In many aspects of science and technology, including the natural sciences, Africa is considered the ‘neglected continent’, due certainly to the endemic problems that have afflicted the development of African countries, including, for example, diffuse poverty, strong economic constraints on research investments, unsafe social and political conditions, and the negative effects of the colonial period. In addition, the whole African continent has been often forgotten in the general reviews on ecological topics although some valuable studies were available, and although Africa is nowadays becoming to be more present in the international scientific scenario in terms of economic investment to scientific research (e.g. Nigeria, see Nature News 2006) and number and quality of scientific papers produced.
In this paper, I review the history and development of community ecology studies on African reptiles, and by performing a meta-analysis of published datasets in independent studies (using null models procedures, see Gotelli & Graves, 1996) also examine what has been learnt about the structure and functioning of reptile communities with these studies. I demonstrate that historically, Africa has played a central role in herpetological community ecology studies, and that the current studies are being still contributing much to our understanding of reptilian community rules.
An historical perspective
Reptiles, especially lizards, have been popular as study models for community ecology since the 1960s and the 1970s, because certain features of their biology (for instance, their ectothermic physiology accomplished with enormously diverse morphologies, physiologies and life-histories) make them appropriate for examining community structure in a matrix of similarities and differences that may allow for generalizations of the results obtained to all groups of animals (e.g. Schoener, 1974; Toft, 1985). In this context, African communities of reptiles have not been marginal for the development and testing of ecological theory. For instance, one of the most relevant papers of the 1970s compared the reptile biomass by continents, and concluded that the biomass of reptile communities in Africa is much depressed, probably due to the impact of large herbivores (Janzen, 1976). This study has influenced lot of later papers, and will be discussed below.
The first authoritative review of community ecology in amphibians and reptiles was written by Toft (1985). She reviewed a total of 88 studies published worldwide, and focused her paper on 43 quantitative studies, mostly of them on lizards and amphibian larvae because, at that time, these organisms were the most popular models in community ecology research. Out of these 43 studies, only two (Inger & Marx, 1961; Poynton & Pritchard, 1976) were relative to African communities, but none of reptiles. The absence of African reptile papers reviewed by Toft (1985) may suggest that studies on African reptilian communities were scarce and peripheral to the general stream of community ecology research, but indeed this was just the outcome of an unsatisfying bibliographic search oriented mainly towards articles written in English. Indeed, since the early 1970s African journals were already publishing small papers of interest for reptile community ecologists (e.g. Broadley, 1974; Western, 1974) and, especially, Robert Barbault published a long series of crucial articles, mostly written in French, on the community ecology of lizards and snakes in Ivory Coast (Barbault, 1971, 1974a,b, 1975, 1976a,b, 1977; just to cite a few). These papers constitute certainly one of the most authoritative series of papers in reptilian community ecology, and likely still present the most complete dataset on the functioning of a specific community of tropical reptiles in the world (see also Scott & Campbell, 1982). Contextually to Barbault’s studies in tropical Africa, other authors investigated reptilian community structure in North Africa (Grenot & Vernet, 1972a,b; Vernet & Grenot, 1972a,b), and Pianka published some of his highly cited papers on desert lizards of southern Africa (e.g. Pianka, 1971; Huey et al., 1974; Huey & Pianka, 1977; Huey, Pianka & Hoffman, 1977; Pianka & Huey, 1978). Thus, although overlooked by Toft (1985), there is no doubt that studies on African herpetofaunas were central for the development of community ecology theory yet since the 1970s. The following decades have seen the publication of further papers by Barbault (e.g. in 1991) and Pianka (e.g. in 1986), where niche theory was examined in much detail, and a series of other contributions from both tropical and nontropical regions (e.g. Hughes, 1983; Nouira, 1983, 1988; Simbotwe, 1984). From the theoretical point of view, the most interesting aspect of the long series of papers cited above is that they followed directly the development and shifts in ecological thought: for instance, whereas papers published in the 1970s were focused on the role of interspecific competition that was thought to be central in shaping the herpetofaunal communities, more recent papers by the same authors included alternative explanations for community structure, including, for example, physiological performance, extrinsic constraints, feedback relationships and indirect relationships among coexisting species, and their deep evolutionary history (e.g. Barbault, 1991).
