Arguing from empirical evidence about the fantasy of overgrazing

A guanaco in Chile. In Chile we do not have commercial ranching, and thus researchers are not concerned by questions of "overgrazing." Photo (c) MRB.

What overgrazing is and who is responsible for it is one of those ecological topics that tends to incorporate a lot of economic assumptions or interests. Overgrazing is, as the name implies, when grazing occurs "too much" and results in a reduction, across years, in total plant biomass. Either the plants are not allowed to flourish because they re being eaten too

herbivory selects for an increase in less-productive, less nutrient-rich plants. Under "natural" circumstances one might expect the herbivores to leave, or to reproduce less when faced with a reduction in available forage. Overgrazing, however, is a phenomenon that seems to only happen when livestock are involved. Livestock generally cannot make their own choices about where they graze or how much they reproduce. Humans want livestock populations to remain stable or, even better, to show increasing productivity. The idea of carrying capacity proposes a concept of optimisation in which productivity of the herbivore (increase in herbivore biomass) is as high as possible given the available vegetation productivity capacity (potential vegetation biomass). Overgrazing is when this optimal level of herbivore-plant biomass conversion ratio is overshot and becomes suboptimal. In places like Argentinian Patagonia, where sheep are raised on a ranching model, overgrazing is not blamed on sheep, but on native wild guanacos (Lama guanacoe). Many ranchers and researchers claim that it is because wild guanacos are not being kept in check by pumas (which are found in low abundance) or human hunting that the carrying capacity has been surpassed and overgrazing is depressing sheep productivity.

In 2019 when I was a mentee associate editor at the British Ecological Society journal Journal of Applied Ecology, I was assigned to a paper that addressed this issue using remote sensing to estimate vegetation productivity. It also used common assumptions from rangeland management analyses to estimate how many animals could be supported at carrying capacity. In this model, the authors assumed that guanacos and sheep are equivalent in what they eat, where they eat, and how they convert it into biomass: they base their calculations on ecologically and biologically substitutable "animal units". They argued that the introduction of sheep and decimation of guanaco populations through competitive exclusion and hunting first led to overgrazing (by sheep), after which the sheep population was gradually reduced to an equilibrium carrying capacity, but then a recent increase in guanaco populations has pushed the landscape into overgrazing again. According to them, the current state of overgrazing is thus not the fault of sheep, and does not call for reductions in sheep, but rather calls for lethal control of guanacos.

As the associate editor I was worried about the assumptions in the remote sensing and modelling work because I am a field ecologist, and I find that remote sensing and modelling can sometimes miss important empirical factors that make a difference to reality. Since the guanaco overgrazing issue is controversial and could directly impact policy, it seemed quite important to pin down this issue. I wanted to find a reviewer with experience in this kind of methodology who could also critically examine the assumptions about "animal units" and carrying capacity vis à vis published empirical results I knew about that said that guanacos and sheep can coexist at large spatial scales precisely because they eat different things in different places with different efficiencies. I was not sure who this reviewer would be, because I am precisely not an expert in rangeland management calculations myself, and because it is difficult to assess whether a rangeland management scientist whose work I don't know has a critical or thoughtful stance or not. Many experts whom I hoped would be able to give this kind of review, and either confirm my suspicion that there was a problem, or assure me there was none, refused to review the paper. The reviewers I ended up with were not interested in the methodology at all, ignored its assumptions, and did not bring up the way it contradicted empirical research.

As the associate editor, before acceptance, I weighed in and asked the authors to incorporate into the Discussion the potentially contradictory published empirical evidence for non-equivalence of guanacos and sheep and thus their potential to coexist. This resulted in a paper that takes one view of this issue in the model and another in the Discussion, in a way that the authors didn't really attempt to make coherent. (I assume that they just wanted to publish the paper, and were willing to say things they didn't entirely believe in the Discussion if that was what it took to get acceptance. I guess we have all done that.) I was also asked by the journal to write a blog about the paper when it was published, in which I glossed over my doubts about the "animal units" used to calculate carrying capacity, and basically just talked about the issues of empirical considerations about guanaco-sheep coexistence that I made the authors include in the Discussion: https://appliedecologistsblog.com/2019/06/13/guanacos-can-coexist-with-commercial-livestock-in-patagonia/

I accepted the paper even though I didn’t like it's methods and approach because at least two people with some level of familiarity with that kind of approach did like it. It seemed to me that, as an editor, my job was not to establish a normative quashing of things I dislike or am suspicious about, but to facilitate an academic forum.

