SOME STRUCTURAL ARE} FUNCTIONAL ATTRIBUTES OF A SEMLARlD EAST AFRICAN ECOSYSTEM The“: {or flu Degree of MI. D. MlCBISAE‘é STATE UNIVERSITY Lawrence Dean Harris @970 _ LIBRARY E Michigan State mversity WWWWWW IWWWWW WWWWWWW wee 3‘ 31293 01006 2937 This is to certify that the thesis entitled SOME STRUCTURAL AND FUNCTIONAL ATTRIBUTES OF A SEMI-ARID EAST AFRICAN ECOSYSTEM presented by Lawrence Dean Harris W has been accepted towards fulfillment ‘ of the requirements for ‘ Ph.D. degree mm W Major professor Date /f yam/‘7 [9) 7/ d / 0-169 .. New} 19"” :JUN 1 2 1995 3! PET-l"— A f? ;__ . . I .. " 4",}:- 3:33-31 V ‘ V . ~- . 5" E‘ '7':¢ I.) 3 .I", 1 \i. " ABSTRACT SOME STRUCTURAL AND FUNCTIONAL ATTRIBUTES OF A SEMIqARID EAST AFRICAN ECOSYSTEM By Lawrence Dean Harris From late 196A through mid 1967 climatological, soils, vegetation and animal studies were conducted in the semi-arid Mkomazi Game Reserve of northeastern Tanzania. An elevational gradient from 230 m in the east to mountain tops of nearly 1600 m above sea level in the northwest underlaid similar rainfall and temperature gradients. Aridity coefficients, based on the different temperature and rainfall conditions alone, were about 50% greater in the central section of the reserve than in the higher elevation northwest. The soils were classified by the American 7Th Approximation to a Comprehensive Classification System and were found to consist of about 75% camborthids (aridisols), 20% pellusterts (vertisols) and 5% miscellaneous types. Soil texture, organic matter content, permeability and profile depth all reflected a gradient of conditions from west to east. The vegetation was categorized into four major types; 1) dry montane forest, 2) bushed and wooded grassland, 3) seasonally inundated grassland, and h) bushland. Grass-forb above-ground standing crop values ranged from approximately 600 gm/m2 in the 500 mm rainfall regions of the hushed and wooded grassland to about 200 gm/mz in the 350 mm rainfall regimes of the central-section bushland. Annual above- ground net production was found to vary from about #00 gm/m2 on previously unclipped plots in the northwest to about 170 gm/m2 in the Lawrence Dean Harris east-central section while denuded plot productivities were about 300 and 150 gm/m2 respectively. Differences in forage-density and ground- cover indices reflected generally poorer rangeland conditions in the central and eastern sections of the reserve. 'While the mean annual large herbivore density ranged from 12 animals/km2 (5,5h8 kg/kmz) in the northwest to about 0.5 km2 (700 kg) in the central and eastern sections, the dry season densities ranged from 23.7 animals/km2 (12,705 kg/kmz) to much less than l/km2 in the eastern sections. Seasonal biomass distribution patterns reflect a large wet season ingress of elephants (Loxodonta africana), zebra (Egggg burchellii), oryx (M M) and Grant's gazelle (Gazella 'ggfli) from adjacent Kenya as well as an eastward movement of herbivores from the dry season water source in the northwest. The eastaweSt density gradient is nearly extinguished during the wet seasons. Both spacial and temporal patterning within the large herbivore array is a major attribute of the animal community structure. Although 22 species of large indigenous herbivores inhabit the reserve, these are partially segregated by their affinities for the different vegetation types. A maximum of 12 and a median number of four'species (1': #.26) were recorded in local areas at any one time. Rhinoceros (Diceros bicornis) were most equitably distributed among the four major ‘vegetation types (niche breadth index = 3.h2) with eland (Taurotragus ‘ggzg),'wart hog (Phacochoerus aethiopicus), giraffe (Giraffa camelopgrdalis) and elephant next in order. Bushbuck (Tragelaghus scriptus), duiker (Sylvicapra grimmia) and buffalo (Syncerus caffer) were the least equitably distributed. Eland, gerenuk (Litocranius 'waIleri), reedbuck (Redunca redunca) and giraffe were most equitably Lawrence Dean Harris distributed through time. Herbivore species diversity was greatest in the bushed and wooded grassland and lower in the Open grassland, bush- land and dry montane communities. The niche overlap (on the habitat dimension) of hartebe'e‘st (Alcelaphus buselapjms), impala (Aepyceros mela_mpus) and ostrich (Struthio camelus) was great (> 0.8, limit = 1.0) while that of bushbuck, klipspringer (Oreotraggs oreotraggs) and duiker with most other species was slight (as low as 0.00, limit = 0.00). Jackals “Lie adustus) reflected the greatest overlap with the herbivores while hunting dogs (m M) reflected the least. From an ecosystem point of view, three of four species were'found to dominate the structure (numbers and biomass) as well as at least one measure of community function, 1.6. energy exchange. About 17.5% of the above-ground primary production (in terms of biomass) was estimated to be channelled through the herbivore-carnivore pathway. The independent effects of elephants, cattle (B_0§_ M) and fire on the vegetation are illustrated as are the combined effects of elephants and cattle and herbivores and fire . SOME STRUCTURAL AND FUNCTIONAL ATTRIBUTES OF A SEMI-ARID EAST AFRICAN EC(BYSTEM By Lawrence Dean Harris A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Fisheries and Wildlife 1970 ACKNOWLEDGEMENTS In Tanzania, Mr. David Anstey, Senior warden of the Game Division 'was instrumental in initiating the study. Messrs. William Dick and IE. 0. Charlton, also Northern Region wardens, provided valuable logistic assistance in carrying out the study. Mr. H. S. Mahinda, the Chief ‘Warden of the Game Division provided encouragement throughout and has kindly authorized use of the data for degree requirements. Dr. H. F. Lamprey, and Messrs. G. S. Child, Patrick Hemmingway, Ano Hooker and ‘Wk In Robinette of The College of African Wildlife Management gave stimulating advice on a number of occasions. Messrs. G. S. Child and Patrick Hemmingway piloted airplanes for me on two occasions. D. G. .Anstey piloted the plane on all but one aerial survey. The U.S. Peace Corps assisted with personal equipment and logistic support while other people too numerous to include gave freely of“their time and advice; I am especially grateful for the field assistance and friendship of Game Scouts v. M. Mushi, H.“ R. mmba,-w: George and J. Kirindo. For them, "Nashakuru sana”. Desmond Vesethitzgerald, Scientific officer, Tanzania National Parks and G. S. Child provided identifications of bird, mammal and reptile specimens. At Michigan State University, Drs. George A. Petrides, Rollin H. Baker, William E. Cooper and Stephen N. Stephenson, all members of my guidance committee, provided enthusiastic support and advice. Dr. Petrides has given editorial comments as well as guidance in certain ii aspects of the field work. I am indebted to W. E. Cooper for his analytical suggestions and constant inspiration. Dr. ‘E.’ P'. Wh‘iteside of the Soil Science Department guided me in classifying and interpreting the soils data. Mr. L. Bowdre of the M.S.U. Museum assisted with small mammal identifications while Mrs. W. H Burke prepared most of the figures. Financial support for the work has derived from the U.S. Peace Corps and the Tanzania Game Division. Fellowships and grants from the U.S. Department of Health, Education and Welfare, The National Science Foundation, The Wildlife Federation, The office of Research Development, M.S.U. and the Society of Sigma Xi have supported the analysis and residence requirements. Finally, I am especially indebted to my wife, Mary Patricia, and family for the hardships incurred and the liberty of conducting the work. iii TABLE LIS T OF MBIlES O O O O O O O 0 O O 0 LIST OF FIGURES LIST 0F.APPENDICES INTRwUCTIONQ'OOOeoeeeeee ILOCATION'AND HISTORY OF THE AREA . . PHYSIOGRAPHY AND CLIMATE Techniques . . Rainfall e e 0 Temperature . . Solar Radiation Relative Humidity ‘Wind 0 Q . O O O Aridity Coefficients The General Climatic Pattern SOILS Techniques GEOLOGY . The Catena Concept and Typical Soil the red and reddish brown soils OF CONTENTS brown and gray brown colluvial soils . . heavy black clays of the valley bottoms Fertility Considerations Discussion VEGETATION . . . . . . . . . Community Descriptions buShlandeeeeeee bushed and wooded grassland grassland iv Sequences O O O O O O O O Page viii 13 lh 15 18 A 20 20 21 21 26 27 28 29 32 31+ 35 39 43 52 52 Page upland dry forest . . . . . . . riparian and miscellaneous types . . . . . . . . . . . . . 54 Net‘mmaryPI‘OduCtioneeeeeeeeeeeoeeeeeee 55‘ RangerAnalySiSt-‘e‘oeeeeeeeeeeeeeeeeeeee 60 Discussion . e . . . . . . . . . . . . . . . . . . . . . . . 63 e e e e e e O\ 03 ANMIS O O O O O O O O O O O O O O O O O O O O 0 Collection and Identification . Numbers, Densities and Biomass Techniques . . . . . . . . ground count transects visibility profiles . . aerial surveys . . . . sample plot study areas sight recording maps . biomass calculations analy318 e e e e e e Rfisults O O O O O O C 0 relative numbers . . denSities e e e e e e absolute numbers . . . relative biomass distribution absolute biomass densities . IMovements and Temporal Dynamics . . . North-south international migrations East-"981': mwements e e e e e e e 0 Seasonal biomass patterns . . . . . Discussion . . . . . . . . . . . . . . O O O O O O ; O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 0 O O O O O O O O O O O O O O I O O 0 O O O O O O O O O O O O 0 O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O I O O 8? C OWN-I TY S TRUCTURE O O O O O O O O O O O O O O O O O O O O O 0 111+ SPOCiGS DiverSity e e e e e e e e e e e e e e e e e e e e e e NiChe Breadthfi-Habitats e e o e e e e e e e e e e e e e e e e 121 NiChe Breadth—Time o e e e e e e e e e e e e e e e e e e e e 122 Habitat PI‘BferenceS e e e e e e e e e e e e e e e e e e 1214' Species Association and Niche Overlap . .4. . . . . . . . . . 128 Bil-5011531011 e e e e e e e e e e e e e e e o e o e e e e e e e 133 m ECOSYS 1“ AS A WHOLE O O O O O O O O O O O O O O O 0 O O O O 137 SUMM O O O O O O O O O O O O O O O O O O O O O O O O O O O O 158 LIIEMNRE CITED 0 O O O 0 O O O O O O O O O O O O O O O O O O O 162 Table 1. 2. 3. 5. LIST OF TABLES Page Seasonal rainfall and aridity coefficient patterns for the Ibaya station in the Mkomazi Reserve. The upper figure of each set refers to the seasonal rainfall in on while the lower figures give the mean aridity coeff. with standard error as calculated by the Thornthwaite method. The lower aridity coeff. limit is -1.0 . . . . . . . . . . . . . . . 17 Mean daily net above-ground primary production for different seasonal periods, rainfall regimes and clipping treatment. The first area receives about 500 mm of rain- fall annually'while the second and third receive about #00 and 350 mm respectively. All values are reported as graIDS/ eieeeeeeeeeeeeeeeeeeeeeeee 59 Indices of'vegetation condition and their standard errors in three areas of the Mkomazi Reserve. The plant density index refers to the mean number of ”hits" on perennial grasses of forbs per 100 samples while the ground cover index refers to all hits on live or dead vegetation. When similar soil types are considered, a gradient of conditions exists from west to east (i.e. Dindira to Kamakou)eeeeeeeeeeeeeeeeeeeeeeoe 62 The mean numbervof animalswrecorded (excluding elephant) by the various counting techniques over the entire course of the study in relation to distance from the western boundary of the reserve. The transect data represent the .mean per km of transect while the study plot data represent themeandenSityperkmz«e‘eeeeeeeeeeeeeeee 87 Numbers and biomass per km2 for the major ungulates in three study areas of the Mkomazi Reserve during wet and dry seasons. The Dindira Study.Area (Fig. 9a) is located in the north- western corner of the reserve and represents an area of permanent water availability. The Mbulu area is located near a semi-permanent water source and the Mzara area in the west-central section of the reserve had only seasonally ‘V‘il‘blemtereeeeeeeeeeeeeeeeeeeee 89 Minimal population estimates of the major animal species inhabiting the Mkomazi Reserve. The numbers of those species for which a range is given represent the dry and wet season values, with the highest number occurring in thawetseasoneeeeeeeeeeeeeeeeoeeeee 90 vi LIST OF TABLES -- continued Table Page 7. Tweaway analysis of variance illustrating significant differences in the density of large animals on the three sample plot areas, as well as significant seasonal differences. Importantly, there is a highly significant interaction effect manifesting the change from the highest dry season density near the permanent water to the lowest density in this area during the wet season (* = sig. at P = .05, *** = P = .001) O O O O O O O O O O O O O O O O O 106 8. Comparative large herbivore density (no./km2) and mean annual standing crop biomass (kg/kmz) for different bush- land areas of East Africa: . . . . . . . . . . . . . . . . 112 9. Tabulated index values of niche breadth on the habitat and time dimensions for the larger animals of the Mkomazi Reserve. The index values were derived from the formula log Pi = -2:Pi leg Pi (see text for terminology).. . . . . 123 10. The percent contribution of the four most dominant species with regard to the mean annual number, the mean annuai biomass, and the mean annual metabolism (see text). , 2% contribution by zebra; **, A% contribution by eland . . . 139 LIST OF FIGURES Figure Page 1. 3 "o 5. The Mkomazi Game Reserve of northern Tanzania lies along the Kenya border between Mt. Kilimanjaro and the Indi‘n Oce‘n . O O O O O O O O O O O O O O O O O O O O O O 0 LP The locations of hill ranges, permanent water sources, rain gauges and soil sample sites in the Mkomazi Game Reserve 0 O O O O O O O O O O O O O O O O O O O O O O O O 10 The catenary sequence of soils along a topographic gradient in the northwestern section of the Mkomazi Reserve near the Dindira Study Area. The abbreviated adjective ”manta” refers to a montmorillonitic clay _ structure, while isohyper. means isohyperthermic . . . . . 3.0. A typical array of exchangeable cation profiles with respect to profile depth and topographic location. The vertical axis of each tier represents profile depth in on while the horizontal axis represents the exchangeable cation level measured as milli-equivalents per 100 gm of soil. The five graphs of each tier correspond to the 5 profiles sampled in the B catena near the Mbula Study Area. B1 was a typical aridisol highest on the slope while B5 represents the vertisolic clay of the valley bottom'............o............. 37 The major vegetation types of northern-Tanzania and southern Kenya. The Mkomazi Reserve is contained in the type known as bush, bushland or ”nyika”. Reproduced from the Division of Overseas Survey D.0.S. 2996 oeeeeeeeeeeeeeeeeeeeeeeeeeM A vegetation-types map illustrating the location and extent of the major vegetation associations within the Mkomazi Reserve. As depicted, the upland dry forest represents approximately 5% of the total area and occurs on most of the higher mountain peaks. In reality, closed canopy forest only occurs on the mountains above about 1000 m elevation although the dominant species occur lower down. Similarly, included in the area depicted as bushland are localized areas of nearly open grass- land (e.g. Maori) resulting from bush suppression. In -general,.the seasonally waterlogged grasslands follow the distribution of the vertisolic clay drainageways . . . #7 viii LIST OF FIGURES -- continued Figure Page 7. 9. 10. a. Typical Commiphora schimpgri and‘gg campestris bushland with Cassia‘gpp., Cordia s . and Grewia spp. subdominant in the central section of the reserve. b. A typical association of bushed and wooded grass- land (near the Dindira Study Area) in the western end of the Mkomazi Reserve. c. Open Pennesitum mezianum grassland occurring on a heavy montmorillonoid clay drainageway in the north- western section of the reserve. Adjacent, higher elevation bushland communities on either side of the corridor 0 0 O O O O O O O O O O O O O O O O O O O O O O O 50 Graphs representing the seasonal changes in the above ground grass and forb standing crop (gm/m2) as well as seasonal net production values for three areas in the Mkomazi Reserve. The dashed lines refer to the stand- ing crap on previously unclipped plots within barbed wire exclosures. The solid lines represent the cumulative net production on sample plots which were clear-clipped before each growing season . . . . . . . . . 57 a. Distribution of the 10 ground transects in the northwestern half of the Mkomazi Reserve. The transects were subdivided into a total of 30 segments for more precise enumeration of habitat preference, density and movement patterns. The locations of the three sample plot areas are also depicted by the symbol 0 . 'b. The aerial transect grid used for the monthly aerial surveys of the Mkomazi Reserve. The starting, turning and terminal points were located at specific topographic features such as waterholes, drainage ditches, rock outcrops and artificial markers. Also included are the locations of the four vegetation exhlosures for the study of net primary productivity and standing crop . . . . . . 73 Visibility profile for the Gate to Ibaya ground transect. The inner profile was used to estimate the density of the small to medium sized herbivores, while the outer profile of considerably greater extent allowed the calculation of density for the larger species such as elephants and herds of eland and buffalo greater than 10 in number . . . . . . . . . . . . . . . . 76 LIST OF FIGURES -- continued Figure Page 11. 13. Visibility profile superimposed upon the circular-road ground transect. Again, because of the greater visibility of large species such as elephant, giraffe and herds of eland and buffalo an additional area of ‘visibility was calculated for these species. The expanded area of visibility in the upper left and upper right hand corners resulted from ascending to hill tops during the transect counts . . . . . . . . . . . . . . . . 78 The Dindira Dam Study area consisted of a 15.1 km plot completely enrrounded by steep mountains of approximately 500 m elevation above plot level. Only three entrance and exit corridors existed and animals likely to be driven from the area by the counting procedure could be enumerated from a small hilltop over- looking the area . . . . . . . . . . . . . . . . . . . . . 82 Quadratic response plots of a trendpsurface analysis of relative biomass density. The surfaces derive from a least-squares multiple regression procedure and may be interpreted as contour maps of biomass density for the respective areas of the reserve. The upper pair of lots were derived from the mean biomass density (kg/km2§ of three dry season aerial counts while the lower pair derives from the mean density of three wet season aerial counts. The left-hand pair of plots is based on total biomass while the right-hand pair refers to large herbivore biomass exclusive of elephants. Since the maps are based on aerial transect data the values only represent the minimum biomass estimate (not absolute) and are therefore best considered as indices to relative biomass distribution. Interpreting the maps is the same as reading a contoured tepographic map. That is, the reference contour (fife...) depicts an arbitrary density isopleth whose value in kg/km2 is given as ”Ref. con.” along with the plot. Density isopleths (contours) denoted by numerals (0-9) represent values higher than the reference contour; explicitly an increase in density by the stated interval amount for each ”edge" of the successively higher numbered bands. As an example consider the lower right-hand map which refers to the wet season without elephant biomass density (WeE). Note the reference contour of 30 kg/ which crosses the reserve in two places in the west central and eastern ends. Each edge of a successively higher numbered band depicts an increase in density by the interval amount (in this case 3 kg/kmz). Therefore the band or contour of 1's represents an isopleth of x LIST OF FIGURES - continued Figure Page 13. 14. 15. 16. (cont.) density between 33 and 36 kg per km2 while in the north- 'western section a short band of #‘s represents a density between 51 and 54 kg/kmz. The density gradient runs perpendicular across the contours. Contours denoted by letters represent decreasing density values for successive letters of the alphabet. Therefore, using the same example, the band of A's depicts a density of between 2# and 27 kg/kmz, while the B contour denotes a narrow zone of between 18-21 kg/kmz. Again, the decreasing density gradient runs perpendicular to the contours. The coefficient of determination (coeff. det.) derived from the least squares analysis if given along with each plot. Along with the example plot (W;E) a value of 0.181 means that l8.l% of the variation in biomass density was explained by the 2nd order regression equation of density on location. The square root of 0.181 yields the multiple correlation coefficient of 0014’26 “111011 is highly Significant e e e e e e e e e e o o._ 93 cubic response surfaces derived from the 3rd order least—squares regression equation of biomass density on location. The data are the same as those used in Fig. 13, but the higher order equation allows a "better fit". Thus, here, 20% of the variation in wet season non-elephant biomass (W;E) is explainable in location . . 95 Fourth degree response surface resulting from the multiple regression analysis of large herbivore biomass density on the x, y location in the Mkomazi Reserve. The data are the same as those used for the quadratic and cubic-responses. The highest coefficient of determination of analysis was obtained with this response (no higher order equations were fitted). Approximately 3#% of the dry season biomass density exclusive of elephant can be explained by knowing the x, y position in the reserve . . . . . . . . . . . . . . . 97 The seasonal relationship of numbers of animals seen along the monthly aerial transects and the mean monthly aridity coefficients of all stations. The curve has been ”Smoothed” by plotting a two-month running mean. The correlation between the "unsmoothed" numbers and the aridity coefficients~was highly significant (P = .02) and the monthly numbers of animals seen was Significantly nonrandom (P_l (n— 12 work. Two main movements caused folding along a north-east axis with cross-folding along a northwesterly trend. The movements have so disturbed the original sedimentary layer that a pseudo-bedded series has been formed. Other faulting created the present tilted block-line form and the large Pangani valley to the west, while the Pangaro valley within the reserve was fermed by subsequent faulting (Tanganyika Geological Survey, 1963). Occasional minor earth tremors suggest continuing move- ment in the present time. The open plains are underlain by Precambrian rocks covered with superficial alluvial Neogene deposits, including some calcareous tuffaceous material derived from the Kilimanjaro volcanicity and other deposits around Lake Jipe. Yet, where extensive erosion has occurred or where the sedimentary rocks have been thrust up, the ancient aneisses, schists and crystalline limestones strikbtthe surface. The Ikongwe hills differ by being composed of a meta-anorthosite with drawn out diopsides forming prominent linear bands. The conspicuous whitish, high sodium-content soils in this area have developed from labradorite (Tanganyika Geological Survey, 1963). ..* C L I MLA T E The factors controlling the overall weather patterns of East Africa are not yet fully known (Griffiths 1962). While some early observers tried to explain the conditions by continental weather fronts similar to those of temperate regions, others associated the phenomena with the Intertropical Convergence Zone (IIT C Z ), a zone of low pressure at the confluence of the NEE and S E trade winds (Thompson 1965, Kimble 1960). Although the seasonal variations in rainfall Seem to cerrelate with the movement of this convergence zone back and forth across the equator, 'variations apparently due to tapography and the great lakes of East .Africa modify conditions to the extent that they can not be said to follow the typical ITCZ pattern (Griffiths 1962). There is little or no evidence to suggest that rainfall in East Africa is associated with any moving air-mass systems and storms are believed to be the result of local developments (Thompson 1965). Since equatorial temperatures characteristically show little variation, variations in rainfall are a dmminant climatic factor, especially in a semi-arid area like the Nkomazi. Further, since the weather patterns are not associated with moving fronts, as is the case in temperate regions, the variance in rainfall is great and the deviations from the mean as well as the intensity of individual storms assume a major role in ecosystem flunction. As mapped by the East African Meteorological Service, most of the Mkomazi area lies between the 50 and 75 cm rainfall isohyets with the northwestern foothill area in the 75-100 cm isohyet zone. According to tha Thornthwaite classification, the area falls within the megathermal l3 14 climatic type (A‘) with a moisture index (D) of -40 to -20 and an annual water deficit of 80 to 100 cm (Carter 1954). TECHNIQUES During this study, climatic observations were made over a period of two and a half years. Eight standard, 5-inch (12.7 cm) diameter, storage type rain gauges were established in the western half of the reserve (Fig. 2). The gauges were buried in the ground with the upper rim extending 4 inches above the surface and the surrounding vegetation which might intercept rain was consistently removed. No gauges (at Ibaya and Kisima camps) were measured each day of rainfall except on a few occasions when no observer was on duty at the end of a month. Other stations were measured less frequently; and since the Kamakota gauge was over 100 km (by road) from the main camp, it was rarely measured more than once per month. Evaporation from the gauges did not appear to be significant since the funnel Opening extended below the water surface. Rainfall data were also obtained from four stations near the reserve, three of which are located on sisal plantations in the higher- rainfall foothills of the nearby mountains. A meteorological station including wet and dry bulb and maximum and Minimum thermometers, a cup-counter anemometer and a calibrated Gunn- Bellani radiometer was established at Ibaya camp in the western end of the reserve (915 m elevation). The thermometers were housed in a standard Stevenson screen which, along with the raingauge, anemometer and radiometer were located in a barbed wire enclosure. The anemometer was positioned 3 m above the ground. 15 'While the data from this station are important for defining local conditions and short-term variation, the long-term records of the East African Meteorological Station at Same, only 5 km from the reserve, have been used to estimate certain parameters. RAINFALL Although 2 years is a minimal period to measure such a variable factor as rainfall, some meaningful results were obtained. The composite monthly rainfall data collected within the reserve along with the data of the Same meteorological station for the same period are given in Appendix I. 1. ‘With the data of the Kisima station excluded because of its proximity to the Kisiwani mountains, a regression analysis of monthly rainfall on the eastewest location of stations yielded an east-west rainfall gradient. From the west end of the reserve eastward, there was a mean reduction of approximately 1 cm of rainfall per year for each 5'km distance; From this it is predicted that the Kamakota station in the center of the reserve would receive 15 cm less rainfall annually than the western end of the reserve. This predicted decrease is slightly less than the empirical results of the study indicate. Rainfall statistics for the eight stations during the period of StUdy'indicated that the calendar year totals lie between 55 and 65 cm for the western end of the reserve, with the Kisima area in the west- centmal section receiving about the same amount because of its proximity tO'the KiSiwani mountains. The more open steppe area around Kisima received about 10 cm less, however, and the short-grass prairie around l6 Maori (20 km further east) received only 40-45 on per year. Along with the rainfall statistics for the Kamakota gauge; the generally xeric conditions of the bush and the absence of drainage lines or erosion gullies suggest that this central section of the reserve must only rarely receive 40 cm of rainfall. Although no raingauges were positioned in the eastern half of the reserve, observations of storm patterns and vegetation conditions suggest a slight increase to 40-45 cm in the eastern end. To evaluate the rainfall conditions during the period of study with respect to temporal trend it is necessary to refer to the long-term data of the four stations outside the reserve (Appendix I. 2.). Based on a composite of 52 years of observations, the 1964, 65 and 66 totals were all below the average for the respective stations. It is therefore concluded that the period of study was relatively dry with respect to the general conditions for the area. Trend analysis of the data from the four stations outside the reserve revealed no significant monotonic or'cyclical trend in any of the data (Cox and Stuart's test for mono- tonic trend and Noether's test of cyclicity, Bradley 1968, pi 174-179). The seasonal pattern of rainfall in the reserve is one of a clear bimodal distribution with peaks of occurrence in March and October (Table 1). The total, however, is not equally distributed. About 50% 0f the annual amount occurs during the ”long" rainy season centered on the vernal equinox while only about 25% occurs during the ”short" rains associated with the autumnal equinox. A further 20% usually occurs in January and February as an extension of the autumnal rains or as an antecedent to the vernal rains. The remaining 5% occurs from June through.September as scattered showers during the long dry season. a a ,.-. a... .n‘].-. once! I a .-a lip... cleat- s-: a... ,-s 17 a. a $5.. a. a. was- tea -_ nee 1| H.mm w.mm howdy puebm o. H oo.HI N. H wa.ou Apoo theeunohv 0.0 H.: map mooq m. « mm.o+ m. H mic- m. a No.0- fissures memes home we: asses mass 0. H oo.Hu o. H oo.Hu Aoewueomv o.o o.o ......I..... he teen nwuwwma wwamwmd mwudwma ,. . a , .. ... condom .. . ... .. ... . ,‘ ,.‘..J..... OOOHLI.”NWHUHH.;AU. .mmooo andpdne nosed age .poneoa opdesnvonona opp hp popodooaeo no sense pudendum Apfi3_.