De-extinction as Artificial Species Selection

D. Turner (2016), “De-extinction as Artificial Species Selection,” Philosophy and Technology (online 18 September, 2016). doi:10.1007/s13347-016-0232-4. De-extinction as Artificial Species Selection Derek D. Turner Connecticut College 270 Mohegan Avenue New London, CT 06320, USA [email protected] Abstract This paper offers a paleobiological perspective on the debate concerning the possible use of biotechnology to bring back extinct species. One lesson from paleobiology is that extinction selectivity matters in addition to extinction rates and extinction magnitude. Combining some of Darwin’s insights about artificial selection with the theory of species selection that paleobiologists developed in the1970s and 1980s provides useful context for thinking about de-extinction. Using recent work on the prioritization of candidate species for de-extinction as a test case, the paper argues that de-extinction would be a form of artificial species selection in which humans influence which species persist vs. go extinct. This points to a serious gap in our ethical theory: Much work has been done to clarify the value(s) of biological diversity, but we also need theoretical guidance for decisions that amount to species sorting, and that will shape the macroevolutionary future. Keywords Artificial Selection, Conservation paleobiology, De-extinction, Extinction selectivity, Macroevolution, Resurrection biology, Species selection 1. Introduction This paper offers a paleontological perspective on de-extinction. I argue that the notion of artificial species selection can help provide important context for understanding what de-extinction might mean. Artificial species selection combines one familiar with one possibly less familiar idea: Darwin’s notion of artificial 1 selection, plus the theory of species selection that paleontologists developed in the 1970s and 1980s.1 The suggestion is that human beings are now playing a role in macroevolution that closely parallels the role that Darwin’s animal breeders were playing in microevolution. De-extinction technology would make this human role in macroevolution very explicit. Here “microevolution” refers to changes such as trends in gene or trait frequencies that occur within biological populations. “Macroevolution” refers to evolutionary changes above the species level. For example, species loss is a trend that would count as a macroevolutionary change, whereas the evolution of lactose tolerance in human populations is an example of microevolutionary change. My claim is that de-extinction, the use of biotechnology to revive extinct species, would be a form of artificial species selection. Section 2 sets the stage for the argument by introducing and diagnosing an ongoing public debate about extinction, a debate that pits optimists like Stewart Brand against the prophets of a sixth mass extinction. Section 3 introduces some macroevolutionary theory and sketches some of the basic issues at stake in the debate about species selection. Section 4 introduces some of Darwin’s thinking about artificial selection and argues that species selection, too, could be artificial. Finally, section 5 applies these ideas to de-extinction. There I develop the argument by focusing on one recent attempt to develop criteria for assessing candidate species for de-extinction. 1 Sepkoski (2012) provides useful context for understanding the emergence of the theory of species selection during what might be called a revolutionary episode in paleontology. 2 Before proceeding, however, I want to stress that I am not going to say much about the ethics of de-extinction, or the question whether we should try to bring back extinct species at all.2 I’ll just say here that I think the ethical issues are quite complex. Since de-extinction would be a special case of restoration ecology (as argued by Turner 2014), much depends on what we think more generally about how to assess proposed restoration and reintroduction projects. It also matters a great deal what story we tell about the value of biological diversity, which is itself a complicated issue and the subject of a vast literature in environmental ethics and the philosophy of conservation biology.3 Nor am I going to say much in this paper about the scientific and technological details, or about how de-extinction would actually work.4 I also want to set aside philosophical questions about species membership. Would a mammoth-like animal created using biotechnology really be a woolly mammoth?5 There is no way to tackle that question responsibly without sorting through the very messy discussion that biologists and philosophers have been having about the nature of biological species. Still another interesting conceptual question that I will not address here is whether extinction is irreversible by definition (see Delord 2007, 2014). If we really could re-recreate a genuine woolly mammoth, should we say that the mammoths went extinct for a while and then rose from oblivion? Or should we say that they never really went extinct, and that they would only really be extinct 2 For introduction to the ethical issues, see Sherkow and Greely (2013), Cohen (2014), or Sandler (2014), as well as the papers collected in Oksanen and Siipi (2014). 3 For a start, see Norton (1987), Sarkar (2005), and MacLaurin and Sterelny (2008). Oksanen (2014) develops the connection between this literature and de-extinction. 4 See Nicholls (2008) and Shapiro (2015) for excellent discussions of the science. 5 Siipi (2014) explores the role that our intuitions about authenticity might play here. 3 once the information required to bring them back is thoroughly destroyed? These conceptual questions are important, but the perspective I defend in this paper does not really depend on how we answer them. One might reasonably wonder if I have not just set all the interesting questions to one side. My goal, however, is to augment the ongoing discussion of all of the above issues by providing a clearer picture of what de-extinction would mean in macroevolutionary terms. This is a case where paleontology has some insight to offer. Perhaps serendipitously, de-extinction could also help to vindicate the notion of species selection. I take my cue here from an offhand remark that paleontologist David Jablonski included in a major review of research on species selection: “[t]oday’s biota appears to be in the midst of a massive experiment in strict-sense species selection” (2008, p. 515). In what follows, I explain why Jablonski’s idea is right and show how it provides a helpful new perspective on the de-extinction debate. The argument of this paper also lines up well with the recent efforts of a few scientists to push for “conservation paleobiology,” an emerging field that seeks to apply insights from paleobiology to the contemporary biodiversity crisis (Dietl and Flessa 2011; Rick and Lockwood 2012; Turner 2016). There are many ways in which methods, data, and ideas from paleobiology might be relevant to conservation efforts. For example, the fossil record contains information about how the geographic range size of different species might respond to climate shifts, and about which species are more prone to extinction during episodes of warming. In the spirit of conservation paleobiology, I argue here that one of the central theoretical concepts of 4 recent paleobiology—species sorting—provides useful context for thinking about deextinction. The larger message of this paper is that extinction rates and extinction magnitudes—the focus of nearly all the ongoing discussion of the biodiversity crisis— are merely part of the ethical picture. Just as important is the issue of which species persist and which don’t. Extinction selectivity matters, both on human timescales, and for the long-term evolutionary future. My claim is that we won’t really have a good account of the ethics of de-extinction unless we get clear about the ethics of artificial species selection. Or to put it another way, we could use some ethical principles concerning interventions in macroevolutionary processes. 