Happily for the state of ecological exchange of ideas, and entirely independently of me, a response to that paper was published the following year:

https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13536

The response, in my view, is an excellent example of reasoning with empirical evidence in ecology. It begins by arguing that guanacos and sheep are not equivalent grazers who each exclusively dominate the entire amount of available plant productivity in any particular area. They argue that on-the-ground studies show that sheep are able to dominate the most productive, nutrient-rich grasslands, while guanacos are displaced to all the "marginal" (unproductive) areas that sheep avoid (this was the literature I wanted the original authors to comment on in the Discussion). They argue that the majority of guanacos, estimated at over 70% in one region, occupy lands that have been abandoned for sheep farming. The remaining percent of existing guanacos who must coexist more closely with sheep, do not use the same bits of landscape or eat the same things as sheep, as they document with citations of empirical studies in the field. Guanacos, unlike sheep, can live without drinking water (they obtain water from the vegetation they eat) and have such efficient physiological processes that they can live on low-nutrient plants: these things have been known since the 1970s at least. Thus, guanacos and sheep coexist by using different parts of the vegetated landscape, at both larger and smaller scales. If guanaco populations have rebounded in recent decades, the commentary argues, this is because sheep ranching abandonment as increased, leaving degraded lands fully available for guanacos.

The commentary further argues that contrary to the conclusion that guanacos are driving overgrazing through their uncontrolled population growth, guanacos are regulated bottom-up (that is, by plant availability and not by predation), and further reduce the probability of population surges and crashes through territory formation and defence. The original paper provides no direct evidence of the rate of guanaco population growth, and the evidence of overgrazing is at a remote-sensing scale, thus a snapshot in time across an entire region. In other words, there is less vegetation productivity than expected, but no one can actually see who is eating it. Given guanaco life history, they simply don't seem like the species more likely to be overeating and overreproducing and then overeating again (sheep by contrast are not territorial). And whatever the guanaco population is doing, this could certainly not be blamed on the lack of predation or hunting, which is not expected to control guanaco population rate. The commentary ends by noting a lack of empirical evidence for the idea that guanacos cause grassland degradation.

In fact, the issue of the ecology of degradation itself in the Patagonian grasslands-- for example the plant dynamics, the soil responses, the role of rainfall regimes, underlying the remotely-observed decline in productivity-- is the big ghost here, which is simply labelled "overgrazing", thus implying that its causes are entirely understood, and not examined in any further detail. The spatial distribution of degradation across grasslands and more marginal habitat types is also entirely ignored, but seems critical to me. If, for example, sheep ranching abandonment was caused by, or led to, shrub increases in some areas, this alone could account for a reduction in measures of plant productivity since shrubs are perennial and woody (while being entirely compatible with guanaco populations). I don't know if that is happening, but similar dynamics could account for elements of degradation and should be better studied.

I think that this commentary is an excellent piece of ecological reasoning, in which a broad set of empirical observations are drawn on to argue about what is possible, what is likely, what kinds of patterns can be produced by certain observed mechanisms, and what effects certain proposed mechanisms would have, given the observed distribution and patterning of species and processes in the place in question. In my view a good ecological argument takes into account both the inter-species variation, flexibilities, and particularities of species and landscapes, and the generalities of mechanisms and processes. With the latter, we can project patterns into the future or past or across scales, but these hypotheses or interpretations will only be as good as our consideration of biological diversity is.

The authors of the first paper have also published a reply to the commentary critiquing it. Briefly, they argue that

• if competitive exclusion allowing coexistence were the norm between guanacos and sheep, ranchers would correctly perceive this and the entire subject of overgrazing would not be an issue

• the animal units assumptions are acceptable because modelling the full complexity of the situation is impossible

• because guanaco populations have sometimes crashed, they must also be able to grow exponentially (and they can fit some data from something that happened once somewhere to a model proving this must be so)

• all of their estimates and models have "hard data" inputs unlike the speculation about "possible" mechanisms drawn on in the commentary

In my interpretation, these arguments are not very good ones. To treat the non-scientific perceptions of ranchers with vested economic interests as having the same level of objectivity and certainty as what they call "hard data" (and models), while treating the empirical observations of field ecologists as mere possibilities, is at best inconsistent. The authors also frequently argue that whenever estimated population numbers surpass modelled population thresholds, real degradation and overgrazing is thus taking place and requires real-life interventions. Their faith in models, estimations, and large-scale averages, and their lack of interest in the complexities of real situations, is in my mind not how theory and empirical observation should be best integrated in ecology. Ecology, to me, is all about the interplay between generalities and specificities: without the specificities it is just fantasy. But I am grateful to the Journal for Applied Ecology for publishing this exchange of approaches and arguments. I wish that this were much more common in ecology today.