3 ‘95q33 Ill W“ IUWWWMMWW 539 3263 LIBRARY Michigan State University This is to certify that the thesis entitled COMPOSITION AND STRUCTURE OF FOREST EDGE IN BEECH-SUGAR MAPLE AND OAK FRAGMENTS IN CENTRAL LOWER MICHIGAN presented by Brian Josef Palik has been accepted towards fulfillment of the requirements for M.S.Jiegree in Botany and Plant Pathology QMW Major professor .‘ Date 26 July 1988 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution ‘IVIESI_J RETURNING MATERIALS: Place in book drop to LIBRARJES remove this checkout from ‘ a your record. FINES w1'Il ' be charged if book is returned after the date Stamped below. 5.59191 1s~ i .- 62.1 Q G , 1&1 “r b9 9.3 a} 1...— £- 13;: ‘V‘ 06139 9 3,991 ‘ ,1 '19" ‘ let“? Jug ZUIfi i039". MAGICz " NOV 2 4 0812 019390 5;; f 91991720011 utévt’, COMPOSITION AND S'I'HJCTURE OF FOREST EDGE IN BEECH-SUGAR MAPLE AND OAK FRAGMENTS IN CENTRAL LOVER MICHIGAN Brian Josef Pal ik A THESIS Subnitted to Michigan State University in partial mlfillment of the requiranents for the degree of MASTER OF SCIENCE Department of Botany and Plant Pathology 1988 COMPOSITION AND STRUCTURE OF FOREST EDGE IN BEECH-SUGAR MAPLE AND OAK FRAGMENTS IN CENTRAL LOVER MICHIGAN By Brian Josef Palik Edge-to—interior gradients of trees and shrubs were examined in four hardwood forest fragments. In general, edge forest was characterized by higher stan densities and a greater number and importance of less tolerant and disturbance—oriented species, relative to the interior. The development and depth of edge characteristics was dependent upon aspect. size, and the disturbance history of a site. Edge conditions were best developed in: 1) an old-growth fragment on the southern aspect; and 2) a very small fragment with a major disturbance in the northern aspect forest. In two formerly grazed fragments edge conditions were less well developed, presumably as a result of edge- species elimination during grazing. Landscape-level managanent plans should consider not only size, but disturbance history and successional status of a site when estimating the amount of interior forest present in a specific fragment. I extend sincere thanks to my advisor, Dr. Peter Murphy. The independence and trust he granted made my successes truly of value. The excellence he displayed in his roles as teacher, counselor, scientist, and friend has inspired me throughout our association. For all that he has done, I am happily indebted. I thank my guidance conmittee manbers, Dr. John Beaman, wine has taught me what it means to be a graduate student and has shown me what it means to be a scholar, and Dr. Gerard Donnelly, whose valued cements and insight have placed me on the road to realistic research goals. My appreciation goes to Dr. Stephen Stephenson for numerous statistical and analytical answers, and Dr. Sheridan Dodge for site information. My friends and col leagues, William Collins, Christine Durbahn, Kimberly Medley, and Janet Salzwedel, provided field assistance. My dear forest friend, Rose—Marie Muzika, provided humane inspiration. I give very special thanks to Janet, my untrierful companion of the last three years. Her tolerance has both anazed and saddened me. I predict a future with Salavedel Acid and will be proud to call the discoverer my friend. iii The refuge provided by my family throughtout my graduate career has kept me relatively sane and I am delighted that they will renain within reach during the next step. Finally, I thank Andrew and Josef, their love of trees has been my light in the forest. iv TABLE OF 00m Ease LIST OF TABLES ................................................... . viii LIST OF FIGURES ................................................. x INTRODUCTION ........... . ........................................ 1 Forest Islands and Edge .................................... 3 The Microclimatic Gradient ................................. 6 Edge Effects .......................................... 6 Aspect Effects ........................................ 7 Empirical Evidence for the Microclimatic Gradient ..... 8 The Vegetaticnal Gradient .................................. 9 Fragment Disturbance and Edge History ...................... 14 Heterogeneity of the Landscape Matrix ...................... 15 Research Hypotheses and Objectives ......................... 15 METHODS ......................................................... 18 Site Selection ............................................. 18 Sampling Dbthcds ........................................... 21 Transact Location ..................................... 21 Vegetation ............................................ 23 SoilMoistureandLight................ ............... 24 Data Analysis .............................................. 25 Vegetation. ..... .................. .. 25 Soil Moisture .and Light ............................... 27 Determination of Edge Depth ........................... 27 RESULTS ......................................................... Soil Moisture and Light .................................... General Edge Structure ..................................... Beech—Sugar Maple Fragments ........................... Oak Fragments ......................................... Individual Species Distribution and Abundance .............. Beech-agar Maple Fragments ........................... Oak Fragments ......................................... The Canpositional Gradient ................................. Species Richness .................. ImportancePercentages ................................ DISCUSSION ...................................................... The Effects of Edge and Aspect ............................. Soil Moisture and Light .................................... General Characteristics of Edge Forest ..................... General Effects of Aspect .................................. Site Specific Distinctions ................................. Beech—Sugar Naple Fragments ........................... Oak Fragments ......................................... Depth of Edge .............................................. Within-and Between—Site Canparisons ........................ Minimum Critical Fragment Size ............................. CONCLUSION ...................................................... APPENDIX A: TCNNSHIP AND RANGE (DORDINATES OF STUDY SITES. . APPENDIX B: SITE DIAGRAI‘B .................................. vi 29 29 36 36 45 50 55 73 93 93 97 109 109 112 118 124 129 129 133 139 143 146 148 148 150 152 152 154 APPENDIX C: THE UTILITY OF DETRENDED CORRESPONDENCE ANALYSIS IN AN EXAMINATION OF EDGE AND ASPECT EFFECTS ........................................ LIST OF REFERENCES .............................................. vii 167 Table 10. 11. 12. LIST OF TABLES Soil moisture (to 12 cm depth) by aspect and date in Site 2. Dunnett's pair-wise comparisons between soil moisture values along the edge to interior gradients (0-50 m) on the northern and southern aspects and in the interior (50 m from all edges) in Site 2. Bartholcmew's test of ordered alternatives for light values along the edge to interior gradients (0-50) on the northern and southern aspects in Site 2. Dunnett's pair-wise comparisons between sapling stem densities along the edge to interior gradients (0-50 m) on the northern and southern aspects and the interior (50 m from all edges) in Site 1. Density per hectare by stratum on the northern and southern aspects (0—50 m) in Sites 1 through 4. Species composition of Site 1. Species canpositian of Site 2. Species composition of Site 3. Species canposition of Site 4. Density per hectare of selected species in the sapling, shrub, and ground layer (g.l.) strata on the northern and southern aspects (0-50 m) in beech-sugar maple fragments. Density per hectare of selected species in the canopy, sapling, shrub, and graurxi layer (g.l.) strata on the northern and southern aspects (0—50 m) in oak fragments. Species richness, separated into canopy arxi non-canopy components, on the northern and southern aspects (edge to 50 m) in Sites 1-4. viii 32 33 37 41 42 51 52 53 54 57 76 95 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Mean species richness in Sites 1 through 4 separated ' into campy and mn—campy species. 98 Correlation between importance percentage of potential canopy daninant species and A) July soil moisture, B) light along the edge to interior gradients (0-50 m) on the mrthern and southern aspects in Site 2. 114 Correlations between light and A) canopy, sapling, and shrub densities, B) July soil moisture along the edge to interior gradients (0-50 m) on the mrthern and southern aspects in Site 2. 116 Correlations between July soil moisture and stem densities along the edge to interior gradients (0-50 m) on the mrthern and southern aspects in Site 2. 117 The contribution (percent) to total species richness in three areas of each fragment by either 1) species indicative of disturbed habitats or 2) species characteristic of a mn-disturbed habitat other than beech—sugar maple forest (in Sites 1 and 2) or oak forest (in Sites 3 and 4). 121—23 The number of species on the mrthern and southern aspects in each site that were either 1) species indicative of disturbed habitats. or 2) species characteristic of a mn—disturbed habitat other than beech—sugar maple forest (for Sites 1 and 2) or oak forest (for Sites 3 and 4). 126 Importance percentages of potential campy dcminant species by stratum on the mrthern and southern aspects in Sites 1 through 4. 128 Density (stars per hectare) in subcanopy strata in Sites 3 and 4. 137 The number of transitional events, smmmed over the campy, sapling, shrub, and ground layers, along the edge to interior gradients on the mrthern and southern aspects in Sites 1 through 4. 141 Eigenvalues by stratum in Sites 1-4. 161 ix LIST OF FIGURES 11911.13. —§9—P e 1. Location of study area and sites. Adapted from Dodge (1984) . 19 2. Transect location and sampling design for a hypothetical forest fragment. Trees (> 10 cm dbh) and saplings (2.5 cm _<_ dbh 5 10 cm) were sampled in the 10 x 5 m plots, shrubs and small trees (<2.5 cm dbh and taller than 1m) weresampled inthe5x3msubplots, thewoodyground layer (< 1 m tall) was sampled in the 1 x 1 m subplots. Three to five 50 m transects were sampled on both the mrthern and southern aspects of a site. Transects were located at least 40 m interior of the western and eastern edges. Three to five plots located 50 m interior of all edges were also sampled in each site. 22 3. Soil moisture profiles (percent of wet weight; to 12 cm depth) fromtheedge to 50monthemrthernand southern aspects in Site 2. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges; OUT = 1 m outside of the edge. The y-axis values are means +/- se; n=5. 30 4. Light profiles (percent of full sunlight) fran the edge to 50 m on the mrthern and southern aspects in Site 2. The y-axis values are means +/- as; n=12. Note the difference in y-axis scale on the two aspects. 35 5. Stem density profiles from the edge to 50 m on the mrthern and southern aspects in Site 1. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. The y-axis values are means +/- as; n=5. 38 6. Stem density profiles fran the edge to 50 m on the mrthern and southern aspects in Site 2. The x-axis values are the upper limits of 5 m distame intervals. The interior (INT.) is 50 m from all edges. The y-axis values are means +/- se; n=5. 43 10. 11. 12. 13. Stem density profiles from the edge to 50 m on the mrthern and southern aspects in Site 3. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. The y—axis values are means +/- se; n=5. Stem density profiles fran the edge to 50 m on the northern and southern aspects in Site 4. The x-axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. The y-axis values are means +/- se; n=3. Sugar maple campy and sapling density profiles from the edge to 50 m on the mrthern and southern aspects in Site 1. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m fran all edges. The y-axis values are means +/- se; n=5. Sugar maple shrub and ground layer density profiles fromtheedgeto 50monthemrthernandsouthern aspects in Site 1. The x-axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m fran all edges. The y-axis values are means +/— se; n=5. American beech canopy and ground layer abundance profiles from the edge to 50 m on the mrthern and southern aspects in Site 1. The x—axis values are upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—axis values are the total number of stems encountered in each 5 m interval and in the interior. The sapling and shrub profiles were similar to the campy and ground layer profiles respectively. White ash campy and shrub abundance profiles from the edge to 50 m on the mrthern and southern aspects in Site 1. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are the total number of stems encountered in each 5 m interval and in the interior. The sapling and ground layer profiles were similar to the canopy and shrub profiles respectively. Redoakcampyarxishrubahmdanoeprofilesfromthe edge to 50 m on the mrthern and southern aspects in Site 1. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—axis values are the total number of stats encountered in each 5 m interval and in the interior. The sapling and ground layer profiles were similar to the campy and shrub profiles respectively. xi 46 48 56 59 61 62 14. 15. 16. 17. 18. 19. Slippery elm campy and shrub abundance profiles from theedge to50monthemrthernandsouthernaspectsin Site 1. The x—axis values are the upper limits of 5 distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y—axis values are the total number of stems encountered in each 5 m interval and in the interior. The sapling and ground layer profiles were similar to the campy and shrub profiles respectively. 63 Basswood campy and ground layer abtmdance profiles from the edge to 50 m on the mrthern and southern aspects in Site 1. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m frcm all edges. The y—axis values are the total number of stems encountered in each 5 m interval and in the interior. The sapling and shrub profiles were similar to the canopy profile. 64 Sugar maple campy and sapling density profiles fran the edge to 50 m on the mrthern and southern aspects in Site 2. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—axis values are means +/- as; n=5. 66 Sugar maple shrub and ground layer density profiles from the edge to 50 m on the mrthern and southern aspects in Site 2. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y—axis values are means +/- se; n=5. 68 American beech sapling (top) and basswood campy (bottcm) abundance profiles fran the edge to 50 m on the mrthern and southern aspects in Site 2. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y-axis values are the total number of stems encountered in each 5 m interval and in the interior. Beech carnpy. shrub, and ground layer profiles were similar to the sapling profile; basswood sapling, shrub, and ground layer profiles were similar to the canopy profile. 69 white ash campy and shrub abundance profiles from theedgeto50monthemrthernandsmthernaspectsin Site 2. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are the total number is stens encountered in each 5 m interval and in the interior. The sapling profile was similar to the campy profile. 70 xii 20. 21. 22. 23. 24. 25. Ironwood campy and sapling abundance profiles from theedgeto 50monthemrthernandsouthernaspects in Site 2. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y—axis values are the total number of stars encountered in each 5 m interval and in the interior. Black oak canopy and ground layer abundance profiles fromtheedge to 50monthemrthernandsouthern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are the total number of st- encountered in each 5 m interval and in the interior. Pignut hickory canopy and shrub abundance profiles from the edge to 50 m on the mrthern and southern aspects in Site 3. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—axis values are the total number of stems encountered in each 5 m interval and in the interior, The sapling profile was similar to the shrub profile. Red maple campy abundance (top) and sapling density (bottom) profiles from the edge to 50 m on the mrthern and southern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y-axis values in the top profile are the total number of stems encountered in each 5 m interval and in the interior. The y-axis values in the bottom profile are means +/-— se; n=5. Red maple shrub and ground layer density profiles fran theedgeto 50monthemrthernandsouthernaspects in site 3. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y-axis values are means +/- se; n=5. Black cherry campy and shrub density profiles from the edge to 50 m on the mrthern and southern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y-axis values are means +/- se; n=5. The sapling and ground layer profiles were similar to the shrub profile. xiii 72 74 75 78 79 80 26. 27. 28. 29. 30. 31. 32. Rosa multiflora shrub and ground layer density profiles from the edge to 50 m on the mrthern and southern aspects in Site 3. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y—axis values are means +/— se; n=5. White oak (top) and red oak (bottom) campy abundance profiles from the ecbe to 50 m on the mrthern and southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are the total number of stems encountered in each 5m interval and in the interior. Red oak sapling and shrub abundance profiles from the edge to 50 m on the mrthern and southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are the total number of stems encountered in each 5 m interval and in the interior. Red maple sapling density profile frcm the edge to 50 m on the mrthern and southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y-axis values are means +/- se; n=3. White ash campy and sapling abundance profiles from the edge to 50 m on the mrthern and southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are the total number of stems encountered in each 5 m interval and in the interior. White ash shrub and ground layer density profiles fran theedgetoSOmonthemrthernandsouthernaspectsin Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all aspects. The y—axis values are means +/- se; n=3. Black cherry sapling and ground layer density profiles frantheedgeto50monthemrthernandsouthern aspects in Site 4. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are means +/- se; n=3. The shrub profile was similar to the ground layer profile. xiv 81 84 85 87 88 89 33. 34. 35. 36. 37. Hawthorn shrub abundance (top) and gray dogwood ground layer density (bottom) profiles frcm the edge to 50 m on the mrthern and southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m interior of all edges. The y—axis values in the top profile are the total number of stems encountered in each 5 m interval and in the interior. The y—axis values in the bottom profile are means +/- se; n=3. Species richness profiles from the edge to 50 m on the mrthern and southern aspects in beech-sugar maple fragments. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m fran all edges. Species richness is separated into canopy species - and mn—canopy species Species richness profiles from the edge to 50 m on the mrthern and southern aspects of oak fragments. The x- axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m fran all edges. Species richness is separated into campy species - and mn-campy species . Importance percentage of potential campy dominant species fran the edge to 50 m on the mrthern and southern aspects in Site 1. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. Potential campy dominant species include American beech and sugar maple. Each percentage is of a total importance value (IV) of 300 in the campy and sapling layers, or 200 in the shrub and ground layers. Total IV is the sum of individual species IV's (relative density + relative frequency 4- relative basal area in the campy and sapling layers; relative density + relative frequency in the shrub and ground layers) in each interval. Inportance percentage of potential campy daninant species fromtheedgeto 50monthemrthernand southern aspects in Site 2. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. Potential canopy dominant species include American beech and sugar maple. Each percentage is of a total inportance value (IV) of 300 in the campy and sapling layers, or 200 in the shrub and ground layers. Total IV is the sum of individual species IV's (relative density + relative frequency + relative basal area in the campy and sapling layers; relative density 4» relative frequency in the shrub and ground layers) in each interval. XV 91 94 96 100 102 38. 39. 40. 41. 42. 43. Importance percentage of potential canopy daninant species fromtheedgetoSOmonthemrthernand southern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. Potential campy dominant species include: black oak; white oak; red maple; sugar maple; and American beech. Each percentage is of a total importance value (IV) of 300 in the campy and sapling layers, or 200 in the shrub and ground layers. Total IV is the sum of individual species IV's (relative density 4» relative frequency + relative basal area in the campy and sapling layers; relative density + relative frequency in the shrub and ground layers) in each interval. 105 Importance percentage of potential campy daninant species from the edge to 50 m on the mrthern and southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. Potential campy dominant species include: white oak; red oak; red maple; sugar maple; and American beech. Each percentage is of a total importance value (IV) of 300 in the campy and sapling layers, or 200 in the shrub and ground layers. Total IV is the sum of individual species IV's (relative density + relative frequency + relative basal area in the campy and sapling layers; relative density + relative frequency in the shrub and ground layers) in each interval. 107 Black cherry ground layer density profile from the edge to 50 m on the mrthern and southern aspects in Site 1. The x—axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. The y-axis values are means; n=5. 111 Site diagrams . 153 DCA ordinations of positions along the edge to interior gradients in Site 1. Symbols correspond to 5 m distance intervals fran the edge to 50 m on the mrthern and southern aspects, (e.g. Sl= 0-5 m from the southern edge and N10= 45—50 m fran the mrthern ecbe). The interior (I) is 50 m from of all edges. Note that positions N1-N5, N8-N10 and I have equal scores in the sapling ordination. 159 DCA ordinations of positions along the edge to interior gradients in Site 1. Symbols correspond to 5 m distance intervals fran the edge to 50 m on the mrthern and southern aspects, (e.g. Si= 0-5 m from the southern edge and N10= 45-50 m from the mrthern edge). The interior (I) is 50 m from all edges. 160 xvi 44. 45. DCA ordinations of positions along the edge to interior gradients in Site 3. Symbols correspond to 5 m distame intervals from the edge to 50 m on the mrthern and southern aspects, (e.g. Sl= 0-5 m from the southern edge and N10= 45-50 m frcm the mrthern edge). The interior (I) is 50 m from of all edges. DCA ordinations of positions along the edge to interior gradients in Site 4. Symbols correspond to 5 m distance intervals fran the edge to 50 m on the mrthern and southern aspects, (e.g. Sl= 0-5 m from the southern edge and N10= 45-50 m fran the mrthern edge). The interior (I) is 50 m from all edges. Note that N1 and N6 were omitted from the campy ordination and N6 was omitted from the sapling ordination. xvii 163 165 INTRODUCTION The postsettlement history of land we in the eastern United States has been marked by large scale deforestation and landscape conversion. Within a span of two to three generations, the once contiguous Eastern Deciduous Forest has been reduced to a patchwork mosaic of isolated forest fragments surrounded by agricultural and urbanized land (Hill 1985; Whitney 62 Scmerlot 1985). Forested areas mmal ly are influenced by natural disturbances ranging from individual treefal ls (Bray 1956; Runkle 1982; Donnelly 1986) to fire, catastrophic blowdown, and insect infestation (Heinselman 1973; Pickett 1979; mm 93 g1. 1983; Sprugel 1984). None of these, however, have the impact that mitten-induced deforestation and landscape conversion have on ecosystem stability and preservation (Curtis 1956; Burgess & Sharpe 1981; Wilcove gt _a_l. 1986). The value of forested areas often involves qualities that are difficult to measure. These imlude watershed stability, climate modification, erosion control, habitat and species diversity, and aesthetic appeal (Bormann & Likans 1979; Hill 1985). Futhermore, forests provide direct econanic contributions through lumber, pulpwood, fuelwood and recreation (Kittredge 1948; curtis 1956; Hill 1985) . Remnant forest fragments are the last strongholds in much of the eastern United States for the ecosystems that once dominated the landscape. Any hope of preserving present landscape characteristics and recovering sane degree of presettlement landscape characteristics must focus on the vegetation that remains. Therefore, a thorough understanding of forest fragments is vital. One of the ways in which the effects of fragmentation are manifest is through the development of edge forest (Ranney 1978). Studies focusing on the role of edge forest in the matrix created through landscape conversion have been limited. Because the effectiveness of potential landscape-level management plans that involving forest fragments may be dependent on an understarxiing of the role of edge forest in the matrix, additional research quantifying the characteristics and extent of edge forest are of value. In its original context edge forest was viewed as beneficial, primarily with respect to wildlife management (Leopold 1933). Attainment of maximum animal species diversity in a given area could be acconplished by increasing habitat diversity. Since edge forest provides a transitional habitat between interior forest and the surrounding landscape, fragmentation of a large forest into several smaller fragments would increase she habitat at the landscape level, and thereby increase species diversity. This has been denonstrated empirically for bird and butterny caammities (Karr 1968; Gates & Gysel 1978; Strelloe & Dickson 1980; Lovejoy _e_t_ _a_l. 1986). The use of mmerous small terrestrial islands, with their high proportion of edge habitat, has been proposed as a possible management 3 approach for maintaining high species diversity (Simberloff & Abele 1976). Unfortunately, this approach is biased towards edge species, i.e. those species that are either not characteristic of, or as abundant in, the original habitat type (Diamond _e_t_ 11. 