The extent of savannah Africa
Global assessments of how much tropical moist forest remains are made routinely, and, in the case of the Brazilian Amazon, monthly. Comparable assessments of tropical dry woodlands and savannahs are few. Moreover, we show that broad-scale global land cover assessments massively underestimate the amount of small-scale land use conversion.
We estimate the original size of savannah Africa to be 13.5 million km2. In 1960, using the human population data sources described above, 11.9 million km2 had fewer than 25 people per km2. The comparable area shrank to 9.7 million km2 by 2000. Sub-Saharan Africa increased its human population by nearly four-fold from 1960 (229 million) to 2010 (863 million) according to CIESEN (2005). The same source expects the population to more than double by 2050 (1.753 billion). Simply, the extent of savannah Africa has surely shrunk considerably in the last 50 years and will likely shrink considerably in the next 40.
In contrast to estimates of moist forest cover, for example, that come with few direct data on the species those forests contain, there are extensive data on large mammals in savannahs. These allow us to estimate what fraction of the remaining savannahs is sufficiently intact to house lions, the ecosystem’s top predator. We estimate this area to be ~3.4 million km2 (Table S1)—only 25 % of the total savannah—highlighting the fact that many low human density savannah areas are nonetheless too small and isolated to support viable lion populations.
Of the roughly 13.5 million km2 of savannah Africa, IUCN classifies about 1.36 million km2 (~10 %) as protected areas, excluding those regions gazetted for timber extraction (IUCN and WDPA 2010). Roughly 1.08 million km2 of this area overlaps with the lion areas. (In other words, substantial areas have protected status, but have lost their lions.) Now, the IUCN categories of protected areas include several that allow extractive use—and that includes hunting. Lindsey et al. (2006) estimate the total area of sub-Saharan Africa devoted to hunting as at least 1.4 million km2, and of this, ~250,000 km2 is in Tanzania. What we cannot easily estimate is the various overlaps between areas with lions, hunting areas, and the various classes of IUCN protected land on a country-by-country basis. Some countries, such as Kenya, do not permit hunting.
To assess lions in Africa, a good map is essential
Total population estimates alone mean little in the absence of knowledge of where lions are. Our maps suggest that lion populations survive in some 67 areas, of which only 15 hold at least 500 lions. While a small fraction of these areas appear to be large and continuous on satellite imagery (e.g. the east of the Central African Republic, southeast Chad, and west South Sudan sub-populations and the Selous and Niassa populations), there are no surveys for several of those areas and their status is uncertain. The majority of sub-populations are small and isolated. Some are so isolated that quibbling over our criteria for land use change would make no difference: the animals that remain are a very long way from individuals that might rescue the population demographically or genetically.
A good map is necessary, but it is not sufficient
An obvious caveat is that areas for which we detect little conversion of savannahs to croplands may still suffer human impacts that make them unsuitable for lions. Over-hunting for trophies, poaching—of lions and of their prey species—and conflict with pastoralists may not have any visual signal to satellites. Even where there are low human population densities and areas designated as national parks, there need not be lions within them. The poor performance of even large protected areas in West Africa is striking. For example: Henschel et al. (2010) surveyed Comoé, West Africa’s largest park, a World Heritage Site that is roughly half the size of Kruger National Park in South Africa. Whilst Kruger holds nearly 1,700 lions (Ferreira and Funston 2010), and much of Comoé looks to be free of human disturbance from the high-resolution imagery that Google Earth provides, Henschel et al. (2010) found no lions, few native mammals, and extensive evidence of poaching and grazing by domestic livestock in Comoé. Size alone does not protect even the largest parks if they suffer poor management (Bauer et al. 2003).
Satellite imagery does pick up recently burned savannahs, sometimes covering hundreds of square kilometres. These could be natural or set by pastoralists to improve grazing. Analyses of the conservation consequences of anthropogenic fires are available for moist tropical forest against a backdrop of protected areas (see Adeney et al. 2009). We have not yet analysed available global data as a means to assess pastoralists’ impacts on the savannahs and how protected areas modify those impacts.
Conversely, we cannot exclude the possibility that lions might still be able to move through areas with land-use conversion, though much experience suggests that they suffer high mortality when they do. For example, Woodroffe (2000) estimated a mean human population density threshold at which lions went extinct of 26 people per km2. Many mechanisms might underpin this threshold, but land-use conversion is the most plausible. The match between her threshold and ours is striking.