In the 1990s and 2000s, the studies on African reptile communities shifted in emphasis in part as a consequence of (i) the increased importance of conservation biology in general and (ii) the dramatic conservation status of several important African environments. Thus, many community studies were poorly oriented in terms of ecological theory, but became strongly focused on conservation matters. This shift in scientific emphasis is clearly seen not only in lots of papers on Malagasy herpetofaunas (e.g. Raxworthy & Nussbaum, 1994), but also in other studies on fragile habitats in coastal Ghana (Raxworthy & Attuquayefio, 2000) and in the oil-rich forest zone of Nigeria (Akani et al., 1999; Luiselli, Akani & Politano, 2006). Nonetheless, several studies have been still being published on aspects of community ecology theory applied to reptiles, and newer study subjects and study regions entered into the scene. In particular, novel study subjects included a variety of issues, for instance the community ecology of sympatric (i) tortoises in arid Zimbabwe (e.g. Hailey & Coulson, 1995; Hailey, Coulson & Chidavaenzi, 1997; Hailey, Coulson & Mwabvu, 2001) and in wet Nigerian forests (e.g. Luiselli, 2003a, 2005, 2006a); (ii) saxicolous lizards in Zimbabwe (Howard & Hailey, 1999); (iii) lizards in Kenya (Wahungu, Mumia & Nzau, 2004; Hardy & Crnkovic, 2006); (iv) small lacertids in northern Africa (Znari et al., 2000; Rouag et al., 2007); (v) chameleons in central and West Africa rainforests (Hofer, Bersier & Borcard, 1999; Hofer, Baur & Bersier, 2003; Luiselli, 2007); (vi) rainforest geckos in Nigeria (Luiselli, Eniang & Akani, 2007; Rugiero et al., 2007); and (vii) snakes from rainforests, mangroves, and derived savannahs in Nigeria (see Luiselli, 2006b, and references therein). All this bulk of studies allow us to have a new, more general look on how communities of reptiles are structured in the African continent.
Basics of resource partitioning
General theory establishes that in living communities the strength of interspecific competition is not equally intense at all climates and latitudes, being generally stronger in tropical than in nontropical climates (Pianka, 1966; Rohde, 1992). In addition, interspecific competition is more intense when the coexisting species are characterized by high density populations, i.e. it is stronger at high biomass levels of competitors (Pianka, 1986). Thus, since biomass is entirely relevant for having competition effects in a living community, and since in Africa it has been supposed that reptilian biomass is depressed (Janzen, 1976), a first key question should be: is African reptile biomass really depressed?
When interspecific competition is strong, the sympatric species tend to partition the available resources to minimize competition effects (e.g. Schoener, 1974; Roughgarden, 1976, 1983). Resource partitioning patterns may derive from the interaction of some categories of causes, including predation, extrinsic and intrinsic constraints on an organism’s performance, and interspecific competition (e.g. see Toft, 1985; Barbault, 1991; Barbault & Stearns, 1991), but in general nonrandom patterns of resource partitioning are associated to communities shaped by interspecific competition (Gotelli & Graves, 1996). These predictions have been widely confirmed by studies on sympatric snakes (Luiselli, 2006b), and are therefore probably valid for other ectotherms as well. So, we may infer that competition is somewhat important in shaping the African reptile communities if most of the studied communities show nonrandom resource partitioning patterns. Based on the theory, we may predict that reptilian communities from tropical Africa will exhibit nonrandom resource partitioning patterns more frequently than those from nontropical Africa, because of the supposed higher strength of interspecific competition. Thus, the second key question will be: are reptile communities from tropical Africa more likely to partition the available resources than those from nontropical Africa? These key questions will be examined in the following text.
Is African reptile biomass really depressed?
After Janzen’s (1976) study, the issue of a depressed biomass in tropical African reptile communities has never been examined in detail by later authors, nonetheless it has been widely accepted and frequently cited. I suggest that this pattern is not widely supported by data, although admit that more research is certainly due to this crucial argument. My reasons for being unsatisfied with Janzen’s argument are as follows:
- 1Large herbivores should have had a very different impact in savannahs versus forests because their abundance is much different among these macro-habitats, so reptilian biomass should be much more depressed in savannahs than in forests. This is because in African tropical forests, mammalian biomass is low, while species diversity is high, compared to more open, savannah tropical habitats (Robinson & Bennett, 2000; Barnes, 2002). African rainforests have, in general, poor soils and high annual mean rainfall and are not as productive as other habitats to support a large mammalian biomass, so habitat type seems very relevant here.
- 2There are some zones and habitat types where the density of snake species (even of a large size) is certainly impressively high (Phelps, 1989; Luiselli, 2006c; Akani et al., unpublished data).
- 3Reptilian biomass seems to be less in pristine remote habitats than in altered forests and farmbush habitats (Madsen et al., unpublished data).