1976). Further, the desirability of high species diversity must be questioned when applied to mature forests of eastern North America. Many of these ecosystems typically have low tree diversity (Levenson 1981; Middleton & Merrium 1985). If the optimal goal of a nanagement strategy is the preservation of whole ecosystems characteristic of the presettlanent landscape, edge habitat must be minimized. This requires recognition and understanding of the effects of edge. The present study examined ecbe-to-interior gradients in the tree and shrub strata of four small fragments of tenperate deciduous forest located in the agricultural landscape of central lower Michigan. Forest Islands and Edge The analogy of isolated terrestrial ecosystem to oceanic islands has resulted in numerous attenpts to explain their patterns of species diversity in terms of island biogeographic theory (MacArtl'mr & Wilson 1967; Miller 8 Harris 1977; Lovejoy g; 9;. 1983, 1986; Harris 1984; Wilcove gt _a_l. 1986). Although direct application of species-area relationships to the vegetation of forest fragments appears invalid (Helliwell 1976; Levetmon 1981; Middleton & Mirriam 1983; Forman & Godron 1981), theories relating species diversity to a canplex function imorporating habitat diversity, disturbance, area, age, heterogeneity of the surrounding landscape matrix, isolation and 4 discreteness of boundaries seen more tenable (Forman & Godron 1981). Fragmentation and continued reduction in forest island size is accompanied by an increase in perimeter/area ratio, increasing the amount of forest ecbe. This favors more shade intolerant plant species Whitney & Somerlot 1985). Leveison (1981) found that fragments of the southern—mesic forest in Wiscorein were subject to retrogressive succeesion below 2.3 ha in area, becaning essentially all-edge forest, composed of intolerant and residual tolerant species. Ranney _e_t _a_l_. (1981), also working in southern-mesic fragments in Wisconsin, found clear distinctions in composition between edge and interior forests. The edge forest, defined as successional ly younger, was characterized by more intolerant species relative to the more closed, light-limited interior forest. These findings point to the negative effects of size reduction, with its corresponding increase in edge effects, on the characteristic vegetation of a particular forest type. Further, the necessity of determining the minimum critical size of an ecosysten required to assure its dynamic equilibrium, or continued successional developnent, becomes apparent (Lovejoy & Oren 1981). Two important components of forest fragmentation, area reduction and isolation, may combine to facilitate edge forest developmnt (Wilcove gt §_l_. 1986). As interfragment distance increases, the pattern of seed input into a fragment may be altered (Curtis 1956). Wind- and manual-dispersed species becane increasingly disadvantaged canpared to bird-dispersed species, because of the relatively low mobility of these dispersal mechanisms (Auclair & Cottam 1971). This suggests that diaspore input into isolated forest fragments may be disproportionately biased towards edge species, a high percentage of which are bird—dispersed (Ranney 1978). Fragment size reductian, with a corresponding increase in edge habitat, assures the availability of safe sites for the bird-dispersed ems species. Conversely, if characteristic interior species are not bird-dispersed their establishment in earlier successional forests may be limited or inpossible given large interfragment distances and a minimal availability of interior safe sites (Levenson 1980). There are two important exceptions to the edge species-bird- dispersed species scenario not addressed by the above authors. Propagules of American beech (Fagus grandifolia Ehrh.1), an important interior species in beech—maple forests (Ranney §_t_ ;a__l_. 1981), are dispersed up to 4 km by bluejays (Johnson 8: Adkisson 1985). Acorns of sane species of oak (ggrcus spp.) are dispersed up to 1.9 km by bluejays (Darley-Hill 6: Johnson 1981). The classification of oaks as edge or interior species is dependent on the forest type in question, e.g. in a beech-maple forest, red oak (Q. _r'_u_br§_ L.) might be considered an edge species (Wales 1972; Ranney 1978), however in mesic oak sites the species is important in the interior forest. These exceptions suggest that, at least for sane species, isolation by itself may not present a major barrier to dispersal. 1 Nanenclature follows Gleason & Cronquist 1963. 6 The Microclimatic Gradient mEffects Although all forests are strongly influenced by the surrounding climatic reginne, the impact is greater on isolated forest fragments as a consequence of their increased perimeter/area ratios (Ranney gt _a_l_. 1981). There is imreased penetration of light into the edge forest relative to the interior resulting fran a large lateral exposure (Hutchirmon & Matt 1976a, 1977). The edge forest may this be subjected to greater net radiation or higher tenperature than the interior, two conditions which in general can increase evapotranspiration (Salisbury & Ross 1969; Rosenberg gt; _a_.]._. 1983). Wind can also increase evapotranspiration (Rosenberg gt .a_.'_1_. 1983), although the laering of air and soil tenperature by wind my sonewtnat dampen the increase (Geiger 1965). Differing wind velocities between edge and interior forest may thus contribute to microclimtic differences. These effects may codeine to establish a couple: microclimatic gradient (Mnittaker 1975) fran outside the fragment to its interior. Wales (1972) and Barney g_t_ a_l. (1981) state that since different species respond characteristically to different microclimtes, a corresponding vegetational gradient my develop. The applicability of gradient analysis to such studies led mittaker to describe the edge-interior forest transition as a microecoclirne (Wales 1972) . The microclimatic gradient my involve both varying light arnd moisture reginnes fran the edge to the interior, such that edge forest might be categorized as beirng less light-limited and more xeric relative to interior forest. The gradient develops as a consequence of decreasing exposure to wind and liglnt, from eche-to-interior, rather than topographic or soil differences within the fragment. A furtlner consequence of the large lateral exposure of the edge forest is the mecharnical effects of the wind. Edge tree crown geannetry is more characteristic of open gram trees than of streamlined, sleder interior trees. The edge trees are buffeted by wind which my cause imreased incidence of branch breakage (Moen 1979). Branch breakage my result in decreased vigor and increased mortality. The potential for catastrophic wirdthrow and windsnap is increased at the forest edge (DeWalle 1983; Franklin & Fonnan 1987). In Amazonian forest fragments of Brazil an increased imidence of tree fall at the forest edge has been reported (Lovejoy g: §_l_. 1986). Even small gaps at that forest edge my provide increased habitat for opportunistic species, above that already provided by the edge condition. Species which are best able to adapt to the increased disturbance regime of the edge my be more inportant in the ecbe forest . Mt Effects The amount of solar radiation and wind to which an edge is emed varies with aspect (Wales 1972). The non-uniformity of climtic corditiorns on the different aspects of a forest fragment suggests that the mgnitude of the proposed microclimtic gradient from edge-to— interior my also vary with aspect. 8 Past studies have demonstrated that vegetational canpcsiticn varies on mrth and south slopes within a forest (cantlon 1953; Zager & Pippen 1977). Although these researchers were not examining edge effects, clearly the arnalcgy of slope to aspect can be made. MiricalgEtvidence for the Microclimtic Gradient Although the existence of a microclimtic gradient my be intuitively obvious, studies which specifically enamined microclimtic variation along edge-tc-interior forest gradients have been limited. The data however, do support the etistence of microclimtic differences between edge and interior forest. Wales (1967) examined several microclimtic variables cnn the mrth and south borders, and in the interior, of an oak-hickory forest in New Jersey. He found that the greatest accumulations of snow occurred within the first 25 meters of the forest on the mrthern aspect. Further, srnow melt was most rapid in the southern border forest. This my translate into seasonal variations in soil moisture and tenperature reginnes between different areas of a forest. Both maximum and minimumair tenperatures, in a profile 5 cm to 2 m above the forest floor, were highest at the boundary of the soutrern aspect forest. The difference was as great as 14° C in the stunner. Light penetraticnn, as percent of full sunlight, generally decreased fran the edge to 25 meters into the forest in the winter and spring leafless conditions. The pattern in the scanner, subsequent to full leaf flush, was not denonstrative of a clear edge-tc-intericr gradient. Haever, in a similar study, August light intesities on the mrthern aspect of a shortleaf pinne (Firms echinata Mill.) stand were found to decrease 9 6596 between the stand edge and 30 m into the forest (Ocsting & Kramer 1946) . In the same shortleaf pinne stand, available surface soil moisture (to 15andepth) imreasedbetweenthestandedgeandpoints4mand 15 m into the forest on the mrthern aspect, although the differences between tlne measurement points were not significant (Ocsting & Kramer 1946). This my reflect the more mesic caditicnn of the mrthern aspect forest, whereas in the forest on the southern aspect, a steeper moisture gradient would be expected. Studies examining wind profiles in forests lnave sham that wind velocities my be reduced 5—15% at the the forest boundary (Kittredge 1948; Hutchiscn 8 Matt 1976b). Reifsnnyder (1955) found reductions of 30—50% at 50 m into the forest (depending on tenperature conditions, height above the forest floor and original wind velocity), in a enall ponderosa pine stand (Pinus pordercsa Dougl.). Although not directly measured, an edge—to-interior gradient is implied in these studies. In the Amazonian forest fragments referred to previously, Lovejoy g3; _a_l. (1986) report increases in relative humidity of 5-20% and decreases in air tennperature of 4-5° C between boundary forest and interior forest. The Vegetaticnal Gradient Studies focusing on the edge—to-interior vegetational gradient that develop in response to the microclimtic gradient in forest fragments have been limited in number. (Neel (1951) studied edge— interior distinctions in several sugar mple—beech fragments in 1 0 Michigan. While both mrth and south aspects were eemined 'm distinction between them was made in the analysis. He reported that edge conditions extended 2—12 m into the forest, and were characterized by a higher sapling desity, relative to the interior. Additionally, there were numerous compositional distinctions between edge and interior forest, such as the restriction of hawtlnorn (Crataggu_s spp.), yellow—poplar (QiCricdendron tulipifera L.) and sassafras (Sassafras albidum (Nutt.) Nees) to the edge forest, and American beech to interior forest. Manny species had a higher frequency of occurrence in the edge forest including: bitternut hickory (Carya cordimmis Wang.) K Koch); pigrmt hickory (9. 9.1.339 (Mill.) Sweet); white ash (Fraxinus annericana L.); black cherry (Prunms serotirna Ehrh.); red oak; and basswood (Tilia americana L). Gysel found that edge-interior compositional differences were also evident in the shrub oonnnunity. Gooseberry (Ribes cyncsbati L.), staghorn sumac (Ring M L.), and three species of m were restricted to edge forest, while the converse was true of witch-hazel (Kamamelis Virginians L.) and mpleleaf viburmnm (Viburmg aoerifolium L.) . Gysel described the open growth crown geonetry of edge trees, some of which had low-growing horizontal limbs extedinng 10 meters in thedirecticnoftheforest boundary. Theincreasedcrmngrowth response of trees bordering forest openings has been examined in detail by subsequent researchers (Trimble & Tryun 1966; Runnkle & Yetter 1987). Finally, similarities in composition between edge forest and recovering gaps were mted. This my reflect a parallel in rmouroe availability in the two situations. 11 Wales (1972), in a study that canpared mrth edge, south edge and interior forest in a mture oak-hickory stand, found that edge forest extended approximtely 20 m into the fragment on the southern aspect, and 10 m cnn the mrthern aspect. Like Gysel (1951), he mted a similarity between edge forest and recovering gaps in horizonntal and vertical structure. Total tree desity, and densities and basal areas of many individual canopy species were highest at or near the edge. These included species characterized as shade intolerant or those capable of vegetative reproducticnn, such as white ash, sassafras, black oak (giercus velutinna Lam.), black cherry and shagbark hickory (Carya ovata (Mill.) K. Koch). The latter two species were not found as large individuals in the interior forest. In addition to the differences in extent of edge forest between aspects, other aspect distinctions included: a greater abundance of less tolerant species on the southern aspect relative to the mrthern aspect; more than three times higher desities of oak seedlings and saplings in the southern edge forest than on the mrthern aspect, or in the interior forest; and higher cover values for blackberry (m al legheniensis Porter) and Japanese honeysuckle (Lonicera 1929115523; Thunb.) in the first 15 m of forest on the southern aspect. Rannney e_t g. (1981) found that in beech-maple forest frments of southeastern Wisconsin, edge forest enteded approximtely 10 m into the interior on the mrthern and eastern aspects, and 20 m on the southern and western aspects. The edge forest supported up to 5096 more basal area than the interior and had higher sten desities. Although these findings do indicate structural variaticnn along the gradient, they should not be misconstrued as suggesting superior site 12 quality of the edge forest because no canpariscn of tree heights, volumes or bimes differences between edge and interior was given. Cannmnnity index values (Curtis 1959) were lower in the edge forest relative to the adjacent interior. This was the result of an edge-to—interior gradient of decreasing importance values for intermediate and intolerant species such as basswood, smooth serviceberry (Amelanchier laevis Wieg.), red oak, white oak (germs 11$ L.), bur oak (Q. macrocarpa Michx.), roundleaf dogwood (m m Lam), hawthorn spp. (Crataggg), hickory spp. (m), willow spp. (_S_a;l_i_.x_), trembling aspen (Mllus trenuloides Michx.), bonelder (Acer m L.), slippery elm (Ulmus rubra Muhl.) and ash spp. (Fraxinus), and increasing importannce values for very tolerant sugar mple and beech and intermediate red mple (Acer rubrum L.). Although they provided nno direct evidence Ranney _e_t_ 11. (1981) stated that edge forest on the southern and western aspects provided a permanent xeric habitat as a result of greater exposure to drying microclimtic conditions, relative to the mrthern and eastern aspects, in an otherwise mesic fragnnennt. They concluded that this condition allowed species of more xeric habitats, such as hawthorn, bur oak, hickory and prickly-ash (_Zgn_thmcylum gmericanum Mill.) to enist in the fragment. This distribitian, rather than being in response to soil moisture differences, my be more of a reflection of the shade intolerarnce of these species, thus they were more abundant in edge forest because it is less light-limited than interior forest. The results of the above study, which are often cited in the literature, and those of Gysel (1951), are potentially misleading in regards to the extent of edge forest penetration. Values reported 13 were averages derived from numerous fragnnents, sites which the autlnors stated were relatively hanogeneous in terms of managenent history and present conditionn. Personal observaticnn indicates that in an urban- agricultural mtrix it is unlikely that this is true, thus warranting an individualistic approach to edge analysis. Examining the influence of different fragment histories and conditions on edge forest characteristics may be requisite for determining how variable is ewe forest across a landscape. Whitney and Runkle (1981) examined edge-interior forest distinncticns in an old-growth beech-mple forest in Ohio. They found that species canposition in the first 10 m of edge forest an the western aspect of the old-growth forest had greater similarity to the interior of an adjacent second—growth forest than to the interior of the old growth forest itself. Coupositicnal characteristics of edge forest included higher densities and basal areas of red oak, shagbark hickory, blue-beech (@inus carolLiniana Falt.), basswood, white ash and ironwood (m vigginiana (Mill.) K. Koch), relative to the interior forest. The canverse was true for sugar maple and beech. No attenpt was made to examine aspect differences. None of the above studies sampled beyond 30 meters from the forest boundary, and although Gysel (1951), Wales (1972) and Phitney & mmkle (1981) did ccnpare edge plots with those of the interior, more extensive linear sampling along the edge-to—interior gradient is necessary to firmly establish the extent of edge conditions in specific forest frwnents. 14 Fragment Disturbance and Edge History An examination of an edge—to-interior gradient must consider the effects of disturbannce, both natural and anthropogenic, on the conposition and structure of the vegetation. Wales (1972) described treefall gaps and patl's as "noise" with the potential to mask edge effects in gradient analysis. Donnel ly (1986) found that black cherry and yellow-poplar (Liriodendron tulipifeg L.) were associated with areas of more open canopy, such as steep slopes and poorly-drained depressions, in a beech-sugar maple forest in southern Michigan. Tie pm of these internal edges may account for edge species in the interior forest, particularly in fragments below a minimum critical size where nno area is very far from edge forest, and consequently the seed rain into the interior may be dominated by edge species (Mmey _e_g 3;. 1981). The management history of a fragment or its edge may influence the canpositionn and structure of tie gradient (Gysel 1951; Wales 1972). For example, grazing and selective tree renoval, by reducing canpetitionn and creating a more open habitat in a fragment large enough to normal ly support interior forest, provide conditions favorable to opportunistic species and may pronote the developnent of edge—like conditions throughout (Auclair & Gotten 1971), thus obscuring any gradient of forest structure and couposition. “ether or not interior conditions can redevelop may depend on the magnitude of tie disturbance and the size of tie fragment. In newly created edge that exposes interior forest to edge conditions for the first 1 5 tine, it may take several decades for edge species to becane established and structural distinnctions to develop. Heterogeneity of the Landscape Matrix In addition to variability induced by individual site history, the characteristics of the surrounding landscape matrix may influence edge structure and conposition (Foruen and Godron 1986). In an agricultural landscape, pesticide and herbicide use, artificial tile drainage, and irrigation practices may influence the developnent and persistence of forest edge in a site specific manner. Natural features of the surrounding matrix inclnding soil structural and drainage characteristics (i.e., rocky or poorly drained soils) and topographic features may also influence edge structure and canposition. Corridors and windbreaks adjacent to a fragnent may alter tl'e biotic and abiotic influences acting on a site, relative to an isolated fragnnent without trese structures (Forman and Godron 1986). Heterogeneity of the landscape matrix may therefore introduce a variable, additional to site specific disturbance and manageuent history, with tie potential to influence edge and fragment characteristics . Research Hypotheses and Objectives This research examined edge-interior forest differences in four forest fragments in central loner Michigan. The fragments differed in forest type and management history. The couposition and structure of 1 6 the tree and shrub strata along edge-to-interior gradients on north and south aspects were analyzed. Soil misture and light penetration were examined on the nnorth and south aspects in one of the fragments. Hypotheses 1. There exists a vegetational gradient from edge-to-interior that develops in response to an assumed microclimatic gradient. At the entrees of this gradient, separate edge and interior forest cannnunities will be distinguishable canpositionally and structurally. The edge forest will be distinguished by a greater abundance of less tolerant or disturbance-oriented species relative to the interior . 2. Given the relatively slow growth rates and long life spans of many dominant canopy species, it is probable that fragments of second- growth forest are of an insufficient age for the canposition of the canopy to reflect an edge-interior forest distinction (Hill 1985; Whitney & Sanerlot 1985). Tlerefore, any tred will be most evident in subcanopy strata that established subsequent to formation of the present fragment boundaries. 3. Tte edge forest on the northern aspect of a fragment will ted to be narrower, and of different canposition and structure, relative to tie edge forest on the soutl‘ern aspect. For this study the assunptian was made that in tenperate forests of eastern North America, eastern and western aspects are internediate to northern and southern aspects with respect to edge conditions. 4. 17 There is a minimum critical size of fragment below which interior conditions will fail to be maintained, resulting in an al l-edge forest . 5. The disturbance regine or managenent history to which a fragnent or its edge has been exposed will influence tne canposition, structure and entent of the edge-to—interior gradient. Specific research objectives include: 1) 2) 3) 4) 5) 6) 7) 8) tie quantification of soil moisture and light levels along the edge-to-interior in a forest fragment; the quantification of the diversity and distribution of woody species relative to the edge-to-interior gradient; the quantitative description of caumnnity structure frann tre edge to the interior; tl'e quantitative determination of the entent of edge forest; a canparison of tl'e edge-to—interior gradients on northern and southern aspects; a determination of the effect of fragment size on gradient coupositionn, structure and extent; an examination of tie effects of fragment disturbance and edge managenent history on gradient composition, structure and entent; ccmparisonn of edge characteristics in beech-sngar maple and oak forest frmnnents . METHODS Site Selection This research utilized four previously studied forest fragments, two each of oak and beech-sugar maple, located in Ingham and Clinton counties in central loner Michigan (Figure 1; APPENDIX A). Dodge (1984) classified these fragnents to forest type and determined that they occurred on sites of similar climate, topography, geomorpi'slogy, soil, and substratum. The adjacent matrix was agricultural for all sites, consisting of eitler pasture, corn, or fallen fields (see Appendix B for site diagrams). The potential influence of netrix structure and managenent history, as well as that of adjacent non— agricultural vegetation, on edge characteristics was nnot addressed by this research. Criteria pertinent to stand selection required that: 1) all vegetational strata were present, i.e. ground layer, shrub layer, understory, and canopy; 2) tie fragment was a remnant of original vegetationn, nnot a newly established stand, or one with newly establisl'ed edges; and 3) fragments had linear edges facing nnorth and south (+/- 10 degrees). Individual sites were chosen for this study specifically for the particular managenent history and present conditionn they represented. 18 19 Shiawassee Clinton mi. 0 5 L 1 O 5 10 km. /“/ ”A (El “,flw fl T x19: '7: 1w ‘ m‘ / -‘ J‘ —— I , I kn; c——_—.