2. Diagnosing the popular extinction debate: Extinction rates are not all that matters Much of the current discussion of our ongoing biodiversity crisis focuses on overall rates of (regional or global) biodiversity loss. I want to suggest that this emphasis leaves out something important. Extinction selectivity—that is, which species persist vs. perish—tends to get lost in discussions of extinction rates and magnitude. And yet the selectivity of extinction matters as much to the future, both the near term (say, the next couple of centuries) and the longer term evolutionary future, as the magnitude and rate of species loss do. One lesson we can draw from paleontology is that influencing the process of species sorting is a way of shaping the evolutionary future. 5 In her recent popular book, The Sixth Extinction (2014), science writer Elizabeth Kolbert explores the idea that we are on the cusp of a sixth mass extinction event, roughly on a par with the “big five” mass extinction events that paleontologists have identified. The most recent of the “big five” occurred 66 million years ago (mya), when a combination of events that very likely included an asteroid collision wiped out the ammonites, the non-avian dinosaurs, and much of the rest of the planet’s biological diversity. The possibility that human activities have initiated a sixth mass extinction event is one that mainstream conservation biologists take seriously. The idea has been around for a while: in 2003, E.O. Wilson referred to the human species as “the planetary killer.” Concern about the sixth extinction is also closely related to the Pleistocene overkill hypothesis, according to which human hunters contributed to the demise of many big animal species, including the woolly mammoth, over 10,000 years ago (Martin 2005). The Pleistocene overkill hypothesis remains controversial, but it has both fed into and drawn life from the idea that humanity is the perpetrator of massive biodiversity loss.6 Thanks in part to Kolbert’s book, the idea of the sixth mass extinction now dominates much of the public discussion of extinction. It certainly lies in the background of public discussion of de-extinction. On the one hand, the urgent need to stave off a mass extinction event could help motivate the use of biotechnology to try to reverse recent species losses. On the other hand, one of the major concerns about de-extinction is that the technology could distract from mainstream 6 For a more skeptical perspective on Pleistocene overkill, see Grayson and Melzer (2003), as well as Wolverton (2010). 6 conservation efforts and make people complacent about biodiversity loss. Either way, the de-extinction discussion is taking place in a context of increasing concern about mass extinction. The “big five” mass extinctions are identifiable only in retrospect, by studying the fossil record millions of years after the fact. As a working definition, Barnosky et al. (2011) define a mass extinction as an episode in which the earth loses 75% of its species in a geologically short time. We should bear in mind, though, that a period of tens or even hundreds of thousands of years could still count as geologically short. What reason is there to think that we are heading into such an event? Barnosky and colleagues (2011) provide a helpful review of the relevant scientific work. One approach, which they adopt, is to estimate rates of extinction for various groups for the last 500 years and project those rates into the future, while making certain assumptions. For example, what if we supposed that all the species listed as “threatened” according to IUCN classifications go extinct over the next century? (This, of course, pessimistically assumes a massive failure of conservation efforts.) Suppose we then calculated the current extinction rate on that assumption, and projected that rate into the future? How long would it take for us to lose 75% of all species, thus crossing the mass extinction threshold? [I]f all threatened species became extinct within a century, and that rate then continued unabated, terrestrial amphibian, bird, and mammal extinction would reach Big Five magnitudes in ~240 to 540 years (241.7 years for 7 amphibians, 536.6 years for birds, 334.4 years for mammals). (Barnosky et al. 2011, p. 55). This sort of model projection amounts to a kind of thought experiment: What would happen if things go very badly, conservation-wise? A geologically rapid sixth mass extinction is a very real possibility. I am skimming over many complications and caveats, but the important point is that scientists take the idea of the sixth extinction very seriously. For example, in an even more recent paper that received attention in the popular press, Ceballos et al. (2015, p.1) claim in no uncertain terms that their estimates of extinction rates “reveal an exceptionally rapid loss of biodiversity over the last few centuries, indicating that a sixth mass extinction is already underway.” Perhaps surprisingly, Stewart Brand has criticized this emphasis on a sixth mass extinction. One might think that Brand, whose Long Now Foundation supports research on de-extinction, would want to frame de-extinction research as part of the larger effort to avert mass extinction. However, in a recent popular essay on extinction in Aeon magazine, Brand (2015) develops several lines of criticism of the emphasis on mass extinction. First, he argues that scientists’ projections might be overly pessimistic. He points out that the high rates of species loss over the last 500 years are due primarily to extinctions that have occurred on islands. We know that island species are especially vulnerable—rats and cats have wreaked havoc among bird populations on many Pacific islands. What if most of the island species that were going to go extinct have already disappeared? That might be a reason for thinking that extinction rates will drop. Second, even if the forward-looking projections are 8 reasonably accurate, species loss is neither necessary nor sufficient for damaging ecosystem health—and ecosystem health, Brand argues, is what we should really care about. We can do a lot of harm to ecosystems by reducing the abundance of wild populations, even if we stop short of causing extinctions. Moreover, some ecosystems seem to do