The Fire in the Wood - cause and effect in species extinctions
Nature abhors simplicity. The one thing most eagerly sought when analysing past events—a clear sequence of causes and effects—is often the one thing denied us. In this way, investigating ancient extinctions is not unlike the process of a crime scene. We may have the scene, the suspects, the end result, yet the smoking gun eludes us. Whatever our data, interpretation and reflection is necessary. Of course, this comparison can only be drawn so far: Quite crucially, the ancient world offers no witnesses. They are unreachable, long dead, and, even were that not the case, key suspects themselves. No, in the case of the missing mammoth and the death of the elephant bird, we are alone.
Beyond all things, our issue is one of data and analysis. We know the species involved—many of them, at any rate—the geography of the extinctions, the general ecological contexts and, very roughly, the dating. The task of the palaeontological investigator is to draw from this jumble of data a cohesive narrative. Many such narratives have been proposed over the years—extinction by climatic swings, disease, asteroid impacts, human hunting. The exploration and clarification of these hypotheses is a large part of this website’s purpose. Suffice to say, we will not resolve them here. Instead, my purpose is rather more specific: Not to discuss the evidence, nor where the bulk of it lies, but rather, the way we interpret it. It is my goal in this article to pursue a single question: What does it mean to say something was the cause of an extinction? Before we begin this exploration in earnest, allow me to paint a simple scenario: A forest, dry and fragrant, perched upon the western slopes of the Sierra Nevada. It is summer, and the weather is blistering hot. Suddenly, a wildfire emerges. From an initial point between two rocky outcrops, it quickly spreads, eventually consuming hundreds of kilometres of woodland before being put out. When investigations are performed, the origins of the fire are traced to a single spot: a campfire, down between the two tall rocks, whose owners had failed to watch it closely. What was the cause of this wildfire? The answer may seem obvious, but do not be too quick. Aridity, heat, wood and fire. All were necessary for the disaster to unfold. And yet, though the causes, plural, were numerous, they evidently are not of equal order. It is this dilemma that we will be exploring.
Figure 1. A raging forest-fire.
Terms of use: This image is licensed under the CC0 1.0 Universal. It is uncredited. The image is unedited and the original can be found here
For James Woodward, a causal relationship between two properties exists when:
“for at least some individuals there is a possible manipulation of some value of X that they possess which, given other appropriate conditions (perhaps including manipulations that fix other variables distinct from X at certain values), will change the value of Y or the probability distribution of Y for those individuals”
Two terms introduced by Woodward are critical for this discussion of extinction-causality: Intervention and invariance. An intervention, put most simply, is a manipulation of the value of property X which is totally independent of property Y. In the context of human impact upon a given species, where humans are the property X and the target species Y, the movement of humans into a new area counts as an intervention upon the property “humans” only if whatever factors led to this migration did not also directly impact the population of the species Y. The reason for this should be fairly plain: If we want to determine whether human entry was responsible for the decline of a species, but if the self-same factor which led to human arrival (say, climatic changes) also independently affected the populations of the target species, this obviously massively complicates the picture.
The invariance of a causal relationship is somewhat different: It is, as Water, C. K. (2007) puts it, the range of values under which a causal relationship holds. Simply put, two properties may influence each other, but only under certain conditions. The example used by the aforementioned author is one of temperature and boiling water—the raising of water’s temperature from 22° to 100° (an intervention) will cause the water to boil, but only under standard atmospheric pressure. Should the pressure be changed, the causal relationship would cease. Invariance describes the range of conditions under which such a relationship holds.
Let us consider a more ecologically relevant example: Two species of non-native mammals are introduced to a pacific island—let us merely term them species A and species B. On this island was found a host of endemic birds, many of them flightless. Now, the ultimate end-result of these introductions was the extinction of 95% of the native birds, including all flightless forms. The question we wish to answer is this: “Which of the two introduced mammals, species A or B, caused these extinctions?”. For simplicity’s sake, we will assume that either both species were fully responsible, or else all of the extinctions could be ascribed to only one of the two species. That is, it is not the case that 20% were caused by species A and 80% by B.
To answer this question, we look at the chronology of events. As it turns out, species A was introduced 100 years ago, while species B arrived 30 years later. In the decades following its introduction, predation by species A severely reduced native bird populations, in many cases greatly restricting their ranges. None of them disappeared completely, however. When, 30 years later, species B was introduced, the results were swift. Within an additional 20 years, all of the aforementioned extinctions had taken place. Now, to disentangle the causal relations here, we must introduce two more pieces of information. Both species A and B have also been introduced to several other Pacific islands; on none of these, however, do they both co-exist. Their exact impacts have varied, yet particularly on those islands nearest to the one studied, in no cases have either species produced extinctions close to the 95%. What, from all of this, can we deduce?
Figure 2. An overview of the hypothetical islands. Diagram by Tristan Søbye Rapp
If this example seems complicated, I do apologise, yet it is substantially less so than any natural situation. Many relevant factors I have left out entirely. Nevertheless, we can draw certain conclusions. Clearest among them is that, in this case, the magnitude of extinctions on the studied island was a result of the co-occurrence of both invasive mammals. Here, then, we see the degree of invariance in the causal relationship—in fact, it is quite limited. In studying the relationship between the introduction of species B and the populations of the native birds, only with the presence also of species A does the predicted impact (near-total extinction) occur. Despite this, both species did have severe impacts on the populations of native birds, to the degree that neither A nor B could be called harmless.
