Cryptic Consumers and the Ecology of an African Savanna

East African savannas have fascinated generations of explorers, scientists, and the public. Nowhere else is there such a concentration and diversity or large, charismatic animals, and nowhere else is ecology portrayed so dramatically. Spectacular migrations of a million wildebeest tracking the rains dominate the Serengeti—Mara ecosystem of Tanzania and Kenya (McNaughton 1976, Sinclair and Norton-Griffiths 1979, Sinclair and Arcese 1995). Lions and hyenas pursue gazelle, zebra, and buffalo in a conspicuous contest between predator and prey (Kruuk 1972, Schaller 1972, Sinclair 1977). And giraffe and elephants shape the landscape by browsing and trampling savanna trees (Cumming 1992, Dublin 1995). The abundance of so many large mammals in East Africa has led to the assumption that interactions among these prominent animals dominate the ecological dynamics of savanna communities. Recent research, however, has shown that there is also a cryptic web of interactions hidden in the tall savanna grasses. This web is centered around a group of herbivorous mammals that share both habitat and forage with their much larger hooved counterparts. These are the small mammals, and until recently almost nothing was known about their importance in savannas of East Africa. The small mammals of East Africa include a diverse assortment of rodents, insectivores, elephant shrews, and rabbits. Aspects of the natural history of many of these inconspicuous mammals have been known for decades, including information about their diets, reproductive biology, and habitat preferences (Delany 1972, 1986, Kingdon 1974, 1997, Packer 1983, Leirs 1995). The most widely reported observation about their biology is that several rodent species periodically undergo dramatic population outbreaks. For example, a report from the Serengeti National Park in Tanzania described an outbreak of grass rats (Arvicanthis) in which the rats were so abundant that “one could hardly avoid stepping on them and many were killed by passing trucks. Survivors were feeding on the dead” (Hubbard 1972, p. 425). In agricultural areas, the multimammate mouse (Mastomys natalensis) reaches population peaks during which it destroys as much as 80–100% of crops throughout its range in sub-Saharan Africa (Taylor 1968, Fiedler 1988, Leirs 1995). Destruction of crops and stores caused by hordes of rodents has repeatedly inflicted heavy costs on humans. The cascading effects of ungulates on small mammals may have profound implications for ecological dynamics in east african savannas Population outbreaks of Arvicanthis and M. natalensis around human settlements have been considered particularly problematic because these species are reservoirs for a number of diseases, including Lassa fever, plague, spotted fever, relapsing fever, leptospirosis, leishmaniasis, and Crimean—Congo hemorrhagic fever (Fiedler 1988, Gratz 1997, Mills et al. 1997, Oguge et al. 1997). Although the ecology of diseases in Africa is poorly understood, studies in other areas have linked increases in disease transmission to increases in population densities of reservoirs for the disease (Mills et al. 1992, Ostfeld 1997). Moreover, in Africa there are reports of disease outbreaks being linked to high population densities of rodents (Monath et al. 1974a, 1974b). This dual pest status for East African rodents as both destroyers of crops and carriers of disease has led to great interest in the population dynamics of particular species (Neal 1977, 1981, Delany 1986, Leirs 1995). Understanding the triggers for rapid population growth could suggest effective ways of predicting, preventing, or mitigating population outbreaks, thus potentially decreasing their impacts on humans (Fiedler 1988, Leirs 1995). Since the 1930s, outbreaks of both Arvicanthis and Mastomys have been linked to heavy rainfall (Taylor 1968, Taylor and Green 1976, Leirs et al. 1994, Leirs 1995), and recent studies suggest that this connection could prove useful in establishing effective rodent control measures around human settlements (Leirs et al. 1996). In North and South America, Europe, and Japan, however, population dynamics of rodents in natural and semi-natural areas are known to be influenced by biotic interactions, especially by predators and competitors (Korpimäki and Norrdahl 1991, 1998, Heske et al. 1994, Meserve et al. 1996, Saitoh et al. 1998). These findings raise the question of whether African rodents are regulated primarily by rains, or whether they are also members of webs of interacting species. Such interactions could affect the abundance of both the rodents and the species with which they interact. These ecological relationships could have important implications for areas of human settlement as well as natural areas with comparatively little human impact. Until recently, however, ecological interactions involving small mammals in East Africa had not been seriously considered. Do herbivorous small mammals compete with large mammals for food? Are small mammals abundant enough to influence the dynamics of other species within savanna communities? Answers to such questions could provide important insights, both for scientists concerned with small mammals as pests of humans and for those concerned with preserving and protecting natural savanna habitats. The Mpala Research Centre (MRC) in central Kenya, where I have been investigating ecological interactions involving small mammals, is located on the equator at an altitude of 1800 m. Rains at MRC, as elsewhere in tropical areas, are highly seasonal. Heavy rains typically fall between April and July and then taper off until a period of lighter rains occurs in October through November. From January through March there is typically no rain at all, although the onset, intensity, and duration of both rainy and dry periods