo .om ou.n . o.mm .ov ou_n - v ocoaom u oc_.o.ou - 22¢ >33...th 0 33a .cocoEcom ‘ —NMVID(ONQOSQ: 3:: s; E :22 53. E ocoom4 11,200 ha. Twenty-five percent lies between 120 and 300 m in elevation, 47 percent between 300 and 600 m, 25 percent between 600 and 900 m, and three percent between 900 and 1075 m (Wadsworth 1951). Site Characteristics Mam—Soils The existence of the Luquillo Mountains, an isolated range of upper Cretaceous monadnocks, was first attributed to folding (Meyerhoff 1933), but more recent work showed that uparching and faulting were the probable causes (Mitchell 1954). The mountains rise abruptly to 1075 m only 8‘km from the ocean, and become increasingly dissected with elevation. The soils are mainly clays with silty clay loams also present. Rocks are common, especially in ravines and large streams. The soils are classified mainly as inceptisols (Beinroth 1971), mineral soils having one or more horizons in which minerals have been altered or removed, but not accumulated to a significant degree. Ultisols, or highly leached soils with a low base saturation, are also present on small areas. The soils are saturated throughout most of the year and contain large amounts of organic matter. Detailed soil mapping has not been accomplished. Climate The abrupt elevational gradient from the coast to the summits profoundly influences the climate of the mountains. Rainfall increases from 2300 mm/yr at 200 m near the northern forest boundary (Crow and weaver 1977) to 4600 mm/yr at La Mina near 715 m elevation (Fig. 2), and averages 4450 mm/yr on Pico del Rate at 1050 m. The mean annual temperatures vary from 23°C to 19°C over the same gradient. Because Figure 2. Climate diagrams (Halter 1975) for La Mina based on data collected from 1935 to 1943 (Wadsworth 1948) and for Pico del Rate from 1969 to 1982 (0.8. Department of Commerce 1970-82). Striped areas indicate periods of water surplus and stippled areas indicate rainfall greater than 100 mm/month. -Lo Mino- 4600 mm/yr 2|.l °C /Y' 7l5 m. Elev. 20g) ~30 qu JFMAMJJASOND Este- 4450 mm/yr -Pico del 18.7 °Clyr IOSO m Elev. -ZOU ~10 JFMAMJJASOND fi Figure 2 8 these measurements were not concurrent, only general trends can be surmized. 'The upper ridges and summits of the mountains are frequently enveloped in clouds, reducing solar insolation on El Yunque peak to just 60 percent of that in coastal areas (Briscoe 1966), and increasing the annual moisture available on the exposed summits by about 10 percent (Baynton 1968; weaver 1972). Transpiration and evapotranspiration decrease with altitude; relative humidity and wind velocity increase (Briscoe 1966; Weaver t al. 1973). The destructive force of hurricanes can impose long-term changes in the natural forest (Crow 1980). September is the month of most frequent occurrence followed by August and July (Fig. 3; Salivia 1972). Over 70 storms have occurred in Puerto Rico since the early 1700’s, many of which probably caused localized damage to vegetation in the Luquillo Mountains. Of these, 13 were A storms (vortex of hurricane passed directly over the island with hurricane force winds) and 10 were B storms (vortex of hurricane may or may not have passed over the island, but only part of the island experienced hurricane force winds), for an average of one storm over at least part of the island every 12 years (Fig. 4). 0f the A storms, four had trajectories directly over the Luquillo Mountains (Figs. 4 and 5). The last of these was San Cipriano in 1932. X£8££££i22. Because certain audiences are better acquainted with particular forest classification schemes (i.e., local names (Table 1) in Puerto Rico, Beard’s (1944, 1949) terminology in the Caribbean, and Holdridge’s (1967) life zone system in continental Central and South America, and on larger Figure 3. Monthly occurrence of 102 hurricanes in Puerto Rico. 1508-1970 (Salivia 1972). Oblique lines indicate hurricanes for which only the month was recorded. These were arbitrarily assigned to the middle of the month. 10 20P- mmcou_.c:: .o w o _ _ .emocmn 22 10" _ 5 mucou_.csz mo confisz m, FL. m. by Thirds Months Figure 3 Figure 4. ll Hurricane record for Puerto Rico from 1700 to 1970 (Salivia 1972). Long vertical I bar - type A storm (vortex of hurricane passes directly over the island with hurricane force winds); intermediate vertical I bar - type B storm (vortex of hurricane may or may not have passed over the island. but only part of the island experienced hurricane force winds); short vertical I bar - type C storm (vortex passed at some distance from the island. which experienced high winds (not hurricane force) and heavy rains); double I bar - two storms in the same year with intensity of second storm indicated by position of second I bar within the first; horizontal line from 1946 to 1981 - years of measurement of tree growth in the Luquillo Forest; black circle - type A storms that passed directly over the Luquillo Forest. 12 S :H HH H H H I'm-l HH PH 3 H m. h. mm 00 Oh Ow On com» 00 q muswwh Linguaa 13 I"ii-Sure 5. Path of type A hurricanes with trajectories directly over Puerto Rico from 1700 to 1970. Solid lines are known traJectories and dashed lines are assumed trajectories based on descriptions (adapted from Salivia 1972). 14 m shaman mam. nnm. ommmz monma uowvmcon “madam cmcmuansnas .>uumonom Hmowaoufi mo musuaumch museum Hmum>mm scum czmum moumawumm H m.mnm m.m mun hue a~a\~av sues“ «use mama omsuma coN-Ons m-unas onm1oom Aan\mav ua=Ho> mm-os on msuos msuoe Amn\~av mots Human mam ms-» o~-w canon Ass cease: unease comm ones coaa coma as\aamsm «Mesa aamm ovmuoaou condonma mummuom uouumm msumucso: caawsvsa was we mumwuom know ca musuusuum vcmum mo comaumaaou .~ macaw 16 Caribbean islands), approximate equivalents are indicated here. Local names are given first. Tabonuco forest is Beard’s lower montane rain forest, and Boldridge’s subtropical wet or rain forest, depending upon rainfall. Colorado forest is Beard’s montane rain forest, and Holdridge’s lower montane wet or rain forest, again dependent upon rainfall. Palm forest is Beard’s palm brake and an edaphic association in each of the aforementioned life zones of Holdridge. Finally, dwarf forest is also called dwarf by Beard and is an atmospheric association in either the lower montane wet or rain forests of Holdridge. Local forest names will be used hereafter to avoid confusion in text. ApprOpriate terminology from the life zone system will be indicated when there is a need to differentiate. Ascending the Luquillo Mountains from a northeasterly direction, four forest types are encountered. The tabonuco forest, which extends from the forest border through about 600 m, is dominated by Dacryodes excels , Sloanea berteriana, and Manilkara bidentata in the overstory. The colorado forest, ranging generally from 600 through 900 m, contains Cyrilla racemiflor , Micropholis chr so h lloid s, and M, garciniaegglig in the canopy. The palm forest, dominated by Prestgea montan , is found on steep slopes and arroyos above 500 m. Finally, the dwarf forest, a gnarled, epiphyte laden, and dense forest type with few tree species including Tabgbuia rigida and 0cotea s athulata, is located on the exposed slopes and ridges above 900 m. The colorado forest has two strata, evergreen foliage, and an abundance of superficial roots (Wadsworth 1949). A representative sample of trees greater than 10 cm in diameter on 4 be contained 60 species averaging 830 stems/ha, 95 percent of which were less than 45 cm in 17 diameter (Wadsworth 1951). Forests similar in physiognomy to the colorado forest, according to the Holdridge system, occupy about 100,000 km2 and constitute a little over two percent of the Caribbean and Central and South American region (Table 2). In addition, several islands of the Lesser Antilles not yet classified by the life zone system contain a relatively small area of "montane rain forest," which is Beard’s terminology for the same or a very similar forest type (Beard 1944, 1949). The most recent classification of the vegetation of the Luquillo Mountains (Ewel and Whitmore 1973) uses the life zone system (Holdridge 1967), and provides a basis for comparison among tropical forests in other areas. Along the same transect mentioned above, the following life zones are encountered: subtropical wet and rain forests at lower and intermediate elevations to the windward, respectively, lower montane rain forest on the summits, and descending to the leeward, lower montane wet forest, subtropical wet forest again, and finally, subtropical moist forest. Other classifications of the vegetation of the Luquillo Mountains are also available (Gleason and Cook 1927; Stehle 1945; Wadsworth 1951; Dansereau 1966). A brief description of the vegetation according to the life zone system follows: The subtropical wet forest life zone - This life zone is located at low to intermediate elevations and occupies 4150 he or about 37 percent of the Luquillo Forest. Mean rainfall ranges from 2300 to 3800 mmlyr. The vegetation tends to form a closed canopy at 20 m and is generally rich in tree species, epiphytic ferns and bromeliads. Orchids are common. WW - This life some is found at intermediate elevations to the windward where it occupies 1420 ha or 13 18 Table 2: An approximation of the areal extent of colorado forest elsewhere in Latin Americalz Geographic Area Area Percent of Country (reference) (ka) total area Caribbean Haiti (OAS 1972) 1.820 6.3 Dominican Republic (Tasaico 1967) 3.215 5-7 Puerto Rico (Ewel and Whitmore 1973) 110 1.2 Jamaica (Gray and Symes 1969)2/ 1.360 11.9 Subtotal 6,505 6.5 Central America Guatemala (Holdridge 1959a) 3.150 2.9 Honduras (Holdridge 19623) 2.430 2.1 El Salvador (Holdridge 1959b) 340 1.6 Nicaragua (Holdridge 1962b) ----- --- Costa Rica (T031 1969) 4,100 7.8 Panama (FAO 1971) ““r 22:. Subtotal 10,020 1.9 South America Venezuela (Ewel et a1. 1968) 8.800 0.9 Colombia (Espinal et a1. 1963) 45.220 4-0 Ecuador (Vivanco de la Torre et a1., no date) 6,490 2.4 Peru (T081 1960) 22,430 1.7 Subtotal 82,940 2.2 Grand Total 99,460 2.3 1/ 2/ Areas determined as Lower montane wet forest according to Holdridge (1967) life zone classification. Information based on climatic and topographic data only. 19 percent of the forest. Mean rainfalls exceed 3800 mmlyr. The vegetation forms a closed canopy at 20 m. Palms and epiphytes are more abundant than in the previous type. The lower montane wet forest life zone - This life zone is located mainly to the leeward of the summit areas and occupies 4300 ha or 38 percent of the forest. Mean rainfall ranges between 2900 and 3800 mm/yr. The vegetation varies from 12 to 15 m.in height and is characterized by reddish-brown, corriaceous leaves. Palms and epiphytes are abundant. Ihgglowerggontage rain forest life zone - This life zone is concentrated on the windward ridges and summits of the Luquillo Mountains and occupies 1180 ha or about 10 percent of the forest. Mean rainfall exceeds 4300 mm/yr. The vegetation is similar to that of the above life zone but has more epiphytes. The dwarf cloud forest association, principally situated on the exposed peaks, has few tree species. The subtropical moist forest 1ifg_gggg_- This life zone is found in a disjunct section of the forest southwest of the main portion and occupies 150 he or about one percent of the forest. Mean rainfall is slightly under 2000 mm/yr. Scattered remnants of the zonal vegetation are characterized by trees to 20 m in height, many of them deciduous during the dry season. Species diversity is high. In the Luquillo Mountains, the number of trees and basal area per hectare tend to increase, whereas mean height of the dominant trees, mean and maximum tree diameter, and the number of tree species all tend to decrease with an increase in elevation (White 1963). Comparative structural information for the montane forests of the Luquillo Mountains is shown in Table 1. 0f the 225 tree species native to the forests of the Luquillo Mountains, 41 percent are also found on the continents (South 20 America, Central America, Mexico and Florida), 28 percent are found in the West Indies but not on the continents, and 31 percent are endemic to Puerto Rico (Little and Woodbury 1976). Past Research in the Luquillo Forest The forests of the Luquillo Mountains rank among the best studied trapical forests in the world with numerous publications on the flora, fauna, and related topics. A complete bibliography on research does not exist but citation of a few major works will point out the extent of the investigations. From 1968 through 1977, beginning with an article introducing the research (Howard 1968), some 17 studies appeared in the Journal of the Arnold Arboretum concerning the dwarf forest in the vicinity of Pico del Oeste. Wadsworth (1970) cited 118 references on geography, climatology, geology, flora, fauna, ecology, archeology, agriculture and forestry for the Luquillo Mountains. Odum and Pigeon (1970) presented a large compendium of information on plant and animal life and the effects of radiation, mainly in the tabonuco forest. A more recent history of research and opportunities for additional investigation (Brown g£_§L, 1983) listed many of the previous works as well as studies completed since their publication. LITERATURE CITED LITERATURE CITED Baynton, R. W. 1968. The ecology of an elfin woodland in Puerto Rico, 2. The microclimate of ’Pico del Oeste.’ Journal of the Arnold Arboretum 49 (4): 419-430. Beard, J. S. 1944. Climax vegetation in tropical America. Ecology 25: Beard, J. S. 1949. Natural vegetation of the windward and leeward islands. Oxford Forestry Memoirs 21: 1-192. Beinroth, F. H. 1971. The general pattern of the soils of Puerto Rico. Transactions of the Fifth Caribbean Geological Conference, Geological Bulletin No. 5. Queens College Press, New York, p. 123-229. Benedict, F. F. 1976. Herbivory rates and leaf properties in four forests in Puerto Rico and Florida. M.S. thesis, University of Florida, Gainesville. 78 p. Briscoe, C. B. 1966. Weather in the Luquillo Mountains of Puerto Rico. Institute of Tropical Forestry Research Paper ITF-3. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 250 p. Brown, S., A. E. Lugo, S. Silander and L. H. Liegel. 1983. Research history and opportunites in the Luquillo Experimental Forest. USDA Forest Service General Technical Report 80-44. Southern Forest Experiment Station, New Orleans, La. 128 p. Crow, T. R. 1980. A rain forest chronicle: a thirty year record of change in structure and composition at E1 Verde, Puerto Rico. Biotropica 12: 42-55. Crow, T. R. and P. L. Weaver. 1977. Tree growth in a moist tropical forest of Puerto Rico. USDA Forest Service Research Paper ITF-22. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 17 p. Dansereau, P. 1966. Studies on the vegetation of Puerto Rico. 1. Description and integration of plant-communities. University of Puerto Rico Institute of Caribbean Science Special Publication No. 1. Mayaguez, Puerto Rico. Espinal, L. S. and E. Montenegro (with J. A. Tosi, Jr. and L..R. Holdridge). 1963. Formaciones vegetales de Colombia; y mapa ecologico de Colombia. Instituto Geografico "Augustin Cadazzi." Bogota. 201 p. +-map. Ewel, J., A. Madriz, and J. A. Tosi, Jr. 1968. Zonas de vida de 21 22 Venezuela; y mapa ecologico de Venezuela. Ministerio de Agricultura y Cria, Direccion de Investigacion, Caracas. 265 p. + map. Ewel, J. J. and J. L. Whitmore. 1973. The ecological life zones of Puerto Rico and the U.S. Virgin Islands. USDA Forest Research Paper ITF-18. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 72 ‘p- FAO. 1971. Inventariacion y demostraciones forestales, Panama. Zonas de Vida, basado en la labor de Joseph A. Tosi. FO:SF/PAN 6. Informe tecnico 2. Rome, Italy. 89 p. + map. Gleason, H. A. and M. Y. Cook. 1927. Plant ecology of Porto Rico. Vol. 7., Parts 1 and 2, p. 1-173. In Scientific Survey of Porto Rico and the Virgin Islands. N.Y. Academy of Sciences, N.Y. Gray, D. M. and G. A. Symes. 1969. Forest inventory map No. 3, life zones, 1:500,000. U.N.D.P. Forestry/Watershed Management Project-Kingston, Jamaica. Holdridge, L. R. 1959a. Maps ecologico de Guatemala, A.C. Interamerican Institute of Agricultural Science, Turrialha, Costa Rica. Holdridge, L. R. 1959b. Maps ecologico de El Salvador. Interamerican Institute of Agricultural Science, Turrialha, Costa Rica. Holdridge, L. R. 1962a. Maps ecologico de Honduras. Organization of American States, Washington, D.C. Holdridge, L. R. 1962b. Mapa ecologico de Nicaragua, A. C. USAID, Managua, Nicaragua. Holdridge, L. R. 1967. Life zone ecology. Tropical Science Center, San Jose, Costa Rica. 206 p. Howard, R. A. 1968. The ecology of an elfin forest in Puerto Rico, 1. Introduction and composition studies. Journal of the Arnold Arboretum 49 (4): 381-418. Little, E. L. Jr. and R. O. Woodbury. 1976. Trees of the Caribbean National Forest, Puerto Rico. USDA Forest Service Research Paper ITF-20. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 27 p. Meyerhoff, H. A. 1933. Geology of Puerto Rico. University of PuertO' Rico Monographs 1. Rio Piedras, Puerto Rico. 306 p. Mitchell, R. C. 1954. A survey of the geology of Puerto Rico. Technical Paper 13, 167 p. University of Puerto Rico Agricultural Experiment Station, Rio Piedras, Puerto Rico. OAS (Organization of American States). 1972. Ecologie, Republique D’Haiti. Washington, D.C. 23 Odum. H. A. and R. F. Pigeon (eds.). 1970. A tropical rain forest. USAEC. TID-24270. Salivia. L. A. 1972. Historia de los tanporales de Puerto Rico y las Antilles (1492 a 1970). Editorial Edil. Inc.. San Juan. Puerto Rico. 385 p. Stehle. H. 1945. Forest types of the Caribbean Islands. Part 1. Caribbean Forester 6 (Supplement): 273-408. Tasaico. H. 1967. Reconocimiento y evaluacion de los recursos naturales de la Republics Dominicans: estudio para su desarrollo y planificscion (con Maps Ecologico de la Republics Dominicans 1:250.000). Pan American Union. Washington. D. C. Tosi. J. A.. Jr. 1960. Zonas de vida natural en el Peru: memoria explicativa sobre el maps ecologico del Peru. Boletin Tecnico. No. 5. Instituto Interamericsno de Ciencias Agricolss de la 0. E. A.. Eons Andina. Lima. 271 p. Tosi. J. A.. Jr. 1969. Republics de Costa Rica. maps ecologico. Tropical Science Center. San Jose. Costa Rica. Vivanco de la Torre. 0.. M. Cardenas Cruz. G. Cortsire Iturrside and J. A. Tosi. Jr. (no date). Croquis ecologico psrcisl del Ecuador IERAC. Quito. Ecuador. 0. S. Department of Commerce. 1970-82. Climatological data - Puerto Rico and the Virgin Islands. Volumes 16-28. National Climatic Center. Asheville. N.C. Wadsworth. .. a. 1948. The climate of the Luquillo Mountains and its significance to the people of Puerto Rico. Caribbean Forester 9: 321-335e Wadsworth. F. H. 1949. The development of the forest land resources of the Luquillo Mountains. Puerto Rico. Ph. D. Dissertation. University of Michigan. Ann Arbor. 481 p. Wadsworth. F. H. 1951. Forest management in the Luquillo Mountains. 1. The setting. Caribbean Forester 12: 93-114. Wadsworth. F. H. 1970. Review of past research in the Luquillo Mountains of Puerto Rico. p. 33-46. In H.T. Odum and R.F. Pigeon. eds. A tropical rain forest. USAEC. TID-24270. Walter. H. 1975. Vegetation of the earth in relation to climate and the soc-physiological conditions. Springer-Verlag. New York. 237 p. Weaver. P. L. 1972. Cloud moisture interception in the Luquillo Mountains of Puerto Rico. Caribbean Journal of Science 12: 129-144. Weaver. P. L.. Mu D. Byer and D. Bruok. 1973. Transpirstion rates in the Luquillo Mountains of Puerto Rico. Biotropica 5: 123-133. 24 Weaver. P. L. 1983. Tree growth and stand changes in the subtropical life zones of the Luquillo Mountains of Puerto Rico. USDA Forest Service Research Paper 80-190. Southern Forest Experiment Station. New Orleans. LA. 24 p. White. H. H. Jr. 1963. Variation of stand structure correlated with altitude. in the Luquillo Mountains. Caribbean Forester 24 (1): l‘6"52e PART I CHANGES IN FOREST STRUCTURE AND GROWTH AFTER HURRICANE DISTURBANCE 25 26 INTRODUCTION Cyclones and hurricanes accompanied by heavy rains occur commonly in the Caribbean and have an impact on forest ecosystems and local economies (Howard 1962; Salivia 1972; Tomblin 1981). When the trajectories of 761 cyclones since 1886 with windspeeds between 40 and 120 km/hr are plotted on a basin-wide map of the Caribbean, all of the islands are obscured by black ink (Neumann g; 51, 1978). About 100 hurricanes passed over or very close to Puerto Rico between 1508 and 1970 (Salivia 1972); those occurring in 1766, 1772, 1867 and 1932 passed directly over the Luquillo Mountains. Defolistion, breakage and windthrow are common effects of hurricanes in forests (Lugo g£_§;, 1983; Wilkinson _£_gl, 1978). The severity of disturbance is related to storm intensity (Sauer 1962), forest structure (Wadsworth and Englerth 1959), and soil conditions (Cremer g;_§l, 1977; Wilson 1976; Wood 1970). Understanding the response of ecosystems to random events such as hurricanes requires aedequate sampling on permanent plots (Bormann and Likens 1979; Callahan 1984). Such monitoring in the tropics today is limited to Venezuela (Veillon 1983) and Puerto Rico (Lugo and Brown 1981; Weaver 1979, 1983), where records are available for 30 and 40 year periods, respectively. Partial analysis of the Puerto Rican data has provided limited information on growth and changes in forest structure and species composition over time (Crow 1980; Weaver 1983). However, this study is the first to provide a detailed analysis of all data sets for a specific forest. 27 In 1928, San Felipe passed over Puerto Rico southwest of the Luquillo Mountains, and in 1931, San Nicolas passed just north of the island. The only precise data available on recovery after San Felipe indicated the elapsed time before certain tree species refoliated in the Luquillo district (Bates 1929). Both presumably caused localized damage within the forest but no description of this damage exists. San Cipriano of 1932 passed directly over the Luquillo Forest with winds of 200 km/hr and rainfalls in excess of 40 cm (Crow 1980). Again, no direct observations of damage in the forest are available. Cyrills racemiflgra is long-lived and reaches large diameters in the colorado forest. The recent loss of several large trees suggested that its regeneration might be associated with past disturbances, particularly destructive hurricanes. Because it survives for long periods, its diameter class distribution might be compared with hurricanes during the past three centuries. Eight permanent plots, established in the colorado forest in 1946-47, were used to test the hypothesis that recovery from substantial damage would cause changes in forest structure and species composition. In addition, growth data for Cygillg were used to estimate the relationship between its diameter and age. The colorado forest type is widely distributed in the Caribbean Basin and in Central and South America (Table 2). A knowledge of its reaction to disastrous hurricanes is important not only to ecologists as they refashion their thinking on the function of disturbance and succession in natural ecosystems (Cairns 1980; Ewel 1980; Pickett and White 1985), but also to foresters whose management policies and silvicultural practices may be modified by these catastrophic events. 28 METHODS The eight permanent 0.4 ha plots used in this study were established in the mid-1940’s to determine diameter growth. Three plots, located on lower slope and valley topography ranging from 650 to 750 m in elevation, occasionally flooded in riparian areas. Four plots, located on lepe topography ranging from 700 to 900 m in elevation, had better drained soils. The remaining plot, on slope topography at 670 m in elevation, was thinned at the beginning of the study. Its basal area was reduced from about 40 to 20 mz/ha providing a balance between large and small individuals for sustained yield. CanOpy trees were allowed at least 2 m of crown freedom on all sides (Wadsworth 1957). Hereafter, the term natural forest will be used to indicate a plot where no silviculture was practiced -- that is, no thinning or stand improvement. The thinned stand will be referred to as thinned. All stands were subjected to natural disturbances such as high winds, heavy rains and hurricanes. All trees 24.1 cm at breast height (dbh, or 1.4 m above the ground) were measured to the nearest 0.1 inch (0.25 cm). Height was estimated visually to the nearest 0.5 m and checked occasionally by abney level on all natural forest plots. Height was not estimated on the thinned plot. Crown classes were recorded as dominant, codominant, intermediate or suppressed (Baker 1950). Dominant trees rise above the general level of the canopy and receive light from above and to some extent laterally. Codominant trees reach the level of the canopy and receive light from 29 above. Intermediate trees are below the general level of the canopy but receive some direct light through gaps. Suppressed trees are overtopped and receive no direct light. Some of the plots were visited again in 1951, 1956 and 1976, but neither tree nor stand measurements were complete. Species were tallied, but height was not recorded and ingrowth (recruitment, or trees that grew into the minimum diameter class) was partially recorded but trees were never tagged. Detailed tree measurements on all plots were made again in 1981. Diameter was measured to the nearest 0.1 cm and height was determined by rsngefinder to the nearest 0.1 m. Untagged trees were assessed by their diameter and position as either old trees that had lost their tags or ingrowth that had not yet been tagged. Missing trees were recorded as mortality. Basal area, volume and biomass were determined for each plot. Heights for use in volume and biomass estimates on the thinned plot at the time of establishment, and for the determinaton of volume and biomass of dead Stems from mortality which occurred in the years 1951, 1956 and 1976, were estimated by a height vs. diameter curve based on data for dicot BINF-Cties from other plots where diameters had been measured and heights e‘t imated simultaneously. Trees that died between measurements were ass“med to continue diameter growth at the rate of the mean tree in the 'tand for half of the interim between the last live measurement and the time when the tree was recorded as dead. With palms, mean height growth ”as used in the same manner. Statistical testing for differences in mean annual diameter increment (HA1) by crown class was analyzed with Duncan’s multiple range test (Steel a nd Torrie 1960). Dicot tree volumes, including branches _>_2.5 cm in 30 diameter, were determined using an existing equation derived in English units for the construction of volume tables in the colorado forest (Wadsworth 1949). This equation would likely be useful for volume estimates in similar forests in the Caribbean region. The equation converted to metric units is: Volume (m 3) - (0.000074931 Dl'8748 H0°9165. The rzfor the original equation is not available, but the relationship was derived from 302 trees of 35 species and had a mean deviation of 14.83 percent and an aggregate difference of 1.07 percent (Wadsworth 1949). These are mensurational measures (Chapman and Meyer 1949) and are defined thus : mean deviation, the average of the differences or deviations of single observations from the means of these observations, without consideration of the sign of the differences (equal to approximately 80 percent of the standard deviation); and aggregate difference, 100 times (the sum of actual tree volumes minus the sum of estimated tree volumes) divided by the sum of actual tree volumes. It indicates goodness of fit 0f sample data to the developed volume table charts. Aboveground woody l>iolnsss estimates were then derived by multiplying individual tree volumes by their respective specific gravities (Jordan and Farnworth 1982). The la'iter were obtained from published sources (Little and Wadsworth 1964, Litt 1e g 2;. 1974) or from the Forest Products Laboratory in Madison, WiseOnsin. A few small, rare or endangered species, were arbitrarily llssigued a value of 0.6 g/cm3, the weighted average specific gravity for the forest. Palm volumes were determined by multiplying cross sectional ‘re‘ by height, then adjusted for biomass by multiplying with-specific gravity. 31 Tree diameters were grouped into 5 or 10 cm classes (except for the smallest and largest classes), tree heights into 3 m classes, and specific gravity into decimal classes, in both 1946 and 1981, to observe trends in forest recovery and growth. Changes in crown classes were also observed. Age of Cyrilg racemiflora trees was estimated as follows (Osmaston 1956): (1) Six 10 cm diameter classes were selected between 20 and 8) cm. The seventh and largest diameter class, >80 cm, was chosen so that its midpoint was equal to the mean diameter of the measured trees within it; (2) For diameter classes _>_20 cm, MAI was determined for all trees within the respective classes. Those trees not attaining a growth rate at least 50 percent of the class mean were eliminated (Webb 1964) under the assumption that they would not grow into the next diameter class; (3) MAI of the ingrowth on the thinned plot was used for the diameter class 0 to 19.9 cm. All trees were assumed to have regenerated at the time of thinning, or to be of maximum possible age, that is, 34 years old. MAI was calculated for all trees. Again, those trees with diameter increment 350 percent of the diameter class mean were used to calculate a new mean for the diameter class under the assumption that those indi-Viduals were the oldest on the plot and most likely to survive into the next diameter class. The new mean was used as the growth rate for trees from 0 to 19.9 cm dbh; (4) The MAI’s for diameter classes 320 cm were then plotted vs. the midpoint (i.e., the calculated mean for all stems) of each diameter class. 2.20 A least squares linear regression was fitted to the data points Cm; 32 (5) Adjusted or fitted MAI’s were read from the curve at the midpoint of each diameter class 320 cm and divided into the width of the class to yield the time in years required for the mean tree to move through the class; (6) The cumulative time for the mean trees to move through the respective diameter classes was then determined; (7) Finally, an age curve was generated by plotting the upper endpoints of the diameter classes on the ordinate axis vs. the cumulative ages on the abscissa. Graphing in this manner facilitates comparison of colorado growth with diameter/age relationships of other species. The age curve was then compared with the diameter class distribution of Cirilla to determine if its regeneration pattern was correlated with past hurricane disturbance in the colorado forest. Percent importance values based on the relative density, relative frequency and relative basal area of each species (Kershaw 1973; Brown and Curt is 1952) were determined for 1946 and 1981 to elucidate species trends. Percent similarity, the summation of the lowest percentages (PErcentage - the number of individuals per species/total number of indiViduals per plot) by species within each pair of stands being compared (Brower and Ear 1977) was used to estimate community similarity amng plots. The Shannon-Wiener Index (Krebs 1972), which is a function of both the number of species (variety) and their relative abundances (equitability), was used to determine species diversity on natural forest plots - A species count was also used. Trends in species composition were studied by following the survival and changes in stem numbers and basal area of ’indicator’ species which re presented a spectrum of ecological roles in the colorado forest 33 community. The species selected (Little and Wadsworth 1964; Little gt; al. 1974) were: Hedyosmum arborescens and Psychotria berteriana, small, short-lived, understory trees maturing at 10 cm in diameter and 6 m tall, found mainly in small openings in the forest; Cecropia peltata and W m, common, intermediate to large-sized pioneer trees, found in secondary forests and in gaps in primary forests in the American tropics, growing to 30 to 50 cm in diameter and 15 to 20 m tall; Miconig lagxigata, a small to intermediate-sized tree, reaching 15 cm in diameter and 8 m tall, widely scattered in the lower part of the colorado forest mainly in small openings; Cyrilla racemiflora, a long-lived, large, canopy tree, occasionally attaining sizes of 100 cm or more in diameter and 15 m tall, whose regeneration is tied to gaps in the mature forest; Cord ia borin uensis, a common, shade tolerant, understory tree, to 15 cm in diameter and 7 m tall, capable of regenerating and growing under a canopy; Prestoea montsna, a shade tolerant canopy palm reaching 15 m tall, most common on steep, unstable soils and along drainages throughout the colorado forest; and Micronholis crysoghylloides and M. W, both shade tolerant, canOpy trees reaching 50 to 60 cm in diamEter and 15 m tall. Rare and endangered species were tallied 'eParately by topography and elevation to elucidate their distributions within the forest. Species nomenclature used was that of "Common trees of Puerto Rico and the Virgin Islands" (Little and Wadsworth 1964; Little t al. 1974). 34 RESULTS Stand Changes Ingrgggh, mortality and number of stems The number of stems at the beginning and end of the study for all natural forest plots combined was virtually the same. The means and standard errors of stem numbers/ha were 1860 1 73 in 1946 and 1858 1 89 in 1981 (Table 3). Ingrowth averaged 20.0 1 2.2 stems/ha/yr and mortality averaged 20.0 _1 0.8 stems/ha/yr. Annual flux for ingrowth and mortality averaged 1.1 percent of the original stem density. When stem density on plots is considered by topography, the four slope plots declined by 6 percent while the two valley plots increased by 9 percent from 1946 to 1981. The. plot with mixed slope and valley t°l>Ography also increased in stem density by 2 percent. In contrast, the previously thinned plot had three times more ingrowth than mortality. Survival of stems measured in 1946 was highest on the thinned plot with over 70 percent while it varied between 50 to 70 Patcent on the natural forest plots. Wi- tg; class distribution The smallest diameter class on the natural forest plots showed a 15 bel'cent decline from nearly 1060 stems/ha to about 900 stems/ha (Fig. 6). In contrast, the 10.0 to 14.9 cm diameter class increased by 33 percent or 100 trees/ha. The remaining classes between 15.0 and 49.9 cm, with the 35 as: H mono «shunne> one a one u mucus amouua> one omens cu m uonm «monou- oua a smacks» c euoum .etaea nn\xasseeoe eea_\»usaqetezc . a sues-ate: .asaeeee eee~\aaeeaeeees - hes-eon cease . n Heeaeuem .mocuaaoo mucus ueouou aeusues an. new use: .eeeea en\2»sa.e.e eeeA\eeeeteeHc . u assesses n c n u .eeeeses — ne.~ n.an so.o e.ns n.aa sown need one Heed H no.” ~.~ue.o~ we.“ u.a«o.e~ n.~e canono_ ceases «Nu—ca fiancee” menu: en.“ a.e~ oa.a n.a~ e.ae mean one ens sees use: on.~ ~.a~ ma.a m.- e.oe meHN mmm. mmm. owes a an.~ o.n~ eo.a e.o~ a.~e -H~ ecu “an amen u a~.a ~.e~ ea.a n.es «.mn need one ewe amen u oo.o e.ea as.“ e.ea e.ne oeaa can one eeo_ ens: Ho.a n.ea eo.a a.ma e.ee new" _mmm mmm mom” a o~.~ e.a~ on.a m.n~ ~.sn noes use was seen u n~.o n.na eo.o e.ea m.ne fines Nee use noes n an.o e.~a eo.o n.oa n.oa ewes Nee Nee Sega a vuuxu nuhxeaerUum nuaxn aux-axeasum u ~oa— susouucn huuaeuuox 0cm“ auscuncn huqneuuox He>w>usm Aesxaleuev sauna-sen loam uenm N ecu uou coauuuaoc lou- .~ca_ on com. seem euega season evaneNeo as. 2.. 5:58 3:23 .232... faces .3. sense .n is 36 Figure 6. Diameter class distributions for seven natural forest permanent plots and one thinned permanent plot in the colorado forest in 1946 and 1981. 37 1-122.“. assist. momma-u 33:85..- I unwound II---?.o.a 7.03... .2952.-- «sumo-u 3.9.35-2.--) on, 53311 p IaquInN Say... I no. 0..—moon bxb, 1.53 noun-.5 522.33-- 0.30. I 0.00. 2 0.00 I 0.8 a 0.00 I 0.00 n. 0.05 I 0.0x. v. 0.00 I 0.00 9 0.00 I 0.00 . 0.0V I 0.00 I 0.3 I 0.0m 0 0.0m I 0.0m k 0.0N I 0.0N u 0.?N 1 0.0N 0 0.0. I 0.0. U 0.: I 0.0. 