&%ooo hufiufinu sees esp seam sandman nosed esp odes: so on daemofiew Hooomeom one oplnoMoa pom some Moe,eaomfih some: one .oeaoeom nausea: 0:9 as ooapepm wheoH one son homespun pooHeHHMeoe seepage pod Hammofioa Hosoneom .H edema 18 To substantiate this pattern the long-term data of the outside stations was referred to. But before such reference is valid it must first be established that the pattern of rainfall for stations within and outside the reserve is similar. Concurrent observations at the Ibaya station (within the reserve) and the Same Meteorological Station were made for over two years and the 28 monthly totals for the two stations are very highly correlated (P<:.001). Correlations of data from other gauges in the western end of the reserve with the Same data are also highly significant (P<:.01):and correlations of gauges in the central section of the reserve with other stations outside the reserve are also significant (P<:.05). It appears, therefore, that the long term Same Meteorological data can be validly used to describe the seasonal pattern of rainfall in the western end of the reserve. The 30 year monthly totals corroborate the bimodality of the pattern. Data fer individual storms suggest that 5-cm rain storms are quite common while only one or two lO-cm storms occur per year. The most intense storm observed was that of February 7, 1967 when Over 27 cm fell at Dindira Dam in approximately 30 hours: Normally the rains of the vernal equinox are more effective because of their mild nature while less penetrating downpours occur more frequently during the autumnal equinox and often cause severe runoff and erosion. TEMPERATURE Full time climatological observations were not possible, and even daily'temperature observations at standard times could not be assured. The maximum and minimum temperature recordings are valid, however, since 19 the mercury'column remains in the most extreme position until reset by the Observer. The maximum recorded temperature for Ibaya camp was 37.8 c while the minimum was 9.4°. The temperature data from the Ibaya station are not sufficiently complete to warrant calculation of mean daily, monthly or annual temperatures. The ambient temperatures were, however, highly correlated with those of the Same station and the afternoon (1600 hr.) temperatures were not significantly different. Data from the Same station are there- fore used as representative of conditions in the western end of the reserve. The mean annual temperature for the two full calendar years of the study (1965-66) was 23.10 with a mean annual minimum of 17.50 and maximum of 29.00 (Appendix I. 3.). Mean monthly temperatures calculated from the daily recordings, are most valuable for determining seasonal patterns; and the data show that July and August are usually the coolest months while the highest daily temperatures occur from December through February; The difference between the mean monthly temperatures for these “seasons, however, was only 5°. The maximum recorded temperature for the Same Station during the study was 36.8° and the minimum was lo.8° (Appendix I. 4.). The greatest absolute range in temperature during any month was 200 while the mean monthly range was 17°. In accordance with Chapman's Rule (a change in the mean annual temperature of approximately 60 for each 1000 meters of elevational Change, Allee e_t 3;]; 1949) it is hypothesized that the central section of the reserve averages 2.50 higher and that the eastern end has tamperatures about 49 higher than those reported above. SOLAR RADIATION As for several of the meteorological instruments, data recordings from the radiometer were too infrequent to warrant quantitative analysis. All measurements fell within the isopleths given by Thompson (1965) however, and the following figures represent elevation and latitude corrected interpolations from his Nairobi and Dar Es Salaam values. The mean solar radiation for the western, central and eastern sections of the reserve are approximately'4.5h x 103, 4.u7 x 103 and n.4o x 103 Kcal/mz/day respectively. The maximal daily values of about 5.7 x 103, 5.6 x 103 and 5.5 x 103 occur in February while the annual daily minima of about 3.4 x 103, 3.5 x 103 and 3.6 x 103 Kcal/mz/day occur during the cloudy periods of July. RELATIVE HUMIDITY The absolute amount of moisture in the air relative to the saturation density at any given ambient temperature is frequently used as a measure of the evaporative power of the atmosphere; This measure is refined considerably when air movement is included in the calculation, but patterns of relative humidity are also instructive as an index to evapotranspiration. Monthly maxima, minima and mean relative humidity figures have been calculated from the 1600 hnwwet and dry bulb thermo- meter readings at the Same station (Appendix I. 5.). The lowest mean monthly relative humidity values (40 to 50%) occur during the short dry season of January and February and the highest values (50 to 60% occur during the vernal rainy season of March-May. Although these mean values 20 21 appear high for a semi-arid area, the afternoon relative humidity rarely falls below'20% durhhg the most severe dry periods. ‘WIND Daily and seasonal wind patterns are important as they greatly affect evapotranspiration rates. Along with the increase in the evaporative power of the air there is a bending and flexing of leaves and stems which probably affects plant losses. The greatest wind movement occurs during the long dry season when a daily run of the anemometer frequently exceeds 160 km. Although daily wind patterns are largely associated with differential rates of heating of the air column and the consequent convection currents, seasonal patterns are determined by the S E and N E trade winds from the Indian Ocean. The S E trades are more noticeable and normally blow over the area from May to October, whereas the N E trades have only a mild effect from November to May. Significant air movement is usually not initiated until around 1200 hrs., but by 1600 hrs. the winds are considerable and gusts up to 15 Kph: frequently occur during the dry seasonh Air movement usually subsides shortly after dusk. ARIDITY COEFFICIENTS It is an oversimplification to describe the seasons of Eastforica as simply rainy or dry (Howe 1953). Integration of the factors discussed above, along with the hours and intensity of solar radiation, soil moisture storage and others produces an overall effect which is greater than the sum of the parts (Jowett _e_t_ al_. 1966). For instance, 22 a week in June with no rainfall is climatologically very different than a similar period in OctOber. Soil moisture is much more reduced during the latter period; and wind, solar radiation, temperature, relative humidity and day length all effect a much greater severity of conditions on the biotic components of the system. Evaluation of this whole array of factors is probably more important when dealing with natural animal populations than with geography or crop science, since animals are dependent upon a host of environmental require- ments over and above the water and nutrient balances which largely control vegetation growth. Furthermorgh doing the simplest quantitative analysis of animal ecology requires the use of some quantitative measure of seasonal climatic conditions. The more of the above considerations that can be integrated into a single index of climatic severity, the more meaningful will be any analytic results. The Thornthwaite classification (l9h8) of climate takes into account several factors other than rainfall while remaining calculable with a limited range of meteorological statistics. His potential evapotranspiration index contains expressions of temperature, day length and radiation, while using 4 inches (10 cm) as the mean available water storage capacity of different soil types. The overall index of aridity is calculated with the deficiency (evapotranspiration minus precipitation) as a percentage of need (potential evapotranspiration). Thus, in periods when precipitation exceeds potential evapotranspiration, there is a surplus of water and the index is positive whereas if evapotranspiration exceeds precipitation, there is a negative balance. The Thornthwaite measure and other equations based on mean temperature have been criticized on several grounds. The most important 23 of these is that they do not account for the lag of temperature behind radiation. But since soil temperatures on or near the equator are characteristically isothermal (weber 1959, Banage and Viser 1967) and since no short term estimates are made here, any error involved would seem at least partially obviated by the conditions of the study. Use of the Penman equations (Penman l9h8, McCulloch 1965) which are presently favored in East Africa (Dagg 1965) requires meteorological statistics not commonly available. Furthermore, these equations have also been criticized as inadequate (Holdridge 1967). In spite of its admitted defects (Thornthwaite and Hare 1965), the Thornthwaite measure of water balance is used here with the major purpose being to derive a comparative index of aridity for different seasons of the year and different areas within the reserve, any constant bias will be of negligible importance. Only rarely does the amount of precipitation exceed the need for any appreciable period of time (Table 1). As expected, however, the aridity coefficients are largest (negative) during the June-September dry season (-l.0 is the limit) when there is no appreciable rainfall and smallest (frequently positive) during the vernal rains. From Chapman's relation of increasing temperature with decreasing elevation and Kenworthy‘s regression equation fer mean annual temperature as a function of elevation in East Africa (Traphell and Griffiths 1960), it can be predicted that the annual potential evapo- transpiration in the center of the reserve is approximately 162 cm (Appendix I. 6.). This is over 10% greater than that of the western end (145 cm). 'When the lower amount of rainfall of the central section is also considered, the mean aridity coefficient for 1966-67 (-.71 t .10) was nearly'50% greater than that of the Ibaya station (-.h9 1 .1h). 2# From this I conclude that the general climatic conditions in the center of the reserve are about 50% more severe than those of the western end. THE GENERAL CLIMATIC PATTERN Beginning with the vernal, lowaintensity rains of February and March the ground is continually moist for a period of approximately 10 weeks. The humidity is sustained at a reasonably high level and the vegetation quickly reaches its asymptotic standing crop. The temperature gradually decreases as the season progresses and the sun moves away from its apogeal position. June is the month of transition into the dry season and south- easterly trade winds begin to blow as the rains subside. The sky frequently remains overcast until late morning early in the month, but clear nights and days prevail later on. By July, the weather turns cool and the year's minimal amounts of solar radiation and minimal temperatures occur. The grass dries out quickly as the winds increase and humidity drops. By August, the countryside is usually heavily burned leaving the blackened grass tussocks to absorb more heat. A period of intense desiccation prevails throughout August and September as the ambient temperature rises, nearly highest daily sun- shine hours and light intensities occur and wind velocity reaches the maximum. In mid October the first thunder showers of the autumnal rains occur and considerable erosion often takes place as a result of the heavy downpours and the lack of vegetation cover. As the vegetation regains its stature, humidity increases and temperatures are moderated by the cloud cover and evaporation of the rainfall. 25 The sporadic rains of November and December are followed by a short dry season of 6 - 8 weeks duration. The lack of rain during this period is accentuated by maximal annual temperatures and amounts of solar radiation (Thompson 1965). Thus in spite of the relatively isothermal conditions, Sharp seasonal differences exist; and there is little doubt that the long dry season from June through October is as severe an environmental stress on the organisms as the winter months of the north temperate regions are on mid-latitudinal species. These seasonal differences are likely to be manifested in various ways by the animals inhabiting the area. Along with the seasonal patterns, the rainfall and temperature gradients within the reserve also must cause differing degrees of environmental stress. In addition to the generally 50% greater mean aridity coefficients of the central section, the soils get shallower toward the central and eastern sections (profile G, was only 50 cm deep). Based on texture, structure and organic matter considerations these soils undoubtedly have lower water holding capacities than those of the west; If quantitatively evaluated, such edaphic factors would force yet a.greater divergence of aridity coefficients in the east versus the west. Similarly, compared to the northwest, the central and eastern sections have greater air movements and less cloud cover which must accentuate the evapotranspiration and overall severity of conditions. In total, climatological observations show appreciable differences bOtween the different areas within the reserve. These conditions surely induce great differences in vegetation and animal productivities as well as in animal densities and movement patterns. S O I L S To date there is no generally accepted classification of African soils or even overall agreement in naming the more important groups (Anderson 1963, Sys 1967). The soils of the region encompassing the Mkomazi Reserve have been variously described by soil researchers as: ”skeletal-montmorillonoid with CaCOB (semi-arid phase)" (Calton 195%); "kaolinite soils" (Spurr l95h); "red soil to calcareous black soil sequence, with intermediate soils rarely containing murram concretions and undifferentiated lowlying grey soil dominant" (Tanganyika Atlas 1955); and "brown to yellow-red sandy clay loans with laterite horizon" (Scott 1962a). In an attempt to unify the soils work being done in Africa and provide a single classification system for the continent, the Commission for Technical Cooperation in Africa (CCTA) sponsored the production of a continental pedological map and explanatory monograph (D'Hoore 1964). In this work, the\soils of the Mkomazi region are classified as "non- differentiated, ferruginous tropical soils (jd)." It seems clear that much of the confusion regarding these classi- fications results from over generalization and inadequate attention to the definition of terms. This, of course, is partially justified by the scale of mapping necessary. More recently, workers have attempted to define specific parameters for ”keying out" various soils (Makin 1969b, USDA 1960, 1967), and the American 7th Approximation to a Comprehensive Soil Classification System U511A 1960, 1967) is gaining wide recognition and support (Makin 1969a, :hfis 1967, 1968, Donahue 1970). I have classified the Mkomazi soils according to the 7th Approximation and, even if the more rigorous 26 ..4 .5. \ 27 classification into families, orders and groups provides little heuristic value, it makes available for comparison the considerable literature on similar, well studied.American soils. TECHNIQUES Seventy-eight samples were collected during early 1967 from 20 sites and 18 profile pits. The locations were chosen to represent the important vegetation and soil associations of the western half of the reserve. TWelve profile pits were dug in three catenary sequences from high on the slope to the valley below (see Figure 2). All pits exceeded a 3 ft. depth and most were dug to 6 ft. or more. Sampling was done according to instructions and procedures described in the USDA Soil Survey Manual (1951). Complete field descriptions including vegetation, surface and profile drainage, moisture, texture, structure, consistence, permeability, organic matter, roots and fauna were recorded. Color was determined by comparison with a Munsell chart. 'When profile horizons were not obvious, sampling intervals of 30 cm were used from the surface downward. Samples were chipped from the profile wall, bagged in cloth, air dried, and sent to the Northern Agricultural Research Center, Tengeru, Tanzania for chemical analysis. 7 Procedures of analysis followed those of Mehlich gtflgl; (1962). Bases were extracted by leaching the soils with normal ammonium acetate at pH 7.0, and exchange capacity was determined by displacing adsorbed amInonia with normal potassium sulphate and by subsequent distillation of the ammonia. Mechanical analysis was by the hydrometer method using a mi’t‘uzure of sodium silicate and sodium hexa-metaphosphate as a dispersing 28 agent. Total and organic phosphorus were determined by the method of walker and Adams (1958), and organic carbon by the Walkley and Black method, assuming 80% recovery of Carbon (Anderson and Talbot 1965). Based on the field descriptions and analytical reports the soils ‘were classified according to the keys in the supplement to the 7th Approximation (USDA 1967). THE CATENA CONCEPT AND TYPICAL SOIL SEQUENCES Milne (1936) noted that where the topography of an area consisted of a repetition of similar crests and hollows, there was also a recurrence of the soil sequences from one slope to another. He then coined the term "catena” for this succession of soil types along topographic gradients and later adopted it as a mapping unit (Scott 1962). A catena consists of three major complexes: the upper slopes termed eluvial; the lower, often concave, slopes termed colluvial; and the valley bottom or lowland termed illuvial or alluvial. This concept seems especially applicable to the Mkomazi as it lies in the region of Efilne's work; and field observations confirm that where climate and parent material are similar these topographical soil sequences are also quite similar (Milne 191w, Burtt 1942, Morison e_t al_- 19118). ‘Ihe pattern is so predictable that Mbrisonqgtflgl;'(l9h8) stated that, ”The monotony consists rather in the repetition of the same limited series of Changes over huge tracts of country." In practical terms, the three zOnes can be recognized by their red to reddish brown, brown to gray brown, and gray to black soil colors respectively. Although specific Parameter values vary considerably in local areas where parent material, cljomate or some other ecological factor has greatly influenced the do‘i'elopment, the concept is empirically sound. 29 Figure 3 illustrates a typical catenary sequence of soils from high on the lepes to the seasonally waterlogged drainageways belowa The nomenclature follows that of the 7th Approximation; and a complete listing of the names derived for the soils sampled along with similar soil series described in the U.S. and its territories are listed in .Appendix II. THE RED AND REDDISH BROWN SOILS The predominant soils of the Mkomazi are the medium-textured red earths (aridisols) which occupy most hills, ridges and fan slopes, and are characteristic of the freely drained areas (Milne 1936, Dames 1959). Ranging from rather heavy'sandy clays to lighter-textured sandy and silty loams they are all low in organic matter (usua11y<:1%) and erosive. Although surface runoff is often excessive, there is usually rapid water infiltration and profile drainage. Although predominantly yellowish-red (5YRh/8), local variations in organic matter and chemical makeup cause slight variations. Red coloring results from unhydrated iron oxide which is unstable under' moist conditions and therefore lost in the less freely drained areas. This coloring is accentuated in thecentral and eastern sections by more severe dehydration of the hydroxides, and it is more prominent due to the lower percentage of grass cover and the absence of a humic layer (Milne l9#7). These soils tend to be shallower and coarse textured high on the slopes, becoming deeper and finer textured with more organic matter lower down. In the northwestern area of the reserve these soils are very deep (h'm or more), but farther east on the Umba Steppe they tend to be shallow with frequent rock outcrops. 3O .annonpaomhnomH enema .aomhnomfi oawmz eonseodhpm made oavwmoadfinospgoa a op macmoa :.psoes o>flpeonpa popew>oannm age .eoa< hoopm mnwbcwm esp ado: obhomom fiamaoxz one we coflpoom pneumoznpnoz one 9H pcofipenw canaeamomop a mzoae mHfiom mo oesosuom haenopao age .m eenwea .3333. 2.5. e_greeseo 2%» coax—.03 «37.2.9. 883:0 3530.9. Bank son—>933 4:08.95 £2;on €38.95 con—30$ €3.28 0:: 6925.25 xx m" . T. m 0332 of we 5:09:52 23 :. mzom 3 85:03 nausea—so 3033 < 32 Typically, the red soils high on the slopes are Camborthids and :manifest eluviation by the absence of significant clay accumulations. But.as the slope decreases, there is an accumulation of silicon clays in an argillic horizon which puts the soils in the suborder of argids as opposed to orthids. If this accumulation of clay particles exceeds 35%, as is the case in the B catena near the Mbula study area, the class- ification becomes Paleargid. The G profiles toward the center of the reserve manifest erosion; and in the case of 62, even though an argillic horizon exists, the percent of clay is sufficiently low to change its classification to Haplargid. In the case of G1, erosion has so truncated the profile that the classification is changed at the ordinal level (to an Entisol); and the shallow, skeletal soil is described as a Lithic Tbrripsamment. In a few local areas the soils have a greater percentage of organic matter (from 1 to 2%) without a change of texture or structure. This tends to increase the color and chroma values and the soils are classified as Mbllic Paleargids. Although a number of factors may influence the amount of organic matter present, it is likely that the combination of lighter grazing pressure, less frequent burning and some- what less erosion explains this increase, at least in the area of the B catena. BRWN AND GRAY BRCMN COLLUVIAL SOIIS Normally, a narrow zone of colluvial soils lies between the red eluvials of the upper slopes and the heavy black clays of the valley bottoms. This band of transitional soil may be of negligible width and only perceptible by textural and structural characteristics or itrmay be ..4 ' .-.. l‘l ..1 33 of considerable extent. This texture may'vary considerably because of the differential deposition from above, but most are fairly heavy textured loams and clay loams with a somewhat sandy surface. They have high conductivities (200-300 mmh/cm.) relative to the more freely drained profiles above, and the cation exchange capacities also increase because of the increased clay and organic matter fractions. By being low in the catenary sequence they have escaped leaching; the profile drainage is slow; and yet they are sufficiently above the lower drainage- ways to escape waterlogging. Regardless of their relative unimportance due to the small area covered, these soils are valuable because of the ”edge-effect” of'high mineral reserves and favorable structural and ‘textural characteristics. Though commonly occurring in the intermediate catenary sequence, these soils at times occupy the whole surface of ridges and slapes and may blend with red soils at arbitrary, but well-defined points in an otherwise unifOrm topography (Milne 1936, l9fi7). The upland gray and gray brown soils are frequently firmly packed clays, but also blend to lighter textured soils with poorly developed structure. Soils of the ”C” profile sequence exemplify the characteristics of the elevated clays of the Mkomazi. Because profile drainage is poor and the expansibility coefficient is low, little dry season cracking occurs. The soils of this series tend to be lighter colored (5YR/l) than the lowland vertisols, and the whole group (particularly C1) is underlaid by an extremely deep (over 7 m) light gray alluvium. The extreme erosibility of these soils is at least partially due to the high amounts of sodium salts causing floculation and a low structural stability. as 34 HEAVY BLACK CLAYS OF THE VALLEY BOTTOMS Commonly known as "black cotton" soils in Africa these clays are usually'limited to drainageways and poorly drained bottomlands. They are products of considerable base accumulation and the formation of montmorillonitic clay particles which have high expansion coefficients. Characteristically containing more than 35% of these 2:1 lattice clay ‘particles, they are plastic and sticky'when wet and wide-cracking (up to 10 cm) and rock-like when dry. Because the lower horizons are subject to pressure when the soils swell, they are compact and nearly impermeable to water (Smith 1965); and since they occupy the drainageways" and depressions, they receive considerable runoff water becoming seasOnally inundated and waterlogged. Despite the fact that their water retention capacity is high the absolute amount of moisture available for plant utilization is very low (Smith 1965). Profiles A , A#’ B5, and D3 are representative of this soil type. All are classified as Pellusterts because of climatic conditions and their color chromas, but each is put in a separate subgroup for various reasons. Although these soils are very rich in the mineral elements of fertility as well as organic matter, salinity and alkalinity are often too great for preferred rangeland vegetation species; and vegetation productivity may even be suppressed by induced elemental deficiencies. Moreover, their utility is greatly limited by their physical characteristics and the gilgai microrelief may well affect ungulate locomotion. FERTILITY CONSIDERATIONS .Approximately 80% of the total reserve area is constituted by the freely-drained red soils (aridisols). All have greater than 50% base saturation in all horizons, and only one (profile E) of the 12 profiles analyzed has less than 85% while seven have 100% base saturation in all ‘horizons. The pH values were approximately neutral for all the samples analyzed with the exception of those in the same profile which was also acidic (pH = 5-6). This was a freely drained profile under short grass prairie and the pH is not unlike that of temperate prairie soils. The mean total milliequivalents of exchangeable bases per 100 grams (cation exchange capacity) in all horizons is 14.2 i 1.9. There is a significant increase (PN:.05) in the C.E.C. between the upper and lower horizons of the profiles and a significant increase (P<:.Ol) from the top of the catenary sequence to the bottom. In general, then, progression down in profile depth or down the catenary sequence results in an increase in exchange capacity and the percent base saturation. A similar pattern holds for organic matter. The mean percentage of organic carbon (x 1.72 = organic matter) for all horizons and all profiles of the red soils was 0.45 t .05. There is a significant reduction from the upper horizons (0.74 t .1) to the lower horizons (0.21 t .03) while there is a significant increase from soils high in the catenary sequence to the bottomland soils. The same pattern also holds for nitrogen in which the mean percentage for all profiles and horizons analyzed is 0.098 t .000 while the top 10 cm samples have a mean of 0.14 i .05 and the 30—40 cm horizon has a mean of 0.06 i .00. Although there is not an appreciable change 35 36 in the percentage N from red soils high on the slopes to those lower down, the vertisols of the valley bottoms have significantly more (0.102 t .00) than the aridisols in the upper catenary positions. The carbon/nitrogen ratio of the red aridisols (7.5 1 .52) is also significantly'lower than that of the clay vertisols (12.2 1 .50) of the valley bottoms. It is more difficult to describe the levels and gradients of specific ions in general terms as there is considerable local variation. Furthermore, the levels of any one cation may be misleading unless 'viewed in relation to the levels of other available cations since the ratio is frequently of great importance. Calcium and magnesium sharply increase in amount in the lower horizons of the profile near the bottom of the slope (profile B4’ Fig. 4), and the amounts in the heavy clay of the valley bottom far exceed those of the profiles above. This pattern also applies to Sodium and occasionally to phosphorus. No appreciable increase of potassium or manganese occurs while progressing down the catenary sequence, causing a considerable shift in ratios of these to other minerals like calcium. As a consequence, the wide K/Ca ratios of the bottomland soils may possibly result in potassium deficiency. Manganese, however, being a divalent ion is less affected by the change in ratio. Although it is generally not available in the lower horizons of any profile, it seems that its distribution reflects an accumulation in the surface layers by plant extraction from the deeper zones and a subsequent recycling. The vertisols of the bottomlands are very rich in total base elements; but as the exchange capacities, base saturations and percentages of organic matter and clay particles increase, so does the 37 Figure 4. A typical array of exchangeable cation profiles with respect to profile depth and topographic location. The vertical axis of each tier represents profile depth in cm while the horizontal axis represents the exchangeable cation level measured as milli-equivalents per 100 gm of soil. The five graphs of each tier correspond to the 5 profiles sampled in the B catena near the Mbula Study area. B} was a typical aridisol highest on the slope while B rep esents the vertisolic clay of the valley bottom.5 EXCHANGEABLE CATION PROFILES m.e.l 1” 9m. Calcium 5 lo ‘ 5 IO 15 5 N 5 ~10 15 15 30 45 60 ‘Mognesium s 2 5 2 2 s 25 - 5 D K A A Potassium m w to ° to lo Manganese .05 a: 15 OJ 01 B: 32 - B: 84 B: 39 concentration of sodium salts. Consequently, many'less freely drained soils have soluable salts at or near the surface. The elevated clays of the C series reflect concentrations of sodium salts; and whereas the accumulation in profile C2 is restricted to the lower horizon, profile Cl is a saline/alkali soil. The C/N ratio widens considerably in the heavy clays and reflects less availability of nitrogen than might be suspected from the absolute amounts present. Empirical results indicate that some nutrient deficiency or toxicity inhibits plant growth on several of these clays; and since Mh is generally low (or notymeasurable as in D3), it seems reasonable to suspect microelement deficiencies at least in the more alkaline heavy clays. DISCUSSION Inasmuch as the nutrient content of vegetation reflects the mineral status of the soil on which it was grown, the soil-wildlife relationship is important. Furthermore, the vegetation communities of East Africa seem extraordinarily closely associated with the underlying soil assoc- iations (Shantz and Math 1923, Burtt 1942, Morison gt 9;, 19%, Phillips 1929, Gillman 1949); and soil considerations play important roles in ecological studies of the region (Swynnerton 1932, Jackson 1954, Boaler 1966, Lang-Brown and Harrop 1962). More specifically, recent studies indicate that East African game distribution patterns are at least partially dictated by soil types, probably mediated through vegetation species composition and/or nutrient status (Petrides 1956, Anderson and Talbot 1965, Bredon 1963). 40 This is neither the place for a detailed probe, nor am I fully qualified to interpret many intricate fertility relationships. But a general discussion is presented relating the results of this study with the overall ecology of the area and with the results of other East African soils work. My results suggest no generally deficient element, and in accordance with other experimental work conducted in East Africawit is unlikely that the grasslands would show large responses to common N-P-K fertilizers (Evans and Mitchell 1962, Mills 1954). Even though increased grassland productivities have been obtained in East Africa by applying sulphate of ammonia (Evans and Mitchell 1962), it is frequently only the interaction and residual effects of treatment combinations which are significant (Evans 1963). As a general condition for East African livestock, Naik (1965) gives minimal levels of calcium and phosphorus as 5 m.e.% and 10 ppm respectively. Based on these levels only two profiles analyzed are deficient in Ca although several are marginal. Both of these are paleargids; but, in contrast, most vertisols of the valley bottoms appear to be only marginal or deficient in P according to the above level. Excesses of'manganese (causing toxicities) and other minor elements have posed problems in East African soils (Chenery 1954), but this is not likely to be the case in the Mkomazi. In fact, excesses of sodium and other salts in local areas are more likely to induce microelement deficiencies. Although the quantitative analyses of extractable elements are believed to be representative of the predominant associations in the 41 reserve, it is important to view the nitrogen figures with some reservation. Semb and Robinson (1969) and others have illustrated that the seasonal fluSh of'mineral nitrogen is particularly important in seasonally'wet and dry soils. Since the soil sampling for this study 'was conducted shortly after the onset of the vernal rains, it is possible that the nitrogen levels are not representative of the full calendar year. Scott (1962b) demonstrated a clear relationship of decreasing exchangeable bases with increasing rainfall in semi-arid areas and a reversal of this trend in areas receiving greater than 75 cm of rainfall. Subsequent to my study, Dr. G. D. Anderson, Soil Chemist of the Northern Region Research Center (Tengeru) extended my sampling work so that the combined Observations are representative of the entire reserve (Anderson 1968). These added samples corroborate the above conclusions;' but, more importantly, they permit interpretation for the full 130 km length of the reserve. Based on these observations, there is no significant correlation of percent base saturation with eastawest location in the reserve or with the presumed rainfall gradient as Scott demonstrated fOrEast Africa in general. Other gradients are more clear: (1) The profile depths of the northwestern foothill area greatly exceed those of the east and (2) the surface hardness and impermeability also increase markedly from west to east. In spite of no consistent gradient in organic matter content, (3) the bushland soils of the east and central sections generally contain less than those of the west. Neither root penetration nor the effects of microfauna are as great in the eastern areas and soil arthropod populations are substantially lower in these soils. Finally, 42 (5) because of their shallow and compact nature, their water retention capacity is substantially lower than the deep profiles of the west. Collectivelytheee results suggest that there might be large scale east- west effects of soils on the biotic components of the ecosystem over and above the differences due to local associations. V E G E T.A T I O N Several lucid descriptive studies of East African vegetation communities (Shantz and Marbut 1923, Phillips 1929, Burtt 1942, Mbrison .Eiuéla.19“89 Gillman 1949) provide considerable insight toward under- standing and describing those communities. ‘When supplemented by more recent classification works (Keay and Aubreville 1959, Trapnell and Langdale—Brown 1962, Pratt _e_1_:_ & 1966, and Aubre’ville 1966), it would seem that terminology, at least, would be well established. This is not the case, however, and because of a reliance on local vernacular such as “nyika”, ”machaka", ”miombo”, and ”mbuga' much of the terminology offers little to those not conversant in Swahili. The descriptive terminology used here largely follows that of the East African Range Classification Committee (Pratt 93 9]; 1966). The Mkomazi Reserve is encompassed by an extensive association of semi-arid bushland which occupies large parts of the Sudan and extends southward through Kenya into Tanzania where it meets the "miombo" wood- land of the more southern countries (Fig. 5). It is usually an assemblage Of woody plants mostly of a shrubby or bushy habit (i.e. branching or forking from the base); and, in the semi-arid and arid regions, these plants possess thorns or spines. Larger, clear-boled trees are dispersed throughout while the grass cover is generally'short and widely spaced providing only basal cover. Because of climatic and soil gradients, as well as the physiographic variation the vegetation of the reserve is quite diverse. The higher mountain areas of the northwest support dry'montane forest while the plains of the western area are covered with bushed and wooded grassland. 