Answer all questions using full sentences and use Part I, Part II, and Part III as your headings to organize your paper. Design your own headings for these main sections, and under each main section, use subtitles to organize your answers to the questions. You should be able to address the requirements of this assignment in 6-8 paragraphs.

- Reference List Required (reading material provided).

- APA Style.

B I O L O G I CA L C O N S E RVAT I O N 1 3 9 ( 2 0 0 7 ) 2 1 9 -2 2 9 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/biocon Carrying capacity of large African predators: Predictions and tests Matt W. Haywarda,*, John O'Brienb, Graham I.H. Kerleya a Terrestrial Ecology Research Unit, Department of Zoology, Nelson Mandela Metropolitan University, Port Elizabeth, 6031, Eastern Cape, South Africa b Shamwari Game Reserve, P.O. Box 91, Paterson 6130, South Africa A R T I C L E I N F O A B S T R A C T Article history: Successful conservation initiatives often lead to rapid increases in large carnivore densities Received 8 October 2006 to the extent that overpopulation occurs. Yet conservation managers have no way of know- Received in revised form ing the carrying capacity of their reserves. Here we derived relationships between the pre- 27 May 2007 ferred prey (species and weight range) of Africa's large predator guild and their population Accepted 22 June 2007 densities to predict their carrying capacity in ten South African conservation areas. Conser- Available online 9 August 2007 vation managers intervened at several of these sites because of evidence of predator overpopulation and these provided independent tests of our predictions. Highly signicant Keywords: linear relationships were found between the biomass of the preferred prey species of lion, Acinonyx jubatus leopard, spotted hyaena and African wild dog, and the biomass of prey in the preferred Conservation ecology and weight range of cheetah. These relationships are more robust than previous work for lion, management cheetah and leopard, and novel for spotted hyaena and African wild dog. These relation- Crocuta crocuta ships predicted that several predators exceeded carrying capacity at four sites, two where Lycaon pictus managers expressed concerns about overpopulation due to a decline in wildlife abundance Panthera and two where carnivores were actively removed. The ability to predict the carrying capacity of large predators is fundamental to their conservation, particularly in small enclosed reserves. Every predator that preys on large, readily surveyed wildlife can have its carrying capacity predicted in this manner based on the abundance of its preferred prey. This will be benecial for reintroduction attempts, threatened species management, overpopulation estimation, detecting poaching and in investigating intra-guild competition. 2007 Elsevier Ltd. All rights reserved. 1. Introduction Reduction in distribution and abundance has led to almost 25% of extant Carnivoran species being threatened with extinction (Ginsberg, 2001). Their conservation ultimately depends upon the accurate assessment of their distribution and abundance to facilitate informed management decisions (Fuller and Sievert, 2001; Gros et al., 1996). In some places, conservation managers have started slowing these declines through translocations and reintroductions (Breitenmoser et al., 2001). Conservation areas where such translocations have occurred are often fenced and heavily managed, and these populations tend to increase rapidly in the absence of threatening processes (Smith, 2006). Managers of such sites are therefore faced with potential overabundance of translocated stock, without knowing the carrying capacity of these species, or the maximum number of individuals that a site can support without causing its deterioration. This is particularly _ * Corresponding author: Present address: Marie Curie Fellow, Mammal Research Institute, Polish Academy of Science, 17-230 Biaowieza, Poland. Tel.: +27 (0)41 504 2308; fax: +27 (0)41 504 2946. E-mail address: hayers111@aol.com (M.W. Hayward). 