—-s . lng ham Figure 1. ( 1984) . Location of study area and sites. Adapted fron Dodge ' ‘11)) 20 Sites 1 and 2 were located on the canpus of Michigan state University. Site 1, Tourney Forest, was an old-gmth undisturbed beech-sugar naple forest (Schneider 1963). Actual area of the forest was 6.07 ha tut only the upland 4.68 ha was studied (APPENDIX B). Linear dimensions of tie sampled forest were 195 m (north—south) x 240 an (east-west). Both the western and eastern aspects of tl'e entire forest were bounded by planted mixed conifers. Site 2, Clever Woodland, was a second-growth beech-sngar naple fragment approximately 2.7 ha in size. Linear dinensions were 150 m (north-south) x 180 an (east-west). Tte site was fornerly grazed by cattle (L. Kramer, former M.S.U. agricultural land manager, personal communication). Age determination, based on growth rings, of individuals of an even-sized subcannopy of sngar maple saplings indicated that grazing was discontinued 18-20 years ago. A fenced off northwest extesion of tie site was actively grazed at the time of this study (APPENDIX 8). Interviews with past and present M.S.U. farm managers revealed that selective tree renoval for firewood from the edge forest was a cannon practice in the early to mid part of this century in this site. There was nno evidence of recent cutting in the forest. Site 3 was a 3.6 ha second-growth oak forest located within the Michigan Department of Natural Resources Rose Lake Wildlife Research Station. Linear dimensions were 240 M (north-south) x 150 m (east- west). This site was released fron steep grazing approximately 45 years ago (S. Me, Dept. of Gepgraphy, Georgia State Univ., personal canmunicaticn). 'n'ere apparently was a limited amount of past tree removal as evidenced by a enall number of deconnposing stunps. The western aspect of tie site was bounded by a hiking path, followed by a 21 10 m-wide strip of ponderosa pine (APPENDIX B). Site 4 was a 1.5 ha second-growth oak forest located approximately 18 km east of East Lansing, Michigan. Linear dimensions were 150 m (mrth-south) x 100 an (east-west). Trere was no evidence of any recent artificial disturbance in tl'e site. However, tnere was a large (20 x 20 m) patch of bracken fern (Pteridium aguilinnmn (L.) Kuhn), conntaining little woody vegetationn, in tle mrth central portionn of the forest, which my suggest a past fire. The site had a strip of forest vegetation approximately 35 m-wide, exteding 100 m eastward from tne soutreast corner of the fragment (APPENDIX 8). Sampling Methods Transect Location In each site a series of three to five, 10 m—wide transects was established on us mrthern and southern aspects. Transects ran 50 m into the forest in a mrth-south directionn. A transect began at the outer boundary of shrub vegetation, tte pre-mantel of Wilmanns & Brunhool (1982; Figure 2). In all sites, transects began interior to tie campy drip line (Bruner 1977). This reflects tl'e managanent history of the edges, in which there has been no recent advancement of vegetationn into tne surrounding matrix. Tlenumberof transectsperaspect,whichdepededonlinearenge legth, numbered five in Sites 1-3 and three in Site 4. The location of tre first transect in each site was determined randomly by choosing anrunber that corresponded to apoint on tl'e east-west connpass line. 22 1-—-10 m——-1 1’3 m“! 1 m DETAIISOFASMPIEPIDT (mttoscale) 1—5 m—n—l :m :_p In: “109 Figure 2. Transect location and sampling design for a hypothetical forest fragment. Trees (> 10 cm dbh) and saplings (2.5 cm 5 dbh g 10 cm) were sampled in the 10 x 5 m plots, shrubs and small trees (<2.5 cm ‘dbln and taller than 1 m) were sampled in the 5 x 3 m subplots, the woody ground layer (< 1 m tall) was sampled in the 1 x 1 m subplots. Three to five 50 m transects were sampled on both the mrthern and southern aspects of a site. Transects were located at least 40 m. interior of the western and eastern edges. Three to five plots located 50 m interior of all edges were also sampled in each site. 23 The location of this point, which was restricted to a position 40 an inside of tre eastern and western forest boundaries, located the transect centerline. Subsequent transect centerlines were placed 20 m apart in whictever direction provided adequate roan. This procedure was fol lowed in Sites 1-3, but was not possible in Site 4, becauseof its enall size. Inthissitethethree 10mx50m transects were placed inlnediately adjacent to one annother, with only a 35 m buffer on us eastern and western aspects. This design violated tie procedure of random sanpling, and may have concurrently introduced edge effects fron tie eastern and western aspects into tl'e sampling. However, this was tne only way to adequately euple a potentially interesting fragment . Vggetat ion Tl'e vegetation was sampled using a series of nested plots to neasure woody ground layer, shrub, understory and campy strata. Each 10 x 50 m transect was divided into ten 5 x 10 m plots, referred to as transect positions (e.g. 0-5, 5-10 45-50) in the Results and Discussion sections. Trees (> 10 cm dbh), and saplings (2.5 cm 3 dbh 5 10 cm), were sampled in the entire 5 x 10 m plot. Shrubs2 (< 2.5 cm dun, > 1 m) were sampled in 3 x 5 m subplots located on either end of tie larger plot. The woody ground layer (< 1 m tall) was sampled in four 1 x 1 msubplots located ineachcorner ofeach10x5mplot (Figure 2). In each plot, species, sten densities of all strata, and 2The shrub size class innclnded individuals of tree species that were of tre specified size, in addition to true shrubs. 24 diameter at breast neight (1.4 m) of trees were recorded. Tl'e interior forest of each site was sampled using 10 x 5 m plots like those described above. These plots were located randomly in an area 50 m or more fron all forest boundaries. The nunber of interior plots corresponded to tne number of transects in that site. These plots are referred to as tte "interior" in the subsequent Results and Discussion sections. The size of each fragment was determined fron U. S. Geological Survey maps. Soil Moisture and Light Measurenents of soil moisture and light were made in Site 2. Gradients on both the mrthern and soutl'ern aspects were ecamined. To determine soil moisture, three 1.5 cm-wide soil cores (to 12 cm depth) were removed fron the middle and end points of a 1 m line located parallel to the long axis in the center of each 5 x 10 m plot on four transects. Additionally, samples were collected fron an area immediately outside of the forest boundary at tie beginning of each transect and in four interior plots. The three cores from each plot were conposited and stored in plastic bags for transport to tke laboratory. Determinations of percent moisture by weight were made gravinetrical ly fol lowing Brower & Zar (1977). Soil moisture was measured on three different dates: 30 April, 1987; 31 July, 1987; and 7 Septenber, 1987. Light penetration, as a percentage of full sunnlight, was measured using a Li-cor quantum photometer under calm, clondless conditions. Measurenents were recorded along three of tie establisted transects at 25 5 m intervals, beginning just inside the forest boundary (neter zero) to the 50 m point in tie forest, for a total of 11 neasurement points per transect. Measurenents were taken at each point on the three transects by walking the legth of the first transect in one direction, us next ajacent transect in tne opposite direction, etc., repeating this procedure until 12 ueasurements had been recorded for each point. Measurements were taken 10 cm above the forest floor, avoiding any obvious light gaps and sun flecks. This technique gives an approximate measure of diffuse light in the forest (Wales 1967). Tl'e procedure, which took approximately one-half hour per aspect, was initiated at 11:00 hrs on 20 Angust, 1987 and 10:30 hrs. on 4 September, 1987, for the southern and mrtrern aspects, respectively. Data Analysis Vegetation Direct gradient annalysis was used to examine patterns in: 1) total density and tree basal area; 2) individual species desity or abundance (number of individuals in the total sampled area at each transect position); 3) species richness (total number of species); 4) individual species distritntios; and 5) synthetic inportance values (Curtis 8: McIntosh 1951) of potential canopy dominant species in each stratum, a a function of transect position (i.e. distannce fronn tte edge) and aspect. Inportance values were suns of relative desity, relative frequency and relative basal area for trees, and relative desity and relative frequency for subcannopy strata. 26 when requirenents of parametric statistics could be met, ANOVA or t-tests were used to test transect position and aspect effects on tre various paraneters enamineda. Unless stated otherwise all data sets analyzed nsing paranetric statistics were square root-transformed in order to nest tl'e assumptions of mrnality and homogeneity of variannces. Graprs and tables report mntransfornned data. Individual transect position differences were examined in as of two ways: 1) orthogonal contrasts of a specific position nean relative to the nean of all positions interior to it conbined on that aspect (excluding the interior position); 2) Dunnett's test for pair-wise comparison with a control (in this case tl'e interior uean) when the interior position uean was substantially different from tte majority of positions on one or both aspects. Manny of tl'e individual species data sets did not meet the assumptions of parametric statistics, even with transformations, because of an encessive number of zeros in tre data (i.e., >2096; J. Gill, Dept. of Annimal Science, Michigan State Univ., personal communication). Further, mn—parametric ANOVA with multiple comparisos could not be used to examine position effects because sample sizes were too small (Soloal En Roth 1981). In true cases, Walsh's approximation, a test of nears robust against non—mrmal data with keterogeneous variances (Gill 1978) was used to examine aspect effects alone. 3 Statistical methods follow Gill (1973). 27 Soil Moisture and Light Soil moisture differences between aspects were analyzed by comparing mans derived by using the 10 position mans on each aspect. Specific couparisos of transect positios against the interior position man were made using Dunnet‘s pair—wise couparison with a control (i.e. tre interior position). Log-transformd light masurements were analyzed using one way ANOVA, separately for each aspect. Specific position differences were analyzed using Bartholomew's test of ordered alternatives (Bartholomew 1961). Since the tine and date of masureuents on tie two aspects were different, no direct statistical conparisonn was atteupted. Regression equations and coefficient were derived for tie light measurements, as a funnction of transect position. Determination of Edge Depth The criterion used to determine tl'e depth of edge forest peetration on tie erm and southern aspects of each site was the location of transitional events alo'g tte edge-to-interior gradient. A transitional event was defined as a relatively abrupt change in a measured attribute within a 5 an interval along tl'e gradient. Attributes inclnded: total sten desities and basal areas (of trees and saplings) in each stratum; individual species distribution or abundance in each stratnmn; species richness in each stratum; importance percentages of potential cannopy daninant species in each stratum; and in Site 2, soil misture and light values. The 5 m interval with tie greatest accumulation of transitional events, 28 relative to the remainder of the gradient, was used to designate the depth of edge forest peetration. A transitional event was attributed to the 45-50 m position if tlne value of tie measured feature was higier of lower in the interior, relative to tie final gradient position. Characterization of a species as an indicator of edge or interior forest conditions was made _a_ Eteriori based on observed distribution and/or abundance patterns. Tie characterization was substantiated by examining ecological and life history traits of the species (e.g. does it occur in disturbed habitats; is it shade tolerant).and by cannparison with characterizations fronn Gysel (1951), Wales (1972), Ranney (1978), Rannrey g; g. (1981), and Whitney and Runkle (1981). Tre distribution and ahndance of edge indicator species in tie interior helped to determine if edge conditions peetrated throughout tte sampled area of a fragment. RBULTS Soil Moisture and Light Soil Moisture The soil moisture profile for April in Site 2 is illustrated in Figure 3. Percent moisture by weight was significantly greater on tie nortrern aspect than tie soutlern aspect (P <.05, ANOVA; Table 1). This relationship also teld without tl'e inclusion of the outside position in tlne ANOVA. Moisture differences along the edge to interior gradients on both aspects were minimal. Only the southern exterior position and position 5-10 m on tie souttern aspect were significantly lower than the interior man (Table 2). Tie July moisture profile depicts both overall lower soil moisture relative to April (Table 1), and a stroger position effect (Figure 3). On tie souttern aspect, percent moisture was significantly lower than tie interior in the positions between the ecge and 25 111 into tlne forest (Table 2). On the northern aspect percent moisture was significantly lower than the interior in tte first 15 m of tie forest (Table 2). Aspect differences were not significant (Table 1). The September soil moisture profile (Figure 3) depicted an overall increase in moisture relative to July, bit not quite back up to April levels (Table 1). Total moisture was significantly higher on the northern aspect, both 29 30 .nua “mm ..\+ name... was 851.5 ovata man. .096 may no £3.38 a a n .50 ammo? :m Eon“ E on we 35: .3235. one .mHmSumucfi. 855% s. at. no 8.22 among mfi mad 839 man; 9s. .u 88 E nuomamm Egusom can Fuentes Bu :0 a on 9. meow 93 .53, Eamon s8 2 3 3:93 am: so E839 82qu Emacs :8 .n magmas 31 APRIL SOUTH NORTH é///////////////¢///V/Aw////ar/fihflnr 22 0 no 1‘0 no" a T , T U U H 0559111155. 0 a o n m rolalfllllllllllllillllll . U noflat/lIll/III’llIII/III‘I/I’IIII/Illl. w m knob/1111115551515 H. w rno......Illllll/I/t/I/I/I/t/I/II. , .oaova.-I/I’I/I’I/Illllllllll . flawlpflctylllllllllllllll/I. w w .. 6.111511115515455. ”loo/fllflflllllllllllllll. Uo/iIIIIII/I/IIIII/II‘II. 5 5 4 4 T. , T. m m 5 5 4 4 o Immmammmnmmnmmnmnmnnmnan o 3 3 n g E g u u ullmmmnnmnmnmmummm m e v n L m g P H u a :mmmmnnnamnummnn e m ... J o n 8 o u N U o - o N L O C 2 C 4 o O O C 2 l ‘ 2 I i 2 2 1 I ...zuomma NEH-PW—O: J-OW 45 so 15 oun‘ III?‘1EI‘£3 IF"(’II 'T‘il! IEI"IIE INT. 45 30 OUT 32 Table 1. Soil moisture (to 12 cm depth) by aspect and date in Site 2. Aspect North South April, 1987 17.72 (.41)* 16.60 (.38) July, 1987 11.72 (.56) 10.47 (.54) September, 1987 15.84 (.42)*** 13.29 (.50) Each value is a mean (percent wet weight) of 10 measurements (at 5 m intervals from the edge to 50 m) on each aspect +/- (se) . * P <.05, *** P <.001 indicates a significantly greater moisture content than the corresponding aspect. 33. .39: .3335” 93 easy page aaucwoaasofim mmz monoumao peep um name any umnu mmuMOwoca so.v m «as .mo.v m «« .ofi.v m * doom passages“ one. no .3233 .335 mason III. III III III III III III III III Ill ** SHSOM hmmfi III l|l Ill ill III III III III III III III. fiuhflz .Ummm III III III u III ... ... a. ... a. a. * ... ... ... a. * ... ... EH58 Roma -1: In- In- nu- --- In- In- a. nu. a. it nusoz sass III III III III III III III III ** III **e nufiflm Emma In- us: nun 1:: in: sun :1: nu: --- nu- us: nuaoz assoc omuma manoa oa-mm mmuom omumm mmuom calms wanes cans muo m.uxm somomd mass may scam mamumz .w 86 as Ammo? 3m sous a one uozmfisw may Ge 93 muowomm Emzuoom dam 593.3: may so .6 onto. 3:33pm honour—w ou mono may 93am modems, wgmwos :om c853 mcomwhmnmsbo mmwsahwmo 93938 .~ manna 34 with (P <.001, ANOVA), and without tl'e innclusion of tie outside position. Position effects were negligible, with significantly lower moisture only in tlne exterior position on the soutlern aspect (Table 2). Light Northern and southern aspect light gradients are illustrated in Figure 4. Tie decrease in diffuse light from edge-to-interior was logarithmic, and was best described by tie fol lowing regression equations for tie nortrern and southern aspects respectively: log(Y*100)=2.25-0.55(logx+10) ; log(Y*100)=3. 12-2.0910g(x+10); wnere Y is percent of full sunlight and X is distannce from the edge in 5 a intervals.‘ The coefficients of determination for the two equations were .935 and .732 respectively (P <.001, f—test). Bartholomew's test of ordered alternatives was used to examine differences in percent light by position. Tie use of this test requires tnat: 1) tte differences of interest are quantitative, (e.g. percent of full sunlight); and 2) tie alternative hypotheses can theoretically be restricted to sequentially inncreasing or decreasing values, although in reality sane variability may be present, (e.g. a decrease in light levels from tre edge to the interior). In such 4Coding Y by multiplication with 100 avoids negative logs; coding X by the adiition of 10 avoids the log of zero at tne edge position. 10 35 I.IC3I11F PERCENT SOUTHERN A3PECT §¥§t\\\\\\\\\.‘ NORTHERN ASPECT §§§§SSSSSSSSSSSS SSSSS§§§§SSS§§ S§SSSS§§§§SSSSSS S§§SSSSS§§§§SSSSS§§ §§§§S§§§§SSS§§ SSSSSSSSSSSSSSSSSS SSSSSSSSSSS§§§§§§ §§S$$SSSSSSSSSSSSS§t" $§§$:CRSSSSSSSSSSSF§ SS§§§S§§§S§§§SS§§§SS§§§\‘ §SSSSSSES§§§§§SSSSSSSSSSSSSSSSSSSSSS§§§§§§§SSSSSSSSSSS§ 4m 1.0 V q ID V (V. a 1N3383d JLEIEDI'I EDGE 40 30 20 10 50 20 30 40 50 METERS FROM THE EDGE IO EDGE Light profiles (percent of full sunlight) fran the edge Figure 4. . to 50 m on the northern and soutl'ern aspects in Site 2. values are means +/- se; n=12. the two aspects. Tie y—axis Note tie difference in y-axis scale on 36 cases oe gains power by using Bartholomew's test (J. Gill, Departmnt of Animal Science, Michigan State Univ., personal couuunication). Tie test functios as a series of ANOVA's, in which the existence of significant position differences is determined in the first ANOVA, fol lowed by tl'e sequential exclusion of treatments with each successive conparison. Tie tests of significance of the ordered alternative couparisos al lows one to infer at wlat position into the forest light levels were nno loger significantly higher tnan the remainder of tie positions interior to it. On the southern aspect the analysis indicated trat there was a significant gradient of decreasing light levels fron tl'e edge to 20 m into tlre forest (Table 3). On tie nnorthern aspect, light levels were significantly greater in the first 10 m of the forest (Table 3). General Barge Structure Beech—mar Maple Fragments Site 1 (4.68 ha) canepy and sapling stem desities were highest at the edge on both aspects (Figure 5). Canopy desity decreased fron the edge to 50 m on both aspects but higher densities exteded farther into the forest on the southern aspect (Figure 5). Contrasts of individual transect position mans against the conbined man of the renainder of the aspect interior to that position showed significantly greater densities in the canopy at positios 0—5 m and 5-10 m north, and 0—5 m through 10-15 m south (P <.01, orthogonal contrasts on untransformed 37 Table 3. Bartholonew's test of ordered alternatives for light values along the one to interior gradients (0—50) on tl'e northern.and southern aspects in Site 2. Meters Fran the Edge Aspect Edges 10 15 20 25 30 35 40 45 50 South“ ’*******nsnsnsnsns- North****"nsnsnsnsnsnsns- ** P <.01 indicates that percent of full sunlight at tlat distannce was significantly greater than tie reminder of the distances interior to it. 38 .0": 3m 1}. 9335 33 35am». ovata one .393 Ham 3.3 a. on ma :2: .8235 m5. may 03 3512’ maxmux one .mamzmpcw muomumwo E n so muga homo: .H ouwm on 30on.... 59333 93 Eguoo 93 do a. on 8 m3 «5 code «.2303 38:3 swam .0 «names 39 MOD“ ”:5. 30¢...- ”GNP“: as on on o. o n n. as an av .sz. as an as n. n 000' 00000 0000 00000. 000N— OOOOVN 0000. smart.— 9.50:0 OOOONM OOOON a. on an as .sz. as an on n. n n n. on on nv .sz. as an as a. n , .mw .4. n. w .. mm o a can .600. can .ooou can IbDOfl 15.802 . GOO" con. CZ..—t(0 >302‘0 0001 can. BNVLOEH 83d SIMS 40 data). Mean sapling density in the forest interior was mrkedly lover tlan most of the positios alog eitler aspect; consequently, specific contrasts were made against the interior man. Desities in fourteen of the 20 positios were significantly greater than in tlne interior (Dunnett t-test; Table 4). Although total desity (tie man of tie 10 positions) was not significantly different between tte northern and soutlern aspects the canopy and shrub strata (Table 5), the Figure 5 does illustrate tl'e greater peetration of edge coditios on tie southern aspect. Canopy basal area showed nno significannt differences by aspect or distannce from He ecge (nnot illustrated). Shrub and ground layer densities showed no significant position treds; however, both strata did show strong differences by aspect (Figure 5). Total shrub density was higlest on tl'e southern aspect (P <.001, t—test on untransformed data), wlnile ground layer desity was highest on tlne northern aspect (P <.001, ANOVA; Table 5). Site 2 (2.70 ha) In geeral, density area treds for tie canopy and sapling strata in Site 2 paralleled those of Site 1. An important difference was the extent of penetration of structural differences into the forest. Density was highest at the edge on both aspects (Figure 6) but significantly greater desities were found only at position 0-5 m south in both strata (P <.01, orthogoal contrasts). canopy and sapling basal area showed no significany differences by distance from the edge or aspect (nnot illustrated). Shrub desity stewed nno aspect trend (Figure 6) but density was significantly higner at position 41 .535 nonsense 93 one? 35. .3333 >Hu3ouwcoem mm: HNDMQHCfl umnu um GMQE mgu Hazy MQUMOHQCH HOO.V m new .Ho.v m *w .mo.v m * was new a w e * new new Sudom «*e we we new we *** SHHDZ owing melov oelmm mmlom omnmo mNION ONImH mHIoH Calm mlo Homage mode 9a. sons 98%: ..n spam Cw Amoco...» Ham 3.3 E 03 .33.”.3ucw may one 3033 Emsusom and anonymous on» so As onlov muomanmno newsman». 0w moon may 9.3on 3333,30 swam ocancmm 533qu 3033800 mmazlpwmm 933:5an .e manna. 42 Table 5. Density per hectare by stratum on the northern and southern aspects (0—50 m) in Sites 1 through 4. Site Aspect Canopy Sapling Shrub Ground layer 1 north 376 (74) 800 (118) 6000 (698) 187100 (14986)** south 456 (86) 952 (233) 9707 (703)** 47850 (3275) 2 north 300 (47) 1564 (102) 5976 (687) 85250 (7841)** south 312 (94) 1556 (183) 6767 (522) 47800 (4863) 3 north 491 (45) 688 (70) 3480 (660) 41980 (3082) south 492 (53) 836 (80)* 7667 (1206)** 44450 (3847) 4 north 333 (54) 1093 (220) 10067 (1029)** 61500 (3759)** south 320 (81) 1593 (127)** 6045 (494) 41375 (2980) Each value is a mean of 10 density measurements along the edge to interior gradient (5 m distance intervals from 0 to 50 m) on an aspect +/- (se). * P <.05, ** P <.01 indicates significantly higher density than the corresponding aspect. ' 43 .0": “mm I\+ 0:00.: mum 8.91.5 mflfil> 05. .8000 Ham 5.3 E on ma A923 hawumucfi one mow who 909:3 mwxmfx any .mam>umuc« moomumao E n no mug: .809 .N wuwm ca 30ng 50580 paw 593.3: 65 co 6 on 3 6606 45 got 62386 38:8 63m .6 0963 44 n- “— 0N nN on mm m? n? .hZ. .hZ. fl! IW>(J Ola—Ola n1 an on an MOON NIP IOfll.flm—mhma n— n n n— 0N an 0000" 00000 00009 OOOON. 00000. 00000— nN n. n n op ON on ODO— OOON coon UZ—Jt‘fl 0001 0* n? .hz. .hZ. 0V 0' on nn nN “N n. 0 mp n >LOZ‘0 000” 0006 0009 OOON. Doon— 000 000 000 DON. 00m. SWIG!" 83d SIMS 45 0-5 m north (P <.01, orthogonal contrasts). Ground layer density pattern paralleled that of Site 1, with a significantly higher total density on the mrthern aspeCt (P <.001, t—test; Table 5) and a lack of a strong position trend (Figure 6). Oak 3mm Site 3 (3.60 ha) (hnopy and sapling densities in Site 3 M no significant position trends (Figure 7). However, total sapling density was significantly higher on the southern aspect (P <.05, ANOVA; Table 5). No significant aspect difference existed in the canopy stratum. Canopy basal area (not illustrated) has significantly higher at position 5-10 11) north (P <.001, orthogonal contrast) but no other position or aspect trends vere apparent. Shrub density showed strong position and aspect trends (Figure 7). Density at position 0-5 m north was significantly higher than positions in the remainder of the northern aspect (P <.01, orthogonal contrasts), as were all positions from 0 to 20 m on the southern aspect (P <.05, P (.10 for 10-15 m, orthogonal contrasts). Total density was significantly higher on the southern aspect (P <.001, ANNA; Table 5). Ground layer density showed no significant position or aspect trends but pcsitim densities were highest exterior to 40 m (Figure 7). Site 4 (1.5 ha) Canopydensitywasgreatestneartheedgembothaspects (Figure 8). Density at position 5-10 111 south was significantly 46 .nuc “mm (\+ mommy: who wooing manna. ma. .mmmow Haw 5.3 E on ma :2: .8335.” 05. 635.8qu 00530:. E n no mug: noon: 05 mum magma, .6..ng 05. .m 03m 5 30093 Fug—6m cam 935.8: 45 so a on B «mom 83 sob 82an Bags 63m .6 585E 47 “N. MOD” ”-9—. 30cm ”EMF”: an 9. .5. av on on a. n n n. on an ac .5. me an an a. n o 000v OOOON 0000 00001 0008' 00000 0000' In»: 350.5 00000 OOOON n a. nu fin n1 .hz. m! an an up n O 0 00" OD" 000 000 GOO GOO IHKOZ DON— GZ-Jfl(fl >503‘0 II (i)! ll con. coup auuaan 83d SISLS 48 .muc “mm (\+ mommE mum. gums. mg...» 05. .890 Ham 6.3 e on ma :2: .8235 05. 05 who game, mwxwlx use .mamggca 005336 E n «0 mafia: moan: .v 33 5.. muowamm Enough can 505.3: «5 8 e on 9. 6664 65 son. 828.6 flags 596 .o 086E 49 n— IhDOfl up a". an nN an n? a! .bZ. .bz. n' ¢W>‘.— DZQOCO 0* an an W00“ NIP :03“. ”CHE: nN n. n a up 0" an OOOON 0000' 00000 00000 1:62 380 . OOOON' 6N up n n up on an 000 000 p OO'N OONM 02—4“. 