43 The major vegetation types of northern Tanzania and southern Kenya. The Mkomazi Reserve is contained in the type known as bush, bushland or "nyika". Reproduced from the Division of Overseas Survey D.0.S. 299E. Figure 5. . é XXIII-m TANZANIA D BUSH LAND LAND a n m A u m um L SCALE Inns." 46 Riparian woodland and dense thicket are interspersed throughout the reserve whereas the heavy clay drainageways support open, seasonally inundated grasslands. COMMUNITY DESCRIPTIONS Collection and preservation of plant specimens was done to become familiar with and describe the important species. Approximately 275 common species were collected and ferwarded to the East African Herbarium fer identification (Appendix III) while duplicate specimens 'were retained and catalogued for a reference and teaching collection. Toward the end of the study community type locations and boundaries were plotted On 1:50,000 maps. These were then reduced to l:250,000 and modified during aerial surveys of the area to provide more accurate delineation. For the purposes of this presentation only four basic communities are mapped (Fig. 6) since division of these into more specific types is largely a subjective evaluation of the effects of animals and man. It was not feasible to depict small areas of riparian woodland, bush thickets or other local variations. BUSHLAND: The most extensive and typical vegetation was bushland which covered nearly all the freely-drained, light textured soils (aridisols) under rainfall conditions not exceeding 50 cm per year. The elevated gray and black vertisolic soils under similar rainfall conditions also supported this community; in total bushland approximated 70% of the reservae 47 Figure 6. A vegetation-types map illustrating the location and extent of the major vegetation associations within the Mkomazi Reserve. As depicted, the upland dry forest represents approximately 5% of the total area and occurs on most of the higher mountain peaks. In reality, closed canopy forest only occurs on the mountains above about 1000 m elevation although the dominant species occur lower down. Similarly, included in the area depicted as bushland are localized areas of nearly open grassland (e.g. Maori) resulting from bush suppression. In general, the seasonally waterlogged grass- land follow the distribution of the vertisolic clay drainageways. —‘ JF‘N‘Y‘ Vegetation map of Mkomazi Reserve li|"1|?"‘. 1.317” I. it" 3' ($115.13 ~ Mamafl§£< gUpIand dry Forest g, ggWooded Grassland v l ;§:-_::Grassland " Mbuga' :Bushland 49 The bushland woody plants are mostly of shrubby habit, having a height of 7 m or less, depending on location. In the central and eastern sections, the can0py is frequently so low that medium sized ungulates (e.g.) impala) can not stand beneath it. Ground cover exceeds 20%, but is usually less than 40% unless approaching thickets (see Figure 7a). In the western areas where trees (i.e. one stem from the base) are the dominant form, Commiphora schilgperi, m 9215124.! A. etbaica and Albizia anthelmintica are the most frequently encountered species. Other species of m, Sterculia, m and Terminalia are also common and locally abundant. The most frequently observed emergent cleareboled (trees are Delonex £233, Adansonia di itata, Eflrina b_1£r_t_i;i_._l and gen—a volkensii. Bushes, shrubs and herbs abound in the understory while the grasses are short to medium height. Comon grasses are perennials of the genera Chloris, Di itaria, S orobolus, Hetero on,” Bothriochloa and Themeda. The bushland of the lower-rainfall areas of the reserve is dominated by bushes and dwarfed tree Species;- This is also usually the case where reasonably well-drained gray and black clay soils occur. The Commiphora A‘schimperii gives way to Q. campestris, M M and 2. M, 23541—8; abbreviata (and Q. longiracemosa and _G_r_el¢_i_g 222. while Cam Aristg‘sppq (M £22., m gpp. and Plat cel ium 3933133 also commonly Occur. The associated shrubs and herbs are again quite varied with the genera Te hrosia, Sericocomopsis, Indigofera and Hermanaia most frequently seen. The grass cover and productivity are usually poor in the drier areas, with Chloris roxberghiana, Cenchrus ciliaris, Sporobolus festivus', S. censimilis, HeteroBogion contortus and Aristida gpp. dominating. 50 Figure 7a. Typical Commiphora schimperi and C, campestris bushland with Cassia gpp., Cordia gpp. and Grewia gpp. sub-dominants in the central section of the reserve. Figure 7b. A typical association of bushed and wooded grassland (near the Dindira Study Area) in the western end of the Mkomazi Reserve. Figure 7c. Open Pennesitum mezianum grassland occurring on a heavy montmorillonoid clay drainageway in the northwestern section of the reserve. Adjacent, higher elevation bush- land communities appear on either side of the corridor. 1‘ w. +£177~4 ..2. ~". A: .» .V r. ~fx’vi- -1 r.- a a- as he I. ”a 52 BUSHED AND WOODED GRASSLAND: In the higher rainfall (3-50 cm) areas of the reserve the bushland is replaced by bushed and wooded grassland and this community covers the more freely-drained eluvial soils on the hill and mountain feathill fan slopes. The widely spaced, but always conspicuous trees and bushes have canopy covers of much less than 20% but they usually stand 10-12 m high (Figure 7b). The more common tree species are AEEEEE tortilis var. spirocarpa, A, etbaica, A, sene al, Platycepgalum zggpgg, 222212 salicifolia, and Malig,volkensii. Other species such as Ziziphus mucronata, Sterculia africana and Cappgris tomentosa occur infrequently. The grassland in these areas reflects high vigor and productivity, partly'because of the added runoff water from above. It usually exceeds a meter in height. Themeda triandra, Heteropggan contortus, Digitaria S22, and Bothriochloa radicans dominate. The more common bushes and shrubs are Combretum Egllgi(second growth), S, aculeatum, Acacia brevispica, Solanum incanum and Thylachium africanum. GRASSLAND: Open grassland areas usually occur on the lowland heavy clay drainageways ("mbugaS”). Because of water catchment from adjacent sloping terrain and the high water retention characteristics of these soils, they are usually seasonally waterlogged and free of tree growth. These grassland drainageways typically'form long, narrow corridors bordered by the bushland of the adjacent well-drained soils (Figure 7c). The dominant grass of these areas is Pennisetum mezianum, while other species such as Dicaanthium papillosum, Dactyloctenium ae tium, Schoenfeldia transiens, Ischaemum 3:33p, Sorghum verticilliflorum and and S, versicolor, Panicumlgpp. and Bracharia Spp, also occur widely. 53 Open grasslands also appear on the higher, more freely drained fersialitic soils, but usually only as seral stages or fire disclimaxes. These grasslands are very different from the seasonally waterlogged ones; and since they occur locally (e.g. around Maori) among the bush- land or wooded grasslands and consist of the same species, they are not differentiated on the map. UPLAND DRY FOREST: Almost all mountainous areas above 1000 m elevation are characterized by a closed canopy forest of 15-20 m height with a substantial growth of epiphytes. Although frequently termed "cloud forest" because of frequent envelopment in clouds, the rainfall is apparently too low to warrant this name. Calodendrum ca nse, Bracgylaena hutchinsii, Clerodendrum hildebrandtii, Albizia harveyi and A, petersiana dominant the canopy; while Sggpgp.dicho amus, Hoslundia o sita, Agggggugggkig, Haplocaelum foliolosum, Lonchocarpus s . and Stgychnos Sp, compose the understory bush layer. The common shrubs are Aspilia mossambicensis, Thylachium africanum and Solanum incanum. Although usually closed, occasionally open glades and less densely crowned areas support growths of tall rank grasses such as Chloris roxberghiana, (Panicum deustum, 2. maxilmlm and modon dactylon. The same species of‘bushes and shrubs usually grow down the mountain slopes to lower levels and frequently ferm dense thickets, especially'in ravines and gullies. While it appears (see Figure 6) that this forest type is of considerable extent, in fact, all the closed forests have an extensive peripheral zone of more open canopy. This is most probably a manifestation of fire encroachment and if fire were prevented the younger 54 age classes would soon fill in the canopy. Therefore, a designation such as ”fire-induced wooded grasslands" as suggested by Anderson (1968) might be justified on a more detailed vegetation map. RIPARIAN AND MISCELLANEOUS TYPES: Smaller areas of riparian woods and ground-water forest are important as game sanctuaries and frequently occur along the seasonal watercourses or in areas supporting a high water table. These forest remnants are dominated by Tamarandus Spgggg, Afzelia cuanzensis, Newtonia hildebrandtii, Terminalia kilimandscharica, .1. prunioides and Zizypgus mucronata while the understory species are commonly )Litpl strickerii, Hoslundia o osita, m bicolor, S. villosum, Ehretia taitensis and Haplocaelum foliolosum. All these species seem to provide palatable dryaseason forage for elephants and other browsers. The smaller rock outcrops and rocky lepes of the mountains are frequently covered by a highly xerophytic shrub, vellozia aequatorialis (or y, Spgkgi) of the Velloziaceae. It apparently needs very little water and is a common species in the driest areas of the reserve. The various areas of saline/alkali soil support interesting vegetation communities. One local area of considerable extent (soil profile D3) supports no vegetation at all whereas several other areas support salt tolerant species. The "miswaki", or ”tooth brush bushes'\ (Salvadora pgrsica and Sgpggg loranthifolia) are found only on these soils and may be used as indicator species. Another highly unique plant of these areas is ASSESS globosa of the Passifloriaceae. Occurring as a giant above-ground potato-like sphere up to 2.5 m in diameter, it has no leaves and is usually'covered by chlorophyllous spines and stems. This is one of the many dry-season water sources utilized by game, 55 particularly eland, but only after it has been "dethorned" by rhino. A similar, but smaller, plant possessing at least some leaves is Pyrenacantha malvifolia. NET PRIMARY PRODUCTION Grassland productivity studies were initiated in late 1965 and continued for 18 months until termination of the study in mid 1967. Four barbedawire 16 x 18 m exclosures were established in the northwestern half of the reserve with the locations representing different grassland and soil types and rainfall regimes. TWO plots (A and B) were centered in each of the exclosures so that a margin of 2-3 m separated the plots from the exclosure wire. Each of the plots was then divided into 12, 6 x l m subplots with iron stakes demarcating the corners of each. During the long dry season, plot A of each exclosure was clipped bare with hedge shears and the above ground standing crop was removed and weighed. Above-ground vegetation was then clipped from one of the 12 subplots of both the denuded and unclipped plots at approximately monthly intervals throughout the following year. The vegetation was bagged, returned to headquarters, and placed in an elevated, open-air, wire mesh drying structure completely covered with corregated roofing. Since the drying structure contained 16 wire mesh bins, the samples could be air-dried for two months before being removed and weighed on a single-beam scales. After clipping and measuring the above—ground standing crOp of the 24 subplots in each exclosure over a 12-month period, the exclosures 56 were reestablished in nearby areas and the studies continued during 1966-67. A marked gradient in standing crop and cumulative net production from the western to the central section of the reserve was established. In the northwestern section the grass-forb above ground standing crap on unclipped plots varied between 200 and 600 gm/m2 (x 10 = Kg/ha) depending on the month of measurement, but with a growing-season asymptote of slightly less than 600 gm/m2 (Fig. 8). This figure fell to approximately 250 gm/m2 in the short grass prairie area in the west central section and to approximately 200 gm/mz in the central section. The cumulative (seasonal) net production on denuded and unclipped plots also follows the same trend. Net above ground production on the denuded plots reached an asymptotic level of about 300 gm/m2 in the northwest, approximately 200 in the west-central and only 150 gm/m2 in the central section. Mean daily productivities (monthly accrua1/# days) for the different areas, seasons and plots have also been calculated (Table 2). Although short duration daily rates exceeded 6 gm/mZ/day in certain plots the overall rainy season mean is 1.93 1 .30 gm/mz/day. There is no significant difference between the rainy season daily productivities for the different exclosures nor for the denuded vs. unclipped plots. On the other hand, there is a marked difference between the cumulative seasonal net production between the denuded and unclipped plots. Whereas the 1966 denuded-plot productivities were about 310,165 and 140 gm/m2 in the three areas from west to east, the respective values for the unclipped plots were approximately 400, 240 and 170 gm/mz. These values are corroborated by the general trend depicted in Table 2. Even though the denuded plots show higher mean daily Figure 8. 57 Graphs representing the seasonal changes in the above ground grass and forb standing crop (gm/m ) as well as seasonal net production values for three areas in the Mkomazi Reserve. The dashed lines refer to the standing crop on previously unclipped plots within barbed wire exclosures. The solid lines represent the cumulative net production on sample plots which were clear-clipped before each growing season. Oran-land standing crop and not productivity MBULA (wooded grassland) 0 1:50qu RAINFALL ,2- "2-3-: . fi’l’ . N. 'I’ —(‘ O b I - ' - - - ~ ~ I: . o . p I O 3004 "" ° - I’ I O . I ’ ' ”'1 7”. I- O IW‘ e . P 70- . so: . co- . ao- . son ° .. I I I I I I I I I I I I I T I MAORI ”V4 (short-grass prairie) . - soo- auooma RAINFALL . __. P “- -‘,"'" b w . . p 0 ° . a . ‘ ' - -"-- zoo-I ’ ,. or f - ’0’ O [I . 0’ 0 . . l 1”" / p O I 70" b C so- . ao- , .. 10‘ . ”-1 h I I I I T I I I I I I KAMAKOTA ml (semi-arid bush) ’7‘. mi OJBOII IAINFALL . . ‘ _ ' e / M“ “---1 _____ —__ ~.—-~-—"( b 0/ e g ' o In“ P m D so- _ .- 1 _ so- P v D u "IAIiIJIJIAfiIoTNFFIJVFIHrAI-I 59 Table 2. Mean daily net above-ground primary production for different seasonal periods, rainfall regimes and clipping treatments. The first area receives about 500 mm of rainfall annually while the second and third receive about #09 and 350 mm respectively. All values are reported as grams/m . Treatment Seasonal period combination Nov-Jan Fequpr May-Jun Jul-Oct Mbula excl. denuded +1.11 +1.61» +1.85 -0. 38 unclipped -0.22 +0.97 +3.08 +0.38 Maori excl. denuded +0.05 +0.71 +1.90 -0.04 unclipped +1.08 +0.82 +1.12 +0.22 Kamakota denuded +0.74 +0.90 +1.20 -0.16 unclipped +0.55 +1.00 +2.38 -0.05 60 productivities early in the season, the unclipped plot values generally surpass them within a couple months. The time lag affecting the unclipped plots is especially important later in the season when production continues later on and frequently buoys the long dry—season mean above the zero point (sustained production early in the dry season exceeded late season attrition). Two of the exclosures were located on water shedding sites while the third (Kamakota) was on a level vertisolic soil. Not illustrated are the data for the fourth exclosure which was located on a receiving site at the bottom of a small hill (near Kisima) in the west central ‘section. The asymptotic standing crop of unclipped plots in this area during the growing season was 780 gm/m2 (i 89.8) while the seasonal productivity reached an asymptote at #15 t 26.5 gm/mz. There is little doubt that these figures are surpassed by the local areas of Panicum and Chloris on similar receiving sites in the northwestern area of the reserve. Appreciably more precipitation occurred in the first half of 1967 than in 1966. The productivities reflect this, as no plot sampled in 1967 appears to have reached its asymptote by June while all plots had already surpassed their 1966 productivities. RANGE ANALYSIS In attempting to quantify the differences in range conditions for various regions of the reserve, range evaluation techniques developed by the U.S. Forest Service (Range Analysis Handbook, Region 2, 1968) were utilized. The technique involves using a .75-inch (1.9 cm) diameter .‘w “.. , u. an. v-1 .1. I- x V 9‘} ‘\.. ‘-:. i .' 4’ 61 iron 100p to measure the frequency of "hits” on various grassland components while walking along compass bearing transects. A ”hit" is that species or item occupying more than 50% of the loop area when the 100p is lowered to the ground every second step along the transect. By always lowering the loop in a guide notch placed in the observer's shoe sole and by not looking at the ground as he walks along the transect, the human sampling bias becomes negligible. A set of 100 such measurements constitutes a sample, and the mean number of times that the loop hits the base of a perennial grass or forb is termed the plant density index. A ground cover index is then derived by subtracting the number of hits on bare soil, erosion pavement (pebbles<2. 5 cm diam.) or rock from 100. Although largely qualitative, an assessment of range trend is made by considering the different values of the plant density and ground cover indices along with the occurrence of litter, erosion pavement, species composition and plant vigor. Index values derived from the results of ten sets of 100 samples each from three transects in the Dindira study area show a mean plant density index of'33.8 and a ground cover index of 92.1 (Table 3). Despite the cloSeness to the permanent water and the severe dry season trampling, the mean incidence of bare ground was only 7.7 per 100 points. Significant changes in the grassland structure occur from the western to the central section of the reserve. In the western area nearly'3#% of the hits were on perennial grass bases while the value for the grasslands around Mbula was 21.5 and only 16.3 for similar soils in the central section. Conversely, the bare-ground values increased from 7.7 in the west to 25.5 around Mbula to 58.4 in the central section (Table 3). The plant density (basal coverage) and ground cover indices 62 dm n saw m4 .... n4: wé n men manee 3333.3. dead avoxaaam 9n n Nd in n mém N.N n 33 2H8 2283...; 8.3 serum 33: o.~ n mam n.m n mi 54 n mém 38 028324 3.3. n83 33: En n he a; 1.. duo NA n mam ennee 038232 8.2 harem 21.83 pnoao>am nowmono Nomad Home“ omhv Hwom no canon» onam noboo ennono hfifimnoe pecan Una nowpdooa .Aapoxaaau op anfipzdn .o.fiv pace on pa93.aonm avenue mnofipfimnoo we pneflvdnm a monooflmnoa one wonky Hwom naflfisdm,so£3_ .nafipavomo> coop no 0>HH so open Had op anomon nomad no>oo ensonw can mafia: madmadm OOH non annom no mommanm Hawnnonom no smpfinz Ho nopa::.naoa on» on mnemon News“ hpfimnoe pnaam 0:9 .obnomom fiaasoxz can no adona oonnv an anonno pnavnapm ndonp endfnofipdpnoo nofipauomo>.mo nonfivnH .m adage .‘w " 'I .‘ . e‘e‘ “’ -: A. 1'".“ , . .... .. . .. e_’. ...h -n .Fn h ‘-b a" .‘e. a‘fi .,"'~ kfi". .I . tn 63 for the seasOnally waterlogged grasslands of the west are higher than the upland grasslands, but this is only infrequently the case in the central section. Along with the higher incidence of bare ground in the central section, the frequenoy of erosion pavement also increases while the frequency of hits on litter decreases. These index values certainly manifest over-grazing and consequent erosion as well as lower rainfall conditions. Since few quantitative range evaluation standards have been deve10ped fer East African rangeland, it is not readily apparent what meaning the index values have other than for comparison of areas within the reserve. But as a crude approximation, the standards established for the foothill shrub community by the U.S. Forest Service (Range Analysis Handbook, Region 2) seem comparable (Hemingwayigt_gl;gl966). From these ratings (excellent, good, fair, poor and v. poor), the grass- land of the western Mkomazi would be considered "good" while that of the central section is rated poor to very poor. DISCUSSION Because of the methodology employed, the net production values reported here are minimal estimates. Because of asynchronous growth patterns of the different species and the inherent inadequacy of a monthly sampling interval, the true seasonal production values are likely to be about 10% greater than seasonal asymptotic values (Weigart and Evans l96h, Golley 1965, Kelly'gt_§l;.l969). A crude correction factor of plus 10% may be used if comparison to more accurate eStimates by rigorous sampling techniques is deSired. a l-‘ _ . l" _oe - ..v ‘ l . ‘1‘- "‘F ..eu Q '79 e..- 64 Taking different rainfall regimes into account, the primary production estimates obtained from this study compare favorably to values Obtained elsewhere in East and Central Africa (Brockington 1961, Harker 1961, Naveh 1968a, b, Anderson and Naveh 1968, McKay'l968). Nearly exact denuded plot predictions are obtainable from the Serengeti' precipitation-productivity curve established by Braun (1969). Of considerably greater interest, however, is the relationship between the denuded and unclipped plot productivities. The "clipping effect”, or the subsequent realization of greater productivities by clipping, mowing, grazing, burning or otherwise removing the above- ground standing crop is a well established range management phenomenon (Stoddart and Smith 1955, Curtis and Partch 1950, Hopkins 1950, Hadley and Kieckhefer 1963, Penfound 1964, and Kucera 93 51; 1967). The fact that this effect was not observed under these conditions is of both pragmatic and heuristic interest and is discussed later. The idea that mowed and burned areas produce a ”flush" of new growth quicker than unburned areas was seemingly substantiated. The quick response of the denuded plots is likely to be a consequence of the increased soil temperature resulting from exposure as well as the greater growth rates generally exhibited.by small or young organisms and populations (voisin 1959). On the other hand, the terminal lag of production in the unclipped plots likely results from the greater moisture retention capacity of the more shaded and cooler substrate under unclipped conditions. The relationships between the different plant communities and the large herbivores are discussed in the pinultimate chapter. However, it is important to stress that primary production per se., or even the 65 vegetation community structure, does not fully explain the inter— relations with the large herbivores. Grazing and browsing successions do exist (VeseyaFitzgerald 1960, 1965, Talbot 1963a, Gwynne and Bell 1968) and therefore several community types are important to single animal species at different times of the year. In contrast to the ideas of Janzen (1967), I found that a highly asynchronous vegetation "leafing" and flowering pattern provided abundant opportunity for temporal patterns of utilization. The large animal species have adapted to and are very dependent upon these patterns. Interestingly, the large herbivores have evolved anatomical and physiological mechanisms for most of the exigencies. The gamut runs from simple adaptations to thorns and spines and the successful utilization of'myrmicophytic Agggi§.species which is generally not the case in the American tropics (Janzen 1966), to the more sophisticated physiological adaptations to strychnine (in the form of Stgychnos £22., Burtt 1929, Lawton 1968) and the many alkaloids of the Solanaceae. ‘Whereas species of Solanum are rarely browsed in the neotropics, (D. H. ‘Janzen, pers. comm.) Solanum incanum and Solanum _sp. (taitense?) appear to provide an important part of eland browse in the Mkomazi. Plant- animal relationships are therefore of obvious importance to an under- standing of the community structure. The effects of fire on the environment are of'major concern to range ecologists the world over. I do not propose to discuss the pros and cons of burning rangelands in any detail here and the interested reader should consult the excellent review articles and over 1000 literature citations included by Shantz (l9#7), Comm. Bur. Past. Res. (1951), Ahlgren and Ahlgren (1960), west (1965), and Daubenmire (1968). 66 Aside from the almost undisputed effect upon bush encroachment there seems to be one other concept of overriding importance. In North America, fire has distinctly different effects on primary productivity depending on the rainfall regime. In areas which receive greater than aboutjéO-60 cm of annual precipitation there is a positive relationship betweenzburning and productivity (Ehrenreich 1959, Hadley and Kieckhefer 1963, Kucera 23.21;.1967). Areas which receive much less than 50 cm precipitation per year almost always reflect a reduced production after burning (Aldous 1931;, 1935, Elwell at. al_._ 1941, Hopkins 932 a_1._ 19%). It remains to be seen how long an enhanced productivity can be sustained by systematic burning in the high rainfall regime, but experimental work in Illinois has established a positive relationship for at least 4 years (Hadley and Kieckhefer 1963). Lay (1956) reported a five-fold increase in dry'matter the first year after burning, but this had declined to a two-fold increase by the third year. Two explanations may account for such a relationship. One, where annual productivity potentially exceeds decomposition, the positive relationship will hold (Olson 1963). Therefore, in the higher rainfall temperate areas it is predicted that there will be a ”burning” or ”clipping” effect since the removal of standing crop facilitates further production. In the lower rainfall regimes (frequently with concurrently higher temperatures) where decomposition potentially exceeds annual production there will be no clipping or burning effect. Thus in 26 years of annual burning trials in central Kansas, Mo Murphy and Anderson (1963) recorded reduced productivities 16 times. These 16 years of reduced production correlated with the years of below average annual rainfall. 67 A second, equally plausible explanation, is that a subtle shift from moisture to space limitation occurs as the higher rainfall regimes are approached. ”Thus available moisture may be a limiting factor in the low rainfall areas and this is further accentuated by herbage removal. 0n the other hand, forage removal from the possibly space-limited stands of the (higher rainfall areas would enhance productivity. Within the Mkomazi, the effect of fire on bush encroachment is well .established (see Figure 22a, p. lh8). Furthermore, from the vegetative exclosure experiments it is clear that there was no positive "clipping effect”. From this it is concluded that annual fires in the Mkomazi are likely to have a degratory effect on production while a l+—5 year burning cycle may be necessary for bush control and the maintenance of highly diverse vegetation communities. u ..~‘ ANIMALS The overriding impression gained by most visitors to the Mkomazi is one of abundant giraffe (Giraffa Scamelopardalis), Coke's hartebeest (Alcelaphus buselaphus) and elephant (Loxodonta africana), with only slightly fewer numbers of impala (Aepyceros melamms), eland (Tauratraggs m) and buffalo (Smcerus caffer). Less ubiquitous, but locally common species are zebra (Eguus burchellii), oryx (M M), steinbok (Raphicerus camestris), gerenuk’ (Litocranius walleri) and Grant's gazelle (Gezella gran—ti). "Rhinoceros (Diceros bicornis), lesser kudu (Strepsiceros imbarbis), not rare.“ Although lion (Pantera _Jfl) is the most numerous large carnivore, leopard (Panthera M33), cheetah (Acinom jubatus), hunting dog (lac—aw M13) and hyaenas (Crocuta crocuta sand Hype—net M) are all present. The more common smaller mammals include mongooses (Herpgstes ichneumon, g. sangu_ineus, Helogale undulata, Eggs _m1_mg9_ and Ichneumia albicauda), gerbils (Tat—ea robusta, Taterillus osgoodi), ground ”squirrel (m rutilus) and the vervet monkey (Cercopithecus aethioEs) while the) monitor lizard (Veranus exanthematicus), puff adder (_B_i_ti1_§ arietans) and black-necked cobra (m nigracolis) are dominant reptiles. Tsetse flies (Glossina spp.) and ground nesting termites (Macrotermes bellicosus, Odontotermes gpp. and others) are conspicuous and ecologically important invertebrates. The avian fauna of the reserve is varied and spectacular. The bushland community supports a great number of colorful and noisy AAdspecieisyrpossiblymore than any other habitat type (Fuggles-Couchman 68 69 1908, Mereau 1935). In numbers of species and individuals, the doves (Columbidae), starlings (Sturnidae), hornbills (Bucerotidae), and weavers (Ploceidae) predominate. The bird nomenclature is that of Mackworth-Praed and Grant 1952. The mammal nomenclature largely follows Swynnerton and Hayman 1951 with reference to Ellerman 1900 for rodents and Best 23 El; 1962 for big game animals. COLLECTION AND IDENTIFICATION Where species habits and habitats are known, important ecological insights can be gained by studying the relative abundance of various species present on an area. Throughout the course of the study, mammal and bird species were collected and identified. Common breakback traps and mist nets provided the bulk of small specimens, but a .010 gauge shotgun with dust shot, a .22 caliber rifle with scope, night-lighting techniques and the analysis of owl pellets were also used. Specimens were sent to various specialists for identification (see Acknowledgements). In total, 233 bird species were identified (Appendix IV. 1.), although the list is admittedly far from complete. Certain taxonomic groups (e.g. sunbirds), habitat-related species such as those of the montane forests, and activity-related species (e.g. nocturnal) are conspicuously absent or only poorly represented on the list. Seventy-eight species of mammals were identified (Appendix IV. 2.) and six other known species are believed to be present, but not positively identified. The number identified is considerably below those of the more well studied areas of East Africa, but as further .1 .