Moreover, the available data on biomass comparisons among different areas are certainly ambiguous. For instance, the biomass of forest floor herpetofauna of Kibale Forest, Uganda (data in Vonesh, 1998) was much less than that of Lac Tissongo and Lombé rainforests, Cameroon (data in Scott, 1982) or Costa Rican rainforests (Scott, 1976), but similar to that of rainforests in Borneo (Lloyd, Inger & King, 1968) and Thailand (Inger & Colwell, 1977), thus showing that generalizations are hardly true or at least still unsupported by unambiguous field evidence. Therefore, for the future it will be necessary to study in much detail the issue of depressed reptile biomass in Africa to verify the generality of Janzen’s (1976) statement and to establish exactly which consequences this factor may have for the African ecological systems. For this moment, I preliminarily suggest that reptile biomasses in Africa are not particularly depressed in a variety (most?) of habitats throughout the continent, and therefore that this factor may have not affected remarkably the functioning of African communities of reptiles.
Are communities from tropical Africa more likely to partition the available resources than those from nontropical Africa?
If reptile biomass in African habitats is not depressed as previously suggested (Janzen, 1976), there is no reason for thinking that the main organization forces governing African assemblages of species are strongly different from those operating elsewhere. Thus, it is predictable that resource partitioning patterns should be more clearly visible in tropical than in nontropical reptile communities as a consequence of an increased competition in tropical climates (Pianka, 1966; Rohde, 1992). I tested this hypothesis by performing a meta-analysis on reptile communities in tropical versus nontropical Africa (Table 1).
|Species||Country||Resource axes tested||Resource axes partitioned||References|
|Freshwater turtles||Nigeria (unpolluted river)||Macrohabitat, microhabitat, diet||Macrohabitat, microhabitat, diet||Luiselli et al., 2004, 2006; Luiselli & Akani, 2003|
|Freshwater turtles||Nigeria (oil-polluted river)||Macrohabitat, microhabitat, diet||None||Luiselli et al., 2004, 2006; Luiselli & Akani, 2003|
|Tortoises (Geochelone pardalis, Kinixys spekii)||Zimbabwe||Macrohabitat, microhabitat||Macrohabitat, microhabitat||Hailey & Coulson, 1995|
|Tortoises (Kinixys erosa, Kinixys homeana)||Nigeria||Macrohabitat, microhabitat, diet, time||Diet||Luiselli, 2003a, 2005|
|Varanus ornatus, Osteolaemus tetraspis||Nigeria||Macrohabitat, diet||None||Luiselli et al., 1999|
|Terrestrial lizards||Cameroon||Macrohabitat, microhabitat (altitude)||Macrohabitat, microhabitat (altitude)||Hofer et al., 1999|
|Terrestrial lizards||Kenya||Diet||Diet||Hardy & Crnkovic, 2006|
|Terrestrial lizards||Ivory Coast||Macrohabitat, microhabitat, diet, time||Microhabitat, time||Barbault, 1987|
|Terrestrial scincid lizards||Ivory Coast||Time||Time||Barbault, 1991|
|Terrestrial scincid lizards||Kenya||Macrohabitat, microhabitat||Microhabitat||Wahungu et al., 2004|
|Terrestrial lizards||Nigeria||Macrohabitat, microhabitat, diet, time||Diet||Akani et al., 2002|
|Chameleons||Cameroon||Macrohabitat, microhabitat, (altitude)||Macrohabitat, microhabitat (altitude)||Hofer et al., 2003|
|Chameleons||Nigeria, Cameroon||Macrohabitat, microhabitat, diet||Microhabitat, diet||Luiselli, 2007|
|Terrestrial lizards||Zimbabwe||Macrohabitat, microhabitat||Macrohabitat||Howard & Hailey, 1999|
|Savanna snakes||Ivory Coast||Diet||Diet||Barbault, 1977|
|Rainforest snakes||Nigeria||Macrohabitat, microhabitat, diet||Diet||Luiselli et al., 1998|
|Cobras||Nigeria||Macrohabitat, microhabitat, diet, time||Macrohabitat Microhabitat||Luiselli & Angelici, 2000; Luiselli et al., 2002|
|Marsh snakes (Natriciteres)||Nigeria||Macrohabitat, microhabitat, diet, time||Diet||Luiselli, 2003b|
|Water snakes||Nigeria||Macrohabitat, microhabitat, diet, time||Diet||Luiselli et al., 2005; Luiselli, 2006c|
|Vipers (Bitis)||Nigeria||Macrohabitat, microhabitat, diet||Diet||Luiselli, 2006d,e|
|Fossorial lizards (Thyphlosaurus)||Namibia||Morphology, macrohabitat, microhabitat, diet||Morphology, diet||Huey et al., 1974|
|Agama lizards||Namibia||Diet||None||Heideman, 2002|
|Terrestrial lizards||Morocco||Diet||Diet||Carretero et al., 2006|
|Terrestrial lizards||South Africa||Macrohabitat, microhabitat, time||Microhabitat, time||Jacobsen, 1982|
|Terrestrial lizards (Sphenops, Scincus, Chalcides)||Egypt||Macrohabitat, microhabitat, thermal||Thermal||Attum et al., 2007|
|Terrestrial scincid lizards (Mabuya)||Namibia||Macrohabitat, microhabitat, diet, time||Microhabitat, time||Huey & Pianka, 1977|
|Terrestrial scincid lizards (Mabuya)||Namibia||Diet||None||Castanzo & Bauer, 1998|
|Terrestrial lizards||Algeria||Macrohabitat, microhabitat||None||Rouag, 1999; Rouag & Benyacoub, 2006|
|Terrestrial lacertid lizards||Algeria||Diet||None||Rouag et al., 2007|
|Terrestrial lacertid lizards||Morocco||Macrohabitat, microhabitat, diet||None||Znari et al., 2000|
|Terrestrial lizards||Tunisia||Diet||None||Nouira, 1983|