0006-3207/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2007.06.018 220 B I O L O G I C A L C O N S E RVAT I O N important when such sites are relatively small, fenced and/or situated in hostile environments where movements of freeranging animals will be necessarily curtailed. Carnivore densities can vary over several orders of magnitude within species, but, in natural ecosystems, generally reect the abundance of their prey (Bertram, 1975; Fuller and Sievert, 2001). Seminal work by van Orsdol et al. (1985) illustrated this relationship with lions (Panthera leo). Such relationships have subsequently been found in cheetah (Acinonyx jubatus) (Laurenson, 1995b) and leopard (Panthera pardus) (Stander et al., 1997). A relationship also exists between tigers (Panthera tigris) and their prey (Karanth et al., 2004) and such relationships have been used to predict the size of reintroduced Eurasian lynx (Lynx lynx) populations (Hetherington and Gorman, 2007). Similarly, the density of grey wolf (Canis lupus) is related to that of its prey, particularly moose (Alces alces) (Peterson et al., 1998, in Fuller and Sievert, 2001). Relationships between predator and prey density apply across the order Carnivora, where 10,000 kg of prey supports about 90 kg of a given carnivore species (Carbone and Gittleman, 2002). In African savannas, predator-prey relationships are related to rainfall and vegetation productivity (East, 1984). While the initial research that identied the relationships between predator and prey density greatly improved our understanding of predator ecology, recent research on prey preferences allows us to investigate these relationships more intensively. For example, lion density was initially linked to the biomass of all available prey species (van Orsdol et al., 1985), while cheetah density was related to the biomass of prey weighing between 15 and 60 kg along with a negative relationship with lion density (Laurenson, 1995b). Leopard density exhibited a signicant relationship with the biomass of prey weighing between 15 and 60 kg (Stander et al., 1997). These relationships are likely to be substantially improved 1 3 9 ( 2 0 0 7 ) 2 1 9 -2 2 9 by using the biomass of preferred prey species or preferred prey weight range of each predator (Table 1). African wild dogs (Lycaon pictus) have never been a common species and there are probably basic ecological reasons for their scarcity (Creel and Creel, 1996), such as competitive limitation by lions and spotted hyaenas (Crocuta crocuta) (Creel and Creel, 1996) or cheetah (Hayward, unpubl. data). Yet there has been no study linking wild dog density with that of their available prey as was intimated by Fuller and Sievert (2001). The use of preferred prey species biomass or the biomass of prey in the wild dog's preferred weight range may yield such predictions (Table 1). Although the Serengeti spotted hyaena population more than doubled during the corresponding increase in blue wildebeest (Connochaetes taurinus) abundance (Hofer and East, 1995), there has been no study to link hyaena density with that of their prey. Again strong relationships may be derived using the hyaena's preferred prey weight range, however given the high degree of dietary overlap with lions (Hayward, 2006), there may be a relationship between hyaena biomass and the biomass of large body mass prey that is preferred by lions (Hayward and Kerley, 2005). No previous study has applied these predator-prey relationships to predicting predator carrying capacity, yet this is precisely the opportunity that the relationship between predator and prey density affords us (Fuller and Sievert, 2001). Here we used data from 22 reserves in eastern and southern Africa over different periods yielding 32 groups of population estimates to examine abundance relationships between predators and their prey using more detailed information on prey choice (Hayward, 2006; Hayward et al., 2006a; Hayward et al., 2006b; Hayward and Kerley, 2005; Hayward et al., 2006c). We then use our new regression equations and the information about prey choice to predict carnivore carrying capacity in ten sites in South Africa where reintroductions were planned Table 1 - Preferred prey species and preferred prey body mass range of Africa's large predator guild Predator species African wild dog Cheetah Leopard Lion Spotted hyaena Preferred prey species Kudu Tragelaphus strepsiceros Thomson's gazelle Gazella thomsoni Impala Aepyceros melampus Bushbuck Tragelaphus scriptus Blesbok Damaliscus