- 70 study is undertaken in the Mkomazi, the list should expand. The mammals so far recorded generally represent the lowland, more extensive, areas of the reserve; and considerable effort should be directed toward the montane, riparian and other less extensive habitats to establish the faunal composition of these communities. The Standard North American trapline (Calhoun 1959) was frequently used to obtain comparative estimates of trapping success and, in general, the success rates (1-2 per 100 trap nights) suggest that small- mammal densities are very low compared to temperate areas where success rates ten times greater prevail. Few quantitative data are available for the assessment of trends in the density of the various populations. It is certain however that at least three large species have been recently extirpated from the reserve. Observations of the greater kudu (Strepsiceros strepsiceros) were made by D. G. Anstey, Game Ranger, in 1952 (Anstey, pars. field notes); and also by the acting warden in official letter no. 051/8/06 of 17 October, 1955. This, along with my finding of a greater kudu horn in the reserve, supports that species' former presence in the reserve. Both the colobus monkey (Colobus angolensis) and the corcodile (Crocodylus niloticus) were recorded by various game department personnel as late as 1957 (annual report, Game and Tsetse Division 1950, District Ranger's report 1957). Neither of these species has been recorded since 1957 and none presently exist in the area. Evidence also suggests that at least two other ungulates formerly occurred in the reserve. Swynnerton and Hayman (1951) report records of sable antelope (Hippotragus niggg) at Lake Jipe and Kisiwani which lie only a few km to the north and south of the reserve respectively and are .\ .- M ”i i- .4‘ 71 connected by the North Pare Mountains which extend through the western end of the reserve. Sable are still occasionally seen east of the reserve. Eastern whiteabearded wildebeest (Egrggn taurinus) were also common in Tanga Province and the area presently occupied by the reserve in the 1930's (Game and Tsetse Division annual report 1932; R. Bradstock pers. comm.). No wildebeest occurred in the area during the fifties and early sixties. Twenty were restocked in 1966. What little evidence exists regarding the cause of these extirpations suggests that heavy cultivation, overgrazing and illicit hunting along the Umba River are responsible for the serious decline in the riparian woodland, permanent water, and thus the demise of the crocodile and colobus monkey. The hypothesis of general habitat degradation and hunting is frequently suggested for the loss of the kudu, wildebeest and sable antelope; yet the introduced wildebeest population is growing after an initial decline and hence the habitat of the western section is apparently suitable for this species. NUMBERS, DENSITIES AND BIOMASS TECHNIQUES Four techniques were utilized to estimate numbers and densities of the larger mammal species of the reserve. These were: 1) ground transects with associated visibility profiles; 2) aerial transects; 3) demarcated sample study areas; and 0) sight-recording maps for the less abundant species. 72 Ground Count Transects: Especially because of the absence of roads or tracks, initial studies were limited to the northwestern section of the reserve. Ten ground transects patterned after Hahn's walking cruises (Hahn 1909) and similar to those used.by'lamprey (1963) were established (Fig. 9a). Four of these, varying in length from 18.5 to 26.5 km, were located in mountainous areas; and animals were counted while walking along cleared and demarcated paths. Six other transects, from 15 to 68 km in length, traversed the open bush and counts were conducted while driving at slow speed in a 0awheel drive vehicle. Each of the ten transects was subdivided into segments for more precise estimates of density patterns, variability, and movements. An attempt was made to conduct each count at monthly intervals, but certain transects were counted more frequently and others were sometimes missed because of the impassible soil conditions during rainy seasons and of non-functional transport. In total, 377 counts were conducted along the ten transect routes, representing over 2000 km walking and 5600 km driving distances. The counting technique was simple. More than one observer was always present; and when driving through the grassland or bush, one or more observers stood in the back of the open vehicle to facilitate animal sightings. After a sighting was made with the unaided eye, 7x02 binoculars were used to count and to classify the animals into sex and size categories. Records were tabulated on standard forms with a system of parentheses and superscripts denoting herd composition. The starting and finishing times were recorded along with the extant water, vegetation, temperature and sunshine conditions for each count. Figure 9a. Figure 9b. 73 Distribution of the 10 ground transects in the northwestern half of the Mkomazi Reserve. The transects were subdivided 'into a total of 30 segments for more precise enumeration of habitat preference, density and movement patterns. The locations of the three sample plot areas are also depicted by the symbolD. The aerial transect grid used for the monthly aerial surveys of the Mkomazi Reserve. The starting, turning and terminal points were located at specific t0pographic features such as waterholes, drainage ditches, rock outcrops and artificial markers. Also included are the locations of the four vegetation exclosures for the study of net primary above-ground productivity and standing crop. LOCATION! OF TIN W0 IIANSICIS . lOCAIIONS OF AERIAL IIANSECI’S unn MAP, :1 “an: "or arms town MABO VEOIIAI’ION Ixcmsuns n u d n. l.. M- ii a- l.‘ 75 Visibility Profiles: To convert the number of animals seen to density figures some estimate of the area surveyed was necessary. Visibility profiles (Figs. 10 and 11) were established for several of the transects; and although it is not suggested that all animals in the area were counted, the numbers seen represent a minimal estimate of the number occurring in the respective areas. These profiles were constructed by sending khaki-clad game scouts with white handkerchiefs in their hip pockets in perpendicular directions from the transect line and subsequently pacing the distance at which they became obscurated by the vegetation. These distances changed with vegetation conditions, of course, and they also varied with the size of the animals involved. Theref0re, two profiles were established with the inside distance applicable to small and medium sized animals and the outside profile representing the area in which elephant, giraffe and herds of eland and buffalo greater than 10 individuals were visible. More-or-less circular visibility profiles were also established for hilltop observations; and after integrating with the linear profile, the sometimes wierd-shaped, areas were measured by a grid overlay of known scale. Aerial Surveys: A dual-seated Piper Supercub airplane was made .available in January 1966 and.monthly aerial surveys of the entire reserve were initiated at that time. A system of 18 permanent transects crossing the reserve transversely and spaced at 6 to 8 km intervals was established (Fig. 9b). The starting, turning and terminal points of the transects were located at specific topographic features such as water- holes, drainage ditches, rock outcrops, or artificial markers along the reserve boundary. Similar features, as well as peculiar trees, Figure 10. 76 Visibility profile for the Gate-to-Ibaya ground transect. The inner profile was used to estimate the density of the small to medium sized herbivores, while the outer profile of considerably greater extent allowed the calculation of density for the larger species such as elephants and herds of eland and buffalo greater than 10 in number. ‘ V! ----- Game transect ........... Area in which If ........... ungulates were ----------- visib|e(1270 ha.) Additional area in ' which elephants were visible (3703 ha.) Blind areas ...... VISIBILITY PROFILE FOR GATE TO IBAYA TRANSECT Figure 11. 78 Visibility profile superimposed upon the circular-road ground transect. Again, because of the greater visibility of large species such as elephant, giraffe and herds of eland and buffalo an additional area of visibility was calculated for these species. The expanded area of visibility in the upper left and upper right hand corners resulted from ascending to hill tops during the transect counts. Visibility Profile for Circular Road Transact Game transect .................. Area in which ungulates were ....... : - ' vmble (3626 ha) "‘3: ...... Additional area in which elephants were vielble(8340 ha) - Blind areas 80 vegetation community boundaries and compass bearings also were used as route markers. Since the total linear distance of the transects was about 560 km and required a flying time of'more than 0% hours, the combined transects were divided into three nearly equal segments. The three segments, totalling approximately 190 km each, were flown on consecutive days at monthly intervals. The counts were normally started about 1% hours after sunrise. Flying speed was held constant at 120 km per hour at a standard altitude of approximately 100 m. All animals seen along the transect routes were tallied on stand- ardized sheets and locations of the larger species, as well as concentrations of game, were plotted on maps. A portable tape recorder was occasionally'used to facilitate recording. Upon sighting large herds or concentrations of animals, the pilot would circle in a counter- clockwise direction and climb to an altitude of 200 m or more while the observer counted and rechecked the number of animals below. After counting such a group, the original position along the transect was regained and the normal census procedure resumed. In addition to the systematic counts along the transect routes, high altitude ”scavenger hunts”were performed later in the day to (further assess the numbers and distributional patterns of the various species. Since these flights were not systematic, the data can not be quantitatively analyzed and can only be used for subjective evaluations. Sample Plot Study'Areas: To provide more accurate estimates of animal densities and seasonal changes three sample plot areas from 10 to 15 km2 Veachwere established in mid 1966. The westernmost of these was located “at the permanent water source of Dindira Dam while one of the others was 81 near a semi-permanent waterhole (Mbula) and the third (MZara) represented an area with only seasonally available surface water (see iFig. 9a). (These areas were demarcated by large drainage gullies, roads, distinct vegetation-soil type boundaries, or in the case of the Dindira area, the surrounding mountains (Fig. 12). Two of these areas contained hills elevated 50 m or more above the surrounding area. Prior surveillance from these vantage points made possible an accurate tabulation of those animals likely to be driven out of the area by subsequent counting activity. Menthly counts (biweekly for Dindira) of these areas were carried out far over a year using a censusing technique similar to that of Petrides (1955). (After initial surveillance, the landrover with observers was driven back and forth over the area at approximately 50 m intervals until it was certain that all animals of reedbuck size or larger had been enumerated. The location, movement and exact composition of all herds was plotted on maps to obviate recounting errors. It is firmly believed that these direct enumerations contained negligible bias or error. With the assistance of several game scouts, simultaneous aerial surveys and ground counts of the samearea were performed on numerous occasions. Thus three censusing techniques could be compared for accuracy: the initial hilltop estimate, the complete ground count, and the aerial survey. Sight Recording Maps: Because of the very low density and the infrequency of observation of several species, sight recording maps were kept. When kept for long-periods it was possible to deduce, within reasonable limits of confidence, the minimal number of certain species 82 .oono one mnnnooano>o mopHHfin Haosm d Bonn ooponoasno on canoe ondoooonm mnnpnsoo onp an done one song no>flno on on hHonnH cadence one oopmflxo mnoofinnoo pfixo one oononpco oonnp haze .Ho>oH pon obono now9o>oao 8 com hHopoSfixonan mo annonQSos moonm an oooQSOnnsm haopoamsoo poam an H.mH o 90 oonmflmnoo oon<.hpdnm Eon onfionfim one .mn onsmnn sea 2:25 he? >>eon seas. lull use. an: 0.0: no.4 nun-m EGO 15.0.20 ..- v L... '0' I - an t ‘ ‘L:‘ 0"- can 80 which was present. Peculiar markings and deformities of individuals were also noted, and in certain cases these animals served to indicate the minimal.movements of the animal involved. The sight location records made during the monthly aerial surveys also provided data for describing migration patterns of the larger species. Biomass Calculations: Biomass is defined as the extant organic material present per unit area. In the following sections, however, the term refers only to animals, and more specifically, to the large herbivores. In dealing with a system in which many large species occur, the total biomass may'be closely approximated by summing only the large dominants. For example, one adult elephant weighs approximately the same as 105 rats; and therefore the total contribution of small vertebrates and invertebrates is probably less than the sampling error involved with only the large herbivores. Further, since the half-life of dead animate tissue is probably less than a day in East African systems, the total animal biomass (alive and dead) is closely approx- imated by the standing crop (live only). The terms biomass and standing crop are used interchangeably therefore, although the figures quoted refer specifically to the latter. Multiplication of size class abundance within species by the approximated weights of each size class is the best available estimate of total biomass per unit area (De V03 1969). In cases where size-class recordings were not feasible, the total species number was simply multiplied by the adjusted mean weight for that species (Talbot 1960). This was the procedure followed for all aerial survey data reported here as well as for buffalo and the three small ungulates: dik dik, steinbok and duiker. 85 All weight estimates were taken from the literature. Even though great variation exists between the estimates reported for different areas and by different personnel, it is believed that modal values have yielded a suitably accurate approximation. When several estimates for different-aged animals were not available, the smaller size class values were derived by reverse extrapolation along a logistic growth curve. The mean species weights, derived for use when the animals were not classified by size, consist of the sum of the products of the size class weight estimates times the mean recorded class frequency of the population. Several compilations of weight estimates (Meinertzhagen 1938, Blancou 1962, Rebinette 1963, Sachs 1967, Ledger 1968) have provided the basis of the weight estimations. Analysis: Subsequent to the collection of'data on standard forms, a tabulated computer codification scheme was established. This systematic game count data were then transferred to approximately 35,000 computer cards for analysis. All computations were performed by Michigan State University's Control Data 6500 facility and an Olivetti programmable deskatOp computer. RESULTS Relative numbers: The western end of the reserve contained greater numbers of'most species than did the central or eastern sections. As - exceptions, however, zebra and oryx tended to utilize the west-central and central portions of the reserve most, while elephants displayed a generally random distribution, with greater numbers in the east during the rainy season. Despite seasonal and species variation, the mean 86 number of animals seen per unit distance (or area) over the entire course of the study clearly was greater in the western end. The mean number of animals seen on all ground and aerial transects in the western end was 9.6/km while the central and eastern sections averaged 6.0 and 0.3/km respectively. This gradient oflrelative numbers is further confirmed by considering the number seen by the three counting techniques within precise zones along an east-west axis through the reserve (Table 0). The two exceptions to the smooth gradient (Table 0) represent the sampling of seasonallyawaterlogged grassland in otherwise bushland conditions. Both areas supported seasonal concentrations of zebra and oryx. Although the eastawest gradient in relative numbers greatly over— shadowed other distributional patterns, there was also a north-south trend. The areas along the Kenya border invariably supported greater numbers than comparable areas along the southern boundarya Thus when none gradient is superimposed upon the other, the overall gradient ran from northwest to southeast. Further evidence of these gradients is included in a discussion of biomass patterns. Densities: The highest game densities of the reserve occurred in the northwestern corner around the permanent water of Dindira Dam. The mean annual density of large herbivores in this area was 12 animals per km2 (Table 0). The density decreased rapidly, however, in all directions. Ten km to the south and east the mean annual densities were 7.2 and 6.3 animals/kmz respectively. The overall density dropped to less than 6 .animals/kmz in the west central section and the bushland of the central 87 m.e N.a o.mn spend oamaom m.0 H m.H N.0 H w.H 0.0 H.0.n 5.0 H 0.: 0.0 + m.m n.0 + 3.0 uncommonp Hdfino¢ 0.0 H :.HH m.H H 0.5 m.0 H 3.5 N.N H n.0H opooocanp onsonm mmnuem omumm mules carom .omuen onuo posses nonsense an an hnaondon snowmen one Bonn oonopnwa . an non hpfimaom- zoos one anomonmon open #on modem onp oHan.poomnonp mo an non nonssz noes one pnomonmom oven poomnonp one .obnomon onp no unconson nnovooz.onp Bonn oonapmfio op cofipeaon an 505nm onp mo oonsoo onfinzo one nobo mosvfianoop manpnsoo muonnab one an Ampnanmoflo mcfiodaoxov moonooon massage mo nonssn zoos one .0 canes 88 and eastern sections supported an annual mean of less than 0.5 animals/ kmzu, With the exception of buffalo, which reflected an extremely clumped distributional pattern and concentrated at the permanent water during the dryseason, most species densities range from less than 1 to about 3.5 animals per km2 (Table 5). The densities for all species combined ranged from dry season values of 23.7/km2 around the permanent water of the northwest to 3.0 in the west-central and much less than l/km2 in the eastern sections (Table 5). 'Absolute numbers: Estimates of the absolute numbers of large animals present in the reserve (Table 6) were deduced from a synthesis of: 1) the monthly totals observed from the aerial surveys and sight recording maps; 2) extrapolations from seasonal mean densities for the different habitats and areas; 3) long term sight records of relatively shortaranging species such as rhino and waterbuck; and 0) empirical knowledge resulting from over 5000 hours of field observations in the area. Even though approximate, a high degree of confidence is attached to these estimates. Elephants were the most numerous animals (Table 6) while hartebeest, buffalo, impala and eland followed in that order. No other non-migratory species numbered much over 250. Whereas Grant's gazelle, oryx and zebra showed seasonally high numbers the resident population of the reserve was much lower. Relative biomass distribution: Since the body weights of the different species vary from a few kg for the smaller ungulates to several thousand for elephant, it is frequently more instructive to analyze biomass patterns rather than numbers. An area which contains 89 sx\mx e.nma e.anem e.nme~ c.0em m.nmen N.mosmn eccscnm aspen mm.e na.m so.“ mn.n oe.m na.m~ .os aspen OHeH 3. N“. ”We m“. 30 ea“ “$.38 e as. o o o mm.n «seen 0 e so. 0 e om. essences: e 0 ac. e e no. sense mm.n so.m me. no. em.m mn.m scoopesasm oo.n mm.m mm.m on. en.m no. snooze e nm. em. 84 84 an. finesse e on. me. e on. em. onnsnnc sm.n na.n mm. No. on. Hm.n essences mm. o o e e no. scene 0 e o o 0 $43 enemas anon: oasnz anemone anon: oddnz onwondm condom 9oz. condom hnm ‘ I w .nopes, engenders haaonomoom 5Hco can o>nomon on» no nonpoom Honpsooupmoz on» an done once: one one oonson.novaz psoaoanon uflaom a neon oopaooa an aono aasn: onH .hpwannodndbe nopa3.pnocoanom «0 done no mpsomonaon one obnooon one no nonnoo anopmoznpnon one an panacea an Adm .wnmv eon< madam annenna one .mnomdom ant one no: wannee o>nomom Hudson: one «a moons modem oonnp an mopedsmnd nofiea one new man non maesonn one anonapz .m oHnue 90 Table 6. Minimal population estimates of the major animal species inhabiting the Mkomazi Reserve. The numbers of those species for which a range is given represent the dry and wet season values, with the highest number occurring in the wet season. ‘ r buffalo 750 oryx 100-000 bushbuck 100 ostrich 250 eland 500 reedbuck 50 elephant 500-3000 .rhino 05 gerenuk 250 'watérbuck 150 giraffe 250 zebra ,100-000 gazelle (Grant's) 150-600 lion 80—100 hartebeest 1000 cheetah 35 impala 600 hunting dog 25 O kudu (lesser) 250 hyaena 91 only a few elephants may support an equal or greater biomass density than another with several hundred smaller ungulates. This tends to be the case in the Mkomazi. Before developing the relative biomass patterns a short description of the analytical technique is given. Mapping of continuous ecological variables offers considerable analytical advantage over attempts to discretize and then test or explain the variation. Several techniques such as the least-squares fitting of polynomial equations exist for such mapping endeavors. The technique used here is that of trend-surface- analysis for which computer programs are available (O'Leary'gtual; 1969). It is an application of multiple regression and has been used extensively in geology, systematics, and more recently, ecology (Sokal 1965, vandermeer 1966, Marcus and vandermeer 1966, Sneath 1967, Gittins 1968). As used here, the north-south and eastawest locations in the reserve are denoted by an X and Y value representing the mean biomass per sz at that point is established. Thus for every datum two independent variates (the X, Y coordinates) and one dependent‘variate (biomass density) exist. To systematize the biomass data a grid of squares 10 km on a side (100 kmz) was drawn onto a profile of the reserve. In so doing a system of th subplots was established. The biomass values for each datum represent the mean density in that area for three aerial counts. Since only the aerial transect data provided complete coverage of the reserve, the biomass densities used in this section are drawn exclusively from those data. Therefore the values given are only minimal estimates and are best treated as index values. 92 Solutions to the regression of biomass density against location can be obtained for a number of different degree equations; Thus, if only a linear response is plotted, the general regression equation will be that of a plane suspended in 3-dimensional space. It will take the form of: A . )’=(3. + (3, X. +G,><,+E where J is the predicted value of the dependent variable (density), (35 the intercept of the plane with the vertical axis, £?1Xi = the partial regression coefficient of density on the X1 coordinate times the X1 distance from the origin, 6 2X2 = the partial regression coefficient of density on the X.2 coordinate times the X2 distance from the origin, and where (E = the residual error of prediction not explained by the two partial coefficients. All lines (responses) deriving from such an equation are linear since they are contained in a flat plane. But generally, linear responses do not provide a very good fit to biological data. Consequently, the percent variation in Z (the dependent variable) explained by the X, Y coordinates (dependent variables) will be low and the % attributed to residual error ((E) will "be high. From this it might be expected that the higher the order of equation becomes [3:g. quadratic (Fig. 13), cubic (Fig. 14), quartic (Fig. 15), etc;7 the greater will be the % variation explained by the equation (X and Y'with their powers and products). The % variation attributed to residual error will decrease correspondingly. From each equation, the regression coefficients, the coefficient of determination and the multiple correlation coefficient are derivable. To evaluate the hypotheses that there was: 1) a dominant eastawest tfiomass gradient; 2) a Subdominant north-south gradient, and; 3) a 93 .psaoamasmam magmas ma neanz wwd.o mo pnowofimmooo nowpoaonnoo oanflpdss onp opaoflh HmH.o mo pooh onodum one .nowpooofl no hpflmsop mo nofipodvo scammonmon noeho mam onp hp ponfioamxo mo: hpwocop mmoaown cw coapoflnob onp mo fia.ma non» mnooa HmH.o mo odHo> o Amlzv wean oamaoxo one npws mnoH< .png nooo nee: macaw sobfim ma mfimhaono moaosvm poooH one Scam mobfinov A.pon .mmooov nofiponfisnopob mo psowowmmooo one .ondopnoo onp op noHSOHvsoaaoa omen pnofipoaw hpfionob wnwooonooc onp .nfiow¢ . Bx\mx Hmlma noozpon mo ones Scene: o mononom adopnoo m one oafinz .man\wx mm new 3N nookpon mo hpfi not o opofiaom o.< mo econ onp .oaaaoxo oEoo‘onp wnfimd .onomoaona .vononmdo onp we oaoppoa obfiooooedm new mosao> hpfimnon mnfloooaoon nnomonmon maoppoa hp nononop mpdopnoo .mneopzoo onp moonoo noHdOfienoanom mash pnofinoam hpfimnop one .men\mx 3m nso Hm nooapon hpfimnon o manomonmon o.: mo econ phonm o soapooo snowmoznpnon onp cw oawns max pom mx on use mm noozpon hpfimnop Ho neoHQOmfi no oncomoamoa m.H mo ndopnoo no econ one oaomoaone .Amsx\mn m omoo mflnp nflv pndoso Hopaopnfi onp kn hpfimnom ca omoononfi no m90flmon neon nononesz nonmwn hHo>Homooodm o mo owno noom .ovco neopooo Ugo Honpnoo poo: onp aw moooHQ 03¢ QH obnomon onp momooao newns Nan\wn on he adopnoo oononomon onp opoz .AMIZV hpflmnon mmoBOfln peonaoao peonpws common no: onp on mammoa nofinz mos pnonwpnmwn noon onp nopwmnoo onBoxo no on .obnon poaonas: nonwfin hHo>Hmmooo5m onp Ho :ompo: nooo hem pndoBo Hopnopnfi nonopo onp an hpfimnop ca ooooaonfi no hapflOflHmNo madopnoo oonohomoa onp nonp nonmfin mosao> pnomonaon Amlov maonossn hp noponop Amazopnoov mnpoHQomH hpfimnom .poam onp npfiz mnoao :.noo .mom: mo cobflm ow Nex\wn cw oSHo> omon3 npofimoofi hpwmzom hhoapfinno so opofimon a .:.i&® nsopnoo oononomon onp .mw 9on9 .moe Ofinmoamomop noadopnoo o wnwnoon mo osom one mfi egos on» mnapoamaoan .moapsnfiapowp mmoEOfln obwpoaon op moOfinnw mo ponovfimnoo noon onomoaonp one one Aopsaomno ponv oposwpoo mooeown ESEfinHE onp psooonaon hano mosaob ony open poomnoap Howaoo no women one ones onp oosfim .opnonmoao mo obfimsaoxo omoBOfin oaobfinaon owaoH op whomoa nwog psonlpnwfln onp oHfins moosoan Honey no women ow mpoan mo nfioa bnonlvmoa onH .meSOO Howaoo noooom nos oonnp mo hpfionop sooa onp Eonm mo>flnon Mama noon on» oHfinz magnoo Hoflnoo sooooo man oonnp mo Amsn\wnv hpwmnon mmoEOfin nooB one 809% nobflaov ones mpoam mo nwom Momma one .obnomon one mo moono obfipoommoa onp mom hpflmnop mmoBOfin mo mama neopnoo mo boponmaopnfl on hos Ugo ondvooonm :ofimmoamon oHQHpHSS moaosUmlnoooH o Scam o>waon mooomn5m onH .yhnfimnon mmoEOfin opwpoaoa mo mamhaono oooMASmlvnoap o mo mpoam omnommon caponvodo .MH oasmwm eo-e-e-o-oeeoeueooaooono .o oboe...u...a...u ..ob.bo.uoa.uoa.b.s.aub.suuu n a '0 an an a o e. up as . o. ‘0‘ Ezemvsvo eve naaunnsaana~nse aa~s~n ...—.vuuoarupnuno «unpaun'dn—n .0. .. C: I G. asses no. CT~€I.I.‘.I .u. recreaoececeoe 3 u.. .ceo¢0cco-saccao¢ o: .. Octa- coco. .. a. C a ... too out ... a. CI 0 «C 0: —'—o-.u- one .. do I .0 :- r—vuvr.u—u-avuu so. .. a. no a...' so. .. no flewws I .... QQ‘VWHHQQ w.- " ...:- ........ .......:..:22.23:: 0!. 3.30.0 ...:::: i waOmmwu U._—(¢O<=O 95 .noepoooa hp oHnonflonxo ma AMsz mmoSOfiD pnonaofiolso: noooom no: me soapofiao> onp mo mom .oaon .osne .apfie noppon: o oonHo noepofico nopno nonmwn onp pen and .mflm ca now: ooonp mo oEom onp one ovon one .cowpoooa no hfifionop mooeofln eo sofipodoo scammoamoa monosvmupoooa nonno man on» gene cobwnop moooendm omzommoh ownno .dH onsmfim «.0. ml’ .3 ...-8 he . unnnneannrnhnus _ per?- ...a..‘. . 212:}...- .2... e o.a.a ..q..‘....< a. ......-.. 1:: q :- q. ‘- ...: ...a.a . .«aaq<.aqq.- . ,1 an e u.a.¢.¢.-.. :2: A a. :1 mmZOawwu 03:0 97 .o>nomon onp cw nowpfimom hV.N one mnwzonx en confieamxo on see unenmoao mo o>fiosaoxo hpwmnon omeEOfin someom en? onp eo fidm eHoveBfiNonma< .Anoppwe one; escapedvo nonno nonmfin onv oonommon mane nae: noneepno mes memeaene onp mo nofipenfienopon eo pnoaoeeeooo poonmfln one .osnop ma mnfievnoo noepedvo onp envenonm nae onozom npflz .wta an n... +.K.®+ex.unm.u8noe onp .Ho m...” ponfieporo onos opHdmon omonp nOHnB Eone coepedvo nowomonmon monedvm neeoa one .momnommon cameo use cepeneedv one now come ooonp me osem one one epen one .o>nomom Heesonz onp 2H soapeooa h .x onp no epflmnon mmeSOHn ono>wnnon omnea mo memeaene scammonmon oHaHpHHE,onp sene mnevflsmon ooeendm omnommon oonwoe npneom .mn oneMnn 1.. _ v , 5.11.2.3: _ . _ _ anaae xxxxxxxx _ _ Z xxazxxxx . . ml! auaouuuc u J 00 3 WI ..m Uuuucoueucu ...-a! . uuuuuguauouvuuc c uu a: aw ...».- sonaauooveuu a uu as aw at: .....aa . 00 a. ...v .3000 e um an .wu unusuaaaaaaaaaaaaaau. a uu .9 was auubsuuaazbaaauaua3:33: H a a a: new. . ...-.8 s .. 2. women $31.33;: a 3.2.2:“... ..u 2. 13333.33?! :— . a. 3 :2. £31.33 :93: 0Q douse“ ... :— 25: BESS: = u —: I". .— 3“... : . : =22}: ’ n. =2 3.: a - uu ...—2:2... ...—ESE? u uuu 2593952. uuuuu ,A". u . us n—nnn nuns: n his. uuu" 2. uuu .853 33 83» 5333333 2...: uuuuuuuua a2... .nnrnnnan— 2323:... nun :— an: an...» an n .- Auk rest: I. 32:2. saga-usiataaxvsxxz ...-.....5}!;. a]: .2; L. . . . / lid}.- ,17 lxxlfilll sewn}; ! | L _ _ r 3...... 3o 3.3.”...53... : u s. t . .I3U1U1uafl 1 I L» . 1.3.3.312. . a w D fatten”... Ct ......uttnuu....tv.. Gib. .....| 3 be uaLui-l hum—ntvtu. . i L: ..LL: L. ul.ua(v,u~ xL. . umZOamws 0.2.340 98 significant difference between the wet and dry season densities, two analyses were required. That is, a wet season plot (lower pair of Figures 13, 14, and 15), and a dry season plot (upper pair of Figures 13, 14, and 15). But since elephants were known to play an important role in the overall biomass pattern, analyses including (left-hand pair of Figures 13, 14, and 15) and excluding (right-hand pair of Figures 13, 14, and 15) elephants were called for. A Consequently,-four density maps had to be prepared to fully'explain the relative biomass distribution pattern-(dry with elephants 9 ENE, dry without elephants = D-E, wet with elephants = WWE and wet without elephants ='W;E). I There is indeed an eastawest biomass gradient within the reserve (Figs. l3, 14, 15). The gradient is most pronounced during the dry season (upper pairs of-maps) and exists regardless of whether or not elephant biomass in included in thefcalculations. During the rainy season (lower-pairs of maps), on the other hand, thewelephant biomass changes the picture completely. When the elephants are included, (WWE), there is a marked depression (low biomass density) in the center of the reserve with increasing densities in both directions. If elephants are excluded from the analysis (W;E), the castaweSt gradient is largely restricted to the northwestern section of the reserve. As mentioned previously, the north-south gradient is subordinate to the east-west gradient. Therefore, during the dry season when the east- west gradient is strongest, the north-south gradient is barely perceptible (upper pairs of maps, Figs. l3, 14, 15). During the rainy Vseason, on the other hand, when the eastawest gradient is less obvious; the north-south gradient appears most pronounced (lower pairs, Figs. 13, 99 14, 15). In the treatment combination of wet season without elephants (34.15:) the neruhuseuth gradient dominates the distributional pattern. In only two of the combinations the linear multiple correlations between density and location was not significant. In all cases, the multiple correlation (“alcoeff. dot.) became highly significant at the 2nd (Fig. 13), 3rd (Fig. 14), and 4th (Fig. 15) degree responses. The percent variation in biomass density explainable by knowing the x, y location in the reserve is given by the coefficients of determin- ation (coeff. det.). Thus, with the 2nd order equation (quadratic response, Fig. 13) only between 8.2 and 18:1% of the variation is explained by location. As the degree of the equation is increased, however, the more local variation can be explained. Thus with the 3rd degree responSe (Fig. 14), between 12.8 and 20% of the variation is explainable by the x, y location. The percent variation explained by the 4th degree response (Fig. 15) is generally greater than 25, and about 34% of the variation in the dry season biomass of herbivores other than elephants is accounted for. Absolute biomass densities: Although the relative biomass distribution maps reflect the overall pattern within the reserve, more precise estimates of the standing crap were obtained from ground transect and sample plot evaluation. Whereas the highest biomass densities approached 13,000 kg/km2 for the Dindira study area during the dry season (Table 5, P. 89), the mean annual biomass for the same area was 5548 kg/kmz. The mean annual standing crops for the Mbula (semi- permanent water) and the Mhara (seasonal water) study areas were 1934 and 707 kg/kmz respectively. The density values derived from the ground transects and.visibility profiles corroborate this trend and suggest 100 that within the western end of the reserve, concentric density isopleths radiated from the Dindira area% At a radius of about 10 km, the mean annual biomass was only 2000 kg/kmz; at 15 km from the permanent water the mean annual density fell to about 1000 kg/kmz. The mean annual biomass over the entire area of the reserve was about 1200 kg/kmz. This amount is equivalent to 0.7 elephants/kmz, and in the eastern half of the reserve elephants make up over 90% of the total biomass. Therefore over a vast area (approximately 750 kmz) all other herbivores combined sum to only about 150 kg/kmz. This is approximately equal to 1 oryx per kmz. MOVEMENT AND TEMPORAL DYNAMICS Previous statements and data have alluded to strong seasonal fluctuations in the numbers of animals in the reserve. As the only system of counts covering the entire reserve, the monthly aerial tran- sect data provide the best overall index to this pattern. Trend analysis of the longereterm ground transect data provides a more accurate assessment of local variation while still further refinement has been achieved by plotting herd locations and game concentrations as they moved across the transect and study plot grid. Seasonal movements of animals in the Mkomazi were characterized by a major north-south movement of animals across the Kenya border and an almost equally strong east-west movement within the reserve. While only the north-south migrations accounted for overall changes in numbers of animals, the east-west movements greatly affected seasonal distribution patterns a 101 North-south international migrations: When the total numbers of animals seen on the monthly aerial transects were plotted against time (Fig. 16), a significantly nonrandom (P<::05) time trend was evident. These monthly totals are also significantly, but negatively, correlated (P = .02) with the mean monthly aridity coefficient for all stations. The highest numbers of animals occurred during the months of highest rainfall. The pattern is, of course, seasonally cyclical. ‘Within the overall annual pattern, two distinct types of migrant pepulations were evident. The elephant numbers fluctuated with dramatic presence-absence pulses while the numbers of zebra, oryx, and to a lesser degree, Grant's gazelle showed more moderate sinusoidal fluctuations. The north-south movements of elephants were characterized by aggregates of’many herds moving as a unit. An eastern population of about 1500 elephants migrates northward across the international boundary during the drier months (Harris 1968, watson 23:31:.1969). Their occurrence within the Mkomazi is highly correlated (P<:.01) with extant vegetation and water conditions as measured on an ordinal scale. The resident population of the eastern half of the reserve is very low during the dry season and thus the presence or absence of the migratory animals causes dramatic fluctuations in total numbers and biomass. A distinct western elephant population, approximately the same size as the eastern, inhabits the region west of the central hill mass. 0f the total number in this pepulation only some 600 migrate seasonally while a considerable number remain in the reserve. These dry-season residents tend to congregate in the mountain foothills within range of Dindira Dam:' The migrants move into the Lake Jipe area of Tsavo Figure 16. 