000' av. u* .3... .hz. a? a! an “N a— un ON a— >§°I¢o 000‘ 0000 OOON u 0000' OOOON 00" 000 000 DON. oom— HHVLOBH Hid SIMS 50 greater than the ranairder of the southern aspect (P <.01, orthogonal omtrast). No significant position or aspect trends in canopy basal area were evident (not illustrated). Sapling density showed no strong position trends (Figure 8) but total density was significantly higher on the southern aspect (P <.01, ANOVA; Table 5). Total shrub density vas highest on the northern aspect (P <.001, ANOVA; Table 5), a result opposite those of Sites 1-3. Significantly higher position densities were found at 0-5 :11, 15-20 m and 20—25 m north (P <.05, orthogonal contrasts; Figure 8). Total ground layer density was highest on the mrthern aspect (P <.001, t—test; Table 5). Although not significant, highest densities were found at the edge positions on both aspects. Individual Species Distributions and Abundance The results prwented below highlight those species whose distribution or abundance patterns illustrated position and/or aspect effects. The species composition for each site, separated into canopy and non-canopy components, is presented in Tables 6-9. It can be assumed that those species listed in the tables, but not treated balm, either showed no position or aspect differences or were so rare that they had little inpact on the analysis. Rare species were defined as those having a frequency of occurrence below 5% in sample plots. For nany species mean density at a position was low but their distribution patterns were important in delineating edge-to—interior and aspect effects. For these species no attempt was made to indicate variability within a position; iretead, the total mmber of individuals per position is reported 51 Table 6. Species composition of Site 1. Carlopy species Acer 93911113 Michx. f. Black maple A. saccharum harsh. Sugar maple (arya cordiformis (Wang.) K. Koch Bitternut hickory Egg wifolia Ehrh. American beech Fraxinus americana L. White ash Prunus serotina Ehrh. Black cherry Tilia americana L. Basswood Ernie alba L. White oak Q. rubra L. Red oak W geciesa Carpinus caroliniana Walt. Hornbeam Cornus florida L. Flowering dogwood Q. alternifolia L.f. Alternate-leaf dogwood Crataggg Eggctata Jaoq. Dotted Haw Crataegp_s sp. Hawthorn Hamamelis Virginians L. Witch hazel Ionicera dioica L. Wild Honeysuckle L. tatarica L. Tartarian Honeysuckle (Btrgg virginiana (Mill.) K. Koch Ironwood Prunus virginiana L. Choke—cherry Ribes americamm Mill. Wild black currant _R. angsbati L. Gooseberry Rosa multiflora Thunb. Rose Rubus allfieghaniensis Porter Common blackberry R. occidentalis L. Black raspberry Sambucus canadensfi L. Conunon elder _S_. Egbens Michx. Red-berried elder Ulmus rubra Muhl. Slippery elm _I_J_. thomasii Sarg. Rock-elm Viburnum opulus L . ' European cranberry-bush y. lantana L. ' . Wayfaring tree Zanthoxylum americarmm Mill. Prickly ash aSpecies that do not typically enter the canopy in the forests of the study region. 52 Table '1. Species composition of Site 2. Canopy'species Acer _niggum Michx. f. Black maple gr rubrum L. Red maple A. saccharum Marsh. Sugar maple Carya cordiformis (Wang.) K. Koch Bitternut hickory §§g2§_grandifolia Ehrh. American beech Fraxinus americana L. 'White ash 'Liriodendron tulipifera L. Yallow'poplar Prunus serotina Ehrh. Black cherry Quercus rubra L. Red oak Tilia amerirana L. Basswood W speciesa Caerinus caroliniana Walt. Hornbeam Celtis occidentalis L. Hackberry Cornus racemosa Lam. Gray dogwood Crataggg sp. Hawthorn Fraxinus quadrangulata Michx. Blue ash Ostrya vigfliiana (Mill.) K. Koch Ironwood Prunus virginiana L. Choke-cherry Ribes gynosbati L. Gooseberry Rubus alleghaniensis Porter Common blackberry Sambucus canadens is L . Common elder _S_. pubens Michx. Red-berried elder Ulmus rubra Muhl. Slippery elm aSpecies that do not typically enter the canopy in the forests of the study area. 53 Table 8. Species composition of Site 3. CEDQPY’SPECiES Acer rubrum L. Red maple IQ. saccharum.Marsh. Sugar maple Carya‘glabra (Mill.) Sweet Pignut hickory Q, ovata (Mill.) K. Koch Shagbark.hiCkory Fraxinus americana L. White ash guglans cinerea L. Butternut Prunus serotiga Ehrh. Black cherry Quercus alba L. White oak .g. rubra L. Red oak Q, velutina Lam. Black oak Tilia americana L. Basswood anr-canopyspeciesa Acer ngguggg L. Boxelder Amelanchier arborea (Michx. f.) Fern. Serviceberry Celtis occidentalis L. Cannon hackberry Cornus alternifoli§_L. f. Alternate-leaf dogwood .9. racemosa Lam. Gray dogwood Corylus americana Welt. Hazel-nut Crataggg sp. Hawthorn Elaeagnus umbellata Thunb. Autumn olive Hamamelis virginiana L. Witch hazel Lonicera tatarica L. Tartarian Honeysuckle Lonicera sp. Honeysuckle Malus coronaria L. 'Wild crab apple Prunus virginiana L. Choke-cherry Rhamnus catharticus L. Buckthorn Ribes gynosbati L. Gooseberry Rubus allegheniensis Porter Common blackberry Rosa multiflora Thunb. Rose Rosa sp. Rose Sambucus canadensis L.” Common elder Sassafras albidum (Nutt.) Nees Sassafras Ulmus rubra Muhl. Slippery elm g; thomasii Sarg. Rockrelm Viburnum acerifolium L. Mapleleaf dogwood Viburnum opulus L. European cranberry-bush Unknown ‘ aSpecieswthat do not-typically enter the canopy in the forests of the study region. ' 54 Table 9. Species composition of Site 4. Canopy'species Acer rubrum L. Red maple A. saccharum Marsh. Sugar maple Carya cordiformis (Wang.) K. Koch Bitternut hickory Q, glabra (Mill.) Sweet Pignut hickory g, ovata (Mill.) K. Koch Shagbark hickory gaggg grandifolia Ehrh. American beech Fraxinus americana L. White ash g, pennsylvanica Marsh. Green ash Jgglans nigra L. Black.walnut Liriodendron tulipifera L. Yellow'poplar Prunus serotina Ehrh. Black cherry Quercus alba L. ‘White oak Q. bicolor Willd. Swamp white oak (Q rubra L. Red oak iNon—canopy’speciesg Amelanchier arborea (Michx. f.) Fern. Serviceberry Carpinus caroliniana Walt. Hernbeam Celtis occidentalis L. Common hackberry Cornus alternifolia L. f. Alternate-leaf dogwood .Q. racemosa Lam. Gray dogwood Corylus americana Walt. Hazel-nut Crataggus sp. Hawthorn Elaeagnus umbellata Thunb. Autumn olive Hamamelis virginiana L. ‘Witch.hazel Lonicera tatarica L. Tartarian Honeysuckle Malus coronaria L. Wild crab apple Ostrya virginiana (Mill.) K. Koch Ironwood Populus deltoides Marsh. Cottonwood Prunus virginiana L. Choke-cherry Rhamnus catharticus L. Buckthorn Rhus gyphina L. Staghornrsumac Ribes gynosbati L. Gooseberry Rosa multiflora Thunb. Rose - Rubus allegheniensis Porter Common blackberry Salix sp. Willow Sassafras‘albidum (Nutt.) Nees Sassafras Ulmus rubra Muhl. Slippery elm g. thomasii Sarg. Rockrelm Viburnum acerifolium L. Mapleleaf Viburnum y, lentago L. Nannyberry V. ogulus L. European cranberrybbush Zanthoxylum americanum.Mill. Prickly ash aSpecies that do not typically enter the canopy in the forests of the study region. 55 In all sites many of the individual species distribution and abundance patterns reflected differential success in resporae to positicm and/or aspect, in one or more strata. Certain species were restricted to, or were most ahmdant in, edge or interior forest, with many also showing aspect responses. This can be interpreted as an adaptation to a specific microclimatic condition created through the interaction of position and aspect. Beech-Ear Maple Fragrants Site 1 (4.68 ha) In the canopy stratum sugar maple was distributed over the entire northern aspect and into the interior (Figure 9). Densities of the first two northern edge positions were significantly greater than the remainder of the aspect (P <.001, orthogonal contrasts of untransformed data). On the southern aspect this species was not found within 10 m of the edge (Figure 9). In the sapling stratum sugar maple was present in all plots throught the forest, however total density (the mean of the 10 positions) was significantly greater on the northern aspect (P <.05, ANOVA; Table 10), and density of position 0-5 111 north was greater than the remainder of the mrthern aspect (P <.05, orthogonal contrast; Figure 9). In contrast, total density in the shrub layer was significantly greater on the southern aspect (P <.01, t-test; Table 10). This was due for the most part to high densities interior of 25 m on this aspect (Figure 10). Total density in the ground layer was significantly higher on the northern aspect (P <.001, ANOVA; Table 10). Low densities on the southern 56 1000 CANOPY SOUTH .00., NORTH COO 4’ ‘00 0 ‘ ‘V‘: ‘ ‘\ ‘\\‘\\“\‘\§‘\\ ‘\\\ _“\_\\! III C 2000 1 .— 0 Ill - f - VVVVV :4 :l: ° 15 25 35 45 INT. 45 35 25 15 5 2 III m n. OAPLINO 3 m NORTH SOUTH h 1..." ‘ U) 12000 ,5’ g 000’ g ---l--rh- - 5 5 d 1 (I 15 25 35 45 INT. 45 35 25 METERS FROM THE EDGE Figure 9. Sugar maple canopy and sapling density profiles from theedge tosomonthenorthernandsouthernaspectsinSite 1. The x-axis values are the upper limits of 5 m distance intervals. The interior (INT.) is 50 m from all edges. The y-axis values are means +/- se; n=5. 57 Table 10. Density per hectare of selected species in the sapling, shrub, and ground layer (95L) strata on the northern and southern aspects (0-50 m) in beechrsugar maple’fragments. Aspect Site Species Stratum. North South 1 Acer saccharum sapling 760 (93)*3 535 (52) shrub 5227 (625) 7380 (684)** 9.1. 178300 (16645)** 17600 (4013) Fraxinus americana sapling 47 (19) 627 (233)* g.l. 1300 (393) 18600 (2330)* Ulmus rubra shrub 53 (20) 793 (201)* Carya cordiformis g.l. 450 (137) 1650 (340)** 2 Acer saccharum sapling 1330 (107)** 848 (109) g.l. 78017 (8192) 41900 (3742)** Fagus grandifolia sapling 80 (29) 252 (66)* Ostrya Virginians sapling 6O (20) 308 (59)** Prunus serotina g.l. 2533 (554)** 600 (381) aThe mean of 50 sample plots per aspect +/- (se). * P NORTH SOUTH I» 1200000 90000" 50000‘ 30000‘ Egg 5 15 25 45 INT. 45 35 25 15 METERS FROM THE EDGE Figure 17. Sugar maple shrub and ground layer density profiles frcm the edge to 50 m on the northern and southern aspects in Site 2. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y-axis values are neans +/- se; n=5. 69 1S SAPLINO NORTH 12 " SOUTH ‘M‘ \ \ \ OEEE 11%.- z 9. t a O n. a: x? In a/ O- y" 2’ a, ., i, -2 - -, - a 1 45 INT. 45 35 15 In I; .. u. CANOPY 0 * NORTH SOUTH 180 ...I < .— O .— 121- so , v O / / , D I- . iwfiJ 5 15 25 55 45 1m. 45 35 '25 15 5 METERS Fnou THE Enos Figure 18. American beech sapling (top) and basswood campy (bottcm) abundance profiles frcm the edge to 50 m on the mrthern and southern aspects in Site 2. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y-axis values are the total number of stens encountered in each 5 m interval and in the interior. Beech canopy, shrub, and ground layer profiles were similar to the sapling profile; . basswood sapling, shrub, and ground layer profiles were similar 'to the campy profile. 7O I=AHIC3PWV ,0 NORTH SOUTH g Q s Q ‘Iss thanks. 5'! 3525155 35 45 INT. 45 SHRUR 4 0 NORTH SOUTH TOTAL JI‘ OF STEMS PER POSITION w 9 W‘\“m“ A A A A v v v . 25 35 45 INT. 45 35 25 15 5 v v u 4 .e METERS FROM THE EDGE Figure 19. White ash- campy and shrub abundance profiles from the edge to 50 m on the mrthern and southern aspects in Site 2. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m fran all edges. The y—axis values are the total number is stems encountered in each 5 m interval and in the interior. The sapling profile was similar to the canopy profile. 71 interior of 35 m on either aspect (Figure 20). Irormod was distributed throughout both aspects in the sapling strata (Figure 20) but total density was significantly greater on the southern aspect (P <.001, Welch's approximation; Table 10). High abundance extended 45 111 into the forest. In the shrub layer the species was confined to positions exterior of 40 m on both aspects, while in the ground layer iroruood was found only exterior of 15 m on the southern aspect. Slippery elm (not illustrated) was confined to positions exterior of 20 m in the canopy and 35 m in the sapling strata, on mrthern aspect. In the lower strata its distribution was somewhat wider on the mrthern aspect; however, abmdance was still highest at the mrthern edge. Red oak (not illustrated) was confined to the first 15 m of forest on southern aspect in the campy and sapling strata, and was found only in the extreme edge position on both aspects in the ground layer. Shrub-sized individuals were not found. In the campy black cherry (not illustrated) was confined to the first 25 m of forest on the mrthern aspect. Sapling distribution extended 45 111 into the forest on the mrthern aspect. In the shrub layer this species was found only at position 0—5 111 south, while in the ground layer black cherry was distributed throughout the mrthern aspect and the most interior positions on the southern aspect. Total ground layer density was also significantly higher on the mrthern aspect (P <.01, Welch's approximation; Table 10). Additional distributions of interest but not illustrated included: gooseberry, which was found throughout the southern aspect; hornbeam, which was found only in the first 25 m of forest on the mrthern aspect; gray dogwood (Corrms racemosa Lam.) which was 72 CANOPY NORTH SOUTH o .. “\‘\‘ N 4m on u: N. u- u u 45 INT. 45 35 25 15 ' 5 15 SAPLING "0"“ SOUTH TOTAL # OF STEMS PER POSITION /, / , «Q 9 ,g g {I Q Q \“X \\ 1 /’l '7” ("I c // /; 5" 7 7' / afigjrlgli.fl- 35 45 INT. 45 5 15,25 01 . 0| METERS FROM THE EDGE Figure 20. Ironwood campy and sapling abundance profiles from the edge to :50 m on the mrthern and southern aspects in Site 2. The x-axis values are-the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m frcm all edges. The y—axis values are the total number of stems encountered in each 5 m interval and in the interior. 73 confined to the extreme edge on the mrthern aspect; aid common elder which was confined to positions exterior of 5 m in the shrub layer and 25 m in the ground layer on the mrthern aspect. OakFrgments Site 3 (3.60 ha) Black oak was nearly ubiquitous in the canopy, showing no strong position or aspect trends (Figure 21). The species was not found in the sapling layer, and was of only minor inportance in the first 10 m of forest on the southern edge in the shrub layer. Ground layer distribution was concentrated within the first 30 m of forest on the southern aspect (Figure 21). Although abundance was not exceptional 1y high, the pattern may be indicative of the envirormental conditions necessary for the establishment of oak regeneration. canopy abundance of pignut hickory showed a strong decrease frcm the mrthern edge to the interior (Figure 22). Highest abundance on the southern aspect was also found near the edge. Total density was significantly higher on the mrthern aspect (P <.001, Welch's approximation; Table 11). In the sapling aid shrub strata, pignut hickory was found throughout the southern aspect, with highest abundance at or near the edge (Figure 22). The species was distributed widely in the ground layer, with no strong position trend evident but abundance was highest in the mid-to-exterior positions (11 both aspects. Inthecanopy,redmapleaburxia1ceincreasedfrcmtheedgetothe interior on the mrthern aspect (Figure 23). On the southern aspect, 74 CANOPY SOUTH NORTH SOUTH INT. 35 '7 ' 25 onouwo LAYER NORTH INT. 25 15 25 45 35 .1 4 2 o 8 4V 3 2 1 o 22:”?— cum mam...» no a. ._<._.o._. METERS FROM THE EDGE Black oak campy ad ground layer aburdaxce profiles fran the edge to 50 m on the mrthern ad southern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals The interior (INT.) is 50 m fran all edges. The y-axis values are the total number of stems encountered in each 5 m interval Figure 21. and in the interior. (positions). 75 CANOPY NORTH SOUTH QGNIN‘fi \\ \\ fi.§‘§‘§.fi‘§ ‘fl.§.&.fi.fl.§ “‘§\ ‘\ \. ‘ . NM§G§t “§ {fix I 01m / 2’ '7' 7 '7 27 ' 4'5 1 I$. ' 4'5 '7 A ' ' 35 25 15 CV NORTH SOUTH TOTAL # OF STEMS PER POSITION 35 2515 5 5 15 , 25 35 45 INT. 45 versus FROM THE EDGE Figure 22. Pignut hickory campy ad shrub abundame profiles frcm the edge to 50 m on the mrthern ad southern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m frcm all edges. The y-axis values are the total number of stars encountered in each 5 m interVal , and in the interior, The sapling profile was similar to the shrub profile. Table 11. 76 Density per hectare of selected species in the canopy, sapling, shrub, and ground layer (g.l.) strata on the northern and southern aspects (edge to 50 m) in oak fragments. Aspect Site Species Stratum North South 3 Acer rubrum canopy 128 (27)*a 20 (9) sapling 338 (61)** 188 (42) g.l. 12100 (2182)** 6250 (904) Carya glabra canopy 116 (20)** 36 (12) Prunus serotina canopy 116 (20) 268 (34)** sapling 144 (24) 368 (47)** shrub 300 (93) 713 (129)** g.l. 800 (271) 3150 (851)** g, vigginiana shrub 573 (230) 2193 (366)** g.l. 5450 (917) 12650 (1945)** Ulmus rubra shrub 80 (39) 213 (50)* g.l. 150 (85) 750 (192)* Amelanchier arborea sapling 36 (12) 84 (18)* shrub 133 (51) 420 (100)* 4 Acer rubrum shrub 544 (138)"‘b 156 (57) g.l. 2438 (719)* 688 (283) Prunus serotina sapling 360 (137) 673 (92)* shrub 1189 (224) 2500 (267)** g.l. 3063 (735) 8500 (1437)** Fraxinus americana sapling so (33) 393 (67)” shrub ‘ 1644 (351)* 722 (180) g.l. 22917 (2128)** 13000 (1628) Rubus allegheiensiS‘ g.l. 12188_(2564)** 1813 (800) aThe mean of 50 plots +/- (se). e mean of 30 plots +/- (se). * P (.05. ** P <.01 indicates significantly higher density than corresponding aspect. 77 the species was confined to positions interior of 25 m, with aburdance low in all positions. Total campy density was significantly greater on the mrthern aspect (P <.05, Welch's approximtion; Table 11). Sapling density increased from the southern ewe to about 45 111 into the forest (Figure 23). Grourd layer density was lowest within the first 10 m of forest on the southern aspect ad highest fran 15 to 25 m on the mrthern aspect (Figure 24). Total densities were significantly greater on the mrthern aspect in both the sapling ad grourd layers (P <.01, Welch's approximation; P <.01 ANOVA; Table 11). In the shrub layer no significant position or aspect differences were evident; Innever, highest densities were fourd in the middle positions on the southern aspect (Figure 24). Red maple density was low in the interior position in the subcanopy strata, while aburdame was moderate in the campy (Figures 23 & 24). Black cherry abundance was highest on the southern aspect in all strata. In the campy, high densities exterded to 50 111, while in the subcampy strata density was highest within the first 30 m of the forest (Figure 25). Total density was significantly higher on the southern aspect in all strata (P <.01, Welch's approximation; Table 11). Choke—cherry distribution was similar to black cherry, exterding 50 111 into the forest on the southern aspect, with highest abundance near the southern edge. Total shrub ard grourd layer densities were significantly greater on the southern aspect (P <.01, Welch's approximation; Table 11). figs; multiflora Thunb. was widely distributed on both aspects in the shrub layer but the patterns of aburdance clearly illustrated an interaction of ecbe ad aspect (Figure 26). Densities were highest 78 c. O CANOPY wonm 5007" 41' I “’2 tl BOO Jm‘n‘ a I: F 15 25 35 45 INT. 45 TOTAL 1? OF STEMS PER POSITION Illll- .2 35 25 15 SAPLING too 4, wonm scum \\. \\.\\‘ x-“\ 4004* STEIIS PER HECTARE a g 2000 \ ‘3‘\ $ A v v f r METERS FROM THE EDGE Figure 23. Red maple campy abundance (top) ad sapling density (bottom) profiles frcm the edge to 50 m on the mrthern ad southern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—axis values in the top profile are the total number of stems encountered in each 5 m interval ad in the interior. The y- axis values in the bottom profile are means +/- se; n=5. 79 2405 SIDD" NORTH SOUTH 1250‘? 500‘ o v v v v v v v v f 5 15 25 35 45 INT. 45 35 25 15 5 45000 GROUND LAYER 350001' STEIS PER HEOTARE NORTH SOUTH zvoorL 15000" 5000“ .15. IL 5 15 25 35 45 INT. 45 35 25-15 5 METERS FROM THE EDGE Figure 24. Red maple shrub ad grourd layer density profiles from the edge to 50 m on the mrthern ard southern aspects in site 3. The x-axis values are the upper limits of 5 m distance intervals . (positions). The interior (INT.) is 50 m frcm all edges. The y-axis values are means +/— se; n=5. 80 coo CANOPY SOUTH 50°" wonm 400 II E! soov I; / Q / zoo» N / Q 1:: Q G 100;» a E / .. .II II = 5 15 25 35 45 INT. 45 35 5 c III I. soon 0’ a SHRUD sourw III I- sooo; NORTH a 24cc" 5: 1'! 13000 g Q 120m- , F I! Q com I! 7" I. ’ I ,9? I! .IIIII’IIQQI! .15.)!!! Q I‘QQQIQQQII o‘ IIIIIIIIIIIN-111111)..IQIUIIQQ 5 15 25 35 45 INT. 45 35 25 15 5 METERS FROM THE EDGE Figure 25. Black cherry campy ad shrub density profiles from the edge to 50 m on the mrthern ad southern aspects in Site 3. The x-axis values are the upper limits of 5 m distance intervals ’ (positions). The interior (INT.) is 50 m frcm all edges. The y-axis values are means +/- se; n=5. The sapling and ground layer profiles were similar to the shrub profile. 81 12000 eunu- SOUTH NORTH 5000" i r jl 30000 II I? 1: I .1 I! I 5000;» _ / .' , I: I v 9! 9 t S I ' I 5‘ I I I I 5 I», I! I; I IIIQII g OIIIlIIvrlIIIIIvIII III'IIIIII. 5 15 25 35 45 INT. 45 35 25 15 5 a: m m 20000 a I onouwo uven E NORTH a 150000 scum 120000 4000“ o - - - - r - - - , - - - - - - 5 15 25 35 45 INT. 45 35 25 5 METERS Fnou THE EDGE Figure 26. Rosa multiflora shrub ad grourd layer density profiles frcm the edge to 50 m on the mrthern ad southern aspects (in Site 3. The x-axis values are the upper limits of 5 m distance intervals" (positions). The interior (INT.) is 50 m from all edges. The y-axis values are means +/— se; n=5. 82 in the extreme edge positions on both aspects, decreasing into the interior. The higher densities extaded farther into the forest on the suithern aspect. In the grourd layer the species was distributed to 50 m on both aspects but it was not fund in the interior position (Figure 26). Campy irdividuals of slippery elm (not illustrated) were mly fund on the southern aspect, while in the subcanopy its distribution exterdedtoSOmonthesouthernaspectad35monthemrthern aspect. Total shrub ad grund layer densities were significantly higher on the southern aspect (P <.05, Welch's approximation; Table 11). Canopy irdividuals of basswood, white ash ad bdxelder (none illustrated) were fund only in position 0—5 m mrth. In the sapling stratum, white ash distribution exterded 45 111 into the forest on the mrthern aspect ad 15 m on the suithern aspect. Serviceberry (Amelanchier arborea (Michx. f.) Fern.; not illustrated) was fund throughout the southern aspect ad in the interior position; however, aburdance was greatest within the first 30 m of forest. 0n the mrthern aspect the species was confined to the first 25 m of forest. Total sapling ad shrub densities were significantly greater on the southern aspect (P <.05, iblch's approximation; Table 11). Additional species whose distributions reflected edge ad aspect affinities hit are not illustrated inclLded: havtlmrn, tamrian hmeysuckle, ad Russia: olive (gaeagnus _tgnbellata TInmb.), which were only fund in the first 25 m of forest on the southern aspect; Viburnum M, which was confined to the first 40 m of forest on the mrthern aspect; ad Lonicera sp. which was distributed to 45 m on the 83 southern aspect. Wayfaring tree ad gray dogwood were only fund exterior of 35 m on both aspects, while gooseberry was confined to the first 20 m of forest on both aspects. Sassafras was confined to the first 5 111 ad 30 m of forest (:1 the mrthern ad southern aspects respectively. Wild crab apple (Malus coronaria L.) was fund in the sapling ad grund layers on both aspects, exterior of 15 111, while blackberry was distribited throughout both aspects ad into the interior, with similar abxndaxce in all positions. white oak regeneration was confined to positions exterior of 10 m onthesouthernaspectadSmonthemrthernaspect. Although rare. sugar maple was fund in all the subcanopy strata. Its distribution was confined to positions interior of 15 m on the mrthern aspect ad 35 m on the southern aspect. Aburdance was greatest at positicn 40-45 m mrth. Site 4 (1.5 ha) Campy irdividuals of white oak were confined to positions interior of 10 m on the mrthern aspect ad 15 m on the suithern aspect (Figure 27). Highest abundance was fund interior of 25 m on both aspects. In the sapling stratum the species was fund througimt the southern aspect ad into the interior, while in the shrub layer it was confined to the first 5 m of forest on the suithern aspect. Red oak was widely distributed in the gromd layer but abLndance was low throughout the sanpled forest. In the sapling ad shrub strata, distributions were limited to the more exterior positions (11 both aspects (Figure 28). In the canopy, red oak aundame was 84 (ZAJIGDPW! NORTH SOUTH 21» 10 5 15 25 35 15 5 o CEAJICIPW! 5;; NORTH sourw TOTAL # OF STENS PER POSITION I g I Q Q Q I I Q Q g Q I Q // / / / / I . ' ‘I z / Yr , | / // ,6; a A A A A A A A A 5 15 25 ‘35 45 INT. 45 "areas FROM THE EDGE Figure 27. White oak (top) ad red oak (bottom) carmpy amrdaice profiles fran the edge to 50 m on the northern ad suithern aspects in Site 4. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y-axis values are the total number of stems encountered in each 5m interval ad in the interior. 85 .APLIN. NORTH SOUTH < b 1!*‘KTOL‘IK‘IG‘LFIL‘OL‘I§‘IC‘L§IK§.L‘ NORTH SOUTH TOTAL Ifi‘ OF STEMS PER POSITION 0' ‘m u“ 15 _25 35 45 INT. 45 55 25 15 METERS FROM THE EDGE Figure 28. Red oak sapling ard shrub aburdance profiles from the edge to 50 m on the northern axd southern aspects in Site 4. The x- axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—akis . values are the total number of stems encountered in each 5 m interval ard in the interior. 