passably well even after suffering lots of extinctions, provided that newcomers replace some of the species that were lost, as has happened on some Pacific islands. For these and other reasons, Brand is skeptical about the value of stewing about the possibility of a sixth mass extinction. My reason for introducing this tension between Kolbert’s emphasis on the sixth extinction and Brand’s reluctance to dwell on gloomy projections is not to try to settle the debate. Rather, I want to highlight an important scientific issue that is largely absent from these popular discussions: namely, the selectivity of extinctions, especially mass extinctions. One thing that we miss by focusing only on the rate and magnitude of extinctions is that it matters a lot which species persist and which go extinct. Selectivity is a major issue regardless of whether or not we are heading into a sixth mass extinction. In developing his argument about species losses on islands, Brand unwittingly appeals to an example of differential extinction risk. Extinction does not happen at random, and some species, such as those living on islands, are more vulnerable than others. The selectivity of extinction can matter a great deal, even on human timescales. Stephen Jay Gould (1989) asked us to imagine rewinding the tape of history back to the Cambrian period, some 500 million years ago. Imagine some small difference in which species persist vs. go extinct. Then play the tape back in the 9 imagination. Gould argued that the downstream evolutionary history would be completely different and that evolutionary history is highly contingent upon the selectivity of upstream extinctions. Gould’s thought experiment also works on human timescales. Rewind the tape of history 150 years. Suppose that North American bison go extinct but that passenger pigeon populations recover after being decimated in the nineteenth and early twentieth centuries. The world would be significantly different, and yet (ex hypothesi) with no difference at all in the extinction rate or the overall amount of biodiversity loss.7 Focusing only on the extinction rates leaves out something of great ecological and evolutionary importance. 3. Species selection Extinction selectivity is an important aspect of our biodiversity crisis that’s often left out of popular discussions of an impending sixth mass extinction. Paleontologists’ notion of species selection is highly relevant to this issue of extinction selectivity. In this section, I’ll provide a sketch of species selection theory. In the next section, I’ll go on to explore how species selection looks when human activities begin making a difference to the differential persistence and extinction of species. My larger suggestion is that species selection theory can go a long way toward helping us think 7 For convenience, I am loosely equating biodiversity loss with species loss. However, most scientists and philosophers would agree that species richness is just one component of biodiversity. For a clear and wide-ranging discussion of the meaning of ‘biodiversity’, see MacLaurin and Sterelny (2008). 10 more clearly about extinction selectivity—the neglected aspect of our biodiversity crisis. Finally, in section 5, I’ll show that de-extinction would be an instance of artificial species selection. Although Darwin occasionally entertained group selection as an explanatory fallback, he almost always thought of natural selection as operating on individual organisms within a biological population. That is, selection occurs when variation in heritable traits makes for the differential survival and reproduction of individuals in a population. Over the years, some paleontologists have explored and defended the idea that something like a Darwinian selection process can operate on whole lineages or species.8 A good place to begin is with the uncontroversial notion of species sorting, or the differential extinction, persistence, and speciation of lineages. Many scientists view species selection as a special kind of species sorting, although there is much debate about what exactly is required for bona fide species selection. We know that species sorting happens in nature. For example, during the Cretaceous-Paleogene (KPg) event 66 million years ago, the non-avian dinosaurs went extinct but frogs and salamanders persisted. Extinction was selective: some made it through the sorting process, while others didn’t. And upstream extinction selectivity makes a huge difference to downstream biological outcomes. Imagine how different the planet would be today if frogs had gone extinct at the end of the Mesozoic, while a few dinosaur species had survived. 8 Turner (2011, Chapters 3 and 4) provides a more detailed introduction to the debate about species selection. Jablonski (2008) reviews the scientific literature on species selection. See also Grantham (2002). 11 Species sorting can be either biased or unbiased. In this context, “bias” does not refer to any prejudice or predilection on the part of scientists or observers (although later on, in section 5, I will be talking about bias in that familiar sense); it has rather to do with probabilities of extinction, persistence, and speciation. The relevant sense of “bias” is best illustrated by coin tossing. Imagine a simple game in which you toss a fair coin. The resulting series of heads and tails is a pattern; the process that generates that pattern is unbiased with respect to the outcomes. Play the same game with a weighted coin, where the probability of heads exceeds .5, and the process will be a biased one. Now imagine a simple model in which each species has a certain probability of extinction vs. persistence with each “turn” or interval. If there is something about the species, such as (say) extreme ecological specialization or small geographic range, that makes extinction more likely, then the process will be biased.9 Of course, it is an open empirical question whether large-scale evolutionary processes are biased or unbiased in this sense. Stephen Jay Gould has, at times, suggested that during mass extinction events, extinction selectivity might resemble a lottery.10 It is probably most helpful to think of the lottery view as a kind of limiting case in which extinction selectivity is due entirely to chance. 9 This way of thinking about macroevolution derives from the work of the MBL group in the early 1970s (Raup, Gould, Schopf, and Simberloff 1973). They developed one of the first computer simulations of macroevolution, and their simulation, known as the MBL model, treated it as an unbiased process. 10 See, for example, Gould (1993). And in the context of his discussion of contingency, Gould (1989, p. 48) writes that “perhaps the grim reaper of anatomical designs is only Lady Luck in disguise.” Turner (2010) argues that the stochasticity of macroevolutionary sorting processes is part of what Gould meant by “contingency.” 