0 0.0 I ..V d «3:23 .02 D 3:350 ova. \\\. neoooa ( ““31 I OON 00v \\\“\\\\\\\\\\\ l “\\“\\\\\\\\\\\\\\ l I can 000 I DON. OOON 83 on] $031110 Isqmnu 38 exception of the 25.0 to 29.9 cm class, also showed an increase in stem numbers ranging from 4 to 18 trees/ha. The diameter classes _>_50 cm showed either a slight decline in the number of stems, or remained about equal. All of the plots, regardless of topography, showed a decrease in the Bulallest diameter class ranging from 11 percent in the valleys through 13 Peteent on the slopes to 23 percent on the mixed plot (Fig. 7). Moreover, all showed an increase in the 10.0 to 14.9 cm diameter class varying from 13 percent on the slopes through 61 percent in the valleys to 66 percent on the mixed plot. The valley plots showed an increase in all diameter classes from I .0 to 49.9 cm whereas the mixed plot remained about equal 01‘ increased in these classes. The slope plots showed more variability, but the trend was toward an increase. Diameter classes 250 cm for all groupings of plots fluctuated slightly between measurements but demonstrated a general decline with mortality of larger stems. In contrast, the thinned plot showed an increase of stems in all 6 iameter classes except for a slight decline in the 35.0 to 39.9 cm class. 43 ight class distribution The trend in height class distribution on the natural forest plots IDal‘allels that of the diameters. The smallest class, 0 to 2.9 m, decreased between 10 and 84 percent, with the smallest decline on the mixed plot (Fig. 8). The next size class, from 3.0 to 5.9 m also showed a deg line, ranging from 35 to 81 percent. All remining size classes increased in numbers of stems, some of them considerably. The most not able gains were in the 9.0 to 11.9 m and 12.0 to 14.9 m classes in the V‘ 1 ley plots . 39 F1sure 7. Diameter class distributions by topography for natural forest permanent plots in the colorado forest in 1946 and 1981. ll'Io Number of Trees Figure 7 200 l 200 600 400 200 ITVTTIWT IZOO 10 600 400 200 ABCDEFG 4O Slope Pl oIs 40 l- 20 Mlaed Slope and Valley PloI 40- H Volley Plots 40 20- ABCDEFG H M JKLMN Legend I l946 DenslIles D 1901 DenslIles "blender Classes (cm )- a 4.1 - 9.9 5 b 23.7“ ZZsz-zannoou SBSBSSfiflSsBs oooooooooboo .'00000..... §3$3$$$$$32 bouoooooooooo JKLMN ---------------- Diomeler Classes -------------- - 41 Figure 8. Height class distribution by topography for natural forest permanent plots in the colorado forest 1946 and 1981. 2.3219 220... 42 ou/ seen, Io Isqmnu w muswwm x a u u o u m d u u o u m < a u w o u m < p n p 0 pl .l p l. 0 .IIIIJ— -J all I I 8. I L e8 L I I 8' 98 -o... a I I can 98 -e... o I e: -e... a a... .9! u I I 80 a... - as e .5 .e.. 9 ed -ed a I L 02. Oi Io 1 f 2... 25:33.... :2... I I Iooo £52.... 3: B 3:23.. :3. I I I Iooe sees-4 L I L o .e... 3:2, ea . 2e... 3:2, 2:. 32.. see.» one.” eel! 43 Because heights were not measured on the thinned plot in 1946, changes of height classes during the period of measurement could not be determined. Crown class distribgtigg_ The dominant and intermediate crown classes on all the natural forest plots decreased by about 70 stems/ha, or 3 percent, from 1946 to 1981, while the suppressed crown class increased by about 175 stems/ha, or 6 percent (Fig. 9). The codominant crown class remained virtually unaltered. Despite these changes, the pattern of crown class distribution in both years is similar; 45 to 50 percent in the suppressed class, 25 to 30 percent in the intermediate class, and 10 to 12 percent apiece in the dominant and codominant classes. In contrast, the previously thinned plot had 60 percent of the stems in the intermediate crown class in 1946 but only about 10 percent in 1981. The suppressed class increased from about 10 percent to over 50 percent in the same period of time. The percentage of stems in both the dominant and codominant crown class approximately doubled. §pecific gravity distributing; About 35 percent of the stems on the natural forest plots in 1946 and .1981‘were found in the specific gravity class from 0.60 to 0.69 g/cm3 ‘Uith another 30 percent between 0.70 and 0.79 g/cm? (Fig. 10). When ‘basal area or volume was substituted for number of stems, however, 40 to 43 percent were found in the 0.50 to 0.59 g/cm3 class in 1946 and about 30 percent in the same grouping in 1981. The 0.60 to 0.69 g/cm3 class had about 28 percent in 1946 and about 35 percent in 1981. 44 Figure 9. Crown class distribution for the natural forest and thinned permanent plots 1946 and 1981. 45 a shaman 3305 c305 3308 5505 m u o m . U o _. - E J N sl. om. “w. I] m T . ..... 08.. v .68.. m. I. I l I I m l J .1 FL oommucaasm a m I MW ooo.~ l 20.3....25 . . IOOQN n 7355330 u u Eocho u o .3938 5595 _mm. D m . o o 3966 5.85 sea. I e n no u 1 - u. o~ fl. nausea Io~ m. T I I 0.. 1 10.. w. l JA 1‘ cm 1. 00. J r. .05 e232... 32¢ None...— .ocs.oz U “u S 46 Figure 10. Specific gravity by groupings (0.25 = 0.2 to 0.29. etc.) for number of stems. basal area. and volume in 1946 (shaded) and 1981 (unshaded) on the natural forest and thinned permanent colorado plots. Slems Percenl of Area of Basel Percent of Volume Percent 47 70- b 60- - 5 L Nciurol Foresl Plots 1mm.“ 0 " Plot 40— - a 30+- F- 20— - .0_E-i i— E14 '1 lzi1flqu 1511141‘ 1’ riai ’r I 50r- i- 40 - 30 L. 20- — 'OLH a] — a] 0' djfl— T T 4F T ' I 'I’ 50 L 40- - 30- - 20— - l0- 3] a] i- .25 .35 .45 .55 .65 .75 .65 .95 .25 .35 45 .55 .65 .75 .65 .95 ----------- Specific Gravity Decimal Midpoinls --------------- Figure 10 48 The weighted mean specific gravity for all trees, based on their respective volumes, showed a 1.9 percent increase from 0.563 to 0.574 g/cm3 between 1946 and 1981 for the entire forest. This estimate includes palms which have a low specific gravity. When palms are excluded from the sample, mean specific gravity for s11 dicots, based on their respective volumes, showed a 3.9 percent increase from 0.608 to 0.632 g/cm3. The thinned plot had fewer specific gravity classes than the natural forest plots mainly because of the selective removal of species. The increase in the percent of stems in both the 0.50 to 0.59 and 0.90 to 0.99 g/cm3 classes reflected the massive influx of Cyrillg ragemiflogg and Clusia kru iana, respectively. The decline in the 0.60 to 0.69 class was due mainly to the mortality of Microphplis garciniaefolig. When considered on the basis of basal area and volume, the pattern in the 0.50 to 0.59 class was reversed because much of the Cyrilla ingrowth was of small diameter, and was offset by the mortality of larger stems in the same specific gravity class. The mean specific gravity for all stems showed an increase from 0.60 to 0.66 g/cm3. The change in mean specific gravity was greater than that on the natural forest plots because of the influx of Clusia and the abscence of palms. Chapges in Species Composition Sprviyal of indicator species Nearly all of the Hedyosmum arborescens and all of the Psychotrig berteriana stems originally found on the natural forest plots in 1946 had disappeared by 1981 (Fig. 11). When ingrowth was added to survival, however, the 1981 populations of these species amounted to 10 and 19 49 Figure 11. Percent survival of indicator species in the colorado forest from 1946 to 1981. Lines indicate survival of original trees. Dot is 1981 population of trees (survival of residual trees and ingrowth). Bars indicate basal area: shaded bars. 1946 basal area; unshaded bars. 1981 basal area survivor and growth on survivors; and obliquely-lined bars. 1981 total basal area. growth of residual basal area and ingrowth. Number of trees/ha are indicated for slope (S). slope and valley (M). and valley (V) plots. 50 2.32283 32215335 oceEoE 3303023.. 22:630. 2 . 23222 2.23325 ooofiocn 2Eou o. . tau .0 we .0 0v .0 we .0 0v .0 we .\ i o e \ L “ I I I IS a l \ L /_ l LOO ./ w [ 0 / 0 - m . I - . I Lom. 87> “ No -> e_e-> 8L, t-> vow-s... . ~.-2- «HE-2.. o -2. No-Soo. hmm-m mS-m om_-m mo-m mm-m I - .. .2 com 2022.2 2.2232: 2oz: 2.2.2.3.. 2.33.395 2 :3 :2 soc—.8535 2a 230 2 2 Scores EsEuouooz - - - . o O O I .4 4 10' I I . I Ion me -> n~ -> I ms ->I ow->I n ->Io~. n. -3; o -x s -2 3-2 I. .2 B -m o -m. o -m mN-m N» -m so seamen (oan Iosoq puo seen) Int-guns Iuaslad 51 percent, respectively, of the 1946 populations. Moreover, the percent importance of Hedyosmum and Psychotria had declined from 1.4 to 0.5 and from 1.5 to 0.7, respectively (Table 4). The 1981 total basal area (growth on residual trees + ingrowth) was 30 percent of the 1946 value for Hedyosmum and 23 percent for Psychotria. Cecropig_peltapg, Didygopanax morototoni, and Miconia laevigaga also declined to 20 percent or less of their 1946 densities. Their percent importance values decreased from 1.2 to 0.7, 0.4 to 0.3, and from 1.5 to 0.9, respectively. Regeneration of these species between measurements ranged from 0 to 10 percent of their 1946 populations. The 1981 total basal areas, however, were higher than for the previous species, and ranged from 80 to 110 percent of the 1946 basal areas. This was due mainly to the larger size and longer survival of these species. Cyrilla racemiflgra lost 45 percent of the stems present in 1946. Regeneration of new stems, however, left the total population of trees only slightly lower in 1981. Total basal area in 1981 was only 65 percent of that in 1946. Since many of the stems that perished were large in diameter, the percent importance declined considerably from 11.9 to 7.5. Cordig,boringuensis showed a survival rate of 80 percent for its 1946 population. With regeneration, this species increased to 130 percent of its original population. Total basal area in 1981 was nearly 130 percent of that in 1946 due to ingrowth and increment on residual stems. Its percent importance also increased from 2.2 to 2.6. In contrast, Prestoea montana, Micrgphglis garciniaefglia, and M, ghrysophyllgides showed survivals ranging from 65 to 90 percent of their 1946 populations. When ingrowth was added to the original stem numbers, the final populations of these climax species ranged from 100 to 120 52 a.e eee.o hae.e eoo.e n._ ~oe.e e~e.e -o.e .aa-«uuuuua.aquuaaau-u o.o eoe.o aao.e noo.e a._ eeo.e n~o.e use.e .uu-aquuua.-«auugu ... cae.e cae.e ._e.e e._ _ae.e nae.e «_e.o .uuquau.uqamummm H.“ nee.e .«e.e e-o.e e.~ eoe.e nae.e nae.e .uaaauuuqqeu.u«uuummumm e.« nae.e o~e.e eee.o ~.~ e_e.e o~e.o one.e .uquauauuquum eaeeeu n.n nee.e eae.o ane.e ~.~ eao.o nae.o a~o.o .uaqauuuuaa-u.uaauauamuuu 9... 9e... .3... .3... 2 3e... no... So... g a” ~.~ n_e.e sue.e nae.e n.n _ne.e nae.e «ne.e .uuumaaquuu.uummmm e.n ame.e o~o.e use.e e.e _no.c nae.e one.e .uaauauuqmumu.ammmmm .6 Se... e36 236 a... 3:... e36 ~86 gage emu-edger. e.o ao~.e cae.e o~_.e n.e ne~.e nae.e -_.e .umuummm.uummuumm ea .8... .36 .26 ea 2.... n3... 5... Nani-ii? 3 a... ..~.e cae.o n...e n.e_ e._.e nae.e e...o .umamuummmmmu-.uwmmmummmmm n.~ he..o ese.e eno.e e.a~ ~o~.e nae.o eno.e .uuemmmroeee.uwaau~m censuses-u usu- daaem successes nuances someones-u seas nee-n successes mud-sen «nausea assumes I I I seams» spay-us. I I I I I I cosmos sang-Hex I I I sequesm ~00— I Bin—53 ocm— I useless-ea: ocu— use so nee-a evacuees-a unsuuua season on nausea: luau season»: one seduces unseen wanna-c was so segues. eeuu non-l sou sue-uncala arouses use n some genes eeuusneu use .euuuoas accuses-a aeol.n« as» use seams» - .—¢m_ use even an uaeuou omeuouou can mu mucus huseaveuu sequence .suaeeee seas-sea .- sue.- .n «a 0035.. 0235.3 :33..— 3- «28.303 023:: .«uuoaov :33: no I.- SF .330..- 3- we no... no.3 «33 I.» «o cognac...- - on .390..- uonaunuua - no 33353 «a a. «33 2:. £00— .: nu 1:. 0«0— 5 03:3 :0) .3900- usuufhusnwnn c « .ccuooso «a. non undue-avouu «o Is. on» no 83359:— . no .30vo so»... o no «289...: I: 3 «08250.5 :33.- .oug - 5.3.!— uouooao 35¢ a .53.: ac 02.3.0 I.» 3 «23:...»- a 3 .0302.- 3- no 33353 no uni:— unuo... .5 «0 83.369:— - on 0302—- 33332. n no 33353 «0 hi: 3:. — c.oo~ ..a« o..~ a." ..~ 9.“ o.~ c.— n.— u.— n.— 5.: «.9 o.— 000.n 000.0 «no.0 a“ 0«0.0 0«0.0 ~00.0 «00.0 «~0.0 n00.0 c—0.0 «00.0 «00.0 n00.0 000.~ ««c.0 0«n.0 0«0.0 c«0.0 0«0.0 0«0.0 0«0.0 .«n0.0 c«0.0 «~0.0 0«0.0 c—0.0 c«0.0 000.— 000.0 0—0.0 Mm...“ ««0.0 nu0.0 n00.0 «00.0 000.0 000.0 «00.0 «00.0 «00.0 0.0.0 0.00n . ~.o« c.9h ...I.m. ~.o o.— o.~ ~.~ ~.~ ~.H a.“ N.— c.— c.— 000.« «n0.0 «00.0 may. «00.0 «00.0 «00.0 900.0 n~0.0 «00.0 0—0.0 000.0 «00.0 «00.0 000.~ «00.0 o—n.0 a. «~0.0 «~0.0 n«0.0 0«0.0 n—0.0 n«0.0 «~0.0 «~0.0 0«0.0 n«0.0 000.“ _00.0 0«0.0 .linm 900.0 000.0 n00.0 000.0 n00.0 «00.0 «00.0 c—0.0 a—0.0 0—0.0 30h nhnouoono unusual-u nIuOuaaa gal” 33 gun—mg 3% 3% 33% «Hanna-Ea 333qu .363. :.d 3.3” gas aslflzh. i0"; :30qu Innulua ..aoaa“.=oo. . .flgaa 54 Figure 12. Diameter class distributions for indicator species on 2.8 hectares of natural forest permanent plots in 19fl6 and 1981. Class A = ”.1 to 9.9 cm; class B = 10.0 to 1fl.9 cm; subsequent classes are in 5 cm increments through class I = 130.0 to 13fl.9 cm. The species indicated by asterisk are also canopy species. 55 _ Hedyosmum arborescens IOO~ ._ r f I I I I I i 50- _ I I : 0L __ .—1 I LV'T‘T' 1 I l T i I : Psychotrlo berteriana 5 I20 ._ r 1 80 - .‘L’ o s 40 - .‘2 ) '6 O b L; C '5 Cecropio peltoto 5 o - .0 I00 r- E 3 z 50. I - I ' I : 0" “— " fiT I 1 I I I : Didymoponox morototoni I : Ioo- - I I I : 50- _ I I I : o- @F‘fi‘w L 1W 41144111 114111 ABCDEFGH ABCDEF ------- I946 ------- -----|98l .---- Diameter Closs Distributions 56 ,1 1 h . _ b b . . b .n m m. .m u u m o .n I O .m a m .m P. .m 1 M m. I I I I .T r - T . . . . b . _ . n. nu nu nu nu nu n. O 5 m R... O 5 0 - ----- ---- -- - ---u.o:Ezu:. Prestoea montana* _-i I T ad . - _ p _ _ . O O 0 O O 5 O 5 W 5 m 5 .0 «086.32 IIIIIIIIIIIIIIIIIIIIIIII l A e c a ---l98|--- 111 J_ l A a c a E ----I94s--- l J J Class Distributions Diameter Figure 12 57 "l T Micropholis chrysophylloides * ZSOF 20°F ISO- 00 50- -------- £03323: «O *- Micrapholis garciniaefalia _— I LILJ‘ I J K L LLLLI A a c a E F a H --------..-- l98l llLll L11 J K A B C D E F G H I --—------- l946---------- Distributions Class Diameter Figure 12 58 Cyrilla racemiflora" l946 CT . _ _ _ _ no n. no no w 4 3 2 I o ------------- 22.232: *0 l98| _ I XY UVW ABCDEFGHIJKLMNOPORST Distributions Class Diameter Figure 12 59 Haenianthus salicitolius 6 4 l 9 5 HI 1|. 1 1. TI 1 r 1 — — P p — 5 0 5 5 2 w 7 w 2 O IIIIIIIIIIIIIIII m_O:D_>_Uc_ «0 m H 9 TI 1! I 1| I a T a L _ _ 5 5 m 7 w 2 0 ..an=2 ------- l4; JKLMNO L l ABCDEFGH JJJJI l Class Distributions Diameter Figure 12 60 percent of the 1946 populations. The 1981 total basal areas were also greater than in 1946, ranging from 125 to over 160 percent of the 1946 values for both species of Microphplis. Prestagg_only increased to 105 percent of the 1946 value because it lacks diameter increment. Percent importance values also increased for each of these species, ranging from 0.4 percent for Prestoea to about 1.5 percent apiece for both species of Micropholis. The abundance of species by topography is also shown in Fig. 11. H d somum, Cyrilla and both species of Micropholis are more common on slope plots whereas Cecro is, Did mo anax, and Prestoea are found mainly on valley plots. The remaining species, Pa chotria, Miconia, and Egggi£_ are well represented on both tapographic positions. The 1946 and 1981 diameter class distributions of these same species are shown in Fig. 12. Bedyosmum and Psychotria stems were largely confined to the smallest diameter class in both years. Cecropia and Miconia showed considerable regeneration in the smallest class in 1946 but the pattern was altered with the loss of stems by 1981. Didmmopgngx showed only a few stems in either year, with none in the smallest diameter class in 1981. The diameter class distribution of Cyrilla showed an influx into the smallest class in 1981, and a decline in the larger diameter classes. Cordia showed the typical negative exponential pattern of shade tolerant species in both years, but it was more pronounced in 1981. The same general pattern was also apparent for both species of Microgholis. The diameter class distribution for Prestoea approached that of a normal curve in both years. 61 Diameter class distributions of canopy species Diameter distributions for 11 of the remaining species that reach can0py size in some part of the colorado forest are given in Fig. 13. Eugenia stahlii and 0cotea moschata have been omitted because each had fewer than 10 stems in 1946 and 1981. Magnolia s lendens, Eugenia boringuensis, Clusia krugiana, Matayba dominguensis and Sgpium laurocergsus showed comparatively even distributions for all but the smallest diameter classes in both years. Haenianthus salicifglius and Tabebuia rigida showed relatively even distributions in 1946, except for the smallest diameter class, but approached a reverse J-shaped curve in 1981. The remaining species, Calycogonium s uamulosum, Crotgn pgecilanthus, Qggtg§_spathu1a§§, Brysonima gadsworthii and gyggi§_fgll§z, showed reversed J-shaped distributions in both years. These results will be discussed in conjuntion with species classifications in Part II. Species sugxiyal on thinned 219; The thinned plot in 1981 had about five times the stem density of Microphglis garciniaefglia and Tabebuia rigigg_and about twice the density of Cyrilla zaggmiflgra as the average for the natural forest plots (Fig. 14). Regeneration after thinning increased the stem density of 9123;; trugiana from 10 to 360 stems/ha, Cyrilla racemiflgra from 84 to 780 stems/ha, and Micgnia tetrandrg from O to 190 stems/ha during the 35-year period of measurement. Haenianthus salicifglius increased from 17 to 54 stems/ha and Tabebuia from 215 to 311 during the same period, and Hicropholis garciniaefglig declined in numbers. 62 Figure 13. Diameter class distributions for canopy species on 2.8 hectares of natural forest permanent plots in l9u6 and 1981. Class A = ”.1 to 9.9 cm; class B = 10.0 to 1h.9 on; subsequent classes are in 5 cm increment through class 0 = 75.0-79.9 cm. Indicator species not included. 63 Croton poecilanthus [111 F1 1 T IIj I I e - m _. . P p _ OOI _ p _ L IOI O u w h 9 I .I o r p S O O u n. w 0 b C O O T 1 l I T I 1 T I 1- 1 _ 1 w 1 F L _ b p - _ _ O O 0 0 O O O 0 0 O O O 0 0 0 5 0 5 4 3 2 l 2 l l I IIIIIIIIIIIIIIIIIIIIIIIIII m.Oaflm>mfl:- .0 «ODE-dz IT*I_TI L1 1 l J ABCDEFGH A a c a E F s H I ------- I945------- ------- l98l --------- Distributions Class Diameter Figure 13 :1 L L fl IL I IL _I r r r I 1 1 r 1 _ I. T. _ I I m _ T .n m .- T 1 I r it p .Pl . s . a . . . h . w m. a S u n .d a. a o .m .m. ' r u .4 a .M .n as m .m .m .m n m m '- B n! E C L 1 1| .L I. ll I I ll I rI .rI I l I I I 1 W 1 r 1 _ 1 4 I[ p . p p . . L p s p O 5 5 c a 2 5 a z - m w a .. - c m. w o --------------------------------- 22.232: .o Congas: --------------- AI 3 C I) E F'IB H II B C I) E F’IB H «mm IssI .------ ------- l946‘------ Class Distributions Diameter Figure 13 65 Matayba dominguensis T 1. 1| I 1 s 1 7 IF . L _. w. r h I . . I r a ' ‘ e a c I. o I. f O u f o I .m m n .W v. 0 M c. u 1. _ . L a b L . _ -. . 5 O 5 o 5 O 5 O 5 o 5 O I. .I 9. .4 I. .I -------------m.e=o_>_oc. .o con—E32 ------------------------ IJ A a c a E ---- 194s ----- A s c a E -----l98l----- Distributions Class Diameter Figure 13 66 22.2:sz was; :zoEoE me spawns I--I---.-:- Em. --I----- ----.---.--.-- ova. --.---.-...- 22.3....zcuwouo4 22.3....zouuoom4 ._...______s____._____.__.__ flldllln h u 4‘! - r7 1 o . ir: . . J 100 m . . . 1 -oo. " N «54:23 3.8002 n m a. 3 J A o w. 1 .32 W- I I. m. 1 $8 my m- S 4 Joan ” n 1 18.. m . u 1 L L can u L . 53225:: Es.cooou«_ou m 1 Lace . 67 Figure 14. Change in number of stems/ha by species on the thinned colorado plot vs. all natural forest colorado plots. In each case. the thinned plot is the first pair of entries by species. 68 III... -- - imo-u-I 5-02 - 52... 5L2 ....... m2 - -- mm_omam \ NE f7 L\\\\\\\X\\\\\\\\\\\\\\\‘ rad-SETS Fm 8.2. 232.8. - E. 222509 0230 I m0 2.33.5200 £23232 I 02 322.22.08.22“. 3.23822 I02 caveats. 2:322 It; 33:29... 2.0:?! Imz mazozuzom «555252.. I m: 0.3.2839. oztau I 5 2.03:... 2220 I 5 Esme-25:9... £2,833.00 I mu .mm. .32... 0c .3622 D 25. .muttoronssz § memos; QUE i-mz ........ mu ....... v6---- 1:8---- I] «H shaman OON Joov loom I 000 .4000. gOON. JaqwnN au/ $3811 ,0 69 Species richness The total number of tree species found on the natural forest plots declined from 88 in 1946 to 83 in 1981 (Fig. 15). Three notable observations may be made. First, the variability of species numbers for plots of the same size is considerable. Second, species numbers tend to converge through time. In 1946, species numbers ranged from 33 through 58 but by 1981 the range decreased to 36 through 49. Third, there is a considerable difference in species numbers according to topography. The slope plots A through D ranged from 33 to 46 in 1946 and from 37 to 43 species in 1981. The mixed plot E remained unchanged at 36 species in both years. The valley plots G and H ranged from 48 to 58 species in 1946 and from 45 to 49 species in 1981. Five fewer species between 1946 and 1981 represents a flux in which 12 species were lost and seven species gained. Of the 12 lost, Cupania merican , Qgggg§_ uidonia, Laplacea ortoricensis, Nectandra antillana, g, sintenisii, and Rheedia portoricensis are more common to the tabonuco forest, and Hibiscus tiliaceous to the coastal lowlands. Cyathea arborea, a tree fern, is more common in the dwarf forest although it occurs in openings at lower elevations. The remaining species, Heterotricbgm gyggsum and Coccoloba swartzii are uncommon in the colorado forest, and two unknowns are assumed to be uncommon colorado forest species. Of the seven species gained, Beilschimiedia pendula and Manilkara bidentata are more common in the tabonuco forest, and grammgdenig_sintenigi1_in the dwarf forest. The remaining species are components of the colorado forest including Hecraneum m d linum, Miconia pachyphyllg, and Miconia mirabilig which are uncommon, and Ouratea striata which is rare. 70 Figure 15. Change in number of species by plot and for the entire forest from 19u6 to 1981. 71 9C,F Entire Forest 30%; Individual Natural Forest Plots 60- Number of Species Thinned Plot 20- IO- 0 J l I I I 46 SI 56 76 8| Years Figure 15 72 On the thinned plot, there were only 24 species in 1946. After a slight decline, species numbers increased to 31 in 1981. Shannon-Wiener indices for all natural forest plots combined showed a decline from 4.81 to 4.33 during the period of measurement (Table 5). For individual plots, indices declined or remained the same in all but one instance. On the thinned plot, the index increased from 2.37 to 2.90 from 1946 to 1981. gare and endangered species Tree species tentatively classified as rare or endangered in the mid-1970’s (Soil Conservation Service and Department of Natural Resources 1973) decreased from 339 to 305 individuals on all 3.2 ha of natural forest and thinned plots during the period of measurement (Table 6). Most of this decline occurred with Magnolia splendens and Ternstroemia luguillensis. No species disappeared and two new species, Ardisia luguillensis and Ouratea striata reached minimum dbh and were recorded on two of the plots. For convenience, information on rare species found during the ordination study (Part II) is reported here. When partitioned by topography, Aggisig_luquillensis, Byrsonim§_wadsworthii, Magnolia splendens, Harliera sintensii, and Xylosma schwanekeanum.were more common on ridges and slopes while §i££g_myricoides was more abundant on slopes and valleys (Table 7). When stratified by life zone, Ardisia, Ma nolia, and Harleiga occurred more frequently in the rain forest than wet forest. When partitioned by elevation, Qi£££_and Xylosma appeared more commonly above 800 m while Brysonima was encountered more frequently below 800 m. 73 Table 5: Shannon-Wiener indices (Krebs 1972) in 1946 and 1981 on the permanent colorado forest plots. Years Plotl 1946 1981 A 3.88 3.77 s 4.23 4.23 c 4.07 3.95 D 3.51 3.63 s 4.03 3.78 G 4.20 4.03 11 4.13 3.93 1 2.37 2.90 A-Hz 4.81 4.33 1Plot sizes in all instances are 0.40 ha. Plots A through H are undisturbed and plot I was thinned. 2Sumof seven 0.40 ha plots or total of 2.8 ha. 74. .uuua :« as oc.o mus cacao HH< .uoa— an vacuum 0:» “coon cw u::bu mucus no pools: as» aw swans: unnum— uam d e.n o.o m.H ~.~ w.o m.o c.c Bassoxuoaaanom mamoa>x ~.o o.n n.HH ~.~ III o.a c.a «amwcouawm aumuauaz m.mH ~.w~ o.o~ ~.<~ o.m ¢.o~ «.mm mcmocoaam aaHocwmz “.9 5.0 c.H III III o.H III «amucmocwm aucavmaaouu III a." ~.o «.2 s.~ III m.o . samumwwm uaaumam ~.o III 5.6 III III III w.o mucussa0AHm anomouo n.o~ ~.- n.nH o.a~ o.n c.c— o.~m «anuuoamvaa anacomhum H.- c.32 m.a_ «.32 ~.n~ o._~ o.m umeaouauaa «snag m.~ m.m n.n Q.H m.c m.¢ o.¢ mausonaasvsd aamfivu< am“: 304 swam ass 20H~m> macaw smog: mmauunm macs mug; mnmwumoQOH uaav coaum>0Hm .auoHa huauomaou amou0u ovauoHou on» so aspen “anxuaoumv moqooaa sou» vouowcavsu was swam .u awash 76 Similarity of the plots Similarity comparisons between pairs of plots in 1946 and 1981 are shown in Table 8. Of the 21 comparisons between natural forest plots, 14 showed greater similarity in 1981 than in 1946 and six showed less similarity. One comparison was the same. In general, the loss of secondary and late secondary indicator species (Fig. 11, top row and Cyrillg) contributed to greater dissimilarity in 14 comparisons whereas the gain in climax indicator species (Fig. 11, bottom row except Cyrilla) contributed to greater similarity in 16 comparisons. The six comparisons that showed increases in dissimilarity had either large negative trends among declining secondary species, or negative trends among both secondary and climax species. The slope plots A through D when compared among themselves showed similarities ranging from 51 to 69 percent regardless of the year; whereas, when compared with valley plots G and 8, they showed similarities ranging from 37 to 57 percent. The valley plots and showed similarities ranging from 69 to 72 percent when compared among themselves in both years. The mixed plot E showed similarites ranging from 51 to 71 percent when compared with the slope plots, and from 48 to 58 percent when compared with the valley plots, regardless of the year. Of the seven comparisons of natural forest plots with the thinned 'pdot, six showed a decrease in similarity from 1946 to 1981 and one showed an increase. Indices ranged from 26 to 45 percent similarity. 77 .330QO some now vacuums-non ”.3303 W I haw-$3.5m ucoouom nausea noon .3 woumsm @38QO mo mod—6mm o>wuooammu ecu casuaa mmwmummouma amazod osu mo Sam mgu ma muoHa aaasmm ozu mo huwumawaam unwouom u III III III III III III III III H OMIeN III III III III III III III : mmIom Nulmo III III III III III III 0 QMIoc qume wmImm III III III III III m nmlmq anmq wmIaq HmIqo III III III III a emqu quqe mthm onImc hmch III III III 0 omINm nmIne mclmc HnIco meIHm «alum III III m wnIme ocImq onImn Nelda moImo mmINm mcINe III < H a U m a o m < uOHm Mama cam beau cu muoaa uamamauon amou0w ovmnoaou ecu How mouavaa huwumawawm unmouom um manna g 78 Growth and Production Diameter growth of select species by crown class The greatest MAI attained by canOpy species with adequate numbers of trees for analysis was 0.30 cm/yr by dominant specimens of Tabebuia rigida (Table 9). Dominant specimens of Ocotea s athulata, Haenignthus salicifolius, and Micropholis chrysophylloides as well as codominant specimens of Croton poecilanthus averaged 20.20 cm/yr. For all species, trees in the dominant or the codominant crown classes were the fastest growing, and trees in the suppressed crown class were the slowest growing. Different 35-year MAI patterns were evident when species were stratified by crown class. One that emerged for Calycogonium aguamulosum and Magnolia splendens was generally slow growth in which the dominant, codominant, and intermediate trees did not grow at significantly different rates from each other, but were significantly faster growing than suppressed trees. A more common pattern shown by Haenianthus ali ifolius, Micropholis arciniaefolia, g, chr so h lloides, Ocotea spgthulatg, and Tabebuia rigidg_was a gradual decline in diameter growth from dominant through suppressed trees. For all species combined, when stratified by crown class, the growth of dominants, codominants and intermediates averaged 3.2, 2.6, and 1.8 times, respectively, the growth rates of suppressed stems. Ratios, however, varied among species. When all survivors, regardless of species, were compared by crown class, a significant decline in MAI was observed between 1946-51 and 1976-81 in the intermediate and suppressed crown classes (Table 10). The dominant crown class also declined in growth, but the difference was not 79 can a» caucuses. Ha>oH una ogu us uaauouuau auuaaoauuauua one uauuueuaasa .uaeu one.» aaauunal a.asuasa .5 5... 63.3.3 .8 83:. .1635... so .213. I . N .uauue pudendum I am— moun: 866 36 .38: 866 . S6 2263 .86 u 26 .363 686 36 6.338 .30.... :4 .53 ~86 66 A.363 .86 u 36 to: :66 a 26 .85 366 36 3...... .55.... .SN 3 686 36 6:. 3 :66 . 36 6.8 3 «86 a 86 .3 3 $66 86 gum—1.5....“ .3. 3 866 36 .2" 3 686 q 36 as: 686 a 86 .2 3 6 86 3.3.3qu 6.2: 866 36 .32. .86 a 36 nan—:86 . :6 .83 266 36 343.33%: .86: 366 36 n.833 A256 4 26 .33 366 a 26 .83 :66 26 33342353436342 ‘36 3 686 36 .E 3 636 u 86 .3 3 .66 s 86 .33 666 :6 3.40.3433 .5 3 ~36 86 a?“ V 266 a S6 633 :66 a 36 .83 .86 36 gag—a .3 3 686 26 .Su 3 «36 a 86 .33 686 u :6 .33 686 26 3663-3435 .5: 686 86 £333 686 u 36 .23 «86 a £6 3.: 3 636 26 gamma-add :85 «86 36 .85 .86 u 36 .33 866 a :6 .8 3 866 26 33 3 3.8.3.... 3 3.33.33 8. 32.138 .3 3.5.8 .321.- NE .23. .586 3 mm . egg! .auana usually-m unsuau Hosanna as» no cacao amass an uamuoaa nuance ua auo>m>usa uaoAInn uo Anetween 2.00 and 5.66 m3/ha/yr on individual plots. In this instance, 82 Figure 16. Basal area changes for all permanent plots from 1946 to 1981. Mean is for all natural forest plots combined. Ingrowth includes recruitment and growth of residual stuns. 2/ha) is shown above Mean annual basal area growth (in m the bars. The top value is for ingrowth alone. The bottcn value sums both ingrowth and mortality as growth. 60 50 '5 .c 40 \ N E o 3 30 < '6 m o 20 cm Figure 16 Legend -UDD l946 Basal Area l946-8l Mortality -O.26 0.4! l946-8l Ingrowth l" Z l98l Basal Area 1:; 3:33 3:33 3233 F V 53211 W Z '333'. '1 0.42 “ r . l j a j 2:2: 3 / W / / / m / / / ’ ” / / / / 4 / / / I / r l / , / / W / l/ y / / / / / / / / / / / / / / F / / / / / / / / / / U L b J Ll Ll : u A B C 0 E G H Mean l -------------Nolurol Foresl-------------------- Than... Plots 84 Figure 17. Volume changes for all permanent plots from 1946 to 1981. Mean is for all natural forest plots combined. Ingrowth includes recruitment and growth on residual stems. Mean 3/ha) is shown above the bars. annual volume growth (in m The top value is for ingrowth alone. The bottom value sums both ingrowth and mortality as growth. 85 2.38 3.74 I l 06 3 Mean 2 3 ----------------Nalural Forest "“"""“""“"‘ TMM'C" 66 G 5. 6 =.====:_=.E=_.=======_==._==__ 25 W nvl. l.7. E a v. N r/l/l/I/l/l/l/l/ll/Ill/Illl/lI?”lili/l/I/l/l/l/l/l/I/t, O. 2 _====_=E===============E===_= D N. m . F n w O I m m m. m I M C d In: M IM 0 1.2 T n o l l l u. v s s w 6 6 6 I e 4 4 4 8 L w m m n U ad I D _ . a . P a p _ O 0 0 O 0 O w 0 5 W 5 O 5 I. I. ?. 9. .l I. .O£\ME. OE:_O> Plats Figure 17 86 the four slope plots averaged 2.40 m3/ha/yr while the two valley and mixed topography plots averaged 4.66 malha/yr. The thinned plot increased 3.74 m3/ha/yr in volume growth with mortality included. Biomass changes The mean increase in aboveground woody biomass, including branches 22.5 cm, on the natural forest plots was 0.67 t/ha/yr (Fig. 18). Individual plots ranged from 0.24 to 1.58 t/ha/yr with the exception of plot D which declined 0.36 t/ha/yr. The four slope plots averaged only 0.22 t/ha/yr while the three plate on valley and mixed topography averaged 1.27 t/ha/yr. The thinned plot accrued biomass at 1.71 t/ha/yr. Again, when mortality was added to ingrowth to determine biomass growth, it ranged from 1.33 to 3.07 t/ha/yr with a mean value of 1.90 t/ha/yr for all natural forest plots. The four slope plots, and three valley and mixed topography plots, averaged 1.45 t/ha/yr and 2.49 t/ha/yr, respectively. The thinned plot accrued biomass at the rate of 2.51 t/ha/yr with mortality included. Age of Cyrilla racemiflora The MAI for Cyrilla based on all available data including short term observations spanning only 5 years was 0.09 cm/yr. When those trees that died during the 35 years of observation were eliminated, the MAI increased to nearly 0.12 cm/yr for all trees regardless of crown class (Table 9). Tree size divided by the MAI provides a crude estimate of tree age. For a Cyrilla 1 m in diameter, an MAI of 0.09 cm/yr yielded an estimate slightly more than 1100 years. An MAI of nearly 0.12 cm/yr yielded an estimate of nearly 850 years old. 87 Figure 18. Aboveground woody biomass changes for all natural forest permanent plots from 1946 to 1981. Mean is for all natural forest plots combined. Ingrowth includes recruitment and growth on residual stuns. Mean annual biomass growth (in t/ha) is shown above the bars. The top value is for ingrowth alone. The bottom value sums both ingrowth and mortality as growth. 88 Legend l946 Biomass I946'Bl Mortality ENS III CED C23 l.58 ZOOr- l946’8llngrowlh 3_07 l98| Biomass 71 / ‘036 / .67 0.24 0.28 0.73 1.53 / I 24 (:90 1.33 |.62 0.99 / - - 7; F 7 x 7 .c F' / 7 / j P _. 7 / l.7l .. / fl / K W x / /' / / / / j / F / / / x / .,. loo~ / x / f 3 / $ +/ / FT / r1 / / / .2 / K W / / 50— / ”l 0 A B C D E G H Mean l ------------- ~---Nalurol Foresl ---------- --- -- -Thinned PHMS Figure 18 89 The mean MAI for the largest ingrowth 520 cm in diameter on the thinned plot averaged 0.36 cm/yr after slower growing stems were removed (Fig. 19; Methods). Elimination of the slowest growing trees in each diameter class 220 cm yielded a smaller data set (Fig. 19, circled points) from which means and standard errors were calculated for each class. These showed a decline in diameter increment with larger diameter classes. The linear regression of these MAI vs. the midpoint of the diameter class for trees 220 cm in diameter was highly significant although the scatter of individual points was considerable. The age curve derived from these data was slightly convex and provided an estimate of 660 years for a 1 m Cyrilla (Fig. 20). 90 Figure 19. Scatter of points for diameter growth by diameter classes for gyrillaLLanemiflgza, Circled points were those used in the weighted least squares regression. Points with standard error bars represent the means and standard errors for the circled points within the respective diameter classes. The mean and standard error at 10 cm is for the ingrowth on the thinned plot. Solid dots indicate growth rates that were discarded in the calculations of tree age. 91 5E“: mommoC BEEEQ ma ouswfim E. 00. » cm 00 l on ow » 00 owl . ) lam ON 0. o A .11 4.4- u. q I .o .14.}-.. _«.1..4«l-«u¢ll.aulldl. _ e e a e a ee . "O. O O O O. O O O O... O. O 9 o 9 000 e e e 0000 e o l O G 9 0w 0 we 0 .Jofio Inxwo 1 *3 N06 «NE 0 . $506.56; a 0 0 so (one o m 1010 aloe umms (AA/wD) 92 Figure 20. Curve of age related to diameter for 912111; racemiflora. The circles represent the upper endpoints of the diameter classes on the ordinate vs. the cunulative ages on the abscissa. 93 IOO 80 E 3 so I— (I) - 0) E .