86 highest near the northern edge (Figure 27). Red maple was widely distributed in all strata but total densities in the shrub ard grourd layers were significantly greater on the northern aspects (P <.05, Welch's approximation; Table 11). Sapling aburdance was greatest interior of 25 m on both aspects (Figure 29). In all subanopy strata aburdance was highest in the interior. Whiteashwasmost aburdantmthenorthernaspect inthecznopy (Figures 30). Aburdanoe was greatest in the first 25 m of forest from the edge. This species was distrimted throughout the entire sampled forest in the shrub ard groutd layers (Figure 31) but total shrub ard ground layer densities were significantly greater on the northern aspect (P <.05, ANOVA; P <.001, Welch,s approximation; Table 11). Conversely, total sapling density was greatest on the southern aspect (P <.001, Welch's approximation; Table 11), with high aburdance exterding 35 m into the forest on this aspect (Figure 30). One campy irdividual of black cherry was fourd in this site, however distributions in the subcanopy strata enoonpassed the entire sampled area (Figure 32). Total densities in the sapling, shrub, ard grourd layers were significantly greater on the southern aspect (P <.05, Welch's approxination; P <.001 ANOVA; P <.01, Welch's approximation; Table 11). Highest densities were fourd in the interior (Figure 32). Hawthorn was distributed 35 m ard 10 111 into the forest on the northern ard southern aspects, respectively, in the sapling layer, ad was fourd scattered throughout the site in the two lower strata (Figure 33). Densities of gray dogwood decreased fran the edge to the 87 1200 .APLINC 1000" III fi soon .— 8 i! z ti 5 0001' won“! I sown-I I n. '7 " 3 II I, II v; E «or I! I! I! I! " II II I [I Q I! I? Q I! Q I! zoo. " a ," / II 7' j' ’ QMQQ M I! . ”II III I' A IIIIIQQIIIIIII I 5 15.125 55 45 INT. 45 55 25 15 5 "ETEBS FRO“ THE EDGE Figure 29. Red maple sapling density profile from the edge to 50 m on the northern ard southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions); The interior (INT.) is 50 m frcm all edges. The y-axis values are means +/- se; n=3. 88 CANOPY 4 1’ NORTH scum 34> 55 45 INT. 15 SAPLINO ‘ 3 " NORTH SOUTH TOTAL II‘ OF STEMS PER POSITION 51> so I. / t I ’ - I I I I . fl / a .A .5 I v f /. ,' / // . . 1 ’ f , l I, /" ., ’ } '1 " / / a IE fie- filllfld- 5 15,25 55 45 INT. 45 METERS Enou THE EDGE Figure 30. White ash campy ard sapling aburdance profiles from the edge to 50 m on the northern ard southern aspects in Site 4. The x-axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—axis values are the total number of stens encountered in each 5 m interval ard in the interior. 89 5000 OHIO- 4000 0 NORTH soum II II . 30000 I' I! 5': It r I' g 20000 l ‘I I. II ‘I E I' Q I! II I! I' a 1‘ Q Q I! I II I' 2 ...... I I I I I I.‘ ,II I‘ I! c QQHH. I! II Qr Q m QQQHI’II II MHIH m uJIIIIIIII.II.....IIIIIII d) 5 15 25 55 45 INT. 45 55 25 15 5 2 III .— 55000 ” NORTH GROUND LAYER II. o _ Sh zsooov _l ' SOUTH PE 2 I! 0 210000 i. 1- ' ; l/x , ¢ ’, I: ,-, {,1 75050 _. Z, I, 11,, 5 45 INT. 45 55 25 15 METERS FROM THE EOGE Figure 31. White ash shrub ard grourd layer density profiles from the edge to 50 m on the northern ard-southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals . (positions). The interior (INT.) is 50 m from all aspects. The y- axis valuesare means +/- se; n=3. 90 T500 541mm: i! 12000 II I! 500" II I. F. gl I' I II II I'. I' QIIIIIII, M 5000 I' I I! [I f [I II II II I' NHL!!! I'I!’ g 0 I? I'Q QQQQQI'QQQ < I' ”I ”HUN” 5 I .QH..QQQHII!HII g .. .I..I IIIIIIIIIIIIIII 5 15 25 55 45 INT. 45 55 25 15 5 z u: ‘ 25000 ”I, oaouuo LAYEN N 34000" .— ” NONTN scum 200000 .I- HIM .rb- 5 15 25 55 45 INT. 45 55 25 15 5 METERS FROM THE EDGE Figure 32. Black cherry sapling ard grourd layer density profiles frcm the edge to 50 m on the northern ard southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m from all edges. The y—ax'is values are means +/- se; n=3. The shrub profile was similar to the grourd layer .profile. 91 5H8”. O I’ SOUTH NORTH TOTAL iI OF STEMS PER POSITION (I d U N U U U n U i . ... a (I (II (I N (I (I 0 O I J I O «_m \a I P ‘m‘ 00000 GROUND LAYER 00000 .. Honm scum C ;3 IE 400001» I o i! m I I?! 30000» I “- 5 ii a 200000 I I E I I I a '°°°°I II I’ I I I / X“ ,' r" ; '4 ;’ I} IIII!’ III 0.14144 .. out 5 '1 5 25 55 45 INT. 45 55 25 15 5 METERS FROM THE EDGE Figure 33. Hawthorn shrub abtmdance (top) and gray dogwood ground layer density (bottom) prOfiles fran the edge to 50 m on the northern and southern aspects in Site 4. The x—axis values are the upper limits of 5 m distance intervals (positions). The interior (INT.) is 50 m interior of all edges. The y—axis values in the top profile are the total number of stems encountered in each 5 m interval and in the interior. "The y—axis values in the bottom profile are means +/- se; n=3. 92 interior on both aspects (Figure 33). In the sapling stratum. slippery elm (not illustrated) was confined to the first 10 m of forest on the northern aspect, while on the southern aspect its distribution extended to 45 m. In the shrub and ground layers the species was found throughout both aspects and in the interior. Blackberry was widely distributed on the northern aspect and into the interior in the shrub and ground layers. Total ground layer density was also significantly greater on the northern aspect (P <.001, Welch's approximation; Table 11). Additional distributions of interest but not illustrated included sassafras, which was most abundant exterior of 45 m on on the southern aspect and cottonwood (ngulus deltoides Marsh), which was found only within the first 25 m of forest on the northern aspect and had highest abundance near the edge. Staghorn—sumac (Rhus typhina L.) and willow (§a_l_igc_ sp.) were confined to position 0-5 m north. Prickly ash was confined to the first 10 m of forest on the southern aspect and the first 25 m on the northern aspect, while hazel—nut (Corlylus americana Walt.) was distributed throughout the northern aspect and in the interior. Several species, which in Sites 1-—3 had distributions limited to more extreme edge positions, had much wider distributions in this site. They included: gooseberry, which was distributed to 50 m on the mrthern aspect and 5 m on the southern aspect; hornbeam, which was found throughout the southern aspect and in the interior in the subcanopy strata; serviceberry, which was widely distributed throughout both aspects and in the interior in the sapling and shrub strata, and had highest abundance interior of 25 :11; wild crab apple, 93 which was distributed to 45 m on the mrthern aspect and 15 m on the southern aspect; and ironwood, which was found within the first 40 m of forest intheshrub layeronthesouthernaspect. Asinthe previous oak site, sugar maple was found in all subcrampy strata. Its distribution was confined to positions interior of 15 m on the northern aspect, and in the interior position. Abundance was greatest in the 45-50 m position on the mrthern aspect, and in the interior. The Canpositional Gradient Species Richnees The total number of species by position for Site 1 is illustrated in Figure 34. Species richness was highest at position 0-5 m south. Further, the southern aspect as a whole had both a higher absolute number of species and a higher mean species richness per position (Table 12). In Site 2, species richness was highest at or near the edge on both aspects (Figure 34). Both the absolute mimber of species and the mean number per position were nearly equal on the two aspects (Table 12). In Site 3, species richness was highest near the edge (Figure 35), with little difference between aspects in absolute or mean species richness (Table 12). Site 4 species richness peaked at 23 about mid—way into the forest on the northern aspect (Figure 35). Higher absolute species richness, and mean richness per position, were also found on this aspect (Table 12). The contribution to species richness is separated into canopy and non—canopy oonponents in Figures 34 and 35 and Table 12. Non-canopy species are defined as those species that do not typically enter the 94 2° 8|TE1 I 3 a. I I I t g I! II, III 3 III? I Zéwwéééaw ’/’/ , II I e “'6 ' Ia ' 0 I.“ n. I’D a. ll. 5 15 25 55 45 INT. 45 35 25 15 5 O z u 3 sure: a :5 z ‘20 NORTH scum I! I I II I I .! NH II 4 9‘?! I i III, III! III”?! UH H“ 4.4 4 4_4 wag; at w a 4 4944 1 i 0‘ 5 15 25 55 45 INT. 45 55 25 15 5 METERS FROM THE EDGE Figure34. Species richness profiles from the edge to 50 m on the mrthern and, southern aspects in beech-sugar maple fragments. aXlS values are the upper limits of 5 m distance intervals. interior (INT.) is 50 m fran all edges. into canopy species - and non-canopy species . Species richness is separated 95 Table 12. Species richness, separated into canopy and non-canopy oanponents, on the northern and southern aspects (echo to 50 m) in Sites 1-4 . Absolute number Mean rmmber of species of species per positiona Species Species Site Aspect Canopy Non-canopy Sum Canopy Non-canopy ‘Sum 1 North 9 8 17 5.2 2.6 6.0 South 9 19 26 6.4 6.1 12.5 2 North 9 6 17 5.6 3.1 6.7 South 6 6 16 4.6 2.6 7.2 3 North 10 20 30 6.3 6.1 14.4 South 9 23 32 5.2 9.0 149 4 North 10 24 34 6.7 11.5 18.2 South 10 . 16 26 5.7 6.4 14.1 aThe mean of 10 positions (5 m distance intervals from the acts to 50 m) on an aspect. 96 .l‘l’l 8 SOUTH NORTH 7’47" ..A'I’." 71"" .l"." "" y"”‘ .‘l”‘ 7"" > i, ‘ 4 4 I5 25 35 45 INT. 45 35 25 ‘5 SITE 4 SOUTH I‘lfllrjlrl' I’ll! 71.71'2‘7 lfleAer' .A'ZIVJIVAIVA'KA' 71.71.71. AICAVJIVA'CA' VHIVAICIVHIVA' 7"" _IVAIKAIKA' I’llrll 7"’,"l 2' " NORTH Dr W ' p ‘ ‘ 20" mmfimmm no cums—.2 15. 25 35 45 INT. 45 35 '25 “evens FROII 'me zone The interior (INT.) is 50 Species richness is separated into canopy species and'mn-canopy species [g . Species richness profiles fran the edge to 50 m on the Figure 35. northern and southern aspects of oak fragments. The x-axis values are the upper limits of 5 m distance intervals. m fran all edges. ~ I 97 canopy in forests of the study area (see Tables 6-9). In general the richrness trends described above resulted from changes in the mmber of non—canopy species (Figures 34 & 35), whereas the mmber of canopy species tended to remain fairly constant. This trend is reflected in the lover standard deviations of the canopy canponent, relative to the non-canopy comment, in all sites (Table 13). An exception is found in Site 2, in which richness patterns were similar for canopy and mn- canopy species conponents. For all sites there were no great differences between aspects in the absolute number of canopy species, mr the mean number of canopy species per position. The difference between aspects, particularly evident in Sites 1 and 4, was a consequence of changes in the mn—canopy species cannponent, similar to the trend for positions. Iggnrtance Percentag§§ An examination of the percent contribution by potential campy dominant species to total importance, by position and stratum, was nade in each site. Potential canopy daninant species were limited to thosespecies thatweremost important inthecanopyat thetinneof this study, or in the case of the oak sites, had the potential to increase in importance in the future given a minimal amount of disturbance. In Sites 1 and 2 potential campy daninant species included sugar maple and beech. Potential canopy dominant species in Site 3 included black and white oak, and in Site 4, red and white oak. Additionally, red and sugar maple and American beech were included as potential campy dominant species in Sites 3 and 4. The inclusion of Table 13. than species richness in Sites 1 through 4 separated into canopy and nonrcanopy species. Site campy species Non-esmpy species 1 5.8 (1.0) 4.3 (2.8) 2 5.1 (1.2) 2.7 (1.5) 3 6.1 (1.0) 8.4 (2.2) 4 6.3 (1.1) 9.5 (2.3) Values are means +/- (sd); n=21 (10 values each on themrthernandsouthernaspectsandoneinthe interior) . 99 Hun Lather three species in the oak forests was based an a substantial literature documenting an increase in importance of these species in more mesic oak sites with the suppression of fire (Ward 1956; Dix 1957; Monk 1961a, 1961b; McClain & Ebinger 1968; Lafer & Wisterflalll 1970; Buell gt a. 1966; Schmelz 95.9;- 1975; Miceli _e_t_al. 1977; Nigh _e_g §_l_. 1985, 1986; Sherwood & Parker 1986 (for sugar maple and beech); Dodge 1987 for sugar maple, beech and red maple; Lorimer 1984 for red naple) . Pignut and shagbark hickory were not included as potential campy dominant species in the analysis. Dodge (1987) proposed that these species may be increasing in importance in the oak forest of the study area, also because of fire suppression. However, their contribution to total importance in oak forests of the region has never been substantial, even in less fire prone sites (Braun 1950), and they are still of mimr inportance today (Dodge 1984, 1987), indicating that their potential to dominate the campy is minimal. In Site 1 importance percentages of beech and sugar maple in all strata increased from the ewe-to-interior on the southern aspect. The trend was particularly pronounced in the subcanopy strata (Figure 36). On the mrthern aspect very high inportance was generally maintained fran the interior to just short of the ethe. Although present, trends in Site 2 were not as pronounced as in Site 1 (Figure 37). Inportance percentage of beech and sugar uaple were generally lowest at or near the edge, increasing into the interior. The trend was most evident in the canopy and sapling strata. In the campy of Site 3, importance of potential campy daninant species on the mrthern aspect showed a moderate increase 100 . awe/amuse Some 5. Ampgma @5650 saw him may 5. 35533 04,3305 + fiancmp m>3m~mu “magma 8:93 can 30:8 05 Ga mmuw Hanan m>3mnmu + Ragga m>3m~mu + rename w>3mamuv m.>H mmwowmm ngw>a§uw mo Esm m5. ma >H H.309 .9“ng gone new Dayan m5 5” CON no .mhgmH 930mm new arm—9.80 may as com mo :52 95$, wusmuuonwfi 133 N no me gaucmouwm comm .maeme human “Em £989 £8295 manage 83me unsung—0p E88 3350.8 .88.... Zn. 69C s on .2 3.2: .8235 9s. 63235 «4.093me E n no muse—am swam: may mum game, max—mix m5. .H muwm 5 muommmm cumsufiom new Emnuuoc 93 :0 e on Cu wave 05. :03 83me #53200 3930 awaucmuoe mo “.390.chme mogupomeH .om magma 101 fi— nN an a? .hZ. hi on ¢W>‘J Ola-ORG “CON ”—9—. IOBL Olmhma nN n. n m a. mu 0 ON 00 00— ON 00— 0231‘“ on m? .hZ. mv mm on n— >EO-‘(0 11430834! 102 Amt/Amen: some aw Amugmm 950.8 98 each may Ca 85:85..“ 045558 + gem—amp m>3m~mu “88H 8:98 Em 30:8 may as 8.8 188 0>Umamn + 35583 “$3.305 + 358.50 955205 m.>H 8508M. H5655?“ «0 56 may ma 5 58.8. .255 956.8 Em harm 95. 5 o8 .8 .555 9538 new ~88 any as com «o A>HV gamer m0cmuh095 eBay 0 no me mmmucmuumd some .025: ummcm pcm £008 :8395 @5505 8.4.0QO #8300 30:8 Hmflcmuom .880 Ham 59G E on 2.. $sz 50235.. me. .3953“: 0055850 E h .45 mug: 5009.. m5. mum gawk, "6.3"me 0:9 .m .38 5" 308mm 588 «am 5250: Bu :0 E on On 000m 05 50.3 85.0QO #58ng 30:8 Hmaucmuoa Mo mmmucmgmm 0058.395 fin when: 103 mp nN an 01 £5: 0' an. ¢w>(d OZOOCO MOON W35. :02& ”Sub“: nN 0' n n n— “N an ON OV on on OO- ON O1 on 00 DO. 0233‘” a? .hZ. 0' an 0N nu >302¢0 ON O! 00 00 OO- ON O! 00 on GO. lNSOHBd 1 04 from edge to interior, while percentages in most of the southern aspect positiora were generally lower than mrthern interior positions (Figure 38). In the subcanopy strata a gradient of increasing importance percentages of potential campy daninant species from the southern edge position to the mrthern aspect was evident, particularly in the sapling stratum (Figure 38). Importance percentages in Site 4 were quite variable throughout the forest in the campy stratum. Percentages were lowest in the exterior portion of the mrthern aspect, and the interior portion of tlne southern aspect (Figure 39). In all subcampy strata, importance percentage of potential canopy daninant species was highest in the interior position, although the increase over other positions was minimal (Figures 39). Contribution by this ecological group to total inportance in both oak sites was quite law in the subcanopy strata, relative to the beech—sugar maple sites, throughout the sampled forest. The difference was most evident in Site 4. 105 40.35055. :000 :w .0596... gone 20 55:0 05.? 5.. 5:083 03.2.30.“ + 35800 03520.” “05033 mcwmmmm «3:0 >mo:8 0:u :H 00am 2803 045530: + 43:08.3 0335305 + 3580p 0455305 m.>H m0wo0mm H0325?“ no son 0,3 ma >H 2809 050?: 990.5 9:0 55:0 05 5.. oom .8 .5085 925.0 Em 50:8 05 5 oom «0 3: 059 083.55 HmHOu 0 no 0“ 093:858 some 50009 58.4.5055 0:0 "023:. H095 “0HQmE 00.“ “#8 05:3 3.8 0.029 ”00.30:... 00380 ”58250 E058 awaufiuom .0080 :0 50.3 E on me .925 50285 05. .0H0>u0u:an 00:8me E n no 055a 50995 05 050 00535 0.4.08.5 0:9 .m 05m :5 300900 5053500 80 50:50: 05 :0 a on Cu 0000 0:u =03 8500mm u:m:«eoc E058 1353.09 no 0000.80.58 00:09:85 .8 0.55; 106 n— 0N an n‘ .bz. n‘ on ¢W>3OI‘U ON 0' 00 00— ON 0' 00 OO— lNEOHSd 107 4052005 0000 0a A000>0H 0000.0 000 00.50 00p 5 000000000 0>30H00 + >§0000 0>30H0u “000?: 003000 000 30000 05 0fl 00.0 H0000 0.230000 + 00000000 0330a?” + 300000 0>30H05 05H 003000 H000H>fl00fi no :50 000 0w >H H0009 000.».0H 000000 000 £0.50 000~ 0.». 00m 00 0000.3. 003000 0:0 >00000 000 0a 000 mo 3: 0305. 000000095 H0000 0 00 0w 003000000 000m 00000 0000005 000 "02mm: 00000 0an 000 3.00 00.“ ”M00 00?? "005.30“ 003000 u000w=00 >00000 H0wu00uom .00000 HH0 :03 E on 0w 00.sz 00fi00u0fl 05. .0H0>u000w 00000000 0 a no 000:: 0000: 00v 000 000.305. 0wx0|x 05. .v 00%0 0w 0000000 00000500 000 0093000 00p 00 E on 00 0000 00y 50.8 003000 0000050 $00000 H0300uom no 003000000 00030095 .00 0000?“. 108 h. an an 0‘ ..p2. n1 NOON Numb. :Ofln- “INF“: nn 0“ 0p 0 n n. 0N on 0‘ .hZ. 0* an “N n. m an on 0v 0v :bzoz on on CW>‘J Dania-no co. oo. o a on on ov 0v on on on on no. 0235‘” >&OZ(O OO— lNEOHSd DISCUSSION The Effects of Edge and Aspect An examination of the data for each site reveals differences betveen edge and interior forest for many of the attributes meaured. Often an edge-to—interior gradient of change in the value of an ‘ attribute was evident. Differences between aspects for many measured attributes were also evident. The edge-to—interior trends were not alvays without variability. Further. the existence of edge and/or aspect effects were not supported by all data sets. A hypothesized edge-to-interior gradient of forest structure and canposition can be idealized as a continuous increase or decrease in sane parameter of interest, e.g. decreasing stem densities from ecb‘e-to-interior. Such a idealized situation assumes: 1) a decrease in the influence of the edge microclimate as interior positions are reached; and 2) the influence of other factors acting on the oanposition and structure of the gradient is miniml. Realization of the latter assmnption is unlikely. Many temperate forests are dependent upon mall scale disturbance, in the form of treefal ls, to promote regeneration of dominant species and maintain populations of less tolerant species that may establish in larger treefall gaps (Forcier 1975; mmkle 1981, 1982; Donnelly 1986). Additionally, "internal ecbes" such as south slopes or saturated soils 109 1 1 0 may increase light availability and serve to maintain less tolerant "edge" species in the forest interior (Donnelly 1986). Anthropogenic disturbance may also influence forest structure. Selective tree removal may function similar to treefal ls, freeing large amounts of resources which can be utilized by pioneer species. Grazing alters forest structure by prauoting the development of an even aged regenerative layer after it is discontinued, and by allcwing the establishment of resistant, opportunistic species (Auclair 8: Cottam 1971). When examining an edge-interior forest gradient, the variability within it, induced by internal edges, may mask any edge-interior forest differences, or introduce variability that may obscure potential gradient structure. For example, consider Figure 40, a profile of black cherry ground layer density in Site 1. The figure implies a positive response of black cherry seedlings to northern interior positions. However, a physical reexamination of this area of the site showed that one sample transect bisected a recovering gap, formed by a black cherry individual, that contained a very high density of black cherry seedlings. It is unlikely that in small fragments, recovering gaps or other internal edges can be avoided and still adhere to the constraints of random sampling. It is therefore necessary to either: 1) treat such instances when recognized as outliers, as was done in the the ease of the ezample by adjusting black cherry seedling densities in the affected portion of the transect to levels consistent with the analogous portions on the other transects; or 2) examine how the interaction of internal edges with lateral edge effect may influence the spatial extent of edge forest. 111 _A A A A A J Y7 V v’ V V v V' v ‘f 5 15 25 35 45 INT. 24000 GROUND LAYER ZOOOOT 'é‘ = 15000-» NORTH SOUTH 8 = c 12000? m n. U) , a 9 E aooo‘r ’ a 6 6 40001- / g “g / / / l -- .I-U, . 45 35 25 15 5- METERS FROM THE EDGE Figure 40. Black cherry groundfilayer density profile from the edge to 50 m on the northern and southern aspects in Site 1. The x- axis values are the upper limits of 5 m distance intervals. The H interior (INT.) is 50 m from all edges. The y-axis values are means; n=5. 112 The second alternative may be most appropriate in small fragments, in which no area is far removed from edge forest. Soil Moisture and Light Sane cautionary remarks concerning soil moisture measurements are necessary. Soil moisture content and availability are not almys synonymous. Textural differences between soil samples with similar moisture content may translate into differences in the amount of moisture available for plant use (Costing & Kramer 1946). In this study no major differences in soil texture were apparent, except for samples collected exterior of the forest boundaries. These samples appeared sandier than those of the interior. This may reflect a decrease in organic matter input at the forest border, or an actual increase in the percentage of sand in the soil. The significantly lower moisture content of soil from these exterior positions may be cauSed by canpositional differences as much as, or more than, the drying influence of the microclimate. The assessment of relatively uniform textural composition of interior samples was qualitative only, and therefore any discussion of soil moisture differences should be viewed with caution, keeping in mind possible differences in moisture availability. The soil moisture and light profiles illustrated the influence of edge and aspect on these two microclimatic variables. The surface soil vas dryer dieing July and there was more diffuse light penetrating into the forest nearer the edge than in the interior. Drier, lighter conditions extended farther into the forest frm the edge on the southern aspect than on the northern aspect. There were 113 no consistent position trends in soil moisture for April and September but the significantly lower moisture on the southern aspect for both dates exemplified the influence of exposure on microclimate. The significant soil moisture differences along the gradient strengthen the findings of Costing & Kramer (1946), who found an increase in soil moisture, although not statistically significant, fran the edge-to—interior on the mrthern aspect of their site. The significant decrease in diffuse light availability, from 0.8 and 8 percent of full sunlight at the northern and southern edges respectively, to 0.2 percent by 50 m into the forest on both aspects, was contrary to the results of Wales (1969) who found no differences in diffuse light from ecbe-to-interior using a similar procedure. Differences in light should have a greater influence on subnenopy vegetation, relative to canopy individuals, since these strata do not normally receive lateral exposure. Additional ly, the campy in Site 2 may have been of insufficient age to have been affected by current edge conditions when many of the individuals established. Differences in surface soil moisture should also affect subcanopy vegetation to a greater degree, since a larger percentage of the rooting zone of the understory strata is likely to to be found within the soil depth sampled (Costing & Kramer 1946). There is some evidence to support the above contention in that importance percentages of canopy dcminant species by position were positively correlated with soil moisture and negatively correlated with light, consistently on both aspects, in the shrub and sapling strata only (Table 14). This suggests that increased light and/or decreased soil moisture may favor less tolerant species. There were 114 Table 14. Correlations between importance percentage of potential canopy daninant species and A) July soil moisture, B) light along the edge to interior gradients (0-50 m) on the northern and southern aspects in Site 2. A) Soil Moisture 8) Light Aspect Aspect North South North South Canopy -.644** .771*** .180 -.801 Sapling . 634“ . 269 — . 269 - . 584* Shrub .679“ .589* -.841*** -.809*** Ground layer .054 -.132 -.534 -.235 * P (.10, ** P <.05, *** P <.01 indicates a significant Pearson product-moment correlation. 