102 The seasonal relationship of numbers of animals seen along the monthly aerial transects and the mean monthly aridity coefficients of all stations. The curve has been "smoothed" by plotting a two-month running mean. The correlation between the "unsmoothed" numbers and the aridity coefficient was highly significant (P = .02) and the monthly numbers of animals seen was significantly nonrandom (P<:.05). zone-1 2500'“ Number of Animals 2 O 7 2 ? F’J v.2 r-a P.‘ p a. ’0. ’0, b.‘ .9 3 900: Wow sane 104 National Park in Kenya. Since the difference between the wet and dry season totals of the western half of the reserve is only twofold, the fluctuations in total numbers are not nearly as great as those produced by the eightfold difference in the east. The north-south movements of the zebra, oryx and gazelle are as seasonally predictable as any but for several reasons their changes in numbers are much less pulse-phasic. First, the distribution pattern of these animals is much less clumped than the aggregates of migrating elephants; they tend to move as small independent herds. 'They also appearrtO‘be much less dependent on surface water and thus their arrival and departure times is less strict. Whereas elephants seem to cross the Kenya border at all points with no specific movement routes, the ingress and egress routes of zebra, oryx and gazelle are more or less restricted to the four areas where open grasslands extend over the Kenya boundary. Therefore, early in the rainy season the number of these species increases on the grasslands along the border (Mbuga ya punda milia, Maori, Kavuma and below Kwamkala ridge in the very eastern end): As the season progresses they gradually mave scuthward.until they reach the southern boundary of the reserve. They inhabit these southern reaches fer 6510 weeks during favorable periods, but as the grazing pressure increases they gradually retreat northward. Their numbers then increase in the northern areas again until they finally leave the reserve for lack of dry season water and forage. East-west movements: Distinct from the north-scuth migrations are seasonal eastawest movements. 0f the animals remaining in the reserve during the dry seasen; the elephant3‘buffalo, giraffe; gazelle and eland all tend to congregate around the permanent water of Dindira Dam. 105 Therefore, when the mean monthly densities of the three sample study areas are analyzed with a tweuway analysis oftvariance; (Table 7), there is a significant difference (Pet.05) between the pooled wet and dry season means as well as a highly significant (Pk:.001) interaction effect. That is, during the dry season the mean density around the permanent water greatly exceeds that of the other two areas. But during the wet-season densities in the two areas away from the permanent water exceed that of the Dindira area. Because of’their highly-clumped distributional patterns and relatively large body size, buffalo movements deserve special mention. Over 90% of the total population occurred in three large herds from 175- 300 animals each. Although the three large herds moved as independent entities, it was frequent that two and sometimes all three of the herds (totalling from 300 to 750) inhabited the Dindira area simultaneously for considerable periods during the dry season. These animals then moved eastward as separate herds during the periods of surface water and ferage'abundance. Although never recOrded as far as the eaStern boundary, '“thflir range extended at least 100 km.to the Kandea area; They also frequently'moved short distances into Kenya; but this was during the wet months, rather than the dry as is the case with other species. There was also an eastewest movement of elephants in the reServe which occurred during the wet season and was not associated with the permanent water of Dindira Dam. In January-February 1965, January 1966, again in March 1966 and in April 1967, elephants concentrated around the northernmost hill mass in the central section of the reserve. These concentrations numbered from 1200-1600; and, judging from their temporary absence elsewhere, they consisted of animals from both the 106 o.nm mm tonne *aan.mn o.nem N eeneeeneeen *e.e «.mem . n neeneem *aam.on e.mmm m eena heave n .e.m.z .n.e eonsem e u z e u z e u z e.mn u we m.e u we n.e u we eeneem m.m n n N.m n n n.a u m up: w n z e u. z e u z N.n u we m.a u we n.omm u an eeneem e.e u x e.e u n o.om u n nee Aomm Henoeeomv Aomm enoneenonweaomv enenz ease: .Aomm enoneanomv Ema dhdflfifin .41! 4| slql . .|.II .Anee. u a u aaa .mo. u a pa .mnm u av .eeneem eo: one wnfinoe eone mane ne hefimnoe emoaoa one oe noee:_e:oneanom one neon hedonoe coweoo hue emonmen one Bone omneno one mndemoefinea eooeeo noeeoenoend eneeemenmem thwen e we onone .hHeneenogEH .moononommee Henomeoo eneeeeenmem me ado: me .eeone eoHn onBee oonne one no edeaene owneH mo eefienoe one an moononoeeee eneOfieHnmeo wadeeneosdae ooneeneb mo eemhaene he:uo;e .e oHnee 107 eastern and western sections of the reserve. After a week or two, the concentrations dispersed and the distribution throughout the reserve returned to normal. Although several other hypotheses might explain this phenomenon, it seems that the concentrations are associated with seasonal breeding patterns (Quick 1965, Laws 1968). Seasonal biomass patterns: The temporal biomass pattern follows that of the numbers closely. Therefore, during the dry season when overall numbers are low, the east-west gradient is steepest (upper pairs of maps Figs. 13, 14, 15). The mean dry-season biomass for the entire reserve reaches a perigee of about 570 kg/km2 during this period while the densities in the western end are at a maximum of nearly 13,000 kg/kmz. During the rainy season, the eastawest gradient is nearly extinguished (lower pairs of maps Figs. l3, 14, 15) and the mean biomass for the entire reserve rises to 1925 kg/kmz. Two salient factors are involved with this seasonal reversal of biomass distribution. During the rainy seasons, elephants constituted about 82% of the total game reserve biomass while only 46% of the dry season biomass was made up of elephants; Since there is a dispro— portionately high ingress of elephants into the eastern end, the overall pattern reflects a large depression in the center of the reserve with highs on either end (WWE, Figs. l3, 14, 15). But of equal importance is the liberation of the other herbivores from the constraints of the dry season water source in the western end. The fact that this occurs was demonstrated by the highly significant interaction effect of areas x seasons (Table 7, P. 106) but is even more substantiated by the almost complete lack of any wet season eastawest gradient when elephants are excluded from the relative biomass 108 calculations (WeE, Figs. 13, 14, 15); From this it is clear that utilization of the artificial permanent water source in the western and plays a major role in the ecology of the reserve. DISCUSSION Based on the east-west gradients of climate, soils, and vegetation established in previous chapters it is not surprising to find that animal density gradients also exist. “It is of interest, however, to note that the total number of animals and biomass in the reserve are greater during the wet season than during the dry. This seasonal pattern is opposite that reported for most African parks and reserves where animals move into the protected areas during stressful periods and move outside during periods of abundant forage and water. In this sense, most of the Mkomazi reserve serves as a wet season liberation area for animals which take dry season refuge in the Tsavo National Park of Kenya. The Opposite pattern, that of moving into protected areas during the dry-season periods of stress applies to the relatively small area in the northwestern section of the reserve, however, where permanent water has been provided. Tb me, this phenomenon reflects the inability of the semi-arid eastern sections to support the wet season densities throughout the year. 'Whereas the forage carrying capacity may be limiting in local areas, it is my opinion that the unavailability of permanent surface water is more crucial. 109 “The word unavailable is chosen judiciously. Less than 30 km from the wet season environs of the migratory zebra, gazelle and oryx flows the permanent water of the Umba River; Yet it is only rarely that wild 'ungulates are seen nearby. If water is such a crucial dry season commodity, the question of why the bushland along the Umba remains baren and depauperate of game animals must be posed. Permanent rivers within other East African parks and reserves largely control the dry season animal distributions. It appears to me that the answer lies almost wholely in the fact that there is heavy usage of this area by illicit Wakwave and'Wapare cattle grazers and hunters. As a consequence, two different effects on the game are evident. Approximately 40% of the entire eastern half of the reserve is seriously overgrazed by cattle (see Table 3, P. 62; Figure 18b; Hemingway gt a_l_._; 1966). At a recommended stocking rate of 6 ha per beast (McKay 1968, Hemingway 23.51;.1966) the southern border areas were estimated to be 15-20 times overstocked in 1966 (Hemingwayigtngl;_l966). There is little doubt, therefore, that the grazing wild ungulates incur the effects of severe forage competition in these areas. But several ungulates (sags kudu, gerenuk and giraffe) rarely graze and since cattle do only limited browsing it would seem that these ungulates would incur little competitive effect from cattle. Probably more important is the behavioral effect that people (particularly hunters) and domestic stock have on the game populations. Although no quantitative data are available, there is little doubt among Game Department personnel that the southern and eastern borders of the ‘reserve support the greatest illicit hunting pressure. Pienaar gtflgl; (1966) have shown that in the Kruger National Park, at least breeding llO herds of elephant significantly'avoid areas developed for touriSm, and that the only recent attacks on tourists have occurred in those areas where elephants had no recourse but to encounter tourists. The results of this study show a significant difference (P<:.05) in the mean herd size of elephants between the eastern and western sections of the reserve. The eastern elephant population has the greater mean herd size, and although subjective, it appears that the eastern elephants are substantially more truculent than those of the west. Both of these parameters correlate with the greater hostility toward elephant on the part of hunters and grazers in the eastern half of the reserve. Elephants have also shyed away from those areas of the Mkomazi in which substantial culling operations were undertaken in 1968 (Barry Turner, Pers. Comm.) These observations are summed up as follows. Altheugh zebra, oryx and gazelle co-inhabit the northern half of the eastern end of the reserve with cattle during the wet seasons, they very rarely inhabit the areas along the Umba River. Although their exodus from the reserve is concurrent with, and appears to be a result of; the drying of surface water; they do not utilize the Umba Rivers Excluding elephants, the 2-3 km strip of bushland adjacent to the permanent water of the Umba River supports the lowest game density of any area in the reserve. This density approaches zero animals per kmz. Elephants, on the other hand, do utilize the bushland along the Umba and since they are essentially browsers it is doubtful that much forage competition occurs with the cattle of the area. They too, move ‘away from the permanent water of the Umba during the dry season. 111 'With regard to the potential utilization of East African rangelands for game crepping or ranching, considerable literature has been compiled on the relative densities of game and the game carrying capacities of different areas. Comparative values for several areas have been compiled and tabulated by Bourliere and Verschuren (1960), Petrides (1963), Stewart and Zaphiro (1963), Talbot gtflgl; (1965), and Pienaar gtflgl;l(l966) (see Talbot gt 31:.1965 for 28 references prior to 1965). Although the game density or standing crop of the Mkomazi does not approach the phenomenally high value reported for the higher rainfall areas of eastern Katanga (Congo) or Uganda (C: 24,000 kg/kmz Bourliere and Verschuren 1960, Petrides and Swank 1966), the values compare favorably to those reported for other Aggie-Common bushland areas under similar rainfall regimes (Table 8). In general, it appears that within the semi-arid bushland association mean annual standing crop values show about a 10—fold increase from only a few hundred kg/km2 in the drier areas (Stewart and Zaphiro 1963) to several thousand under higher rainfall conditions (Potts and Jackson 1952, Talbot 1963). This same'range oftbiomass standing crops occurs in the Mkomazi under ' apparently similar vegetation conditions and suggests that some measure of overall climatic conditions (e.g. actual evapotranspiration) may be the best correlate for predicting carrying capacity and mean standing crop. Little mention has been made of the cattle numbers of biomass within the reserve. While conducting the monthly aerial game surveys cattle numbers and locations were also recorded. From the compiled data the cattle numbers were known to exceed 3000 during certain rainy seasons and even during the driest months they numbered in the hundreds. eons eHHenonom ooone sene .onon:ooHo mood thenoeemnoo one onon coon.menweo: coeeaeeoo one * mama wm.m oHeeeo nee: neat nan aged 84 333 profit. Haeeneen eeneunee Bo wmlmm .eneflnmon .o>nomom euesonz enonnwasooleeoeo< eenesnee eooz.eewoeohnoenm ewma hennaeq omoa .o n .o awanxfiZIeeoQ enne>em eeoeo4 eeneunee ewma hennaen owed .o e .o o>nomom onemnenee ennebeo eeoeo< mwma onenQeN ehnon Haeeneen so me .0 one enesoem mow N .o amneeem enaez. noun omenm enomom nwma onenaew ehnom Haeenfien so on .o nae enezoem Nan nm.a .e amneeam Homenem ooone new: noon hnm neeaeemez ecefinmen enema eonHee moon ewneazeelehnon enonafiaaoolefioeo4 mama noonoeh eeneunee nosnnnone one meeom oamm no 3 .o .emnehnenm enonnfieeoolewoeod oononomom mono meanneem heeonon noeeeooq omee eeeenem .eoenme eeem eo meone usednosn enonoeeee now A \wnv mmeaoen mono maeeneem deduce neoa nee Amax\.osv mofiefionoe onebennon owneH opeeen .0 .m oHnee 113 The mean annual number of cattle supported within the reserve was estimated to be 1500 :l: 10%. Using a mean population weight estimate of 225 kg (Deans _e_t_ ah 1968), the additional biomass to be included with the wild ungulate standing crop is 337,500 kg or 103 kg/km. In discussing the severe habitat degradation of the Tsavo National Park of Kenya (contiguous with the Mkomazi), Glover (1963), Bourliere (1965) and Laws (1969) conclude that a density of 0.4 elephants per km2 (1 per miz) is approximately the sustained carrying capacity of the semi- arid bushland. Although a considerable area of the Tsavo Park (East) receives less rainfall than the Mkomazi, it is unlikely that the Mkomazi carrying capacity is much greater than 0.5/king. It is therefore hoped that the Game Department will pay special attention to the potential problem of too many elephants and pursue a viable management policy as exemplified by the preliminary culling of 300 elephants in 1968. C O MWM U N I T X, S T R U C T U‘R E Tropical ecosystems have long been known for their highly complex structure, the intricate interrelations between components and their high organismic diversity (Wallace 1878). Although these attributes are commonly ascribed to the wet trOpics, East African ecosystems reflect the same high species diversity; and because of the array of large animals, they have gained considerable notoriety. Much ecological theory has been directed toward the questions of tropical community structure and how these systems support such high diversities (Klopfer and MacArthur 1961, Connell and Orias 1964, Pianka 1966). Without implying causation, most theoretical arguments reduce to the hypothesis that, in general, tropical organisms have narrower niches. In other words, they generally manifest greater specificities for abiotic and biotic conditions than their temperate counterparts. With specific reference to East Africa, studies by Talbot and Talbot (1962), vesey-Fitzgerald (1960, 1965), Lamprey (1963) and Field (1968) suggest considerable ecological or niche separation of the large herbivores on the food resource alone: A similar condition exists for the large predatory array CWright 1960, Kruuk and Turner 1969, Mitchell gtnglg_l965, Hirst 1969). Darling (1960) has described other aspects of the ecological separation of the large ungulates and the work of Lamprey (1963) provides still farther insight. Hofmann (1968) has elucidated internal anatomical differences which correlate with and possibly govern feeding habits. Although this study was not purposely designed to elucidate the ecological separation of the species array, the quantitative analysis of results provides insight into this phenomenon. 114 115 Conceptually, the Mkomazi community may be portrayed by means of a species-environment matrix (Fig; 17). The total area of the reserve has 'been roughly divided into four habitat types, namely, dry montane and riparian forest, tallgrass savanna (bushed and wooded grassland), open grasslands and bushland. A measure of species diversity for each habitat provides insight into the large animal constituancy of each. By considering each species' relative occurrence in each of the major habitats a crude measure of ecological separation is obtained. However, the utilization of the various habitats is time dependent; and therefore if the total period of study is divided into wet, dry, and transitional periods an estimate of temporal separation can be made. An index of niche breadth on the "time dimension” will therefore reflect an inverse measure of a species' specificity of seasonal conditions. Clearly, the number of resources (dimensions) upon which a species' specificity could be evaluated is very large. This line of reasoning soon leads to the Hutchinsonian concept of each species occupying an n-dimensional hyperspace (Hutchinson 1957, 1965). Finally, it seems clear that all species are not discretely separated (Lamprey 1963, Field 1968) along any dimension: That is, closely related species tend to "overlap" one another with respect to food, habitat utilization, temporal patterns, etc. This leads to the concept of niche overlap or the quantitative expression of the similarity of different species. SPECIES DIVERSITY A primary problem associated with the study of community diversity is its measurement. The species diversity of a community or ecosystem 116 .Aexoe oomv moeoomm noeo en ooesnenenoo eeeenen one CH onowee>nomno Heeoe eo nowenomonm one mnenoeemnoo en oHneneeeno haeoeon we heemnobeo ooeoomm onobfinnon owned o.eeefinen noeo eo onsmeos < .neeeonn once: Henomsoe .ooeoomm noeo mo endoeoa e mo>wnoe mofinowoeeo oaee oonne one mo noeo me oonnnooo noen: onoeeePnomno Heeoe mo mnoeenomonn one mneeenaebo to: esn eHanoe oeem one ween: .oobenoo we neoeonn_onoen .ooeoomm noeo mo ondmeos e .Aexoe . ooov em woeemnuu n .m woe eadenoe one new: oghe eeeenen noeo cw onoeenomona one mnweenaebo hm .mnoeenomonm mnehneb 2e neooo meoenoa oeHe oonne one moahe eeeenen nohes nsoe one 2e ooeoomm mo mnoeesnenemeo one .XHneeB enoenonw>nojheenseEoo eeeaonz one mo noeeeueaeseaoonoo .eH onewem lhuuntn .................. lushbnst .............. nutdflk ................. lattes: ............... . Lluuattugknul ..... - latte ................... Species diversity per habitat type niche breadth, of species on the liabltat dimension 118 may be simply expressed as the number of species present or some complex relationship between the number of species and the numbers of individuals per species. A simply tally of the species present may show two greatly dissimilar communities to have the same "diversity” since no condideration is given to possibly differing abundances. More sophisticated indices include Fisher's “bKF index (Fisher, Corbet and Williams 1943); Simpson's ”‘XP index (Simpson 1949); Margalef's "d" indices (Margalef 1957) and others. Currently, use of the Shannon4Wiener information indices are gaining wide acceptance because of their comprehensive but simplistic nature (Margalef 1963, Pielou 1967, Lloyd et.al;'1968). In short, the underlying theory is that of defining the number of ”bits” of information or binary choices necessary to fully identify any element in an array. Thus, from an "information" point of view, the index relates to the uncertainty involved in predicting which species will be encountered by a random sample from the community (Lloyd gt_§l:.l968). From this it is established that any monospecific array of elements will have the lowest possible value since only one choice is necessary to identify any element. It does not hold, however, that an array with the largest number of species of elements will necessarily have the largest index value. For if S species of elements total N individuals, the maximum value would only be achieved if all elements were equitably distributed among the S species, that is, N/S individuals per species. 0n the other hand, the minimum value will be achieved when all N;S+l elements are of the same species and the remaining 8-1 species are represented by only one individual each. 119 Clearly, there are two components to the index; the number of species, and the equitability with which individuals are distributed among the species. The mathematics of the indices are adequately developed and explained by a number of authors including Margalef (1957), Pielou (1967), Lloyd and Gherlardi (1964) and Lloyd gt,él:.(l968). Using Sterling's approximation to large factorials, the equation for individual diversity is: s H'=-E PilogPi i=1 Where Pi = the prOportion of the total elements (individuals in this case) which occur in the ith species of the array. The logarithms may be taken to any convenient base although base 2 is commonly used. This is the case below. This index is sensitive to unequal sampling, however, and therefore the values reported here are derived by calculating Pi = in; /'£ X H}. where Ni 3 the number of the ith species occurring in each game count along a segment. The mean numerical species diversity for all game count segments representative of the various habitat types was: montane 0.74 i .10 bushland 1.12 f .00 shortgrass plains 1.46 f .22 tallgrass savanna 1.68 i .10 Although there is no clear diversity gradient with east-west location in the reserve pg£.gg;, it appears that there is a direct relationship between diversity and utilizable net primary production. 120 When the size or weight difference between the species of a community is very large, some question may be raised regarding the usage of numbers as Opposed to biomass. For example, since an elephant weighs roughly 500 times as much as a dik dik, it seems illogical to treat the two as equals. It has been proposed, therefore, that the proportion of total biomass contributed by the ith species be used as the Pi rather than the proportion of total numbers (Dickman 1968). But theoretical consideration of the "clumped" distribution of biomass units and other aSpects (valiela, in prep.) may negate the validity of using biomass preportions in the standard information equations. Nonetheless, the mean individual biomass diversities for the various habitats were: montane 0.50 t .00 bushland 0.74 t .10 open grassland 1.18 t .22 wooded grassland 1.31 i .00 The relationship between the different habitat values does not "Change appreciably irrespective of whether‘numerical or biomass values are usedt The main difference between the two is that the biomass indices are all reduced substantially from the numerical index values. This reflects a less equitable distribution of biomass between the species than do numbers. However, since the relative values do not change appreciably, it must be that the large species were approximately evenly distributed between the four habitat types. If there had been greatly disparate distribution of the large species between the habitats, the difference between the numeric and biomass values would have been large. 121 Despite the relative similarity of numerical and biomass index 'values above? there seem to be marked differences when considering changes in time. Numerical diversity indices for a number transect areas show a tendency to increase during the wet season. This trend was not universal, however, and it does not hold for the reserve as a whole. 0n the other hand, the biomass diversity indices tend to reflect the opposite response. The monthly values calculated for the entire reserve show a marked decrease during several rainy periods. Uhdoubtedly this was due to the ingress of large numbers of elephant and the consequent shift in biomass prOportions. NICHE BREADTH -- HABITATS Whereas the concept of the ecological niche is old (Grinnell 1917), the quantitative evaluation of niche breadths is relatively new (Levins 1968). The contemporary indices of niche breadth are simple expressions which attempt to quantitatively evaluate the specificity of an organisms resource utilization or tolerance. Indeed, any measure which quantitatively evaluates the "spread" or breadth of conditions over which an organism ranges could be used as a niche breadth measure; Along with most measures of dispersion, the index value increases as the 'breadth of the distribution increases. Therefore, the index represents the inverse of specificity; and if a species has a high niche breadth value for habitats, it would be expected to be a cosmopolitan species. Again, with minor alteration, the information theory derivate used for the diversity indices may be applied to quantify niche breadth (Levine 1968). Let Pi of the formula: h log B' = -§E_Pi log Pi i=1 122 represent the proportion of observations on a species made in the ith habitat as opposed to the proportion of individuals representing the ith species as was the case with the diversity index. Then the result, B', is a measure of the niche breadth of the species on the h different habitats considered. That is, it provides a measure of the uncertainty involved with predicting in which of the h habitats the species will next be encountered. But as mentioned above, unequal sampling Obviously biases the value. Therefore Pi is defined here as Xi/Z Xi where Xi equals the mean number seen per segment per unit time in the ith habitat type. By taking the antilog of log B', the limits of the index are adjusted to yield a maximum of 4.0 when a species was equitably distributed between the four habitat types and a minimum value of 1.0 ‘when it was only seen in one habitat. The index values for the large ungulates range from a maximum of 3.42 for rhinoceros to a value of 1.22 for bushbuck (Table 9). The five most cosmOpolitan (equitably distributed) species were rhinoceros, eland (331), wart hog (3.22), giraffe (3.11) and elephant (3.09) while the five most restricted species were bushbuck (1:22), duiker (1263), buffalo (1.71), wildebeest (1.75) and reedbuck (1.97). NICHE BREADTH -- TIME If the total observations for each species are categorized by Season rather than habitat, the niche breadth index reflects the SPecies' temporal distribution. A low index value suggests that the Preportions of total observations made during each season were disparate and"‘.the species probably reflects large seasonal density fluctuations. 123 Table 9. Tabulated index values of niche breadth on the habitat and time dimensions for the larger animals of the Mkomazi Reserve. The index values were derived from the formula log Pi = -ZPilog Pi (see text for terminology). niche breadth on three seasons (15x :3) niche breadth on four habitat gypes (lzxafl) rhinoceros (3.42) eland (2.988) eland (3.31) gerenuk (2.975) wart hog (3.22) reedbuck (2.974) giraffe (3.11) giraffe (2.973) elephant (3.09) wart hog (2-953) gerenuk (2.92) rhino (2.947) dik dik (2.86) ostrich (2.934) hartebeest (2.79) Gazelle (2-933) gazelle (2.54) kudu (2.908) zebra (2.52) hartebeest (2.898) ostrich (2.46) elephant (2.895) kudu (2.45) cryx (2.867) impala (2.36) impala (2.806) oryx (2.31) dik dik (2.679) steinbok (2.10) steinbok (2.568) waterbuck (2.03) zebra (2.509) klipspringer (1.98) buffalo (1.897) reedbuck (1.97) waterbuck (1.708) wildebeest (1.75 ) wildebeest (1.592) buffalo (1.74) duiker (1.000) duiker (1.63) klipspringer (1.000) bushbuck (1.22; bushbuck 1.000 lion (1.90 lion 1. 