vary considerably from year to year (Keesing 1998a). The vegetation at MRC is dominated by the whistling thorn tree (Acacia drepanolobium), five species of grasses (Themeda triandra, Pennisetum mezianum, Pennisetum stramineum, Brachiaria lachnatha, Lintonia nutans), and several forbs (Aerva lanata, Helichrysum glumaceum; Young et al. 1998). This habitat harbors a rich diversity of ungulates, including elephants (Loxodonta africana), giraffes (Giraffa camelopardalis), Grevy's and Burchell's zebra (Equus greveyi and Equus burchelli), hartebeests (Alcelaphus buselaphus), buffalos (Syncerus caffer), elands (Taurotragus oryx), steenbucks (Raphicerus campestris), oryx (Oryx beisa), and Grant's gazelles (Gazella granti). Carnivores are also common, including lions (Panthera leo) and an assortment of smaller predators, such as mongooses (Ichneumia albicauda, Herpestes sanguineus) and black-backed jackals (Canis mesomelas). Patas monkeys (Erythrocebus patas) and vervet monkeys (Cercopithecus aethiops) are being studied in the area (Isbell et al. 1998), and a classic study of sexual selection in long-tailed widowbirds (Euplectes progne) was conducted nearby (Andersson 1982). A current study is investigating interactions among several species of ants and the whistling thorn tree (Young et al. 1997, Stanton et al. 1999). MRC is more representative of most of East Africa than are the famous game parks, such as the Serengeti of Tanzania and the Maasai Mara of southern Kenya. Ungulate densities are lower at MRC than in the parks, and because the land is outside park protection native grazers share much of the area with livestock (LWF 1996). The importance of conserving wildlife in areas outside of reserves and national parks is currently the subject of intense debate in East Africa (McRae 1998). Of particular concern is how to resolve the apparent conflicts between humans and their livestock, on the one hand, and populations of native ungulates, on the other, given that some scientists and pastoralists believe that native grazers compete with cattle for food (Prins 1992, Happold 1995, Swift et al. 1996). Because livestock and native ungulates coexist at MRC, this site is an ideal location for studies of the interactions among these two groups of grazers. In August 1995, the Kenya Long-Term Exclusion Experiment (KLEE) was established at MRC to investigate such interactions and to consider the separate and combined effects of domestic and wild ungulates on vegetation (Young et al. 1998). The experimental removal of native large mammals also simulates reductions in ungulate abundance caused by the encroachment of human populations, which is a critical conservation issue throughout East Africa (Happold 1995). KLEE excludes several combinations of native ungulates and livestock using six treatments (various combinations of +/− elephants and giraffes, +/− cattle, and +/− other native ungulates; Figure 1), which are established through a combination of regulated herding for livestock and electrified fencing for native ungulates. Because livestock are continuously accompanied while they graze at MRC, their movements can be controlled by herders. Native ungulates are excluded using two types of fencing, one of which excludes all ungulates (Figure 2), and the other of which excludes only the biggest ungulates (giraffe and elephants). Each treatment area is large—200 m × 200 m (4 ha)—and there are three replicates of each of the six treatments. Schematic diagram of the Kenya Long-Term Exclusion Experiment large-mammal exclusion design. Each treatment area is 4 ha (200 m × 200 m), and there are three replicates of this block design. The focal treatments for the study of small mammals (i.e., all ungulates allowed access or no ungulates allowed access) are shaded. One of the Kenya Long-Term Exclusion Experiment fences at the height of the dry season. This treatment access only by livestock, which and through the The trees in the are whistling thorn Because the of so much ecological in East Africa has been on ungulates, one question about small mammals is how their abundance and diversity are by large Because many small mammals are or it that the removal of herbivorous large mammals, such as and the of food for some small the densities of those small mammals removal of the large This of is the classic to whether are for 1983, the KLEE fences were being in (Young et al. 1998), I established in the of two of the large-mammal exclusion treatment that excluded all ungulates and the control which excluded (Figure Each and there were three replicates of each The most small on all the to be the a rodent (Figure that is in East Africa had been reported in high abundance (Keesing MRC, of the small mammals in small mammals the mouse the mouse and several species of The is the most small at the study of this was the with which to the exclusion of ungulates. January 1996, 4 the fences were were more abundant where there were no large mammals than they were on control (Figure the the so that by August there were as many the fences as in the control areas (Figure has been during the in abundance that are in In the of this habitat for there are no ungulates. of the in the and of ungulates. was in August 1995, the large-mammal exclusion fences were in abundance between with and ungulates are on a of could ungulates the of the habitat for The most and In the the removal of large mammals could an in the of in the The of has been to be important for protecting small mammals from predators, particularly et al. 1976, and which are abundant at predators at the site include small black-backed and and of these predators was excluded by the KLEE fences (Keesing 1998a). ungulates were be to these predators, which could to an in their In the ungulate removal could an in the or of food to the This suggest that ungulates and are for food in this