11S significant correlations in other strata but not consistently for both variables on both aspects. There were no significant correlations between soil moisture and light arnd any canponent of species richness. It is reasonable to attribute higher light levels nearer the edge to lateral light penetration alone, given the significant regressions of light on distance from the edge. Alternatively, more light may have been available if foliar cover near the edge was less than in the interior. Cover was not measured but sten densities in the campy, sapling, arnd shrub layers were positively correlated with light levels on both aspects (Table 15), which may mean there was increased foliar cover at the edge. Thus, supporting the existence of lateral light penetration effects. An explanation for the soil moisture pattern is more complicated. Certainly it is reasonable to assume that if more light is available temperature may be increased and surface soil may dry out to a greater degree, and in fact July soil moisture and light were negatively correlated on both aspects (Table 15). However, higher stem densities, particularly in the shrub and ground layer where a high percentage of roots may be concentrated in the surface soil, may deplete soil moisture reserves to a greater degree than in the interior forest. Significant negative correlations between shrub densities and soil moisture on both aspects existed (Table 16). The high sten densities in turn may be in response to increased light levels. In general, the microclimatic parameters measured and the current vegetative structure of Site 2 were not as well synchronized as might be predicted if a lateral gradient in microclimate has indeed been the 116 Table 15. Correlations between light and A) canopy, sapling, and shrub densities, 8) July soil moisture along the edge to interior gradients (0 to 50 m) on the northern and southern aspects in Site 2. Aspect North South A) Canopy .672" .694**" Sapling -.012 .648" Shrub .825*** .634" B) -.594* -.696** * P <.10, ** P <.05, *** P <.01 indicates a significant Pearson product-moment correlation. 11'? Table 16. Correlations between July soil moisture and sten densities along the edge to interior gradients (0-50m) on the northern and southern aspects in Site 2. Aspect Stratum North South Canopy -. 564* -.498 Sapling .247 -.112 Shrub -.844*** -.736** Ground layer . 613* . 294 * P <.10, ** P <.05, *** P <.01 indicates a significant Pearson product—moment correlation. 1 1 8 major factor controlling the developnent of the vegetation along the edge-to—interior gradient. However. given the history of grazing in the site this may not be surprising since subcanopy vegetation would have been suppressed for sane time after grazing was discontinued. The correlation of microclimate with density and importance percentage of campy daninant species in the shrub layer may reflect a greater influence of edge microclimate on a specific size class of vegetation. The lack of correlation in the ground layer may reflect a greater ability of early tree regeneration to tolerate a variety of microclimatic conditions. or it may reflect an independent response of this stratum to the physical environment, relative to the upper strata . General Characteristics of Edge Forest ' Although edge forest structure and canposition can be quite site specific, as will be examined in a subsequent section, sane generalizations can be made. Edge forest was characterized by higher stem densities in the canopy stratum. Basal area shaved little or no position trend. A similar trend was found in the sapling stratum of the beech-sugar maple sites. Conversely, in the oak sites sapling density was lower in the edge forest. This may reflect specific differences between forest types, i.e., age of tin edge. microclimate, wispy structure, etc. Shrub layer stem density was higher in the edge forest, with the exception of Site 1, where there was no clear distinction between edge and interior forest. Finally, in the ground layer there were no 119 strong differences in total stem density between edge and interior forest along the gradient. The lack of distinction between edge and interior forest in the ground layer may be an expression of the independent response of different vegetative layers to the physical environment. Rogers (1981), McCune and Antos (1981), and Dunn and Stearns (1988) have presented results that illustrate this type of independent response among strata in various forest types. Higher stem densities in the upper strata of the edge forest is consistent with the results of past studies (Gysel 1951; Wales 1972; Bruner 1977; Ranney 5 a1. 1981). The lack of a consistent trend in basal area was in contrast to the results of Barney _e_t §_l_. (1981) who found higher tree basal area in the edge forest. Although time constraints prevented a quantitative examination of forest height along edge-to—interior gradients, personal observation suggested that edge forest stature was generally less than that of interior forest. This difference is expressed in the convex profile often evident in forest fragments. final ler stature may be a reflection of the younger age of an edge that was allowed to advance from its original position into the surrourxiing matrix. Alternatively, or perhaps in addition to an we factor, the analler stature of ecbe forest may be a result of the reduced height growth of edge trees, relative to those of the interior. Jacobs (1954) and Neel and Harris (1971) found that trees allowed to naturally sway in the wind, and those that were nechanical 1y shaken, had reduced height growth and greater radial incranent, relative to trees prevented from moving. The higher wind velocities incident on the forest edge may increase movement of edge trees, relative to those of the interior, 120 resulting in smaller statured trees in the edge forest. This in turn may influence diffuse light penetration into the forest, since analler statured trees, coupled with the low grading horizontal limbs characteristic of edge trees (Gysel 1951), may increase shade just interior to the edge, relative to this area in a newly established edge. Species composition of edge forest will of course depend on forest type, age of a site, and its history (whitey & Rurakle 1981). Neverthelees, general trends encompassing all sites were evident. Edge forest was characterized by a greater contribution to total richness by species that do not typically reach the canopy in the light-limited interior forest. Additionally, edge forest contained a greater percentage of species cannon to disturbed habitats, and species that were not characteristic of the forest type of each particular site. Table 17 sumarizes this trend, showing that in general a higher percentage of species found between 0 and 10 m were species characteristic of disturbed or early successional habitats, relative to those between 40 and 50 m (Barnes and Wagner 1986; Voss 1985). The contrast was even more striking between edge forest and positions interior to 50 m (Table 17) but the interior species' totals were derived fran 7596 lees sampling area than the 0-10 m and 40-50 m totals and should be interpreted with caution. The distinction was even more pronounced when in addition to disturbance species, non-characteristic species were included (Table 17). Although species richness and composition were influenced by edge conditions, these attributes say nothing about the influence of 1 221 Table 17. The contribution {percent} to total species richness in three areas of eac: fragrant by either 1} species indicative of disturbed habitats or 2) species CoifiCtEIIS via ... of a non-iisturhed habitat other than bee: -sugar naple fcr es: (in sites 1 and 2) or oak forest (in Sites 3 and l). Disturbance Species Percent Ion-characteristic Specres Eercent Sun Prunus serotina 33c g. virginiaza Crataegus sp. Rosa rcltifiora Lonicer a tat..i mrthoxvlua are Rubus ails; sen; pa 3V3 er & 3 b canun 515 i ‘ u U Viburnun'opulus 10 i3 Sanhucus canadensis Prunus SEIOtlZi 17 2' v1rg13;a3a Sanbucus canadensis 8 25 Interior“ ‘Prunus serotina 13 13 43-50 a ?runus serotina 24 9. Virginia's anbns ailsgzenzsnszs Cornus racezcsa Prunus serotina 33 ,,P. Virginaana Eornus racezosa ‘J I.” Sanbucus canadensis 11 Praxrnus q.adrajcfiia: a . I O ‘A, w Interior ‘1232 Tab} 17. {cont'd.} 3 5-10 3 Rosa multiflora 43 Viburnum opulus 7 Cornus racenosa Sanbucus canadensis Prunus serotina g. virginiana Rubus allegheniensis Kleaqnus unbellata Sassafras albidur EEEE requndo Crataegus sp. Malus coronaria Lonicera tatarica Rosa sp. 43-56 a Prunus serotina 2? Viburnum opulus 5 32 g. virginasa gage sultzflora Rubus ailegneniensis Crataegus sp. 3353 sp. '0‘ 1 2h} 'Prnnus serotina 41 g. Virginians Corrus racenosa Crataegus sp. auhus allegheniensis Zanthcxylua aaerzcanua Populus deltoides ghus txphrna Malus coronaria flaw- 3953 nultiflora ll 40-50 R Prunus serotina 33 g. Virginiasa Cornus racezosa Malus coronarza Crataegcs sp. Corylus anerzzana Ruhus allscnenzsnsis Sassafras 51:;ds 33 Interior Prunus serotina 33 -lifii SS Corylus anericana Ruhus‘glleghsnlessis EMeters iron the site; Each aspects conhined. :Eiity neters frsa all edges. ‘Percent of the total hunter of species in that area of 1 24 different species on forest structure. On the other hand, a synthetic species importance value is a weighted measure indicative of the influence of a particular species on forest structure. Geisequently, the stratal profiles of importance percentages of potential canopy daninant species gives insiglnt into the structure of the forest. The decreased importance of potential campy dominant species at or near the ecbe is suggestive of an earlier successional status of the edge forest, relative to interior. A categorization of edge forest in the sites of this study as successional, relative to interior forest, supports the conclusions of Wales (1972), Ranney gt §_l_ (1981) and mirey and Runkle (1981), who noted similarities between edge forest and earlier seral stages in the regions of their studies. In the context of this analysis an "edge species" can be defired as ore that has greater inportance in the edge forest than it typically does in the interior forest. General Effects of Aspect Branples of differences in canmmity structure and canposition induced by extremes in microclimate on a local scale include: different erposures on a mountain; exposed ridges and sheltered val leys; and slopes within a forest (Cantlon 1953; Niering 1953; Zager & Pippen 1977). As in the above eamples, the northern and southern aspects of small tenperate forest fragments may be subject to microclimatic extrenes, caused by differences in intensity of solar radiation, wind velocities and amounts of precipitation (Wales 1967, 1972; Ramey 1978; Ramey gt 3.1. 1981). Intuitively, any distinctions 125 in comuunity canposition and structure resulting fran microclimatic differences would be most extrene on these aspects. The predicted result of this assumption is a greater influence of edge effects in southern aspect forest, since the edge microclimate presumably penetrates farther into the forest on this aspect. The influence of aspect on total sten desity was variable among sites. Often high desities at or near the edge extended 5 to 10 m farther into the forest on the southern aspect. Beyond this point responses were variable. This variability resulted in higher total densities on the southern aspect in some irstances and on the northern aspect in others (Table 5). The variability in desity trends by aspect between strata arnd forest type may be reflective of an independent response of structural layers and different species to environmental conditions. Alternatively, the differences may have resulted from site specific disturbance events or managenent practices. Differences in species richnees between aspects did not show any consistent trend amng sites for either total number of species or for non-canopy species (Table 12). The only consistent trend in species richress was a similar number of canopy species on both aspects of all sites. Differences between aspects in the number of successional and disturbance oriented species was also variable among sites (Table 18). The number of non-characteristic species was low in all sites and did not vary mach between aspects (Table 18). The contribution to total inportance by potential carepy dominant species did reflect an aspect distirnction. In general, percent inportance (mean of all positions on an aspect) contributed by canopy 126 Table'18. The number of species on the northern and southern aspects in each site that were either 1) species indicative of disturbed habitats, or 2) species characteristic of a non- disturbed habitat other than beechrsugar maple forest (for Sites 1 and 2) or oak forest (for Sites 3 and 4). Disturbance NOnrcharacteristic Site Aspect Species Species 1 North 2 1 South 7 1 2 North 3 2 South 3 1 3 North 11 2 South 11 2 4 North 13 1 South 9 0 127 dominants was higher on the northern aspect in all strata (Table 19). Deceptions to this trend were found in the canopy of Sites 2-4 and the shrubandgroundlayersinSite 2. Theexceptionsinthecanopymay be reflective of the original canposition of the forest when size reduction to the present fragment size occurred. Oneof thehypothnesesof thisresearchwasthnatseccndaryforest fragments are too young for the campy to danorstrate strong edge or aspect effects, and that trends would be most evident in the subcanopy strata. The exceptions noted in the canopies of Sites 2-4 lend support to this hypothesis. The loner inportance of canopy doninants on the southern aspect in Site 1, an old-growth site in which adequate tine for edge developnent has elapsed, further supports the hypothesis. The additional exceptions noted for Site 2 are most likely related to the managenent history of the site (see site canparisons below). The canpositicnal differences between aspects based on inportance of potential canopy doninant species parallels the edge-to-interior trends noted previously. In the southern aspect forest, as in edge forest in geeral, less tolerant species contribute more to forest structure than do potential canopy dominants. The recognition of sore aspect distinnctions points out the loss of refinenent in an analysis that is restricted to edge effects alone. For eanple, Phitrey & Runkle (1981) demonstrated marked edge-interior forest differences on a western aspect; however, their findings would have been much mre informative if differences between aspects were also addressed. The same criticism applies to Gysel (1951), who canbired mrthern and southern aspects in his analysis. Table 19. Inportance percentages of potential canopy dominant species by stratum on the mrthern and southern aspects in Sites 1 through 4. 128 Aspect Site Stratum Nbrth South 1 Canopy 97.28 (1.82) 55.08 (11.81) Sapling 96.73 (1.41) 68.01 (8.53) Shrub 71.97 (2.34) 51.40 (4.77) Ground layer 777.34 (2.21) 30.88 (5.23) 2 Canopy 47.06 (10.66) 73.59 (9.24) Sapling 82.50 (4.17) 67.11 (5.54) Shrub 84.77 (3.23) 87.44 (3.55) Ground layer 71.15 (2.66) 74.98 (2.25) 3 Canopy 50.49 (5.31) 50.61 (3.21) Sapling 51.27 (5.91) 21.29 (6.22) Shrub 27.15 (3.81) 16.31 (2.29) Ground layer 26.35 (2.62) 19.53 (1.75) 4 Canopy 51.11 (10.97) 72.99 (8.19) Sapling 20.27 (4.74) 17.46 (4.56) Shrub 13.90 (1.80) 7.00 (1.98) Ground layer 9.69 (1.38) 6.33 (1.79) Each value is a uean of'the importance percentages in 10 positions (5 In distance intervals from the edge to 50 m) on an aspect +/- (se). Potential canopy donninant species include: American beech, sugar maple, red maple, annd red oak (Sites 1-4); black oak (Site 3); and white oak (Site 4). Importance percentage is of 300 (total importance value) in the canopy'and.sapling layers or 200 ithhe shrub and ground layers. Total importance value is the sum of individual species importance values [relative density + relative frequency (shnrub and ground layers) + relative basal area (canopy and sapling layers)] on.an aspect. 1 29 Site Specific Distinctions Beech—Snyar Maplejgrents Site 1 This site best enenplified edge and aspect distinctions. In the upper strata sten desities were geeral ly higher near the edge and on the southern aspect. The stark contrast between edge and interior sapling desities on both aspects was particularly demonstrative of the lateral penetration of edge effects. Highest species richness was foundinthefirst 30mof forestonthesouthernaspectandthe increase in richness was due primarily to non-canopy species. The low contribution to total inportance by sugar maple and beech in the edge for all strata stresses the canpositional distirnction between edge and interior forest. Other striking examples of this difference were reflected in the distributions of slippery elm, red oak, and white ash, species which had greatest abundance in the southern edge positions. The lack of a strong species response on the northern aspect reflects the narrower breadth of edge forest in response to the more interior-like microclimatic conditions on the northern aspect. The restriction of beech to more interior positions en both aspects in all strata suggests that for this species edge conditions. may act to limit its distribution to more mesic interior forest, a result consistent with the findings of m1 (1951) and Ranney e; _a_;_. (1981). Beech hes the ability to produce vigorous root sprouts (Curtis 1959; Ward 1956, 1961; Forcier 1975). Since the canopy 1 30 distribution of beech in this site was restricted to interior positions, vegetative stens in the lower strata would be restricted to locations occupied by the root system of the canopy individuals. This suggests that the absence of beech in the understory of the ecterior positions may be caused by a lack of parent roots. This is of course acircularargument; if therearenoparent roots therecanbenno sprouts to produce parent trees. However, it seens unlikely that the current wspy distrihntion of the species is a result of stochastic events alone. Even if parent roots extended into the exterior positions the importance of vegetative sprouting may be minimal (Ward 1961). In a beech-sugar maple forest in southwestern Michigan Donnelly (1986) found that a very low percentage of beech st- were of sprout origin. Seedling regeneration was evident in the interior forest of Site 1, indicating that regeneration was not solely depedent on vegetative propagationn. Given the old-growth status of the site, adequate time has elapsed for beech establishment by seed thronghout the entire forest, yet this has failed to happen on the forest margins. Although beech is a shade-tolerant species (Baker 1949) it is dependent on the liberation of resources by treefall gaps for recruitnent into larger size classes (Forcier 1975; Donnelly 1986). If recruitmentandincreasedgrowthinthegapisinresponseto increased light availability, one might suspect that conditions at the edge might be favorable for beech. However potential canpetition fran very high desities of sugar maple stems on the nnorthern aspect and the extreme edge conditions on the southern aspect, and/or canpetition fran the less tolerant species on the southern aspect, may inhibit 1 31 beech establishment in these areas. Sugar maple, whnich is also dependent on light gaps for recruitnent (Forcier 1975; Runkle 1981; Donnelly 1986), may be inhibited by the sane conditions as beech on the southern aspect. Although present in the subcanopy strata, desity was very law in the ground layer relative to the northern aspect. Individuals may fail to recruit into the canopy, perpetuating the exclusionn of the species fran this stratum on the southern edge. The high stem desities in the canopy and sapling strata on the northern edge suggest a response to the increased light availability. In geeral there was a strong trend of species sorting along the edge—to-interior gradient and/or aspect in this site, reflecting the developnent of extensive edge conditions as would be expected in an old-growth forest fragment. Site 2 The general structural patterns in this site were similar to Site 1: highest canopy and sapling densities at the edge; highest total density in the shrub and ground layer on the southern and nnorthern aspects respectively, with no treds along the edge to interior gradient evident. However, exceptions reflective of the site's past history were evident. The increased canopy desity characteristic of edge forest (Site 1 this study; Ranney _e_t _a_l. 1981; Bruner 1977) did not exted as far into the forest as in the previous site. This may be a consequence of past managenent practices which included selective tree removal fran the edge forest, or it may reflect a lack of sufficient 1 32 tine, since reduction of fragment size, for the developnent of strong edge-interior differences. Although sapling density was highest at the edge, desities were generally very high throughout the sampled forest, with no significant difference between aspect positions and the interior, as in Site 1. Additionally, total sapling desity in this site was significantly higher then Site 1 (P < .001, t-test). The high sapling desities likely resulted fran the establishment of a cohort of seedlings and sprouts after grazing was discontinued. The canpositional patterns in this site were also reflective of its past history. Total species richness was lower than in Site 1. In a forest that had been actively grazed, excluding all but the most resistant or unpalatable species, richnees would be expected to be low for sone tine after release fran grazing. In geeral, species were not as strongly sorted along the edge- to-interior gradient. For example, basswood, slippery elm, red oak and black cherry, which shaved strong edge responses in Site 1, did so only moderately in this site. This reflects the undeveloped nature of edge vegetational conditions. As in Site 1, percent importance of sugar naple and beech were generally lower in the exterior positions in all strata, reflecting the decreased ability of the interior campy daninant species to sustain themselves in edge forest, or under the previous managenent conditions of this site (i.e., grazing). Havever, beech was found closertotheedgeonbothaspects, andsugarmapleonnthesouthern aspect, than in Site 1. In the canopy this distribution may be remnant fran a tine when the site was larger, such that the current edges were once interior forest, whnile in the snibcannopy strata it may 133 reflect the lack of strong canpetition frcm edge species. Beech distribution in this site nay suggest a nere recent isolation, relative to Site 1. Highest species richness was found in the entree edge positions on both aspects, where extrafragment diaspore inp.nt would be nest intese (Ranney 1978; Ranney g: g; 1981; Levesonn 1981). Thnis, plns higher total species richnees on the nerthern aspect and the higher importance percentage of potential canepy daninant species on the southern aspect in the subcanepy strata, sugest that post grazing species sorting by aspect has been minimal. Oak_Fragnents Site 3 Moderate structural treds delineating ache and aspect differences in the canepy and sapling strata were present in this site. These included highn canepy desities near the mrthern edge and higher total sapling desity on the southern aspect. Desity profiles in the shrub layer best exenplified edge and aspect effects in this site. Densities were highest between 0 and 5 m on the nerthern aspect and O and 20 n on the southern aspect. This supports the hypothesis that subcanepy strata in secondary forests are more likely to reflect age and aspect effects then is the canepy. A lack of position response in the ground layer was consistent with patterns in the previous sites. The distributionn of individual species indicated sorting in response to edge and aspect differences. Pignut hickory regeeration 134 was confined to positions near the southern edge, even theugh canepy individuals were nest abundant on the nerthern aspect. The optimal requirements for establishment of this species include nesic soil conditions (Fowel Is 1965). However, its subcanepy distribution in this site nay snggest a requirenent for a less light—limited condition than that of interior or nerthern aspect forest. The distribution of red maple reflected differential response to edge and interior conditions. Highest abundance in the canepy, sapl ing and ground layers were found in the mrthern interior positions. On the southern aspect the distribution was restricted to the interior portion of the gradient. Subcanepy abundance in the interior was low relative to the nerthern aspect positions. This could be interpreted as a positive response by the species to entesive edge conditions on the nerthern aspect. However, the restricted distribution of the species on the southern aspect, and the successful recruitnent into the canepy, as evidenced by moderately highn interior abundance, suggest that the optimal distribution of the species is in interior forest. The increased importance of red naple in nesic oak forests, presumably in response to fire suppression, has been reported by several werkers (Lorimer 1984; Dodge 1987). The distribution in this site suggests that the potential for this happening in the southern edge forest may be limited, a situation which in time nay serve to stregthen edge-interior forest distinctions in the site. The distributies of species such as black cherry, serviceberry, hawthern, dogwood and blackberry in this site supported their designation as edge indicators (Gysel 1951; Wales 1972; Ranney _e_‘g a_l. 1 35 1981; Wnitney 6: Runkle 1981). The confinement of sugar naple to the nerthern aspect and the interior of this site may parallel the location of mesic microsites. Auclair and Cottam (1971) have said that these microsites may be required for successful establishment of sugar maple in oak forests. Black oak was the dominant canepy species in this site but regeneration was minimal. This may be the result of an increase in oak wilt disease or possibly an increase in conpetition from additional species which establish after the suppression of fire (Auclair 8: Gotten 1971). In the southern edge forest of Site 3, canpetitionn nay preclude the recruitnent into larger size classes, but it nay net preclude the initial establishment of oak regeeration, as suggested by the concentration of oak in the ground layer in this area. Initial establishment nay be favored by a xeric microclinnate or the increased availability of light on the southern edge. Such a situation nay have sane similarity to post-fire seedbeds with warner, drier soil and nere light incident on the forest floor, conditions often required for successful oak reproduction (Monk 1961; Nigh gt _a_l_ 1985). Site 4 The highn canepy desity in the exterior positions of the forest was characteristic of edge ceditions. However in the subcanepy strata there was a lack of strong treds along the edge-to-interior gradient. Additienal ly, total shrub desity was higher on the nerthern aspect, in contrast to Sites 1-3. Total canepy desity and basal area were significantly lower in this site than that of Site 3 1 36 (P <.01 and P <.05, t—test). Further, subcanepy desities were higher than in Site 3 (Table 20). The lower canepy density and basal area nay have resulted in an increase in light availability to the lover strata, possibly explaining the high desities in the subcanepy strata (a situation analogous to lateral edge conditions). This may have been particularly true on the nerthern aspect, hence the high shrub desity relative to the southern aspect. Individual species distributions in the site reflected the deep penetration of edge conditions into the forest on both aspects. Red maple sten desity in the subcanepy strata was generally low, relative to Site 3, and its distribtionn was sporadic. This suggests that conditions favorable for red maple survival nay be limited, similar to the situation in the edge forest of the previous site. The distribution in the canepy layer does net reflect any innhnibition of the species by edge conditions as in the previous site. Harever, the present distributionn may coincide with establishment at a tine when the forest was larger, such that the present locations of canepy individuals were onnce in interior forest. The widespread abundance of black cherry, with highn subcanepy densities on both aspects, further illustrates the great extent of edge conditions in the forest. This species is rare in the canepy, but may be expected to inncrease in importance in the future given the current situation in the understory. The high position abundance and wide distribution of white ash, particularly in the subcanepy strata, net only serves to illnstrate the extent of edge conditions in the forest, but also points to a potential shift in canepy daninanoe to this species. Other characteristic edge species found throughout the Table 20. Density Sites 3 and 4. 