3 jackal (1.64) jackal (1.000) hunting dog (1.00) hunting dog (1.000) (1.00) (1.000) hyaena hyaena 124 Only patterns applicable to the reserve as a whole can be described here: Consequently, only the aerial transect data were used in the calculations since only they represent the entire reserve. Because of the few seasonal categories, the limits of the index are 1.0 and 3.0. Small species which were observed only once or a few times on the aerial transects reflect the lower limit. Thirteen other species showed index values greater than 2.8 (Table 9). Because of this clumped array of values near the upper limit, the index difference between eland which‘reflected the greatest temporal stability (2,988) and Grant's gazelle (2.933) which was notably migratory is only 0.055. HABITAT PREFERENCE With compensation for unequal sampling and the differential visibility in the various habitat types, the relative proportion of times a species was observed in the various habitats may be used as a crude index to habitat preference. Correction for unequal sampling may be made by simply reverting to the mean number seen per unit distance Wlflie the differential visibilities can be partially corrected by <3onverting to the number observed per unit area surveyed. Still, it is IhmprObable that the success of sighting animals in different habitats can be completely equalized and therefore a relatively constant bias 'toward the more open vegetation types seems inevitable. ‘Within the 22 species array of large herbivores, there appears to be ‘3 gradient of preferences from montane and riparian.forest to cpen grass- 1and conditions (Fig. 18). Since the observations are here categorized iJTto four major habitat types, 25% of all observations would be expected IRDI each habitat type if no affinities or preferences were operative. Tilerefore if the proportion of observations made on a species exceeds 125 25% some degree of preference is implied. Further, if the majority of observations (proportion >035) occurred in a single habitat type it seems that the species involved was strongly attracted to that habitat type. Five of the 22 species (bushbuck, duiker, rhinoceros, waterbuck and buffalo) were observed in montane or riparian forests greater than 25% of the time. Three of these species (bushbuck, duiker and rhino) occupied these environs the majority of the time. Twolve species were observed under bushland conditions more than 25% of the time but only two species' (dik dik and kudu) were found under ' these conditions the majority of the time. Three other species (gerenuk, eland and klipspringer) reflected subStantial affinities (240%) for bushland but were not recorded there a majority of the time. Nine species; eland, elephant, gerenuk, hartebeest, kudu, oryx, reedbuck, wart hog and zebra manifested some degree of affinity for the Various types of grasslands, but only klipspringer was observed under these conditions a majority of the time. Klipspringers, although reflecting high proportions of occurrence in the grassland and bushland categories, are very restricted to rock outcrOps and small hilltops under relatively open conditions. Since these local conditions are largely contained in surrounding areas of grass or bushland, the reported proportions are artifacts of the necessarily gross categories. The concept of the "edge effect" or the utilization of the ecotonal area between vegetation types is of great importance in the Mkomazi as elsewhere in East Africa (Lamprey 1963). Impala, ostrich and the introduced wildebeest were all observed under ecotonal conditions greater than 50% of the time with greater than 40% of the buffalo, 126 Figure 18. Observations of the 22 large herbivores occurring in ' the Mkomazi Reserve show varying degrees of segregation into the different habitat types. There is, however, a general gradient from those which occurred most freq- uently under montane or riparian conditions (lower left) to those which were observed most frequently in grasslands of varying types (upper right). We of total observations ln each habitat type . Montana Bushland Ecotonal Grasslands u: Bushbuck Dulker Rhino Waterbuch Butlelo le Dllt Kudu Geronuk Eland Giraffe Stelnbuck Gazelle Wildebeest Impala Ostrich Hartebeest Elephant Wart Hog Zebra O ryx Reedbuck Klipspringer 128 gazelle and hartebeest observations being recorded under these conditions. Eight other species utilized the transition zones greater than 25% of the time. The many advantages of such behavior are well known, but close proximity to the different food and cover types seems to be a dominant factor. SPECIES ASSOCIATION AND NICHE OVERLAP Because of the decidedly non-normal frequency distributions and the high incidence of tied observations, standard correlative techniques are generally not valid analytical measures for the data at hand. Therefore, a measure of niche overlap has been applied to the species occurrence frequency distributions. If two species were found in the same proportion on all game count segments over the entire course of the study, the measure of correlation between the two should approach the maximal limit. ‘Similarly, a measure of niche overlap on the habitat dimension would reflect a maximum value under such conditions. The equation fer niche overlap used here is that of Horn (1966). His paper should be consulted for the derivation. If the total number of Observations made on a species is subdivided into categories representing game count segments upon which the observations were made, then the number occurring on the 1th segment is denoted by xi and the sum of the xi (representing the total observation) is denoted by X. That is 2X1 = X. Similarly, when the total observations ofla second species are appropriately categorized, the number in each category is denoted yi and 129 that total observations of species 2 is Y. When the observations are categorized in such a manner the ”overlap" of the two species (i.e. the degree of similarity in the categorized observations) is given by the equation: R _ x +Y lo X'+Y°)- X°lo x - YilogYi ° - (X +'Y) i 22 +.%) § 1 2 § 1 'Y 03 - 0g - 03 Since we are only interested in the ratios of these measures rather than 'the values themselves, the logarithms may be taken to any convenient base. The limits of overlap deriving from the above equation vary from 0 when the two species being considered are completely distinct with respect to distribution, to 1.0 when the two species are identical with respect to proportional distribution. A similarity matrix of R0 values representing the degree of overlap between all species combinations is then generated and since 26 species (22 herbivores and 4 carnivores) are donsidered here, the matrix is square and of order 26 x 26. Since the degree of overlap between species 1 and species 2 is identical to that between species 2 and species 1 the matrix is symmetrical and only a diagonal half need be considered. Similarly, when the various proportions of a species are compared to itself they are seen to be identical, and therefore all elements along the matrix diagonal reflect the limit of 1.0 (Fig. 19). Biologically, two salient features derive from such a similarity matrix. One, those species pairs reflecting high degrees of overlap tend to exist in the same places in about the same preportions. Such a relationship might imply some positive association or facilitation between the two. Secondly, those species pairs manifesting low index Values showed little spacial overlap which might imply antagonism or competitive exclusion. 130 Figure 19. Elements of the similarity matrix shown here represent the degree of niche overlap as measured by the proportional occurrence on the different segments of the game count grid. For example, species 1 and species 2 (hartebeest and impala) reflect a spacial niche overlap of 0.86 where the limits of the index are 0.0 and 1.0. A value of 0.0 would be obtained if the species were 100% spatially isolated and were never observed on the same game count segment. A value of 1.0, on the other hand, would be obtained if the two species occurred on all segments in exactly the same proportions. The order of the species has been arranged to segregate the ones of greatest niche overlap in the upper left corner of the matrix while those reflecting little overlap are Segregated toward the lower right. Species Number hartebeest l impala 2 ostrich 3 steinbok h giraffe 5 gazelle 6 wart hog 7 dik dik 8 eland 9 elephant lO gerenuk ll zebra 12 oryx 13 kudu 14 rhinoceros 15 reedbuck l6 wildebeest l7 waterbuck 18 buffalo l9 duiker 20 klipspringer 21 bushbuck _2§;_ jackal 23 lion 24 hyaena 25 hunting dog 26 1 2 3 4 5 6 7 0 0 1011121314'151611101920212223242026 '13 9% 1 1 2 .06 1 3 .04 .02 1 4 .00 00 .12 1 5. .05 .01 .03 .12 1 6 63 1.1 03 .15 .15 1 . 1 .16 1160.60.16.621 - 0 ..12'64 61 39 .14 66 .61 1 9 6069.62.51.116096641 10' 6550.58.50.13J35056101 11 .00 60 ‘62 .59.63' 63 35 .66 .55 .51 1 . 1 .12 .60 62 .53 .69 .19 .52 .54 .69 .14 .50 1 - .61 .51 '.61 .51 .61 .60 .61 69 .51 60 50 .10 1 14 .69 63.53 .29 .69 .50 .42 62 .53 61 .49 .43 .49 1 15 .14 66.52 .41 .61 .53 .45 .35 .51 .69 .36 .50.34 40 1 16 .11 .15 .54 .40 .66 .53 Ab .33 .39 .52 .34 .30 .26 .40 .11 1 11 .10 .64 .44 .21 .53 .51 .20 04 41.63 .05 .49 .12 .04 .46 43 1 10 64 .51-A0 .29 .40 .39 .32 .34 .42 66 .33 90 34.34 .45 34.35 1 19 .54 90 .35 .29 .42 .35 .26 .23 30.52 22 .44.24 .21 31 .51 .14 44 1 - 20 .53 45 30.32 46 .35 .31 .20 .36 61.21 .41 .34 .26 .46 .30 .20 .61 .30 1 21 .60 .59.01 .14 .31 .25 .22 .23 .20.39 .24 .20 M .55 .14 .13 04.30 .19 .23 1 22 40 .23 .10 .12 .26 .12 .10 .10 .21 .51 .05 .24-10 .15 .41 .25 .11 .54.36 .61 .13 1 23 .04 .82 .10 .60 .13 .16 .59 .43 .65 .60 .53 .14 .62 .29 .39 .29 .32 .51 .51.40 .01 .01 1 24 .10.14 .53 .50 .56 60 .44 .35 .53 .60 .41 60.49 .31 .36 .41 .40 .40 .53 .30 .39 .20 .66 1 25 61.13 .61 .5160 .50.59 .54 21.35.51 .11 .41 .41 .44 .55 .00 .19 .13 .26 .12 .09 .13 .22 1 26 .51 .16.“ .10 .14 .49 .05 .03 .14 .51.05 .64 40 .06 .27 .01 .16 .00 .16 .02 .00 .01 .07 _.12 .00 1 132 Not discernable from the matrix given are the potential temporal interactions of species. That is, it may be inferred from a high index value that two species occupied the sample areas in approximately equal proportions. But the reader is cautioned against the inference that they inhabited these areas simultaneously. .Analysis of the 33 monthly matrices is required for such interpretation. The highest overlap value found by this analysis was 0.861 obtained from the occurrence records of impala and hartebeest. Other herbivore associates with hartebeest and impala which reflected a high degree of associatiOn were ostrich, steinbok, giraffe and gazelle (see upper left hand corner of Figure 19). Jackal manifested the highest degree of association between a carnivore and the herbivores mentioned above while lion reflected only slightly lower index values. It is a common observation that many ungulates do'not scatter or flee from the area when most predators are present. On the other hand hunting dogs frequently drive the potential prey animals from location upon arrival. The index values between hunting dogs and most herbivores are much lower than the three other carnivores and imply a distinct negative association. Whether or not an antagonistic behavior was actually involved can not be definitely established however. Low index values reflecting spacially separated distributions between the large herbivores were conspicuous for a number of species. The values between wildebeest and klipspringer (.Ofl), dik dik (.04), gerenuk (.05) and kudu (.0#) as well as between bushbuck and gerenuk (.05) and klipspringer and ostrich (.07) all fell below .10 reflecting a strong lack of overlap in their distribution. Among the dominant species, buffalo seemed to reflect a slight negative association 133 (values-<350) with most other species while elephants manifeSted about a 60% overlap with most other'speciest Zebra, oryx and Grant‘s gazelle showed 68 to 80% overlap in their distributions. DISCUSSION In the proceeding chapters emphasis has been placed on the description of abiotic and biotic components of the ecosystem. The purpose of the present chapter is to illustrate patterns of organization within the large animal community. 0f the 22 large herbivores considered, three species were recorded a majority of the time in the montane and riparian forests, two species reflected a majority of observations under bushland conditions and three species were recorded in ecotonal conditions greater than 50% of the time. When the observations can be more discretely classified into vegetation subtypes such as tallgrass, drainageway, bushed and wooded grass, and shortgrass; the description of habitat preferences and ecclogical separation will be greatly enhanced. Of the fourteen species which were not recorded more than 50% of the time in any one habitat examples of both habitat ”specialists” and ”generalists” exist. Klipspringer, for example are Very discriminating and only'occupy rock outcrops. Reedbuck only seem to occupy the rank Panicum maximum and Chloris roxberghiana grass swards in and around gullies, drainageways and groundwater seeps. waterbuck are similarly 1 restricted, but frequently travel considerable distances across open, shorter grasslands in pursuit of water and therefore were frequently recorded under these conditions. 134 The generalists such as elephant and giraffe may wander freely throughout several major habitat types while maintaining a selective diet of particular species or forage types. At this level of analysis the complete ecological separation of the species is not possible. Consequently, the habitat overlap of'many species appears great. For example, the high degree of overlap between impala and hartebeest on the habitat dimension could be greatly refined if food species preference within the habitats were considered. Superimposed upon the spatial patterning of species is the possibility of temporal patterning. Considering the relatively isothermal environmental conditions and the physiological adaptations to water stress (Taylor 1968, 1969), it appears that more opportunity for temporal patterning exists in tropical communities. Thus a cursory purusal of the game count records shows that the greatest number of large herbivore species ever seen on a single segment (representative of a local area within one habitat type) was 12 and the median number was four (I = 11.26). The maximum of 12 occurred around the Dindira Dam area during the dry season when animals were concentrated to the greatest degree. The pattern of north-south and eastawest movements associated with the Mkomazi fauna seems to substantiate a high degree of temporal patterning. Four species, elephant, gazelle, oryx and zebra all reflect a dry season exodus from the reserve. Of those animals remaining in the reserve during this period elephant, zebra, gazelle, eland, giraffe and buffalo all show marked increases in numbers around Dindira Dam. But, surprisingly, the numbers of hartebeest and impala around Dindira Dam decrease during this period of intense utilization. They move to slightly more distant areas still accessible to the water. 135 Similarly, there appears to be a temporal interaction between the numbers of impala and gazelle in certain local areas of the northwestern section (e.g. that covered by the Gate-to-Ibaya transect). During periods when impala numbers were high in this area, the numbers of gazelle tended to be low; the obverse also held. From the habitat point of view, the diversity indices cf the four major communities reflect an important overall trend. From the limited data available, there is a strong positive relationship between the mean large herbivore diversity (both biomass and numerical) and the accessible net primary productivity of the major plant communities. "Within the lowland vegetation communities there seems little doubt that the hushed and wooded grasslands of the northwest are generally more productive than the short grass plains and that the bushland is least productive of all (see Figure 8 p. 58). It may be recalled that the herbivore diversities of the different habitats were significantly different with hushed and wooded grassland being greater than shortgrass which was greater than bushland. 'It is further'postulated that even though the dry montane forests are surely'more preductive than the bushed and wooded grasslands; a high percentage of the production is either limited to the canopy and inaccessible to the large herbivores or it is in the form of unutilizable cellulose and lignin. Therefore, the large herbivore diversity is understandably low. Connell and Orias (l96h) and Pianka (1966) have hypothesized that a positive correlation between productivity and diversity might hold ‘generally while Nargalef (1963, 1968) presented arguments in favor of an inverse relationship. Confounded within these arguments, are the ideas 136 of specific trophicllevel production and diVersity as opposed to the ecosystem as a whole. It is of considerable importance to mention at this time that although the precision with which biological phenomena can be described is usually enhanced by the use of mathematical formulations, the accuracy of the descriptions is not necessarily increased. This is of particular importance here since an attempt has been made to use quantitative analytical techniques on a body of basically descriptive data. The fact that klipspringers were shown to exist largely in buShland and grassland and have a relatively wide niche breadth on the habitat dimension is not an inadequacy of the analytic techniques, but rather an inadequacy of the sampling design. For optimal results it is essential that the modes of analysis be borne in mind when establishing the sampling regimes. 'T”H E ‘E C‘O SWY‘S T E M?' A S ’ A W H19 L E The onus of the ecologist is the integration of seemingly disjunct facts, figures and empirical knowledge into a coherent whole. This whole is frequently more meaningful than the sum of the parts. Further- more, there are a number of factors, influences and interactions which have meaning when a system is viewed as a whole, but which are not apparent to the autecologist or even the pepulation or community analyst. For example, the analysis of the Mkomazi animal community by"itself provided a valuable description of the spacial gradients in denSity and biomass. Without the insights gained by considering the abiotic, vegetative and human aspects, however, the factors causing those effects would remain little more than speculation. Similarly, the formulation of ecological questions without a full appreciation of the frame of reference could lead to spurious results. Hopefully, sufficient evidence is presented in the present chapter to substantiate the claim that large scale ecological surveys must be viewed in an ecosystem context. Although the concept of species dominance is presently in disfavor among many ecologists (MCNaughton and WOlf 1970), it appears that, in general, a relatively few ”dominant” species account for the bulk of the numbers and biomass in most communities. In other words, a few species predominantly define the structure. Seemingly of more interest, how- ever, is the concept of "functional arrays" of species or the idea that most ecosystem functions are also mediated through a few species. If 'this is the case, and we wish to further analogize with physical syStems, we might label this subset of species which performs the bulk of the system's function (e.g. energy, nutrient, or mass transfer; productivity; 137 138 decomposition etc.) as the "energy processors”. Thus, with regard to any particular'function (e.ga energy transformation), a few species perform the bulk of the function and all other species might be categorized as a ”control” or "signal processing" component. Although this array of ”control" organisms plays a minor rele in the system's function it performs a major control or regulatory activity. These "control" organisms probably constitute the majority of the inherent redundancy characteristic of biological communities. They provide the system's'longeterm stability similar to the intricate guidance controls of a space vehicle, or in crude form, the furnace thermostat. Specifically, it is postulated that a few species of‘the Mkomazi ecosystem dominate the numbers and biomass and, in general, portray the system. It is further hypothesized that a few species process the bulk of the energy, mediate the bulk of the nutrient'flow, and exert a major effect upon the environment. These postulates seem substantiated by the following statistics. During the period of study four large mammal species accounted for between 60 and 90% of the mean annual numbers on the'various‘study areas and on the reserve as a whole (Table 9): Four species also accounted for between #4 and 96% of the total biomass. Simce maintenance and sustained work metabolism appear to be linear functions of the basal metabolic costs of homeothermic animals (Brody 1905, Kleiber 1961, Hemingsen 1960 in Lamprey 1963), an index to community metabolism is derivable from the mean weights of the species involved. Even though the standard metabolic function 70 kg‘75 (Maynard and Loosli 1956, National Academy of Sciences 1966) may underestimate the true fasting catabolism of indigenous East African ungulates 139 as as a mm mm aw mm om mm mm mm am am mm mm mo coapspanp Inoo Hopop HH . ** * NH sownpmo 6H 6H m m 0H mm as ea caeaaa m mm wH m mm NH m 2H wm em ma pmoonopaon m m m as as oeceaam es «H mm a efiaoeem no on mm an as He -mm mm m m ecceaoae as me .uu we NH ms oHceecn S S S 6 6 e r r r a a a m a a .H m a a .H m a a .n a .... a m a .... m .m a e a m m "m _ mm nu m “m mu m um “m _u amfiaonopoa Hodges mmoeoan Houses Hones: Hansen mofioomm zoos F zoos M sooe& .ocmao Hm soapcpfiapcoo “a .**,mmanou an soapsnaapsoo wm .* .Apxoe oomv Emfiflonmvoa access some one one ammoEOHn dosage some one .nopesn Hodges some one op Eamon. EL”: moaoonm passage pmofi. Room 05 mo :Oflpopflpcoo psoonom one 00H canoe 140 (Rogerson 1968), this is of little consequence here since it is only being used as an index to total metabolism. .Multiplying the estimated basal metabolism of each species by the appropriate density and summing over all species present yields an estimate of the total large herbivore catabolism. From this it is concluded that four species account for 77 to 91% of the total large herbivore energy transformation throughout the reserve (Table 9). This is further corroborated by the species-specific productivity estimates which are not finalized at this time. As previously established, however, spatial and temporal patterning of the total species array suggests that different species dominate the functional processes in different regions, and at different times in the same region. Therefore, annual statistics of specific areas or even seasonal statistics derived from large areas are not the most accurate reflection of the month to month or seasonal properties of specific communities. Although the numbers of feur species (buffalo, elephant, hartebeest and zebra) must be summed to account for'90% of‘the dry season totals in the Dindira Study area, two species (buffalo and elephant) constitute approximately 90% of the total biomass and metabolism (Fig. 20). Buffalo completely abandon this area during the wet season and impala replace them as an integral species. Along with elephant and hartebeest they constitute approximately 90% of the total numbers and biomass while processing 93% of the energy flow. From this it seems that the bulk of the Mkomazi large herbivore Species do not contribute significantly to the major structural or functional attributes of the system and their role must be that of signal processors or controllers for the system. Chew and Chew (1970) Figure 20 e lhl Mean seasonal species contribution to the numbers, biomass and large herbivore metabolism of the Dindira Study Area in the western end of the Mkomazi Reserve. Although the dry season biomass density of animals was approximately three times as great as the wet, the dry season biomass density was over six times that of the dry. Importantly, only three or four of the 22 large herbivore species constitute over 90% of the numbers and biomass of the large herbivore community and process an equally high percentage of the energy transfer. DINDIRA STUDY AREA DRY SEASON WET SEASON Zlotol Number 7: total Biomass \ \ \ \\\R§\\\‘ 7. 1010' Metabolism [:1 01111.11. _ Honcho-s1 Elephant» Impala - 12 other 800- 143 draw a nearly identical conclusion from their study of mammals in a desert shrub community: They state, ”Mammals are not important in the energy turnover -- their importance must lie in the specific controlling actions on the plants and other components.” This concept does not imply that they are of less importance to the biological system, but only that their role is less obvious and probably more involved with the system's homeostasis than in the processing of nutrients 0r energy. In other words, thesesspecies may exert considerable influence on the abiotic component (e.g. soil structure, microrelief‘and microclimate) as well as other biotic components (e.g. the control of bush encroachment or vegetation species composition). It is likely that they also have a major influence in balancing the composition of the biological System. 'For example, aside from the beneficial effects of competition, they undoubtedly play a major role in supporting the predator component which. serves as the ultimate control on the herbivores (Errington 1963, Buckner 1966, Hirst 1969). Ranging from nearly imperceptible homeostatic effects'under‘ ”normal" conditions (VeseyeFitzgerald 1960, 1963, Paine 1966; Paine and Veda 1969, Harper 1969) to conspicuous habitat degradation in certain instances (Petrides and Swank 1958, Buechner and Dawkins 1961,-Glover 1963), large herbivores effect considerable pressure on their 'environment. The homeostatic effects are rarely discernable unless the herbivores are reduced in numbers or removed and therefore it is difficult to measure or interpret. Evidence for the later effect (i.e. habitat degredation) was available, however, and is presented below. A Although factors are uSually confounded in nature to the extent that single-factor-effects are rarely measurable, a form of experi- 144 mentation can be accomplished by isolating the factors in time or space. By considering all other factors relatively constant, the independent effects of elephants, cattle and fire can be described. Elephant densities in the Mkomazi area are relatively high (up to l/ka/yr in local areas) and there is considerable reason to believe they are a major factor in changing their environment (Buechner and Dawkins 1961, Glover 1963, Bourliere 1965, Pienaar gt_§l:.l966, Lamprey Eluéla.1967)r Under the drier bushland conditions in the Mkomazi, the majority of the elephant foodstuff is browse and therefore this animal would be expected to have its greatest effect on the woody component of the biota (Fig. 21a). The effect of trampling on grasses and herbs is restricted to watering points and is negligible. Aside from water holes there is no area in the Mkomazi where the natural populatiOns of game alone have had a perceivably degrading effect on the grass-forb component. The ”elephant-effect" is therefore believed to be mainly restricted to the arborescent vegetation. The southern three-fourths of the entire eastern half of'the‘ reserve (within watering distance of the Mbaramu-Umba River; 15 km.) has been subjected to intense cattle grazing for over a decade. In the southern border areas, there is essentially no game (<:100 kg/kmz) and therefore any vegetation degradation is certainly due to cattle. Since browse provides only a small percentage of cattle forage; the grass-forb component of these areas all manifest the effects of severe overgrazing (Table 3 area 4, p. 62; Fig. 21b; Hemingway'gtugl; 1966), while there has been no apparent degratory effect on the woody vegetation (Fig. 21b). It is therefore concluded that the "cattle-effect" is largely limited to the grass-forb component of the vegetation. Figure 21a. 1115 The ”elephant-effect" seems to be that of reducing the woody, arborescent component of the vegetation. Although normally less diligent, elephants had felled and completely devoured this baobab (Adansonia digitata Bombacaceae) in less than 2 weeks time. In no area free from cattle grazing was there any substantial degradation of the grass-form component of the vegetation by elephants or other indigenous herbivores. Figure 21b. The ”cattle-effect" is that of severely degrading the grass- forb component of the vegetation while having a benign to positive effect on bush standing crop and production. About 750 km2 along the south—central and eastern boundaries of the reserve are overgrazed to the extent shown. 147 In the east-central section of the reserve (around Kamakota) the annual density of both cattle (1 cow/3-4 ha) and elephants (1.0/km2) is high. The combined effect is expected to be displayed by both components of the vegetation, therefore, and there is obviously such an effect (Fig. 21b). The single factor "fire-effect" can be demonstrated by controlled burning in areas where cattle and game have been excluded or by natural fires in areas where animals do not occur. In August, 1966'a range fire burned extensively in the Mkomazi and the fire line representing the point where it was extinguished vividly portrays the effect of one hot fire on bush suppression (Fig. 22a). The combined effects of cattle, elephants and fire on the vegetative component of the system are well known to field ecologists in East Africa, but only partial documentation exists. Mr. D. G. Anstey, formerly of the Tanzania Game Division, has kindly provided a series of photographs of the extant vegetation conditions around Dindira Dam at the time of its construction in 1957 (left hand column, Figb‘23)t Matched photographs were taken in 1967 after 10 years of the combined 'effects of fire and elephant usage (right hand column, Fig. 23). The over-riding difference is that the area is now more open with considerably less bush and tree cover than was extant in 1957. Mr. Anstey, having known the area intimately for nearly 20 years, assures me that, to a greater or lesser extent, the depicted change applies to the whole reserve. 'The management implication Should be obvious; if the bushed and wooded graSSland structure is to be maintained, fire, elephant, and possibly other herbivore management will be required. 148 The "fire-effect" is one of obvious bush suppression. Figure 22a. Hillsides as steep as that depicted are rarely grazed or browsed by indigenous herbivores and the depicted effect is that of one fire in August, 1966. The sharp fireeline manifests the effect one hot burn may have on bush control. Figure 22b. The combined effect of cattle overgrazing and elephant browsing pressure in the central section of the reserve. There is essentially no "fire-effect" involved here as the area will not sustain a bush or grass fire. Figure 23. 150 Ten year time-lapse photographs illustrating the combined effects of fire and elephants along with dry season concentrations of other herbivores around Dindira Dam. The left-hand column of photographs depict the vegetation as it existed in 1957 during construction of the artificial perm- anent water supply (photos courtesy of D. G. Anstey). Matched photographs taken in 1967 appear in the right-hand column and illustrate the vegetation condition 10 years later. Although the hillside vegetation appears more sub- stantial in the 1967 photograph of the bottom pair, this is an artifact of a higher power lens and better resolution. The removal of the bush thicket of the foreground is a significant change toward a more open grassland. Vegetation changes surrounding an artificial water supply irom 1957 to 1967 ..v ...» ‘__, $ ' ‘ q o I o b ‘ s ' ’ ' ‘ q“.“o"..-'-‘ " ' . I.‘ ‘u|"‘ ‘9. \ua. ~Ri“ ~°' . .>q 0 a o d - ‘ 1 q.\\- u‘!“~~'x. 152 As previously described, the complex of large herbivores is partitioned temporally as well as by habitat type: There was also shown to be a highly significant interaction between season and area when measured as tOtal biomass density (Table 7, p 106). A salient factor in this interaction involves the seasonal differences in the utilization of specific areas. Thus, herds of impala and hartebeest retreat from the Dindira Dam area as buffalo, elephant, eland and zebra concentrate there during the dry season. Impala and hartebeest move back into the area during the rainy months. Similarly, as many of the impala move onto the MZuRune area (10.15 km south of Dindira) during the dry season, Grant's gazelle numbers show a large reduction in this area. In certain areas the pattern of seasonal change in Species utilization is so striking as to suggest that an overall ”grazing strategy” might be operative. 'Working with the interaction between different stocking rates and grazing systems in New Zealand, McMeekan (1960) and McMeekan and walshe (1963) found that higher dairy cattle productivities per unit area were achieved at high stocking rates than at the low rates, but significant 'interactions were found to existn At high stocking rates, best results wereobtained from a controlled rotational grazing scheme, while at low stocking rates continuous grazing produced highest productivities. The same significant interaction has now been described for East African domestic stock (Walker 1968,'Walker and Scott 1968 a, b). 