Resource availability and habitat preferences of three African ungulates

Biological Conservation 54 (1990) 357-365

Resource Availability and Habitat Preferences of Three African Ungulates Raphael Ben-Shahar Department of Zoology, South Parks Road, Oxford OX1 3PS, UK (Received 5 November 1989; revised version received 15 February 1990; accepted 27 February 1990)

ABSTRACT The relationships between resource availability and the requirements of ungulate species were approached by assessing the habitat preferences of three ungulate species, namely roan Hippotragus equinus, sable H. niger and tsessebe Darnaliscus lunatus, in relation to the classification of their available habitats. Cluster analysis distinguished between five habitat types. Roan and tsessebe were generally associated with habitats forming a large portion of the total area, whereas sable were associated with confined habitats. Selectiveness and possible resource limitation were, however, indicated for tsessebe as they attained higher correlations with particular habitat features which were relatively scarce in the study area.

INTRODUCTION The quantification of associations between animals and their environment has received considerable attention during the last three decades. Research has been orientated towards the resource selection of animals (Alldredge & Ratti, 1986), while the approach to resource availability has generally been related to the estimation of the carrying capacity (Moen, 1973). The subject is particularly important for the conservation of vulnerable animal species or for controlled utilization of prolific species. In this paper, I relate habitat preferences of three ungulate species to habitat features of their environment, and then evaluate the potential of the area to sustain a particular species. 357 Biol. Conserv. 0006-3207/90/$03"50© 1990ElsevierSciencePublishersLtd, England. Printed in Great Britain

358

Raphael Ben-Shahar

The roan antelope Hippotragus equinus is endangered in South Africa, whereas the sable antelope Hippotragus niger and tsessebe Damaliscus lunatus are rare (Smithers, 1986). All occurred historically in the region but disappeared or were severely reduced in numbers during the last four decades (Ben-Shahar & Skinner, 1987). This study took place in 1985, two years after the re-introduction of 10 roan antelope, 26 sable and 8 tsessebe. Its goal is to provide guidelines for the future conservation of the species.

STUDY AREA Animals were located in a fenced section (approximately 2400ha) of a private nature reserve, situated in the Waterberg region, northwest Transvaal (23 ° 51' S, 28 ° 16' E). The topography in the region varied from undulating rocky hills to elevated plateaus at 1100-2300 m above sea level. The area is characterized by abundant perennial watercourses and rivers. Summer rainfall ranged between 650 and 900 mm. Average daily maximum temperatures were 32°C in January and 22°C in July; the average minimum was 18°C in January and 4°C in July (Scholze, 1965). The region is classified as part of the moist savanna biome (Huntley, 1982), dominated by woody plant species such as Terminalia sericea, Burkea africana and Combretum spp.

METHODS

Habitat and animal sampling The area was divided into a grid system of 19 plots whose shape and size were determined by geomorphological features. Each plot was then subjected to an analysis of edaphic factors, vegetation composition and vegetation structure. Two soil samples from the top 200ram were taken in each plot, the location being determined according to the variations in soil colour throughout the study area. The samples were analysed for particle size composition, acidity levels, and organic matter (% carbon) by the WalkleyBlack method (Jackson, 1985). Magnesium and calcium contents were determined by ion extracting with ammonium acetate solution (Black, 1965). Grass species composition was measured according to the Dry Weight Rank method (t'Mannetje & Haydock, 1963). Four hundred 1-m 2 random quadrats were sampled in each plot. All species present in the quadrat were

Habitat preference of African ungulates

359

recorded, after which the three most common were ranked on the basis of their estimated dry mass contribution. The proportion of quadrats in which each species occurred in first, second and third place was calculated. These values were then weighted by multiplying with constants which produced the relative contribution of each species to the total grass biomass. The sum of these values, for each grass species, provided the percentage contribution of that grass to the total mass of the sward in the sampling area. Woody vegetation composition and abundance were determined by the Point Centred Quarter method (Mueller-Dombois & Ellenberg, 1974); 32 points in each plot were sampled, recording the four woody plants which were the nearest to the sampling point. Plant species abundance was then calculated and converted to the form of percentage contribution to the total biomass of grass or tree species, respectively. Tree density was estimated by the Third Nearest method, which measures the distance from the transect point to the third closest tree (McNeill et al., 1977); 32 points in each plot were measured. Animal sightings were recorded daily from transects during 1985 and summed to give an annual estimate. Sighting indices for each species on the different plots were obtained by traversing the plots using different directions and several paths to provide a random base for the final estimates. Sighting indices were formulated since the direct plotting of animal sightings on a habitat map was likely to be biased due to stratified sampling and changes in vegetation density obstructing vision. The indices included the following variables: size of plot (A), the total number of animals sighted within a plot (B), number of transects per month (N), and visibility factor (P) measured according to the Density Board method (de Vos & Mosby, 1969) using: S i = Bi x P i / N i × A i