|Amphisbaenians||South Africa||Diet||None||Webb et al., 2000|
I used for this meta-analysis only studies, published in peer-reviewed international journals or in university dissertations, that (i) explicitly tested resource partitioning and interspecific competition hypotheses in African reptile communities and/or that (ii) provided data-sets fully re-analysable by statistical procedures (see below). Therefore, I excluded from analysis several studies that, although interesting and data-rich, investigated aspects of community structure different from resource partitioning and competition (e.g. Raxworthy & Attuquayefio, 2000, etc.). As done in earlier reviews (e.g. Toft, 1985), I included in the analysis both studies considering all important resource dimensions (i.e. habitat, food and time, see Pianka, 1986) and those deliberately selecting certain dimensions although others may be important.
Datasets were inspected to find nonrandom structure of the various communities along four resource dimensions: macro-habitat, micro-habitat, food and time. To evaluate whether each community was structured randomly or not, I contrasted the actual data matrix as given in the original literature source with random ‘pseudo-communities’ generated by Monte Carlo simulations (Gotelli & Graves, 1996). I used the EcoSim software (Aquired Intelligence Corp., Kesey-Bear, VT, USA) to calculate overlap indices and to generate Monte Carlo simulations. I parameterized resource items data as presence versus absences. As too many zeroes in the matrices might distort the error levels while too often reject structure, I fixed zeroes prior to any analyses. This was also justified by the fact that the different sizes of the species within each community (e.g. Luiselli, Akani & Capizzi, 1998) may justify a fixing of zeroes (Pianka, 1986). Pianka’s (1986) overlap formula was calculated for all communities, and the original species utilization matrices were randomized by shuffling the original values among the resource states. I used two randomization algorithms (RA2 and RA3) of Lawlor (1980), as they are particularly robust for niche overlap studies (Gotelli & Graves, 1996). RA2 tests for structure in the generalist–specialist nature of the resource utilization matrix by conserving guild structure, but destroying observed niche breadth (Gotelli & Graves, 1996). RA3 tests for guild structure by conserving niche breadth for each species, but destroying guild structure manifested by the zero structure of the resource utilization matrix (Gotelli & Graves, 1996). For each pair of species, 3 × 104 random Monte Carlo permutations were generated. This number of permutations is enough to avoid algorithm biases in calculations (Lehsten & Harmand, 2006). Niche overlap values were calculated for each of these randomly generated matrices, and species-pair and community-summary statistics were computed (Friggens & Brown, 2005). Actual overlap values were then compared to the distributions of the expected values. Structure was assumed when Pobs<exp = 0.05 or less in at least one niche dimensions, either with RA2 or RA3 (Gotelli & Graves, 1996). In all cases, equiprobable resource use was a priori assumed in the analyses, unless the studies reviewed explicitly reported resource availability data.
However, before presenting the results of my meta-analysis, it is necessary to note the potential problems associated with this type of analysis. First of all, it is well known that in too small matrices Poisson errors might obscure possible structures. I checked this with a highly structured simulated matrix with five species and two resources, and EcoSim was unable to find this structure at the 5% error level. Thus, it is possible that some such cases may have occurred in my analysis. The use of standard software and standard algorithms, despite their wide use in previous literature and their previous tests of robustness (Gotelli & Graves, 1996), may have some inherent problems associated to the different structure of the various datasets. Differences in data quality among studies may be potential sources of type II error rates; however, this is an aspect that is nearly impossible to take into account in these comparative studies, where the reviewer is forced to consider all studies as ‘nearly equivalent’ in terms of data quality. In any case, I used the same level of resource identification for all studies, and this should have much reduced the risk of biases coming from different data quality among studies. Last but not least, structure has a temporal dimension (Gotelli & Graves, 1996). Data obtained over longer time periods give averaged values and might obscure short turn dietary differences. Thus, it is clear that the differences in the duration of the various studies may have introduced unpredictable biases in the results. However, also in this case only a very few studies on African reptile community ecology have been conducted over long time (for instance, Barbault’s studies in Ivory Coast), thus it is likely that for most cases this has not been a problem with my synthesis.