How to protect lions?

Lions: just big kitties really!

There have been a couple of lion stories in the news in the last week or two, and enough interest in them that I felt compelled to write something. First there was a paper by Craig Packer and many coauthors about lion populations in Africa, their current declines, and the possible role of fencing in protecting them. Then, shortly after, there was a letter in the New York Times by Tanzania's own Director of Wildlife, asking the US government not to list the lion as endangered, as lion hunting is crucial to their conservation in Tanzania.

Tarangire Lioness - a surprisingly resilient population.

So, I thought it would be useful to bring the two issues together and look at some of the facts behind them. So, let's start by looking at Craig's paper on lion populations. As several others have already noted, this is an important paper - it has collected as many medium and long term population data as could be found, from lion populations all across Africa and received a lot of attention (click the links before and here!). in some places like Serengeti (where Craig, his coworkers and his predecessors have been busy recording lions annual for a very long time) the data are good. In other places, like Waza NP in Cameroon, the data are much patchier, but by collecting everything in one place we can see for the first time quite how much the population has been suffering in recent years. And there's a clear picture painted of large-scale, continent-wide declines. This isn't really anything new (though sad to see it so plainly), and will come as no surprise to anyone who knows anything about African conservation.

Lions are still outside protected areas in much of Tanzania!

Where the story gets more interesting is in the analysis of the population trends. As well as population trends, the authors have estimated how many lions there should be in a protected area, based on the number of herbivores and (where that's not available) the rainfall and nutrient availability. They can therefore compare observed lion densities with expected densities and identify populations that are below capacity, even if the trends aren't changing much. And alongside each of the population trends the authors have gathered a set of additional information for each protected area: human density in the surrounding area, an index of official corruption, operational budget, whether or not there is commercial hunting, area protected and other details of the protected area management, including whether or not the protected area is fenced. Using these pieces of information it is fairly straightforward to identify correlations between population trend and conservation actions. This analysis identified a number of correlations: both the degree to which lion densities matched potential lion density and the trend in lion population are positively correlated with whether the protected area is fenced, whilst the degree to which actual and potential lion populations matched was also correlated with the management and the trends correlated (negatively) with whether the park was run by the state (bad) or privately (good).

Pale blue line is total lion density, dark blue is adult
density, red is estimated maximum density

This is potentially an important result. Regular readers will probably have noticed that I'm not generally a fan of fences, so data suggesting they're of major benefit needs some careful examination. I actually had a look at these data a while ago, as Craig was kind enough to send me an earlier draft of the paper, which enabled me to reach my own conclusions, of which I'll share a few here. Those who know me won't be surprised to know that the first thing I turned to were the data themselves. In the Supplementary Material there's a wealth of information (which is fantastic to see), including plots of the population densities in individual reserves. I've cut out the Serengeti information here so you can see (one of the best) examples. When I look at this, I can see that lion density as increased over time - some of that is due to the return of the woodlands, and the increase in herbivore populations - and that it appears to be pretty much at the theoretical maximum at the moment, but we can see fluctuations that relate to drought and disease. So it seems slightly surprising to me that the important table that summarises these results (Table S1) suggests populations are only at 85% of capacity and that the trend is a slight negative one. (The reason for this, of course, is that they only use the last 10 year of data, and the fact that the log scale is important is ignored when calculating the percentage) but what was obvious to me from this figure and, particularly the other, much more patchy datasets, is that any estimate of trend and/or population is only a rough estimate, with lots of annual variation and measurement error, but the variation in estimate is not considered in this analysis - it is (incorrectly and unnecessarily) assumed we have perfect knowledge of both population and trend, something I've often criticised. If this uncertainty is appropriately considered, I'm quite convinced there will be absolutely no significant correlation of any of the measured variables with population trend (though the correlation with population size might remain - but I'm equally sceptical here, because if there really are lion populations nearly 300% higher than the theoretical maximum density (as they appear to be from Table S1), then I'm sure the theoretical maximum density is incorrectly calculated, no matter how strong published correlations may be. This is a particular concern since fenced sites tend to have lower predicted lion densities than unfenced sites).

This statistical issue aside (and the related one that assuming exponential growth when populations are manifestly not on the exponential part of a logistic growth curve is strange), there are other, more profound issues here too. For me, the crux of the matter is that these are correlations, and correlation alone can never identify causation. The two strongest correlations, between the degree to which populations are at capacity and fences and management budgets (illustrated left - black dots are fenced reserves, red unfenced). Note that the black dots really are higher than the red ones, but note that (with the exception of only seven points) they're also the highest budget sites. They're also the smallest sites, tend to have lower estimated maximum lion densities (they're mostly in drier and less productive regions) and tend to be in the same geographical region. Even using modern statistical methods it's very difficult to untangle correlations when things are so profoundly tied together. But think, too, about how these correlations might (if they're real) actually be caused. The authors suggest that fenced populations should suffer less from poaching due to 'edge effects': fewer people within the boundaries of the protected area, and fewer animals leaving the safely protected zone. But some of the same authors have already noted that in a number of populations included in this study these edge effects aren't actually seen. It's also hard to explain why a protected area effect hasn't be identified too, if edge effects are important: large sites have relatively low edge to core ratios. Hand how does spending more really save lions? If it's all spent on anti-poaching then maybe so, but that's not the case (most spending in the higher budget parks like Kruger are traditionally not spent on anti-poaching - though there's a lot of extras on rhino at the moment - but on road development and maintenance, staff salaries and general management cost, etc., etc.). So it is again very hard to see how, once a minimum spend on anti-poaching is achieved, additional funding has any additional impact (though I agree completely that many parks are chronically underfunded and as we can see from the figure, they're also unfenced). Both problems suggest to me that these correlations are unlikely to really be causal, which means that spending more or fencing more is not necessarily going to improve matters.