dorcas phillipsi Impala Thomson's gazelle Grant's gazelle G. granti Springbok Antidorcas marsupialis Impala Bushbuck Common duiker Sylvicapra grimmia Blue wildebeest Connochaetes taurinus Buffalo Syncerus caffer Gemsbok Oryx gazelle Giraffe Giraffa camelopardalis Plain's zebra Equus burchellii Nil, but high (69%) overlap of preferred prey of lions Preferred body mass range is based on 3/4 of adult female body mass. Preferred prey body mass range (kg) 16-32 and 120-140 Reference Hayward et al. (2006c) 23-56 Hayward et al. (2006b) 10-40 Hayward et al. (2006a) 190-550 Hayward and Kerley (2005) 56-182 Hayward (2006) B I O L O G I C A L C O N S E RVAT I O N or had occurred and where wildlife census data was available, but were not used in the derivation of our new relationships. These predictions were tested at sites where managers had expressed concern about carnivore overpopulation or had intervened due to declines in prey abundance. It is only with knowledge of predator carrying capacity that informed conservation management decisions can be made (Fuller and Sievert, 2001), such as predator reintroductions or removals, implementation of fertility control, or plans for park expansion. 2. Materials and methods We reviewed the literature using electronic databases (Current Contents, Biological Abstracts, Web of Science), libraries and reference lists of other papers, and tabulated data on predator density and prey abundance at individual sites from the savanna ecosystems of southern and eastern Africa. This information was converted to biomass km2 using 3/4 of adult female body mass (following Schaller (1972) to account for sub-adults and young preyed upon) estimates from Stuart and Stuart (2000) (see Appendix). Several studies were excluded because they were from extremely different habitat types, such as Afromontane forest (Sillero-Zubiri and Gottelli, 1992) and/or produced outlying results when plotted alongside other studies at that site at similar times (Dunham, 1992, 1994; Eloff, 1973; Kruger et al., 1999; Mills et al., 1978; Mizutani and Jewell, 1998). The methods used to gather these data varied among studies, however, like Creel and Creel (1996), we did not conduct post-hoc corrections to account for this as it was considered to be too subjective. If more than one density was recorded for an individual species in a decade, then either the mean of these was used or the estimate with the most accurate measure of prey abundance relating to it when obviously erroneous estimates were present. Like other authors (Creel and Creel, 1996; Grange and Duncan, 2006), the majority of sites used were relatively unaffected by humans, however several have been fully or partially fenced (e.g. Hluhluwe-Umfolozi, Kruger), others have culls or hunting (e.g. Kruger, Selous) or pastoralism (Ngorongoro), and other populations were reintroduced (e.g. Hluhluwe). Where reintroduced populations were included, a sufcient time (15 years) was left to allow the populations to attain carrying capacity. We then regressed predator density against the biomass of signicantly preferred prey and the biomass of prey within each predator's preferred weight range (see Introduction) using data presented in the Appendix and Table 1. These preferred prey species and weight ranges were calculated in previous studies on lion (Hayward and Kerley, 2005), leopard (Hayward et al., 2006a), cheetah (Hayward et al., 2006b), African wild dog (Hayward et al., 2006c) and spotted hyaena (Hayward, 2006). Given the high degree of dietary overlap between lions and hyaenas (Hayward, 2006), we also tested for relationships between hyaena density and that of the preferred prey of lions. We also regressed data presented in van Orsdol et al. (1985) on lion, in Stander et al. (1997) on leopard and Gros et al. (1996) on cheetah to derive predictive equations with which to compare those relationships derived from our work on prey preferences. We also used the equations calcu- 1 3 9 ( 2 0 0 7 ) 2 1 9 -2 2 9 221 lated by Carbone and Gittleman (2002) to compare their predictive accuracy. We then used these equations to predict the potential predator population density and size at ten sites where large predators have been or are being reintroduced using the wildlife census data presented in Tables 2 and 3. Essentially, these