12 If one possibility is that species sorting is an unbiased, stochastic process, another possibility is that various factors could influence a lineage’s probability of extinction or speciation. This could lead to a biased process in which different species have different probabilities of extinction vs. persistence. We have already seen that where you live might matter: species on islands might have higher extinction risk. Body size could matter, too. Having a larger body size could make a species more vulnerable to extinction.11 Perhaps, when times get really bad—as for example during the end-Cretaceous asteroid catastrophe—species that are capable of burrowing and hiding underground might have a higher probability of making it through. Another idea, which goes all the way back to the nineteenth century paleontologist, Edward Drinker Cope, is that ecological specialization makes a species more vulnerable to extinction, an idea that remains important in conservation biology (Gallagher et al. 2015). Intuitively, it makes sense to talk about some species having higher risk of extinction than others; indeed, conservation biologists routinely try to analyze the extinction risks of current populations so as to mobilize resources most effectively. However, as soon as we start talking about different species having different probabilities of extinction, persistence, and speciation, we are in effect saying that species sorting is biased, and are very close to talking about whole species having differential fitness. 11 This idea that body size increases extinction risk has a long history in paleontology. It’s closely related to the “Lilliput effect,” which is an observable pattern in the fossil record (Harries and Knorr 2009). The fauna that show up in the fossil record after mass extinction events tend to have smaller body sizes than those that lived just prior to the mass extinction. 13 One of the first scientists to spell out the theory of species selection was the paleontologist Steven Stanley: In this higher-level process species become analogous to individuals, and speciation replaces reproduction. The random aspects of speciation take the place of mutation. Whereas, natural selection operates upon individuals within populations, species selection operates upon species within higher taxa, determining statistical trends. In natural selection types of individuals are favored that tend to (A) survive to reproduction age and (B) exhibit high fecundity. The two comparable traits of species selection are (A) survival for long periods, which increases the chance of speciation, and (B) the tendency to speciate at high rates. Extinction, of course, replaces death in the analogy (Stanley 1975, p. 648). Species selection theory gained ground in paleontology in the 1970s and 1980s, at a time when many other evolutionary biologists were committed to gene selectionism. And it raises deep questions about the relationship between macro- and microevolution (Grantham 2007). Proponents of species selection have often described their project as the “hierarchical expansion of evolutionary theory.” Scientists and philosophers have not always agreed about what precisely is required for species selection. The broadest possible view just identifies species selection with biased species sorting. This is sometimes referred to as the “emergent 14 fitness” view.12 The crucial idea here is that the fitnesses of individual organisms and the fitnesses of whole species can pull apart. For example, natural selection working within a species might favor, say, larger body size, or greater ecological specialization. But these are traits that might increase the extinction risk (hence, reduce the fitness) of the species as a whole. This independence of the species-level fitness from the fitnesses of the individual organisms is what is meant by “emergent fitness.” Some scientists have defended more stringent views about what’s required for species selection. One concern about the emergent fitness approach is that the traits that make a difference to species-level fitness—say, by increasing or decreasing extinction risk—might still be traits of the organisms in question, rather than traits or characteristics of the species. For example, if burrowing underground makes it more likely that a species will make it through an asteroid catastrophe, then that might enhance species-level fitness. But it’s the individual organisms that do the burrowing, not the species. This worry has led some scientists to defend a narrower “emergent character” approach.13 According to this view, species selection in the strict sense only occurs when some emergent trait or character of the species as a whole is 12 Lloyd and Gould (1993) defend a version of this emergent fitness view. For an early version, see Arnold and Fristrup (1982). Okasha (2006, pp. 207-8) similarly sets a fairly low bar for species selection. For a nice characterization of the emergent fitness view, see also Gould (2002, p. 241). 13 Elizabeth Vrba (1983, 1984) has been a leading defender of the emergent character approach. See also Vrba and Gould (1986). Gould (2002, pp. 656ff.) has a helpful discussion of both approaches, but he ultimately sides with the more permissive emergent fitness view. 15 making a difference to species-level fitness.14 Perhaps the best known effort to document species selection in this strict sense is David Jablonksi’s research on geographic range size of marine invertebrates (Jablonski 1987). Crucially, geographic range is a trait of the species as a whole, not of the individual organisms. These conceptual questions about how much is required for bona fide species selection remain unsettled. But for present purposes that won’t matter too much. My main claim is that de-extinction would be a kind of artificially biased species sorting. We might call that “loose sense” or “broad sense” species selection—or species selection according to the more permissive emergent fitness view—while allowing that it remains open whether de-extinction would count as species selection in the strict sense. There is a lot of disagreement among scientists about how much species selection might matter in evolutionary history. On the one hand, there are scientists like Gould who see it as a major factor. On the other hand, many scientists take the view that species selection could happen, but they quite reasonably wonder how badly we need this add-on to standard evolutionary theory. Species selection could be one of those rare or unusual natural processes—possible in principle, but on the whole, not very central to our understanding of evolutionary theory. Perhaps we can explain most of what needs explaining without invoking species selection at all. On this mildly dismissive view, species selection is a respectable idea whose partisans have tended to exaggerate its scientific importance. Here the discussion tends to get 14 One major challenge for this view is to clarify the relevant sense of ‘emergence” (Grantham 1995, 2007). 16 bogged down in the details of potential case studies of species selection (for a comprehensive review of the research, see Jablonski 2008). However, even those who take a more dismissive view must acknowledge that human activities can bias the sorting process. Indeed, de-extinction—if scientists succeed in pulling it off—would count as a clear-cut case of broad sense species selection. 