where S i = preference index of an ungulate species in plot i (1... 19). Habitat classification A cluster analysis based on a non-overlapping approach (Spurway, 1981) was used to classify the plots. Means and deviation of each habitat parameter were calculated across all the plots (Table 1) by selecting a plot and then scanning the remaining plots which had the closest values to the first chosen. A standard deviation was calculated while the means were transformed to form the weighted sum of the two plots. These values were used in turn to select the next similar plot. Thus, small deviation values are derived for closely related plots--the higher the deviation, the less similar the plot is to those previously grouped.

Raphael Ben-Shahar

360

TABLE 1 Habitat Variables Measured in a Private Nature Reserve in the Waterberg Region, Northwest Transvaal, South Africa Variable Slope (°)~ Rockiness (% cover)a pH ° Clay (% contents) ° Calcium (pg/g) Magnesium (/~g/g) Organic matter (%C) ° Distance to water (m)a Tree density (no./ha)a

Grass species (% dry weight) Brachiaria nigropedata C vnodon dact)'lon Digitaria eriantha Enneapogon scoparius Heteropogon contortus Panicum maximum Schizach)'rium sanguineum Tree species (% composition) Acacia karroo Acacia nigrescens Elephanthorriza burkei Ziziphus mucronata Burkea africana

Mean

SE of mean

6.02 51"05 5-29 lif0 590.42 76'36 0.67 894.73 512"52

1.19 5"66 0"21 10.21 272.34 19'80 0'10 150.00 90'39

2'33 2"14 2'73 6"28 3'28 5.80

0"62 1-53 0"53 3'41 1'21 1"55

5"13

1.12

2-05 4-02 2.60 1.51 7'31

1"97 2'63 0"61 0"61 2"32

a Variables used for correspondence analysis.

Habitat preferences Habitat preferences were examined using correspondence analysis (Greenacre, 1984), using a data matrix containing the frequencies of occurrence of antelopes derived from the sightings in relation to the different habitat categories (Table 2). The presentation of results was simplified by analysing the grass species composition separately (Ben-Shahar, 1987). Further reduction of the matrix was possible by omitting the grass height variable which showed poor correlation with antelope distribution. Similarly, positive correlations between soil magnesium and calcium to soil clay contents (r = 0"56 and r = 0"66 respectively, p < 0.05), enabled the omission of these variables. The interpretation of the analysis is based on the distribution and relationships between the variables, represented as points along the principal axes. The importance of the points in correspondence analysis is

Habitat preference of African ungulates

361

TABLE 2 Habitat Variable Categories Used in Correspondence Analysis of Habitat Preferences of Roan, Sable and Tsessebe in a Private Nature Reserve in the Waterberg Region, Northwest Transvaal, South Africa Variable Slope (o)

Rockiness (% cover)

pH

Clay (% contents) Organic matter (%C) Distance to water (m)

Tree density (no./ha)

Abbreviation

Range

SI $2 $3 RI R2 R3 PH 1 PH2 PH3 SAND LOAM Ol 02 WI W2 W3 D D2 D3

0-2 3-9 10-20 0-30 40-60 70-100 4.1-5.0 5"1-5"7 5-9-7.6 4'0-7.0 8-0-21.0 0.3-0.9 1-0-2.3 0-200 250-1 000 I 100-2 500 100-200 201-800 801-1 300

indicated by their absolute (ABS), or relative (REL) contribution. The former indicates the contribution of the variable to the formation of the axis whereas the latter identifies the variables which lie close to the plane of the axis.

RESULTS The cluster analysis grouped the 19 plots in relation to five habitats (Table 3) which could be distinguished in the field. Habitat I was characterized by gentle slopes with medium to high rock cover (50-70%) and sandy acidic soils with fairly dense tree cover. Habitat II consisted of flat terrain with low rock cover on sandy soils with sparse tree cover. Habitat III corresponded to steep rocky slopes on loamy soils and with high tree density. Habitat IV described the riverine vicinity with a dense tree cover on sandy soils. Habitat V was represented by one fiat and open plot with few rocks on clay soil, and had been previously cultivated. The derived deviation values suggest that

Raphael Ben-Shahar

362

TABLE 3 Cluster Analysis Grouping of Habitat Variables for 19 Plots in a Private Nature Reserve in the Waterberg Region, Northwest Transvaal, South Africa

Habitat

I

II

III IV V

Plot

B E C Q R H M L P I J K O S N F G D A

Area

Slope

Rock

(ha)

(deg.)