‘3 40 o 20 200 400 600 800 Age (years) Figure 20 94 DISCUSSION The Natural Forest Plots Hurricane-induced changes in forest structurg_§nd cogpositiog_ The observed trends in forest structure, species composition and forest growth during the 35-year measurement period, can be correlated with hurricane disturbance between 1928 and 1932, and subsequent forest recovery. Hurricane San Felipe, with maximum winds of 240 km/hr and 36 hours duration, passed southwest of the Luquillo Forest in 1928, and San Nicolas, with 140 km/hr winds and 40 hours duration, paralleled the north coast of the island in 1931 (Salivias 1972). These storms probably had some impact on the Luquillo Forest. San Cipriano of 1932, however, with 200 km/hr winds, more than 43 cm rainfall and 18 hours duration, had a trajectory directly over the Luquillo Mountains (Crow 1980; Fig. 5). This storm caused considerable defoliation, windfall, stem breakage, and mortality. A later hurricane, Santa Clara of 1956, traversed the island southwest of the Luquillo Mountains and caused only slight, localized damage within the Luquillo Forest (Wadsworth and Bnglerth 1959). Forest recovery begins as a mosaic of patches of different ages and gaps of different sizes, followed by colonization and growth, building, and maturity, to be renewed again by treefalls (Whitmore 1975; Brokaw 1985). In the colorado forest, most treefalls, or major forest damage, are concentrated into a short period of time during and after recurrent hurricanes. Immediately after the hurricanes, fallen debris including 95 leaves, branches and entire trees adds a considerable organic load to the ground surface (Lugo _£__;, 1983) and simultaneously allows light to reach the forest floor. Increased solar insolation, reduced transpiration, and direct contact of the biotic components of the soil and litter with fallen debris, creates an environment favorable for increasing the rate of decomposition of the debris and increasing nutrient availability (Bormann and Likens 1979; Master and Trappe 1984; Canham and Marks 1985). In response to improved environmental conditions, secondary species increase in prominence (White 1979) by means of a regeneration niche (Grubb 1977; Fox 1977) and help to conserve nutrients that otherwise might be exported due to heavy rains. Elsewhere, the abnormal production of seeds after storms (Bates 1929; King 1945; Webb 1958), regeneration in gaps of various sizes (Whitmore 1974), and the growth of secondary forest in patchy arrangements (Browne 1949) have been suggested as adaptive features that enhance the survival of pioneer and secondary species within the natural forest. Although this second phase of forest recovery, early colonization and growth (Brokaw 1985) including the response of residual stems to altered forest conditions, was not assessed directly, evidence of this phenomenon is available from the first plot measurements of 1946. In that year, 55 percent of the total stems were found in the smallest (diameter class (Fig. 6). Moreover several short-lived pioneer and early secondary tree species such as Bedyosmum ggborescens, Psychotrig herteriana, Cecropia eltata, Didmpanax morototoni, and Miconia Ilaeyigata were numerically more abundant at that time than in 1981 (Table 4 and Fig. 11). The third stage of recovery, that of building, is characterized by a gradual decline in the number of secondary species coupled with the 96 ingrowth and continued deve10pment of longer-lived, shade tolerant understory and canopy species like Cordig_borinquensis, Eggstoe§_montana, Calycogonium s uamulosum, and both species of Micropholis (Fig. 11). The observed shifts in the distributions of diameter, height, crown and specific gravity classes are all consistent with the steady trend of forest recovery (Figs. 6, 7, 8 and 9). The decline of stems in the smallest diameter and height classes is due in part to growth and competition among survivors, as well as the mortality of short-lived secondary species. Intermediate size trees, usually very vigorous, increase in numbers because of additions from smaller diameter classes as well as the continued survival of trees within the class. The general decline in the largest diameter classes is due to senescence and mortality of large trees that survived the hurricane. The increased number of suppressed stems in 1981 as compared with 1946 indicates that the forest is being gradually dominated by larger stems, or is ’closing’. The trend toward higher specific gravity for the entire forest reflects the gradual decline of light weight pioneer and early secondary species as well as the increased dominance of larger, heavier, shade tolerant, climax species. This occurs despite an influx of palms, which have a low specific gravity. The stem density for the entire forest remained approximately equal although the increases on the mixed plot and valley plots balanced the overall decline on the slope plots (Table 3). A priori, one would expect a decline in stem numbers immediately after the hurricane, followed by a maximum 15 to 20 years later in the second phase of recovery, as was observed in the tabonuco forest (Crow 1980). Subsequently, during the building phase, stem.numbers should decline due to increasing competition among survivors. This trend is apparent on the slope plots to different 97 degrees, although one of the four plots remained essentially stable in numbers. The forest-wide equilibrium in ingrowth and mortality during the period of measurement is due to the offsetting trends on slope and valley plots as well as the fact that the colorado forest is a dense formation characterized by small-size, slow-growing trees (Wadsworth 1951; Weaver 1983). Change during recovery within this forest is not as rapid as in the tabonuco forest where an 18 percent decline in stem numbers was observed between 1951 and 1976 as the forest approached steady-state (Crow 1980). Regardless, an increase in stem numbers on the valley plots during the building phase merits attention. Two phenomena may account for the differences between slopes and valleys. The first is that the valleys support larger trees, particularly in height (Fig. 8). When these die, they create large openings, or growing space for saplings. Six trees with dbh’s >90 cm.died during 35 years on valley plots but none that size were lost on the mixed or slope plots (Fig. 7). Moreover, the ingrowth of palms on the mixed plot and valley plots averaged 150 trees/ha compared with only 50 trees/ha on the slope plots during the period of measurement. The average number of palms in 1981 on the mixed plot and valley plots was 380/ha whereas on the slope plots it was only l40/ha. Palms are not only shade tolerant, slow growing and persist for long periods (Bannister 1970), but also require less space to mature than the larger dicot species, commonly growing to maturity in their shade. The six percent decline in the number of tree species for the entire forest is also a trend evident during the building phase (Fig. 15). Individual natural forest plots, however, show increases, decreases, and stable numbers of species, yielding ambiguous trends if considered alone. 98 Species diversity is a topic that has received considerable review in the literature with several authors postulating greater diversity in mature forest (Budowski 1965; Margalef 1968; Odum 1969). Others have suggested that recurrent disturbance should provide more niches (Sloan Denslow 1985) and subsequently greater species diversity, and that species richness will be the highest in communities subject to intermediate levels of disturbance (Connell 1978; Pickett and White 1985) or in regimes where disturbance is more frequent than the time required to competitively exclude certain species (Huston 1979). The most notable decreases in species’ numbers are on both valley plots located near the transition between tabonuco and colorado forest at 650 m in elevation. The disturbance of this area during and after the hurricane led to colonization from both forest types giving these plots a higher species richness about 15 years after the storm. It has been suggested that a landscape under a stable disturbance regime should develop a dynamically stable mosaic of patches of different ages (Sprugel 1985), and to this could be added a dynamically changing number of species at any particular locale. Perhaps the most interesting trend in the colorado forest is the apparent convergence of species’ numbers on individual plots during the period of measurement. This suggests that after disturbance, localized species richness may vary considerably because of the status of the patch before disturbance, the amount of disturbance, the particular locale, and the availability of seed, among other factors, and that during recovery, species numbers tend to converge as the stands mature. For all natural forest plots combined, the floristic richness appears greatest a few years after disturbance during the phase of colonization and growth, when both primary and secondary 99 species are components of the stand. During the building phase, the gradual dominance of shade tolerant species leads to decreases in abundances of secondary species and may locally exclude some of the uncommon species. The increase in similarity among two-thirds of the natural forest plots (Table 8) reflects the increase of indicator primary species as well as many additional changes in similarity among the most common species. In general, the decline of secondary and late secondary species led to greater dissimilarity between plots. Similarity changes 22 percentage points occurred for 20 species including the first 17 species and Clusia krugiana and Miconia tetrandrg in Table 4, and Guettarda valenzuela, a species not listed. The species which provided the most frequent and largest percent changes, in descending order, were Prgstoea nt na, Calycogonium s uamulosum, Micrgphglis hr so h lloides, Cyrilla racemiflora, and Qggggg_ W- Except for Magnolia s lendens, which may attain sizes of 1 m in diameter and 20 m in height, the remaining 11 species tentatively classified as rare or endangered in 1973 (Table 6 and 7) are small, usually not exceeding 15 cm in diameter and 10 m.in height. Qi;;;_ gygicgidgs is found locally between 600 and 750 m elevation in the Luquillo Mountains, frequently in valley topography. Maytenus glgngaga, Eggenig sii, and Xylgsgg scgganeckeanum are found from about 300 m to the summits. The remaining species are most common in the dwarf forest on the exposed summits of the Luquillo Mountains and largely survive on ridges and upper slapes within the colorado forest. 100 Grogth Mean annual diameter growth for all stems in the colorado forest (Table 9) is slower than in the tabonuco forest of Puerto Rico (Wadsworth 1951, 1952; Tropical Forest Experiment Station in 1953; Crow and Weaver 1977; Weaver 1979); only the subtropical dry forest in the southwest (Weaver 1979) and the dwarf forest on mountain summits (Weaver 1983) have slower mean diameter growth rates. In the dry forest, growth is clearly t _l. 1978) and in the dwarf forest limited by a lack of moisture (Lugo by the adverse growing conditions of summit areas (this study, Part III). Elsewhere in the humid tropics, including 13 sites ranging from 20 to 2100 m in elevation with annual rainfalls between 1960 and 7500 mm (Leigh and Smythe 1978), diameter growth for the trees measured exceeded that of the colorado forest. Many temperate forest trees with limited growing seasons have more rapid annual diameter growth (USDA Forest Service 1965) than colorado forest trees. The pattern of similar diameter growth by crown class, except for suppressed stems, of some species, suggests that once light is available, rapid growth is possible. Another pattern with a decreasing trend in growth from dominants through suppressed trees suggests that maximum diameter growth is attained by larger trees with good exposure to light. The second pattern is particularly interesting because it substantiates the utility of simple crown class designations as indicators of growth potential for a period as long as 35 years. The 35 percent decline of diameter growth on the trees that survived the 35-year period of measurement (Table 10) probably indicates that the forest building phase is ending and the mature phase is beginning. 101 Similar declines in the rates of biomass and nutrient accumulation have been observed in both natural forests and plantations following the phase of most rapid growth (Vitousek 1985). Growth in basal area, volume, and biomass may be estimated from differences in the standing crop of trees in 1946 and 1981, providing an estimate of net change, or by summing net change and mortality over the same period. The second method yields a better estimate of actual stand production because it includes all trees that exploited growing space within the stand. Mortality is not estimated in most studies because they are of short duration and little mortality is evident, or because it is not considered important. Omission of mortality underestimates actual forest production, and is analogous to excluding intermediate thinnings in studies of plantation production. Mean basal area changed little on all natural forest plots. This indicates a long-term dynamic balance between the loss of secondary species and mortality of large stems that survived the passage of the hurricane, and ingrowth and increment on numerous residual stems in the small and intermediate diameter classes. Because measurements such as diameter and basal area do not reflect height changes, they are of limited use as indicators of forest production unless correlated with them. The mean net volume growth was 1.06 malha/yr during 35 years, and was positive on six of the seven natural forest plots suggesting recovery from past forest disturbance. In steady-state conditions, individual plots might show minor increases or decreases in standing volume, but mean net volume change would approach zero. The notable variability in growth rates between slope and valley topography may be partly attributable to better site conditions and partly to different species’ compositions. 102 Valley plots in colorado forest, especially at lower elevation, contain Dacryodes xcelsa, Manilkara bidentata, Buchenavia ggpitatg, Matayba dominguensis and Miconia laevigata from the tabonuco forest, all of which attain larger size than most colorado forest species, and tend to grow at faster rates. Several other estimates of net volume growth in previously disturbed tropical moist and wet forests are in the range between 1 to 5 mglha/yr. Virgin tabonuco forest in Puerto Rico recovering from disturbance by San Cipriano had a net volume growth of 2.5 m3/ha/yr (Wadsworth 1957). Partially cut, or disturbed rain forest in Malaysia, yielded 1 to 3 m3Iha/yr, rain forest in Nigeria 2, and dipterocarp forest in the Philippines from 2.9 to 4.3 m3/ha/yr (Johnson 1976). Tropical high forest in Uganda appeared incapable of yielding >1.4 mglha/yr under any management system involving multiple cutting, nor >4.2 m3/ha/yr under a single harvest system (Dawkins 1959). Although these data are not directly comparable because of differences in forest composition and structure as well as the nature of past disturbance and forest measurement, they do demonstrate the generally low volume accumulations in previously disturbed natural forests. Net aboveground woody biomass accumulation was also positive on six of the seven natural forest plots, with a mean rate of 0.67 t/ba/yr. Again, the valley plots G and B exceeded the slope plots, averaging 1.41 t/ha/Yr. These plots are located only 2 km horizontally and 200 m vertically from a 0.72 ha tabonuco plot that was monitored for 33 years after the same disturbance. Net aboveground woody biomass accumulation on the tabonuco plot was estimated at 1.50 t/ha/yr (Crow 1980). 103 Net aboveground woody biomass accumulation rates for secondary and mature natural forests elsewhere in the American tropics are presented along with those determined in this study in Fig. 21. Site details are presented in Appendix Table 1. All net woody biomass accumulation values for mature forest plots are clustered in the range of 0.5 to 3.5 t/ha/yr. Secondary net woody biomass accumulations are more variable, but except for the youngest stands, exceed those in the mature forest, ranging up to 12 t/ha/yr in tropical moist forest. Because all of the secondary forest sites were cleared before establishment of the growth plots, their biomass accumulation rates should approach the maximum for the colonization and growth, and early building phases of natural forests after disturbance. The Thinned Plot Changes in forest structurg_gnd composition induced by thinnigg Thinning in 1947 eliminated diameter classes 250 cm (Fig. 6), reduced the number of species to about 60 percent of the average for the natural forest (Fig. 15), removed seed sources from the plot area, and provided growing space for the residual trees. The 50 percent reduction of basal area, which included canopy and understory stems, probably simulated the most destructive effects that could be inflicted on forest structure and composition by a hurricane. The combined effects of the 1766 and 1772 type A hurricanes that passed directly over the Luquillo Mountains may have caused similar destruction in certain localities. Records of recovery on the thinned plot begin after two disturbances, one by hurricanes, and the second by silvicultural treatment.i Because the increment of residual stems was slow and growing space remained available (Weaver 1983), a massive influx of new stems increased stem density on the 104 Figure 21. Net biomass growth rates (aboveground woody biomass) for secondary and mature natural forests in the American tropics: (1) tropical moist forest (Salas 1978; Folster. Salas and Khanna 1976); (2) tropical moist forest (Ewel 1971); (3) subtropical moist forest (Snedaker 1970); (4) tropical wet forest (Galley sisal, 1976); (5) tropical moist forest (Uhl 1980); (6) subtropical wet forest (Crow 1980); (7) lower montane wet forest (Tanner 1980); and (8) lower montane wet forest -- colorado forest (this study). Forest formations are Holdridge (1967) life zones. 105 7 777 6 / / .w s -m 3 3 3 32 I 3 15 4 2 3 3 I2 3 4 5 35 LI . b O 5 0 5 0 taxes). cozosvoca €663 959.0 26.: Mature forest Age of stand (years) Figure 21 106 plot by nearly 40 percent (Table 3; Fig. 6). The five species that increased considerably in numbers (Fig. 14) were not considered secondary or late secondary indicator species (Fig. 11, top row of species and Cyrilla), except for Cyrillg. The increases in abundance of Tgbebgig rigida and Cyrilla racemiflor; can be attributed to the proximity of seed sources. Both of these species along with Haenianthug salicifgligs, Clusia kru iana, and Miconia tetrandrg appear capable of exploiting small gaps within the colorado canopy. Moreover, of the seven new species recorded on the thinned plot during the period of measurement, Cgcropig eltat , Alchornia latifolia, and Psychotria berteriana are pioneer or secondary species that had largely disappeared on the natural forest plots, and Ficus trigon§;g_and Marliera sintenisii are uncommon within the colorado forest. The general decline in similarity between the natural forest plots and the thinned plot during recovery (Table 8) is related to the thinning itself and its subsequent effect on the regeneration of several species (Fig. 14). The greatest declines in similarity were caused by Calycogpnium sguamulosgg_and 0cotea s athul ta, both of which increased on the thinned plot, but not in proportion to the influx of new stems (Fig. 14). The largest increases in similarity between the thinned plot and valley plot H, the only natural forest plot which became more similar to the thinned plot, were for Cyrilla racemiflora, Miconia tetranda, and Tabebuig rigigg, Valley plot B was the only natural forest plot on which Cyrilla regenerated well. 107 EMA The initial basal area, volume and aboveground woody biomass on the thinned plot were only 54, 40, and 43 percent, respectively, of the means for the natural forest plots at first measurement. Net growth of all three parameters was positive and higher than on any individual natural forest plot, with the exception of volume on valley plot 6. This comparatively rapid and sustained development is partly attributable to more ingrowth (Weaver 1983) during the colonization phase, and a prolonged building phase due to the initial thinning. The apparent anomaly with greater biomass but less volume growth than on plot G is caused by the differences of the specific gravities of the species common on each of the plots. Palm, with a specific gravity of 0.26 g/cm3, is common on plot G whereas Clusig.krugiggg_and Micropholis garciniaefolia with specific gravities of 0.90 and 0.64 g/cm3 are abundant on the thinned plot. Mean annual diameter growth for all surviving stems averaged 0.10 cm/yr on the thinned plot (Weaver 1983), a value which is not significantly different from the mean annual increment for all species combined, regardless of crown class, on the natural forest plots (Table 9). Age and Dyggmics of Cyrilla racemiflgga Regeneration of Cyrillg is stimulated by openings within the forest. Seedling regeneration on the natural forest plots is low (Fig. 38, Part II) showing that it does not regenerate as well in shaded conditions as do known climax species. This evidence is corroborated by the paucity of 108 suppressed trees which were found on the natural forest plots in 1946 (Table 9). The distribution of Cyrilla trees by diameter class on the natural forest plots in 1946 (Fig. 12; see also Appendix Fig. A-l) strongly suggests that regeneration was concentrated in relatively short intervals at some time in the past, possibly after major disturbances (Jones 1945; Knight 1975; Lorimer 1980). In contrast, the 1981 distribution for Cyrilla shows a disproportionate number of stems in the smallest diameter class (Fig. 12). Most of the change in distribution was caused by the ingrowth of Cyrilla into smaller diameter classes after the 1932 hurricane, but there was a slight decline of stems in the larger diameter classes as well. The most valuable information derived from the availability of two diameter class distributions in different years is that conclusions regarding regeneration strategies based on a single observation may be erroneous if past disturbance of the forest is not considered. The abundant regeneration of Cyrilla on the previously thinned plot supports the hypothesis that Cyrilla requires a gap to regenerate. Whereas the natural forest plots showed a general decline of Cyrilla stems during the measurement period (Fig. 11), the previously thinned plot had nearly a lO-fold increase (Fig. 14). Initial growth of the stems was rapid compared to the residual stand, with many narrow-crowned trees exploiting small openings in the canopy. The convex pattern of diameter growth shown by gygillg (Fig. 20) is not characteristic of shade tolerant climax trees. For shade tolerant species, there is considerable variability of diameter increment in the seedling and sapling stages (Brown and Matthews 1914, Brown 1919, Richards 109 1952). In closed forest, diameter increment of these species commonly increases with tree size, yielding a sigmoid growth curve during the life cycle of the tree. An initial slow phase of seedling growth is followed by rapid increment approaching linearity during youth and early maturity, gradually slowing, and finally ceasing altogether in senescence (Whitmore 1975). The rapid growth of Cyrilla in the juvenile stage assures its early establishment in the canopy and is followed by slower growth in mature stages. This pattern is similar to that suggested for Didymopanax rototoni, a known secondary species (Crow and Weaver 1977). Because Cyrilla attains large size within montane areas of the Antilles where tree diameter growth averages about 0.10 cm/yr (Wadsworth 1951; Weaver 1982, 1983), the age of the largest trees may approach 1000 years. Using the synthesis technique, a tree 1 m in diameter was estimated to be 660 years old. Critics of the technique, however, point out that validation is difficult, and that confirmation of the age of any particular tree is impossible. Although this is true for the individual tree, records of hurricanes that passed directly over Puerto Rico from the 1500’s to 1970 (Salivia 1972) allows comparison with the diameter class distribution of Cyrilla and its estimated diameter growth rate. According to the record, no hurricanes passed near Puerto Rico from 1673 to 1713 (Salivia 1972). and none passed directly over the island from 1713 to 1737 (Fig. 4). In 1738, two storms, either type A or B, passed over the south coast where damage was reported. When a temporal scale derived from the age curve (Fig. 20) is superimposed over the 1946 Cyrilla diameter class distribution (Appendix Figure A-l), the dates of major hurricanes known to have passed directly over the Luquillo Mountains coincide with observed concentrations of stems. In 1766, San Jenaro and 110 San Marcos, and in 1772, San Agustin, all type A or B storms, passed over the mountains. These storms probably did considerable damage to the forests which resulted in the notable increase in the 30 to 40 cm diameter class (Fig. 12). The fact that the storms were concentrated in a brief period of only six years, and that many trees in the forest may have been senescent after nearly a century without disturbance, probably accentuated their effect. Although its trajectory was not directly over the Luquillo Forest, Santa Ana of 1825 may rank as the greatest of recorded hurricanes. It caused considerable damage to the central and eastern end of the island, and probably disturbed the forest. San Narciso of 1867 was another very damaging storm whose effect may be detectable in the 20 to 30 cm diameter classes. The effects of the most recent hurricane that traversed the Luquillo MOuntains, San Cipriano of 1932, are not noticeable in the lowest 1946 diameter class distribution. This may be in part due to the limited amount of time for regeneration and growth to 4.1 cm, the lower limit of the smallest diameter class. A recent study of stand dynamics in the subtropical wet forest of the Luquillo Mountains during the period between 1943 and 1976 (Crow 1980) implicated this hurricane in species composition changes over time including the presence of secondary tree species a few years after the storm followed by a gradual increase in dominance of climax species with stand recovery. It should be pointed out that information on trajectories, intensity and duration of past hurricanes, in particular those before the 20th century, is vague. At least the hurricane record does not detract from 111 the hypothesis that regeneration of Cyrilla is tied to recurrent climatic events and stand disturbance. The loss of Cyrilla in the larger diameter classes and the current lack of seedling regeneration are probably attributable to a 54-year hiatus in hurricane disturbance. This decline in Cyrilla is a major concern of wildlife biologists working to save the rare and endangered Puerto Rican parrot. Once found throughout the island, the parrots are now confined to the colorado forest of the Luquillo Mountains where they use hollow Cyrilla trees for nesting sites. One possible method to increase Cyrilla tree density in the Luquillo Mountains evident from this study is selective thinning favoring Cyrilla trees in the residual stand to serve as a seed source. Additional field observations of Cyrillg throughout the Luquillo Mountains have shown that trees are capable of growing to large size within 1 m of each other. The largest trees are found at low elevations in the colorado forest, or at high elevations in the the tabonuco forest. Extremely large trees, in excess of 2.5 m, have been seen only twice. These trees have an oblong shape and a major bifurcation two or three meters above the ground. Mereover, the intervening diameter classes between those reported in this study (Fig. 13) and the largest specimens, that is, between 1.3 and 2.5 m, have not been seen in the field. These observations lead to the conclusion that large Cyrillg trees found in Puerto Rico, and presumably elsewhere in the Antilles, are the result of growth and development on favorable, protected sites at the lower limits of their elevational range. Furthermore, extremely large Cyrillg trees probably result from the coalescence of proximate trees. If this is true, it is doubtful that the largest trees live much beyond 1,000 years. 112 On occasion, fallen Cyrilla trees have develOped new crowns from already exisiting branches thus ’walking’ to a new microsite (Bodley and Benson 1980) and surviving into a new generation! Moreover, trees with damaged crowns have been observed to sprout from the base. These characteristics are combined with the capability to regenerate and grow into small openings in the canopy, and persist for what appears to be about 1000 years for the largest individuals. Hurricane Damggg_§nd Recovery in the Luguillo Mountains Hurricane damage is a function of numerous factors among which storm characteristics (wind velocity, total rainfall, and duration), storm recurrence, vegetation features (tree size, stem density, crown growth patterns, and tree specific gravities) and soils and topography all play a role. Sufficient information is available from this and other studies to propose a comprehensive hypothesis regarding hurricane damage and recovery in the forests of the Luquillo Mountains. The principal storm event, in this instance, San Cipriano, was already described. The soils in the Luquillo Mountains are mainly clays: The vegetation, for the most part, lacks an emergent story which has been attributed to recurrent hurricanes (Odum 1970). With ascent in the mountains, the soils become wetter, notably so above the rather abrupt transition between tabonuco and colorado forest (Wadsworth and Bonnet 1951). Tree size declines along the same elevational gradient (Wadsworth 1951), and at any particular elevation in both the colorado and dwarf forests, trees are tallest in the valleys and smallest on ridges, with greatest stem densities on ridges. In contrast, 113 the tabonuco forest has some of its tallest trees on upper slopes and ridges (Wadsworth 1953; Crow and Grigal 1980). In tabonuco forest, windfall is most prevalent in valleys where the soils are wet and the trees have large crowns, whereas on ridges, breakage prevails (Wadsworth and Englerth 1959). The small, dense vegetation on upper slopes and ridges in both colorado and dwarf forest is characterized by woods with high specific gravities, which should reduce both uprooting and breakage (Lawton 1984; Putz §t_§l, 1983), resulting in damage mainly by defoliation. Indeed, this forest type was designated as "hurricane hardwood" long ago (Murphy 1916). The lower slopes and valleys in the upper forests probably suffer more windthrow because trees are larger and the soils wetter on that topographic position. In summary, the gaps created by storm damage should not only diminish in size with an increase in elevation, but also should decrease in size from valley through slopes to ridges at any particular elevation. The greater occurrence of gegggpig pgltatg, a known pioneer, in valley tapography of colorado forest (Fig. 33, Part II), lends support to this idea. Moreover, the greater net aboveground biomass increment on valley plots as compared with slope plots (Fig. 18) may partially reflect greater windthrow in valleys. Recovery after hurricanes in the tabonuco forest is by colonization of large pioneer trees such as Cecrgpia eltata, Didymopagax morototoni, Alchornea l tifoli , and Alchorneopsis portoricensis along with smaller trees in the Rubiaceae and Melastomataceae. The number of stems peaks after 15 to 20 years (Crow 1980) and then declines due to competition among the survivors. In the colorado forest, Cecropia and Digymopangx also colonize but the trees are smaller and grow slower. The number of stems fluctuates widely several years after disturbance, generally 114 declining on slapes and increasing in valleys, the latter because of palm growth. No direct observations of hurricane damage are available for the dwarf forest. However, recovery on two sites, one that had been clearcut and the second which had been cleared by a plane wreck, showed a dense growth of grass and ferns with some sprouting after six years (Byer and Weaver 1977), a condition that has prevailed for 17 years. Vegetative reproduction was observed elsewhere in the same forest (Nevling 1971). It may be surmised, then, that major windthrow would be detectable many decades after a hurricane. Indeed, the formation of "alpine meadows" in isolated areas within the dwarf and upper colorado forest may be related to landslides and windthrow caused by hurricanes. Once formed, their recovery may span centuries. The fact that they are rare, and that secondary tree species are largely absent (Byer and Weaver 1977), indicates that uprooting of trees and the creation of large gaps in the dwarf forest are uncommon. The building phase in the tabonuco and colorado forest is characterized by a shift in the number of stems to larger diameter and height classes, as well as a shift from intermediate to suppressed crown classes. Because growth is more rapid in tabonuco forest, the changes occur more rapidly. In the dwarf forest, localized damage could result in a greater number of stems due to sprouting, but stem number changes should be minimal. Species richness in the tabonuco and colorado forests varies from one locale to another with disturbed areas containing greater relative abundances of secondary species. As these forests recover from hurricane damage, secondary species decline in number and primary species increase in dominance. In the dwarf forest, damage is not as extensive, secondary 115 species are largely absent, and tree species richness remains unaltered during recovery. Biomass increment in the tabonuco forest is initially rapid and later declines producing a convex recovery pattern (Crow 1980). In colorado forest where damage is not as extensive, and secondary species are fewer and slower-growing, the recovery pattern is probably linear. The dwarf forest recovery pattern is surmised to be linear as well, but slower than in colorado forest. In the event of another hurricane in the near future, damage and recovery of the forest should be similar to that described for San Cipriano. Storm attributes and the interval between storms would influence the amount of forest damage and recovery processes. If the Luquillo Forest continues to mature for many more years, however, senescence, mortality, and individual tree falls during periods of unusually heavy rainfall and high winds should play a prominent role in creating gaps and initiating recovery on a smaller scale. 