are our predictions of each sites' carrying capacity for each predator based on the available food resources in individual years. The sites seeking estimates of carrying capacity for large predators were fenced reserves in South Africa's Eastern and Western Cape Provinces that ranged in size from 70 to 3410 km2 (Tables 2 and 3). Addo Elephant National Park (Addo) is located 72 km north of Port Elizabeth in the Sub-tropical Thicket biome that supports dense thickets dominated by Portulacaria afra alongside grasslands derived from past agricultural practices (Vlok et al., 2003). Lions, spotted hyaenas and a leopard were reintroduced to the Main Camp section of Addo in 2003 and 2004 (Hayward et al., 2007a; Hayward et al., 2007b). There are plans to reintroduce lions into the Darlington and Nyathi sections of Addo (fully fenced and separate from Addo Main Camp) when wildlife densities attain sufcient levels to support a small population. The Greater Addo Elephant National Park (GAENP) includes these areas as part of the planning regime of a much larger reserve that will conserve thicket, savanna, grassland, fynbos, nama karoo and forest biomes (Boshoff et al., 2002). Shamwari Game Reserve is 40 km east of Addo and supports similar vegetation types to Addo (Vlok et al., 2003). Lion, cheetah, African wild dog and leopard have been reintroduced here since 2000, and the 2004 population estimates were 15 lion, 10 wild dog, two leopard and six cheetah (Hayward et al., 2007b). The Karoo National Park is 500 km north of Cape Town and is situated within the Nama Karoo biome and preparations are underway to reintroduce lion there. The Mountain Zebra National Park is 100 km north of Addo in the Nama Karoo biome. Cheetah were reintroduced there in 2007 (Hayward et al., 2007b). These sites support different vegetation communities however the habitat of large predators is dependent upon adequate prey (Hayward et al., 2007c; Karanth et al., 2004) and these sites support species diversity similar to sites throughout the rest of southern and eastern Africa. We compared our predicted population estimates with those made for the proposed GAENP (Boshoff et al., 2002). These published population estimates were based on theoretical area requirements of each predator rather than available food resources and hence provide independent estimates. Finally, we tested our predictions of large predator carrying capacity using sites where declines in prey species led to predator management. Eight lions were reintroduced to Madjuma Lion Reserve (15 km2; 2442 0 S; 2758 0 E) in 1996 and immediately caused declines in blue wildebeest which ultimately led to the removal of the lions (Power, 2002). Lions were reintroduced to Phinda Resource Reserve in 1992 and managers removed 30 between 1996 and 1998 due to similar wildebeest declines (Hunter, 1998). By 1995 there were 13 lion and 21 cheetah (Hunter, 1998). Cheetah predictions were tested on Phinda also, although no cheetahs were removed by managers due to their high mortality rate, but there was a precipitous decline in common reedbuck Redunca arundinum 222 Table 2 - Wildlife densities (# km2) at prediction sites Site Mass (kg) Shamwari 2001 Habitat Thicket, transformed grasslands Nama karoo Nama karoo 187.46 700 Addo 185 2004 2002 2003 2004 2002 2003 2004 2002 0.02 Nyathi Darlington 2004 Thicket, grasslands Thicket 134 0.03 1.49 0.23 0.02 1.22 0.42 0.04 1.40 0.45 0.03 1.07 0.05 1.04 0.06 1.13 2004 70 GAENP 2004 Nama karoo 90 0 0.02 0.40 0.03 0.19 0.02 0.25 0.69 0.83 0.92 0.71 0.77 0.87 1.27 1.35 1.52 0.39 0.41 0.39 0.08 0.07 0.14 4.81 0.33 0.54 0.64 0.50 0.91 0.78 1.03 4.57 8.44 5.27 0.54 0.74 0.64 0.35 0.27 0.34 0.56 0.52 0.65 1.43 1.54 1.95 0.60 Thicket, grassland, fynbos, savanna, forest, nama karoo 3410 0.20 1.88 0.16 4.85 1.67 1.68 0.18 4.85 1.67 1.84 0.18 5.33 1.44 4.81 0.61 0.23 0.29 0.82 0.49 0.75 0.75 2.49 4.81 0.56 0.26 0.33 0.96 0.49 0.74 0.85 3.72 4.93 0.59 0.28 0.36 0.85 0.43 0.69 0.96 4.75 1.65 0.28 5.33 1.44 0.43 4.93 0.66 0.26 0.46 0.96 0.49 0.84 0.11 5.21 0.19 4.85 0.64 0.18 4.27 0.13 0.15 4.81 0.18 0.19 5.26 0.23 0.29 0.22 0.69 0.20 1.47 0.59 0.75 0.21 1.63 0.69 0.75 0.16 1.87 0.75 0.85 0.21 1.87 0.85 0.11 1.73 0.96 0.11 0.06 0.01 0.07 0.01 0.06 0 2.41 1.63 1.48 1.44 1.79 0.74 3.31 0.04 2.54 0.06 1.19 0.04 