4. Artificial species selection For the most part, the paleontologists working to develop models of species selection have not given much attention to the possibility of humans doing the selecting at the level of whole species. That omission is only natural, because paleontologists tend to be interested in evolutionary patterns and processes in the deep past, long before humans even evolved. But with species selection theory in place, there’s no reason not to consider the possibility that human agency could matter to species selection processes. Much as Darwin argued that selection at the level of individual plants and animals is sometimes artificial, it might make sense to say that species selection could also be artificial. Famously, the first chapter of Darwin’s Origin of Species (1859) is all about pigeons. Darwin focused on pigeons, in part, because he thought there was good evidence that all of the various fancy pigeon breeds—the pouters and tumblers, the fantails and Jacobins—descended from a single wild common ancestor. He tried to warm readers up to the idea of natural selection by showing how cumulative selective breeding over many generations could produce varieties so different from one 17 another that they might almost count as different species. Darwin saw that many of the features of the fancy pigeon breeds would reduce fitness in the wild, and argued that these traits exhibit adaptation “not indeed to the animal’s own good, but to man’s use or fancy” (1859, pp. 29-30). Selection, as Darwin understood it, is the biased sorting of individuals in a population; selection is artificial when human activities help to generate the bias. Interestingly, there are different ways of thinking about what makes a process or a product artificial. Some writers emphasize intent (Hilpinen 1992). On such a view, an artifact must have been made with some intent, or for some purpose. So for example, a tool made by a non-human animal could count as an artifact. And some by-products of human activity, such as as a flake that was a waste product of stone tool making, would not count as artifacts. This is probably not how Darwin was using the term ‘artificial,’ because he explicitly allowed that artificial selection is often unconscious. “But, for our purposes, a kind of Selection, which may be called Unconscious, and which results from every one trying to possess and breed from the best individual animals is more important. Thus, a man who intends keeping pointers naturally tries to get as good dogs as he can, and afterwards breeds from his own best dogs, but he has no wish or expectation of permanently altering the breed. 18 Nevertheless, I cannot doubt that this process, continued during centuries, would improve and modify any breed …” (pp. 34-5).15 The breeder who chooses to breed his best dogs might not be fully conscious of what makes those dogs seem the best. Subtle aesthetic biases and preferences come into play. Over time, breeders could modify the breed without ever meaning to do so. Clearly, Darwin does not think that intent to bring about change is necessary for artificial selection. Where selection is unconscious, the breeder might not even be able to foresee the longer-term results. It seems that what makes selection artificial in these sorts of cases is merely the fact that humans are influencing the process. In what follows, I’ll use the term ‘artificial’ in the way that Darwin did, as a way of highlighting human influence on historical processes. Clearly, humans can and do bias the process of species sorting in various ways. In some cases, the biasing is open, explicit, and conscious. For example, according to the Centers for Disease Control, in 1986, 3.5 million people in 20 countries suffered from Guinea worm disease.16 Today, according to recent news reports, that number is down to two confirmed cases. The difference is that 15 Sterrett (2002) argues that the two types of artificial selection in Darwin’s work— methodical vs. unconscious—closely parallel his distinction between natural selection based on the principle of divergence vs. the principle of extinction. In her view, the analogy between artificial and natural selection has more structure than many realize. This is a plausible reading of Darwin. I’d argue that it’s also important that all natural selection is unconscious, so that in pointing out that some artificial selection is unconscious, Darwin is eroding the difference between the two. 16 Information about the Guinea worm eradication program is available from the Centers for Disease Control at http://www.cdc.gov/parasites/guineaworm/gwep.html, retrieved 17 August, 2015. 19 governments and international organizations have worked together on a systematic eradication campaign. Since the Guinea worm cannot survive without human hosts, this is a case of deliberate, planned extinction. At the same time, we humans go to great lengths to save the species we like. At one private game reserve in South Africa, wildlife managers have taken to injecting poison into rhino horns (Smith 2013). The poison—an insecticide usually used to kill ticks—is mildly toxic to humans, and will induce nausea and diarrhea in anyone who consumes the horn in powdered form. The wildlife managers are also injecting colored dye as a warning to poachers. Consumers of poached rhinoceros horn, beware. Whatever you think of the ethics of these two cases, both involve human actors going to great lengths to sort species—in the one case, for extinction, in the other, for persistence. They are fairly obvious cases of artificial species selection, at least in the broad sense. Notably, the cases of selective breeding that Darwin cared most about involved cumulative selection, with small changes in the population adding up over many generations, and with generations of breeders selecting with at least some degree of consistency. Because species sorting takes place at a much larger temporal scale, it’s not clear there has been enough time for the same sort of cumulative directional selection to take place at the species level. Even if we suppose, in keeping with the Pleistocene overkill hypothesis, that humans have been in the business of causing extinction for over 10,000 years, it’s not at all clear that there is long-term, cumulative selection taking place at the species level. If anything, an earlier bias against big animals may have been replaced by a bias in their favor. This, then, marks one significant difference between individual-level artificial selection and 20 artificial species selection: it’s not clear that the latter has had enough time to produce the kind of cumulative directional selection that Darwin saw when he considered breeds of dogs or fancy pigeons. With Darwin’s notion of artificial selection in mind, we can say that species selection is artificial when human activities make a difference to species-level fitness, or to different species’ prospects for persistence and/or speciation. ‘Anthropogenic’ might even be a more apt term than ‘artificial.’ One potential worry about this notion of artificial species selection is that it’s too broad. This approach would seem to imply that virtually any niche construction on the part of humans, any activity that transforms nature, could count as artificial species selection. For example, if human activities destroy some of the habitat of a particular species, that could increase its extinction risk by some small amount, making this a case of artificial species selection. One could, alternatively, insist on a narrower conception of artificial species selection, one that requires intent or conscious planning. In a case where human activities just happen to contribute to the differential persistence and extinction of species, we might not want to call that artificial species selection.17 Darwin’s own view of artificial selection, however, is extremely broad. While it’s true that his examples of artificial selection, even unconscious selection, all involve animal breeding, he gives no principled reason for restricting artificial selection to 17 This issue of intent might matter for ethical reasons, even if we don’t think intent is necessary for selection to be artificial. As one anonymous referee pointed out, there could be cases where people foresee that their activities will increase the risk of extinction, but do not intend for the extinction to happen. In such cases, the principle of double effect might apply. 21 animal breeding contexts. Once you allow that artificial selection can be unconscious, it’s hard to see why the notion should be restricted to domesticated plants and animals. There are plenty of other contexts in which humans, without intending to cause evolutionary change or even foreseeing that change, impose selection pressures on other populations. Human fishing imposes a downward selection pressure on body size in many fish species; agricultural pesticide applications impose severe selection pressures on insect populations. In these sorts of cases, we end up with smaller bodied fish and pesticide resistant insects—the very opposite of what people might want. But it’s hard to see why we should not consider these to be cases of unconscious artificial selection. If we take a broad view of what counts as artificial selection at the individual level, it’s hard to see why we shouldn’t take a similarly broad view at the species level. Although I favor a broader conception of artificial species selection, we’ll see in section 5 that de-extinction would count as artificial species selection even on the narrowest possible conception. For de-extinction would involve conscious human decisions about which species have futures and which don’t; de-extinction researchers are good macroevolutionary analogues of Darwin’s animal breeders. A second potential worry is that by making too much of the distinction between natural and artificial selection, we single out humans in a way that’s problematic. Arguably, artificial selection is just a special case of natural selection. And of course humans are not the only species that exerts selection pressures on other populations. Selection is selection, and which species is applying the selection pressure might seem incidental. We don’t need to suppose, however, that the 22 distinction between artificial and natural selection tracks anything having any scientific or metaphysical importance. The main reason for singling out a certain kind of (species-level) sorting and calling it ‘artificial’ is to underscore the fact that when human activities bias species sorting processes, that creates and ethical and a policy problem. Once we realize that we’re engaged in artificial species selection, we need to think about what sorts of principles might guide our macroevolutionary interventions. Because small upstream differences—differences in which species persist vs. go extinct—can make a huge difference to the subsequent course of evolutionary history, just how we influence species sorting processes is something that matters a great deal. Saying that species selection is artificial is a way of delineating an ethical problem. One way of thinking about the biodiversity crisis is to think of humans as inadvertently increasing the extinction risk of vast numbers of species, all at the same time. Putting things this way would be to shift the focus back to extinction rates and magnitudes (section 2). Suppose, by way of a thought experiment, that we increased the extinction risk of every species on earth by the same amount. Would that be a case of artificial species selection? It might not, since species selection, like individual level selection, is all about differential persistence and termination. As it happens, though, human impacts on other species really are differential. We’re increasing the extinction risks of some groups (e.g. amphibians) more than others. And some species (say, coyotes), have probably had their prospects improved by human activities. So the ongoing biodiversity crisis really is an example of artificial species selection at work. 23 5. Assessing candidate species for de-extinction: A case of artificial species selection What does all of this have to do with de-extinction? In sections 3 and 4, I argued that the notion of artificial species selection gives us a helpful way of thinking about the ongoing biodiversity crisis. Along the way, I’ve taken fairly expansive views of species selection (equating it with biased species sorting) and artificial selection (treating selection as artificial whenever there’s any human involvement.) I’ll now show that this notion of artificial species selection also provides a helpful way of framing questions about de-extinction. Thinking clearly about de-extinction means seeing it as an exercise in artificial species selection. One pressing issue for proponents of de-extinction is how to assess candidate species. Which species should be prioritized for de-extinction research? This matters a great deal because the technical problems that need to be solved will vary from one species to the next. Some prioritization needs to happen before investing in deextinction research. Seddon, Moehrenschlager, and Ewen (2014) offer one reasonable approach to this prioritization problem. Examining their proposal will help to make it clear just why de-extinction would be a form of artificial species selection. To start with, Seddon, Moehrenschlager, and Ewen (hereafter, SME) treat de-extinction as a special case of species reintroduction or translocation. They argue that the criteria for determining which species would be good candidates for deextinction should resemble the criteria for reintroducing a species that is only locally 24 extinct. They begin, accordingly, with the 2013 IUCN Guidelines for Reintroductions and Other Conservation Translocations and develop a set of ten standards, in the form of questions, for assessing candidates for de-extinction. The questions (listed in Table 1) speak for themselves. More interesting, for present purposes, is what SME do with those questions. They start with a pool of 20 candidate species (Table 2). Their source for this initial pool is none other than the Long Now Foundation, Stewart Brand’s organization. SME then select three species from this initial pool: the Xerces blue butterfly, the Yangtze River dolphin (a.k.a. the baiji), and the thylacine (a.k.a. the Tasmanian wolf). They proceed to apply their ten questions to each of these candidate species, with the