(% cover)

121-8 121"0 110"4 115'7 75"4 82' 1 171"5 83"4 76"5 165"6 148"2 160"4 203"8 217"0 237'5 75-6 74'3 67'8 82"3

7 5 3 9 7 7 9 10 20 1 1 4 2 l 2 10

11 0 1

60 60 50 70 60 60 60 70 70 40 30 50 40 40 30 90 90 0 0

pH

Tree density

Deviation

(no./ha) 4"55 4"80 5"40 5-45 4'90 5"95 4-50 4-10 5"95 6"05 4"90 4"75 4.25 4'90 4"45 6"50 6-55 4'95 6"95

368 323 279 1 156 551 548 575 963 658 268 115 179 963 176 102 1 317 1 105 836 104

2"4 2'6 8'6 6"3 7'3 7'9 9"6 28'2 41"8 5'1 5-8 4"3 4"0 12"8 194-8 17"1 52'5 545-3

habitats I and II are closely related (Table 3) whereas features of habitats III to V differ considerably when compared to any other habitat. The correspondence analysis in Fig. 1 indicates that the point for the roan antelope correlates with areas far from water sources (W3) on sandy soils (SAND), with low organic matter contents (O1); they show a weaker affinity for occurrence on moderate slopes ($2). Tsessebe show close correlations with low rock cover on acidic soils that coincide with fiats or moderate slopes, close to water sources (Fig. 1). Sable habitat preferences include areas with increasing slope and more rocks, less than 1 km from water sources. These habitats have soils rich in organic matter and with higher pH levels. My analysis demonstrated the affinity between the distribution of any of the three ungulates and the changes in vegetation density. The absolute correlations of the vegetation variables to the axes were low and did not coincide with the ungulate points. The ordination of habitat variables along the first axis (Fig. 1) supports the segregation of the plots by cluster analysis. The second axis separates two habitat types. Habitats on sandy acidic soils (habitats I, II and IV) are situated on the positive values of the first axis, whereas habitats III and

Habitat preference of African ungulates

0.5

363

R3

0.4 0.3 0.2

(~ND $3

0.1

W3 P2

D1 O1

0

( t ~

~2 $1

D2 02 LOAM

-0.1

m® P1

R~

W!

W2p3 -0.2

1

-0.7

I

-0.5

!

I

-0.3

I

I

I

-0.1

0.1

I

I

0.3

I

I

0.5

I

I

0.7

Fig. 1. Correspondence analysis o f habitat preferences of roan (ROAN), sable (SABL) and tsessebe (TSES) in a private nature reserve in the Waterberg region, northwest Transvaal, South Africa. Abbreviations of habitat variables appear in Table 2.

probably V, which are characterized by higher pH and organic matter contents (Table 3), are situated left to the origin of the axes.

DISCUSSION The relationships between preferences of animals to an 'objective' classification of the habitats in a particular area might give an indication of the balance between habitat requirements and resource availability. Although habitat categories and the animals' requirements are subjectively chosen by the observer, it is nevertheless possible to indicate some important factors that influence the occurrence of animals in a particular area. Topographical features and soil characteristics were used to explain the differential distribution patterns found for the ungulates, thus overlooking the influence of vegetation composition as a potential environmental factor. While correlations between the abundance of plant species and the occurrence of ungulates could be obtained (Ben-Shahar & Skinner, 1988), a detailed analysis tends to be somewhat misleading when data on the feeding habits of the related animal species are insufficient or lacking. Instead, the indication of an array of edaphic factors favouring the occurrence of an ungulate species is also likely to signify a habitat with a food source. Previous studies related a change in vegetation structure, namely the encroachment of shrubs, to a negative effect on roan antelope (Joubert, 1976;