137 (stems per hectare) in subcanepy strata in Stratum Site 3 Site 4 Sapling 743 (55) 1400 (208)M Shrub 5438 (795) 8037 (687)* Ground layer 42752 (2390) 52024 (3162)* Values are means +/- (se); n=21 (10 values on each aspect and 1 in the interior). * P <.05, ** P <.01 indicates the mean is signnificantly greater than the corresponding stratum in site 3 (Welch's approxination) . 1 38 site included blackberry, wild crab apple, slippery elm, red oak, gray dogwood and hanthern (this sttdy; msel 1951; Wales 1971; Ranney _e_t _1. 1981). Altheugh evidence for extesive edge effects on both aspects were evident, conditions on the nerthern aspect appeared to be of greater nagnitude, possibly indicating disturbance beyond that caused by lateral edge effects alone. The open nature of the canepy in and around the bracken fern patch was indicative of a past disturbance such as fire or multiple tree blow dam. Matever the cause, the creation of an "internal edge" in such a snall forest patch has contributed to serious degradation of original forest structure. Although this is a snall forest fragnent with evidence of extensive edge conditions, it did contain a small population of sugar maple in the understory. However, its distribution may be indicative of limited availability of suitable sugar maple habitat in the forest. Seedlings were only found interior to 50 m on both aspects. In the shrub layer individuals were found between 15 and 50 m on the nerthern aspect and in the interior, but with highest ahundance between 45 and 50 m and in the interior. Saplings were found only between 45 and 50 m on the nerthern aspect, and in the interior where abundance was highest. There was one canepy individual of sugar maple in the nerth central interior area of the forest. The center of greatest abundannce in the subcanepy strata was net far fran what is nest likely the parent tree; this was, therefore, an area of potentially high diaspore input. Samaras of other species in the. genus have been found to be dispersed 50-100 m in a 10 km per hour wind (Matlack 1987). 139 Presunably, the potential dispersal distance for sugar maple is similar. This suggests that the potential for site-wide distribution exists. In the context of this study, the area of conncentrated stgar mple regeeration, the nerthern interior portion of the forest may be least affected by edge microclimate, relative to the remainder of the forest. The decreased influence of edge conditions my provide sngar maple with an advantage in this part of the forest. It is tempting to speculate on the potential of this population to drive the succession of this oak site to a sugar maple-daninated forest. However, the extensive edge conditions and snall size (limited interior conditions) may prevent this frcm happening. Depth of Edge Sane indication of the depth of edge penetration in each site was evident from the site-specific discussion. A nere quantitative neasure can be garnered by determining where along the edge to interior gradients the greatest nnumber of transitionn events occurred. I have attempted to attach a degree of significannce to as nanny of these events as possible. Haever, given the small sample sizes associated with nanny of the individual species, and the high announnt of variability of manny attributes examined, this was net always possible. Further, the presence of a species in a particular area of the forest and its total absence in all other areas precludes the assignment of an alpha valte to the difference. Additionally, net all events my lead to the sane conclusion. For eample, profiles of total density 140 in the canepy, sapling, and ground layers of a site my lead 'to one conclusion conncerning edge depth, whereas the patternn in the shrub layer my lead to a very different connclusion. This was certainly true for sons of the data in this study. Rather than taking the exceptions to indicate negation of any ecge effects it my be best to treat then as outliers. As discussed earlier, stochastic and deterministic processes will act on the forest at different tines and in different fashions, such that an expectation of strict concordance between all data sets is net realistic. lather, it isthetreds that energe fromthedataenmssethatsheuldbe considered as nest relevant to a correct interpretationn. The four fragments examined in this study represented four very different conditions and histories. The depth of edge forest peetration was dependent on age and history of a particular site. Site 1 (4.68 ha) The number of transitionn events in all strata Stunned by position indicated that edge conditions peetrated at least 50 m into the forest on the southern aspect and 10 m on the mrthern aspect (Table 21). The extent of edge on the southern aspect was 20-30 m greater than has been reported in the literature to date by: Gysel 1951; Wales 1971; Bruner 1977; Rannney 1978; and Barney _e_g _a__l_. 1981, whe examined similar forest attributes along edge-to-interior gradients of 30 m or less. The entesive peetration of edge conditions on the southern aspect in Site 1 might reflect the old-growth, undisturbed status of the site; a condition which my have provided sufficient tine for .ssessssw sss sh sssssdssss ssssssss ass sssdms ss ssesssss sssdssssss sass . ss.ms ssssssss ss ssssss Hsssssssssss , .sshssss ssssssss ssssss “sessssss «s ssssssssss ssssssssse ass ass assessss sass assassssss s sssssssss.ss ssdssss sues “sssssss ssss assss sums ”sssssesss ssssssss sss sass“: sssssss ssssss sssssssse ssss us usssss see .ssasssss sss ss ssdss sss ss sssssssss ssssssss s as sssss sss sssssssss as ssssscssss ssss s ss. . ssssssfis sh sssmss Hmssssesssss us ssssss sss .Hsssssss sssssses ssss sessss ssssss ssssssss ss ssss .sss ssssss s sasssssss sssss .sssse “s sssssess Hess sssssss so cssssssm ssssdsss esssss asessssss es ssssssssss ssssssssss ussssssss sssssss usssmsssss as essssss .sssssssnssss sshssss Hsssssassn ss “essssss essss uss ssssss s ms sssssss ss sssss asssssssssss a has :1 sh . e e E: has: 2:“ “a '4 "1 ND - «'3 VII" m a. \D m.s OH.m mfi.su o~.mu m~.s~.sm.m~ mm.sm ss-mm ms.ss om.ms sazn sm.m« ms.ss s..mm mm.sm om.m~ m~.o~ °~.mH mH.s~ o~.m m.o seem hsssm suds: sssm sss sssm sesssz .s sssssss d ssses ss sssssss ssssssss ass ssssssss sss ss sssssssss usessssd ss ssss mss sssds .sasess ssssss ass .sssss .sssssss .esssss sss ssss ssssss .ssssss fisssssesssss es ssssss ssh .sN ssass 142 extensive edge developnent. Alternatively, it my simply reflect the greater extent of sampling along the edge—to-interior gradient in the current research, relative to previous studies. If the assumption is made that the extent of edge forest on the eastern and western aspects is intermediate to the two extrenes (approximtely 30 m) then the amount of potential edge forest in this site, assuming a fragment the size of the sanpled upland portion, was 2.25 he or 48.07% of the total area. Site 2 (2.70 ha) In this site the number of transition events indicate that edge conditions penetrated to about 5 m on both the nerthern aspect and the southern aspects (Table 21). The disturbannce history of the site, and the young age of the edges, has prevented the developnent of much distinction between edge and interior forest, or between the two aspects in depth of edge. Assuming internediatepe'etration of edge conditions on the eastern and western aspects, approximtely .32 he or 11.85% of the site is edge forest. Site 3 (3.60 ha) In Site 3 edge conditions peetrated to 5 m on the nerthern aspect and approximtely 50 m on the southern aspect (Table 21). A spatial estimte of edge forest in this site, seeming an internediate edge entent on the eastern and western aspects of 22.5 m, is 1.66 ha or 46.04% of total area. 143 Site 4 (1.5 ha) In this site edge conditions ectended to at least 50 m on both aspects (Table 21). An assumption of 50 an edge penetration on all aspects suggests that the entire site is edge forest since linear dinensions were only 150 x 100 m. This conclusion is supported by the high number of edge species and the low importance of potential canopy dominant species in the interior. (Table 21). Within and Between Site Canparisons Two of the objectives of this research, investigated throngh between—site oonnparisons, were to ascertain what influence, if any, site history and forest canpositionn had on the developnent of edge forest. An additional objective was to determine, through within-site canparisons, the effect of aspect on edge forest development. The results show that the entent of edge differed by aspect, but the magnitude of the difference and the aspect of greatest edge developnent were dependent upon site history. Site cannposition appeared to be less of a factor influencing edge characteristics. The analogy between ecge forest and recovering treefall gaps for certain structural features is not new (Gysel 1951; Wales 1972) but it has not been enteded to include the similarity between southern and northern ewe, and large and small gaps. The manner in which structure and canpositionn differed between aspects of Site 1 illustrated this well. On the mrthern aspect there was high a density of sugar maple regeneration, with only a mininal contribution fran more shade intolerant species. This shows sane similarity to a 144 small canopy gap which is filled primarily by recruitment of the dominant canepy species surrounding it. In the southern aspect edge forest, stem desity was also high, but here the inportance of less tolerant species was much greater. This is similar to a large canepy gap in which light demanding species are able to establish and successfully canpete with the dauinant canopy species of the interior forest (Runkle 1982). The development of these conditions was dependent on the relatively undisturbed nature of the site, which has allowed the vegetative differences resulting fran microclimtic variation on the northern and southern aspects to be ecpressed. In contrast, the current edge conditions in Site 2 are a result of past disturbance which resulted in minimized edge-interior distinctions in the tree and shrub cannmnnities. The current cannopy species distributions suggest establishment in more interior conditions when the site was larger. The patterns in the understory are a result of grazing which would have suppressed all but the most unnpalatable or resistant species. The establishment of vegetatian subsequent to the release from grazing would be through the most readily available nears, i.e. senual reproduction and vegetative propagation of the resident canepy species. Not until adequate time had elapsed for dispersal of species fran the local source pool into the site would the remnant vegetation face any competitive pressures from species better adapted to edge conditions. Additional consequences of grazing that may influence the developnent of edge-interior vegetational distinnctions include soil ocmpactionn and root damage. Presunably this site is in the early stages of edge developnent as evidenced by low species richness, the minimal entent 145 of edge conditions, and the current distribution of sugar maple and beach on the southern aspect, relative to Site 1. In Site 3 the depth of edge conditions was similar to that of Site 1. However, canopy distinctions were net strong, which snggests that the tine elapsed sirnce the cessation of grazing (40—45 years) has been adequate to allow the development of recognizable edge conditions and aspect differences in the understory strata only. Site 4 was perhaps the nest interesting in terms of the developnent of edge conditions. The greater entent of ache forest on the nerthern aspect, relative to Sites 1-3, may be a consequence of the ready access of edge species propagules to the naturally occurring disturbance on the nerthern aspect in this site, essential 1y creating internal edge conditions that have coalesced to create an al l-edge forest. The characteristics that serve to distinguish edge forest fron interior forest, i.e. higher stem desities, higher species richrees, a greater number of non-canopy species, and increased importance of less tolerant species were important in both the beech-sugar naple and the oak sites. In addition to compositional differences innherent in the two forest types, there were also sane differences in edge forest composition. For enample, red and white oak served to distinguish axe conditions in the beech-sugar maple sites. The presence of thnese species in the canepy of the oak sites obviously could net be interpreted in a similar nannner. However, the wide distribution of red oak in the ground layer of Site 4, which may have been in response to nere xeric conditions, or a greater availability of light throughout the site, was used to illustrate the entent of 1 46 edge in the site. An important edge indicator in both forest types was the restricted distribution of nere shade tolerant species to nere light-limited and/or mesic microsites in the forests. Many edge species were connen to both forest types. For eample, black cherry and cheke—cherry, white ash, and slippery elm were characteristic edge species in all sites. The similarity of edge composition noted in forest types of this study parallels the similarity of edge composition in oak-hickory and maple-basswood forests stndied by Wales (1972) and Rannney (1978) respectively. Minimum Critical Fragment Size Strict interpretation of the results of this study suggests that fragnents below 1.5 ha in size will eventually beccme a1 l-edge forest. This figure is consistent with that generated by a computer simulation of edge effects in forest fragnents (Ranney 1978). Leveson (1981) suggested that for mple-basswood forests of southeastern Wisconsin fragnents larger than4hanaybenecessarytoabsorbrandom disturbance and still maintain their characteristic structure and function. Given the entent of edge reported in Sites 1 and 3 of this study I snggest that this nay net be large enough. Although edge species were found in all strata, their presence was particularly evident in the subcanepy strata. These species included less tolerant canopy trees that nay seesce and die before ever reaching the canepy, and shrub species that will never be very abundant under a closed canepy. In a small forest fragnent these edge species are never very far from any portion of the forest, inclnding 147 the interior. In a large, contiguons tract of forest, the effects of disturbannces large enough to open up the canopy (e.g. multiple treefalls) and allow the establishment and success of edge species are dampened by the sheer enpanse of closed canepy forest. However, in a snall forest fragnent such disturbannce would occupy a greater percentage of total area. The accessibility of thnese "internal edges" to opportunistic species of the true ewe may pranote the coalescing of edge conditions throughout the fragnnent. 0n the basis of this hypothetical situationn it may be desirable to identify two categories of critical fragnent size; absolute and ecologically sound minimum critical size. The forner describes a size large eneugh to potentially contain some interior forest, the latter, a size large enough to sustain interior forest in the face of random disturbance. Ecologically sound minimum critical size is obviously the nere desirable attribute in a fragmented landscape. An enamination of the frequency and distribution of disturbance in small forest fragments nay prove useful for determining ecologically sound minimum critical size. Such infornation could be used to assess the stability of interior forest in Sites 1 (5.85 ha) and 3 (3.6 ha), both of which contained a large spatial extent of edge forest. WION Suntan-y This study enamined edge effects in the tree and shrub strata of four small forest fragments. The sites differed in disturbance history, size, and compositionn. In general, total sten desities in the canepy stratum were highnest at or near the edge, decreasing into the interior. Sapling desity fol loved a pattern similar to the canopy layer in the beech-sugar naple sites, while in the oak sites sapling desity generally increased fran the edge to the interior. In the shrub and ground layers desities differed nere by aspect than distance fran the edge. Abundance and distribution patterns of individnal species may have been in response to microclimatic differences between edge and interior forest. Species richness decreased from the axe to the interior, primarily as a consequence of the reduction in the number of non-canopy species. The inportance of less tolerant and disturbannce oriented species was greater in edge forest, relative to interior forest. An "edge species" was defined as one that has greater importannce in the edge than it typically does in the interior. . The depth of edge forest peetrationn varied with aspect, disturbance history, and size of a fragment. In Site 1, an old- growth, undisturbed beech-sugar maple fragment, edge conditions 148 1 49 penetrated 10 and 50 m into the forest on the nerthern and southern aspects respectively. The depth of edge on the southern aspect was 30 m greater than previous literature reports for similar forests. This my reflect the greater linear entent of sampling from the edge that was done in the current study. Alternatively, it my be a consequence of the old-growth conditionn of the site, in which adequate tine had elapsed for the developnent of strong edge-interior differences. In Site 2, a beech-sugar mple site released fran grazing approximtely 18 years prior to this study, the developnent of edge conditions was minimal on both aspects. This my reflect a lack of adequate tine for area species dispersal into the site and sorting along the edge—to-interior gradient or by aspect or both. A microclimtic gradient of decreasing diffuse light and increasing soil neisture from the edge-to-interior was present in this site; however, cannunity attributes and the microclimtic gradient were net well correlated. The lack of strong coupling between microclimte and vegetation was attributed to the disturbannce history of the site. In Site 3, a second-growth oak fragment released fran grazing approximtely 40 years ago, edge depth was similar to Site 1. The greatest enpression of edge conditions was in the sapling and shrub layers. The time elapsed sinnce grazing was discontinued apparently has been adeqnate for the development of strong axe-to-interior distinnctions in the understory, and presumably distinnctions in the canepy will beccme stronger in the future. An all—edge condition was present in Site 4, a very small (1.5 ha) oak fragment. This resulted fran an interaction between size (which allowed ecge species easy access to all areas of the forest, “150 and limited the amount of interior forest present) and disturbance, which created a large area of "internal" edge on the northern aspect. The geeral characteristics of edge forest, i.e. high sten densities, inncreased species richness, and increased importanee of less tolerant and disturbance—oriented species, were represented in all sites regardless of forest type. Although there were compositional differences between edge and interior forest of beech- sngar mple and oak sites, many characteristic edge indicator species were shared by both forest types. The results of this study have sham that size and disturbance interact to determine the aneunt of edge forest present in a particular fragrant. The potential for entesive penetration of edge conditions into the forest as found in this study, coupled with the accessibility of disturbances in interior forest to edge species, suggest the need to differentiate between absolute and ecologically sound minimum critical fragnent size when estimting long-term availability of interior forest in a fragmented landscape. Future Work Long term monitoring of the sites in this study would add to the understanding of forest edge dynamics. Questions to be addressed include: hnow stable is the well developed ecge forest an the southern aspect of Site 1; will the developnent of the edge-to—interior graiient in Site 2 proceed as predicted; in what direction will forest developnent proceed given the al l-edge conditions of Site 4? An enamination of edge-interior differences in the herbaceous flora of 151 these fragments may indicate different dimensions to edge forest thnan thesewhichwerederivedbyenaminingtreesandshrubs. The monitoring of sugar maple populations in isolated oak fragments may help to mess the potential for successional developnent to sugar maple forests in a fragmented landscape. Finally, a detailed examination of the microclimatic gradient including potential evapotranspiration, relative humidity, wind velocity, and air and soil temperature in addition to soil moisture and light would help to solidify our understanding of the abiotic factors influencing the vegetational gradient. APPENDICES APPENDIXA TOWNSHIP AND RANGE COORDINATES OF STUDY SITES Site 1 (Tourney Forest): NE 1/4, SE 1/4, 580. 