habitat and that the removal of the ungulates caused an in their the on several of the of mouse not to have been to a of ungulates were of on the and the fences were that there was no in the of with or ungulates for year were not more to predators ungulates were (Keesing of not on treatment control of was ungulates were then have from one to the in these areas as to areas ungulates. to be the more for the of mouse on of and of the on treatment and control of by in the two areas were from the which the that during This is a of the area of an on from which ungulates had been excluded (Figure for were not although they a (Figure where were as because ungulates were (Figure each mouse this had no on were not in the two treatments. were not because could have the In other where ungulates were they This observation that food was for in the of ungulates. to be with ungulates for by small mammals within a on with and ungulates. on ungulates than on with ungulates. were for all at during a all the August 1995). For each were using a of with and treatment as The exclusion of ungulates the abundance of small mammals as well as their the removal of ungulates in a in the number of small mammals on the (Figure of that was to increases in the number of with ungulates, of the on ungulates that to number of small mammals on with and ungulates August between were on a the for each block The for most of the small mammals in each The species diversity of a both species number of and species of the that is of each A with high species have one of the species is much more abundant than the The both of these measures in a most of the as using this was the whether or not ungulates were (Figure at the of the diversity on the control Arvicanthis and were not at all on control although both species were on ungulate exclusion The of these two species on control may have been the of an dry in of small mammals at MRC typically as the dry from January through March (Figure and by the the rains in 1996, densities of Arvicanthis and on the with ungulates may have been so that these two species of small mammals on with and ungulates August Since August 1997, diversity has been on the with ungulates than on the ungulates for August 1996, an dry may have caused of some species on with ungulates. on the two of was using a of the rapid in diversity in August 1996, diversity to in both the and of ungulates (Figure Since that however, there has been a diversity in the of ungulates than in their with ungulates have of the species in the in For example, the rains, a of are on control than on ungulate exclusion that ungulates are the in this in through than through MRC, ungulates, especially elephants (Figure vegetation and the their and these have and from Such in increases in some such as a which is abundant only in areas the rains especially such as elephants and may that diversity through their effects on vegetation on This is in with the which that of one or a in diversity African by elephants and other large mammals may in species diversity of small influence the ungulates are is lower and diversity is although diversity has the to ungulates are small mammals, especially are more abundant and diversity is mammals, are important of dynamics in this large mammals or could increases in the densities of small mammals have important for the savanna Although heavy impacts of rodent populations on crops have been in Africa Leirs 1995), effects of rodents on native savanna vegetation are In herbivorous rodents can and species within and interactions may between African small mammals and savanna communities. these I established exclusion fences in August and these with control The fences are each m × of m and m the to small mammals from to the of the animals from The exclusion fences were located within the fences that ungulates because elephants fences they had access to effects of small mammals on vegetation the effects of high densities of small mammals, especially The exclusion of high densities of small mammals in a rapid and dramatic on was on both experimental and control (Figure by the of the year the in small mammals was than that in to which small mammals had the of the this had to a that was through the of the year (Figure of small mammals on in was on m × m with and small The of on on small mammals than on with small mammals, and this was on a of rainfall in and in an in on all The of small mammals on is given that the the small has been reported to be an with a for 1974, Although a could have an on such an be to more than the I especially because vegetation at the study site not to be of the species and are than the of at the study then a at and I conducted on in which they were an assortment of a and an of in The almost vegetation by and this species is primarily a at during some which may its rapid on This for vegetation was apparent in the habitat as where many of vegetation to those by in habitats. In both the feeding and of they in the both forbs and grasses and preferences for species in mammals may also influence the of vegetation at the study the are for and a can be by more than five in to other species of small mammals (Keesing Because many animals their from a food the be more than those from the (Keesing 1997). A study that experimental were at than those from the m of the of of the were at of m from the of the only of the were (Figure 1997). Because are both and it is not whether these to native of removal studies using of the on small mammals of the of a and that there is in removal which to be to the of rodents Recent has that of within and Young in intense by small mammals has the to influence both the and the of this The of from as a of the from the of the each of six removal was by at and from the of the to a of m. the of the of on vegetation with it is to the by on the The of removal with on a of that affect