364

Raphael Ben-Shahar

Wilson & Hirst, 1977), whereas sable and tsessebe could be found in a wide range of habitats, from open grassland to woodland on different vegetation types (Child et al., 1972; Grobler, 1973). My results showed that in the study area, change in vegetation structure has little effect on the distribution patterns of roan, sable or tsessebe. Tsessebe were influenced by other factors, as they occupied an area of flats with sandy and acidic soils and low rock cover (habitat II), which comprised 40% of the total available area. Roan antelope occupied this habitat as well but also ventured onto rocky slopes further away from water resources (habitat I), covering an area that comprised 87% of the total available. Habitat diversity in the fenced section where the animals were confined to a large extent represented the conditions which prevailed in the total reserve area and throughout the Waterberg region. When the potential of the fenced section to sustain the three ungulate species is estimated, sable show an affinity to the third and probably fifth habitat types. Tsessebe, however, are more strongly associated with habitat variables than are sable, suggesting that the latter can utilize a wider range of habitats and resources. There are several drawbacks for this type of approach. First, the analyses rely on the premise that habitat use by the ungulates reflects a state of optimal requirements. Thus, if preferred habitats in the study area are not optimal for the species, it will not show in the analysis. Secondly, the adequacy of an assessment depends on the selection of appropriate habitat variables. Further improvement can be made, for example, by the inclusion of preference ratings of food items. Thirdly, there is no indication of the relationships between animal species, be they other ungulate species or predators. This limitation can be partly overcome by incorporating other ungulate abundance and predators as habitat variables. These improvements in turn could define more precisely the balance between resource availability and requirements in the reserve and facilitate the implementation of management practices.

ACKNOWLEDGEMENTS I would like to thank Drs H. Dott for providing the cluster analysis programme and N. Fairall for commenting on the manuscript. Financial assistance was provided by the Endangered Wildlife Trust, Legamed Pharmaceuticals, Bob Blundell Memorial Trust and the Council for Scientific and Industrial Research. Lapalala Wilderness provided accommodation for the period of fieldwork. This is a publication of the Mammal Research Institute, University of Pretoria.

Habitat preference of African ungulates

365

REFERENCES Alldredge, J. R. & Ratti, J. T. (1986). Comparison of some statistical techniques for analysis of resource selection. J. Wildl. Manage., 50, 157-65. Ben-Shahar, R. (1987). Grasses and habitat relationships on a sour bushveld nature reserve. Vegetatio, 72, 45-9. Ben-Shahar, R. & Skinner, J. D. (1987). The conservation of roan, sable and tsessebe in the Waterberg region. Aft. Wildl., 42, 95-6. Ben-Shahar, R. & Skinner, J. D. (1988). Habitat preferences of African ungulates derived by uni- and multivariate analyses. Ecology, 69, 1445-51. Black, C. A. (1965). Methods of Soil Analysis, 1. Physical and Mineralogical Properties including Statistics of Measurements and Sampling. American Society of Agronomy, Madison. Child, G., Robbel, H. & Hepburn, C. P. (1972). Observations on the biology of tsessebe, Damaliscus lunatus lunatus, in northern Botswana. Mammalia, 36, 342-88. De Vos, A. & Mosby, H. S. (1969). Habitat analysis and evaluation. In Wildlife Management Techniques, ed. R. H. Giles. The Wildlife Society, Washington, DC, pp. 135-71. Greenacre, M. J. (1984). Theory and Application of Correspondence Analysis. Academic Press, London. Grobler, J. H. (1973). Aspects of the biology, population ecology and behaviour of the sable Hippotragus niger niger (Harris 1838) in the Rhodes Matopos National Park, Rhodesia. Arnoldia (Rhod.), 7(6), 1-36. Huntley, B. J. (1982). Southern African savannas. In Ecology of Tropical Savannas. Ecol. Stud. No. 24, ed. B. J. Huntley & B. H. Walker. Springer-Verlag, Berlin, pp. 101-19. Jackson, M. L. (1985). Soil Chemical Analysis. Prentice-Hall, Englewood Cliffs, NJ. Joubert, S. C.J. (1976). The population ecology of the roan antelope Hippotragus equinus equinus (Desmarest 1804) in the Kruger National Park. DSc thesis, Pretoria University, Pretoria. McNeill, L., Kelly, R. D. & Barnes, D. L. (1977). The use ofquadrat plotless methods in the analysis of tree and shrub component of woodland vegetation. Proc. Grassld Soc. Sth. Afr., 12, 109-13. Moen, A. M. (1973). Wildlife Ecology. Freeman, San Francisco. Mueller-Dombois, D. & Ellenberg, H. (1974). Aims and Methods of Vegetation Ecology. Wiley, New York. Smithers, R. H. N. (1986). South African red data book--terrestrial mammals. S. Aft. Nat. Sci. Progr. Rep., No. 125. CSIR, Pretoria. Spurway, N. C. (1981). Objective characterization of cells in terms of microscopical parameters: an example from muscle histochemistry. Histochem. J., 13, 269-317. Scholze, H. (1965). Climate of South Africa. Government Printer, Pretoria. t'Mannetje, L. & Haydock, K. P. (1963). The dry-weight-rank method for the botanical analysis of pasture. J. Br. GrassM Soc., 18, 268-75. Wilson, D. E. & Hirst, S. M. (1977). Ecology and factors limiting roan and sable populations on South Africa. Wildl. Monogr., 52, 1-111.