aim of illustrating how the assessment process might work. SME’s initial pool of candidates reflects some very interesting human biases. Consider that the pool includes no plant species at all, and only one invertebrate species. The list includes two big marine mammals, in addition to the Yangtze River dolphin. All the other species are terrestrial. There are no reptiles or amphibians on the list at all. The list of terrestrial birds and mammals seems biased toward the large: elephant birds, dodos, moas, and great auks all make the first cut, as do aurochs, woolly mammoths, mastodons, and saber-toothed cats. Surprisingly, since their stated goal is to develop a procedure for assessing candidate species for de-extinction, SME say nothing at all by way of justification for their selection of an initial pool of candidates. It’s a familiar idea that people often exhibit a bias toward charismatic megafauna, and perhaps also a bias toward creatures more similar to us (e.g. toward other mammals over amphibians and reptiles). People are likelier to invest a great 25 deal in conserving big-bodied species (think African elephants) and also in bringing them back (think woolly mammoths). The preference for bigger animals is, however, just one example of how human biases can influence species level fitness. The particular nature of the bias does not actually matter much for the argument I want to make. The argument is more general than that: any system at all for prioritizing candidates for de-extinction will, by definition, create a biased species sorting process. Nor is this a special feature of de-extinction. We are already in the business of prioritizing threatened species for conservation investments, which means that we are already in the business of biasing species sorting processes. Any standards at all that we use to help direct our allocation of resources for either conservation or de-extinction will bias the sorting process. I have said that my goal was not to issue a normative verdict as to whether we should pursue de-extinction technology. Of course, the “should we do it?” question is the obvious one. Without denying the importance of that question, the argument that I have developed suggests a somewhat different way of framing the ethical and policy issues. I want to suggest that the human involvement in macroevolutionary sorting processes is a policy issue. 6. Conclusion In this paper, I have argued (section 2) that one thing missing from current popular discussions of extinction is the issue of extinction selectivity. And the 26 selectivity of extinction, as contrasted with both the rate and magnitude of biodiversity loss, matters on both human and geological timescales. Paleontologists, however, have long been interested in the selectivity of extinction, including (perhaps especially) mass extinction events. In section 3, I offered a brief introduction to species selection theory as it was developed in paleobiology in the 1970s and 1980s. The central idea is that species sorting processes are sometimes biased. In section 4, I suggested that several of Darwin’s observations about artificial selection apply just as easily to species selection as to familiar cases involving the selective breeding of individual plants and animals. Finally, in section 5, I argued that de-extinction would be an instance of artificial species selection. It’s not the only instance. As David Jablonski observed, we are in the middle of a huge experiment in macroevolutionary species sorting, and the sorting taking place now is largely anthropogenic. We need to treat de-extinction as one aspect of this great sorting. De-extinction is not merely a response to biodiversity loss; it is a way of shaping the evolutionary future. Darwin sought to soften up resistance to his ideas by starting with relatively easy cases of unconscious artificial selection and then pointing out that natural selection works in precisely the same way. Species selection remains somewhat controversial among evolutionary biologists and paleontologists, but it’s hard to see why, once we come to appreciate anthropogenic bias in ongoing species sorting processes. The obvious way to defend (broad sense) species selection is to start with the uncontroversial cases of artificial selection. Whether de-extinction would count as species selection in the strict sense (i.e., in the sense of the emergent character approach) is a more difficult question. It’s not clear whether our biases (e.g. in favor 27 charismatic megafauna) are targeting properties of the species are emergent in the right sort of way. But there can be no doubting that species selection in the loose sense is an important factor in evolution. We are, after all, doing the selecting. The perspective offered here does not yield any clear verdict as to whether we should try to develop de-extinction technology, or whether, once developed, we should use it. What it does do is provide important context for that ethical discussion. So far, most of the discussion in environmental ethics and the philosophy of conservation biology has focused on the value of biodiversity and the need to stop the loss of biodiversity. That is an important issue, but it builds in the assumption that what really matters is how much biodiversity we have. So extinction rates and magnitudes become the main issue. Indeed, the whole point of clarifying the value of biodiversity, presumably, is to make the ethical case for reducing extinction rates. But extinction rates and magnitudes, while important, are only part of the story. What we really need, if I am right, is an ethic for species sorting, but this is also unmapped philosophical territory. Thanks to the efforts of environmental ethicists, we now have a fairly good idea what to say about the value(s) of biological diversity, but we do not have much theory about how to make decisions that will shape the macroevolutionary future of the planet for millions of years to come. De-extinction needs to be understood as the technological manipulation of macroevolution. 28 Acknowledgments An earlier version of this paper was presented at the ISHPSSB meeting in Montreal, Canada, in July 2015. I am grateful to colleagues at that meeting (especially Leonard Finkelman, Markku Oksanen, and Helena Siipi) for their ideas and critical feedback. The paper has also benefitted from comments from Russell Powell, as well as two anonymous reviewers for this journal. References Arnold, A.J., and K. Fristrup (1982), “The Theory of Evolution by Natural Selection: A Hierarchical Expansion,” Paleobiology 8: 113-129. Barnosky, A.D., et al. (2011), “Has the earth’s sixth mass extinction already arrived?” Nature 471: 51-57. Brand, S. (2015), “Rethinking Extinction,” Aeon. Available at http://aeon.co/magazine/science/why-extinction-is-not-the-problem/, retrieved 1 August, 2015. Ceballos, G., Ehrlich, P.R., Barnosky, A.D., Garcia, A., Pringle, R.M., and T.M. Palmer (2015), “Accelerated Modern Human-Induced Species Losses: Entering the Sixth Mass Extinction,” Science Advances 1(5). Available online at http://advances.sciencemag.org/content/advances/1/5/e1400253.full.pdf. Retrieved 2 August, 2015. Cohen, S. (2014), “The Ethics of De-Extinction,” Nanoethics 8: 165-178. Delord, J. (2007), “The Nature of Extinction,” Studies in History and Philosophy of Biology and Biomedical Sciences 38: 656-667. 