of may in have cascading for this savanna These trees a in the of the vegetation at the study site (Young et al. 1998). the trees are widely although there is in their (Young et al. 1998). are important in savannas et al. 1994, and because the of vegetation is savanna trees than in areas et al. as a of and and the vegetation could animals, which the through of and Because is in large by ungulate and because of by small mammals to be by a ungulates to grass through their effects on small ungulates be to of tree which could of tree ungulates. The of these on removal and that the effects of by small mammals on savanna vegetation can be those of the ungulates. small mammals to a large for the exclusion of the in this of the of vegetation in areas with and ungulates of ungulate vegetation the fences not from that on control This may the of of small mammals caused by ungulates and the reductions in vegetation this of the effects of small mammals on vegetation that small mammals at MRC have to those of other small mammals of each small at a of (Equus for at a of only such as these are the for the between and of mammals The of small mammals in areas with ungulates is on this the of is by of ungulates. In the of ungulates, the of small mammals is than the in the of ungulates. The in the to that by of ungulates. ungulates are small mammals have the to for of ungulates. this to the of ungulates excluded by the Although the of ungulates at the KLEE site has not been of ungulates throughout a area of in which MRC is led to an of of ungulates (LWF 1996, 1998a). among habitat types within the (LWF 1996). the at the KLEE site is lower than this and the of the small mammals to be ungulates are and small mammals are at their is of that of their larger by Although small mammals are they may be almost as important to the dynamics of African savanna vegetation as the more conspicuous and and The findings in this that savanna dynamics at MRC are dominated by large mammals, in some ungulates are densities are diversity to and the impacts of small mammals on the are (Figure ungulates are or densities diversity is lower more and the can have a profound on the and of the (Figure of by small ungulates are they a high of which in a for small small mammals have a small on ungulates are or they little or no in a high for small These small mammals then for by the large mammals, in to that large mammals are These cascading effects on the in the of ungulates may have profound implications for ecological For example, or native ungulate populations in East Africa have reductions in native ungulate populations small then densities of small mammals, particularly rodents, could to transmission of both human and livestock diseases for which rodents are And high densities also the of predators that on small mammals, including and small high densities of small mammals for by native ungulates, then native ungulates not range for livestock because small mammals could forage for cattle and other grazers. more on African it is whether the interactions among small mammals, large mammals, and vegetation at MRC are The habitat in which has been conducted is in East Africa. The and vegetation that the KLEE site more than of the ecosystem (Young et al. and are in many other of East including National Park and the of Serengeti National Park (Young et al. 1998). dynamics to those described in this at these other has not been is it known whether are in savannas with and The at MRC may from that of other areas of East Africa in at one Although the mouse is the most abundant small at MRC, it has not been reported as abundant at the other for which abundance has been (Delany Moreover, the mouse has not been reported as a pest species. it as a for human or livestock diseases to be Such a is given that for other disease the species in the is a highly for and diseases (Mills et al. 1992, et al. 1994, Ostfeld The two most small mammals throughout East at around human are Arvicanthis and M. both of which at the KLEE the Mastomys in abundance ungulates were excluded (Keesing 1998a). the densities of both Mastomys and Arvicanthis are lower than the of in both the and the of ungulates. is at where these two species dominate the small ungulates affect them as as they affect at This could be critical to in studies because of the of these species as agricultural pests and disease important area for is whether livestock and native ungulates affect small abundance and diversity in of the on the effects of ungulates on small mammals in other has on livestock et al. et al. Heske and 1991, and and it is that heavy affect vegetation and can dramatic in the species and densities of small livestock have these effects in East Africa is not In of the interactions between ungulates, small mammals, and vegetation described in this and the about between native and domestic ungulates, impacts of cattle is critical for effective of wildlife and habitats. I currently investigating the separate and combined effects of livestock and native ungulates on small mammals at in other areas of East Africa are of the ecology of African savannas have much of the for the of species interactions within ecological communities. The of these studies have on the most conspicuous members of savanna communities. some of the most important and interactions may be Understanding cryptic in may prove for and conserving both natural and habitats. I the of the of the of Kenya and the Kenya for their which has been for the of this on East African small mammals has been by the National the National the of the of for the the for and the of the of and the of at the of This could not have been conducted the of the Kenya Long-Term Exclusion for which and great are to the the National and the Mpala has been by and the of and has been by the Mpala Research Centre and the of in This was by the of and five