Finally, even within suitable habitat, lion densities vary greatly (see Chardonnet 2002). Densities of prey also vary widely when considering the variation in rainfall and soil type across lion range (Coe et al. 1976; East 1984; van Orsdol et al. 1985; Hayward et al. 2007).
Lion population estimates
Our lion population estimate of 32,000 lions is higher than the population estimate by Bauer and Van Der Merwe (2004), but lower than the estimate by Chardonnet (2002).
The differences between these estimates do not represent an increase in lion populations in Africa, but are a direct result of different methods. There is evidence of strong declines and even extirpation of lions in some range countries. Especially in West and Central Africa, declines have been dramatic and conservation measures are urgent. While lions are protected in some of the lion areas, in many they are not, and in others they are hunted. While user-communities express the desire to manage lions sustainably, achieving that for any long-lived species is problematic. Several studies raise concerns about the impact of trophy hunting on lion densities and demographics (Yamazaki 1996; Loveridge et al. 2007; Davidson et al. 2011, Becker et al. 2012.). As noted above, the area devoted to lion hunting is large and Lindsey et al. (2006) emphasise the importance of hunting zones for protection of lions and their habitat.
How credible are the lion estimates?
Lions have low densities, large ranges and low visibility and are intrinsically difficult to count accurately. Few of the studies we report involve statistically justified surveys. The data we report are mostly “expert opinions”. They are controversial, yet we cannot simply pretend they do not exist. We now address their strengths and weaknesses.
The process that produced estimates of lion numbers involved people with widely different experiences and motivations. Some estimates were produced at meetings where they were hardly questioned, politely assuming equal expertise to keep the process going and reporting that they were “working figures.” The IUCN-sponsored workshops had delegates that were both biologists and politicians. However dedicated and well intentioned the participants, there is at least the potential for numbers to reflect wishful thinking or national policies that put a positive spin on numbers to ensure continued funding support. Countries across savannah Africa receive disproportionate funding for conservation from the World Bank, for example (Hickey and Pimm 2011).
Bauer and Van Der Merwe’s report (2004) went through peer-review and the IUCN reviews (IUCN 2006a, b) embraced broad-scale consultation with a wide variety of sources. These two quality control mechanisms were used to a lesser extent by sources producing national estimates from the sport hunting industry (Chardonnet 2002; Chardonnet et al. 2009; Mésochina et al. 2010a, b, c; Pellerin et al. 2009).
Globally, assessments of natural resources by user-communities are consistently more optimistic than independent estimates (Pimm 2001). Whether trophy hunters and the reports they fund also consistently inflate lion numbers to ensure continued business should be detached from any heated rhetoric and viewed simply as the legitimate scientific question that it is.
Table S1 shows that various studies by Mesochina et al. (2010a, b, c), Chardonnet (2002), Chardonnet et al. (2009) and Pellerin et al. (2009) constitute the majority of the putative lions (~55 %). These studies all had similar methodologies for estimating lion populations. They used estimates from scientific studies for areas that had them (e.g. call-in stations, individual identification etc.). For areas without, they used questionnaires and interviews to determine the frequency of lion presence within the past 5 years. They developed an equation to estimate density based on the closest, well-established density figure as the baseline and corrective factors to alter that density. Tailoring the equation for each specific area based on a variety of factors, density estimates and hence overall population numbers were generated for all areas with lion presence. We find this method scientifically debatable but we do see value in presenting the speculative results of this user-community along with the other data and provide an alternative estimate that includes them.
Certainly, these methods could overestimate both lion range and numbers. Since these reports affect over half of all lions, they greatly affect the global population estimate. This concern precipitated the generation of a global population estimate with and without the hunter-funded numbers (Table 1). With the user-community funded reports, the total number of lions increases by about 8 %.
For specific examples, IUCN (2006a) estimated 5,500 lions in the Selous, 4,500 in the Ruaha—Rungwa areas and 3,500 in the Serengeti and Mara. These total 13,500 lions. In contrast, Mesochina et al. (2010b) estimated these numbers at 7,644, 3,779 and 3,465, respectively, for a total of 14,888. These IUCN estimates are 8 % lower than those the user-community funded.
In sum, the numbers are broadly similar and, given the substantial uncertainties in lion counts, surely indistinguishable. Clearly, we need many other such independent comparisons if we are to draw more detailed conclusions. This applies a fortiori to Tanzania where the numbers are highest and where there are many uncertainties.