Finally, (well, I have to stop somewhere - there's lots more I could question, like the wisdom using the price of a North American to estimate the cost of fencing in elephants in remote places where people use chain-link fences to make snares!) I just want to put a slight different spin on the whole paper. Instead of a message of disaster, which has been quite widespread, let's look carefully at the results we see here. The bad news: "Nearly half the unfenced lion populations may decline to near extinction over the next 20–40 years". However, let's look at which populations those are - it seems (a) they're the populations already nearly extinct (Mole in Ghana) or massively depleted (Katavi in TZ) and (b) they're numerically small, relative to those populations that are not predicted to go extinct (Serengeti, Selous, etc). I've not actually done the sums (because I'm on the wrong computer), but I'd hazard a guess that we're looking at no more than 10% of the total population - hardly the doom and gloom story I've seen this study cast as elsewhere, rather it seems like we can be confident there will still be viable lion population (maybe in reduced number, and maybe with some local extinctions) in 20-40 years time. Whilst I'd like to see a rosier picture for lions too - and certainly we can do better - this isn't the apocalyptic paper I thought it must be.

Do I want to shoot one? No. But does shooting them help save them?

So, do we need fences everywhere? I'm not yet convinced - all the negative consequences I've talked about before remain, I fundamentally disagree with Craig that simply because fenced areas can support lions they maintain an intact and fully functional ecosystem and the evidence presented here in favour of them is very weak. I'm not willing to risk what we know to be a sub-optimal strategy on the basis of this paper alone.

And as for the Director of Wildlife, do I think lion hunting is required to protect them? I'm not convinced of that either - certainly the currently poor regulation is causing more problems than it is solving. And certainly $75M over 4 years is tiny compared to the $1.6Billion estimated for 'non-consumptive' tourism revenue. But hunting does protect lots of habitat that it would be extremely difficult to get tourists to visit, and $75M is still not to be sniffed at. But the argument is subtler: we're not suggesting stopping hunting completely, just stopping hunting lions. What would that do to the hunting industry in Tanzania? I can only guess, as I doubt the data exist. I do know that the 60% of the hunting market identified by the Director of Wildlife is probably an overestimate - maybe 60% of tourist hunters are American, I'm very doubtful 60% of all tourist parties hunt lions, though maybe someone can tell me. Even if they did, I can't see them hanging up their rifles simply because lions are no longer on the menu: that didn't happen when Rhino came off the trophy list (even though you can still hunt them elsewhere in Africa).  And whilst I know that many areas currently hunted are never going to be major tourism sites, the example of Grumeti, where a game reserve has turned into a stunning tourism operation, demonstrated that at least some of the prime areas can effectively, and probably with increased revenue, be converted to other tourism types. Of course, I do agree with the DW that a sustainable harvest of lions is possible in Tanzania (even though I don't agree with his rosy picture!), it's just a shame it's not happening now.

Main references:

Packer, C., Loveridge, A., Canney, S., Caro, T., Garnett, S., Pfeifer, M., Zander, K., Swanson, A., MacNulty, D., Balme, G., Bauer, H., Begg, C., Begg, K., Bhalla, S., Bissett, C., Bodasing, T., Brink, H., Burger, A., Burton, A., Clegg, B., Dell, S., Delsink, A., Dickerson, T., Dloniak, S., Druce, D., Frank, L., Funston, P., Gichohi, N., Groom, R., Hanekom, C., Heath, B., Hunter, L., DeIongh, H., Joubert, C., Kasiki, S., Kissui, B., Knocker, W., Leathem, B., Lindsey, P., Maclennan, S., McNutt, J., Miller, S., Naylor, S., Nel, P., Ng'weno, C., Nicholls, K., Ogutu, J., Okot-Omoya, E., Patterson, B., Plumptre, A., Salerno, J., Skinner, K., Slotow, R., Sogbohossou, E., Stratford, K., Winterbach, C., Winterbach, H., & Polasky, S. (2013). Conserving large carnivores: dollars and fence Ecology Letters DOI: 10.1111/ele.12091

PACKER, C., BRINK, H., KISSUI, B., MALITI, H., KUSHNIR, H., & CARO, T. (2011). Effects of Trophy Hunting on Lion and Leopard Populations in Tanzania Conservation Biology, 25 (1), 142-153 DOI: 10.1111/j.1523-1739.2010.01576.x