'Within the Mkomazi, the Dindira area represents the most heavily stocked region and the grazing pattern is cyclical to the point of appearing "rotational". Other areas removed from the permanent water support much lower densities of herbivores; but with different species moving in and out, the numbers remain relatively constant throughout the r' .‘b 153 year. The ”coincidence” of these patterns seems almost teological and is deserving of further appraisal. The alternative hypothesis of competitive exclusion in certain areas but not in others is only testable by evaluating the interspecific competition coefficients derived from a system of Lotka-Volterra equations. Such coefficients are available from my data and are held for future analysis. It has been postulated that the high species diversity of tropical areas might derive from the inherently greater productivities of this region (Connell and Orias 1964, Pianka 1966). Hypothetically, since net primary production is greater, the areas could support a larger number and a greater array of animals in the higher trophic levels. A similar argument in favor of game cropping asserts that greater forage specificity on the part of each herbivore species allows a greater portion of the net primary production to be channeled upward through the herbivore food chain than is the case in low diversity, high latitude regions. It is unknown what percentage of the total annual energy budget flowed through the North American bison and allied herbivore component. But within contemporary biotic regimes, it appears that less than 10% of the total net primary production flows through the herbivore component in most temperate regions ('Odum e_t 9.1.: 1962, Golley 1960, Chew and Chew 1970, Slobodkin gthgl;_l967). Petrides and Swank (1966) estimated that 9.5% of the available net production in a small area of Queen Elizabeth Park, Uganda was consumed by elephant alone. In a study of indigenous ungulate energy utilization, Rogerson (1968) states that, "the results suggest that the eland and wildebeest ‘would require from 20 to 30% more metabolizable energy than would 154 cattle, and since the efficiency with which these animals used the digestible energy for metabolic purposes is similar, then a corresponding greater food intake would be required by the eland and wildebeest.” Therefore, it seems that the standard food requirements established for domestic cattle might reasonably be used to crudely estimate the food intake per ruminant stock unit in the Mkomazi Reserve. The standard maintenance diet for range cattle is estimated to be about 20 lb. of air dry forage per day for a 1000 lb. range cow (Stoddart and Smith 1955). This makes no allowance for growth and reproduction, however, and therefore 25 lbs. (11.3 kg) per stock unit/per day is used here. From this it is estimated that the mean annual Standing crop of herbivores on the Dindira Study area (5,5h8 kg/kmz) would consume 50,497 kg of air dry forage per year. The estimated net annual \L production of this area is about 3 x 105 kg/km2 and therefore the percentage of net production consumed by large herbivores alone is 16.8%. But it is well known that elephants are very "rough" feeders and they consume much greater quantities of forage per unit body weight than do the ruminants (Benedict 1936); This, along with the added grazing pressure of the other small herbivores is likely to raise the percentage utilization to at least 17.5%. This is a substantially greater amount than reported for temperate areas. The idea that a greater percentage of the total energy is transferred through the herbivore-carnivore pathway in East African systems is not surprising to field ecologists of the area. For, surely, one of the most striking features of these systems is the seemingly depauperate decomposer fauna. Studies have shown much lower arthropod numbers in these areas than are commonly found in temperate regions (Salt 1952, Madge 1965) and it seems that chemical oxidation and termites 0 155 are the major decomposers of the East African semi-arid regions. Hopkins (1966) reports that ”wood disappearance was caused by termites rather than microorganisms”, and consequently, ”wood decay on the savanna site took about half as long as on the forest site." Although the alternating wet and dry conditions are believed to be important factors in limiting the decomposer populations, the full explanation remains obscure (Visser 1969). Not only are termites of indisputable value in effecting decomposition, but they¢a1so play other roles in the ecology of these areas (Kemp 1955, Murray 1938, Pendleton‘l9fil, Hesse 1955, watson 1962, Glover gtflgl;.l96h). There is almost always an accumulation of base elements in the area immediately surrounding their mounds. These nutrients have a great effect on local plant associations and provide centers of radiation from which succession takes place (Thomas l9h1, Myers 1936). The termites provide a major source of food for the aardvark (Orycteropus 2:23), and other mammals, as well as many birds, while the mounds are of importance in the territorial behavior of ungulates and a valuable refuge for mongooses and other small mammals. Their direct and indirect importance to the Mkomazi ecosystem can not be overemphasized. In conclusion, the interrelations of the various system components can not be overly stressed. Although there is a slight gradient of solar energy input to the system from west to east, it is doubtful that this is of direct importance to the overall functioning of the system. The solar input of about 1.65 x 106 Kcal/mz/yr. provides the driving force behind the system's structure and function while obvious gradients ‘in the other abiotic components provide the major constraints within 156 which the biological components must functiOn. water, sesentially the carrier'of'nutrients and energy up the trophic ladder, seems to be the major constraint both directly and indirectly; Just as there is a difference between food consumption and fOod utilization, so is there a difference between the quantity of precipitation and the quantity of utilizable water. The distribution and intensity of rainfall as well as temperature, soil runoff, permeability, and the other soil parameters which affect the water retention capacity play major roles in determining the system*s characteristics. Being dependent upon the climatic patterns and pedogenesis, the exchangeable ion profiles provide the second major constraint upon the biological components. It was pointed out that Mn is localized in the upper few cm of the soil profile and is likely to be deficient in several Na-saturated vertisolic soils. Mn is prObably the most important micro-element affecting animal productivity. The symptoms of only mild deficiencies are lowered milk production, retarded growth and ataxia as well as lowered reproduction per se. Manifestations of these abiotic constraints on the biological components of the system appear at the first trophic levels Along with the gradient in physiognomic form from west to east, net primary production and general range condition followed the same pattern. Now, along with the greater water stress from west to east, the herbivore component is further limited by lower primary productivities. As a result, herbivore density, diversity, energy transferral and time of occupancy of the central and eastern areas was considerably reduced. The interrelations of abiotic and biotic components are not 'unidirectional however. The herbivores have considerable influence on 157 the structure of the vegetative and soil components. This is a clear example of "feedback” or ”control" and adds validity to the analogy between biological and physical systems. Man‘s interference with the energetics of biological systems (e.g. the increase of productivity) will most probably necessitate an involvement with control. Thus, just as the utilization of monospecific agronomic practices usually requires the concurrent use of pesticide controls; so will the construction of artificial water supplies and other production oriented game management ‘activities require concurrent controls: The 1968 elephant culling operation in the Mkomazi was just such a ”control:" HOpefully3 such forsight will continue to be a part of the overall systems management strategy. S U M M.A R'Y The Mkomazi Game Reserve of semi-arid northeastern Tanzania was established in 1951 as a guidnpgg_gug negotiation for the former Pare Reserve which was ”dereserved” in 1950. Since its establishment, the boundaries have twice been retracted and human pressure continues to be high. An elevational gradient from 230 m above sea level in the southeast to about 800 m in the northwest underlies much of the biological _ variation of the area. Superimposed upon, and partly a consequence of, the elevational gradient; there is a decline in annual precipitation from about 55-60 cm in the northwest to only 35-h0 cm in the east- central section. The annual rainfall pattern is sharply bimodal and although precipitation is by far the most important climatic factor, sufficient importance is attached to temperature, wind and solar radiation to warrant utilization of some more descriptive climatic index. USing Thornthwaite's measure as a comparative index, the climate of the east central section of the reserve was estimated to be at least 50% more arid than that of the northwestern section. Along with the elevational and climatic gradients, soil profile depth, organic matter content, permeability and water retention capacity all generally decrease from west to east. No general soil fertility gradient was elucidated. Most of the bottomland soils are saturated with sodium salts and along with seasonal waterlogging these highly expansive montmorillonite clays appear to be of less overall value than the more freely drained, but lower mineral status soils higher on the slopes. While many of the upland soils contain only marginal levels of calcium (by domestic livestock standards), several of the lowland 158 159 vertisols contain only marginal phosphorus reserves. The high sodium levels may induce microelement deficiencies in the lowland soils. The vegetation was categorized into four major types with dry montane covering the mountain t0ps and bushed and wooded grasslands occupying the freely drained fan slopes of the northwestern section of the reserve. :Agggingommiphora bushland replaces the bushed and wooded grassland in areas receiving less than about 50 cm annual precipitation. Covering approximately 70% of the total area, this community typifies the reserve. Open corridors of seasonally inundated grasslands occupy the bottomland vertisols and constitute nearly 20% of the reserve area. ‘Annual net primary production follows the abiotic gradients and varies from about 400 gm/m2 in the higher rainfall areas of the north- ‘west to approximately 170 gm/m2 in the east-central section. Judging from plant density, ground cover and other indices, rangeland condition is also substantially better in the northwest. In concordance with the abiotic and.vegetation production gradients, animal density also varies from west to east. The mean annual large herbivore biomass density of the northwest is approximately 5,550 kg/km2 while that of the central and eastern sections drops to about 1000 kg/kmz. There was a shift in species composition from west to east however, and although elephants constituted less than 50% of the total large herbivore biomass in the northwest, about 90% of the eastern section biomass was contributed by elephants. Consequently, although the difference in biomass density from west to east was only five-fold, there was nearly a lZ-fold difference in numbers of herbivores per unit area. The dispropOrtionately large seasonal fluctuations in elephant numbers also caused seasonal shifts in total biomass composition. 160 Whereas only 46% of the dry season biomass was contributed by elephants, they constituted 82% of the wet season biomass. There appeared to be a direct relationship between utilizable primary production and herbivore diversity. The bushed and wooded grasslands of the northwest supported the greatest diversity while the open grassland, bushland and dry montane supported successively lower diversities. Based on the relative frequency of occurrence in four general habitat types the total species array was found to be segregated by habitat preferences. Spatial and temporal patterning was further elucidated by means of a niche breadth equation. 0f the total herbivore community, rhinoceros were found to be the most equitably distributed among the various habitat types and thus this species reflected the greatest index value of niche breadth on the habitat dimension. Eland, wart hog, giraffe and elephant were the next most broadly distributed species. Eland observations were found to be the most equitably distributed in time while gerenuk, reedbuck, giraffe and wart hog were next in order: Certain of the species reflected seasonal migration patterns (e.gt elephant, buffalo, zebra, oryx and gazelle) and along with generally lower niche breadths on the time dimension, these species are largely responsible for a highly significant interaction effeCt of season and space on biomass density. Areas which support the greatest wet season densities generally support the lowest dry season densities and vice versa. A measure of niche overlap was used as a quantitative expression of species association. From the similarity matrix of association (overlap) coefficients for all two-species combinations it was found 161 that hartebeest and impala reflected the greatest overlap on the habitat dimenSion while klipspringer and bushbuck reflected the leaSt overlap. 0f the predators, jackals manifested the greatest distributional overlap with herbivores while hunting dogs showed the least overlap. Since only'four species of large herbivores accounted for 80-90% of the total numbers and biomass and 85-90% of the energy exchange, it was concluded that the system's major structural and functional attributes were dominated by a very few species. Further, since collectively, the 16-18 ”nondOminant” large herbivores accounted for only 10-15% of the conSumer level numbers, biomass, productivity and energy exchange; it is hypothesized that they funCtion mainly as "signal processing” or "cOntrol” mechanisms for the system. In closing, this Study'has explicitly shown the interrelations of abiotic and biotic gradients in ecological systems, and suggests how they may affect large herbivore behavioral patterns such as migration and grazing. The concept of dominance of ecological systems by a few species was verified. And finally, support is given to the whole- syStems approach to ecclogiCal problems since considerable insight was gained by describing and evaluating the various components concurrently. LITERATURE CITED Ahlgren, I. F. and C. E. Ahlgren 1960. Ecological effects of forest fires. Bot. Rev., 26:483-533. Aldous, A. E. l93fl. Effect of burning on Kansas bluestem pastures. Kans. Agr. Exp. Sta. Tech. Bull., 38, 65 p. Aldous,.A. E. 1935. Management of Kansas permanent pastures. Kans. Agr.‘Exp. Sta. Bull., 272, Lill- p. Allee, W} 6., A. E. Emerson, 0. Park and K. P. Schmidt 19h9. Principles of animal ecology. W} B. Saunders, Philadelphia, 837 p- Anderson, B. 1963. Soils of Tanganyika. Tanganyika Min. Agr. Bull. 16, 36 p. Anderson, G. D. 1968. A reconnaissance survey of the land use potential of the Mkomazi Game Reserve and an appraisal of factors affecting present and potential land use and productivity in its environs. Tanzania Min. Agr. mimeo report, 41 p. Anderson,.G. D. and Z; Naveh 1968. Promising pasture plants for'northern Tanzania: Vt Overall comparisons of promising plants. E. Afr. Agr. For. J., 3h(1);84-105. Anderson, G. D. and L. M; Talbot 1965. Soil factors affecting the distribution of the grasSland types and their utilization by wild animals on the Serengeti Plains, Tanganyika. J. Ecol., 53:33-56. Anstey, D. G. 1956. Mkomazi Game Reserve. Oryx, 3(4):183-185. Aubrerille, A. 1965. Principes D'une systémati ue des formations végétales tropicales. Adansonia, 5 2):l53-l96. 162 163 Banage,'w. B. and S. A. Visser 1967. Soil moisture and temperature levels and fluCtuations in one year in a Uganda soil catena. E. Afr. Agr. For; J., 32(1): 450-455. Benedict, F. G. 1936. The physiology of the elephant. Carnegie Inst., Publ. 474, 'Washington, 302 p. Best, G. A., F. Edmond-Blanc and R. C. Witting (eds£g 1962. Rowland ward's records of big game. 11 ed. (Africa). Rowland ward, London. Blancou, L. 1962. Mensurations et poids de quelques ongules detchad et de la Republique centrqufricaine. Mammalia, 26(1):84-106. Boaler, S. B. 1966. The Ecology of Pterocggpus angolensis D. C. in Tanzania. Min. of Overseas Dev., Overseas Res. Pub. No. 12, London, 128 p. Bourliere, F. 1965. Ecological notes on the Tsavo. Afr. Wild Life, 19:u7-1o7. Bourliere, F. and I. verschuren 1960. Exploration de Parc National.Albert. Fax 1. BruSsels. Bradley, J. V. 1968. Distribution-free statistical tests. Prentice-Hall, Inc., Englewood Cliffs, 388 p. Braun , H a Mo 1969. Grassland productivity, p. 15.l7. In Serengeti Research (Institute annual report. Tanzania National Parks (Arusha). Bream, R. M. 1963. The chemical composition and nutritive value of grasses from semi—arid areas of Karamoja as related to ecology and types of soils. E. Afr. Agr. For. J., 28(3):134-l42. Brockin on, N. R. 19 1. Studies of the growth of a Hyparrhenia dominated grassland in Northern Rhodesia. J. Brit. GrasSland Soc., 16(1):54-64. Brody, S. 1945. Bioenergetics and growth with special reference to the efficiency complex in domestic animals. Reinhold, New York, 1023 p. Buckner, C. H. 1966. The role of vertebrate predators in the biological control of forest insects. Ann. Rev. Entomol., 11:449-470. 164 Buechner, H. K. and H. C. Dawkins 1961. Vegetation change induced by elephants and fire in.Murchison Falls National Park, Uganda. Ecology, 42(4):?52-766. Burtt, B. D. 1929. A record of fruits and seeds dispersed by mammals and birds from the Singida District of Tanganyika Territory. J. Ecol., l7(2):351-355. Burtt, B. D. 1942. gomeugast African vegetation communities. J. Ecol., 30: 7-1 . Calhoun, J. B. 1959. Revised sampling procedure for the North American Census of Small Mammals (NACSM), Population Dynamics of Vertebrates, Release No. 10, Mimeo, Administrative Publ., U.S. Dept. Health, Ed. & welfare. Calton , We E e 1954. An experimental pedological map of Tanganyika. 2nd. Inter- Afr. Soils Conf., 237-240. Carter, D. B. 1954. Climates of.Africa and India according to Thornthwaite's 1948 classification. .Publ. in Climatol., VII(4): 453-479. Chenery, E. M. 1954. Minor elements in Uganda soils. 2nd Inter-Afr. Soils Conf., 1157-1160. Chew, R. M. and Alice E. Chew 1970. Energy relationshi s of the mammals of a desert shrub ‘ (Larrea tridentatag community. Ecol. Mbnogr., 40(1):l-21. Commonwealth Bureau of Pastures and Field Crops , 1951. Management and conservation of vegetation in African Bull. 41, 97 p. Connell, J. H. and E. Orias 1964. The ecological regulation of speCies diversity. Am. Natur., 98(903)=399-4l4- Curtis, J. T. and M. L. Partch 1950. Some factors affecting flower production in Andropogon gerardi. Ecology, 31:488-489. Dagg, M. 1965. A rational approach to the selection of crops for areas of marginal rainfall in E. Africa. E. Afr. Agr. For., 30(3): 296-300. Dames , Te’ We Go 1959. The soils of Pangani Valley. F.A.O. Rep. No. 970 to Tanganyika Gov. F.A.O. Rome. 165 Darling, F. F. 1960. Wild life in an African territory. Oxford Univ. Press, London, 160 p. Daubenmire , R. .1968. Ecology of fire in grasslands, p. 209-266, in J. B. Cragg (ed.) Advances in Ecological Research 5. Academic Press, New'York. '\ Deans, R. J., C. M. Teague, L. G. Signell, B. M. John and M. E. IKnickerbocker 1969. Livestock and range improvement in Masailand: Tanzania. U.S. Dep. Agr. and U.S. Agency for Intern. Dev., washington, 156 p. Decken, C. C. von der 1869. Reisen in Ost-Afrika in den Jahren 1859 his 1965, bearbeitet von Otto Kersten. Leipziz and Heidelburg, 4 v. D'Hoore, J. L. 1964. Soil map of Africa with explanatory monograph. ‘C:C.T.A. (Comm. Tech. Coop. Afr.). Lagos, 205 p. \ De Vos, A. " 1969. Ecological conditions affecting the production of wild herbivorous mammals on grasslands, p. 137-183, in J. B. Cragg (ed.) Advances in ecological research 6. Academic Press, New York. DiCkman , Me 1968. Some indices of diversity. Ecology, 49:1191-1193. Donahue, R. L. 1970. Soils of equatorial Africa and their relevance to rational agricultural development. Inst. Intern. Agr., l'fi:ch. St. Univ., E: Lansing, 52 p. Ehrenreich, J. H. 1959. Effect of burning and clipping on growth of native prairie in Iowa. J. Range Mgmt., 12:133-137. Elwell, H. M., H. A. Danie1;and F. A. Fenton 1941. The effects of burning pasture and native woodland vegetation. Okla. Agr. EXp. Sta. Bull. B-247, 14 p. Ellerman, J. R. 1940. The families and genera of living rodents. Brit. Mus. Nat. Hist. London. 3 vol. Errington, P. L. 1963. The phenomenon of predation. Am. 501., 51(2):180-l92. 166 Evans, A. C. 1963. Soil fertility studies in Tanganyika II continued applications of fertilizer on the red and red-brown loams of the Nachingwea series. E. Afr. Agr. For. J., 28(1):228-230. Evans, A. C. 1963b. Soil fertility studies in Tanganyika III a- on the Kikungu “‘ and Luseni soil types of the lake and western regions. E. Afr. Agr. For. J., 28(1):231-239. Evans, A. C. and H. W. Mitchell 1962. Soil fertility studies in Tanganyika I - improvement to crop and grass production on a leached sandy soil in Bukoba. E. Afr. Agr. For. J., 27(1):l89-196. Field, Ce Re 1968. A comparative study of the food habits of some wild ungulates in the Queen Elizabeth National Park, Uganda, preliminary report. Symp. 2001. Soc. Lond., 21:135-151. FiSher, R. A., A. S. Corbet and C. B. Williams 1943. The relationship between the numbers of species and the number of individuals in a random sample of an animal population. J. Anim. Ecol., 12:42-58. Forbes, S. A. 1887. The lake as a microcosm. Bull. Peoria Sci. Assoc., p. 77-87, reprinted in Illinois Nat. Hist. Soc. Survey Bull., 15 1925, Po 537-550- Fuggles-Couchman, N. R. 1948. The habitat-distribution of birds of northern, eastern and central Tanganyika with field keys. IV -- birds of the thorn-bush. Tanganyika Notes Rec. 38:31-68. Gillman, C. 1949. A ve etation-types map of Tanganyika territory. Geogr. Rev., 39(1 37-37 + map. Gittins, R. 1968. Trend-surface analysis of ecological data. J: Ecol., 56(3): 845-869. Glover, J. 1963. The elephant problem at Tsavo. E. Afr. Wildl. J., 1:30-39. Glover, P. E., E. C. Trump and L. E. D. wateridge . 1964. Termitaria and vegetation patterns on the Loita plains of Kenya. J. Ecol., 52:367-377. Galley, F. B. 1960. Energy dynamics of a food chain of an old field community. Ecol. MbnOgr., 30:187-206. 167 Golley, F. B. 1965. Structure and function of an old-field broomsedge community. Ecol. Mbnogrt, 35(1):1l3-l37. Griffiths, J. F. 1962. The climate of East Africa, p. 77-87. In Es'W. Russell (ed.) The natural resources of East Africa. D. A. Hawkins Ltd. , NaiI'Obie Grinnell, J. 1917. The niche relationships of the California thrasher. Auk, 34: 427-433 . Gwynne, M. D. and R. H. V. Bell 1968. Selection of vegetation components by grazing ungulates in the Serengeti National Park. Nature, 220(5165):390-393. Hadley, E. B. and B. J. Kieckhefer 1963. Productivity of two prairie grasses in relation to fire frequency. Ecology, 44:389-396. Hahn, He Ce, Jr. 1949. A method of censusing deer and its application in the Edwards Plateau of Texas. Texas Game, Fish and Oyster Comm., Austin, 24 p. Hahnel, L. von 1890. Ostaquatorial -- Afrika zwischen Pangani und dem neuentclecten Rudolph -- See; Ergebnisse der Graf S. Telekischen expedition 1887-1888. Petermann's Mitteilungen, Gotha. Harker, K. W. 1961. A comparison of standing hay production from grasses at Entebbe. E. Afr. Agr. For. J., 27(1):49-51. Harper, J. Le 1969. The role of predation in vegetational diversity. Symp. Brookhaven Natl. Lab., 22:48-62. Harris, In Do 1968. Population dynamics of the African elephant in the Mkomazi Game Reserve, Tanzania, East Africa. M.S. Thesis. Mich. State U., 54 p. Hemingway, P., A. Cormack and W. L. Rebinette; 1966. Appraisal of range condition on the Kalimawe controlled area, p. 7-9, in D. G. Anstey, Kalimawe in relation to Mkomazi Game Reserve. Rept. to Reg. Dev. Comm., Kilimanjaro Reg., Moshi. Ref. No. TA/G/GR/MR. HeSee, P. R. 1955. A chemical and physical study of the soils of termite mounds in East Africa. J. Ecol., 43(2):449-461. 168 Hirst, S. M. 1969. POpulations in a transvaal lowveld nature reserves Zool. Afr., 4(2):199-230. ‘ Hofmann, R. R. “ 1968. Comparisons of the Rumen and omasum structures in East African game ruminants in relation to their feeding habits. Symp. 2001. Soc. Lond., 21:179-194. Holdrid e, L. R. 19 7. Life zone ecology. Trapical Science center, San Jose. 206 p. Hopkins, B. 1966. Vegetation of the Olokemeji Forest Reserve, Nigeria, IV the litter and soil with special reference to their seasonal changes. J. Ecol., 54:687-703. Hopkins, H. H. 1954. Effects of mulch upon certain factors on grassland environments. J. Range Mgmt., 7:255-259. Hopkins, H. H., F. W. Albertson and A. Riegel -:1948. .Some effects of burning upon a prairie in west-central Kansas. Kans. Acad. Sci. Trans., 51:131-141. Horn, He So 1966. Measurement of "overlap" in comparative ecological studies. Am. Natur., 1002419-424. Howe, G. M. 1953. Climates of the Rhodesias and Nyasaland according to the Thornthwaite classification. Geogr. Rev., 43(4):525-539. Hutchinson, G. E. 1957. Concluding remarks: Cold Spring Harbor sympz, 22f415-427. Hutchinson; G. E. ' 1965. The ecological theater and the evolutionary play. Yale Univ. Press, New Haven, 139 p. Huxley, J o 1961. The conservation of wildlife and natural habitats in central and east Africa. UNESCO. Place de Fontenay, Paris, 113 p. Jackson, G. 1954. Preliminary ecolo ical survey of Nyasaland. 2nd Inter-Afr. Soils Conf., 679- 90. Janzen, D. H. ‘ L 1966. Coevolution of mutualism between ants and acacias in Central America. Evolution, 20(3):249-275. 169 1967. Synchronization of sexual reproduction of trees'within the dry season in Central America. Evolution, 21:620-637. Jowett, Ba and P. 0% Eriaku 1966. The relationship between sunshine, rainfall and crOp yields at Serere Research Station. E. Afr. Agr. For. J., 31(4): 439-440. Keay, R. w. J. and A. Aubre’ville 1959. Vegetation map of Africa south of the Tropic of Cancer. Oxford Univ. Press, Oxford, 24 p. Kelly, J. M., P. A. Opstrup, J. S. Olson, S. I. Auerbach and G. M. van Dyne 1969. Models of seasonal primary productivity in eastern Tennessee FestuCa and Andropogon ecosystems. Oak Ridge Natl. Lab. 4310, U.S. Atomic Energy Commission, Biology and Medicine UC-48, 305 P- Kemp, P. B. 1955. The termites of north-eastern Tanganyika: their dietrubution and biology. Bull. Ent. Res., 46:113-135. Kimble, G. H. T. 1960. Tropical Africa, Vol. 1, Land and Livlihood. Doubleday and Co. Inc., Garden City, New York, 2 v. Kleiber, M. 1961. The Fire of Life. John Wiley and Sons, New Yerk, 454 p. ICLopfer, P. H. and R. H. MacArthur 1961. On the causes of trapical species diversity: niche overlap. Am. Nat., 95:223-226. Kruuk; He and Me Turner 1969. Comparative notes on predation by lion, leopard, cheetah and wild dog in the Serengeti area, East Africa. Mammalia, 31(1):1-27. Kucera, C. L., R. C. Dahlman and M. R. Koelling 1967. Total net productivity and turnover on an energy basis for tallgrass prairie. Ecology, 48:536-541. Lamprey, H. F. 1963. Ecological separation of the large mammal species in the Tagangire Game Reserve, Tanganyika. E. Afr. Wildl. J., 1: 3-92. \mprey, He Fe “ 1964. Estimation of the large mammal densities, biomass and energy exchange in the Tarangire Game Reserve and the Masai Steppe in Tanganyika. E0 Afr. Wildle Jo, 2:1-[4'60 \. \ 170 lamprey, H. F., P. E. Glover, M. I. M. Turner and R. H: V. Bell 1967. Invasion of the Serengeti National Park by elephants. E3 Afr; Wildl. J3, 53151-161. Lang-Brown, J. R: and J. Fr Harrop 1962. The ecology and soils of the Kibale grasslands, Uganda. E. Afr. Agr. For. J., 27(1):264-273. Laws, R..M. 1968. A preliminary report (to the Tanzania Government) on three hundred elephant culled in the Mkomazi Game Reserve in March and April 1968. Typescript. 5 p. Laws, R. M. 1969. The Tsavo research project. J. Reprod. Fert., Suppl. 6: 495-531- Lawton, R. M. 1963. Palaeoecological and ecological studies in the northern province of Northern Rhodesia. Kirkia, 3:46-77. Lawton, R. M. 1968. The value of browse in the dry tropics. E. Afr. Agr. For. J., 34(1):227-230. Lay, D. W. 1956. Effects of prescribed burning on forage and mast production in southern pine forests. J. For., 54:582-584. ledger, H. P. 1968. Body composition as a basis for a comparative study of some East African mammals. Symp. 2001. Soc. Lond., 21:289-310. Le Roy, A. 1893. Au Kilima-ndjaro. (Transl. from French.) Sanard et Deragean, Paris. Levins, EL 1968. Evolution in changing environments; some theoretical explorations. Princeton Univ. Press, Princeton, 120 p. Linderman, R. L. 1942. The trophic-dynamic aspect of ecology. Ecology, 23:399-418. Lloyd, M. and R. J. Ghelardi 1964. A table for calculating the "equitability" component of species diversity. J. Anim. Ecol., 33:217-225. lloyd, M., J. H. Zar and J. R. Karr 1968. On the calculation of information-theoretical measures of diversity. Am. Midland Nat., 79(2):257-272. Mackworth-Praed, C. W. and C. H. B. Grant 1957; Birds of Eastern and Northeastern Africa. Longmans, Green and Cox, New York, 2 vol. 171 Mbdge’ De So 1965. Leaf fall and litter disappearance in a tropical forest. Pedobiologia, 5:273-288. Makin, J. 1969a. The use of identification keys in distinguishing individual soil categories. E. Afr. Agr. For. J., 34(4):416-418. Makin, J. 1969b. A feature card system for data retrieval and its application to soil research. E. Afr. Agr. For. J., 34(4):418-421. Marcus, L. F. and J. H. Vandermeer 1966. Re ional trends in geographic variation. System. 2001., 15%1):l-13. Margalef, R. 1957. Lateoria de la informacion en ecologia. Mem. R..Acad. Barcelona, 32(13):373-449. English translation by W. Hall (1957). Information theory in ecology. Gen. System, 3:36-71. Margalef, R. 1963. On certain unifying principles in ecology. Am. Nat., 97: 357-37“. Margalef, R. 1968. Perspectives in ecological theory. Univ. Chicago Press, Chicago, 111 p. Maynard, L. A. and J. K. Loosli 1956. Animal nutrition. MCGraw Hill, New Yerk, 484 p. McCulloch, J. s. c. 1965. Tables for the rapid computation of the Penman estimate of evaporation; E. Afr: Agr. For. J., 30:286-295. MCKay, A. D. 1968. Rangeland productivity in Botswana. E. Afr. Agr. For. J., 34(2):l78-193. McMeekan, C. P. 1960. Grazing management. P. 21-26, Proc. 8th Intnatl. Grassland Congr. McMeekan, C. P. and M. J. walshe 1963. The inter-relationships of grazing method and stocking rate in the efficiency of pasture utilization by dairy cattle. J. Agr. Sci., 61:147-163. MbMurphy, W. E. and K. L. Anderson 1963. Burning bluestem range-forage yields. Trans. Kans. Acad. SCia’ 66:h9‘510 172 McNaughton, S. J. and L. L. Wolf 1970. Dominance and the niche in ecological systems. Science, 167(3915):131-139. Mehlich, A. ,. R. J. Kempton, A. Pinkerton and W. Robertson 1962. Mass analysis methods for soil fertility evaluation. Scott Agr. Lab., Kenya Dept. Agr., Nairobi. Meinertzhagen, R. 1938. Some weights and measurements of large mammals. Proc. 2001. Soc. Lond. A., 433-439. Mills, .W. R. 1954. A review of recent trials with fertilizers in Uganda. 