116 SUMMARY Periodic damage and recovery from hurricanes constitute an integral phase of forest growth and development in the Caribbean. Recovery of the colorado forest on seven natural forest permanent plots in the Luquillo Mountains of Puerto Rico from 15 to 50 years after the passage of San Cipriano is characterized by: (1) ingrowth and mortality rates of about 20 stems/ha/yr, or 1.1 percent/yr of the number of stems at first measurement; (2) shifts to larger diameter and height classes. This includes a decline in the smallest diameter class, and the 2 smallest height classes, accompanied by increases in intermediate diameter classes, and all remaining height classes. The largest diameter classes showed minor declines caused by the loss of larger trees, mainly Cyrilla acemifl a; (3) a 5 percent decline of stems in the intermediate crown class accompanied by a 7 percent increase in the number of suppressed stems. Dominant stems also declined slightly while the codominant stems remained nearly equal; (4) an increase from 0.563 to 0.574 g/cm3 in the average specific gravity of the forest including palms. Dicots alone increased from 0.608 to 0.632 g/cm3; (5) a change in species composition with the loss of pioneers and secondary species and increasing dominance of climax species; 117 (6) a decline in species numbers from 88 to 83 on all plots, or 2.8 ha combined, with a convergence of species numbers to about 35 to 45 on most individual 0.4 ha plots. The Shannon-Wiener diversity index also declined from 4.81 to 4.33 for all plots combined, and decreased on all but one of the individual plots; (7) mean diameter increment on all surviving stems of 0.16 i 0.08 cmlyr for dominants (D), 0.13 1 0.01 for codominants (C), 0.09 1_0.09 for intermediates (I), 0.05 1 0.00 cmlyr for suppressed (S) stems. Two basic growth patterns were shown, namely, D - C - I > S, and a gradual decline in annual diameter increment from D through 3, with at least three significant differences; (8) a decline in diameter increment between 1946-51 and 1976—81 on surviving dominant, intermediate and suppressed stems, and an increase in diameter increment on surviving codominant stems; (9) little change in basal area, with a mean and standard error of 41.2.: 1.4 m2/ha for all plots combined in 1946 and 1981. Slope plots remained equal or declined while valley plots increased slightly in basal area; and (10) an increase in standing volume from 216.7 :_12.1 to 253.8 1 14.7 m3/ha and standing biomass from 121.5,: 8.2 to 145.0 1;7.2 t/ha for all plots combined, yielding net growth rates of 1.06 m3/ha/yr and 0.67 t/ha/yr for volume and biomass, respectively. Again, growth on the valley plots exceeded that on the slope plots. A thinned plot that had not only been disturbed by the hurricanes but also had been reduced in basal area by 50 percent at the time of first measurement showed the following differences when compared with the natural forest plots during the same period of measurement: 118 (l) 20 percent less mortality and 260 percent more ingrowth of stems than on the natural forest plots; (2) an increase in the number of species from 24 to 31 and an increase in the Shannon-Wiener diversity index from 2.31 to 2.90; (3) a greater increase in numbers of Clusia kru i na, Miconia tetrandra, and Cyrilla racemiflor; than on the natural forest plots; (4) greater changes in crown class distribution than on the natural forest plots with a 47 percent decline in intermediate stems and a 40 percent increase in suppressed stems; and (5) a positive net basal area increment averaging 0.42 m2/ha/yr with net volume and net biomass increments also positive, averaging 2.38 m3/ha/yr and 1.71 t/ha/yr, respectively. Moreover, data from both of the above groupings of plots disclosed that: (1) 12 tree species tentatively classified as rare and endangered for Puerto Rico in 1973 were found in the colorado forest, the majority of these on ridges and upper slopes at higher elevations; and (2) the age of a 1 m Cyrillg racgmiflgrg, the largest and most prominent tree species within the forest, is about 660 years old. Also, the diameter class distribution of Cyrillg was tentatively correlated with previous hurricanes whose trajectories passed directly over the Luquillo Mountains around 1770. Hurricane damage within the Luquillo Mountains appears to vary considerably by forest type, topography, and elevation. Windthrow is more common in humid valleys particularly at lower elevations where trees are larger. Breakage, in turn, is more common on upper slopes and ridges where trees are better anchored, in particular, in colorado and dwarf 119 forests, where these topographic positions are occupied by densely-packed, small and heavy~wooded species. Recovery patterns in the forests after hurricanes are also distinct. In the tabonuco forest, fast-growing pioneers and secondary species rapidly invade gaps in the tabonuco forest resulting in a convex recovery curve for biomass accumulation. In the colorado forest where gaps are smaller and the secondary species slower growing, biomass recovery appears linear for several years. In the dwarf forest at the summits, breakage creates small openings that are filled by vegetative reproduction or sprouting of climax species rather than regeneration of secondary species. In instances where damage is severe, recovery may take more than a century. LITERATURE CITED LITERATURE CITED Baker, F. S. 1950. Principles of silviculture. McGraw-Hill Book Company, Inc., New York. 414 p. Bannister, B. A. 1970. Ecological life cycle of Euterpg glgb9§a_ Gaertn. p. B 299-314. In H.T. Odum and R.F. Pigeon, eds. A tropical rain forest. USAEC, TID-24270. Barden, L. S. 1981. Forest development in the canopy gaps of a diverse hardwood forest of the southern Appalachian Mountains. Oikos 37:205-209. Bates, C. 2. 1929. Efectos del huracan del 13 de septiembre en distintos arboles. Revista de Agriculture de Puerto Rico 23:113-117. Beard, J. S. 1944. Climax vegetation in tropical America. Ecology 25(2):127-158. Beard, J. S. 1949. Natural vegetation of the Windward and Leeward Islands. Oxford Forestry Memoirs 21:1-192. Bodley, J. H. and F. C. Benson. 1980. Stilt-root walking by an iriateoid palm in the Peruvian Amazon. Biotropica 12(1): 67-71. Bormann, F. H. and G. E. Likens. 1979. Pattern and process in a forested ecosystem. Springer-Verlag, New York. 253 p. Briscoe, C. B. and F. H. wadsworth. 1970. Stand structure and yield in the tabonuco forest of Puento Rico. p. B. 79-90. In H. T. Odum and R. F. Pigeon, eds. A tropical rain forest. USAEC, TID-24270. Brokaw, N. V. L. 1985. Treefalls, regrowth, and community structure in tropical forests. Chapter 4, p. 53-69. In S.T.A Pickett and P.S. White, eds. The ecology of natural disturbance and patch dynamics. Academic Press, Inc., New York. Brower, J. E. and J. H. Zar. 1977. Field and laboratory methods for general ecology. Hm. C. Brown Company Publishers, Dubuque, Iowa. 194 p. Brown, R. T. and T. T. Curtis. 1952. The upland conifer-hardwood forests of northern Wisconsin. Ecological Monographs 22: 217-234. Brown, W. B. 1919. Vegetation of Philippine Mountains. Manila, Philippines. 120 121 Brown, W. H. and D. M. Matthews. 1914. Philippine Dipterocarp forest. Philippine Journal of Science 9 (Section A): 413-561. Browne, F. G. 1949. Storm forest in Relantan. Malayan Forester 12:28-33. Budowski, G. 1965. Distribution of tropical American rain forest species in the light of successional processes. Turrialba Byer, M. D. and P. L. Weaver. 1977. Early secondary succesion in an elfin woodland in the Luquillo Mountains of Puerto Rico. Biotropica 9:35-47. Cairns Jr., J. 1980. The recovery process in damaged ecosystems. Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan. 167 p. Callahan, J. T. 1984. Long-term ecological research. Bioscience 34:363-367. Canham, C. C. and P. L. Marks. 1985. The response of woody plants to disturbance: patterns of establishment and growth. Chapter 11, p. 197-216. In S.T.A Pickett and P.S. White, eds. The ecology of natural disturbance and patch dynamics. Academic Press, Inc., New York. Chapman, H. H. and W. E. Meyer. 1949. Forest mensuration. McGraw-Hill Book Company, Inc. New York. 522 p. Connel, J. H. 1978. Diversity in tropical rain forests and coral reefs. Science 199:1302-1310. Cremer, R. C., B. J. Myers, F. van der Duys and I. E. Craig. 1977. Silvicultural lessons from the 1974 windthrow in Radiata Pine plantations near Canberra. Australian Forestry 40:274-292. Crow, T. R. 1980. A rainforest chronicle: a 30-year record of change in structure and composition at El Verde, Puerto Rico. Biotropia 12(1):42-55. Crow, T. R. and D. F. Grigal. 1980. A numerical analysis of arborescent communities in the rain forest of the Luquillo Mountains, Puerto Rico. Vegetatio 40(3):135-146. Crow, T. R. and P. L. Weaver. 1977. Tree growth in a moist tropical forest of Puerto Rico. USDA Forest Service Research Paper ITF-22, Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 17 p. Dawkins, H. C. 1959. The volume increment of natural tropical high-forest and limitations on in improvements. Empire Forestry Review 38:175-180. Ewel, J. 1971. Biomass Changes in early tropical succession. 122 Turrialba 21(1):110-112. Ewel, J. 1980. Tropical succession. Biotropica Supplement 12(2):l-95. Ewel, J. J. and J. L. Whitmore. 1973. The ecological life zones of Puerto Rico and the U. S. Virgin Islands. Forest Service Research Paper ITF-18, Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 71 p. Foggie, A. 1945. On girth increment and age of trees with special reference to management in irregular tropical forest. Empire Forestry Journal 24:176-191. ' Folster, H., G. de las Salas and P. Rhanna. 1976. A tropical evergreen forest site with perched water table, Magdalena Valley, Colombia. Biomass and bioelement inventory of primary and secondary vegetation. 0ecology Plantarum ll(4):297-320. Fox, J. F. 1977. Alternation and coexistence of tree species. The American Naturalist 111: 69-89. Golley, F. B., J. Ewel and G. I. Child. 1976. Vegetation biomass of five ecosystems in north western Colombia. Tropical Ecology 17:16-22. Grubb, P. J. 1977. The maintenance of species-richness in plant communities: the importance of the regeneration niche. Biological Review 52:107-145. Holdridge, L. R. 1967. Life zone ecology. Tropical Science Center. San Jose, Costa Rica. 206 p. Howard, R. A. 1962. Volcanism and vegetation in the Lesser Antilles. Journal of the Arnold Arboretum 43(3):279-311. Huston, M. 1979. A general hypothesis of species diversity. American Naturalist 113:81-101. Johnson, N. E. 1976. Biological opportunities and risks associated with fast growing plantations in the tropics. Journal of Forestry 74:206-211. Jones, E. W. 1945. The structure and reproduction of the virgin forest of the north temperate zone. New Phytologist 44:130—148. Jordan, C. F. and E. G. Farnworth. 1982. Natural vs. plantation forests a case study of land reclamation strategies for the humid tropics. Environmental Management 6(6):485-492. Rershaw, R. A. 1973. Quantitative and dynamic plant ecology, second edition. American Elsevier Publishing Company, Inc., New York 308 p. 123 King, H. C. 1945. Notes on the three cyclones in Mauritius in 1945: their effect on exotic plantations, indigenous forests and on some timber buildings. Empire Forestry Review 24:192-195. Knight, D. H. 1975. A phytosociological analysis of species-rich trapical forest on Barro Colorado Island, Panama. Ecological Monographs 45:259-284. Krebs, C. J. 1972. Ecology: the experimental analysis of distribution and abundance. Harper Row, Publishers, New York. 694 p. Ladrach, W. 1976. Rendimientos y comportamiento de la regeneracion natural. p. 54-64. In Concesion Forestal Bajo Calima 1959-1975. Carton de Colombia, Cali, Colombia. Leigh, E. G. Jr. and N. Smythe. 1978. Leaf production, leaf consumption, and the regulation of folivory on Barro Colorado Island. p. 33-50. In G. G. Montgomery, ed. The ecology of arboreal folivores. Smithsonian Institution Press, Washington, D. C. Little, E. L. Jr. and F. H. Wadsworth. 1964. Common trees of Puerto Rico and the Virgin Islands. Agricultural Handbook No. 249. USDA Forest Service, Washington, D. C. 548 p. Little, E. L. Jr., and R. 0. Woodbury. 1976. Tress of the Caribbean National Forest, Puerto Rico. USDA Forest Service Research Paper ITF-20. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 27 p. Little, E. L. Jr., R. 0. Woodbury and F. H. Wadsworth. 1974. Trees of Puerto Rico and the Virgin Islands Second Volume. Agricultural Handbook No. 449. USDA Forest Service, Washington, D. C. 1024 p. + illustrations. Lorimer, C. G. 1980. Age structure and disturbance history of a southern Appalachian virgin forest. Ecology 61:1169-1184 Lugo, A. E., M. Applefield, D. J. Pool and R. B. McDonald. 1983. The impact of Hurricane David on the forests of Dominica. Canadian Journal of Forest Research 13(2):201-211. Lugo, A. E. and S. Brown. 1981. Ecological monitoring in the Luquillo Forest Reserve. Ambio 10(2-3):102-107. Lugo, A. E., J. A. Gonzalez-Liboy, B. Cintron and K. Dugger. 1978. Structure, productivity, and transpiration of a subtropical dry forest in Puerto Rico. Biotropica 10:278-291. Margalef, R. 1968. Perspectives in ecology theory. University of Chicago Press, Chicago, IL. 111 pp. 124 Maser, C. and J. M. Trappe. 1984. The seen and unseen world of the fallen tree. USDA Forest Service General Technical Report PNW-164. Pacific Northwest Forest and Range Experiment Station, Corvallis, OR. 56 p. Miller, D. F. S. 1952. A girth increment study of Baikiaea plurijggg in Northern Rhodesia. Empire Forestry Review 31:45-51 0 Neumann, C. J., G. W. Cry, E. L. Caso and B. R. Jarvinen. 1978. National Climatic Center, 0.8. Department of Commerce, National Oceanic and Atmospheric Administration, Ashville, N.C. 170 p. Nevling, L. I. 1971. The ecology of an elfin forest in Puerto Rico, 16. The flowering cycle and an interpretation of its seasonality . Journal of the Arnold Arboretum 52: 586-613. Odum, E. P. 1969. The strategy of ecosystem development. Science 164(3877):262-270. Odum, H. T. 1970. Rain forest structure and mineral cycling homeostasis. p. H 1-52. In H. T. Odum and R. F. Pigeon, eds. A tropical rain forest. USAEC, TID-24270. Osmaston, M. A. 1956. Determination of age/girth and similar relationships in tropical forestry. Empire Forestry Review 35:193-197. Pickett, S. T. A. and P. S. White. 1985. The ecology of natural disturbance and patch dynamics. Academic Press, Inc., New York. 472 p. Prince, A. J. 1973. The rate of growth of Greenhart (Ocotea godiggi) Schomb. Commonwealth Forestry Review 52:143-146. Putz, F. E., P. D. Coley, R. Lu, A. Montalvo, and A. Aiello. 1983. Uprooting and snapping of trees: structural determinants and ecological consequences. Canadian Journal of Forest Research 13:1011-1020. Richards, P. W. 1952. The tropical rain forest: an ecological study. Cambridge University Press. Cambridge, England. 450 p. Rosero, P. 1979. Some data on a secondary forest managed in Siquirres, Costa Rica. p.209-210. In G. de las Salas, ed. Workshop agro-forestry systems in Latin America. Turrialha, Costa Rica. Salas, G. de las. 1978. Carare-Opon el ecosistema forestal. Serie Tecnica No. 8 CONIF. Bogota, Colombia. 87 p. Salivia, L. A. 1972. Historia de los temporales de Puerto Rico y 125 las Antillas. Editorial Edil, Inc., San Juan, Puerto Rico. 385 p. Sauer, J. D. 1962. Effects of recent tropical cyclones on the coastal vegetation of Mauritius. Journal of Ecology 50: Sloan Denslow, J. 1985. Disturbance-mediated coexistence of species. Chapter 17, p. 307-323. In S.T.A Pickett and P.S. White, eds. The ecology of natural disturbance and patch dynamics. Academic Press, Inc., New York. Snedaker, S. 1970. Ecological studies on tropical moist forest succession in eastern lowland Guatemala. Ph. D. thesis, University of Florida, Gainsville, Florida. 131 p. Soil Conservation Service and Department of Natural Resources. 1973. Rare and endangered plants of Puerto Rico. San Juan, Puerto Rico. 85 p. Sprugel, D. G. 1985. Natural disturbance and ecosystem energetics. Chapter 19, p. 335-352. In S.T.A Pickett and P.S. White, eds. The ecology of natural disturbance and patch dynamics. Academic Press, Inc., New York. Steel, R. G. D. and J. H. Torrie. 1960. Principles and procedures of statistics. McGraw-Hill Book Company, Inc., New York. 481 p. Tanner, E. V. J. 1980. Studies on the biomass and productivity in a series of montane rain forests in Jamaica. Journal of Ecology 68:573-588. Tomblin, J. 1981. Earthquakes, volcanoes and hurricanes: a review of natural hazards and vulnerability in the West Indies. Ambio 10(6)340-345. Tropical Forest Experiment Station. 1953. Thirteenth annual report. Caribbean Forester 14:1-33. Uhl, C. 1980. Studies of forest, agricultural, and successional environments in the upper Rio Negro region of the Amazon basin. Ph. D. thesis, Michigan State University, East Lansing, Michigan. 201 p. USDA Forest Service. 1965. Silvics of forest trees of the United States. Agriculture Handbook No. 241. Washington, D. C. 762 p. Veillon, J. P. 1983. E1 crecimiento de algunos bosques naturales de Venezuela en relacion con los parametros del medio ambiente. Instituto de Silvicultura, Universidad de los Andes, Merida, Venezuela. 122 p. Vitousek, P. M. 1985. Community turnover and ecosystem nutrient 126 dynamics. Chapter 18, p. 325-334. In S.T.A Pickett and P.S. White, eds. The ecology of natural disturbance and patch dynamics. Academic Press, Inc., New York. Vivanco de la Torre, 0., M. Cardenas Cruz, G. Cortsire Iturraide, and J. A. Tosi, Jr. (no date). Croquis ecologico parcial del Ecuador IERAC. Quito, Ecuador. Wadsworth, F. H. 1949. The development of the forest land resources of the Luquillo Mountains, Puerto Rico. Ph. D. Dissertation, University of Michigan, Ann Arbor. 481 p. Wadsworth, F. H. 1950. Notes on the climax forests of Puerto Rico and their destruction and conservation prior to 1900. Caribbean Forester 11:38-47. Wadsworth, F. H. 1951. Forest management in the Luquillo Mountains, 1. The setting. Caribbean Forester 12:93-114. Wadsworth, F. H. 1952. Forest management in Luquillo Mountains, III. Selection of products and silvicutural policies. Caribbean Forester 13:93-119. Wadsworth, F. H. 1953. New observations of tree growth in Tabonuco Forest. Caribbean Forester 14:106-111. Wadsworth, F. H. 1957. Tropical Rain Forest. p.13-23. In FAQ Forestry and Forest Products Studies No. 13, Tropical Silviculture. Rome, Italy. Wadsworth, F. H. and J. A. Bonnet. 1951. Soil as a factor in the occurrence of two types of montane forest in Puerto Rico. Caribbean Forester 12:67-70. Wadsworth, F. H. and G. H. Englerth. 1959. Effects of the 1956 hurricane on forests in Puerto Rico. Caribbean Forester 20:38-51. Weaver, P. L. 1979. Tree growth in several tropical forests of Puerto Rico. USDA Forest Service Research Paper SO-152. Southern Forest Experiment Station, New Orleans, LA. 20 p. Weaver, P. L. 1982. Monitoring of stand changes and growth in the natural forests of Puerto Rico. p. 43-59. In Proceedings of the Eight Symposium on Natural Resources, Department of Natural Resources, San Juan, Puerto Rico. Weaver, P. L. 1983. Tree growth and stand changes in the subtropical life zones of the Luquillo Mountains of Puerto Rico. USDA Forest Service Research Paper 80-190, Southern Forest Experiment Station, New Orleans, LA. 24 p. Weaver, P. L., M. D. Byer and D. Bruck. 1973. Transpiration rates 127 in the Luquillo Mountains of Puerto Rico. Biotropica 5(2):123-l33. Webb, D. B. 1964. The growth of Khaya radif lia, Chlorophora xcelsa, Afzelia africana and Pycanthus angolensis in Kurame in southern Zaria province. Department of Forest Resources Technical Note No. 30. Nigeria. Webb, L. J. 1958. Cyclones as an ecological factor in tropical lowland rain-forest, North Queensland. Australian Journal of Botany 6:220-228. White, H. H. Jr. 1963. Variation of stand structure correlated with altitude, in the Luquillo Mountains. Caribbean Forester 24(1):46-520 White, P. S. 1979. Pattern, process, and natural disturbance in vegetation. The Botanical Review 45: 229-299. Whitmore, T. C. 1974. Change with time and the role of cyclones in tropical rain forest on Rolombangara, Solomon Islands. Institute Paper No. 46, Commonwealth Forestry Institute, Univerty of Oxford, England. 77 p. Whitmore, T. C. 1975. Tropical Rain Forests of the Far East. Clarendon Press, Oxford, England. 282 p. Wiley, J. W. 1985. The Puerto Rican Parrot and competition for its nest sites. p. 213-223. In P.J. Moors, ed., Conservation of island birds : proceedings of a symposium, 1982. ICBP World Conference, Cambridge, England. ICBP Technical Publication 3. Wilkinson, R. C., R. W. Britt, E. A. Spence and S. M. Seiber. 1978. Hurricane-tornado damage, and insect infestations of slash pine. Southern Journal of Applied Forestry 2:132-140. Wilson, H. H. 1976. The effect of the gale of August 1975 on forests of Canterbury. New Zealand Journal of Forestry 21(1):133-l40. Wijesinghe, L. C. A. 1959. A study of the girth increment of Manilkara hexandra Dubard (Palu). Ceylon Forester 4:219-224. Wood, T. W. W. 1970. Wind damage in the forest of Western Samoa. Malayan Forester 33:92-99. APPENDIX TABLES Appendix Table l. 128 Biomass accumulation rates for secondary and mature natural forests in the American Tropics. Growth rate Codel Life2 Age Biomass Source No. zone (yrs) (t/ha/yr) l T-mf 2 4.89 Salas 1978; Folster, 5 12.02 Salas and Khanna 1976 16 11.05 2 T-mf 2 3.93 Ewel 1971 4 8.00 6 6.07 3 ST-mf l 1.30 Snedaker 1970 2 2.33 3 5 . 12 4 4.67 5 5.80 6 6.15 7 5.12 8 6.71 9 6.54 10 4.67 4 T-wf 2 6.62 Golley et al. 1976 4 11.11 5 T-mf 0.8 0.18 Uhl 1980 1.8 3.47 6 ST-wf M3 1.67 Crow 1980 7 LM-wf M3 o . 5 Tanner 1980 1 I o 2 I O 3.5 3 8 LM-wf M o . 67 this study 1Data coded to Fig. 21. 2Life zones (Holdridge 1967): T-wf, Tropical wet forest; T-mf, Tropical moist forest; ST-mf, Subtropical moist forest; ST-wf, Subtropical wet forest; LM-wf, Lower montane wet forest. 3 M - mature forest APPENDIX FIGURES 129 Appendix Figure A-1. Diameter class distribution of Cy:illa,racgmi£1923, on the natural forest plots within the colorado forest in 1946. The time scale indicates the years to which the diameter class corresponds as well as the years that type A storms passed directly over the Luquillo Mountains. according to the synthesized age curve. The shaded portion of smallest diameter class. that is. < 4.1 cm in diameter. was not tallied in the field. Percent of Total Stems 130 Time Scale for l946 Distribution IS“ 1890 I776 |C42 MOS 1283 L_*_._.I._k A; l L 1 32 67 72/36 20 r- . i—-+ |946 Distribution N5- _ — H)- 5 .. 0‘ , 4[ l 0 IO 20 so so so so 70 so so IOO no I20 I30 Diameter Classes (cm) Appendix Figure A-l PART II STRUCTURE AND COMPOSITION OF CLOSED CANOPY COLORADO FOREST RELATIVE TO ENVIRONMENTAL GRADIENTS 131 132 INTRODUCTION Relationships between vegetation and environmental gradients are difficult to demonstrate in tropical forests because structural complexity and species richness compound the tasks of sampling and interpretation (Hall and Swaine 1976). Moreover, the gradients themselves are often complex, and in montane habitats include elevation, aspect, tapography, and associated moisture differences, among others. The existing forest classification schemes for the Caribbean Islands (Beard 1944, 1949) including Puerto Rico (Wadsworth 1951; Little and Wadsworth 1964; Ewel and Whitmore 1973) emphasize forest physiognomy and common tree species but contain little information on variation in species composition within the recognized forest types. Gradient analysis is the interpretation of vegetational response to spatial variation in environmental factors. The advancement of numerical analysis beginning with the ordination of rain forest in Borneo (Ashton 1964a, 1964b) and the classification of rain forest in Australia (Webb g;_ al, 1967a, 1967b) provided an impetus to the study of gradient analysis in several tropical forests (Table 11). These studies disclosed numerous relationships including forest composition with topography (Ashton 1964a; Austin and Graig-Smith 1968; Ashton 1976), vegetation structure with altitudinal gradients (Webb gt_§1, 1970), forest composition with reproductive biology (wong and Whitmore 1970), gibbon behavior with forest types and structure (Whitten 1982), and the influence of animals on forest 133 zvsum mwzu cash Humane use zone mama nausea mmwfi uammxc mam :Hozxmzao mood semen mm was momemm mnofi Has: 3.3 @585 can 3mm onmfi cwxsom cuss nwm.mm new: naem_ .waeoz .Hm a. new: «mom nouuoum van muonscz onmfi causm< “cum" muosuficz mam macs II.IINxa~ ensues: somH .Hw um squmnuwcnc II.I|.wem~ suwsmlwwmuo mam awuma< NNAL .Hw hm caum=< “seems .mseaL eouem< ammu0m awn» mcmucos ummp0m away mcmucoe umsoa umwu0m :fimu vcmavoos mazm>mm accm>mm ummuow swan msoamwumvlwsmm mam Hmcommom comumum>m macm>mm ummu0m awn» Dachau awe» unseen came mamasofl amouow camp amouom away encased umauom away ammucm away ammuom away oowm ouuosm ouwx cuumsm nausea .ommuoaoo ouumm moweouuooz mwumwwz mwumwwz mmumwaz amaze muwum< Hmuucmo .mvamzm muasm< mwfimuumn< .ummoo .3 memuuma< .z mwamuumn< xmamumw mamas: .m>umma¢ sommm muumsbm .wzmuamz mmcmHmH aoeoaow macaw cannon magmam umam can mwm< mmumv mam muocu=< can. unchan— cameo; cowwom mummuom Haowmouu am mommaacm Hmowuaesc mafia: mawvnum mo mamasnm .HH manna 134 composition (Ashton 1976). The effects of soil on species composition within a particular forest, however, were less conclusive (Ashton 1964a, 1976; Austin g£_§l, 1972; Knight 1975; Newbery and Proctor 1984). In the Luquillo Mountains, two attempts were to made to discern the effects of environmental gradients on species distributions. One was a numerical analysis in tabonuco forest which disclosed eight major forest types that were strongly correlated with past land use and topography (Crow and Grigal 1980). The other was a transect with a fixed aspect on ridge topography, mainly in colorado forest, which indicated variation in tree species abundances according to elevation (White 1963). Additional relationships remained unknown because of the complex nature of the environmental gradients pointing out the need for further work in the forest. In addition to the lack of information on tree species distributions, little was known about the regeneration patterns and seral stages of canOpy trees in the colorado forest. Tree species reflect a gamut of characteristics that comprise their life history strategies to assure survival in a complex, montane environment. These include morphological differences in seed size and wood specific gravity as well as ecological differences in the capacity to regenerate and grow into canopy openings or through a shaded understory. Analysis of these characteristics would provide a basis for the tentative classification of tree species according to seral stage. The development of management alternatives for timber, wildlife, and rare plant species, and other forest resources, is dependent on-a knowledge of forest structure and species distribution patterns. Likewise, 135 regeneration patterns and seral stages of the component species must be known (Dawkins 1958; Whitmore 1984). 136 METHODS Existing trails were used to locate 25 groups of three plots each on different topography scattered between 620 and 970 m elevation within the colorado forest. Groups of plots (i.e., ridge, slope, and valley) were spaced at about 50 m increments in elevation. They contained closed-canopy vegetation, and had varying aspects, slopes, and exposures within the groups (Fig. 1). The definition for topography in this section differed from that in Part I. The lepe, valley and mixed designations used in Part I referred to the predominant topography of the plot. In this section, ridge plots were located entirely on convex t0pography, valley plots were entirely on concave or level topography, and slope plots were located on nearly uniform slopes without concave or convex features. After preliminary observations, a 50 x 10 m plot size and a minimum stem diameter of 4.1 cm*were selected for sampling. The plot size was a tradeoff between a unit large enough to include a representative sample of the most common tree species and small enough to remain entirely on the topographic feature. The short stature of the forest and the inclusion of small trees in the samples were important factors facilitating the use of small plots. On each plot, all trees 24.1 cm in diameter at breast height (dbh, or 1.4 m above the ground) and rooted within the plot were measured to the nearest 0.1 cm using a diameter tape. Height was measured to the nearest 0.1 m using an optical rangefinder. Crown classes were recorded as dominant, codominant, intermediate or suppressed (Baker 1950). Tree 137 volumes, including branches 22.5 cm in diameter, and above ground woody biomass, were derived in the same manner as outlined in the Methods section of Part I. On each plot, percent forest cover was estimated by taking four densiometer readings at each of two stations spaced about 20 m apart, to substantiate that the canopy of the vegetation was closed. Moreover, at each station, two soils samples to 50 cm depth were collected using a standard soil probe. Soil contents were combined by station and placed in standard collecting bags. Bulk density was calculated from the dry weights of the samples and the volume of the soil probe. The samples were then ground with a mechanical pulverizer and passed through a 2-mm sieve. Organic matter concentrations were determined using the Walkly-Black method (Black 1967). An unbalanced two-way analysis of variance (ANOVA) including topography (ridge, slope or valley) and life zone (wet or rain), with a continuous elevation covariate, was used to explore variance in forest parameters. The parameters tested for each plot were: the number of trees, the number of species, the mean height of dominant and codominant stems, basal area (mZ/ha), volume (m3/ha), aboveground woody biomass (t/ha), soil bulk density (g/cm3), percent organic matter, and soil organic matter content (kg/m2). In the data analysis, the main effects of life zone, topography, and elevation were considered along with interaction. Initially, for each parameter (Y), the complete model was: 138 Y - life zone + topography + elevation + (life zone x topography interaction) + (life zone x elevation interaction) + (tapography x elevation interaction) + (life zone x tapography x elevation interaction) + e. The null hypothesis of the complete covariance model was that the interaction terms involving elevation were not significant, or specifically, that the difference between (topography x elevation) - (life zone x elevation) - (topography x life zone x elevation) - zero. If the F value for any of the parameters under consideration was significant, then it followed that elevation could not be used as a covariate. In all instances where F values for interaction were not significant in the complete covariance model, these terms were eliminated and a reduced model was developed. Within each plot, a centrally located strip of 20 subplots 1 m2 in size was used to sample seedling regeneration of tree species that reached canopy size. All seedlings between 15 cm and 1.5 m tall were recorded by species. Seedlings of tree species characteristic of mid-elevations in the tabonuco forest (Wadsworth 1951) were eliminated from consideration because they were not major components of the colorado forest whereas seedlings of tree species most abundant in the transition zone at about 600 m were included. Seedlings of tree species common to the dwarf forest were also included because exposed slopes and summits where they occur may be found within the elevational limits of the colorado forest. Aspects of the life history strategies of the 20 canopy species were assessed by comparing the abundance of seedlings and understory trees to canopy trees, and by determining seed sizes and tree specific gravities (Smith 1970). These four indices, which reflect the capacity of individual 139 species to develop under closed canopy, were used to differentiate between primary and secondary tree species. The capacity of canopy trees to produce seedlings, and their capacity to grow into the understory (defined as stems between 4.1 and 7.5 cm dbh and in the suppressed crown class), were termed success. The following definitions were used: (1) Seedling success of canopy trees: relative density of seedlings of each canopy species under forest shade divided by the relative density of canopy-size trees of the same species. Relative density ratios based on all 20 canopy species were used to reflect success with regard to all species, and not just per species. (2) Understory success of canopy trees: relative density of understory trees of each species divided by the relative density of canopy-size trees of the same species. (3) Seed size: one half the length plus the width of the disseminule. (4) Specific gravity: the ratio of weights of a given volume of oven-dried wood to that of an equal volume of water. Each species was ranked from one (most secondary) to 20 (most primary) for each of the four factors. Higher ratios in both seedling and understory success indicated the capacity to survive in shade whereas lower ratios indicated intolerance to shade. Large seeds, dispersed mainly by gravity or animals, characterize primary species whereas small, usually wind-dispersed seeds, are typically borne by secondary species (Budowski 1965; Gomez-Pompa and Vazquez-Yanes 1974). Wood specific gravity is also an index to life history strategy in that primary species grow slower and have denser wood than secondary species (Budowski 1965; Smith 1970). Conversely, secondary species require more rapid growth to avoid shade and their wood densities tend to be less than those of shade tolerant species (Lawton 1984). 140 After ranking each tree species according to each factor, a composite species score was derived by the following relationship: Species score ' ((2 x understory ranking) + seedling ranking + seed ranking + wood density ranking)/5. Greater emphasis was given to understory success as compared to other components of the composite score because it showed the capacity of the seedlings to survive and grow in shaded environment. These scores, in turn, were ranked from 1 to 20. Other complementary information was also compared with the above to aid in the classification of species: (1) Population structure: Reversed J-shaped diameter class distributions are indicative of those species capable of reproducing and growing in the shade, and those with greater proportions of stems in intermediate size classes, or equal distribution of size classes, are indicative of species that reproduce in openings or gaps created by recurrent disturbance of the forest (Knight 1975). (2) Differential survival after disturbance: Rapid increase in basal area and numbers a few years after disturbance followed by decline is characteristic of pioneer species. A gradual accumulation of basal area, and possibly stem numbers, several years after disturbance, characterizes primary species. Gap species may give varied trends depending on their growth rates and longevity. Reciprocal averaging and polar ordination techniques outlined in Ordiflex (Gauch 1977) were selected to determine the similarity of species’compositions on plots from the dozen or so multivariate techpiques available (Gauch 1982). Reciprocal averaging accommodates a wide range of community variation and simultaneously provides species and sample configurations. Polar ordination, in turn, provides several options for 141 data analysis (Gauch 1973) making it a versatile technique (Cottam.g§_§_, 1973). Ordination was favored over classification because relationships between species and the environment were being explored. Stem densities were used as basic input data for reciprocal averaging. For polar ordination, stem densities in the form of percent similarity were employed along with a double standardization option and both program and user selected endpoints. Species with fewer than 30 occurrences were eliminated from each data set. This break-off point eliminated about 55 percent of the total species. However, each of these eliminated species constituted less than 0.4 percent of the total number of stems, or less than an average of 0.4 stems per plot. Elimination of rare species is supported in the literature because they contain little ecological information and their incidence is a matter of chance (Gauch 1977). Use of canopy species alone was sufficient for a classification of tropical and subtropical forests in Australia (Webb g;__l, 1967b) and less than 25 percent percent of the species were adequate for a gradient analysis on the Solomon Islands (Austin and Grieg-Smith 1968). Different combinations of input data were run in the ordinations and different endpoints were tested with polar ordination. Included were: forest-wide tree species and sample plot ordinations; composite sample plot ordinations (that is, combined ridge, slope, and valley plots at the same elevation within the same life zone) for the entire data set; ordinations of sample plate by life zone; and ordinations of species by topography (ridge, slope or valley) without regard to life zone. The last two ordinations limited the sample plot data sets to those plots located on an elevational transect over a ridge between El Yunque and El Toro peaks (Fig. 1, Part I). 142 Rain forest was located to the windward and wet forest to the leeward on the transect. Finally, stem densities of 20 common colorado forest trees grouped in topographic composites (ridge, slope, and valley plots), were plotted by elevation and aspect on a transect over the ridge between El Yunque and El Toro peaks. The number of stems per species were also plotted by tapography without regard to elevation or aspect. All species were identified using the taxonomic descriptions and keys for the trees of Puerto Rico (Little and Wadsworth 1964; Little g;_al, 1974). These references also provided information on seed sizes and tree specific gravities. 