The 67 lion areas contain some populations that are large, stable, and well-protected—and so likely to persist in the foreseeable future. They also contain those that are so small, isolated, and threatened that only immediate, energetic conservation measures can offer any hope for their survival. And, of course, there are lion areas that are everywhere in between. How one groups areas across this continuum is inevitably arbitrary. Our approach is to use three classes: strongholds, potential strongholds, and the remainder. Broadly, these correspond to areas where management appears to be working (but we should always be vigilant), where immediate interventions might create a viable population, and where present management clearly is not working.
Our threshold of 500 (see “Methods” section) comes from Björklund (2003) who assessed the risk of inbreeding in lion populations due to habitat loss. He determined that, “…to sustain a large out-bred population of lions, a continuous population of at least 50 prides, but preferably 100 prides, with no limits to dispersal is required.” We took the average lion pride as containing approximately five adults (Bauer et al. 2008). Of course, the numbers of prides to avoid inbreeding is itself an arbitrary number, not a genuine threshold. (Simply, the fewer males who contribute genes to the next generation, the more inbred the population will be.) Moreover, the mean pride size is smaller in West and Central Africa, so the W-Arly-Pendjari population might also sensibly qualify as a stronghold. (We consider it a potential one.)
From the data derived in the lion population assessment, as well as the World Database on Protected Areas (IUCN and WDPA 2010), we considered only those lions found within existing protected areas including those with IUCN categorization that allow hunting, to count towards the minimum viable population. The Tarangire lion area of Tanzania, has an estimated 700+ lions, but only ~200 in protected areas with IUCN categories I–VI. The rest are found in non-designated hunting areas that do not qualify towards stronghold status.
Finally, only lion areas that are contained within LCUs having stable or increasing lion population trends as per the IUCN (2006a, b) are lion strongholds. The single exception to this rule is the Tsavo/Mkomazi lion area (Maasai Steppe LCU), which IUCN cites as having decreasing numbers. However, while lion numbers are declining outside of protected areas, we believe that lions within the parks are usually well protected and in sufficient numbers to meet the criteria. This criterion also has its uncertainties, for in some parks—Kafue National Park in Zambia, for example—poaching of lion prey may be a cause of concern for the lion’s long-term persistence. IUCN’s statement that the populations here are “stable” may be optimistic. Similarly, intense hunting outside protected areas can also affect those populations within the reserves (Woodroffe and Ginsberg 1998). These caveats accepted, the broad conclusions of our Table S1 remains: approximately 24,000 lions are in strongholds, about 4,000 in potential ones, but over 6,000 lions are in populations that have a very high risk of local extinction.
This is not the place to review management options for lions, the forces that threaten them, or savannahs in general. We restrict our comments to issues that arise from the mapping and assessments we have presented.
Lion numbers have declined precipitously in the last century. Given that many now live in small, isolated populations, this trend will continue. The situation in West Africa is particularly dire, with no large population remaining and lions now absent from many of the region’s national parks. Central Africa is different in that it has a very large contiguous lion area centred in the Central Africa Republic. In view of reported declines, it still does not qualify as a stronghold. Populations in these regions are genetically distinct (Antunes et al. 2008; Bertola et al. 2011). Keeping these populations from extinction will require conservation efforts well beyond the dismal performance of the region’s other protected areas. (The World Conservation Congress, 2012, issued a formal resolution Res 5.022, specifically supporting mammal conservation initiatives in these regions, http://www.iucn.org/about/work/programmes/global_policy/gpu_resources/gpu_res_recs/)
Hunting areas are extensive, so the fate of lions depends on how well user-communities manage them. The same principle applies to lions within protected areas, with responsibility falling on protected area managers to secure these populations. Finally, lions also occur well beyond protected areas, and how well one manages lion-human conflict will determine persistence there. Yet, conflict outside protected areas can affect lion persistence within (Woodroffe and Ginsberg 1998). Good protection within a protected area is not sufficient if there is unrelenting killing of lions outside it.
Central Africa may have sizable lion and prey populations, but they are poorly known, even by African standards.
That said, independently verified census data, using statistically repeatable techniques are the rare exception, not the rule, across even relatively well-studied East and Southern Africa. The situation is particularly acute for Tanzania, which holds a large fraction of the world’s lions.
Repeated mapping of areas which have at least the potential for lions because of their low human impacts may provide the only quantifiable measures of how savannah Africa is shrinking from the lion’s viewpoint. This is necessary, but definitely not sufficient. The lack of repeated, statistically credible lion counts, for well-defined areas is a striking omission, one that must be rectified if we are to assess not only the trends in lion numbers, but our success in reversing their declines.