However, whilst both A and B are therefore undeniably causes of the population declines, and by extension, the final extinctions (extinctions being merely a population-decline carried to its logical end), the impacts of their introductions cannot be said to be identical. It was the intervention of introducing species B that actually triggered the extinctions. Comparison with surrounding islands points to a counter-factual where, without the introduction of species B, most if not all of said losses do not in fact occur. Granted, a comparable counter-factual where species A is never introduced likewise sees less extinctions, but in the actual event, it was the introduction of species B which constituted the ultimate causal trigger, even though the presence of species A was a necessary precondition for this event to take place.
In such a context, species B may be called the agent of extinction, in the same sense that a match could be considered the agent of ignition, even though factors such as the material of the matchbox and sufficient dryness of the air are obvious co-acting causes. Of the match and the matchbox, it is clear that, although both must necessarily be present for ignition to take place, it is the striking of the match which is the ultimate cause.
At length, we can turn our focus back to the megafaunal extinctions, and the degree of human agency therein. Two major points may be derived from all that we have hitherto covered. The first is that the existence of a confluence of causes affecting a given population does not preclude the existence of an actual cause explaining said population’s ultimate collapse. Moving from our idealised island example to the actual ecosystems of our world, complexity increases massively. Looking specifically at species of megafauna, numerous factors may affect their populations, from inter- and intraspecific competition to predation pressure and disease. Significant among these factors are the broader environmental conditions—weather, soil-fertility, water-availability and more. These are dependent (not exclusively, but in large part) on the climate, and as such, climatic impacts on the populations of animals are inevitable.
It may be no great surprise, then, that many researchers, observing the synchronicity of climatic swings with megafaunal declines, have proposed these as the ultimate causes of extinction. Examples of such reasoning include Nogués-Bravo et al. (2008) on modelling the ranges of mammoths (Mammuthus primigenius), de Oliveira et al. (2020) on the litoptern Macrauchenia patachonica, and many others. Just as common if not more are papers which, establishing causal links between both human arrival and climatic shifts, conclude that both factors in tandem caused the extinctions. The issue with such conclusions should be quite apparent by now: Merely establishing that multiple factors were at play in the megafaunal extinctions is not sufficient—indeed, it is inevitable. To conclude from this that the ultimate cause was synergistic is equivalent to noting that both arid weather, dense vegetation and an unchecked campfire were responsible for the raging wildfire. Such a conclusion is not wrong, per say, yet it seems fair to note that this “mixed aridity-and-campfire”-explanation would be viewed as pedantic at best. The issue here is one of counterfactuals—had the wood been dry, but there been no campfire, the fire would not have happened. On the contrary, it is not unthinkable, given the dense vegetation and other compounding factors (say, the presence of particularly flammable plants) that a fire could have still occurred, even had the weather been wetter. By the same token, the mere observation that both climate and humans impacted megafaunal populations tells us little. There were massive climatic swings at the onset of, before and throughout the last glacial period. These caused no comparable extinctions.
The complexity here may be seen by returning to our burning forest one last time. Let us paint two superficially very similar, yet crucially different scenarios. 1) a campfire is made and managed poorly. Subsequently, a fire breaks out in the surrounding vegetation and begins to spread quickly. At this point, a wind picks up, carrying the flames further into the wood, exacerbating the fire. 2) A campfire is made and managed appropriately. The fire does not spread. Then, suddenly, a fierce wind picks up, carrying the flames into the surrounding shrubbery, which subsequently catches aflame, beginning a wildfire. Viewed from the outside, say, from a town below the forest, these two scenarios would appear almost identical. In both cases, a wind picks up, a forest-fire follows, and a campfire is discovered at the point of origin. The factors are the same, the end-result the same. Indeed, without precise knowledge of the chronology of events, it would likely be impossible to tell the scenarios apart. Yet the difference is this: In the first scenario, the wildfire preceded the wind. It was already spreading and would by all indications have gotten just as bad eventually regardless. The wind merely exacerbated a disaster already unfolding, hastening the spread of the flames. In the second scenario, there would have been no wildfire if not for the wind. In the first example, the poorly managed campfire is the actual cause of the wildfire. In the second example, the actual cause is the wind.
How can this be applied to the megafaunal extinctions? Like so: When viewing the prehistoric world, we are the townsfolk, gazing at the wildfire from afar. We know the factors—climate, soil, competition, human hunting—and the end result—extinction. What we do not know, or rather, what is far more difficult to discern, is the actual order of events. Did human hunting merely present the backdrop upon which climatic shifts were the ultimate causes of extinction (the second scenario), or were humans always going to wear down the populations of megafauna in the end, with the climate playing at most an exacerbating role? (the first scenario). The difference may seem slight, yet it is not. The question at hand is this: Removing either humans or climatic changes, do the extinctions still happen? Which intervention, the shift in climate or the introduction of humans, was the ultimate agent of extinction?
It must be noted that there is more, much more, that could be said on this subject. The examples given here do not cover all facets of the topic, and in their over-simplicity, some readers may find them lacking. Yet my purpose was not to resolve all questions—that would be the work of several articles, if not a full thesis. Rather, I hope merely to provoke conversation. What has here been discussed is philosophy of science, and as such, not beyond the reasonable purview of scientists. It is important, of course, that we are attentive and empirical in our treatment of data, yet we must also be honest. There are times when the evidence falls short, and mere extrapolation is not enough. There are times that call for interpretation, and yes, even philosophy. We would not claim a bout of dry air ignited a forest fire; let us be certain we do not make the same mistake, when looking at disasters older and far greater.