143 RESULTS AND DISCUSSION Forest Cover Canopy closure for the plots sampled ranged from 81 to 97 percent with a mean and standard error for all 75 composite readings of 90.8 1 0.5 percent (Table 12). No statistical differences were observed by life zone, elevation or topography. Cover percentages in this range of values indicate closed canopy with occasional small openings among the branches of trees. W The increase in number of species according to topographic position was most pronounced for cumulative areas 55000 m2 (Fig. 22). Thereafter, additional plots increased the total number of species at a slower rate. Tree species averaged 15/plot and did not demonstrate significant relationships with components of the covariance model (Table 13). However, certain trends were apparent. The greatest mean number of species was found on ridge topography and the least on valley tapography, with species numbers in the wet forest greater than those in the rain forest for comparable topographic positions. With an increase in elevation, the average number of species/plot declined (Fig. 23). The topographic variation in species’ occurrence with the greatest numbers on ridges may be related to the lower forest canopy and better drainage conditions on these sites. Many small, rare tree species survive on ridges in the colorado forest (Table 7, Part I). In contrast, fewer 144 Table 12: Summary of percent cover in closed canopy stands in the colorado forest Percent cover Factor Mean 1 SE (n)-L/ Topography Ridge 91.9 -_i-_ 0.9 (25) ' Slope 90.8 i 0.8 (25) Valley 89.6 i 0.9 (25) Life Zone Rain 90.1 i 0.8 (30) Wet 91.4 i 0.6 (45) Elevation 615 to 790 m 90.9 i 0.7 (37) 800 to 970 m 90.6 i 0.8 (38) Mean 90.8 i 0.5 (75) 1/ SE - standard error and n a number of observations. 145 Figure 22. Tree species-area relationships for 25 valley. 25 slope. and 25 ridge plots 500 m2 in size in the colorado forest. All stems were 2.4.1 on dbh. 146 A AA AA 70— A‘ ‘A O G O 0 A A o o o o 0 AA 60— ‘A e e A 0 . e ‘1 C)o-c> m.mm~ An.a~v o.cm~ Am.n~v m.o- macaw <.ocm An.wev c.w~m Am.smv <.emm owvwm .I m E as: o \m A ;\m V H > «.mq Illa: ~.mq IIIII m.me annoy ~.em ah.~v n.—m Ao.mv n.5m huHHm> m.oe Ao.mv c.me Am.~v m.n< macaw ~.mm A—.mv m.nm Am.mV m.~m owvwm .I m a some man \s A :\N v H m ~.N~ IIIII m.- lulu: ¢.- Hauofi o.e~ Am.cv w.m~ Am.ov ~.e~ hmHHw> —.- Aw.cv a.- Ao.ov m.- macaw o.- Aa.0v n.- Am.ov ~.- oumfim .I AEV unwwon some \m qsfim IIIIII mmdm IIIIII mmgm Hence Nun“ Am.moHv emofi Am.~m~v mowfi zuHHm> exam A~.momv comm A~.amv NNNN macaw «mew AN.~omv woqm Aa.om~v ccmw warm: \m Ae;\.ozv sawmeme some N.m~ lull: m.m~ lulu: ~.m~ Hayes w.- Am.ov ¢.~d Am.~V H.q~ muHHm> m.m~ Am.~v «.mfi Am.~v ~.w_ macaw ~.s~ AN.~V ~.m~ Aa.Hv H.o~ owmwm .l A.ozv newcomm sous \N IIIIIII nouum wummnmum.H.:maz.lllllullll can: unseen seam unsuom um: hammuwonoa vacanaoo meow om“; nouoenuma unseen \H zzmmquAOu can mean mafia >9 mnmuoemnmm ammuom mo moaam> .m~ canny 148 .m + Auaxav + Na + a u » .muowoumne .ucmuwmfiawfim was Haves mumHmEou a“ m sues aofiuumuouafi 02 .I \N .Hmvoa an» was man moanedvaa mHHmu may now one muouuu mummamum .xmmumumm am up vouaowmam aw Ha>mH newsman no mzu um seawuomumu new no mucuumm you muaauwwwcwwm can x cm :uws macaw mum muouomm sesame: maowuomuoucm .hznmquQOu I H .ocrs mafia I NA .umuaemuma unau0m u w HH< .mmHnEmm 0“ was mean omwa swan way can cowuwmoa owzamumOQOu some you monEmm mg mm: anon mafia use use .cowum>mflm I m can “cowumuoa mewsoHHou ecu mm: sedan mmuuoneu maovoa re (*3 M moo 00—4 Mx‘l’ffi «.mq~ o. macaw enema A~E\wxv nouuca owaewuo Hence suHHm> macaw emcee \m ANV Hugues omammno annoy suHHu> macaw amuse I. 50 \e An 3 32:2. is» Hence maHHm> macaw amuse 1.xae\sv awesome \m AeoaaaseouV .mz «Hams 149 .su I Aunouom :«muv > was um I AmmoHOH umav N ”Na ha mamhamam mumumamm 03» .ouowouosa .Hmmoa ouuamaoo cw uamoauwcwwm was AmeAv 3 .«Amxav + «m + «a . Asmohos cause u was “Range + m + a u Asmosou Case » "NA an mumhaaam oueuaaam can .ou0u0u05H .Hovaa uuaaaaou :« uaao«mwawam one: Amxaxugv was .AmeAv .Amxav \m .samxav + m + «a I Ammouow campy » mam “Amxsv + m + a I Aumou0w usav w “NA an emphases mumumnum o3u .oHOHmuenh .Homoa ouoamaoo asu a“ manna coauomuouaw unmouuwawwm any one: Amxnaxav can Amxav I Ammauou away » l mamas; menu mom momhamaa oumumaom oau .uHOHmumsa 3 .eu I Ammouom campy r use new .sm I Aumuuom awauv » vac um I AmmuHOH usav w I muamaoc menu now "NA an .Humoa uuoaqaou unu aw auou cowuumuouaw uamuuuuawam muao saw no: AmeAv \m Aeoaaasaouv .n_ magma 150 Figure 23. Distribution of data points in the covariance model I = T + LZ + TxLz + E where Y = the number of species per 500 m2. T = topography. L2 = life zone. and E = elevation. None of the component terms of the model was significant. 151 .e. 3:23; 002 on. can co. oo- 02. on. 80 I11 1 a d a d q 0 e 3:; I 8.3 a o 4 .8... 5.... a 4 959... o e as. I I I II II o. I o o U DO 4 D Iu I O D 4 O I 0‘0 I d I and d 1n. 4 a 4 40 o a on I4 0 D O o 4 [on 4 I 4 04 a 4 4 4 4 run a a a ---10|dz|ll 00; led sspsds 001], so Isqmnu——— . MN shaman 152 species are adapted to the valley plots where wetter soils and occasional flooding preclude them. In comparison, tree species diversities in the Smokies of the United States were highest an intermediate sites along a moisture gradient (Whittaker 1965). The extremely high rainfalls in the colorado forest would render all topographic positions wet, but the ridges would be the most mesic because of drainage. The decline in species numbers with an increase in elevation is a common phenomenon in both temperate and tropical plant communities (Beard 1949; Glenn-Lewin 1977; Wadsworth 1951; Whittaker 1956, 1960). With ascent from the tabonuco through the colorado to the dwarf forests of the Luquillo Mountains, 0.4 ha plots average 45 to 55, 35 to 45, and about 15 tree species, respectively. The lower average number of species in rain forest as compared to wet forest may be due to environmental factors. All rain forest sites are to the windward where the effects of greater rainfall and cloud cover are similar to those of increased elevation in that solar insolation is reduced, soil water content is increased, and organic matter mineralization is slowed down. Tree species common to lower elevations are not as competitive as high elevation species in the rain forest. Consequently, species richness is reduced. The comparison of numbers of tree species in this study with other research plots in the colorado forest is confounded by sampling technique and plot size. The 500 m2 plots used in this study varied from four species on a high elevation wet forest valley plot to 28 species on a mid-elevation wet forest slope plot (Fig. 23). For all 75 temporary plots combined, or 3.75 ha in total, 87 species were found (Appendix Table 2). Virtually the same number of species was found on the combined 2.8 ha of the 153 seven permanent plots (Table 4, Part I). Individually, the permanent 0.40 ha plots ranged from 35 to nearly 60 species for all stems 24.1 cm in diameter (Fig. 15, Part I; Weaver 1983). The permanent plot with the most species was located near the transition between the colorado and tabonuco forests and contained several species from the lower forest. In earlier work, about 60 species 210 cm in diameter were reported on 4 ha of contiguous virgin colorado forest (Wadsworth 1951). Previous estimates of species richness from 0.4 ha plots in the colorado forest, although valid for the plots sampled, only contained 1/3 to 1/2 of the total species of the forest. This is, of course, to be expected. The plot samples estimated alpha diversity, or that found within a community, whereas the forest-wide sample included beta diversity, or the relative extent of differentiation of communities along an elevational gradient (Whittaker 1965). The comprehensive design and larger area used in the sampling of temporary plots included a number of rare species that increase the richness estimate. Stem Density The mean stem density for all 75 plots was 2174 trees/ha and was significantly related to elevation in the rain forest but not in the wet forest (Table 13; Fig. 24). Individual plot values ranged from 1060 to 4660 trees/ha. Although there were consistent differences by topographic position with valley plot densities averaging 75 percent and 70 percent of slope and ridge densities, respectively (Table 13), these were not significant because of considerable variation among plots within each life zone. Ultimately, the covariance model was reduced to a simple linear regression of tree density vs. elevation in each life zone. 154 Figure 24. Distribution of data points in the covariance model I = E where Y = the number of stems/ha and E = elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in the rain forest. but not in the wet forest. Topography of plots is indicated. 155 1 5000 - > 3 z a ‘ 3000- A 2 ,2 n A D A a9 o 86% D U 0 All ! O "' Uf p. >~ Iooo ‘98 0 O a i- M q.- Il I 1 _L_ .‘2 600 700 soo soo lOOO (L N 5 Legend 0 m to Image 5 El Slope 0 Valley '0 0 In 5 a F i 3 sooop 2 m *- 3 4000- A - 2 O o u. E 3000 e E 3 z a: 2000 l I IOOO | l . . . . 600 700 800 900 lOOO Elevafion m Figure 24 ( ) 156 The decline in stem density reflected the elevational gradient sampled, varying from fewer stems at lower elevations where trees were larger to greater densities at higher elevations where trees were smaller. Moreover, the topographic variability in stem numbers also reflected a gradient in stem size with the largest trees in valleys and the smallest trees on ridges. The mean tree density for all plots was high reflecting the selection of closed canopy stands for the ordination study. Other work in the same forest using 0.40 ha plots showed mean stem densities ranging from 1635 to about 2190/ha with an average of 1860/ha (Table 3, Part I; Weaver 1983). The 0.40 ha plots contained some gaps created by the loss of canopy stems. Tree Size Gradient in thg Forest The mean height for all dominant and codominant stems on the 75 plots was 12.5 m.and was significantly related to elevation in both life zones (Table 13; Fig. 25). Individual plot mean heights ranged from 8.1 to 17.2 m. Although there were consistent differences among t0pographic positions within the life zones (Table 13), these differences were not significant because of considerable variation among individual plots. Ultimately, the covariance model was reduced to a simple linear regression of tree size vs. elevation in each life zone. At low elevations in the colorado forest, some of the tallest trees were found on ridges where Dgcrygdgs gxgglsg, Manilkgrg higgpgggg, ngggi§_ ggggii, and other species from the tabonuco forest, occasionally grow. At higher elevations, ridges are often exposed and develop a "dwarf” vegetation (Beard 1944, 1949; Wadsworth 1951) because of the adverse conditions. At 157 Figure 25. Distribution of data points in the covariance model I = E where I = the mean height of dominant and codominant trees on the 500 m2 plots and E = elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in both the wet and rain forest. Topography of plots is indicated. 158 O .m .m D A D 0 Am A» o D A O o 1W .0 V. 40 A a m «up. a 9 mm.“ D c RSV O L ADO .. A... 1m 4m A m an o o _ _ _ _ _ W e . W n w m s s s .33... .03 .33... fem lllfqocooqok 3 BE «Econ ton eczema; 32:5 3...: 222.. coo: 111 (ml Elevation Figure 25 159 all elevations, trees on the better protected valley plots were among the tallest. The mean basal area, volume and biomass for all 75 plots combined were 45.4 mzlha, 252.9 m3/ha and 148.2 t/ha, respectively (Table 13). Neither topography nor elevation explained a significant portion of the variance for basal area differences in the wet forest whereas topography and the interaction between topography and elevation did in the rain forest (Fig. 26). Volume and biomass results were similar. Again, in wet forest, no significant trends were apparent. In rain forest, however, topography, elevation and the interaction between both were significant (Figs. 27 and 28). The comparatively low basal areas and volumes in valleys were partially caused by differences in stem densities (Table 13) and species’ compositions. At mid-elevations in particular, valley plots are dominated by palms which averaged about 15 cm in dbh whereas slopes and ridges at the same elevation contain proportionately more hardwood species that occasionally attained much larger diameters. One notable exception was a wet forest valley plot near 800 m elevation which contained a large Cyzilla racemiflgra (Fig. 26). In a forest with generally small diameter stemm, a few large trees influence plot values considerably. At higher elevations, the increase in numbers of stems and decrease in mean heights of the canopy trees are more pronounced within the rain forest as compared with the wet forest. Tapographic differences in tree size diminish and the forest is more uniform in size, in particular, in the rain forest (Figs. 25-28). Elsewhere in the Caribbean, including the Lesser Antilles (Beard 1944, 1949) and Jamaica (Asprey and Robbins 1953), a similar decline in forest stature over an elevational gradient was observed. 160 Figure 26. Distribution of data points in the covariance model I = T + E + TxE where Y = the basal area (ma/ha) on the 500 m2 plots and T = topography. E = elevation. and TxE = interaction. according to life zone. The topography and interaction components of the model are significant at alpha = 0.05 in the rain forest. but none of the components are significant in the wet forest. |.. >01 £2 a' o 2 IL a~ o I: 3 o p. 31 > An ‘- o 0. N E O O K) b o a. a .z A? E o o I- 4 ‘5 m a “T5 I Ch 5’ 0 ° u. I! I: lo '0: Figure 26 161 lOO- ° 80-’ Ah so - 13 El A_£}_, ° “‘lfldge A Valley 40 W Slope O o D 20-' ° 0 l 1 l l 600 700 800 900 1000 Legend Al Ffldge 13 Eflope C) VaHey lOOb Siope VaHey lfldge l 600 700 800 900 1000 Elevaflon (m) 162 Figure 27. Distribution of data points in the covariance model I = T + E + TxE where Y = aboveground woody volume (ma/ha) per 500 m2 plot and T = topography. E = elevation. and TxE = interaction. according to life zone. All components of the model are significant at alpha = 0.05 in the rain forest. but none of the components are significant in the wet forest. 163 500- | A A r- A 4 | 3 00’ A D A I 3 m DA 0 u. _—=AE z: 300 W A0 — Ridge " Valle 3 " £3903 419— J 3" Slope, - 3 200- o D D A 3' ° 0 a Q o n o A o o D O t- lOtip ’s D o— 1 1 1 l _j .2 600 700 800 900 1000 a N E 0 Legend CD in A Ridge I. D 51090 3 0 Valley ‘6 i m 600 E g 500 3 '5 .. y. at 4C") 0 C L 3 .‘2 2 300 | c: O : a: zoo 100- l l 1 l 600 700 800 900 1000 Elevation (ml Figure 27 164 Figure 28. Distribution of data points in the covariance model Y = T + E + TxE where Y = aboveground woody biomass (t/ha) per 500 m2 plot and T : topography. E = elevation. and TxE = interaction. according to life zone. All components of the model are significant at alpha : 0.05 in the rain forest. but none are significant in the wet forest. 165 I a I :- soo- Q C b g‘ a; 400- °. 0 .3 3 o 2 u. 300- A .9 _ 7, 083 0A a A A a g z 2°° a: A i m... “e Goo A 0 L” A +23. 3311’; O '00 ”arr- tr 0 a DD CD " o o u) o J 44 1 l a 600 700 soo soo lOOO E \ Legend It £3 Ridge 3 D Slope E 0 Valley .2 1m >~A - 'U i 500- 'U C a * g 3 400 g L z 5.: a 300 < c c .- O 3 K 200 2 I 100 i i i . . I 600 700 800 900 1000 Elevation (ml Figure 28 166 Biomass values in valleys are lower than on slopes and ridges, and the trends are more pronounced than those for basal area and volume. This is because of the lower specific gravities of trees that characterize the canopy species in valleys. Presoeg montana, the palm, at 0.26 g/cm3, the lowest of any tree within the forest, and Cecgopia peltatg and Sapigm laurocerasus, at 0.29 and 0.38 g/cm3, respectively, are less dense than canopy slope and ridge species which range between 0.50 and 0.70 g/cm?. Similar topographic differences in specific gravity between tree species in valleys and those on slopes and ridges were observed in the montane forests of Costa Rica (Lawton 1984). The basal area and volume values for these plots, and consequently the derived biomass values, exceeded previous estimates for the forest (Table 1, Part I; Figs. 16 and 17, Part I; Weaver 1983). This is because plot selection precluded disturbed areas and plot size was small. Qgganic Mgtter of Forgst Soils The mean bulk density, percent organic matter, and organic matter content in the top 50 cm for all 75 plots combined were 0.92 g/cm3, 9.6 percent and 33.8 kg/mz, respectively (Table 13). In the rain forest, none of the model components accounted for a significant portion of the variance whereas in the wet forest, only elevation significantly explained the variance for all three soils’ parameters (Figs. 29-31). Mean organic matter percent and mean organic matter content were both greater in the rain forest than in the wet forest (Table 13). Other factors being equal, organic matter of soils in humid forests should increase with a decrease in temperature or an increase in rainfall, or with a decrease in the temperature/rainfall relation (Brown and Lugo 1982). A priori, s 167 Figure 29. Distribution of data points in the covariance model I = E where Y = soil bulk density (glans) and E = elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in the wet forest. but not in the rain forest. 168 1.50 - I 1.25 I ‘5 l.00 I so 5 > o g ‘5 0.75 O 2 z 3 3 0.50 - °' 0 o p. 2 0.25 — .D 3 1 24 J J 3 600 700 800 900 1000 a. 5 Legend ”E A R100. 5: D S lope o 0 V0 Hey I‘ E o 1.501- 0 .A 8 3 1.25 - m 0 A D U o o :2 *- ° A o g 3 1.00 w— CA I 3 9A A m- _ I “' 0.75 -° 0 06 o A c 0 ° CI 1 " CM50P- a: 0.25 - E1 1 J l 1 500 700 800 900 1000 Elevafion (ml Figure 29 169 Figure 30. Distribution of data points in the covariance model I = E where Y = percent organic matter in soil and E : elevation. according to life zone. The elevation component of the model is significant at alpha 2 0.05 in the wet forest. but not in the rain forest. 170 800 900 1000 700 600 30- » 0 2 «ache... .0; IIIIII 2:20.30... 3 Legend Ridge Slope Valle A D O 1 900 L 800 1 700 1000 30L .3... .3 3.. 3:22 a m 2 7.30... Eu: 0. com: o IIIIIII (ml Elevation Figure 30 171 Figure 31. Distribution of data points in the covariance model Y = E where Y = soil organic matter content (kg/m2) and E = elevation. according to life zone. The elevation component of the model is significant at alpha = 0.05 in the wet forest. but not in the rain forest. 172 70 - A 60 - | a i s I so .. 3 i u. E .. a. 0 g 3 O .- h D II S '0600 700 soc 900 l000 E". : Legend 0 ‘5- A Ridge E D Slope ‘ 0 Valley 0‘ 8 :3 2 70 r- o A 2 .(E) 60 '- A a D D g «5 so . c: a a It 0 - 8 O D O m on .3 ll. 40 "5‘ A m i; I '5 30 r a I 3 A ° . A 0 A A | 20 - A l0 1 ‘ L l l 500 700 000 900 IOOO Figure 31 Elevation (ml 173 decrease in bulk density, and an increase in organic matter content and percent, would be expected with an increase in elevation in the colorado forest. With an increase in elevation, temperature decreases and both rainfall and cloud cover increase, raising the soil water content and retarding the mineralization of organic matter (Grubb 1971). Increased organic matter, in turn, leads to a decrease in soil bulk density. The gradients in the wet and rain forests, however, are not the same, and the differences between them may account for the bulk density and organic matter trends observed. First, precipitation is greater in the rain forest which is located largely near the summits and to the windward of the Luquillo Mountains, than in the wet forest which is found mainly to the leeward of the mountain peaks. Moreover, in the rain forest, the distribution of rainfall is more uniform averaging about 4500 mm/yr whereas in the wet forest it ranges from near 4000 mm/yr at high elevation to levels approaching 3300 mm/yr at lower elevation. Second, there may be a difference in temperature gradients between rain and wet forests as well. Cloud build-up is common at 600 m to the windward of the mountain peaks, or in the rain forest life zone. Here, where condensation is occurring, rising air should cool at the moist adiabatic rate. To the leeward, after dissipation of the clouds, warming should occur at the dry adiabatic rate, or roughly twice as fast, yielding a more pronounced temperature gradient. This phenomenon, or rain shadow, is characteristic of areas to the leeward of mountains lying in the path of moisture laden winds from the ocean (Rummey 1968). Another interesting observation is that the amount of organic matter on slopes exceeded that on ridges or valleys (Table 13). Wind and heavy rainfall in the Luquillo Mountains are probably contributing factors to this 174 phenomenon. Wind on the more exposed ridges could distribute leaf litter to the leeward of crests as was observed on steep topography in central New York State (Welbourn g; 2;. 1981). Moreover, runoff from the ridges during heavy rains is rapid and should be instrumental in moving organic matter downslope where it may be trapped locally in route or deposited in valleys. In the latter topograhic position, it is washed downstream during peak flows. Recent work adjacent to one of the plots used in this study showed that the downslope export of organic matter ranged from 1 to 4 g/mz/day (Lugo and Brown 1981). Oggination 9f Arbozescent Vegetatign Several reciprocal averaging and polar ordinations were computed from various combinations of sample plots and species. The most informative ordinations employed a large complement of species. This was probably because of the variety of habitats studied including plots to the windward (rain forest) and leeward (wet forest), at high and low elevations, and on different topographic positions. In most instances, polar ordination program and user selected endpoints yielded similar patterns. Ordination 1 - sample plogs fgr the entirg forest The first reciprocal averaging ordination contained all 75 plots and 39 species and yielded a sample, or plot ordination (Fig. 32). The first axis extending from the upper right corner toward the arc of plots in the lower left was topographic and partitioned valley plots from slope and ridge plots, with few exceptions. The second axis was elevational and extended from the lower portion of both valley and combined slope and ridge groupings to the top of the graph. A contour line was used to separate plots above 175 Figure 32. A reciprocal averaging sample plot ordination comprised of 75 plots based on 39 tree species for the entire colorado forest. Two configurations are separated by the dashed line. Group A contains mainly valley plots ranging from low to high elevation. Group B contains largely slope and ridge plots. again ranging from low to high elevation. The 800 m contour. the midpoint elevation. is indicated. 176 32.5 c. 2.2m nee 003: e322 E0: 20:35 2.32033. ~m 93mg 4 30.. :3 ..o... 82m a 82: :53. - m to... 3:2, 5...: - < 53> I P... 4 I 4 :2... I n I 4 .25. 4 m... 4 :52... m. ‘ U D . m 4 4 W. I 0 / I 4 1 e / / / I u. / 0 I / e / I I 4 4 n I / I .. fl I o e o e e e / 4 I :58 H e / I .. o o o s I o 4 w. e 4 e / I 4 0 £000 0 / I4 :2: I I I / 4 :3: I / I 177 and below 800 m. The noticeable dips within both plot configurations were comprised of high elevation leeward plots. Ordinationggl: species distribution for the entire forest The second reciprocal averaging ordination produced simultaneously with the first yielded a configuration for the 39 species (Fig. 33). The axes were the same as in the first ordination. For section A in the upper right corner, each of the five species was recorded in valley tOpography more than 65 percent of the time. The eight species in section B, in contrast, were found in valley topography only from 30 to 60 percent of the time. Indeed, some occurred more frequently on other topographic positions, as indicated. The 26 species in section C were found predominantly on slapes and ridges. In summary, six species occurred predominantly in valleys, five on slopes, 18 on ridges, and 10 species showed distributions without a clear majority of occurrences on any particular topographic feature. In all groupings, species along the bottom of the graph were more common at lower elevations and those at the top at higher elevations. In ordination 1, the prevalence of Prestoea montan , regardless of elevation, aspect, or forest classification, was the main feature that characterized the valley plots. Other species that were common in valleys included Sapium laurocerasus and Citharexylum caudatum at low and high elevations, respectively, along with Cecropia peltata and Croton poecilanthus. The ridge and slope plots that grouped with the valley plots were both located at high elevation to the windward in an area bordering a palm.brake. Conversely, the valley plots that occurred within the slope and ridge configuration of plots were located at high elevations bordering dwarf 178 Figure 33. A reciprocal averaging species ordination comprised of 39 tree species on 75 plots for the entire colorado forest. Three configurations are separated by solid lines. Group A contains species predominantly found in valleys. from low to high elevation; group B. species with 30 to 65 percent occurrence in valleys. from low to high elevation; and group C. species with combined ridge or slope occurrence greater than 75 percent. from low to high elevation. See legend for more information on topographic occurrence of species. 179 >e_.o> o. 2.2m ace «of: no.2: E0... .5305 azaoaoao... Incas. I..I.._z‘ 32.92.33: uuuuwuoooa‘ Booze‘ ‘ sets... 2:39.33 *obA .0320 .I III; a so... Inns“! :3...-:::...» ’IIaon .502; I I 2.2.2.3.... 1.: A £52. a l :3332. u.lu..........3.23.233: o now-ezuocov: tea... 3.9-II :28 I. ”wanna... x. I‘ .2: no: .ioum. 3.9..» .23.. I o 2...... a... .909... .883 3.2.» ca...» I 2:25 .gN. 3.0:» :3: I 2.8.... :22... ozuom. o.¢.oao.4* 2.3.8‘ 2.2-3......» 99.3....II 93.3....- * U 2.2.232... cocoa; ‘ al.232- 2.3.0:... 0 ‘ :2: 39:8 2.92:3... 2.30.50..- ..:2.I¢..oa 2.0.39.2! 4 29:0 .2300... 9.0.3.390. 3.282.... .x. 3.3.. 2....u‘l I (3.: ‘ 0.3....35 3:332... : 3.3.3.8 25.-3.30: X. 32.333‘ nos.ou.;...* loos-lea: accuse:- e..o..oeI 30...... 2.5.3202... .395 3:0 Ia.ceoeou.eu* ‘ I :II; 39......3.‘ :93... 2:3 “ices-=30 I330 I... .2. 030...... 2.3qu I I III..I.I 3.....- II.oIIeu I 0.2-one». 12:-.2. :u Ice..2.oa *1 2.35:; :39... 32.-sect: :21". 2...qu XX. mm ounm.m [ouounnela up"... 0; MO‘] won wegpmg 180 forest and were dominated by hardwood species characteristic of that elevation. In ordination 2, the species along the bottom of the graph were more common to the leeward 8 and W aspects than to the windward, for the elevations sampled in the study. The species in the middle were common to mid-elevations to the leeward and in lower to mid-elevations on the windward plots. The species at the top of the graph were found in the highest elevation plots regardless of aspect. Ordination 3 - composite plots The third reciprocal averaging ordination contained a composite sample of ridge, slope, and valley plots grouped according to elevation, or 25 composite plots and 39 species (Fig. 34). The species and plot data were the same as for the first two ordinations. Four major configurations of plots were observed: leeward wet forest plots to the W at low to mid-elevations, leeward wet forest plots to the S at low elevation, windward rain forest plots at low to mid-elevations, and windward rain forest plots at mid- to high-elevations. In general, the rain forest plots were grouped to the right side of the graph. In the first grouping, all plots were found to the leeward W and showed a gradual gradient in elevation from the left to right. In the second grouping, all plots were found to the leeward 8 except #20 which was located at 860 m to the leeward W. Grouping 3 was mainly lower elevation plots to the windward. However, two leeward plots were found. One was located at about 910 m.to the W, and the second was located at about 700 m to the H, but considerably farther N in a more humid area of the mountains. Grouping 4 was mainly windward high elevation plots in the rain forest. The 181 Figure 3h. Reciprocal averaging sample plot ordination for 25 composite groupings of ridge. slope and valley topography at the same elevation. based on 39 tree species for the entire colorado forest. Groups 1 through u are partitioned on the basis of life zone (coincident with aspect) and elevation. as indicated. Exceptions are explained in text. 182 . 3.2 4.. 5.3.... .o... is... .5: s... a .II.‘ :3 I 93.9.4 IOO-.0n0 .II. IL .0. .I. IcIIIIJ- : [000000. .IIIII I38 .Ut— 2:2... .3 I896nI :28 SI .u. 2!: . a. [2.8. .IIIIu I38 .08. i5! 2 31:1 '3 ‘suoas mm "duos cm muswam 183 exceptions were windward wet forest plots 14 and 15, at 900 m and 850 m, respectively, and plot 18 which was located to the leeward W at 960 m. In summary, the plots sorted largely by aspect and elevation with exceptions for high elevation leeward plots. Ordination 4 - leeward wet forest plots to the west The fourth ordination was polar with program selected endpoints that contained a subset of 27 plots and 38 species located in wet forest to the leeward (Fig. 35). The subset produced both topographic and elevational groupings. The topographic groupings ranged from valley plots at the top of the diagram through slope plots in the middle to ridge plots at the bottom. The elevation groupings ranged from (740 m on the left, through 750-860 m in the middle, to 2900 m in elevation at the right of the graph. A similar polar ordination for windward rain forest plots produced a less decisive configuration and is not shown. Other ordinations Reciprocal averaging ordinations of species grouped by topography, that is, separate ridge, slope and valley plots, were also conducted. Each ordination included 27 plots to the windward of El Yunque (simultaneously rain forest plots) combined with 30 plots to the leeward (simultaneously wet forest plots). Species ordinations were similar to ordination #2 (Appendix Figures A-Z, A-3, and A-4). Inceptisols and ultisols are found in the colorado forest. Species ordinations using these soil groupings however, produced no trends in species composition by substrate, confirming earlier studies of a similar nature (Wadsworth and Bonnet 1951). In other words, the extent to which 184 Figure 35. Polar sample plot ordination for 27 individual leeward wet forest plots based on 38 tree species. Solid lines separate different elevational groupings of plots and dashed lines separate different topographic groupings of plots. 185 .60—9050 3.3.3.0; 20003 E 30. 325.4 .:§.I :=£.. -ecoooa wagpnag :gud baboon; mm shaman 185 2.2.3.9 3.32.2.0; 232.‘ ...:._I ..:.> . -eeoooa wagpcag aqua 0160001 mm mn:m.m 186 soil factors can be predicted by species composition data is still not certain (Knight 1975). Species distributions The occurrence of major colorado forest tree species is shown by tapography in Figure 36. Distribution by elevation, and aspect, which is largely coincident with life zone, is shown in Figure 37. Prestoea mont n , Croton poecilanthus, Cecropia pgltata, Cithgrexylum caudatum, and Sapium laurogerasus are all more common in valleys than in other topographic positions. Prestoea is found in excess of 125 stems/ha throughout the forest but appears more abundant between 800 to 900 m to the windward, and at 750 to 800 m to the leeward. Croton is also found throughout the forest and shows peaks with >100 stems/ha between 700 to 800 m to the windward, and at 750 to 850 m to the leeward. Cecropia fluctuates at low levels averaging about 10 to 20 stems/ha throughout much of the forest. Cithgrgxylum averages about 15 stems/ha, but has greater abundances between 800 to 900 m, whereas Sapigm is found at levels below 15 stems/ha, except for one area at 750 m to the leeward. Qggtgg_s athulata, Micropholis chr so h ll'des, Eugenia boringuensis, Cygilla r cemiflor , ngnianthus salicifolius, Miconia tetrandra, Mataybg dominguensis, flyrgi§_f llax, and Brysonima wads orthii, all show greater abundance on ridges than other topographic positions. 9295;; is common from 650 m to the windward through 900 m to the leeward where it ranges from 75 to 230 trees/ha. Micropholis has some stems to the windward but is clearly more abundant to the leeward below 800 m where it ranges from 175 to about 300 stems/ha. Eugenia shows a definite peak between 850 m to the windward through 950 m to the leeward where it ranges from 150 to nearly 500 187 Figure 36. Distribution of the major tree species on 3.75 ha of colorado forest by topographic position (V = valley. S = slope. and R = ridge). The specific gravities given in g/cm3 are discussed later in text. 188 Presioea montana “’00— Cal to M m . o F— azuazulozum Micropholis Tabebuia 500 O 26 O 74 oarclniaefolia ridgla x... o 8 E .m Ocotea Croton 500- spathulaio Micropholis poecllanthus ~-—— Cardin 11 in II id l____ 0'62 borinquensls c rysop ° es l25- ISL-_m 0'58 0.50 0 Pi _L. 3 c m ‘ " Eugenio Y' o . b — 0- ISO borlanensls racemiflora c'::|::|ana Hoenianthus ._ 0.75 0.53 salicifollus ° 75- 0.8: s. 0.90 3 [—1 E O I z Miconia M . re a IOOh- tetrandra MO'OVDO ’fallax dominguensis O 7 fl 50- ' afl— Magnolia 0-7 0.75 solendens Cecropla 50 Brysonima oeliaia . wadsworthii . Cliharexylum Saplum 0.54 I—i caudatum laurocerasus 25- 0.29 {L - 057 w ' 0.38 V S R V S R V S R V S R Figure 36 Topographic Position 189 Figure 37. Elevational distribution of common colorado forest tree species to the windward and leeward of El Yunque peak. Each point represents the mean value for a ridge. slope. and valley composite plot at a given elevation. gavclnioeiolio Iicvoohoils a” ..I:. o’ ... : .... g ’ g Q ’ 3 s . 3 a I Ig‘g i .z” 1% t P”/ 4’.____.:__,._..'._._ 42...... \ "a \r:"~. g : \. § 2 I}, ° “ i '3' .. '6 190 IOO on] so”; go noun” in) Elevation Figure 37 191 I b o I I I 3 g I :l 3 2 2 . 2 o ‘- «are 46.: II. I 35': O‘CO :0..- tons. _. I 3-‘h 0 .- 2%20 3 . 5 t 0223 u 8 u u 0..... I.... .0" WI «Guard 400 such mu sou; lo noun" 600 700 l000 (In) Elevation Figure 37 Legend 192 Croton poocllonflun O O O 0 O ‘ 0. b .\ O .0 i o {h J ’/ . D ’1’ O l” .0. I : l .” '0'. _ ’4’, '0 8 | ’ .g. . ’ I ’ ~ I O C 00......1 g “\ .0 00....O..... O \- O z D .......ooo;:h"~ “ 3 o .3 — ‘6 = 3 3 s . 8 U a i g 5 a '- C o o g 'E.2 = i O b 9 h 0 U 40 - Soc 1 I § § 3 on] son; go aqua" Elevonon Figure 37 193 Figure 37 194 .E. 39):.- 80. .3233; 000 I 'I 20:2. 2.9.0.0 8.3.2.3 lat-203:0 22.022 2.823 232...: 2.2:! 93.904 22.2..) 1 an. um unnwwm mu sun, :0 ANN"! 195 stems/ha. Cyrilla is most common between 600 through 700 m to the windward where it ranges from 80 to over 200 trees/ha, and from 750 to 950 m to the leeward where it varies from 50 to about 200 stems/ha. figenignthus averages about 20 stems/ha from 600 to 850 m to the windward, and about 100 stems/ha from 800 to 950 m to the leeward. Miconia levels are low to the windward, averaging from 5 to 10 stems/ha between 700 to 900 m. To the leeward, it averages about 75 stems/ha between 650 to 800 m. Hataybg is limited to the leeward side where it varies from 25 to 140 stems/ha between 650 to 850 m. flyggi§_has a bimodal distribution ranging between 10 to 50 stems/ha to the windward and from 15 to 70 stems/ha between 700 and 950 m.to the leeward. Finally, Brysomina shows a bimodal distribution with about 20 stems/ha between 600 through 900 m.both the windward and leeward. Calycogonium sgugmplosum, Micropholis garciniaegglia, Tabebuia rigida, ng§i§_krugiana, and Magnolia splendens all show abundant distribution on slopes and ridges with a slightly higher number of stems per hectare on the former. Calycogonium is a very common tree that shows its greatest abundance at highest elevations where it reaches >650 stems/ha. Migrgphglis appears bimodal, with >135 stems/ha between 600 through 800 m to the windward, and between 850 to 950 m to the leeward. Tabebuia gradually increases from 50 stems/ha at 600 m to the windward to >750 stems/ha at 900 m. To the leeward, it declines rapidly from >300 stems/ha at 950 m to 10 stems/ha at 900 m. Clusia is found from 600 m on the windward to 700 m to the leeward, but shows a definite peak of 350 stems/ha at 950 m to the leeward. Magnolia has a bimodal distribution ranging from 10 to 75 stems/ha between 600 through 850 m to the windward, and from 10 to >50 stems/ha between 750 through 950 m to the leeward. 