29 Delord, J. (2014), “Can We Really Recreate a Species by Cloning it?” in The Ethics of Animal Modification and Recreation: Reviving, Rewilding, Restoring, edited by Markku Oksanen and Helena Siipi, Palgrave Macmillan, pp. 22-39. Dietl, G.P., and K.W. Flessa (2011), “Conservation Paleobiology: Putting the Dead To Work,” Trends in Ecology and Evolution 26(1) Gallagher, A.J., Hammerschlag, N., Cooke, S.J., Costa, D.P., Irschik, D.J. (2015), “Evolutionary theory as a tool for predicting extinction risk,” Trends in Ecology and Evolution 30(2): 61-65. Gould, S.J. (1989), Wonderful Life: The Burgess Shale and the Nature of History. New York: W.W.Norton. Gould, S.J. (1993), “The Wheel of Fortune and the Wedges of Progress,” in Gould, S.J., Eight Little Piggies: Reflections in Natural History. New York: W.W. Norton, pp. 300-312. Gould, S.J., and E. Lloyd (1999), “Individuality and Adaptation Across Levels of Selection: How Shall We Name and Generalize the Unit of Darwinism?” Proceedings of the National Academy of Sciences 96: 11904-11909. Grantham, T. (1995), “Hierarchical Approaches to Macroevolution: Recent Work on Species Selection and the ‘Effect Hypothesis’,” Annual Review of Ecology and Systematics 26:301-326. Grantham, T. (2002), “Species Selection,” in M. Pagel (ed.), Encyclopedia of Evolution. Oxford: Oxford University Press, pp.1086-7. Grantham, T. (2007), “Is Macroevolution More than Successive Rounds of Microevolution?” Paleontology 50(1): 75-85. 30 Grayson, D.K., and D.J. Meltzer (2003), “A Requiem for North American Overkill,” Journal of Archaeological Science 30: 585-593. Harries, P.J., and P.O. Knorr (2009), “What Does the ‘Lilliput Effect’ Mean?” Palaeogeography, Palaeoclimatology, Palaeoecology 284: 4-10. Hilpinen, R. (1992), “On artifacts and works of art,” Theoria 58(1): 58-82. Jablonski, D. (1987), “Heritability at the Species Level: Analysis of Geographic Ranges of Cretaceous Mollusks,” Science 238-360-363. Jablonski, D. (2008), “Species Selection: Theory and Data,” Annual Review of Ecology, Evolution, and Systematics 39: 501-524. Jones, K.E. (2014), “From Dinosaurs to Dodos: Who Could and Should We DeExtinct?” Frontiers in Biogeography 6(1): 20-24. Jørgensen, D. (2013), “Reintroduction and De-Extinction,” Bioscience 63(9): 719-20. Kolbert, E. (2014), The Sixth Extinction: An Unnatural History. New York: Henry Holt and Company. Lloyd, E., and S.J. Gould (1993), “Species Selection on Variability,” Proceedings of the National Academy of Sciences 90: 595-599. MacLaurin, J., and K. Sterelny (2008), What is Biodiversity? Chicago: University of Chicago Press. Martin, P. (2005), Twilight of the Mammoths: Ice Age Extinctions and the Rewilding of America. Berkeley, CA: University of California Press. Nicholls, H. (2008), “Let’s Make a Mammoth,” Nature 456: 310-314. Norton, B. (1987), Why Preserve Natural Variety? Princeton, NJ: Princeton University Press. 31 Okasha, S. (2006), Evolution and the Levels of Selection. Oxford: Oxford University Press. Oksanen, M. (2014), “Biodiversity and the Value of Human Involvement,” in Oksanen, M. and H. Siipi, eds., The Ethics of Animal Recreation and Modification: Reviving, Rewilding, Restoring. Palgrave MacMillan, pp. 150-169. Oksanen, M. and H. Siipi, eds. (2014), The Ethics of Animal Recreation and Modification: Reviving, Rewilding, Restoring. Palgrave MacMillan. Raup, D.M., Gould, S.J., Schopf, T.J.M., and D. Simberloff (1973), “Stochastic Models of Phylogeny and the Evolution of Diversity,” Journal of Geology 81: 525542. Rick, T.C., and R. Lockwood (2012), “Integrating Paleobiology, Archaeology, and History to Inform Biological Conservation,” Conservation Biology 27(1): 45-54. Sandler, R. (2014), “The Ethics of Reviving Long Extinct Species,” Conservation Biology 28(2): 354-360. Sarkar, S. (2005), Biodiversity and Environmental Philosophy: An Introduction. Cambridge: Cambridge University Press. Shapiro, B. (2015), How to Clone a Mammoth: The Science of De-Extinction. Princeton, NJ: Princeton University Press. Seddon, P.J., Griffiths, C.J., Soorae, P.S., and D.P. Armstrong (2014), “Reversing Defaunation: Restoring Species in a Changing World,” Science 345(6195): 406412. Seddon, P.J., Moehrenschlager, A., and J. Ewen (2014), “Reintroducing Resurrected Species: Selecting DeExtinction candidates,” Trends in Ecology and Evolution 29(3): 140-147. 32 Sepkoski, D. (2012), Rereading the Fossil Record. Chicago: University of Chicago Press. Sherkow, J.S., and H.T. Greely (2013), “What if Extinction is Not Forever?” Science 340: 32-33. Siipi, H. (2014), “The Authenticity of Animals,” in M. Oksanen and H. Siipi (eds.), The Ethcis of Animal Re-Creation and Modification: Reviving, Rewilding, Restoring. Palgrave MacMillan, pp. 77-96. Smith, D. (2013), “South African Game Reserve Poisons Rhino’s Horns to Prevent Poaching,” The Guardian (4 April 2013), available online at http://www.theguardian.com/environment/2013/apr/04/rhino-hornspoisoned-poachers-protect. Retrieved 17 August, 2015. Stanley, S. (1975), “A Theory of Evolution Above the Species Level,” Proceedings of the National Academy of Sciences 72(2): 6467-650. Sterrett, S. (2002), “Darwin’s Analogy Between Artificial and Natural Selection: How Does it Go?” Studies in History and Philosophy of Biological and Biomedical Sciences 33: 151-168. Turner, D. (2010), “Gould’s Replay Revisited,” Biology and Philosophy 26: 65-79. Turner, D. (2011), Paleontology: A Philosophical Introduction. Cambridge: Cambridge University Press. Turner, D. (2014), “The restorationist argument for extinction reversal,” in The Ethics of Animal Modification and Recreation: Reviving, Rewilding, Restoring, edited by Markku Oksanen and Helena Siipi, Palgrave Macmillan, pp. 40-59. 33 Turner, D. (2016), “Conservation Paleobiology,” Extinct: The Philosophy of Palaeontology Blog, 21 March 2016, available online at http://www.extinctblog.org/extinct/2016/3/16/conservation. Vrba, E. (1983), “Macroevolutionary Trends: New Perspectives on the Roles of Adaptation and Incidental Effect,” Science 221: 387-9. Vrba, E. (1984), “What is Species Selection?” Systematic Zoology 33: 318-328. Vrba, E., and S.J. Gould (1989), “The Hierarchical Expansion of Sorting and Selection: Sorting and Selection Cannot be Equated,” Paleobiology 12: 217-228. Wilson, E.O. (2003), The Future of Life. New York: Vintage Books. Wolverton, S. (2010), “The North American Pleistocene Overkill Hypotheses and the Rewilding Debate,” Diversity and Distributions 16: 874-876. 34 Table 1. Questions for assessing candidate species for de-extinction. Reprinted from Trends in Ecology and Evolution 29(3), P.J. Seddon, A. Moerhenschlager, and J. Ewen, “Reintroducing Resurrected Species: Selecting DeExtinction candidates,” pp. 140-147, 2014, with permission from Elsevier. Table 2. List of candidate species for de-extinction. Reprinted from Trends in Ecology and Evolution 29(3), P.J. Seddon, A. Moerhenschlager, and J. Ewen, “Reintroducing Resurrected Species: Selecting DeExtinction candidates,” pp. 140147, 2014, with permission from Elsevier. 35