30, T.4N., R.1W. Site 2 (Clever Woodland): SW 1/8, W 1/4, Sec. 31, T.4N., R.1W. Site 3: NE 1/8, 99 1/4, Sec. 23, T.5N., R.1W. Site 4: SW 1/8, SW 1/4, Sec. 19, T.4N., R.2E. 152 153 Figure 41. Site diagrams. Site 2 Site 1 L—lSOm—n hi1 pazeifi AtaAnioe pasture --150m corn planted conifer L--195 m-—— run]- ----.1 gesture ---240 m snegnuoo paiuetd pasture APPENDIX.B SITE DIAGRAMS Q' E‘ h'most /N 'o H ,3 u———-240 m -——J on *4 r m E 3:. .5. o E a) H to O m r4 0 ‘H i #1 aund esorapuod hiking path 154 APPENDIX C THE UTILITY 0F DETREND- GJRRESPONDENCE ANALYSIS IN AN EXAMINATION 0F EDGE AND ASPECT EFFECTS Microclimatic differences resulting fran the interaction of position (distance fran the edge) and aspect influence forest structure at a particular point in a fragemnt. If the influence of other variables such as tree-fall gaps, or topographic and edaphic differences, can be minimized, point— distinctive vegetative structure may be ordered along an assumed edge—to—interior microclimatic gradient. The vegetative changes associated with this microclimatic gradientcanbethought ofasacanuunitycontimnum inthesenseof Curtis (1959), but on a microscale of variationn. Patterns of spatial variation in community structure can be examined objectively, without direct reference to an environmental gradient, using indirect gradient annalysis by canparative ordination (Barbour 31; g. 1980; Whittaker 1967). Ordination is used to singlify multivariate data, and mathenatical ly express ecological differences between species or sites, through an examination of the covariance of sane measurable attribute (e.g. abundance, frequency, inportance value). The ordination can then be used to delineate the environmental gradients that are influencing caummity structure. When used in ccnnjmnction with a direct examination of structural and 155 1 56 canpositional changes along assumed environmental gradients, ordination can provide corroborative evidence for testing a hypothesized ecocline. In the present study direct gradient analysis was used to measure the linear extent of vegetative edge effects on northern and southern aspects in four different forest fragments. Ordination was used not so much as corroborative evidence in this stndy, but to aesess its applicability to an examination of changes in alpha diversity along edge-to-interior gradients . Detreded Correspondence Analysis Detreded correspondence analysis (DCA) is an eigenvector ordination technique based on reciprocal averaging (Hill 1973; Hill 6: Gauch 1980). I will not attempt to present a detailed description of the algorithm of DCA, but only a brief sunmnary of the methnod (see Hill 1973, 1979; Hill & Gauch 1980; Gauch 1982 for details). Data sets to be analyzed using DCA consist of records of abundance for a set of species (i.e. species scores) in a set of samples. Sample scores are defined as the mean score of all the species found in that sample. Newspeciesscoresarethendefinedasthemeanofthescoresofthe samples in which a species occurs. This "reciprocal averaging“ is repeated until species and sample scores stabilize to a solutionn that is indepedent of the original species scores. The final sample scores are scaled so that the lowest sample score is zero. This vector of sample scores defines the first solution of the ordination, i.e. the first ordination axis. Additional solutions exist, which are 1 57 derived with the use of matrix algebra as detailed by Hill (1973), and correspond to subsequent axes of the ordinations. The ultimate goal of the analysis is to delineate actual environmental gradients based on the arrangenent and separationn of samples on the ordination axes. For each axis an eigenvalue is geerated, which is a measure of the amount of variation explained by that axis. Eigenvalues decrease with each subsequent axis. Axes whose eigenvalues are much less thnan the highest eigenvalue are not likely to be of much significance. It is generally the first two axes of a DCA ordination that are most interpretable in terms of environnnental gradients. The legth of an ordination axis is a measure of species turnover along the gradient (Hill & Gauch 1980). Its value is the nnean standard deviation of species scores in the data set. The longer the axis the greater the dissimilarity of samples. Samples on a gradient longer than 4 ad will generally have no species in cannon (Hill 8: Gauch 1980). Methods DECORANA, a microcomputer version of DCA (Hill 1979) was used to ordinate positions (i.e. 0-5 m north 45-50 m north, interior, etc.) by stratum in each site. Initial species scores consisted of square root-transformed importance values. The use of a transformation that serves to reduce large disparities between species in the measured attribute often helps to avoid a swamping out effect as a result of daninance by one or two species (Hill & Gauch 1980). In sane cases not all species present in a particular position were inclnded in the ordination. Rare species that have little real 1 58 significance in connmunity structure can often influence the ordinnatian by their presence, particularly when the range of species scores is minimized with a transformation. Rare species can be downweighted, an option of DECORANA, or omitted. Hill & Gauch (1980) suggest that species with frequencies less than 2096 in the entire data set should be omitted frann the ordination. This criterion was adopted for this study. Results and Discussion Site 1 Ordination plots for each stratum are illustrated in Figures 42 and 43 and first and second axis eigenvalues are listed in Table 22. The predaninant trend of the ordinations in all strata was the sane. Positions were ordered along the assumed microclimatic gradient on the first axis, fran the entreme southern edge positions to a canbined grouping of southern interior and the bulk of the northern aspect positions. In the campy, shrub and ground layers the extrene northern edge position was associated with the interior to mid positions on the southern aspect. From the canopy to the ground layer, this axis accounted for 70, 51, 30, and 3196 of the variation in the data, respectively. The second axis separated positions by aspect in the upper two strata, however eigenvalues were lav. Although sane variation is evident, in geeral the ordinations indicate that a gradient of edge effects exteds approximately 45 m into the forest on the southern aspect, and 5 m on the nerthern aspect. Further, the internediacy of position 0-5 m nnorth between the extrene southern edge 159 N1 140 130 _ CANOPY 120-1 110 '4 100- 22 ZO’NM tozzUI (Dbl 90H 80 — s1 70 — S|7 60_ #3 50 _N10 54 4O - $3 30 _ 58 N4 $1506 59 20— 0 4O 80 120 160 200 240 280 AXIS ll $2 100 ~ 59 51 N7 80 -‘ $3 S4 70 .1510 N557 60 -— $5 50 _- 40 -4 30 — 55 20~ 10 - . Ell-N5 SAPLING I I I I I I I I I I I I I I I I O 20 40 60 80 100 120' 140 160 180 200 AXIS l Figure 42. DCA ordinations of positions along the edge to interior gradients in Site 1. symbols correspond to 5 m distance intervals fran the, edge to 50 m on the northern and southern aspects, (e.g. Sl= 0-5 m from the southern edge and N10: 45-50 m fran the mrthern edge). The interior (I) is 50 m from of all edges. Note that positions N1—N5, N8-N10 and I have equal scores in the sapling ordination. 160 100 N8 SHRUB 90-4 N6 SS 80-— N10 510 52 60- $3 58 N1 $6 S7 S4 40'— N2 N7 50:— N5 30-— $1 59 N3 N9 20-— “De I N4 0 I I I I I I I I I I I n I I I I I O 20 4O 60 80 100 120 140 160 180 AXIS II 130 $1 120- 1H34 100-— $9 $8 80-— N1 70-— so — 57 51° N2 .. 36 N3 50 ‘ N6 N7 N5 N8 4 _ N10 0 N4 30-— 20- 33 $4 $5 anouuo LAYER O l I T’ I I I I I I I I I I I I I I I O 20 4O 60 80 100 120 140 160 180 AXIS I Figure 43. DCA ordinations of positions along the edge to interior gradients'in Site 1. Symbols correspond to 5 m distance . intervals fran the edge to 50 m on the northern and southern aspects, (e.g. Sl= 0-5 m from the southern edge and N10: 45—50 m from the northern edge). The interior (I) is 50 m from all edges. 161 Table 22. Eigenvalues by stratum in Sites 1-4. Eigenvalues Site Stratum 1st axis 2nd axis 1 canopy .695 .209 Sapling .514 .133 Shrub .301 .062 Ground Layer .306 .106 2 Canopy .546 .179 Sapling .186 .060 Shrub .278 .095 Ground Layer .205 .113 3 Canopy' .192 .095 Sapling .214 .059 Shrub .196 .124 Ground Layer .112 .067 4 Canopy .433 .314 Sapling .445 .124 Shrub .241 .142 Ground Layer .202 .090 Eigenvalxmenof the 3rd and 4th axes were less than.2nd.axis”walues in.all.caeuen 162 and the interior is evident. These results are similar to these derived throughn direct gradient analysis. Sitg 2 Eigenvalues were lav for all but the first canepy axis (Table 22). The ordinations plots for this site (net illus.) failed to separate positions into an interpretable edge-to—interior gradient in any of the strata. The extrene edge positions on both aspects were generally separated fran the renainder of the positions in the subcanopy strata. This is exactly the interpretation that was derived from direct gradient analysis, i.e. 5-10 m of ecge forest on both aspects . Site 3 Sapling and shrub ordination plots are illustrated in Figure 44. Eigenvalues were low for all axes (Table 22), however an examination of the figure shows slight separation of the first six southern aspect positions in the sapling layer and the first five in the shrub layer. Further, the two extrene edge positions on the northern aspect were widely separated from the renainder of the positions in the shrub layer. These examples illustrate that although low eigenvalues may result from an ordination with re interpretable gradient as in the previous site, they do net negate the existeee of a partial gradient. The expression of the edge-to-interior gradient in the sapling and shrub strata is consistent wi th the finding of direct gradient analysis. The failure of the ordination to detect edge effects on the southern aspect to the extent that direct gradient analysis did was AXIS II 100 90 80 7O 60 50 40 30 20 10 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 163 N10 58 N4 N9 59 S7 N3 510 N1 N5 52 N6 $5 $6 SAPLma $1 $3 $4 80 100 120 140 160 N5 N2 SF N9 N146 N1 - 7 $5151 5%; N§7SS $10 I I 5359 $2 $1 20 40 I I 60 80 100 I I I 120 140 T I I 160 180 AXIS I Figure 44. DCA ordinations of positions along the edge to interior gradients -in Site 3. Symbols correspond to 5 m distance intervals frann the edge to 50 In on the northern and southern aspects, (e.g. Sl= 0-5 m from the southern edge and N10= 45-50 m from the northern edge). The interior (I) is 50 m fran of all edges. 1 64 likely caused by the exclusion of species with frequencies below 20%. In the direct analysis these species were included in the tahxnlation of transitional events, since their distributions were obviously associated with edge and aspect effects. Site 4 The canopy and sapling ordination plots for Site 4 are illustrated in Figure 45. The first canepy and sapling axes account for 43 and 45% of the variannce respectively (Table 22), however re interpretable tred was evident on this or the second axis. The positions are widely separated along the gradient, indicative of a high aneunt of dissimilarity between than. There was also a lack of any interpretable treds in the shnrub and ground layers (net illus.). Positions were again widely separated fran one anether. The lack of any easily interpretable gradient trends and the dissimilarity between positions can be interpreted as an indication of an all-edge condition of the forest, as was determined using direct gradient analysis. Conclusion The interpretability of the DCA ordinations was depedent on the condition of each particular site. In Site 1 the edge-to-interior. gradient was developed to such extent in all strata that the ordinationn was able to extract nearly the sane treds as did the direct analysis. In Site 3 the ordination onnly partially extracted the gradient to the extent of the alternative nnnethed. However, the expression of the gradients in the subcanepy strata was consistent AXIS II 260 240 220 200 180 160 140 120 100 80 60 40 20 200 190 180 170 160 150 140 130 120 110 100 90 80 70 60 50' 4O 30 20 10 165 $8 _ N4 CANOPY “ $3 $2 4 -57 S 55 N8 fi ' N2 51 "1 N7 N9 N3 ‘ N10 59 - $10 _ N5 I I I I I I I 556 I I I o 40 so 120 160 200 240 _ N3 N10 1 N9 _ ' N1 _ 55 N3 - 55 L1 510 $2 _ 53 N7 _ 57 S4 59 _ 58 N5 N2 “N4 _ SAPLING $1 I I I T I I I I I I I I I T 0 4o 80 120 150 20.0 240 280 320 AXIS I Symbols correspond to 5 m Figure 45. DCA ordinations of positions along the edge to interior gradients in Site 4. distannce intervals frann the edge to 50 m on thenorthernn and southern aspects, (e.g. Si= 0—5 m from the southern edge and N10= 45-50 m fran the mrthern edge). The interior (I) is 50 m from all edges. Note that N1 and N6 were anitted from the canopy ordination and N6 was omitted frcnn the sapling ordination. 166 between netheds. Gauch (1982) suggests the use of principle canponent analysis for the ordination of sites with low dissimilarity. This methed may have been nere appropriate for Site 3 and perhaps all situations in which treds in alpha diversity across alert environmental gradients are being examined. The ordination results for Sites 2 and 4 exannplified the necessity for first evalnating an analytical technique against a known data set before applying it to a new situationn. There was re § 213.1212 basis for interpreting the results as representative of little vegetative edge developnent in Site 2 and of extesive edge conditions in Site 4. Hanever, the patterns observed, i.e., ne easily interpretable gradients, low eigenvalues and correspondingly lap: and high dissimilarity between positions in Site 2 and 4, may now be used as base line comparative information in future studies. LIST OF REFERENCES LIST OF REFERENCES Auclair, A. N. and G. Cottam. 1971. Dynamics of black cherry (Prurnus serotina Erhr.) in southern Wisconsin oak forests. Ecological Monegranis 41: 153-177. Baker, F. S. 1949. A revised tolerance table. Journal of Forestry 47: 179-181. Barbour, M. G., J. H. Burk and W. D. Pitts. 1980. Terrestrial plant ecology. 'ne Benjamin/Cnmmings Publishing 00., Ire. Barnes, B. V. and W. H. Wagner Jr. 1981. Michigan trees, a guide to the trees of Michigan and the Great Lakes regionn. The University of Michigan Press, Ann Arbor. Bartholomew. 1961. Journal of the Royal Statistical Society. 23: 239—281. Bornann, F. H. and G. E. Likes. 1979. Pattern and process in a forested ecosystem. Springer-Verlag, New York. Braun, E. L. 1950. Deciduous forests of eastern North America. New York: The Free Press. Bray, J. R. 1956. Gap replaceent in a maple—basswood forest. Ecology 37: 598-600. Brower, J. E. and J. H. Zar. 1977. Field and laboratory methods for geeral ecology. 2nd ed. W. C. Brown Publishers, Dubngue, Iaa. Bruner, M. C. 1977. Vegetatian of forest island ewes. Thesis. University of Wisconsin, Milwaukee, Wisconsin, U.S.A. Buell, M., A. N. Langford, D. W. Davidson and L. F. Ohmann. 1966. The upland forest continuum in nerthern New Jersey. Ecology 47: 416-432. Burgess, R. L. and D. M. Sharpe (Editors), 1981. Forest island dynamics in nan-dominated landscapes. Springer-Var lag, New York. Cantlon, J. E. 1953. Vegetation and microclimates on nerth and south slopes of Cushetunk Mountain, New Jamey. Ecological anegraphs 23: 241-270. 167 168 Curtis, J. T. and R. P. McIntosh. 1951. An upland forest continuum in the prairie-forest border region of Wisconsin. Ecology 32: 476-498. Curtis, J. T. 1959. The vegetation of Wisconsin. Madison: University of Wisconsin Press. Barley-Hill, S. and W. C. Johnson. 1981. Acorn dispersal by the blue jay (Cyanocitta cristata). Decologia 50: 231-237. DeWalle, D. R. 1983. Wind damage around clearcuts in the ridge and valley province of Pennsylvania. Journal of Forestry 81: 158-159. Diamond, J. M., J. Terborgh, R. F. Nnitcomb, J.F. Lynnch, P. A. Opler and C. S. Robbins. 1976. Island biogeography and conservation: strategy and limitations. Science 193: 1027-1032. Dix, R. L. 1957. Sugar naple in forest succession at Washington, D.C. Ecology 38: 663-665. Dodge, S. L. 1984. Soil texture, glacial sedinennts, and woodlot species conposition in neth Ingham County, Michigan. Dissertationn. Michigan State University, Fast Lansing, Michigan, USA. 1987. Presettlenent forest of south—central Michigan. Michigan Bot. 26: 139-152. Donnnel 1y, G. T. 1986. Forest cannposition as determined by canepy gap dynamics: a beech-maple forest in Michigan. Dissertation. Michigann State University, East Lansing, Michigan, USA. Dunn, C. P., G. R. Guntenspergen and J. R. Dorney. 1983. catastrophic wind disturbance in an old-growth hennlock—hardwood forest, Wisconsin. Canadian Journal of Botany 61: 211-217. Dunnn, C. P. and F. Stearns. 1987. Relationship of vegetation layers to soils in southeastern Wisconsin forested wetlands. Anerican Midland Naturalist 118(2): 366-374. Fons, W. L. 1940. Influence of forest cover on wind velocity. Journal of Forestry 38: 481-486. Forcier, L. K. 1975. Reproductive strategies and the co—occurrence of climax tree species. Scieee 189: 808-810. Forman, R. ‘1‘. T. and M. Godron. 1981. Patches and structural components for a landscape ecology. BioScience 31(10): 733-740. 1986. Landscape ecology. John Wiley and Sons, New York. 169 Fowells, H. A. 1965. Silvics of forest trees of the United States. United states Forest Service, Government Printing Office, Washington, D.C., USA. Frannklin, J. F. and R. T. T. Forman. 1987. Creating landscape patterns by forest cutting: ecological consequences and principles. Landscape Ecology 1(1); 5—18. Gates, J. E. and L. W. Gysel. 1978. Avian nest dispersion and fledging success in field—forest eootones. Ecology 59: 871—883. Gauch, H. G. 1982. Multivariate analysis in cannnunity ecology. Canbridge University Press. Geiger, R. 1950. The climate near the ground. Harvard Unniversity Press. Gill, J. L. 1978. Designn and analysis of experiments in the animal and nnedical sciences. Iowa State University Press, Mes, Iowa. Vol. 1. Gleason, H. A. and A. Cronquist. 1963. Manual of vascular plants of nertheastern United States and adjacent (anada. Welland Grant Press, Bostonn. Glock, W. 8., R. A. Studhalter and S. R. Agerter. 1960. Classification and multiplicity of growth layers in the branches of trees at the extreme leer forest border. Smithsonian Miscellaneous Publications 4421. Gysel, L. W. 1951. Borders and openings of beech—maple woodlands in southern Michigan. Journal of Forestry 49: 13-19. Harris, L. D. 1984. The fragnented forest. Unniversity of Chicago Press. Heinselman, M. L. 1973. Fire in the virgin forests of the Boundary Waters Ganee Area, Minnesota. Quaternary Research 3: 328-382. Helliwell, D. R. 1976. The effects of size and isolation on the conservation value of wooded sites in Britain. Journal of Biogeograrhy 3: 407-416. Hill, D. B. 1985. Forest fragmentation and its implications in central New York. Forest Ecology and Managenennt 12: 113-128. Hill, M. 0. 1973. Reciprocal averaging: an eigenvector methed of ordination. Journal of Ecology 61: 237-249. 1979. DECORANA-A FORTRAN program for detreded correspondence analysis and reciprocal averaging. Ecology and Systenatics, Cornell University, Ithaca, New York 14850. 170 and H. G. Gauch. 1980. Detreded correspondence analysis: an inproved ordination technique. Vegetatio 42: 47-58. Hutchinson, B. A. and D. R. Matt. 1976a. Beam enrichment of diffuse radiation in a deciduous forest. Agricultural Meteorology 17: 93-110. . 1977. The distribution of solar radiationn within a deciduous forest. Ecological Monographs 47: 185-207. Jacobs, M. R. 1954. The effect of wind sway on the form and developent of Firms radiata D. Donn. Australian Journal of Botanny 2: 35-51. Johnson, W. C. and C. S. Adkisson. 1985. Dispersal of beech nuts by blue jays in fragmented landscapes. Anerican Midland Naturalist 113: 319-324. Karr, J. R. 1968. Habitat and avian diversity on strip-mined land in east central Illinois. Condor 70: 348-357. Kittredge, J. 1948. Forest influences: The effects of woody vegetation on clinnate, water, and soil. Dover Publications, Inc. Iafer, N. G. and W. A. Wistedahl. 1970. Tree canposition of Dysart Woods, Delmont Counnty, Ohio. Castanea 35: 302—308. Iatiner, W. M. 1950. General neteorological prinnciples. In: Handbook on aerosols, U. S. Atannic Energy Camission (Chapt. 2). Leopold, A. 1933. Gene Managenent. Charles Schrihrner's Sons, New York. Leveson, J. B. 1980. The southern-nesic forest of southeastern Wisconsin: species composition and cannunity structure. Contributions in Biology and Geology, Milwaukee County Public Nansen, Milwaukee, Wisconsin. . 1981. Woodlots as biogeograninic islands in southeastern Wisconsin. Puss 13—41 i_n_: R. L. Burgess and D. M. Sharpe, editors. Forest island dynamics in man-dominated landscapes. Springer-Ver lag, New York. Iorimer, C. G. 1984. Development of the red naple understory in nertheastern oak forests. Forest Science 30(1): 3-22. Lovejoy, T. E. and D. C. Oren. 1981. The minimum critical size of ecosystems. Pages 7—12 Q: R. L. Burgess and D. M. Sharpe, editors. Forest island dynamics in nan-dominated landscapes. Springer-Verlag , New York. 171 Lovejoy, T. E., R. 0. Bierregaard, J. M. Rankin, and H. 0. R. Schubart. 1983. Ecological dynamies of tropical forest fragnents. Pages 413—421 _ig: S. L. Sutton, T. C. Whitmore, and A. C. Chadwick, editors. Tropical rain forest: ecology and management. The British Ecological Society. Lovejoy, T. E., R. 0. Bierregaard Jr., A. B. Rylands, J. R. Malcolm, C. E. Quintela, L. H. Harper, K. S. Brown Jr., A. H. Powell, G. V. N. Powel, H. 0. R. Schubart and M. B. Hays. 1986. Fiche and other effects of isolation on Amazon forest fragnents. Pages 257-285 _ig: M. E. Soule, editor. Conservation biology: the science of scarcity and diversity. Sinauer Associates, Irnc. , Sunderland, Massachusetts. MacArthur, R. H. and E. 0. Wilson. 1967. The theory of island biogeographny. Princetonn University Prses, Prinnceton, New Jersey. Matlack, G. R. 1987. Diaspore size, shape and fall behavior in wind- dispersed plant species. American Journal of Botanny 74(8): 1150-1160. McClain, W. E. and J. E. Ebinnger. 1968. Woody vegetation of Barber Woods, Edgar County, Illinois. American Midland Naturalist 79: 419-428. Miceli, J. C., G. L. Rolfe, D. R. Pelz and J. M. Edgington. 1977. Brannfield Woods, Illinnois: woody vegetation changes since 1960. Merican Midland Naturalist 98: 469-476. Middleton, J. and G. Merriam. 1983. Distribution of woodland species in farmland woods. Journal of Applied Ecology 20: 625—644. 1985. The rationale for conservation: problems from a virgin forest. Biological Conservation 33: 133-145. Miller, R. I. and L. D. Rarris. 1977. Isolation and extirpation in wildlife reserves. Biological Conservatien 12: 311-315. Moen, A. N. 1974. Turbulence and the visualization of wind flow. Ecology 55: 1420-1424. Monk, C. D. 1961a. The vegetatien of the William L. Butcheson Memorial Forest, New Jersey. Bulletin of the Torrey Botanical Club 88: 156-166. 1961b. Past and present influences on reproduction in the William L. Hutcheson Menorial Forest, Nev Jersey. Bulletin of the Torrey Botanical Club 88: 167-175. Neel, P. L. and R. W. Harris. 1971. Motien-induced inhibition of elongation and induction of dormancy in Ligiidambar. Science 173: 58-59. 172 Niering, W. A. 1953. The past and present vegetation of High Point State Park, New Jersey. Ecological Monographs 23: 127-148. Nigh, T. A., S. G. Pal lardy and H. E. Garrett. 1985. Sugar maple- environnment relationnships in the river hills and central Ozark Mountains of Missouri. American Midland Naturalist 114: 235-251. 1986. successional treds in oak-hickory forests of central Missouri. Program of the IV International Congress of Ecology (Abstract) pg. 256. Oosting, H. J. and P. J. Kramer. 1946. Water and light in relation to pine reproduction. Ecology 27(1): 47-53. Ranney, J. W. 1978. Edges of forest islands: structure, canposition, and importannce to regional forest dynamies. Dissertation. The University of Tennessee, Knoxville, USA. , M. C. Bruner and J. B. Levenson. 1981. The importance of edge in the structure and dynamics of forest islands. Pages 67—97 _in: R. L. Burgess and D. M. Sharpe, editors. Forest island dynamics in man—dominated landscapes. Springer-Verlag, New York. Reifsnyder, W. E. 1955. Wind profiles in a small isolated forest stand. Forest Science 1(4): 289-297. Rosenberg, N. J., B. L. Blad and S. B. Verma. 1983. Microclimate: the biological environment. John Wiley and Sons. Runkle, J. R. 1981. Gap regeeration in sane old-growth forests of the eastern Unnited States. Ecology 62(4): 1041-1051. 1982. Patterns of disturbance in some old-gravth nesic forests of eastern North Anerica. Ecology 63(5): 1533-1546. and T. C. Yetter. 1987. Treefalls revisited: gap dynamies in the southern Appalachians. Ecology 68(2): 417-424. Salisbury, F. B. and C. Ross. 1969. Plant physiology. Wadsworth, Belmont, California. Schmelz, D. V., J. D. Barton and A. A. Lindsey. 1975. Donaldson's Woods: two decades of change. Proceedings of the Indiana Academy of Science 84: 234-243. Schneider, G. 1963. A twenty-year ecological investigation in a relatively undisturbed sugar uaple—beech stand in southern Michigan. Dissertation. Michnigan State University, East Lansing, Michigan, USA. 173 Sherwood, P. T. and G. R. Parker. 1986. Effects of endogenous and exogenous disturbance on tree species replacement patterns within the eastern deciduous forest. Program of the IV International Congress of Ecology (Abstract) pg. 310. Simberloff, D. S. and L. G. Abele. 1976. Island biogeography theory and conservation practice. Science 191: 285-286. Sokal, R. R. and F. J. Rohlf. 1981. Biannetry. 2nd ed. W. H. Freean and Company, San Franncisco. Sprugel, D. G. 1984. Desity, biomass, productivity, and nutrient- cycling changes during stand deveth in nave-regeerated balsam fir forests. Ecological Monographs 54(2): 165-186. Strelke, W. K. and J. G. Dickson. 1980. Effect of forest clear-cut edge on breeding birds in east Texas. Journal of Wildlife Nanagenent 44(3): 559-567. Trimble, G. R., Jr. and E. H. Tryon. 1966. Crown encroachment into openings cut in Appalachian hardwood stands. Journal of Forestry 64: 104-108. Voss, E. G. 1985. Michigan flora. Pt. 2, Dicots (Saururaceae- Cornaceae) . Cranbrook Inst. of Sci. Bull. 59. Regents of The University of Michigan, Ann Arbor. Wales, B. A. 1972. Vegetation analysis of north and south edges in a mature oak-hickory forest. Ecological Monographs 42(4): 452-471. Ward, R. T. 1956. The beech forests of Wisconsin-changes in forest composition and the nature of the beech border. Ecology 37: 407-418. 1961. Some aspects of the regeerationn habits of the Anerican beech. Ecology 42: 828-832. White, P. S. 1979. Pattern, process, and natural disturbannce in vegetation. Botanical Review 45: 229-299. Whitney, G. G. and J. R. Runnkle. 1981. Edge versus age effects in the development of a beech-mple forest. Oikos 37: 377-381. Whitney, G. G. and W. J. Somerlot. 1985. A case study of woodland continuity and change in the American midwest. Biological Conservation 31: 265-287. Whittaker, R. H. 1967. Gradient analysis of vegetation. Biological Review 49: 207-264. 1975. Connnunities and ecosystens. Macnnnillan Publishing Co. 174 Wilcove, D. S., C. H. Mclellan and A. P. Dobson. 1986. Habitat fragmentation in the tenperate zone. Pages 237-256 in: M. E. Soule, editor. Conservation biology: the science of scarcity and diversity. Sinmier Associates, Inc., Sunderland, Massachusetts. Wilmannns, 0. and J. Brunn-hool. 1982. Irish mantel and saum vegetation. Journal of Life Science, The Royal Dublin Society 3:165-174. Zager, P. E. and R. W. Pippen. 1977. Fifteen years of change in a southwest Michigan hardwood forest. The Michigan Botannist 16(4): 201-211. "llfiflilfihfljhhhlfiimhhh7flhfi“