196 Cordi§_boringuens;g is unique in that its abundance by topography is essentially equal. It occurs throughout the forest ranging from 15 to 275 stems/ha and shows a peak from 750 to about 900 m to the leeward. In summary, Cglycogonium, Tabebuia, Eu enia, and Clusia show unimodal peaks at higher elevations. 0cotea is more abundant to the windward and at high elevations to the leeward. Prestoea, Micropholis gargjniaefolig, Cr ton, Cyrillg, M rci , Br somina, and Magnolia have biomodal distributions. Prgstoea reaches higher elevations to the windward while the remaining species with the exception of Bgysomina reach higher elevations to the leeward. Brysonima appears evenly distributed with regard to elevation on both the windward and leeward. Mico ia, Micropholis ghrysophyllgides, Haenianthus, Sa ium, and Qg;dia_are found on both aspects, but show greater abundances to the leeward. Mayayba was found only on the leeward, and Citharexylum and Cecropig were found in low numbers throughout the forest. The differences in plot groupings in ordinations 1, 3, and 4 are related to complex environmental factors. Rainfall at the windward edge of the Luquillo Forest, at about 200 m elevation, averages 2300 mm/yr and increases to 3560 mm/yr at 450 m elevation (Fig. l; Crow and Weaver 1977). Continuing the windward ascent to the lowest elevations of colorado forest at about 600 m, rainfall averages over 4000 mm/yr, and at La Mina located at 715 m, it averages 4600 mm/yr (Fig. 2, General Introduction; Wadsworth 1948). Pico del Este (Weaver 1972; Brown g;_al, 1983) and Pico del Oeste (Baynton 1968) located to the southeast of La Mina receive similar annual rainfalls. El Yunque slightly to the west receives about 4000 mm/yr (Briscoe 1966). The latter 3 sites average about 1050 m.in elevation. Descending to the west, Rio Grande at 475 m receives 3300 mm/yr (Briscoe 1966) and El Verde at 450 m receives 2920 mm/yr (Smith 1970). Both of these 197 sites are located at high elevation in tabonuco forest. Descending to the south, Rio Blanco 4 at 550 m, just outside the colorado forest, receives 3870 mm/yr (Brown g£__l, 1983). The last site, Cubuy, to the southwest of the forest at 450 m elevation, only receives 2000 mm/yr of rainfall (Schmidt and Weaver 1980). The lowest elevation colorado forest at slightly over 600 m elevation to the leeward on the western part of the forest receives about 3500 mm/yr of rainfall. This is considerably less rainfall than received by the windward sites at the same elevation. Mean annual temperatures measured along with the rainfall also vary with elevation. At the windward edge of the forest, temperature averages about 23°C. This declines to 18 or 19°C at the summits and then increases again to 23°C and the leeward edge of the forest. Two serious problems emerge with the climatic data, however. First, many of the rainfall records are not concurrent and therefore do not provide direct comparisons of the rainfall gradient within the mountains. Second, except for the La Mina site, the weather stations are located in forests above or below the colorado forest. The characteristic base of trade wind cumulus clouds in the Caribbean ranges between 600 to 750 m (Baynton 1968). In essence, this includes the entire colorado forest of the Luquillo Mbuntains, and is a major factor defining the border between the tabonuco and colorado forest. The duration of cloud cover increases with elevation, and in summit areas such as Pico del Oeste, the forests are enveloped in cloud cover for more than 60 percent of the daylight hours (Baynton 1968). Incident solar radiation on the mountain summits, for example El Yunque (Briscoe 1966) and Pico del Oeste (Baynton 1968), is reduced by 40 percent of that received on the north coast of the island. 198 With regard to ordination 3, greater rainfall and cloud cover to the windward may be the critical factors separating low elevation plots to the windward from those of the leeward. Several tree species common to the tabonuco forest are scattered near the transition with the colorado forest. These species appear to occur at higher elevations to the leeward than to the windward (Fig. 33). Indeed, in Beard’s (1949) excellent summary of Caribbean forests, the occurrence of dwarf forest at lower elevations to the windward is noted. The grouping of plots by elevation is a function of the combined effects of rainfall, temperature, and cloud cover gradients within the mountains. The fact that high elevation leeward plots group with windward plots is related to the interaction of these factors near the summit areas. Rainfall on Pico del Este, the eastern-most peak in the Luquillo range, was slightly greater on high elevation leeward slopes than either on the summit or on high elevation windward slopes (Weaver 1972). However, the greater interception of cloud moisture on the windward slopes and summits delivered practically the same amount of moisture to the soil surface on all three sites. These same high elevation leeward slapes are classified as lower montane wet forest (Ewel and Whitmore 1973) yet separate with lower montane rain forest in the ordination. It appears that the small area encompassed by these plots is rain forest and not wet forest. Actual rainfall and cloud moisture interception data for this particular area, El Toro peak, however, are not available. Ordination 4 reflects elevational and topograhic gradients. Again, the combined effects of rainfall, temperature, and cloud cover appear responsible for the elevational sorting of plots. The topographic sorting may be related to another complex of variables including a moisture gradient 199 ranging from better drained upper slopes and ridges to more humid lower slopes and valleys. Other important factors related to the topographic gradient include better light conditions and greater exposure to winds on ridges, higher organic matter content on sIOpes, and recurrent flooding of valley environments. However, while individual plots sorted by topography to the leeward, only valley plots were grouped separately to the windward. This phenomenon may be attributable to the more pronounced effects of topographic differences in slightly drier montane conditions that characterize the leeward plots. The effects of less rainfall and cloud cover combined with greater solar insolation to the leeward allows scattered species from the next lower forest to survive. Their occurrence is helpful in establishing topographic trends. Rageneratign 9f Canopy Species 0f the 20 species that reach canopy size in some part of the colorado forest, 0cotea s athulata, Tahabuia ri id , Clusia krugiana, and especially Eugenia b rin uensis, are more common in the dwarf forest. The remaining 16 species occur either primarily in the colorado forest, or in the transition zone at about 600 m elevation between the colorado and the tabonuco forests. Prestoea montana, the palm, had the greatest number of seedlings averaging more than 23,000/ha (Table 14). Seven other species had more than 1,000/ha and four species showed no regeneration whatsoever. The number of plots on which regeneration was found generally paralleled seedling abundance. To partially surmount problems of analyzing seedling data based on a single sampling of subplots, several other factors including seedling success, understory success, seed size, and wood specific gravity of the 20 200 Table 14. Summary of regeneration characteristics of canopy species in the colorado forest Frequency of Species Seedlings/ha occurrence in plots1 Brysonima wadsworthii2 47 6 Calycogonium sgaamulggga. 2,487 45 Cecropia peltaga 0 0 Clusia krugiana2 13 2 Croton poecilanthus 1,240 30 Cyrilla racemiflora 33 2 Didymopanax morototoni 20 2 Eugenia boringuensis2 3,413 21 Eugenia stahlii 1,440 18 Haenianthus salicifolius 5,453 40 Magnolia_splendeaa_ 0 O Matayba dominguensis 767 20 Microoholia_chgysgnhvllgidaa 3,720 25 Microphglig sarciniaefelia 547 29 Music fallaxz 273 13 Ocotea moschaaa_ 0 ‘ 0 agate; W2 2 .460 44 Prestoea mantana 23,147 68 fiaaiaa_lauroaeraaaa 0 0 Tabebuia gigggg? 240 16 1Based on 75 plots 20 m x l m.an which regeneration was found. 2Small trees that may reach the canopy locally. 201 canopy species were assessed (Fig. 38, panels A, B, C, D). The species were arranged from the lowest cumulative scores at the left to the highest scores at the right (see Methods). A priori, one would expect each of the factors to increase from left to right, that is, from species with mainly secondary life history strategies to those with primary life history strategies. With some exceptions, these general trends are apparent. Seedlin2,and understory success Eugania atahlii is uncommon as a mature tree, but it has a comparatively high relative density ratios of seedlings and understory trees to canopy trees. Slow growth may eliminate the species as the forest continues to close after major disturbance. In contrast, Haenianthua salicifglius shows a high ratio for seedling success but a low ratio for understory success. In this instance, the seed is relatively large and germinates in the shade but seedling mortality is high. Its survival may be tied to gaps within the forest. The ratios for both seedling success and understory success for Cecropia peltata are the least for any of the species studied. It regenerates and grows in openings. A comparison of the seedling and understory relative density ratios provides an insight to the survival strategy of the respective species. Didyaopanax morototoni, Matayba domin uensis, Micropholis chrysophyllgides, and Haenianthus salicifolius all show ratios of 0.5 or more units greater in the seedling than in the understory classification (Fig. 38). This probably reflects a greater capacity to germinate and grow into seedlings than to compete and survive into later stages of development. Sapium lau c r us, 0cotea moschata, Tabebuia i ida, Calycogonium s amulosu , and Byzsgniga waasworthii all show ratios 0.5 or more units greater in the understory than 202 Figure 38. £3ne1_A; The seedling success of canopy tree species based on the seedling relative densities of canopy species divided by the relative density of canopy size plants of the same species; £3nal_fl; The understory success of canopy tree species based on the understory relative densities of canopy species divided by the relative density of canopy size plants of the same species; £3na1_g; Seed size of canopy tree species. defined as the sum of half the length + the width of the disseminule; and £3na1_n; Wood specific gravities of canopy tree species. The species are arranged in the same order as they appear in Table 18. 203 Specific Gravity (glm’) 03' ”- . i- 0 (en) Seed Size 03'50 mm arcane 5 5 3 .l 5. O .II]J1IJIJI.W. a _ __ Relative Density Ratia- Understary zueoscetaa 2.336 22:50: eoaouo :22. 2:33.. nszazuzau 2.5.3.50: «32:23.2. c330 32223335 2.23222 zeta-.30! 0.5.3qu no.2: 29;: E32363: £3,303». 00 2.2025209 «223222 ocoEeE owe—moan 032. 2.52.3. nice-3552. c.1202 Eugene... 0330 0:203... 2.3.0 manganese. Esiom Relative Density Ratio - Seedlings aeoeeoan 2.230! 22232: announcing 32:53.: a. _ .;u 20:... eacaueu P o 4 223a Species 204 in the seedling classification. This trend may reflect greater capacity to persist in shaded conditions once the seed has germinated. The remaining species have seedling and understory ratios that are approximately equal, or arbitrarily, within 0.4 units of each other. Saed siaa,of canopy species Seed sizes ranged from about 2.8 cm for 0cotea spataulata to about 0.15 cm for Cacrppia peltaga (Fig. 38). Known secondary species such as Cearppia peltata and Didymopanax morptptoni have among the smallest seeds recorded, whereas known primary species such as both species of Micpppholis range in the upper half of the seed size values. Other species for which seral stages have been suggested include Cypilla pacemiflora as a late secondary species (Part I of this study) and Ppestoea pantana as a shade tolerant, primary species (Bannister 1970). The seed data (Fig. 38) agree with these classifications. Seed size varies over a range of 10 orders of magnitude (Harper aa_al, 1970) if compared on a world-wide basis. Observations in temperate climates have shown that species capable of establishing in the shade have heavier seeds than those germinating in openings, and that species growing in advanced seral stages have heavier seeds than those in successional stages (Salisbury 1942; Baker 1972). Similar observations have been made in neotropical forests (Budowski 1965). Large-seeded plants are better adapted to establishment under shade than small-seeded plants because of their access to seed reserves, and their capacity to penetrate litter and humus after germinating (Canham and Marks 1985). 205 Specific gravity of canopy spacies’ woods The specific gravities of dicot canopy species range from a very light 0.29 g/cm3 for Cecropia peltata to a very dense 0.90 g/cm3 for glaaia_ krugiana (Fig. 38). The palm, Prestoea montana, has a specific gravity of 0.26 g/cm9. Usually, slower-growing, shade tolerant climax trees, have higher specific gravities than faster growing, shade intolerant pioneer and secondary species (Budowski 1965; Smith 1970; Lawton 1984). Environmental factors have also been implicated in wood density. Species growing on windy sites, such as high elevation ridge species, have denser woods than those on sheltered sites (Fig. 13; Lawton 1984). Ppastoea and Cecropia, which are most common in valleys, have specific gravities below 0.30 g/cm3, and Sapiaa laur cerasus, Citharexylum c udatum, and Q;pppp_ oecilanthus, also common in valleys, have specific gravities between 0.38 and 0.60 g/cm3. In contrast, Qapaaa,spa§hulata, Eugania paginguensis, and Haenianthus sal'cif lius, all common on ridges at higher elevations (Fig. 13), have specific gravities ranging from 0.62 to 0.81 g/cm3 (Fig. 13). The mean specific gravity of the five common valley species is 0.42 g/cm3; for the nine species most commonly found on ridges, 0.68 g/cm3; and for the remaining six species found predominantly on slopes and ridges, 0.69 g/cm3. Strength is a function of the amount of material in a given volume (Wangaard 1953; Forest Products Laboratory 1955), and exposed trees should have lower rates of windfall and breakage if denser (Lawton 1984). It is interesting to note that Tababuia ri id , common an exposed slopes and summits, occasionally blows down and forms new trunks from existing branches. Its specific gravity at 0.57 g/cm3 is the lowest of all species common to high elevation ridges. 206 Mortality rates are also correlated with wood density (Lawton 1984). Denser woods on ridges would, other factors being equal, lower the rate of gap formation. Recent investigations in the summit forests of the Luquillo Mountains showed that secondary tree species were notably lacking an cutover areas and the site of an airplane wreck (Byer and Weaver 1977). Detailed observations of hurricane damage according to forest type and topography in the Luquillo Mountains are lacking. Bates (1929) noted defoliation and broken crowns after San Felipe, but these observations appeared to be confined to accessible sites at lower elevations. A plausible hypothesis for the colorado forest would be that upper slope and ridge forests, mainly composed of more densely packed (Table 11), smaller (Fig. 25), and heavier trees, would be less affected by hurricanes than valley forests. Tentative Lifa History Classification pf Canopy Trees Table 15 contains a tentative classification of the life history strategies of colorado forest canopy species. The species are ranked according to the cumulative scores derived from.the relative density ratios of seedlings and saplings to canOpy trees, seed size, and wood density (Table 15, footnote 1). Ideas regarding life history classifications for forest trees (Budowski 1965; Whitmore 1975; Pickett 1983) and the influence of gap sizes on species which colonize them (Oriana 1982; Hartshorn 1978) have been used for the following groupings. Shaaa iptplerant pipneer spagies Caaropia peltata and Diayaopanax mprptptoni are typical pioneer species (Vasquez-Yanes 1980). These species exhibit diameter class distributions in 207 .833 3 n .3 u 3 3 3 3 3 «m .u 3.8.93.3 3:33. 3 .833 3 n no i «Z 3 8 .3 3 n n 33:52:. 338 3 SEE an n 3 i n 3 2 3 3 a. a. 323.. 359.“. 3 «33.... «n + on + 9.: 3 3 S N a an 2.33333 oazucaacoe: 2 ¢u\m\o as +. a + New ON a ma NH H n ansucaauuoom nauouu a“ EBB n... + .3 + 2n 2 2 3 3 a _. uoBodanabfi 332383: 2 no\w\o aN + m + on m o a ma a an «usuuozmoma analommwn ed 323 an + R + on 3 3 S a _. a 5:8 38;: 2 no\m\o ma 0 N + me 0H N ca ma H H lamenaaaavm Inqeomoohaao Nd no\m\u NN + n + ado NH «a a Na H n caucuaaucwwmmu awaozaouuwx ~— mEB n + m + 3... 3 NH 3 2 «z ..,2 33:8. 333: S 53:03 3 + an + me o 3 N 3 an «a a3 3 3338. o «233 83+ 8? n .3 S 3 o m s. 323238 2.53. a .3833 R + 3 .. a a 3 a 2 a. a. 3288 338 a «283 3? 2 + S 8 a o n m .m 3 o ¢o\o\ma nom+ ma + o— e m N o m N anamuauousaa saunam n «233 o .. an . m3 5 a n a «u «m 83:33: 333...: a .8310: on n .3 n 3 n m N3 N m m accuououoa genome-$3: n mo\u\ma cc i m I HmH m n n m em 2 auoaufiaauau aaauuhu N 533 3 i mu n Nu N 3 3 3 u an 333% «Ecuuoo ~ dflhfl Hanna 03000 dwau 0¢G~ oaouuauuuamnaau maneum no awcnzu «Aocmuv mason nhuqmcov mango nwcuavaom HhMOumuuvca Nanaunauuuawv awcwxaau moguaam 0>HUQHGOH “Guam—0m HO uflngz V003 UOQm OHUQH hadOCO—u 0>fiuwdwm DOGHU HQUOIQHD .uuouOu ovmuoHou may :« zaocmu ecu :umou umsu moaooam mono mo AEOuuon cu acuv muuaaua umoa cu homecouum umoa aouu wcwxaam .mu wanna 2(H3 .IoIsI nu acumenopbo no INAIAIo I a nunIlaoNI>Iv vaI suabuu Inauocin aIu I nu «huuuaunl nan vouaavou sIa I so "ucuaaoun anusu «noququIqu new IeIzn I a uncquuIIIqu new IvIsI uo nIa I «no .BONquocomou new nIu I u unauaaoun acouoo «II-«Ho I u «huIacouoI Iqu I ma .uoo:o«n I m «IquuoH no unuaaouu quuh .H qum louu noquluowaw sauna. no vIIIa InaIgu .H qus .c IMAIH luau IuIa .mn ouauuu no nogna nuaalaqu .An. can «3 .Iuduv I>uau vInIgoun vquI>Iu I I« couuaa«qu«v I n III “Ipuau uIaIqun vIIuI>Iu I ucdguIOHQmI acuuaauquuv I In a couu:a«qu«v IIqu noun-Ida I IUIlauII cu ucoauuuuaIcu no uaqqun IuIv I a "Ragga IInIAv acquanuquav uIluo: I I: «coauaauuunuv NIluoc haouulaxounnI I z uncuuanuquuv cI>I NHIuII«I0unaI I u «IIIHo quolqu uII—NIII Isa cu III». «a non-s: quIqu >3~I¢0uuuoaounauv I saw: couusawquqv aI>I hNIuIldxouaAI I an .onosuox IIIV .n\AIu¢«x=Iu aqucou v09: + quI vIII + unwavIII + Aunqchu huoquIvas I va I unaaqu IquIam .Auoaaduaoov .n3 «an-a 209 which the number of stems per class is approximately equal. Seeds are small and easily dispersed by wind. They may germinate in partial shade but are incapable of surviving there. The wood is light in weight, and growth is initially rapid with a decline after reaching the canopy (Crow and weaver 1978). The trees are usually short to intermediate in longevity. These species are most common after major disturbance although the occasional death of large trees may provide adequate space for their survival in the forest. As the forest recovers from major disturbance, less growing space is available for regeneration of these species, and their numbers gradually decline (Fig. 11 Part I). Gap 0; late secondary species These species may have diameter class distributions that are equal, that contain a greater proportion of stems in intermediate size classes, or that are approximately equal with a disproportionately large representation of stems in the smallest diameter class. Their survival is tied to local disturbance, usually as gaps in mature forest. Seeds are usually small and the wood tends to be intermediate in weight. Volume growth is fairly rapid until the trees are in the canopy, providing that light is adequate. These species are most common several years after disturbance. gapigm_laurgcer§sus and Clusia krugiana both demonstrate comparatively low relative density ratios of seedlings and understory trees to canopy trees, but show an increase in stem numbers and basal area from 1946 to 1981. Both species are intermediate in size and capable of rapid growth with adequate light. The regeneration of glg§i§_on the previously thinned plot was considerable (Fig. 14, Part I). Cyrilla racemiflora and Magnolia splendgns also show low seedling and understory canopy ratios but decline in 210 stem numbers and basal area from 1946 to 1981. These are much larger and longer-lived trees than the former. Neither demonstrated particularly rapid increment even when exposed to light (Table 9, Part I). It may be that the disturbance associated with the 1932 hurricane was not sufficient to provide large enough gaps for their regeneration, growth and survival in large numbers. Again, on the thinned plot (Fig. 14, Part I), the growth of Cygillg in the smallest diameter class is rapid (Fig. 19, Part I). Hgtayba dominguensis is intermediate in size in the lower part of the colorado forest where its relative density ratios indicate higher seedling than understory success. Its growth is fairly rapid in the upper portions of the tabonuco forest (Crow and weaver 1977) and its basal ares increase in the colorado forest indicates similar growth. Although capable of regenerating in the shade, gaps are required for the species to mature. Fast growth enhances its success. 8 t e s ecies - These species usually have a negative exponential or reversed J-shaped diameter class distributions. The seeds tend to be large and capable of germinating and developing in the shade. The wood is usually heavy or dense, and growth rates slow to intermediate unless the trees are favorably located with regard to light. Trees range from intermediate to long-lived in age. Without major forest disturbance, either human or natural, these trees would tend to increase in basal area and possibly numbers during recovery. Q;g;gg_ oecilanthus, Micropholis chgysgphylloidgg, g, garciniagfglig, Brysgnimg gadsggrthii, Cglyggggniug sggagglosug, Hyrgig {51155, and chteg spathglgtg are the best examples of shade tolerant climax species. All have reversed J-shaped diameter class distributions and show intermediate to high seedling and understory to 211 canopy ratios. Eugenia stahlii and Qggtga_moschata may also fit this category but are extremely rare within the forest. Shelterwood plantings of both species showed good survival after several years (Marrero 1948). Six-year-old Qggggg averaged 0.6 cmlyr dbh growth and 0.7 m/yr height growth. Nine-year-old E enia, however, only averaged 0.3 cm/yr in diameter at the tree base and 0.1 m/yr in height. Tabebuia rigida is capable of surviving in the shade as indicated by its high understory to canopy ratio, but has been observed in previously cleared areas at high elevations. Its diameter class distribution is nearly equal in larger classes with a disproportionate number of stems in the smallest class. It is capable of rapid diameter increment when exposed to sunlight (Table 9, Part I). nggtgea mgntana demonstrates several characteristics of a shade tolerant climax species. The exceptions are related to the fact that the plant is a monocotyledon: the near-normal diameter class distribution which reflects the range of stem sizes and lack of cambial growth, and the extremely light weight of the wood. The palm’s exceptionally low specific gravity distorts its position in the classification scheme (Table 15). Its "slow growth, long survival, apparent shade adaptation, and high moisture requirement suggest that Euterpe [now Prestoea] is adapted to conditions inside the forest and is a normal component of rain forest vegetation, not a successional species as previously assumed" (Bannister 1970). Although particularly common near streams, Prestoea is abundant throughout the colorado forest. Haenianthus salicifolius is capable of growing in shade but field observations of this species show that it is tied to gaps for survival. fiagnianthus has a comparatively low relative density ratio of understory to 212 canopy plants for a shade tolerant species. Moreover, neither its 1946 or 1981 diameter class distribution is truly reverse J-shaped. Diameter growth is rapid in all crown classes exposed to direct sunlight, and even fairly rapid when shaded (Table 9, Part I). Eugenia boringuensis is common at higher elevations where its seedlings germinate under shade. It has also been observed in areas previously cleared of trees, growing through thick tangles of grass and ferns. Of the shade tolerant species, Prestoea is capable of establishing and growing slowly in the understory until gradually reaching the canopy. Eugenia stahlii and Ocotea moschata are difficult to classify because they are rare in the forest. Their growth into mature trees is probably benefited by gaps. Croton, Micropholis a ciniaefolia, Cal c ni m, gyggig, and Byrsgnima are capable of surviving in the understory for long periods but benefit from gaps to reach the canopy. The remaining species, Tabebuia, Eugenia borin uensis, Qgg£g§_spathulat§, Micropholis chgysgphyllgidgs, and Haenianthus all appear to require gaps to grow into the canopy. The last two would require larger openings to mature because they are larger trees. Although this phase of the study emphasized regeneration of canopy trees under closed forest, many of the observations are relevant to the growing body of knowledge regarding the importance of gaps in the success of seedling establishment and penetration of trees through to the forest canopy (Whitmore 1975; Hartshorn 1978; Picket 1980; Pickett and White 1985). Of the 20 tree species that reach canOpy size in some part of the colorado forest, two were classified as shade intolerant pioneers, and five were grouped into gap, or late secondary species. The remaining 13 species were classified as shade tolerant, but of these, five benefitted from gaps and 213 five required them to mature. Two species were difficult to classify, but probably require gaps to mature. Only one species, P toea, is capable of routinely growing through closed canopy. In summary, then, at least 70 percent of the species require gaps to either establish seedlings, or grow into mature stems. These results compare favorably with lowland rain forest in Costa Rica where 75 percent of the canopy species required gaps to mature (Hartshorn 1980). In relatively open forest sampled in the transition between tabonuco and colorado forest, gaps caused mainly by the fall of large Cyrillg gaggmiflora trees averaged 220 m2 (Perez Viera 1986). These gaps are intermediate in size between the 90 an2 average for lowland rain forest in Costa Rica (Hartshorn 1978) and a 600 m2 gap in Dipterocarp forest of Halesia caused by the fall of adjacent trees (Poore 1968). With an average crown-to-bole ratio of 20:1 in the colorado forest, the fall of a single, large tree, 50 cm in dbh, could create a gap approaching 80 m2. However, since 95 percent of the stems at mid-elevations in the colorado forest are less than 45 cm in dbh (Wadsworth 1951), most gaps are smaller. Moreover, many gaps are caused by the slow death of a tree, a branch at a time, in which case, the surrounding vegetation occupies a portion of the gap as it is created. In contrast, much larger openings could form as the result of multiple tree falls caused by major disturbances such as hurricanes. The regeneration and growth of Cyrilla racemiflora and giggi§_krugiang trees during 35 years on the thinned colorado plot (Figs. 14, 16, 17, 18, Part 1) demonstrate the potential of gap species to exploit available growing space. In other forests, it has been shown that emergents and upper canopy species require forest openings for establishment (Jones 1956; Whitmore l 975). 214 The presence of gaps and their variation in size are factors that influence subsequent establishment and growth of trees (Whitmore 1978; Denslow 1980; Pickett 1980). At one extreme is closed canopy forest under which tolerant tree species regenerate and remain as seedlings until a gap forms (Whitmore 1978). Without gaps, they may persist for years, but ultimately, most will perish. At the other extreme, large gaps create favorable conditions for large pioneer species which have rapid growth rates. Existing regeneration may respond poorly or die due to the altered environmental conditions. Small gaps, on the other hand, favor the growth of advanced regeneration (Whitmore 1975, 1978). The classifications assigned to the aforementioned colorado forest species are tentative. As pointed out by Whitmore (1975), each species is unique and many are capable of surviving under different environmental conditions. Variation in elevation, aspect, and topography within the colorado forest, as well as normal gap formation and that caused by major hurricane disturbance, provide a gamut of potential environmental conditions. To these might also be added differences of light penetration through tree canopies varying in foliage density, potential allelopathic reactions about which virtually nothing is known for tropical forest, and physiological capabilities of tree species of different ages and in different developmental stages to survive, or respond to altered conditions. 215 SUMMARY The sampling of 75 closed-canopy plots with an average canopy cover of 91 percent throughout the colorado forest of the Luquillo Mountains of Puerto Rico showed that stand structure and species distributions were related to environmental factors. A covariance analysis of forest structure on 500 m2 plots including the number of species, the number of trees, mean height of dominant and codominant trees, total basal area (mzlha), total volume (m3/ha), and total biomass (t/ha) by two life zones (wet and rain) and three topographic positions (ridge, slope, and valley) with elevation as a covariate showed that: (1) the number of species averaged lS/plot and was not significantly related to any of the environmental factors; (2) tree density averaged 2174 stems/ha and was significantly and positively related to elevation in the rain forest, but not in the wet forest; (3) tree height averaged 12.7 m and was significantly and inversely related to elevation in both the rain forest and the wet forest; and (4) basal area, volume and biomass averaged 45.4 mzlha, 252.9 mg/ha and 148.2 t/ha, respectively, and were significantly related to topography and elevation in the rain forest, but not in the wet forest. The same covariance analysis of soils data showed that: 216 (1) bulk density averaged 0.92 g/cm3 and was significantly and inversely related to elevation in the wet forest, but not in the rain forest; and (2) soil organic matter percent and soil organic matter content averaged 9.6 percent and 33.8‘kg/m2, respectively, and were significantly and positively related to elevation in the wet forest but not in the rain forest. Reciprocal averaging and polar ordinations of sample plots showed that: (1) the 75 individual plots partitioned into two major groupings which were valley plots, and slope and ridge plots combined. These groupings were based mainly on the relative abundance of Prestoea mont n , the palm; (2) composite plots derived from the grouping of complementary ridge, slope, and valley plots at the same elevation, sorted into four configurations. They were leeward (west) wet forest plots between 650 and 800 m in elevation, leeward (south) wet forest plots between 630 and 660 m, windward (northeast) rain forest plots between 650 and 750 m, and windward (northeast) rainforest plots between 800 and 960 m in elevation; and (3) the leeward (west) wet forest plots partitioned into groupings which included ridge, slope, and valley plots between 640 and 740 m in elevation, ridge, slope, and valley plots between 750 and 860 m, and ridge, slope, and valley plots greater than 900 m in elevation. Reciprocal averaging also showed configurations for 39 tree species according to elevation and topography, indicating species preferences in a complex, montane habitat. Five species were largely confined to valleys, eight were observed in valleys from 30-65 percent of the time, and the remainder were tallied on ridges or slopes more than 75 percent of the time. 217 The occurrence of 20 canopy species with regard to elevation showed four species with unimodal peaks at high elevation, one species with a unimodal peak to the windward, and seven species with bimodal peaks to the windward and leeward. Five species were found on both aspects, but were more common to the leeward. Three species were about evenly represented throughout the forest. Histograms of occurrence for these same species by topography showed that five were found mainly in valleys, nine had greater occurrence on ridges than other topographic positions, and six occurred about equally on both slopes and ridges. The regeneration of 20 canopy species showed that the palm, Prestoea montana, was the most successful species with an average of nearly 23,150 seedlings/ha. Seven other species had more than 1,000 seedlings/ha and four species had no regeneration whatsoever. Eight species had fewer than 770 seedlings/ha. Relative understory and seedling density ratios, seed size and wood density were used to rank the canopy species from most secondary to most primary in characteristics. The groupings recognized were: shade intolerant pioneers, two; shade intolerant gap or late secondary species, five; and shade tolerant species, 13. Of the latter, one species was capable of growing to maturity in shade, seven benefited from gaps, and five required them to mature after regeneration. Thus gaps in the canOpy are critical to the regeneration and growth of nearly all of the colorado forest species. LITERATURE CITED LITERATURE CITED Ashton, P. 8. 1964s. A quantitative phytosociological technique applied to tropical mixed rain forest. Malay Forester 27:304-317. Ashton, P. S. 1964b. Ecological studies in the mixed Dipterocarp forests of Brunei State. Oxford Forestry Memoirs No. 25. Ashton, P. S. 1976. Mixed dipterocarp forest and its variation with habitat in the Malayan lowlands: a re-evaluation at Pasoh. Malaysian Forester 39:56-72. Asprey, G. F. and R. G. Robbins. 1953. The vegetation of Jamaica. Ecological Monographs 23:359-412. Austin, M. P. and P. Grieg-Smith. 1968. The application of quantitative methods to vegetation survey II. Some methodological problems of data from rain forest. Journal of Ecology 56:827-844. Austin, M. P., P. S. Ashton and P. Creig-Smith. 1972. The application of quantitative methods to vegetation survey. III. A re-examination of rain forest data from Brunei. Journal of Ecology 60:305-324. Baker, F. S. 1950. Principles of silviculture. McGraw-Hill Book Company, Inc., New York. 414 p. Baker, R. G. 1972. Seed weight in relation to environmental conditions in California. Ecology 53:997-1010. Bannister, B. A. 1970. Ecological life cycle of Euterpg globosg Gaertn. p. 299-314. In R. T. Odum and R. F. Pigeon, eds. A tropical rain Bates, C. 2. 1929. Efectos del huracan del 13 de septiembre de 1928 en distintos arboles. Revista de Agriculture de Puerto Rico 23:113-117. Baynton, R. W. 1968. The ecology of an elfin forest in Puerto Rico, 2. The microclimate of "Pico del Oeste." Journal of the Arnold Arboretum 49:419-430. Beard, J. S. 1944. Climax vegetation in tropical America. Ecology 25(2):127-158. Beard, J. 8. 1949. Natural vegetation of the Windward and Leeward Islands. Oxford Forestry Memoirs 21:1-192. Black, C. E. 1967. Methods of soil analysis. American Society of 218 219 Agronomists Monograph No. 9. Madison, Wisconsin. Bouxin, G. 1976. Ordination and classification in the upland Rugege forest (Rwanda, Central Africa). Vegetatio 32:97-115. Briscoe, C. B. 1966. Weather in the Luquillo Mountains of Puerto Rico. Institute of Tropical Forestry Research Paper ITF-3. Institute of Tropical Forester, Rio Piedras, Puerto Rico. 250 p. Brown, S. and A. E. Lugo. 1982. The storage and production of organic matter in tropical forests and their role in the global carbon cycle. Biotropica 14(3):161-l87. Brown, 8., A. E. Lugo, S. Silander and L. H. Liegel. 