4.76 0.10 0.02 1.58 2.46 0.34 2.83 0.74 0.30 2.65 0.77 0.37 0.48 1.36 2.54 0.53 1.43 2.79 0.41 0.79 2.54 1.36 1.86 2.15 0.11 0.04 0.44 0.29 0.01 0.19 1.24 0.90 0.21 0.08 0.71 2.43 0.11 0.29 3.12 1.25 9.58 6.25 0.12 0.14 7.52 0.39 0.02 1.43 0.92 0.36 0.13 0.33 0.93 0.33 0.35 0.35 0.57 0.45 0.30 0.35 0.78 0.62 0.37 0.98 0.38 0.68 0.74 0.37 1.23 0.33 0.29 0.63 1.47 3.78 0.75 4.76 0.12 0.02 0.65 0.07 0.57 0.34 6.34 0.04 0.01 0.75 0.52 0.01 0.36 0.01 0.04 0.54 0.06 0.49 1.64 0.30 1.90 1.63 0.31 1.61 1.99 0.15 1.62 2.22 2.22 0.22 0.27 0.79 4.16 26.08 0.63 0.01 0.53 0.30 0.03 0.03 0.15 5.90 0.23 0.68 0.75 0.79 0.05 0.51 0.15 0.43 0.08 Shamwari data comes from distance estimates of walked transects (J.O'Brien, pers. comm.), Greater Addo Elephant National Park (GAENP) come from predictions of future populations sizes by Boshoff et al. (2002) and remaining data from South African National Parks aerial census data (G. Castley, unpubl. data). These data can be converted to biomass (kg km2) by multiplying by 3/4 of the adult female body mass of each species (mass) based on body masses given in Stuart and Stuart (2000) for mammals and Schaller (1972) for ostrich. All sites are fenced. 1 3 9 ( 2 0 0 7 ) 2 1 9 -2 2 9 0 0.02 0.01 0.01 0.94 0.27 4.96 1.33 0.35 4.53 0.37 0.28 0.31 0.13 0.27 0.60 0.82 3.50 0.03 2003 B I O L O G I C A L C O N S E RVAT I O N Baboon 12 Blesbok 53 Buffalo 432 Bushbuck 46 Bushpig 46 Duiker, blue 3 Duiker, common 16 Eland 345 Elephant 1600 Gemsbok 158 Giraffe 550 Grysbok 7 Hartebeest 95 Hippopotamus 750 Impala 30 Klipspringer 10 Kob 45 Kudu 135 Nyala 47 Oribi 14 Ostrich 70 Reedbuck, bohor 32 Reedbuck, common 32 Reedbuck, mountain 23 Rhinoceros, black 800 Rhinoceros, white 1400 Sable 180 Springbok 26 Steenbok 8 Vervet monkey 3.5 Warthog 45 Waterbuck 188 Wildebeest, blue 135 Zebra, plains 175 Zebra, mountain 179 2003 Mountain Zebra 2000 Area (km2) 2002 Karoo Years B I O L O G I C A L C O N S E RVAT I O N 3. Table 3 - Wildlife densities (# km2) at test sites Site Years Madjuma Madjuma 1997 1998 Phinda 1995 Pilanesberg 1997 Savanna 15 Savanna 170 Savanna 70 Habitat Area (km2) Baboon Blesbok Buffalo Bushbuck Bushpig Duiker, blue Duiker, common Eland Elephant Gemsbok Giraffe Grysbok Hartebeest Hippopotamus Impala Klipspringer Kob Kudu Nyala Oribi Ostrich Reedbuck, bohor Reedbuck, common Reedbuck, mountain Rhinoceros, black Rhinoceros, white Sable Springbok Steenbok Vervet monkey Warthog Waterbuck Wildebeest, blue Zebra, plains Zebra, mountain 0.11 0.04 0.44 0.19 1.24 0.90 0.21 0.39 5.33 2.33 5.00 5.73 11.36 2.13 0.53 1.48 12.49 2.43 0.54 0.60 0.46 0.01 0.36 0.01 3.47 1.73 5.12 1.67 8.53 3.69 0.33 5.73 3.12 1 3 9 ( 2 0 0 7 ) 2 1 9 -2 2 9 0.03 0.03 0.43 Madjuma data comes from Power (2002), Phinda from Hunter (1998), and Pilanesberg data from van Dyk and Slotow (2003). Where stated these density estimates were derived from aerial censuses. These data can be converted to biomass (kg km2) by multiplying by 3/4 of the adult female body mass of each species from body masses given in Table 2. All sites are fenced. (Hunter, 1998). Lion were reintroduced to Pilanesberg National Park in 1993 (van Dyk and Slotow, 2003) and exceeded 50 individuals in 1998 (Tampling and du Toit, 2005). Four years later, excessive lion predation had led to declines in the blue wildebeest population of 45%, eland by 76%, waterbuck by 67% and kudu by 65% (Tampling and du Toit, 2005). We compared our estimates of lion carrying capacity with the number of lions in Pilanesberg and related these to the timing of the prey population declines. 223 Results The population density of each large African predator was signicantly related to the biomass of signicantly preferred prey and/or the biomass of prey in their preferred weight range (Table 4). The relationships that explained the greatest amount of variance in the data for each predator were biomass of preferred prey of lion (r2 = 0.626), leopard (r2 = 0.833), spotted hyaena (r2 = 0.487) and wild dog (r2 = 0.523), and preferred weight range for cheetah (r2 = 0.519; Table 4). We consider these our best methods of predicting the population size and carrying capacity of each predator. We quote these results hereafter, although both preferred weight range and preferred prey biomass provide very similar predictions (Pearson's R 0.95, p