2nd Inter-Afr. Soils Conf., 1133-1141. Milne, G. 1935. Some suggested units of classification and mapping, particularly for East African soils. Soil Res., 4:183-198. Milne, G. 1936. A provisional soil map of East Africa. Crown Agents Colon., London, 34 p. Milne, Go 1947. A soil reconnaissance journey through parts of Tanganyika Territory, Dec. 1935 to Feb. 1936. J. Ecol., 35:192-265. Mitchell, B. L., J. B. Shenton and J. C. Muys 1965. Predation on large mammals in the Kafue National Park, Zambia. Zool. Afr., 1(1):297-318. Moreau, R. E. 1935. A synecological study of Usambara, Tanganyika Territory, with particular reference to birds. J“. Ecology, 23(1):l-43. Morison, On Go To, A. On HoyleJand J. F. HOpe-SMpson 1948.. Tr0pical soil-vege ation catenas and mosaics. J. Ecol., 36:1-84. Murray, T. M. 1938. An investigation of the inter-relationships of the vegetation, soil and termites. S. Afr. J. Sci., 35:288-297. Myers, J. G. 1936. Savannah and forest vegetation of the interior Guiana pldteau. J. Ecol., 24:163-184. Naik, A. H. 1965. Mineral status of grasses and soils of part of Mbulu district of Tanzania in relation to animal health. E. Afr. Agr. For. J., 30(3):175-l82. a! r 173 National Academy of Sciences, National Research Council 1966. Biological energy interrelationships and glossary of energy terms. Publ. 1411, washington, D. Co, 35 p. Naveh , Z c 1968a. The determination of agro-ecological site potentials of \\_ Tanzania Masailand - a challenge for the International Biological Program. Proc. Symp. Recent Adv. Trop. Ecol., p. 601-611. Naveh , Z 0 1968b. The ecological basis of pasture production in northern Tanzania. Proc. Symp. Recent Adv. Trop. Ecol., p. 685-696. Odum, E. P. 1962. Relationships between structure and function in the ecosystem. Jap. J. Ecol., 12(3):108-118. Odum, E. P., C. E. Connell and L. B. Davenport 1962. POpulation energy flow of three primary consumer components of an old-field ecosystem. Ecology, 43:88-96. Odum, H. T. 1967. Energetics of world food production. P. 55-94 in The World Food Problem. President's Sci. Advis. Comm. III, washington. O'Leary, M., R. H. Lippert and O. T. Spitz 1969. Fortran IV and map program for computation and plotting of trend surfaces for degrees 1 through 6. St. Geol. Survey, U. Kansas, 48 p. 013011, J. So 1963. Energy storage and the balance of producers and decomposers in ecological systems. Ecology, 444322-331. Paine, . ROI To 1966. Food web complexity and species diversity. Ame Nat., 100: 55-75. Paine, R. T. and R. L. Vadas 1969. The effects of grazing by sea urchins on benthic algal populations. Limnol. and Oceanog., 14:710-719. Pendleton, R. L. 1941. Some results of termite activity in Thailand soils. Thai Sci. Bull., 3(2):28-53- Penfound, W. T. 1964. Effects of denudation on the productivity of grassland. Ecology, 45:838-845. Penman, H'e‘ Le 1948. Natural evaporation from cpen water and grass. Roy. Soc. (Endon), Proc., As, 1933120-1460 174 Petrides, G..A. 1956. Big game densities and range carrying capacities in East «\\_~ .Africa. Trans. N. Am. Wildl. Conf., 21:525-537. Petrides, G. A. and W. G. Swank 1958. Management of the big game resource in Uganda, East Africa. Trans. N. Am. Wildl. and Nat. Resources Comf., 23:461-477. Petrides, G. A. and W. G. Swank 1965. Population densities and the range-carrying capacity for large mammals in QUeen Elizabeth National Park, Uganda. Zool. Africana, 1(1):209-225. Petrides, G. A. and W. G. Swank 1966. Estimating the productivity and energy relations of an African elephant population. Proc. 9th Intern. Grassl. Conf., 831-842. Phillips, J. Ft V. 1929. Some important vegetation communities in the central province of Tanganyika. S. Afr. J. Sci., 26:332-372. Pianka, Ee Re 1966. Latitudinal gradients in species diversity: a review of concepts. Am. Nat., 100(910):33-46. Pielou, E. C. 1967. The use of information theory in the study of the diversity of biological pOpulations. Proc. 5th Berkeley Symposium on Math. Stat. and Prob., 4:163-177. Pielou, E. C. 1969. An introduction to mathematical ecology. John Wiley and Sons, Inc., New Yerk, 286 p. Pienaar; Us de ve, Pa'van wyk and N. Farrall 1966. An aerial census of elephant and buffalo in the Kruger National Park, and the implications thereof on intended management schemes. Koedoe, 9:40-107. POttS,. We He. and Ce He Ne JaCkson 1952. The Shinyanga game destruction experiment. Bull. Ent. Res., 43:365-374. Pratt, De Je Po, Je Greemay and Me De Gmne 1966. A classification of East African ran eland, with an appendix on terminology. J. Appl. Ecol., 3(2§:369-382. Quick, H. 1965. Ecology of the African elephant. Biol. Sci. curric. study pamphlet 21. D. C. Health & Co., Boston. 175 Robinette, W. L. 1963. 'Weights of some of the larger mammals of Northern Rhodesia. The Puko, 1:207-215. Rogerson, A. ‘\‘ 1968. Energy utilization by the eland and wildebeest. Symp. 2001. “ Soc. Lond., 21:153-161. saChS ’ Re 1967. Liveweights and body measurements of Serengeti game animals. E. Afr. Wildl. J., 5:24-36. Salt, Ge 1952. The arthropod population of the soil in some East African pastures. Bull. Entomol. Res., 43:203-220. SCOtt , Re Me 1962a. The soils of East Africa, p. 67-76. In E. W. Russell (ed.) The natural resources of East.Africa. D. A. Hawkins, Ltd., Nairobi. Scott, R. M. 1962b. Exchangeable bases of mature, well drained soil in relation to rainfall in East Africa. J. Soil Sci., 13:1-9. Semb, G. and J. B. D. Robinson 1969. The natural nitrogen flush in different arable soils and climates in East Africa. E. Afr. Agr. For. J., 34(1): 350-369. Shanta, H. L. 1947. The use of fire as a tool in the management of the brush ranges of California. Cal. State Div. Form, 156 p. Shanta, H. L. and G. F. Marbut 1923. The vegetation and soils of“Africa. Amerw Geogr. Soc. Res. Ser. No~ 13, New Yerk, 263 p. plus maps. Simpson,;E.1H.: 1949. Measurement of diversity. Nature, 163:688. Smith, G. D. 1965. Lectures on soil classification. Pedologie, spec. no. 4. Slade, N. A. and R. D. Anderson 1970. Is wildlife management an ecological science? The Wildl. Soc. News, 129:38. Slobodkin, L. B., F. E. Smith and N. G. Hairston 1967. Regulation in terrestrial ecosystems, and the implied balance of nature. Am. Nat., 101:109-124. A \fi“ UV. . ( I‘) _ e~ (I) rn 176 Sneath, P. H. A. {1967. Trendasurface analysis of transfermation grids.- J. 2001., 151:65-122 . Sokal, R. R. 1965. Statistical methods in systematics. Biol. Rev., 40:337-391. Spurr, A. M. M. 1954. A basis of classification of the soils of areas of composite topography in Central Africa, with special reference to the soils of the southern highlands of Tanganyika. 2nd Inter- Afr. soils conf., 175-190. Stewart, D. R. M. and D. R. P. Zaphiro 1963. Biomass and density of wild herbivores in different East African habitats. Mammalia, 27:483-496. Stoddar‘t, L. A. and A. D. Smith ’ 1955. Range Management. MbGraw-Hill, New Yonk, 433 p. Swynnerton, Ce F. Me ' 1936. The tsetse flies of East Africa. Roy. Entomol. Soc., London. Swynnerton, G. H. and R. W. Hayman 1951. A check list of the land mammals of the Tanganyika territory and the Zanzibar protectorate. J. E. Afr. Nat. Hist. Soc., 20(6-7):274-394. Sys, C. 1967. The concept of ferallitic and fersiallitic soils in central Africa -- their classification and their correlation with the 7th approximation. Pedologie, 17(3):284-325. Sys ,, Ce 1968. Suggestions for the classification of tropical soils with lateritic materials in the American classifications Pedologie, 18:189-198. Talbot, L. M. 1963. The biological productivity of the tropical savanna ecosystem, p. 88-97. In Ecosystems and Biological Productivity. IUCN 9th technical meeting (Nairobi). Morges. TalbOt’ Le Me 1963b. Comparison of the efficiency of wild animals and domestic livestock in utilization of East African rangelands. Publ. N. S., 1:328-335, I.U.C.N. Morges. Talbot, L. M. 1964. The concept of biomass in African wildlife research. Mammalia, 28(4):613-6l9. 177 Talbot, L. M. and Martha H. Talbot 1962. Food preferences of some East African wild ungulates. E. Afr..Agru Fore J., 27:131-138. Talbot, L. M.,'W. J. A. Payne, H. P. Ledger, Lorna D. Verdcourt and Martha He TalbOte 1965. The meat production potential of wild animals in Africa, a review of biological knowledge. Commonwealth Bur. of Anim. Breed. Genetics. Tech. Commun.~16, 42 p. Tanganyika Atlas 1955. Government Printer, Dar Es Salaam. Tanganyika Game and Tsetse Division Annual Reports 1933, 1934, 1950. Government Printer, Dar Es Salaam. Tanganyika Ministry of Lands, Forests and Wildlife, Survey Division 1963. Geological Survey maps. Dodoma. Taylor , C e Re 1 ‘~ 1968. The minimum water requirements of some East African bovids. Symp. 2001. Soc. Lond., 21:195-206. TaYlor, Ce Re 1969. The eland and the oryx. Scientific American, 220(1):88-95. Thomas, Ae Se 1941. The vegetation of the Sese Islands Uganda -- an illustration of edaphic factors in tropical ecology. J. Ecol., 29:330- 353. Thomas, Ae Se 1943. The vegetation of the Karamoja District, Uganda, an? illustration of biological factors in tropical ecology. J. Ecol., 31:149-177. Thompson, B. W. 1965. The climate of.Africa. Oxford Univ. Press, Oxford. Thornthwaite , C e. We 1948. An approach toward a rational classification of climate. Geogr. Rev., 38:51-94. Thornthwaite, C. W. and F. K. Here' 1965. The loss of water to the air. Meteor. Mbnogr., 6(28):l63- 180. Trapnell, C. G. 1959. Ecological results of woodland burning experiments in Northern Rhodesia. J. Ecol., 47:129-168. 178 Trapnell, C. F. and I. Langdale-Brown 1962. The natural vegetation of East Africa, p. 92-102. In E. W. Russell (ed.) The natural resources of East Africa. D. A. Hawkins, Ltd., Nairobi. Trapnell, C; G; and J. F. Griffiths 1960. The winfall-altitude relation and its ecological significance in Kenya. E. Afr. Agr. J., 207-213. U. S. Department of Agriculture 1951. Soil survey manual. anbk. No. 18. U.S. Government Printing Office, washington, 503 p. U. S. Department of Agriculture F‘— l960. Soil classification -- a comprehensive system 7th approximation. Soil Conserv. Serv., washington, 265 p. U. S. Department of Agriculture 1967. Supplement to soil classification system 7th approximation. Soil Conserv. Serv., washington, 207 p. “a. U. S. Forest Service 1968. Range environmental analysis handbook. Rocky Mbuntain Region (2), FSH 2209.21 Denver. vandermeer, J. H. 1966. Statistical analysis of geographic variation of the flathead minnow, Pimephales promelas. Copeia, 3:457-466. Vesey-Fitzgerald, D. F. 1960. Grazing succession among East African game animals. J. Mammal., 41:161-172. Vesey-Fitzgerald, D. F. 1963. Grasslands in ecosystems and biological productivity. Proc. 9th Tech. Meet., I.U.C.N. Nairobi. Merges. Vesey-Fitzgerald, D. F. 1965. The utilization of natural pastures by wild animals in the Rukwa Valley, Tanganyika. E. Afr. Wildl. J., 3:38-48. Visser, S. A. 1969. Environmental factors influencing the occurrence, distribution and activity of’micro-organisms in soils, with special reference to waterlogged environments. E. Afr. Agr. For. J., 34:336-341. Voisin, A. 1959. Grass productivity. Philos. Libr. Inc., New York, 355 p. Walker , B e .1968. Grazing experiments at Ukiriguru, Tanzania. II. Comparisons of rotational and continuous grazing systems on natural pastures of hardpan soils using an "extra-period Latin- square change-overdesign." E. Afr..Agr. For. J., 34(2): 235-244. 179 ‘Walker, B. and G. D. Scott 1968a. Grazing experiments at Ukiriguru, Tanzania. I. Comparisons of rotational and continuous grazing systems on natural pastures of hardpan soils. E. Afr. Agr. For. J., 24(2): 224-234. ‘Walker, B. and G. D. Scott 1968b. Grazing experiments at Ukiriguru, Tanzania. III. A comparison of three stocking rates on the productivity and botanical composition of natural pastures of hardpan soils. E. Afr. Agr. For. J., 34(2):245-255. Walker, T. W. and A. F. R. Adams 1958. Studies on soil organic matter. I. Influence of phosphorus content of parent materials on accumulations of carbon, nitrogen, sulphur, and organic phosphorus in grassland soils. Soil Sci., 85:307-318. wallace,-A. R. 1878. Animal life in the tropical forests, p. 270-311. In A. R. wallace, Tropical nature and other essays. Reprinted in Natural selection and tropical nature. 1895. MacMillan, New York. Watson, Je Fe 1962. The soil below a termite mound. J. Soil Sci., 13(1):46-51. watson, R. M., I. S. C. Parker and T..A11an 1969. A census of elephant and other large mammals in the Mkomazi region of northern Tanzania. E. Afr. Wildl. J., 7:11-26. WBber, Ne Ae 1959. Isothermal conditions in tropical soil. Ecology, 40(1): 153-154. Weis S”, P'e A. 1958. Condensed transcript of the conference, p. 184. In Gerard, R. W. (ed.) Concepts of biology. Publ. 560 Nat. Acad. Sci., Nat. Res. Coun. washington. weSt’ 0e \\ 1965. Fire in vegetation and its use in pasture management with special reference to tropical and sub tropical Africa. Commonwealth Bur. Pastures and Field Craps. Mimeo, Pub. 1, 53 p. 'Wiegart, R. G. and F. C. Evans 1964. Primary production and the disappearance of dead vegetation on an old field in southwestern Michigan. Ecology, 45:49-63. Willoughby, J. C. 1889. East Africa and its big game. London. Wright, B. s. 1960. Predation on big game in East Africa. J. Wildl. Mgmt., 24(1): 1-15. [If APPENDIX APPENDIX I Composite monthly rainfall statistics in cm for the eight stations in the Mkomazi Reserve and the Same Meteorological Station from March, 1965--June, 1967. f -——— I. 1. Mbnth Station mz¢m s: g 8 g a E g a E Q E! 6‘3 .75 Mar .88 .70 Apr .42 .66 May .83 1.0 Jun 1 1.0 Jul 9 .86 .88 Aug 6 1.0 1.0 1.0 .97 .97 Sep 5 1.0 1.0 1.0 .80 .71 Oct 1.0 .78 1.0 .46 .21 Noc 1.0 0.0 + .05 .13 .41 Dec 1.0 .84 1.0 .76 .81 .77 .75 Jan .14 .47 .72 .77 .36 .42 .46 Feb .46 .20 .03 .18 .68 + .47 + .10 MAr .38 .25 + .48 .50 .77 .93 .6 Apr .55 .85 .93 .91 .86 .55 .15 Rhy' l 1.0 1.0 1.0 .57 .91 .95 .86 Jun 9 I .93 .95 .93 1.0 .95 1.0 1.0 Jul 6 I 1.0 .95 1.0 1.0 1.0 1.0 .91 Aug 6 1.0 1.0 1.0 .93 1.0 1.0 .93 Sep 1.0 .90 1.0 .96 1.0 .44 .28 Oct 1.0 .53 1.0 .63 .61 .75 .81 Nov 1.0 1.0 .66 .73 .88 .73 .94 Dec .84 .87 .79 .94 .96 1.0 1.0 Jan 1.0 1.0 1.0 .50 .34 + .53 .81 Feb 1 .95 .65 .72 .62 .23 .41 .86 Mar 9 + .56 .06 .40 +1.51 + .08 + .62 .45 Apr 6 .15 .27 .49 .35 + .59 + .42 .92 May 7 .89 .91 .87 .89 1.0 .95 .97 Jun vfiwamflmmm ofighe .Ofienospnomhromfi .poxHE .hemoa mean we pfimeewmmfinhoe ownpwa .oflahonpnomhnomw .poxfls .Hmpofloxm hpnmm Ho 828.8028 3.5 ...-€22.56.“ .8892 £58- 28 a pfimhmoamm efinhe .Oflsnonphomhflomfi .pfiow .pcxwa .onfim m pnopmdaaom capo .oaanonpnmmhgomfl .oapwnoaafinoepsos .ozam hao> mm vagpaonswo oaHHoz .owenonvnomhnomfi .oapfiSOHHwnonzos .ozfim Na pawneoamm Ofinha .Ofisnonpnonhgomfi .poxfis .hEmOd.osz HQ pacpwfldaom oapcm caspospaonhnomfi .oapd20HHaeoapzos .onfim knob mm pwmnaoawm efinhy .oaehoapnonhsomfi .poxda .ozfim am pfimeaoflam ofiemy .oaenospaonhnomfi .poxae .hEmoa .csfim mm pflmnwoamm OfiHHoz .Ofisnozpnonhgomfi .poxfis .onfim mm pfimnmoamm afiHHoz .Ofiancgpaonhsoma .poxds .ozfim Hm pumpmsaacm awake .owenongeomhsomfl .oapfinoflafinospnos .oeww :4 peopmsafiom oaBongo .Ofisnonpnomhgomfi .OfipwnoaafinoEpsos .oswm m4 pfimnmmawm efimhy .efisnonpnoghsomfi apoxaa .onfim Nd 35.8.28 35 6.5.5.8268 88.8 .25 H4 .Amwmfl .owma .«.Q.m.av sopwhm noa4weamfimmmao Haem.o>Hmsonohmsoo 8 ea scavwewxonmmw Jpn zeefinoad esp an.pofiMfimmeo we meom weasoxz can we mnfipmfia 4 .H .HH HH NHszmm< APPENDIX III III..1. A vegetation species list compiled from specimens collected in the Mkomazi Reserve and identified by the East African Herbarium, Nairobi. ..Gramineae Andropogon schinzii Hack. Aristida adscensionis L. Aristida lommelii Mez Bothriochloa glabra (Roxb.) A. Camus Bothriochloa radicans (Lehm.) A. Camus Brachiaria defloexa (Schumach.) Hubb. Brachiaria eruciformis Griseb. Brachiaria lachnantha (Hochst.) Stapf Brachiaria leucacrantha (K. Schum.) Stapf Brachiaria serrifblia (Hochst.) Stapf Cenchrus ciliaris L. Chloris mossambicensis K. Schum. Chloris roxburghiana Schult. Chloris virgata Sw. Gymbopogon afronardus Stapf Cymbosetaria sagittifolia (A. Rich.) Schweickt. Cynodon dactylon (L.) Pers. Cynodon plectostachyus (K. Schum.) Pilg. Dactyloctenium aegyptium (L.) Beauv. Dichanthium pappilosum (A. Rich.) Stapf Digitaria macroblephara (Hack.) Stapf Digitaria mombasana C. E. Hub . Digitaria remoti luma (Forsk.) Beauv. Digitaria rivae Chiov.) Stapf Digitaria setivalva Stent Dinebra retroflexa (vahl) Panz. Echinochloa haploclada (Stapf) Stapf Enneapogon cenchroides (Roem. & Schult.) C. E. Hubb. Enneapogon elegans (Nees) Stapf Enneapogon sp. Enteropogon macrostachyus (A. Rich.) Benth. Eragrostis aethiopica Chiov. Eragrostis aspera (Jacq.) Nees Eragrostis caespitosa Chiov. Eragrostis rigidior Pilg. Eragrostis superba Peyr. Eriochloa meyeriana (Nees) Pilg. Eriochloa nubica (Steud.) Thell. Eustachys paspaloides (vahl) Lanza & Mattei Heterocarpha haareri Stapf & C. E. Hubb. Heteropogon contortus (L.) Roam. & Schult. Ischaemum afrum (J. F. Gmel.) Dandy Latipes senegalensis Kunth Leptocarydion vulpiastrum (De Not.) Stapf Leptochloa obtusiflora Hochst. Leptochloa panicea (Retz.) Ohwi Lintonia nutans Stapf Microchloa kunthii Desv. Panicum coloratum L. Panicum deustum Thunb. Panicum infestum Anderss. Panicum maximum Jacq. Panicum sp. Pennisetum mezianum Leeks Rhynchelytrum repens (Willd.) C. E. Hubb. Rhynchelytrum setifolium (Sta f) Chiov. Rhynchelytrum villosum (Parl. Chiov. Rottboellia exaltata L.f. Schmidtia bulbosa Stapf Schoenefeldia transiens (Pilg.) Chiov. Setaria homonyma (Steud.) Chiov. Setaria incrassata (Hochst.) Hack. Setaria sphacelata (Schumach.) Stapf & C. E. Hubb. Sorghum versicolor Anderss Sorghum verticilliflorum (Steud.) Stapf Sporobolus consimilis Fresen. Sporobolus festivus A. Rich. Sporobolus filipes Napper Spordbolus fimbriatus Nees var latifolius Stent Sporobolus pyramidalis Beauv. Sporobolus sp. near pyramidalis Beauv. Tetrapogon bidentatus Pilg. Tetrapogon tenellus (Roxb.) Chiov. Themeda triandra Forsk. Tragus berteronianus Schult. Tripogon abyssinicus Steud. Urochloa mosambicensis (Hack.) Dandy Urochloa sp. Cyperaceae Cyperus a10pecuroides Rottb. Cyperus bulbous vahl Cyperus distans L.f. Cyperus exaltatus Retz Cyperus obtusiflorus Vahl Kyllinga oblonga C.B.C1. Maniscus circumclusus C.B.C1. Mariscus leptophyllus (Hochst.) C.B.Cl. Mariscus pseudovestitus (C.B.C1.) Kukenth. Mariscus taylori C.B.Cl. var. taylori Commelinaceae Aneilema aequinoctiole (P. Beauv.) Kunth Commelina sp. ... ”"1 '—-_l Liliaceae Aloa sp. Anthericum ?moni1iforma Rendle Asparagus asiaticus L. Asparagus racemosa Willd. Gloriosa simplex L. Ornithogalum donaldsonii Rendle Ornithogalum sp. Velloziaceae Vellozia aequatorialis Rendle Vellozia spekei Bak. Mbraceae Ficus pretoriae B. Davy Ficus sp. Polygonaceae Oxygonum sinuatum (Mbisn.) Dammer Amaranthaceae Achyranthes sp. Aerva lanata Juss. Aerva persica (Burm.) Merr. Alternanthera sessilis R. Br. GentemOpsis rubra (10pr.) Schinz Cyathula erinaceae Schinz Digera muricata (L.) Mart. Pupalia lappacea Juss. Sericocomopsis.grisea Suessenguth Sericocomopsis pallida (S. Mbore) Schinz Aizoaceae Gisekia pharnaceoides L. Portulacaceae Calyptrotheca taitensis (Pax & vatke) Brenan Talinum caffrum Eck. & Zey. ”file a u ‘ ‘LWOI Capparidaceae Boscia angustifolia A. Rich. Boscia salicifolia Oliv. Cadaba farinosa Forsk. subsp. adenotricha Gilg & Bened. Cadaba ruspolii Gilg Cadaba sp. Capparis tomentosa Lam. Cleome stenopetala Gilg & Benedict Maerua grantii Oliv. Maerua Kirkii Thylachium africanum Lour. Mimosaceae Acacia ancistroclada Brenan Acacia brevispica Harms Acacia bussei Harms ex Sjostedt Acacia etbaica Schweinf. subsp. platyoarpa Brenan Acacia mellifera (Vahl) Benth. Acacia senegal (L.) Willd- var senegal Acacia seyal Del. var. fistula (Schweinf.) Oliv. Acacia stuhlmannii Taub. Acacia tortilis (Forsk.) Hayne subsp. spirocarpa (Hochst. ex A. Rich.) Brenan Acacia zanzibarica (S. Mbore) Taub. Acacia sp. no flowers or pods Albizia anthelmintica Brongn. Albizia harveyi Albizia petersiana (Bolle) Oliv. Albizia sp. Dichrostachys cinerea Newtonia hildebrandtii (Vatke) Torre var. hildebrandtii Caesalpiniaceae Afzelia.cuanzensis'Welw. Cassia abbreviata Oliv. subsp. beareana (Holmes) Brenan Cassia longiracemosa Vatke Cassia mimosoides L. Cassia singueana Del. Delonix elata (L.) Gamble Tamarindus indica L. Tylosema fassoglensis (Kotschy ex Schweinf.) Torre & Hillcoat Papilionaceae Abrus schimperi Hochst. ex Bak. subsp. africanus (vatke) Verde. Craibia brevicaudata (Vatke) Dunn. subsp. brevicaudata Crotalaria sp. Dalbergia melanoxylon Guill. & Perr. Erythrina sp. Indigofera schimperi J. & S. var. baukeana (vatke) Gillett Indigofera spinosa Forsk. Indigofera zenkeri Bak. f. Indigofera sp. Lonchocarpus eriocalyx Harms Lonchocarpus sp. Neorautenenia pseudopachyrhiza (Harms) M. Redh. Ostryoderris stuhlmanni (Taub.) Bak. f. Platycelyphium voense (Engl.) H. Wild Sesbania sesban (L.) Merrill var. nubica Chiov. Tephrosia ?interrupta Hochst. & Steud ex E. Engl. Tephrosia pumila (Lam.) Pers. Tephrosia villosa (L.) Pers var. incana (Roxb.) Bak. Vigna fragrans Bak. f. Vigna reticulata Hook. f. Zygaphyllaceae Tribulus terrestris L. (specimen without flowers) Balanitaceae Balanites sp. Rutaceae Calodendrum capense (L.f.) Thub. Fagara sp. Vepris uguenensis Engl. Burseraceae Oommiphora caerulea Commiphora campestris Engl. Commiphora schimperi Commiphora sp. aff. C. mollis (Oliv.) Engl. Comm phora sp. Meliaceae Malia volkensii Guerke Trichilia sp. Wm Malpighiaceae Acridocarpus zanzibaricus A. Juss. Euphorbiaceae Acalypha ciliata Forsk. Acalypha fruticosa Forsk. Croton dichogamus Pax Euphorbia systyloides Pax Phyllanthus amarus Schum. & Thonn. Phyllanthus maderaspatensis L. Ricinus communis L. Anacardiaceae Lannea alata Engl. Lannea stuhlmannii (Engl.) Engl. Salvadoraceae Dobera loranthifolia (warb.)'Warb. ex Harms Salvadora persica L. Icacinaceae Pyrenacantha malvifolia Engl. Sapindaceae Haplocoelum foliolosum (Hiern) Bullock Rhamnaceae Ziziphus mucronata Willd. Vitaceae Cissus rotundifolia (Forsk.) vahl Tiliaceae Corchorus trilocularis L. Grewia bicolor A. Juss. Grewia fallax K. Schum. Grewia tembensis Fres. var kakothamnos (K. Schum.) Burret Grewia tenax (Forsk.) Fiori Grewia villosa Willd. Malvaceae Abutilon guineense (Schum.) Bak. f. Hibiscus micranthus L. Hibiscus vitifolius L. Sida cordifolia L. Sterculiaceae Hermannia exappendiculata (Mast.) K. Schum. Hermannia oliveri K. Schum. Melhania ferruginea A. Rich. Sterculia africana (Lour.) Fiori Sterculia appendiculata K. Schum. Passifloraceae Adenia globosa Engl. Thymelaeoeae Gnidia latifolius (Oliv.) Brenan Rhizophoraceae Cassipourea malosana (Bak.) Alston Combretaceae Combretum aculeatum Vent. Combretum molle R. Br. ex G. Don. Terminalia kilimandscaharica Engl. Terminalia prunioides Laws Terminalia spinosa Engl. Vernonia cinerascens Sch. Bip. Plumbaginaceae Plumbago zeylanica L. Loganiaceae Strychnos spp. material on loan Apocynaceae Adenium obesum (Forsk.) Roem. & Schult Convolvulaceae Astripomoea hyoscyamoides (Vatke) Verdcourt Ipomoea pestigridis L. var. longibracteata vatke Impomoea wightii (wa11.) Choisy Bombacaceae Adansonia digitata Boraginaceae Cordia ovalis R. Br. Cordia rothii Roem. & Schult Ehretia amoena Klotzsch Ehretia taitensis Guerke Heliotropium eduardii Martelli verbenaeceae Clerodendrum hildebrandtii Vatke Lantana rhodesiensis Moldenke Premna oligotricha Bak. Premna resinosa (Hochst.) Schauer Premna sp. Vitex strickeri vatke & Hildbr. Labiatae Aeolanthus repens oliv. Basilioum polystachion (L.) Moench. Hemizygia fischeri Guerke Hostundia opposita vahl Leucas glabrata R. Br. Orthosiphon parvifolius vatke Pycnostachys umbrosa (vatke) Perk. Solanaceae Solanum incanum L. Solanum sp. nr. taitense vatke Scrophulariaceae Striga asiatica (L.) O. Ktze Striga latericea Vatke Acanthaceae Barleria diffusa (Oliv. Lindau Barleria ramulosa C. B. Cl. Barleria sp. Blepharis integrifolia (L.f.) E. May. Crossandra mucronata Lindau Dyschoriste hildebrandtii Lindau Justicia flava vahl Justicia glabra Roxb. Pseuderanthemum hildebrandtii Lindau Thunbergia affinis S. Mbore Rubiaceae Gardenia sp. Pentanisia auranogyne S. Mbore Pentas parvifolia Hiern Psychotria kirkii Hiern Rytigynia sp. Compositae Aspilia mossambicensis (Oliv.) Wild Brachylaena hutchinsii Hutch. Haarera alternifolia (O. Hoffm.) Hutch. & E. A. Bruce Helichrysum glumaceum DC. Lactuca capensis Thunb. Microglossa oblongifolia O. Hoffm. Vernonia cinerascens Sch. Bip. Vernonia pauciflora Less. H APPENDIX IV.1. Recorded bird species for the Mkomazi Reserve Struthionidae Ostrich Podicipidae African little grebe Phalacrocoracidae White-necked cormorant Long-tailed cormorant Pelecanidae White pelican Pink-backed pelican ‘ Ardeidae Grey heron Black-headed heron Purple heron Great white egret Buff-backed heron Squacco heron Night heron Little bittern Scapidae Hammerkop Ciconiidae European white stork European black stork 'Woolly-necked stork Open-bill stork Saddle-bill stork Marabou stork Threskiornithidae Wbod ibis Sacred ibis African spoonbill Anatidae African pochard Red-bill duck Knob-billed buck Egyptian goose ‘ Spur-winged goose Sagittariidae Secretary bird Aegypiidae Ruppell's griffon vulture White-backed vulture Lappet-faced vulture White-headed vulture Falconidae European lesser kestrel Pygmy falcon ‘ Tawny eagle wahlberg's eagle African hawk-eagle Martial eagle Crowned hawk-eagle Long-crested hawk-eagle Lizard buzzard Black-chested harrier eagle Grasshopper buzzard Bateleur eagle African fish eagle Augur buzzard Gabar goshawk Pale chanting goshawk Mbntagu's harrier Harrier hawk Phasianidae Crested francolin Scaly francolin Yellow-necked spurfowl Halmeted guinea-fowl Kenya crested guinea-fowl vulturine guinea-fowl Otididae Kori bustard Jackson's bustard Buff-crested bustard Black-bellied bustard Hartlaub's bustard Crested bustard Burhinidae Spotted stone curlew Jacanidae African jacana Charadriidae Three-banded plover Crowned lapwing Senegal plover Blacksmith plover Black-winged stilt Scolopacidae Little stint Green sandpiper wood sandpiper Greenshank Glareolidae Heuglin's courser Bronze-winged courser Pteroclididae Black-faced sandgrouse Columbidae Pink-breasted dove Red-eyed dove Ring-necked dove Laughing dove Namaqua dove Emerald-spotted wood dove Green pigeon Cuculidae Cuckoo Red-chested cuckoo White-brewed coucal Musophagidae Violet-crested touraco White-bellied go-away-bird Psittacidae Orange-bellied parrot Coraciidae European roller Lilac-breasted roller ‘ Rufous-crowned roller ’ Broad-billed roller Alcedinidae Giant kingfisher Brown-hooded kingfisher Stripedckingfisher Meropidae European bee-eater Madagascar bee-eater Carmine bee-eater Little bee-eater Bucerotidae Trumpeter hornbill Black and white-casqued hornbill Grey hornbill Red-billed hornbill Yellow-billed hornbill Von der Decken's hornbill Jackson's hornbill Crowned hornbill Ground hornbill Upupidae African hoopoe Phoeniculidae Green wood hoopoe Scimitar-bill Abyssinian scimitar-bill Strigidae Scops owl Pearl-spotted owlet Spotted eagle owl Caprimulgidae European nightjar Donaldson-Smith's nightjar Freckled nightjar Gaboon nightjar Coliidae Blue-naped mousebird Capitonidae Brown-throated barbet Spotted-flanked barbet Red-and-yellow barbet D'Arnaud's barbet Indicatoridae Greater honey-guide If Picidae Nubian weodpecker Cardinal woodpecker Bearded woodpecker Apodidae Common swift Alaudidae Singing bush lark Red-winged bush lark Rufous-naped lark Flappet lark Fawn-coloured lark Notacillidae African pied wagtail Eastern yellow wagtail Golden pipit Pycnonotidae Dark-capped bulbul White-eared bulbul Muscicapidae Spotted flycatcher South African black fly- catcher Puff-back flycatcher Paradise flycatcher Turdidae Olive thrush t Red-tailed ant thrush European rock thrush EurOpean wheatear Pied wheatear Capped wheatear Cliff chat Stonechat Red-capped robin chat Spotted morning warbler White-winged scrub robin White-starred bush robin Sylviidae Olivaceous warbler European wood warbler Red-capped forest warbler Crombec Grey-backed camaroptera Zitting cisticola Rattling cisticola Winding cisticola Tiny cisticola Hirundinidae European swallow Red-rumped swallow Mbsque swallow Striped swallow Campephagidae White-breasted cuckoo-shrike Dicruridae Drongo Prionopidae Straight-crested helmet-shride Laniidae White-crowned shrike Long-tailed fiscal shrike Red-backed shrike Red-tailed shrike Slate-coloured boubou Black-backed puff-back shrike Black-headed bush shrike Blackcap bush shrike Sulphur-breasted bush shrike Black-fronted bush shrike Grey-headed bush shrike Rosyzpatched shrike Nicator shrike IT Paridae White-breasted tit Oriolidae EurOpean golden oriole African golden oriole Black-headed oriole Corvidae Cape rook White-necked raven Sturnidae wattled starling Golden-breasted starling Red-wing starling Fischer's starling Hildebrandt ' s starling Superb starling Yellow-billed oxpecker Red-billed oxpecker Zosteropidae Yellow'white-eye Nectariniidae variable sunbird Amethyst sunbird Scarlet-chested sunbird Collared sunbird Ploceidae BuffalO‘weaver Red-billed buffalo weaver White-headed buffalo weaver Stripe-breasted sparrow weaver White-brewed sparrow weaver Grey-headed sparrow Parrot-billed sparrow Yellow-spotted petronia Layard's black-headed weaver Chestnut weaver Black-necked weaver Red-headed weaver Red-billed quelea Cardinal quelea Yellow'bishop White-winged.widow-bird Green-winged pytilia African fire-finch Red-rumped waxbill Black-rumped.waxbill Cordon bleu Red-cheeked cordon-bleu Pint-tailed whydah Fischer's whydah Paradise whydah Broad-tailed whydah Fringillidae Brimstone canary APPENDIX IV.2. Recorded Mammal Species for the Mkomazi Reserve Insectivora Macrose lididae E1 phantulus sp. Rhynehocyon cirnei Soricidae Crocidura sp. Crocidura sp. Chiroptera PterOpodidae .Rousettus angelensis Epomopherus sp. Nycteridae Nycteris thebaica Hipposideridae Hipposideros caffer Mblossidae Tadarida aegyptica Primates Lorisidae Galago senegalensis Cercopithecidae Papio cynocephalus Cercopithecus aethiops johnstoni Cercopithecus mitis kibonotensis Hominidae Homo sapiens Pholideta Manidae Mania temminckii spectacled elephant shrew chequered elephant shrew shrew shrew rousette bat epauletted fruit bat large-cared hollow-faced bat lesser leaf-nosed bat mastiff bat bush baby yellow'baboon Kilimanjaro green monkey Kilimanjaro blue monkey modern man ground pangolin Lagomorpha Leporidae Lepus capensis abbotti Rodentia Bathyrgidae Heliophobius spalax Hystricidae Hystrix galeata Sciuridae Paraerus ochraceus Xerus rutibus sativiatus Gliridae Graphiurus murinus iMuridae Lemniscomys barbarus Lemniscomys griselda Mastomys natalensis MuS‘minutoides Acomys wilsoni Acomys cahitinus Gerbillus pusillus Tatera robusta Taterillus osgoodi Carnivora Canidae Canis familiaris Canis adustus notatus Lycaon pectus lupinue Otocyon megalotis Carnivora Mustelidae Mellivara capensis sagulata Viverridae Genetta genetta Civettietis civetta civetta Herpestes ichneumon Herpestes sanguineus Helogale undulata Munges murgo colonus Ichneumia albicauda ibeana Abbott's cape hare Blesmol African bush squirrel African ground squirrel African dormouse Taita striped grass mouse Taita single-striped grass mouse shamba rat Pygmy mouse spiny mouse Taita pygmy gerbil gerbil gerbil domesticated dog East African side-striped jackal East African wild dog East African bat-eared fox East African honey badger Neumann's genet African civit greater grey mongoose lesser mongoose dwarf mongoose East African banded mongoose East African white-tailed mongoose Hyaenidae Proteles cristatus termes Crocuta crocuta Hyaena hyaena dubbah Pelidae (Felis lybica taitae Caracal caracal nuficus Leptailurus serval hindei Panthers pardus fusca Panthera 1eo massaicus Aoinonyx jubatus Tubulidentata Orycteropodidae Orycteropus cper Proboscidea Elephantidae Loxodenta africana knochenhaueri Hyracoidea Procavidae Procavia johnstoni Perissodactyla Equidae Equus asinus Equus burchellii Rhinoceratidae Diceros bicornis Artiodactyla Suidae Petamochoerus porous Phacochoerus aethiopicus Giraffidae Giraffa camelopardalis Bovidae Strepsiceros imberbis Tragelaphus scriptus Taurotragus oryx Bos taurus Syncerus caffer Sylvicapra grimmia Masailand aard-wolf spotted hyaena striped hyaena Taita wild cat caracal Ukamba serval Bengal leopard Masai lion cheetah aard vark East African elephant rock hyrax domesticated ass Burchell's zebra Cape black rhinoceros bush-pig warthog Tanganyika giraffe lesser kudu bushbuck East African eland domesticated cow Cape buffalo bush duiker Kebus ellipsiprymnus Redunca redunca Onyx beisa Alcelaphus busolaphus Oreotragusroreotragus Raphicerus campestris Rhynchotragus kirkii Aepyceros melampus Litocranius walleri Gazella granti Capra hireus Ovis aries Swahili common waterbuck Bohor reedbuck fringe-cared oryx Coke's hartebeest klipspringer Tanganyika steinbok Taita dik dik Tanganyika impala Gereruk Grant's gazelle domesticated goat domesticated sheep on ---- ---- ---- --- --- --- moo weapon: m ---- ---- ---- --- --- --- -.xo.n on on nn an o: om oH euoooeo on no on on on on n- .uoaae oma Nu- nn- one om on on gou- ooN nnm onm _ onm owe oo on .43.. oom oom oo- oHN --- --- --- pmooeoo-az nae now n2 nmm oma oo oo 0163.3... on on on no o: no on so: pan: 9 2 Q Q ...... l... I: .883. nmoa onN- onma onm- one oon nmm monsoooaou oe nn no on on om on scope... nu «ma oma nma on no oa eoanpmo oe- one one na- oo- oo oo Kano on oo om oo- on o: om sene 3H nun: nun: III: III III nun nomcfinnmmfiax oe on no on no om o- .H.aaa owe oofl oma one oo on no omoonoosue one ono oom ooHH oon oon on- memo... em on on nm no om oH ascouom o: on no on. o: on n- sea...» nma- ---- ---- ---- --- --- --- usages-o oon ooo no: onn on- on o: ease. ma ---- ---- ---- -.. -.. In- hexane n.o ---- I--- ---- ..- In: .1: see see on ---- .... ---- un- uun- ...: assesses on: nan oon one In: tun In: o-weeon owe . vases cause“ odes HHH HH H -mofloonm .mmsaog: _emmsaoqd .7 added vases anode mweae anode .xno: wasp ca poms mommedo omen can now co>am hHgo one moeae> .ondpeaopaa one scam capesdupno mopasapmo pump can paomonmon heap .m:0fipeadoaee mm¢50fin new onepeaopfifl cap an poms haaeuonom omonv seep «med one no>am mosaeb can gweonpad .oehomom nueeoxz can no nonconw newnea any new wav mopeafivmo pawns: camfioonm oawm .m .>H