1983. Research history and opportunities in the Luquillo Experimental forest. General Technical Report 80-44. Southern Forest Experiment Station, New Orleans, LA. 128 p. Budowski, G. 1965. Distribution of tropical American rain forest species in the light of successional processes. Turrialba 15:40-42. Byer, M. D. and P. L. Weaver. 1977. Early secondary succession in an elfin woodland in the Luquillo Mountains of Puerto Rico. Biotropical Canham, C.D. and P.L. Marks. 1985. The response of woody plants to disturbance: patterns of establishment and growth. In S.T.A. Pickett and P.S. White, eds. The ecology of natural disturbance and patch dynamics. Academic Press, Inc. New York, NY. p. 53-69. Cottam, C., F. C. Goff and R. H. Whittaker. 1973. Wisconsin comparative ordination. p. 193-221. In R. E. Whittaker, ed. Ordination and classification of communities. Dr. W. Junk b. v., Publishers. The Hague, Netherlands. Crow, T. R. and D. F. Grigal. 1980. A numerical analysis of arborescent communities in the rain forest of the Luquillo Mountains, Puerto Rico. Vegetation 40(3):l35-146. Crow, T. R. and P. L. Weaver. 1977. Tree growth in a moist tropical forest of Puerto Rico. USDA Forest Service Research Paper ITF-22. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 17 p. Dawkins, R. C. 1958. The management of natural high forest with special reference to Uganda. Commonwealth Forestry Institute Paper No. 34. Imperial Forestry Institute, Oxford. 155 p. Denslow, J. S. 1980. Cap partitioning among tropical rainforest trees. Biotropica 12 (supplement): 47-55. Ewel, J. J. and J. L. Whitmore. 1973. The ecological life zones of Puerto Rico and the U. 3. Virgin Islands. USDA Forest Service Research Paper ITF-l8. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 71 p. 220 Forest Products Laboratory. 1955. Wood handbook: basic information on wood as a material of construction with data for its use in design and specification. USDA Handbook No. 72. Superintendent of Documents, Washington, D.C. 528 p. Gauch, R. G. 1973. A quantitative evaluation of the Bray-Curtis ordination. Ecology 54(4):829-836. Gauch, B. G. 1977. Ordiflex, a flexible computer program for four ordination techniques: weighted averages, polar ordination, principal components analysis and reciprocal averaging, release B. Ecology and Systematics, Cornell University, Ithaca, N. Y. 185 p. Gauch, R. G. Jr. 1982. Multivariate analysis in community ecology. Cambridge University Press, Cambridge. p. 1-42. Gauch, E. G. Jr., R. R. Whittaker and T. R. Wentworth. 1977. A comparative study of reciprocal averaging and other ordination techniques. Journal of Ecology 65:157-174. Glenn-Lewin, D. C. 1977. Species diversity in North American temperature forests. Vegetatio 33:153-162. Gomez—Pompa, A. and C. Vazquez-Yanes. 1974. Studies of the secondary succession of tropical lowlands: life cycle of secondary species. p. 336-342. In Proceedings of the First International Congress of Ecology: structure, functioning and management of ecosystems. Centre for Agricultural Publishing and Documentation. Wageningen, Netherlands. Greig-Smith, P., M. F. Austin, and T. C. Whitmore. 1967. The application of quantitative methods vegetation survey. I. Association analysis and principal component ordination of rain forest. Journal of Ecology 55:483-503. Grubb, P. J. 1971. Interpretation of the ’massenerhebung effect’ on tropical mountains. Nature 229(5279):44-45. Grubb, P. J. 1977. The maintenance of species-richness in plant communities: the importance of the regeneration niche. Biological Review 52:107-145. Hall, J. B. 1977. Forest-types in Nigeria: an analysis of pre-exploitation forest enumeration data. Journal of Ecology 65:187-199. Hall, J. B. and M. D. Swaine. 1976. Classification and ecology of closed-canopy forest in Ghana. Journal of Ecology 64:913-951. Harper, J. L., P. R. Lovell and A. G. Moore. 1970. The shapes and sizes of seeds. Annual Review of Ecology and Systematics 1:327-256. Hartshorn, G. S. 1978. Tree falls and tropical forest dynamics. p. 221 617-638. In P. B. Tomlinson and M. H. Zimmermann eds. Tropical trees as living systems. Cambridge University Press, England. Hartshorn, G. S. 1980. Neotropical forest dynamics. Biotropica 12 (supplement): 23-30. Jones, E. W. 1956. Ecological studies of the rain forest of southern Nigeria. IV. The plateau forest of the Okumu Forest Reserve. Part II. Journal of Ecology 44:83-117. Knight, D. E. 1975. A phytosociological analysis of species-rich tropical forest on Barro Colorado Island, Panama. Ecological Monographs 45:259-284. Lawton, R. O. 1984. Ecological constraints on wood density in a tropical montane rain forest. American Journal of Botany 71(2):26l-267. Little, E. L. Jr. and F. R. Wadsworth. 1964. Common trees of Puerto Rico and the Virgin Islands. Agricultural Handbook No. 249. USDA Forest Service, Washington, D. C. 548 p. + illustrations. Little, E. L. Jr., R. O. Woodbury and F. E. Wadsworth. 1974. Trees of Puerto Rico and the Virgin Islands Second Volume. Agricultural Handbook No. 449, USDA Forest Service, Washington, D. C. 124 p. + illustrations. Lugo, A. E. and S. Brown. 1981. Ecological monitoring in the Luquillo Forest reserve. Ambio 10(2—3):102-107. Marrero, J. 1948. Repoblacion forestal en el Bosque Nacional Caribe de Puerto Rico: experiencias en el pasado como guia para el futuro. Caribbean Forester 9:148-213. Newberry, D. McC. and J. Proctor. 1984. Ecological studies in four contrasting lowland rain forests in Gunung Mulu National Park, Sarawak. IV. Association between tree distribution and soil factors. Journal of Ecology 72:475-493. Onyekwelu, S. S. C. and J. C. Okafor. 1979. Ordination of a savanna woodland using woody and herbaceous species. Vegetatio 40(2):95-100. Orians, G. E. 1982. The influence of tree-falls in tropical forests in tree species richness. Tropical Ecology 23(2):255-279. Perez Viera, I. E. 1986. Tree regeneration in two tropical rain forests. M.S. thesis. Biology Department, University of Puerto Rico, Rio Piedras, PR. 97 p. Pickett, S. T. A. 1980. Non-equilibrium coexistence of plants. Bulletin Torrey Botanical Club 107:238-248. Pickett, S. T. A. 1983. Differential adaptation of tropical tree species to canopy gaps and its role in community dynamics. Tropical Ecology 24(1):68-84. 222 Pickett. S. T. A. and P. S. White. 1985. The ecology of natural disturbance and patch dynamics. Academic Press. Inc. New York. NY. 472 p. Poore. M. E. D. 1968. Studies in Malasian rain forest. I. The forest on triassic sediments in Jengka Forest Reserve. Journal of Ecology Proctor. J.. J. M. Anderson. P. Chai. and H. W Vallack. 1983. Ecological studies in four contrasting lowland rain forests in Gunung Mulu National Park. Sarawak. I. Forest environment. structure and floristics. Journal of Ecology 71:239-260. Ramsay. D. McC. and P. N. de Leeuw. 1965. An analysis of Nigerian savanna. IV. Ordination of vegetation developed on different parent materials. Journal of Ecology 53:661-77. Rumney. O. R. 1968. Climatology and the world's climates. The Macmillan Company. Collier-Macmillan Ltd.. London. England. 656 p. Salisbury. E. J. 1942. The reproductive capacity of plants. Studies in quantitative biology. Bell and Sons. London. 244 p. Schmidt. R. and P. L. Weaver. 1981. Tree diameter increment in the subtropical moist life zone of Puerto Rico. Turrialba 31(3):261-263. Smith. R. F. 1970. The vegetation structure of a Puerto Rico rain forest before and after short-term gamma irradiation. p. D 103-140. In H. T. Odum and R. F. Pigeon. eds. A tropical rain forest. USAEC. TID-24270. Vazquez-Yanes. C. 1980. Notes sobre la autoecologia de los arboles pioneros de rapido crecimiento de la selva tropical lluviosa. Tropical Ecology 21:103-112. Wadsworth. F. H. 1948. The climate of the Luquillo Mountains and its significance to the people of Puerto Rico. Caribbean Forester 9:321-335. Wadsworth. F. H. 1951. Forest management in the Luquillo Mountains. I. The setting. Caribbean Forester 12(3):93-114. Wadsworth. F. R. and J. A. Bonnet. 1951. Soil as a factor in the occurrence of two types of montane forest in Puerto Rico. Caribbean Forester 12:67-70. Wangaard. F. F. 1950. The mechanical properties of wood. John Wiley and Sons. Inc. New York. NY. 377 p. Weaver. P. L. 1972. Cloud moisture interception in the Luquillo Mountains of Puerto Rico. Caribbean Journal of Science 12:129 -144. Weaver. P. L. 1983. Tree growth and stand changes in the subtropical life 223 zones of the Luquillo Mountains of Puerto Rico. USDA Forest Service Research Paper SO-190. Southern Forest Experiment Station. New Orleans. LA 24 p. Webb. L. J.. J. G. Tracey. W. T. Williams and G. N. Lance. 1967a. Studies in the numerical analysis of complex rain-forest communities. I. A comparison of methods applicable to site/species data. Journal of Ecology 55:171-191. Webb. L. J.. J. G. Tracey. W. T. Williams and O. N. Lance. 1967b. Studies in the numerical analysis of complex rain forest communities II. The problem of species sampling. Journal of Ecology 55:525-538. Webb. L. J.. J. G. Tracey. W. T. Williams and G. N. Lance. 1970. Studies in the numerical analysis of complex rain forest communites V. A comparison of the properties of floristic and physiognomic structural data. Journal of Ecology 58:203-232. Welbourn. M. L.. E. L. Stone and J. P. Lassoie. 1981. Distribution of net litter inputs with respect to slope position and wind direction. Forest Science 27: 651-659. White. M. R. Jr. 1963. Variation of stand structure correlated with altitude. in the Luquillo Mountains. Caribbean Forester 24(1):46-52. Whitmore. T. C. 1975. Tropical rain forests of the far east. Clarendon Press. Oxford. England. 282 p. Whitmore. T. C. 1978. Gaps in the forest canopy. p. 639-655. In P. B. Tomlinson and M. H. Zimmerman. eds. Tropical trees as living systems. Cambridge University Press. Cambridge. Whitmore. T. C. 1984. Tropical rain forests of the Far East. second edition. Clarendon Press. Oxford. 352 p. Whittaker. R. H. 1956. Vegetation of the Great Smoky Mountains. Ecological MonOgraphs 28:1-80. Whittaker. R. H. 1965. Dominance and diversity in land plant communities. Science 147:250-260. - Whitten. A. J. 1982. A numerical analysis of tropical rain forest. using floristic and structural data. and its application to an analysis of gibbon ranging behavior. Journal of Ecology 70:249-271. Wong. W. I. and T. C. Whitmore. 1970. On the influence of soil properties on species distribution in a Malayan lowland dipterocarp rain forest. Malayan Forester 33:42-54. APPENDIX TABLE S 224 Nm.e can menu: mameommeauon cannon NN No.6 N .cmwux Nauuumzm maoaooooo ON Nn.N omd eenu=.mmmmwmnx.mwesao mg No.o N sens: + wsux A.aomwuuv.mmmmmmmmmmmm.mmmnmmm @— Ne.o em .4 aaumoseo asahxeumnuao Nd co.o m .uao>.a:HHMJQGuUmE asuumoo cu mn.c co .A mumuaoe munouooo mg No.o N .An>< cameosmwaw mousomwmweo em do.o H .mm mwuumo>amw mwueoemo mu co.c n menu: A.Am>v assesses ss>.=.su=m a 85 a .a econ a .533 :5 338m 3325828 m m—.c N~ efima< A.:ouu«umv mameeaaasvsa equated e oc.o m cope: easemfimsuno mezuaua¢ n No.o N can»: mamaooauOuuom mamnoocuonoa< N o~.c cc .mm muflouuuea oeeuonoa< d cameo ueoouom macaw .oz Heuoa newsman a mowooam .oz 5...... .mosum nouumewouo onu :N ooemoesnm mofiooem menu mo unqumHH Hmowuenmned< .N oases xqoeone< 225 MN.o no.N mn.o mH.o mc.o mo.o «0.9 co.o hm.o mc.o mc.c «N.m mm.o Hw.o NH.o do.o No.9 wo.N an.c ow.N Nm.~ mm.n mg mod NN meN we 00 ed CNN Ne «MN mag on .3m maumflhophfl gONUUm nfi .eomsuo usasouauaaau useucmscume Ne .suwm .< maddesucoam> moumuuoau ~¢ .uem>.mmmwmmmmm mousse o: no: menu: fig gunman—dud on .aumz «fimfieouewe.uqqflu mm H£m>.MMMMMMMMfifdddHN mm .a gum. dad on mean: + many .xmuoewu.flflflmmwm;mflmmmmm mm :H a g cm .xmuomNM.flwmmmwmm.mwmmmmm.mm 633.5 a g Nm H53 neg 1mg an .nomauu Mama 3 cm 5%: + .988 .32 was; on semaeum .o.e fiesta: .mv.mmwwmwwmmmw.mmmmmmmmmwm mu socmam + .oeoon .A.mmmmmmmm.mmmmmmmmmwm.NN zen-5 + was. 333% g cm Ham> a 3 N :H 301mm g «N :sfiam .u.n.q.mmmmmmm.mmmmmxu mu can»:.mmmmmmflflmmmm.mmmmmu.NN Ameseaueoov N canoe awesome< 226 co.o~ mN.e ~N.H ~o.o m~.c Hm.o ~N.0 No.0 Nm.H an.o No.9 No.6 eo.o NN.o N—.o em.c No.0 Nu.o No.0 MH.o mo.H mam own oqg Ha Ne Nd QOH mN fin ma o“ wN CH HH ~N~ ouumam mwaowomfieaoumw maaosmouoflz ouuoqm mooaoflflmznomhuno mHHonmouowz non .n A.3mv enmemuou eweoowx .Un A.3mv mcwmmun canoe“: .awoo maahnamnumn mfiaoowz .un A.Av eumwa>oea mecca“: .omaom “Homespun mamowaoz unwwua .0 Aumonv afieaamowwae seasoned: cameowewaoo mammumx .xmuomwx afimfieouefim muowaumz .>mno A.Un .dv mumuaoown mumxawcmx menu: mcovaeamm mHHoewmz .nnocx A.namAv mwmcumwmwaov muowuoeaa .nuaom A.vomnv coupon muoxH mafia: muo> eweH 3:: 3.3 a finale couuaum Acmnuav Hfimaaouewm onH nomwuu movwoamwouoowm wean .aaxmz .anm> moans: quH .comh assesses» abaamaom .muom.uuduuu.udduquflm co mm No No co mm mm mm on mm em mm Nn an on me no me we me «e Amoncfiucoov N edema xfiozoqm< 227 em.o OH.o No.0 mm.w mo.o em.o mm.o HH.o no.o do.o No.o ao.oH do.o NN.o do.m 06.0 mo.o «o.o Ho.c ma.o o~.~ wN ewe mN cm mm an mom.a ma Ode Nm AH mm 53.5 + mag 3.3.5 .8 «mousse “REESE. 3 .awoo «Human: dammemuuoa em enmesx A.3mv muowfiammaen mauumemmuuoa mm amen: emanau mannonma cognate A.onv_maeaeaaumume assemesa undone eeewmouuon moemoam .mmoa mammuooousma aswmmm deny: + menu mameoowuouuom mfiuoaooeom dean: mwmeooauouuomvmaooozm not .>mm +_NH:M moafiwmuuom mocmmem .oa ecmwuouumn daemonomem .maonoaz nacho .MV mdmueoa mooumoum Auden «Hannah monsoowaom menu: Hausa: mamoauo no: mamasnumme mououo we: mammooHuouuom mouooo no: A.cmwezv eumnomoa mouooo no: A.amv :oHNMoosoH mouooo .on A.smv memeaaemm when»: .on Auaomv sxuaeoe neon»: Nm Nm om mm mm mm on an em MN NN an ON me we no em .09 .30flm xmaamw mauuwfi no nooseauooov N manna seasoned 228 .ssms .wm.mm.mauusa “seam assesses: can manage "outaom \H oo.ooa mma.m maaaoa o~.o m eon»: Adena: + msuxv.aaduu&mummmmmm.wamoaux hm No.0 N .am eueoaue>ao xoua> om Aoosewueoov N canes newcome< APPENDIX FIGURE S 229 Appendix Figure A-2. Reciprocal averaging ordination: species ordination of 27 windward and 30 leeward plots and 30 species on ridge topography. 230 2......5 .ee.......u ‘ ‘ 4 3...... ......u q 32.3.3.3. 2...... .32.... 22...: C 2.2.... 2.3.8 23...... 23......» 3.... 3...... . 4 . u 53...}... 2.826 :0... ‘ 3.3.2.309 2.2.... 5......- .e..e.... C 3.8.2.3... ..........I 4 ‘ ........x< 1 2.2.2.... ‘ ’ ‘ .........a ‘ ‘ :0... 3...... 2.3.1.... 3.5.2.3.... .......o‘ 2:... .. :3. funny...”- £313.... 22...... 5.3.3.... 22...... 2322‘ 2...... 3.. 2.3.3! 8:2... .3... .. 9 [3.38.2.- e.......u....3 .. I... .23.; ‘ (3.33:8 33.2.3-3! .21.... I I 8:2... .3 .23.... - c .833... o...“ d 9...... .3... d 2...: d :2... 28.... 22...... 2:23... 2.3.... ........4 32.232... 223‘ ‘ 33.5.3... 5.2.4 23...... 3...... ‘ NI< ounwwm xfioaomn< 231 Appendix Figure A-3. Reciprocal averaging ordination: species ordination of 27 windward and 30 leeward plots and 26 species on slope topography. 232 2 £23.22... 2.3.5 3...... 2.. 3.3.23 3......m .2... I 0 3.335122 a. 3.4.2.3....» 3.3.33-2: .933... u a 3......w I... 63...... .. ( ecu... ...o... I ......u .2... I 23...... 23...... 22...... 3...... 3.30 I I 3.2.2.3.. c.3223... I .239 E...oo3...u I I .2... 2...! x 2.2.... 22...... 2.3.! I V3332... 2.30 as... I I 32.82:... I I :32. 2...... I 333...... 2... 2...... 22...... 3.3.. 3.... 32...! 253.83.. I 33...... 3...»...L 22.3.30 I I .a . o .. abney»... 33.: .........I .33.... I 2.2.... I 2.2.33. \ 2......- 32:23.. 2.2.35 .eeo..o>..u 2.2.229... 2.3.22! _ 2......" 83.2.... 2.35.... mud mane“. awesome... 233 Appendix Figure A-h. Reciprocal averaging ordination: species ordination of 27 windward and 30 leeward plot. and an specie. on valley topography. 234 .......o ..........u 3.... 33.2.3 .83.... v... .832...» .23.... so... ... u ........u-2l .. B... .2262! 3.333-...- .....3.... u o ........u 3.. .223... .. C 96.6. .I. 080.500.53.- .3...“ :1... ......”u 0 2.2.3.3... .223... O 2.2.... 2.3.... O 22...... 2 .3... I 322.233.... 0 2......28 .......2... 3.2.2.3.. .2... 2.2.2.3.: I .229. . . I 2 3. h 2.2.2.2.... 2.3.92... 2.2.00 222...... 8.3.. /...2...2..a 333...... 2... 3.2.33.0. 5.3.0.3590. I I 32.2.... .........a. 22.33... 2...... I 223.... 3. / 23...! 2.30.5 1.39... .23.. I 5.3.22... o 3...... 3.3.9 32...... 3.32... I 22... 2.2.30 2.2.23. I 22...... ..< «.3... xfivam..< PART III FOREST STRUCTURE AND PRODUCTIVITY 235 m 236 I NTRODUCTI ON The diversity of plant life in the tropics, where plant communities range from the lowlands through montane areas to timberline and paramo, in .humid and dry environments, exceeds that of other latitudes. 0f the world’s 116 life zones according to Holdridge’s (1967) classification of plant formations, 66 are located within the tropics. Nearly half of these support forests. Moreover, tropical forests constitute about half of the world’s forests on an areal basis (UNESCO 1978) and make significant contributions to human welfare, yet remain poorly studied. For example, information on total annual net primary productivity (total NPP), or the rate of storage of organic matter in excess of respiratory utilization during the measurement period (Odum 1971), was available from only 20 tropical forests about a decade ago (Murphy 1975). The purpose of this study was to determine NPP in the colorado forest by summing the rates of litterfall, herbivory, and biomass increment (Medina and Klinge 1981; weaver _t__l, 1986). Over the gradient from low elevations to the summits of the Luquillo Mountains, temperature decreases, and cloud cover and rainfall increase, resulting in a decline in forest structure and complexity (Baynton 1968; weaver g;__l, 1973; Wadsworth 1951). The existence of information on forest structure and dynamics for each of the forest types in the Luquillo Mountains including the tabonuco forest at 450 m in elevation, through the palm brake (Frangi and Lugo 1985) and the colorado forest (this study) at 750 m, to the dwarf forest at 1050 m in elevation (Weaver g; .1. 1986), provided a rare 237 opportunity to compare changes along a 600 m elevational gradient in the subtropical latitudes of the Caribbean. 238 METHODS The re-measurement of the permanent plots and their data analyses were described in the methods of Part 1. Soil data collection and analyses were outlined in the methods of Part 11. Leaf area index (LAI) measurements were made on two permanent plots located at mid-elevations in the colorado forest, slope plot B, and mixed topography plot E, once during April which is drier than most months (Fig. 2, General Introduction), and again in June. Of primary interest was the LAI of the woody vegetation since ground vegetation may have been disturbed by previous tree measurements. A modified plumb-bob technique was used (Benedict 1976). Fifteen sampling stations were spaced about 10 m apart within each stand. At each station, a ladder was positioned against the nearest tree and an extension pole was used to lower four plumb-bob lines 900 apart through the canopy to the ground. All leaf contacts were recorded. Because ground area is reduced to a point, each contact per line represents 1 mzof leaf surface per 1 mzof ground surface. An adjacent area was also selected and the measurements were repeated at 10 stations. In this instance, canopy vegetation was partitioned between trees and epiphytes, and the ground vegetation among herbs, grasses and bryOphytes. Litterfall was collected periodically for two years starting in January 1981. Twenty 0.25 m2 baskets were placed about 10 m apart in the center of both permanent plots. During the first year they were emptied twice per month and during the second year on a monthly basis. 239 Litterfall was partitioned into leaves, wood, fruits and flowers, and miscellaneous categories, and dried to a constant weight and recorded. Loose litter, or unincorporated detritus composed of leaves, other plant materials and woody fragments, was collected monthly for one year starting in January 1982. Twenty 0.25 m2 collections were made from areas adjacent to both permanent plots. All samples were partitioned into woody and miscellaneous categories, dried to a constant weight, and recorded. Herbivory was evaluated in two ways. Standing herbivory, or the amount of herbivory evident at a particular moment, was determined once by sampling the eaten portions and total areas of 1439 leaves on 153 branches of 35 trees. Twelve of the most common species were represented. In order to convert standing herbivory estimates by species to a value representative of the entire colorado forest, a weight factor was used. First, the relative density and relative basal area of each species sampled were determined. These percentages of density and basal area were based on the species composition of all seven of the natural forest permanent plots. Next, these percentages for density and basal area were summed, and then divided by two to yield a value that was weighted by both the number of individuals (density) and size of the trees (basal area). The standing herbivory for the forest species not sampled was assumed equal to the weighted mean of the 12 species that had been sampled. These weighted species values were converted to ecosystem values by multiplying by the LAI of the arborescent vegetation. Herbivory rate, or the amount of herbivory during a given period, was determined by sampling 418 leaves on 111 branches of 41 trees. Nine of the most common species were represented. In this instance, fully 240 developed young leaves without evidence of herbivory were selected. Each was numbered and tagged with plastic tape that was stapled across the midrib on the underside of the leaf. After 90 days, the leaves were removed from the trees and the amount of herbivory was determined. Herbivory rate for the ecosystem was estimated using the same weighting procedure mentioned above. Leaf weight was determined by removing 30 leaves for each of 18 common species. Leaves were then placed in numbered bags and dried to a constant weight. The same weighting procedure outlined above was used to convert individual species values to ecosystem values. For standing herbivory, herbivory rate and leaf weight, leaves were selected from trees of different crown classes, and at different heights within the canopy. Leaf area in each instance was determined with a Li-Cor Model 3100 Area Meter. In all cases, weights were to the nearest 0.01 g and areas to the nearest 0.01 cmz. Seventeen dicotyledenous trees ranging in size from 6.5 to 36.7 cm in diameter and representing seven different species were selected for biomass determinations. For sampling purposes, biomass was partitioned into trunks, branches >2.5 cm, branches 52.5 in diameter, and leaves. we: weights were determined in the field. For each tree, four sample discs of the trunk taken at different heights and six sample discs of branches >2.5 cm were oven-dried along with four samples of branches 52.5 cm and leaves. Leaves were dried for three days and the remaining materials for six weeks at 70°C, at which time no decrease in dry weight was observed. Sample dry/wet weight proportions were then used to determine the dry weight of all classes of material. Regressions in which dry weight was expressed as an allometric function of tree diameter squared 241 times tree height were developed for leaves, branches, trunks, aboveground woody material and total biomass (leaves and woody fractions combined). The regressions were used to determine the standing biomass of the leaf and aboveground woody compartments for all measured trees. Root biomass was not sampled. Rather, it was estimated as 30 percent of the aboveground woody biomass. This value has been used frequently for humid forests when root data are lacking (Murphy 1975). The value slightly exceeds the estimate of 22 percent determined for root biomass on a palm plot (Frangi and Lugo 1984) adjacent to one of the colorado forest plots used in this study. The percent root biomass estimate on the palm plot, however, may be low because of poor drainage and the dominance of palms which characterize these sites. In all instances, the above regression equations were compared with data for the tabonuco forest (Ovington and Olson 1970) using standard statistical techniques (Freese 1967). Significance was defined as P 50.05. Cecropia peltata and Didymoganax morototoni, both very light woods, were removed from the data set of the tabonuco forest so as not to distort the regression estimates. Net primary productivity (NPP) was estimated as the sum of litterfall, herbivory, and biomass increment. Herbivory was determined as the annual herbivory rate times the leaf biomass. Biomass increment during the 35-year period was partitioned into the accrual of the leaf, the aboveground woody, and the root compartments. Net biomass increment for leaf and aboveground woody compartments was determined as the 1981 standing biomass minus the 1946 standing biomass divided by 35 years. The increase in root biomass was estimated as 0.3 times the aboveground net woody increment. 242 Palms were treated separately. Total palm biomass (leaves, stems, and roots) was estimated by the equation In Y - 3.85 + 2.91 In X where Y is the dry weight of the palm in grams and X is its height in meters (Bannister 1970). Palm trunk biomass was estimated independently as basal area times height times the specific gravity. These values were added, respectively, to total aboveground biomass and aboveground woody biomass estimates. The weight of palm leaf stalks, inflorescences, and roots was determined as the residual of total palm biomass minus palm trunk biomass, both derived by the aforementioned relations. 243 RESULTS Leaf Characteristics in thgvforg§£_ Leaf area index There was no significant difference in LAI for trees measured on the same plot in different months, nor was there a significant difference in tree LAI between plots measured in the same month. Therefore, all data were combined into a single estimate of 3.01 mzlm2 (Table 16). Epiphytes averaged 0.30 mzlmz, herbs and grasses 1.18 m2/m2, and ground bryophytes 0.47 m2/m2. The total for all vegetation was 4.96 mz/mz. Leaf weights Mean oven-dry leaf weights for dicotyledons ranged from 0.06 g for Qi£;§_to 1.61 g for giggig. Palm leaflets (Prestoea) averaged 1.75 g (Table 17). Mean leaf areas for dicotyledons ranged from 8.10 cm2 for Qi£;§_to 135.89 cm2 for Cordi , with palm leaflets at 236.90 cmz. Specific leaf areas for the dicots ranged from 36.01 cm2/g for giggig_to 163.73 cmzlg for Cordia, with palm leaflets at 135.37 cm2/g. The mean oven-dry leaf weight values for all dicotyledons combined was 0.67 g; the mean area was 56.38 cm2; and the mean specific leaf area was 90.96 cmzlg. The weighted mean specific leaf area for all dicotyledons, a value more representative of the forest, was 67.79 cmzlg. All dicotyledonous leaves were within Raunkiar’s mesophyll size class (Richards 1966). The palm leaves were classified as megaphylls. 244 Table 16. Leaf area indices by compartment in the colorado forest Compartment Number LAI, mzlm2 of lines Mean 1 SE Canopy Trees1 230 3.01 (0.12) Epiphytesz 40 0.30 (0.14) Ground Herbs + grasses2 40 1.18 (0.16) Bryophytes2 40 0.47 (0.08) Total -- 4.96 (----) 1 This figure is derived from 60 measurements on each of 2 plots made in April and June plus an additional 40 measurements on a third plot in June. Neither plots nor dates showed significant differences, therefore, the data were combined. Three epiphytes and herbs + grasses and bryophytes were determined only on the third plot which had not been disturbed by previous dbh measurements. 245 .m~.~m a“ muoammoa same no“: mono mama ouwuuonm counwfiua 0:3 I. A .m<.nm can .~<.oo .m~.c .m~o>«uounmou .oua monan> \ on» .vovnaocfi aw sewn onu cons .AmoOumoumv muofiwmoa Egan ovaauxu mouuonm Han you wanna vouswwoaca 039 \m. .Eoumhmouo umouou ovmuoaoo any yam mono mama uwmwoonm some USu wawafiauOuov How mo=Hm> mounwfioz \m .oau an emuasac Aumonom :H vmmnamm noun Amman Hmuou mo uaouumn + umouom :H voamamm maoum Hmuou mo unmouoav so comma muswwoz \I .umuuow o:u manna: mHu>oH uaououuav scum coxmu mowooam you mo>mo~ on no woman moumaaumu \m \ma~.~o o.m~ ‘[\mc¢.om . - newsm.en . . wumwe.o cmm=\amnoa mm.e~ mm.o so.am Aoo.nv Na.sw Amo.oy. mc.u seamen massages s~.o~m as.~ am.mm~ Aow.mv oa.om~ a~a.ov m~.a «causes amonmmsm om.oa ms.o N¢.am ass.av oa.o~ Amo.ov mm.o mnmflscnmaw mmnouo m~.~ No.o am.sad Amm.ov -.mm Aoo.ov -.o coasxousma monouo so.mea mm.m o~.ms Ae~.av oa.mm Amo.ov om.o masocmmscsonmw assessors“: s¢.moa 05.2 oo.am Aam.ov as.- Amo.ov os.o maeaoaassaOmaugu maaoeaonuaz wa.N ~o.o Na.aoa Awe.sv on.am Aco.ov am.o mnecmnnmu assess: as.em No.o mo.a¢ Aa~.ev na.~ma Aeo.ov os._ msmuamaam asaocwmz om.n so.c so.sma ass.av N~.os Ado.ov om.o mcmommnonsm==asoaem= s~.oa sw.o No.mm Amm.ov em.~a Amo.ov a~.c maaaowausamm msencmacmmx n~.mm m~.c oo.mm_ Am~.ov oa.m Aoo.ov oo.o masseusnsa manna as.os es.c sm.aoa Amm.~v ~a.mm Amo.ov ~n.c mamammaaasa mamaocnama ~¢.o~a ma.~ s~.om Asm.ov am.o~ “do.ov hm.o macawasmomn «Haasau aa.a~a an.“ m~.sm Ao~.Nv om.om Amo.ov ea.o mannamaaumoachUonu. an.ao~ No.o -.mo_ Rah.mv mm.mma Aso.ov mw.o mamamssasnon «assoc o~.sa as.o so.om Aom.mV no.5m Aoa.ov so.a mamswsnx masses mn.~s~ Ne.~ an.~a Asa.av mm.o~ aao.ov a~.o aamoasamaam anacowoosamo mm.ma ~_.c s~.aaa Ao~.av em.m~ Ado.ov H~.o Hanunosmnms manganese Amx 50v Ioaam mono wauaov AmEUV Amv «Moa ofimwoonm \muouomN mmum mama mou< unwaoa has vuuswwoz unwaoz uawaoonm mm H can: kumwmo3== mmfiuoam \~ .umou0m ovmuofloo on» unonwaounu v=30m mofiooam :oeaoo now mowumauouonumnu moon .5“ manna 246 Litterfall ggd_§oose Litter The litterfall data from the two plots were not significantly different on a monthly basis, and were therefore combined into a single monthly estimate of litterfall. The same was true of the loose litter data. The mean annual rate of litterfall was 680 g/m2 (Fig. 39). Leaves constituted 74 percent of the total, wood fall about 18 percent, and flowers, fruits and miscellaneous fragments accounted for 8 percent. Leaf fall exhibited a peak between March and June (Fig. 40), but monthly rates did not exceed 75 g/m2. Flower and fruit fall showed a peak in July (Fig. 41) with generally higher values from April through August. Whody litterfall showed considerable monthly variation (Fig. 40) and miscellaneous litterfall did not demonstrate any consistent trends (Fig. 41) . The mean standing crop of loose litter was about 875 g/m2 (Fig. 42). The first three months of the year showed a notable decline in loose litter, while the rest of the year it ranged between 800 to 1000 g/mz. A peak in loose litter high is evident in June and July just after peak leaf fall. Because the loose litter values were low during the first three months of the first year of measurement, loose litter was sampled the following year during January, February and March. Again, these values were consistently lower, and compared favorably with the results of the first year. 247 Figure 39. Total litterfall for the colorado forest by month. from January 1. 1981 through December 31. 1982. n}- w: :50: llllnlull~o2 lllllll _.. llllllll SEIIIII O 2 O 2 4 S u.- q. 0 z O m 4 3 5 S < 2 5 o _ q _ _ _ . a _ _ _ _ _ a _ _ _ _ _ . o i r l r r M N.CII ILON T I. I m w ML cnOrI M” I.O¢ r m w m .. 901% low _I J H e w W H eol W e W loo . - .e - 0:: mac. ozx. ~E\o wwswwm smug pJOpUOlS pun suosw ssnwoga wbgan- Ma 249 Figure 40. Leaf litterfall and woody litterfall for the colorado forest by month. from January 1. 1981 through December 31. 1982. 250 oe seamen 3:592 Illnllullu~o2|llulll llIlulllISSIIIIII ozom<352<2usozomweq es sauna-ea lane ago was noses: e as caused-sou as: cuss-u used can .eonwueaw saga an \u .9)» ha escapee Aueeuou ea wean-Is sens Hosea usuou we assess: + ueeuou an usual-e sleu- ueuou no unsouemv so wee-a enemas: .I 2 N6 I... N.“ N18“: 24 was.” I... .l I. as! I. 8.2 1- I... 9: -1. «as 2.: .5 an :38 .6 3... 3. Sent.“ N2 v.3 SN 2 N «and anal-H QN _N.N fin SNHNN 85“.}: W NN \u c N «mad as a... 86 N5 8;.qu 86 at: 2 a N mafia amused 9N a: n4 No.o.+.N.o :6 #6 SN «N c enamel-an as». ..N 8... N.N 35.3... 8... No.1 NNN ON .. fig 3 a... :6 1: :NHNNN 3.." #43 an 3 N a flag 9N «N... o; .N.o.+.N.N SN #5.” NN o N g .N. s. 3 N.N 24 in Saws... “N... “NAN SN 2 .N a a Na :6 3.. NNéflN 3; H92 9: 3 N a a N... 9.: s... 26H“... 2.. HQ: 2 .N N a 34mm N5 :6 RN Sofia 34 Hi? 2: 3 n a 52.8 .12 Na; n... NioNN SA H12 2 2 .N a g ueeuou an season- »A A Jwv ANJWV buosmauoz \m.uouosm buo>mauo= m+m mm+m seamen meson-nu mocha useuuom usuwefi umuouom sauna Hausa Illnmdsdluafllll [34' .3825 .ueoHOu ensuedou use aw eewuems sens Nu you sum-Ia huo>uaues uswunium N n— sun-H 257 .eeoeeu we enema-Cu Ines ecu one goes»: a as caused-soc as: edge-u used was .eoawuamu mass as \m .ueuesoueu use one: ass-ea munch \u .emee ooxseuee ensue no .eesae» easy one essay .euese sauna scanned assuage: sea-n \fl .3 I. N... 3... #6 N3 No.2 .1. I. I as... I. .2 1. Ne... e.NNN a. NNN N. :38 .3 z... N.N 36... N.N 9N “Na: 3 NN N an; e NN.N a... 36MB... ENNHNNNN he. 4.. N dida- nun-flu N... 3... N... No.e.+.8.e N.N “12 N 2 N J..-15.3 3&8 N6 NN.N Na saws... ..N “NAN 2 3 N 42% fig N... NE. 3. Bang... N.N MN...” N S N 4333 3385.3 N... :4 N... Saws... N... .35 S 3 N Seams-Jaw. m.and a :6 a... saws... .3 #3. m3 NN N a 836 N.N NNé N.N 2.2.2; N.N flNN N... NN N del-mag... :38 N.N 8N N.N NNéflza 3... No.3 N. N N Nazi's... lam-ads: unseen no ceases. an Ammmw Ammmw hhosuauea nous-u sausages: mm sou-m neuow ass-ea cease-um eemwm N83... 33.: N88... 3.1 N3.— JmuoNJqs-m 8.1.... .333 .eueu co nose unseen coeueucu as» an seasons eeuu seas you ens-Ia sues uuosuaue: Na— swash 258 (Figs. 43 to 47). The r2 values were all 29.89 with the exception of that for leaves which showed a value of 0.60. The comparison of least squares linear regressions for leaf. branch. and total aboveground woody biomass. at the 95 percent level. between colorado and tabonuco forests. yielded significant differences. The regressions for trunk and total aboveground biomass. however. were not significantly different. Biomass estimates for the leaves. aboveground woody. and root compartments are shown in Table 20. Palm estimates were derived with other equations (see Methods). .Nst_Primarx_frndustinn_inzzl The total annual NPP for the colorado forest between 1946 and 1981 using the definition that NPP is equal to the sum of litterfall. forest herbivory rate. and biomass increment (including leaf. aboveground woody. and estimated root increment) was 7.83 t/ha/yr (Table 20). Litterfall accounted for nearly 87 percent of the total. biomass increment more than 10 percent. and the herbivory rate nearly three percent (Table 20). A simple model depicting the storage of organic matter aboveground and in the soil to a depth of 50 cm is shown in Figure 48. The four measured compartments. contained 55.584.6 g/mg. or 555.8 t/ha. Of this. about 70 percent was stored in the soil or on the soil surface. and the remainder was found in leaves and in aboveground wood. The litter 4 turnover rate was once per 1.28 years. or 0.78 times per year. 259 Figure 43. Leaf weight expressed as an allometric function of D28. Tabonuoo data were obtained from Ovington and Olson (1970) and the colorado data from this study. 260 10' l ’f / a / a ll / ‘ - IO . ‘ 4’4/ :: ‘L l/Lfirfl/ 2' 05 -:-: + .. ‘// r it .. / , 5 ‘4 z/ : , I" 5 u——— ‘ /( / ‘KV /KJ q d/ ' LEGEND / / COLORADO: O——-—-——-C / . turnsszsuooo Inx ‘ .2-o.sos TAIONUCOZA— — —A n. Y-- s.oas+o.szsmx .2:- 0.717 0.1.0: .0. L1 L Lim‘ .0. tozu,.N3N1 Figure 43 261 Figure 44. Branch weight expressed as an allometric function of D23. Tabonuco data were obtained from Ovington and Olson (1970) and the colorado data from this study. Oven Dry Weight 262 .0! M /a r . / .1 w r“ ’ . / a e / A” ,1 .fi: a z . ./ /' ‘b ‘ C / I! ,0 / w" / / J/ / / LEGEND If COLORADO: O—-—-—‘ ‘ g, lnYI-8.381+|.24Oln x .Z-o.s93 /f raaouuco: a——_a a a , Inv--s.4usu.ozunx . ‘ .2-o.77s . 4P/' - L,1_J_1144, 1 L L 1 L 10' no' no‘ (Oathcmzn) Figure 44 263 Figure 45. Trunk weight expressed as an allometric function of Dan. Tabonuco data were obtained from Ovington and Olson (1970) and the colorado data from this study. 16‘ 264 LEGEND COLORADOZ O—-—O lab-3193+ 0.916111! '3 - 0.951 TAIONUCOI ‘— —- A II Y'-3.50’ 00.96. In M r2 O 0.972 lllLLlL L 1 L Figure 45 to‘ (0210. cmzn) 265 Figure 46. Total aboveground woody tree weight (branches. trunks) expressed as an allometric function of D23. Tabonuco data were obtained from Ovington and Olson (1970) and the colorado data from this study. Oven On 266 10’ 7 "L /,/ f" a // /.. e no" 1 nc LEGEND coLouoo: o———-—e InYt- 3.536 +I.OIO In x .2 -0.sss usouuco: ‘— .— —a In 7- - 3025409701.. 1: r2 - 0.970 . 4 l l L LLL L 1 l L L I0' 10' 10‘ 10’ (Dan, enznl Figure 46 267 Figure 47. Total aboveground tree weight (trunk. branches. and leaves) expressed as an allometric function of D23. Tabonuco data were obtained from Ovington and Olson (1970) and the colorado data from this study. 268 .9! r/ ’ L /. "' IO ’ O I N. I O s . ' i a 9 . / a a C : a o 10 / 9’ a f ‘ > 4 LEGEND COLORADO: .————-C u. v-- 3. 47s +1.00s lax .Zooaas YADORUCOifi— -— A It” .- 3.153 +o.ssslnx .2 - 0.969 -_4 :1:1.1L11L .L L I 1,L IO' IO' 10‘ 10 Figure 47 269 .mummm mM\Am:0Numnvu an vw>wumv noon .mmmaowp xcnuu I mmmEoNn Ease Hmuouv Naoum vo>fluon .mucmauuenaoo nusuo ouafi oumumnom ou manammoaau huauamnv < .muoou cam .muucuumuuon0H .mxamum mood Baum m .nuaouw uoou moaowh mums» mm up vovN>Nv m0000u0LMNn .HwaH was mama 0N Amaco mucouvv mecca venouw0>onm onu mo non mm counaaumu e .u>\m:\u H~.o I we x mm.m n Amuse huo>Nnuosv x Afiwoa 0N mmmaown mead weavcmumv n mums» nm\AocmH mmmaoannfimmfi mmmEOva “coaumasoamo mmz N .mmz mo cowueanoamo mnu aw mom: mosam> uaomcommvcH H oo.oo~ mw.n Hmuoa No.H wo.o wn.m mw.o menopause Baum um.~ m~.o «c.5e mu.~e «Amuoofivv muoom nu.» mm.o e~.eo~ m~.ee~ Hmuounsm ...n u--- NN.N «N.N NuaNsav n--- u--- NN.NNN NN.osN AmNoUNev N2503 vasoumo>oa< No.N NN.o NN.o u--- Nsuo>Nnuum mw.ow ow.o ow.o null Hamwuouuaa cu.o No.0 mm.m ms.e Nmo>m0A llllfl£\ullll mmz Hmuoa u>\mn\u ~wo~ cqm~ Mouumm m0 N mmz new» fl .umouom ovmuoaou use now Ammzv muN>Nuonvoua hamfiaua as: no coaumfinoamo voHNcuoa .oN sassy 270 Figure 48. Mean annual organic matter (O.M.) budget for two plots of 0.40 ha in size in the colorado forest. 271 :32, 3.25.5 .e N»\~E\.8.o o T ~E\.2.o o O venom; we shaman 272 DISCUSSION Forest structure and dynamics, including NPP, are related. The assessment of biomass increment in tropical regions is usually based on two successive estimates of total forest biomass made at the beginning and the end of the study period. Likewise, herbivory rates are related to leaf characteristics, and litterfall, the major component of NPP, reflects the entire aboveground structure of the forest. For these reasons, trends in forest structure are considered along with those of forest dynamics over the 600 m elevational gradient in the Luquillo Mountains. .n---.- The number of trees per unit of land area and the weight of leaves per unit of leaf surface increase (the specific leaf area decreases) with elevation in the Luquillo Mountains (Table 21). The first trend is due to the greater number of smaller trees found at higher elevation. The second trend is because of a change in leaf structure with elevation. In general, montane leaves are smaller, have greater outer epidermal wall and lamina thickness, and have a lower non-palisade to palisade ratio (Tanner and Kapos 1982). In contrast, canopy height, typical diameter range, forest volume and biomass, leaf area index and species richness, all decline with elevation. The fact that there is a greater reduction in forest canopy height than in aboveground biomass in montane forests was attributed to .vovaaoxu Eden .veamlmm newsman you use: mounwuoaeau aoNN INNN NSN so? :23 N33... one»: :8 and ace can can usxeouuoma sons .3 .8 «on cc: .389 3.3 158. NS 3 .. - . NNNN NENSV at. as: 398% NN.N-N TN uNN e7.» NN£N5 use: .3. No.8 NNN N.N . 012 CNN 3:: 3852 N8.— m N8 :2 as: No2 3:3 333.. .38.. _..—853...: N2: oNN oSN u.eoNN :5st 833 334.32 amass S as: .8 gel: 3th 3:. N33 .24 .84 .84 «Ric 93 «N8... .3. N325 «TN .8-.. a: «2.8 3 233.. 22.8 NSNJN ca: 0838 .82 N63: 3:839 Auuemav Aoemuodoov Anaemv Aooacoaeav chanson unseen neon accuses were; umouou as: essence nosed ueeuow as: amounouuoam Neuauoauum .mcwsucaoz onawnvsg ecu canoe: mumouou soon you cuauuauum venue mo commune-00 .NN snows 274 couum>oao I sea as modules n me some .HH uuem .un shaman: In an nor! as Gag no» .mmaN one; use «unsung Iu nN aou .ohon Isvox moauomm no mo mama udmadm .aha— Into” chad uuuvocomw cha— lave: SS. ..N|. mm 2:3.» cNNN souNo ea. cosmeNsoN Acowueoueaaaoo Neeomuma .huanvooz “nom— u0>e03 Ncnm— guacamvmzv meavaue neuu>0m0 usualmwv cu Bu e N..~ha— uo>m03v nanosaNe aN no N.» .NNNN case use Nuauneu nonoasNe :N no «.4 .eNNN sauce Houseman :N 30 «_A .nca— u0>eozu .NeuacNucoov NN oNsae 275 the interaction of a decrease in canopy height with an increase in basal area (Edwards and Grubb 1977). Comparisons of soil organic matter are more difficult because sampling was done at different depths (Table 21). The mean for the t0p 25 cm in tabonuco forest was 420 t/ha while those in the palm brake to 100 cm depth and the colorado forest to 50 cm depth were 316 and 328 t/ha, respectively. The mean for three values at highest elevation for colorado forest near 960 m, essentially at the border of dwarf forest, was about 550 t/ha (Fig. 31, Part II). Earlier studies (Lyford 1969) showed from 10 to 202 organic matter in dwarf forest soils, estimates corroborated for the sites at highest elevations in this study (Fig. 30, Part II). Leaf biomass for small stems in the tabonuco forest is higher than that of small stems in the colorado forest (Fig. 43). For larger trees, the leaf biomass estimates converge. Leaf biomass is related to many factors, among them the leaf area index, tree growth patterns, and the average leaf dry weight per unit area of leaf surface (Table 21). The most common large trees in the tabonuco forest regression were Qgggygdgg excelsa with an average dry weight for leaves of 162 g/m2 (Odum 1970). The largest trees in colorado forest were gygillg gaggmiflggg, Microphglig ggrcinigefolig, and M. ghrysophylloides with average dry weights for leaves of 177, 206 and 172 g/m2, respectively, all of which exceed the dry weight of Qgggygdgg leaves (Table 17). Moreover, Qgggygdgg frequently has a low crown-to-bole ratio and has unbranched holes for several meters while both species of Micropholis are shade tolerant, tend to branch low, 276 and bear leaves throughout the crown. These leaf and branching characteristics account for the convergence of the regression lines. Branch biomass in the colorado forest is greater than branch biomass in the tabonuco forest, in particular, for larger trees (Fig. 44). Earlier accounts of this forest (Wadsworth 1951) described the trees as dense and ’branchy.’ In comparison, numerous stems in the tabonuco forest, particularly in the intermediate crown class, are tall and slender with comparatively small crowns concentrated at the tops of the trees. Trunk weight is greater in tabonuco forest than in colorado forest for larger trees although the converse is true for very small stems (Fig. 45). The average specific gravity of the species used in both regressions was 0.6 glcm3 with a range from 0.45 to 0.82 in the tabonuco forest and from 0.53 to 0.70 in the colorado forest. However, despite the overall similarities, Egggygggg, a tall, straight tabonuco forest dominant, was the only tree greater than 30 cm in diameter in the tabonuco regression. The proportionately greater trunk volumes of these large trees probably account for the trends. Total aboveground woody biomass (Fig. 46) and the total aboveground biomass (Fig. 47) regressions for both forests are very similar. The reason for their similarity is due to the compensating effects of greater branch biomass in the colorado forest and greater trunk biomass in the tabonuco forest for trees of the same size class. 277 £9;gg§_gyp§mics ggd,Productivit1 Dynamic features of the colorado forest are compared with other forest formations within the Luquillo Mountains in Table 22. Mortality rates in the palm and tabonuco plots measured during 30 years exceeded ingrowth while in the colorado plots measured during 35 years they were equal. These trends probably reflected differences in disturbance and recovery responses to past hurricanes. In the lower forests, trees are larger and Openings caused by treefall provided "growing space." Competition among numerous fast growing stems caused greater mortality. In the higher forests, trees are smaller and damage was mainly by breakage creating smaller openings. Tree growth was slower. Moreover, a considerable proportion of the ingrowth was by palms which are shade tolerant and require less growing space to mature. Because of the short duration of measurement in the dwarf forest (Weaver 1983), comparable records were not available. Loose litter tends to increase with elevation in the Luq0illo Mountains, but there are notable exceptions in the dwarf forest at the summits and in the palm forest. Comparison with the only other collections made along an elevational gradient on a single tropical mountain, Rhao Luang in Thailand, also showed an increase in loose litter with elevation except at the summit (Fig. 49A). In contrast, an increase in elevation in the Luquillo Mountains leads to a decline in litterfall with the notable exception of the palm forest (Fig. 49 B). A similar decline in litterfall is evident over an elevational gradient on Gunung Mulu in Sarawak. 278 1 Nmm.m smm.~ Nn.oN N.NN Amuoou +.eesonwo>oamv Nance zmfl.o x-.o xc.~ xwh.o Doom eoN.N ee.N 0N.Na N.oa eaaouwupoao Nance 9&2: 8 95 ems.o NN.o oN.N NN.N Nua\se\uv Nassau» amaaoam eoa.o NN.o --- HeNN Naa\.a\Nse Nassau» masNos e.uNo.o .cN.c .NN.c emN.o Ana\aov abacus use use: eoa.o Na.o oNe.N a.aoNN Nausoauas uuusNe N.NNN.¢ NN.o I-I NN.o Nae\se\uv use» Nuo>Nesmm --- N -.. NN any Naosanuua maNeamnm esN.s 0N.N us.N ac.e Naa\sv hasuNN «moss --I cN.o eN.o NN.N NaomewNNsuNNz . h «H.H em.o uwsum Ame av NNN av Na.o 5N.o umaONe III NN.~ cm.o eN.N noes N¢.N No.N eN.e sa.s News c~.m m.o om.m nae.» Aah\e:\uv Haemaouumg III om omm sum An%\o£\m50uev huwaouaoz --- oN sNN .NN Naa\aa\sassov eusoameN A «H035 Ace—whoa 00v c565 33523.5 0..—sumo w umouou odousoa peso: umouom Dos assumes posed amouom um: Hoownouunam swans»: .mcwmuoaox 039.95 as”. 9.333 somehow. anon you nowadays mucus mo 83.3930 .2 0.33. 279 cu wanes sauna mmz uoou wouuawumo + mmz acaouwu>onn gauchfl Amsmg >cnuszv mm: ecaouwm>onu n n.c I mmz noon vmumawumux omma soHOH muoaa Hmuu>om Bonn muaammu momma cuuozmumz can mouuwumu NmmH .fimma nuuoamuazp snag muauzwuwsm can save» cum“ uumvmcunm cs¢H aauou newuw>wau a coca um muOHa “emaa .Hw um um>mm3v munch n pom vuuamwos tam cowuu>uao E_onh um muonn umwm~ omaa van wwcmumu ~ma~ uwm.mm :auauxn muaom on new kusmaua can cowuu>~ao E cme um muoaa «mead uu>uu3u “woo: vcsouwo>ona Huuoa N uuuuwa «mooH\a~auumuumA~ Avuscwuaoov Nu «Hang 280 Figure “9. A - Estimates of loose litter storage in montane tropical forests over an elevation gradient on the same mountain: dots - Khao (Mt.) Luang. peninsular Thailand (Yoda and Kira 1969); triangles - Luquillo Mountains. Puerto Rico (Table 22). Circled triangle. palm forest. B - Estimates of litterfall in montane tropical forests over an elevation gradient on the same mountain; dots - Gunung Mulu. Sarawak (Proctor 31,31, 1983); triangles - Luquillo Mountains. Puerto Rico (Table 22). Circled triangle. palm forest. 281 30+- (0) ’5 3 \ :; zo- b 2. 3 0 I0- (I! O o 4 O I I l l 500 IOOO ISOO ZOOO l5- (b) ’2 > \ a t\\\\’— 8 2 lab- E :3 z: 5- .3 500 IOOO I500 2000 Figure 49 Elevofion (m) 282 A gradual increase in elevation is accompanied by cooler, wetter, and cloudier environmental conditions where the breakdown of organic matter in the loose litter is slowed (Grubb 1971; Edwards 1977). The noticeable dip in loose litter at the summits is partially due to the smaller size, and lower productivity of the dwarf forest. It may also reflect some loss, or distribution, of litterfall to the leeward by winds in the exposed areas (Welbourn gt 9;. 1981). The lower loose litter concentrations in the palm forest, in this instance, situated in a valley adjacent to a stream, reflects the low specific gravity of palm tissue at 0.26 g/cm3 as well as frequent flushing of the forest floor by floods removing organic matter from the soil profile (Frangi and Lugo 1985). The reduction of litterfall production with an increase in elevation correlates with a decline in tree size over the same gradient (Table 21). The exceptionally high litterfall production in the palm forest was attributed to the prevalence of palm which has a rapid turnover of large leaves combined with comparatively little wood production, due to the lack of secondary tissue growth (Frangi and Lugo 1985). Litterfall and loose litter accumulations in the colorado forest also demonstrate temporal variation. The higher leaf and total litterfall observed between April and June occur just after the comparatively drier months of January through March (Fig. 2, General Introduction). They also coincide with a period of higher average wind velocity. January through March have average wind velocities between 6 and 11 km/hr while April though June have average wind velocities between 14 and 19 km/hr (Briscoe 1966). The unusually high value for leaf and total litterfall in March of 1981 may be the result of high winds over the span of a day or two. 283 Loose litter accumulations are notably less during the first three months of the year, a period that coincides with lower rainfall, lower average wind velocity, and less cloud cover. This period characterized by less litterfall input and generally drier and clearer conditions may foster greater biotic activity and facilitate the decomposition of organic matter. Elsewhere, the annual variation in litter accumulations has been considerable with climatic events accounting for much of it (Spain 1984). An increase in the leafy portion of litterfall from 57.3 percent of the total litterfall in tabonuco forest through 74.3 percent in the colorado forest to 79 percent in the dwarf forest is evident (Table 22). This positive correlation of the percentage of leaf in litterfall with altitude has been observed elsewhere (Spain 1984). It appears related to the decrease in specific leaf area from tabonuco through colorado to dwarf forest combined with a decrease in the aboveground woody biomass (Table 21). Again, the palm forest is an exception because of the extremely large size and weight of the leaves, and generally low woody production. It should be pointed out, however, that similar trends were not apparent on Gunung Hulu in Sarawak (Proctor g; al. 1983) where the leafy portion of litterfall varied between 52 and 64 percent over an elevational gradient, but did not display elevational trends. An earlier summary of production of total litter for 226 forests of the world yielded estimates between 0.22 and 4.19 g/mzlday, with cool temperate forests averaging about 0.93 g/mzlday (Bray and Gorham 1964). It also indicated that year to year variation in tropical rainforest litterfall was low in comparison to other forest types, that both total and leaf litterfall were significantly and negatively related to altitude and latitude, and that leaf litterfall was significantly related to mean 284 rainfall. A recent summary of mesic forests with a continuous canopy cover (Jordan and Murphy 1978) showed that the rate of litter production decreases in direct proportion to the decrease in light available during the growing season along a worldwide gradient. Similarly, within the Luquillo Mountains, a gradient exists. Between El Verde and Pico del Oeste (Pico del Oeste is located near Pico del Este and has similar climate and vegetation), a distance of 8 km horizontally and 550 m vertically, litterfall decreased by about 30 percent whereas solar insolation between the north coast and Pico del Oeste declined by about 40 percent (Briscoe 1966; Baynton 1968). Comparison of colorado litterfall values with predictive equations showed excellent correlation of total annual litterfall with the temperature/precipitation ratio proposed by Brown and Lugo (1982). A previosuly suggested relationship between the common log of aboveground NPP vs. the common log actual evapotransportation (Rosenzweig 1968) overestimated litterfall. This discrepancy may be due to the inefficiency of actual evapotranspiration as a measure of productivity in very humid tropical environments with frequent cloud cover, generally cooler leaf temperatures, and frequent.heavy rainfalls. Much of the moisture evapotranspired in wet forests follows the evaporation route because many of the rainfalls are light (Wadsworth 1948). or is lost as runoff during heavy rainfalls. Litter turnover rate appears to decrease with elevation in the Luquillo Mountains, reflecting the cooler temperature, greater rainfall, and slower breakdown of organic matter. Litter turnover varies among forest types, but in the moist tropics, it usually takes one year (Olson 1963; Golley gt al. 1975). 285 Standing herbivory and herbivory rate records are incomplete and thus difficult to compare. It appears, however, that both may decline slightly with elevation. It has been suggested that ecosystems with high NPP may have higher herbivory rates because of greater amounts of available food and more favorable climatic conditions for insects (Benedict 1977). In Panama, mature leaves of pioneer species were eaten three to 10 times more rapidly than mature leaves of ’persistent’ species (Coley 1982). Growth of these species is usually faster, and less energy is expended on defensive mechanisms and, therefore, the leaves are more palatable. Moreover, with higher elevations in the Luquillo Mountains, the leaves become smaller and thicker, and may be more difficult for insects to consume. Also the climatic limits for insect infestation may be approached in the dwarf and other high elevation forests. It is probable that short leaf lifespan, high NPP, low energy investment in leaves, high palatability, and high herbivory rates are all related (Benedict I976). Diameter, volume and biomass increment all decrease with elevation as do aboveground NPP and total NPP estimates. The unique exception is that of the palm plot with a comparatively high value for each category. Wetland ecosystems have been shown to have high NPP rates (Brown and Lugo 1982). In general, the NPP rates in the forests of the Luquillo Mountains are low in comparison to other tropical moist, wet and rain forests (Murphy 1975; Medina and Klinge 1982; Brown and Lugo 1982), most of which fall in the range of 5 to 32 t/ha/Yr. Environments characterized by high precipitation and humidity, intense soil leaching and generally lower temperatures are those in which productivity may be limited (Murphy 1975). 286 The nggillg Fgrgst Contiguum The structural and floristic impoverishment of forests with an increase in elevation on small, wet, tropical mountains, such as the isolated, coastal mountains in the West Indies, has been attributed to the Massenerhebung Effect (Richards 1966; Grubb 1971), a phenomenon first recorded in the European Alps. Many investigators have forwarded ideas to account for this phenomenon in Puerto Rico and elsewhere, among them, saturated soils and reduced root respiration (Holdridge 1967), impeded soil drainage (Wadsworth and Bonnet 1951), physiological drought (Beard 1944), high winds in summit areas (Roy 0. Woodbury, pers. comm.), leaching of the soils combined with high fog incidence (Baynton 1969), reduced transpiration rates (Beard 1944; Odum 1968; Leigh 1975), fog, high soil water content and reduced mineralization of organic matter (Grubb 1971), and shallow soils (Gleason and Cook 1926). The soils are not shallow in most areas of the Luquillo Mountains so this explanation to account for a decline in size and species richness of the forests does not seem satisfactory. Soil leaching has also been questioned. The high cation exchange capacity characteristic of high elevation soils rich in organic matter should hold nutrients so that they would not be rapidly leached (Lyford 1971). Physiological drought, or ”the difficulty of drawing water from the soil coupled with increasing atmospheric power of evaporation from the leaves” (Beard 1941) is an idea that seems contrary to recent observations. Transpiration rates have been measured (Gates 1969; Weaver g; 5;. 1973; Medina g; 5;. 1981) and are generally low. With the frequent cloud cover in the summit areas, relative humidity is high, saturation pressure deficits are low (Odum 1968), and there is a positive addition of cloud water to the moisture 287 balance (Kerfoot 1968; Baynton 1969; Weaver 1972). None of these observations supports physiological drought. Yet the leaves are distinctly coriaceous and xeromorphic in appearance, and presumably adapted to occasional high winds and pronounced changes in environmental conditions characterized by cloud cover one moment and intensive solar isolation another. Saturated soils and impeded drainage have been observed (Wadsworth and Bonnet 1951), and such conditions would influence root aeration. Roots in the higher elevation forests are largely superficial (Wadsworth 1951; Lyford 1971) and spreading on the forest floor. Recent work with roots of food crops like tomatoes has shown that these plants decrease water use when oxygen is excluded from the root zone (Sojka and Stozy 1981). As the diffusion rate of oxygen in the soil decreases, the stomates close, independent of other factors such as soil water or light intensity. Forests at higher elevation in the Luquillo Mountains were called the "hurricane hardwood” type (Murphy 1916) suggesting that wind might play a role in their growth and survival. Specific reference was made to the contorted trees characteristic of the summits. Average wind velocities for El Yunque peak at 1050 m in elevation do not exceed 19 km/hr in any month (Briscoe 1966) although gusts of 100 km/hr were observed on El Toro peak at 1065 m during a period of normal winds elsewhere on the island (Wadsworth 1948). Hurricanes have had an influence on the species composition and dynamics of the colorado forest (Part I). Despite this influence, observed windfalls at higher elevations are rare, and gaps in the forest are not common. Indirect evidence of this is also available in 288 that pioneer tree species are notably lacking in the dwarf forest (Byer and Weaver 1977) and are not abundant in the colorado forest. Among the possible adaptations to minimize the effects of hurricane damage in higher elevation forests is the development of denser woods. On windy sites, tree species with dense wood should experience lower rates of mortality, windthrow and snapping than less dense species (Lawton 1984). The mean specific gravity of trees considered to be climax for the upper forests average about 0.65 g/cm3 in the dwarf forest and about 0.60 g/ cm3 in the colorado forest, the latter figure applying to slopes and ridges. In valleys, the average specific gravity is lower. Consequently, growth would be slower due partly to the investment in wood density and partly, perhaps, to the demands for nutrients to produce the denser wood. Another factor related to high winds is the energetic cost of replacing foliage after storms. Direct observation of the effects of Hurricane Eloise of September 1975 showed an eight percent defoliation of the standing leaf area in the dwarf forest (Benedict 1976). Despite the fact that the trajectory of the storm was at a considerable distance from the island, the effects were noticed in the high elevation forests. Relatively cool temperatures, frequency of fog and heavy rainfalls characterize the higher elevation forests, and interact to retard the mineralization of organic matter. It has been suggested that these environmental factors may effectively lower the elevational level at which shorter forest, mainly controlled by temperature, would occur (Grubb 1971, 1977; Weaver g; 5;. 1986). Ascent of the Luquillo Mountains shows a gradual impoverishment of the forests both in stature and species composition. Indeed, the progressively cooler, wetter, and cloud-covered forests of higher elevations appear to constitute a stress environment for 289 trees, reflected in a continuum that is compressed into a narrow altitudinal range that extends from tabonuco forest at lower elevations to dwarf forest at the summits. 290 SUMMARY A total net primary production (NPP) of 7.83 t/ha/yr was estimated for the colorado forest at 750 m elevation in the Luquillo Mountains of Puerto Rico. NPP was determined by summing litterfall (6.80 t/ha/yr), herbivory (0.21 t/ha/yr), and net biomass accrual above and below ground (0.82 t/ha/yr). Several structural and dynamic components of the colorado forest were also estimated: (1) a total leaf area index of 4.96 mzlm2 partitioned into aerial components of tree leaves (3.01) and epiphytes (0.30), and ground components of herbs and grasses (1.18) and bryophytes (0.47); (2) a mean leaf weight for 17 dicotyledonous species of 0.67 g with a mean leaf area of 56.4 cm2 (mesophyll size class) and a mean specific leasf area of 91.0 cmzlg of leaf surface; (3) an aboveground volume (trunks and branches) of 220 m3/ha and aboveground woody biomass of 130 t/ha; (4) an estimate of standing herbivory of 5.1 percent of the leaf area and herbivory rates of 4.0 percent of total leaf area per year, both on a forest-wide basis; and (5) a mean litterfall rate of 680 g/mzlyr with a mean standing crop of 875 g/m2 of litter and a turnover rate of 0.78 times per year. Information on the structure and dynamics of other forest types in the Luquillo Mountains was compiled and compared with the results of this study to highlight the effects of a 600 m elevational gradient. 0f the 291 structural features considered, the number of stems per hectare, basal area, and soil organic matter, all increased with an increase in elevation whereas canopy height, range of tree diameters, forest volume and biomass, leaf area index, specific leaf area, and species richness declined. 0f the dynamic features studied, ingrowth and mortality of trees, litterfall, loose litter, standing herbivory and herbivory rates, litter turnover, growth in diameter, volume and biomass, aboveground woody NPP, and total NPP, all appeared to decline with an increase in elevation. The structural and floristic impoverishment of the forest types with ascent, and their compression into a narrow altitudinal range, reflect the effects of numerous environmental factors among which the relatively cooler temperatures, frequency of fog, and heavier rainfalls interact to produce saturated soils and a retardation in the mineralization of organic matter. LITERATURE CITED LITERATURE CITED Bannister, B. A. 1970. Ecological life cycle of Euterpe globos; Gaertn. p. B 299-314. In R.T. Odum and R.F. Pigeon, eds. A tropical rain forest. USAEC, TID-24270. Baynton, R. W. 1968. The ecology of an elfin forest in Puerto Rico, 2. The microclimate of "Pico del Oeste." Journal of the Arnold Arboretum 49:419-430 . Beard, J. S. 1941. Montane vegetation in the Antilles. Caribbean Forester 3(2):61-74. Beard, J. S. 1944. Climax vegetation in tropical America. Ecology 25(2):127-158. Benedict, F. F. 1976. Herbivory rates and leaf properties in four forests in Puerto Rico and Florida. M.S. thesis, University of Florida, Gainesville. 78 p. Bray, J. R. and E. Gotham. 1964. Litter production in forests of the world. In J. B. Cragg, ed., Advances in ecological research 2:101-157. Briscoe, C. B. 1966. Weather in the Luquillo Mountains of Puerto Rico. Institute of Tropical Forestry Research Paper ITF-3. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. 250 p. Briscoe, C. B. and F. E. Wadsworth. 1970. Stand structure and yield in the tabonuco forest of Puerto Rico. p. B 79-90. In H. T. Odum and R. F. Pigeon, eds. A tropical rain forest. USAEC, TID-24270. Brown, S. and A. E. Lugo. 1982. The storage and production of organic matter in tropical forests and their role in the global carbon cycle. Biotropica 14(3):l6l-187. Byer, M. D. and P. L. Weaver. 1977. Early secondary succession in an elfin woodland in the Luquillo Mountains of Puerto Rico. Biotropica 9:35-47. Coley, P.D. 1982. Rates of herbivory on different tropical trees. p. 123-132. In E.G. Leigh Jr., A.S. Rand and D.M. Windsor, eds. The ecology of a tropical forest: seasonal rhythums and long-term changes. Smithsonian Institution Press, Washington, D.C. Crow, T. R. 1980. A rainforest chronicle: a 30-year record of change in structure and composition at El Verde, Puerto Rico. Biotropica 292 293 Edwards, P. J. and P.J. Grubb. 1977. Studies of mineral cycling in a Montane Rain forest in New Guinea. 1. The distribution of organic matter in the vegetation and soil. Journal of Ecology 65:943-969. Frangi, J. L. and A. E. Lugo. 1985. Ecosystem.dynamics of a subtropical floodplain forest. Ecological Monographs 55:351—369. Freese, F. 1967. Elementary statistical methods for foresters. Agriculture Handbook Mo. 317. USDA Forest Service, Washington, D. C. 87 p. Gates, D. M. 1969. The ecology of an elfin forest in Puerto Rico, 4. Transpiration rates and temperatures of leaves in cool humid environment. Journal of the Arnold Arboretum 50(1):93-98. Gleason, H. A. and M. Y. Cook. 1927. Plant ecology of Porto Rico, Vol. 7, Parts 1 and 2. p. 1-173. In Scientific Survey of Porto Rico and the Virgin Islands. New York Academy of Sciences, Mew'York. Golley, F. B., J. Ewel and G. I. Child. 1976. Vegetation biomass of five ecosystems in north western Colombia. Tropical Ecology 17:16-22. Grubb, P. J. 1971. Interpretation of the ’Massenerhebung Effect’ on tropical mountains. Mature 229(5279):44-45. Grubb, P. J. 1977. Control of forest growth and distribution on wet tropical mountains: with special reference to mineral nutrition. Annual Review of Ecology and Systematics 8:83-107. Holdridge, L. R. 1967. Life zone ecology. Tropical Science Center. San Jose, Costa Rica. 206 p. Jordan, C. F. and P. G. Murphy. 1978. A latitudinal gradient of wood and litter production, and its implication regarding competition and species diversity in trees. The American Midland Naturalist 99(2):415-434. Kerfoot, 0. 1968. Mist precipitation on vegetation. Forestry Abstracts 29:8-20. Lawton, R. 0. 1984. Ecological constraints on wood density in a tropical montane rain forest. American Journal of Botany 71(2:261-267. Leigh, E. G. Jr. 1975. Structure and climate in tropical rain forest. Annual Review of Ecology and Systematics 6:67-86. Lyford, W. H. 1969. The ecology of an elfin forest in Puerto Rico, 7. Soil, root, and earthworm relationships. Journal of the Arnold Arboretum 50(2):210-224. Medina, E. and H. Klinge. 1981. Productivity of tropical forests and 294 tropical woodlands. Chapter 9. In 0. L. Lange, P. S. Nobel, B. Osmond and H. Ziegler eds. Part 12 D, Physiological Plant Ecology, IV, Ecosystems processes: mineral cycling productivity and man’s influence. Springer-Verlag, Berlin, Germany. Medina, E., E. Cuevas and P. L. Weaver. 1981. Composicion foliar y transpiracion de especies lenosas de Pico del Este, Sierra de Luquillo, Puerto Rico. Acta Cientifica Venezolana 32:159-165. Murphy, L. S. 1916. Forests of Porto Rico: past, present, and future. USDA Bulletin No. 354. Washington, D.C. 99 p. Murphy, P. G. 1975. Net primary productivity in tropical terrestrial ecosystems. p. 217-231. In H. Leith and R. H. Whittaker, eds. Primary productivity of the biosphere. Springer-Verlag, New York. Odum, E. P. 1971. Fundamentals of ecology. W.B. Saunders Company, Philadelphia, PA 574 p. Odum, H. T. 1968. Work circuits and systems stress. p. 81-138. In H. E. Young, ed. Primary productivity and mineral cycling in natural ecosystems. University of Maine Press, 0rono. Odum, H. T. 1970. Summary: an emerging view of the ecological system at El Verde. p. I 191-289. In H. T. Odum and R. F. Pigeon, eds. A tropical rain forest. USAEC, TID-24270. Odum, H. T. and J. Ruiz-Reyes. 1970. Holes in leaves and the grazing control mechanism. P. I 69-80. In H. T. Odum and R. F. Pigeon, eds. A tropical rain forest, USAEC, TID-24270. Odum, H. T. and R. F. Pigeon eds. 1970. A tropical rain forest. USAEC , TID-24270 . Olson, J. S. 1962. Energy storage and the balance of producers and decomposers in ecological systems. Ecology 44(2):322-331. Ovington, J. D. and J. 3. Olson. 1970. Biomass and chemical content of El verde lower montane rain forest plants. p. H 53-77. In H. T. Odum and R. F. Pigeon, eds. A tropical rain forest. USAEC, TID-24270. Proctor, J., J. M. Anderson and H. W. Vallack. 1983. Comparative studies on forests, soils, and litterfall at four altitudes on Gunung Mulu, Sarawak. The Malaysian Forester 46:60-75. Reagan, D. P., R. W. Garrison, J. E. Martinez, R. B. Waide, and C. P. Zucca. 1982. Tropical rain forest cycling and transport program: phase 1 report. Center for Energy and Environment Research and the University of Puerto Rico, San Juan, Puerto Rico. 183 p. Richards, P. W. 1966. The tropical rain forest. Cambridge Press, England. 450 p. 295 Rosenzweig, M. L. 1968. Net primary productivity of terrestrial communities: prediction from climatological data. The American Naturalist 102(923):67-74. Spain A. V. 1984. Litterfall and the standing crop of litter in three tropical Australian rainforests. Journal of Ecology 72:947-961. UNESCO. 1978. Tropical forest ecosystems. UNESCO/UNEP/FAO. Paris, France. 683 p. Wadsworth, F. H. 1948. The climate of the Luquillo Mountains and its significance to the people of Puerto Rico. Caribbean Forester Wadsworth, F. H. 1951. Forest management in the Luquillo Mountains, I. The setting. Caribbean Forester 12(3):114. Wadsworth, F. H. 1952. Forest management in the Luquillo Mountains, III. Selection of products and silvicultural policies. Caribbean Forester 13:93-119. Wadsworth, F. H. and J. A. Bonnet. 1951. Soil as a factor in the occurrence of two types of montane forest in Puerto Rico. Caribbean forester 12:67-70. Weaver, P. L. 1972. Cloud moisture interception in the Luquillo Mountains of Puerto Rico. Caribbean Journal of Science 12:129-144. Weaver, P. L. 1983. Tree growth and stand changes in the subtropical life zones of the Luquillo Mountains of Puerto Rico. USDA Forest Service Research Paper 80-190. Southern Forest Experiment Station, New Orleans, LA 24 p. Weaver, P. L., M. D. Byer and D. Bruck. 1973. Transpiration rates in the Luquillo Mountains of Puerto Rico. Biotropica 5(2):123-133. Weaver, P. L., E. Medina, D. Pool, K. Dugger, J. Gonzalez-Liboy and E. Cuevas. 1986. Ecological observations in the dwarf cloud forest of the Luquillo Mountains of Puerto Rico. Biotropica. 18:79-85 Welbourn, M. L., E. L. Stone and J. P. Lassoie. 1981. Distribution of net litter inputs with respect to slope position and wind direction. Forest Science 27:651-659. Yoda, K. and T. Kira. 1969. Comparative ecological studies on three main types of forest vegetation in Thailand V. Accumulation and turnover of soil organic matter, with notes on the altitudinal soil sequence on Khao (Mt.) Luang, peninsular Thailand. Nature and Life in South-East Asia 5:83-11. APPENDIX TABLE 8 296 .n... ..... 2.8 3.8 8... S... 8.. 2...... ... a... "_..de 33...... S... 8... 8... n... a... a... s... 8.... 1. ... «mylu 83...... 8.... 8.8. 2.... 2.8. .3... B... ...8 8...... .6. .8. 338.888.... 8.288... 2... «n... 8... an... a... 8... 8... 2...... .... a... 8.8.1888... 3.2.8.... .1... 8.8 8... 8... 8.. a... 8.. 8...... .... a... 33.2288... 8.2.8.... 8.... 8.... 8.... 8.... 8.... .n... 8.. 2.8.. a... .3. 8.8.8.2.... 8.2.8.3.: 2.... 8.... 8... 8.8 8... 8... 8.. 3...... .... s... 88.3.. 88.... 8.... .....n ...:. 8.8. 2.8. 8... a... 8...... 1.. .6. 8.8.8... 1...... 3.... 8.8. 8... 8... 3.8 8... .... 8.88 .... in. 93.88... 3...... 8.8 8.8 8.8 8... ..... 8.. a... 8.... o.. N... .8888... 3...... 8... a... a... 8... 3... 8.. ...u 2...... .6. a... 33.8.... 3...... 8... 8... 8... 8.. 8.. 8.. 8.. 2.... ... o.. 3.2.8... 3...... 8.... 8.8. 8:... 8... 8... 8.. 8.. 8.8.... .3. .... 358......— 8.25 ...8 8.8 S... .n... 8... .... .... 8.8.. N... .5. 388...... 8.8. a... ..... z... ..... 8.. 8.. 8.. 2...... .... o... 2.88.8.2. 8.96 a... 8... 8... 8.. .1. 3.. 8.. 8.8.. a. m. 84.13% a 8.. 8.. 3.. a... r... a... 8.. .3... ... a... .18..ch 1.38 .9... ..m n... m. n... 1 .3... .83 .8... .284... .25... 2. .3. my: 23.: H.933... a... .8888. 8:. .38... .3... _..... 3.8... .o Sal... . n .2... 5.8....