Edited by James O’Connell, University of Utah, Salt Lake City, UT, and approved April 9, 2013 (received for review February 12, 2013)
Around 88 large vertebrate taxa disappeared from Sahul sometime during the Pleistocene, with the majority of losses (54 taxa) clearly taking place within the last 400,000 years. The largest was the 2.8-ton browsing Diprotodon optatum, whereas the ∼100- to 130-kg marsupial lion, Thylacoleo carnifex, the world’s most specialized mammalian carnivore, and Varanus priscus, the largest lizard known, were formidable predators. Explanations for these extinctions have centered on climatic change or human activities. Here, we review the evidence and arguments for both. Human involvement in the disappearance of some species remains possible but unproven. Mounting evidence points to the loss of most species before the peopling of Sahul (circa 50–45 ka) and a significant role for climate change in the disappearance of the continent’s megafauna.
Explaining Pleistocene faunal extinctions remains one of the most challenging problems in the prehistory of Sahul (1⇓–3). The vast majority of extinctions across geological time are wholly attributable to climate-related factors (4), but claims that some, or even all, Pleistocene extinctions of large-gigantic vertebrates (Fig. 1) in Sahul were the consequence of human activity have generated particularly robust debate. Polarized views have emerged to account for the mode and timing of these events (2, 5⇓⇓⇓⇓⇓–11). A paucity of empirical data; shortfalls in radiometric dating; and, until recently, a limited appreciation of the paleoenvironmental record (7, 12) have placed considerable constraints on the ability to resolve “who or what” was responsible for these extinctions. Given these limitations, assertions such as “...the question is no longer if, but rather how, humans induced this prehistoric extinction event” (ref. 13, p. 563) are premature.
Thylacoleo carnifex, the 100- to 130-kg marsupial lion with massive “bolt cutter-like” cheek teeth and the most powerful bite for its size of any mammalian carnivore, was a formidable predator of large animals. (Drawing by Peter Schouten.)
Sahul—mainland Australia, Tasmania, and New Guinea—comprised up to ∼11 million km2 at glacial maxima. Although dominated by an expansive desert core (14), environments ranged from periglacial in Tasmania to tropical in New Guinea (Fig. 2). It was against a backdrop of deteriorating conditions leading up to the Last Glacial Maximum [LGM; in Marine Isotope Stage (MIS) 2], ∼28–19 ka, that the first people crossed the biogeographic divide (Wallacea) to enter Sahul at ∼50–45 ka (15, 16). It seems likely that the size, distribution, and density of human populations in pre-LGM times have been overestimated (17), although these were evidently behaviorally modern people (18, 19).
Sites mentioned in the main text. The continental limits of Sahul are defined by the −130-m bathymetry line, and Sahul incorporates Australia, New Guinea, and Tasmania. The Nullarbor Plain caves contain middle Pleistocene fossil deposits. Kings Creek is located in the eastern Darling Downs region.
As commonly used in the context of extinctions in Sahul, the term megafauna refers to an arbitrary compilation of relatively large mammalian, reptilian, and avian taxa, ranging in size from ∼10 kg or less up to >2,000 kg (20⇓⇓–23). In addition to Diprotodon optatum and Thylacoleo carnifex (Fig. 1), other well-known marsupial “giants” included the 230-kg kangaroo Procoptodon goliah (24), the tapir-like Palorchestes azael, and the bull-sized Zygomaturus trilobus. Among nonmammalian megafaunal species were the massively built flightless bird Genyornis newtoni (25), the anaconda-like madtsoiid snake Wonambi naracoortensis, and the 5-m-long mekosuchine crocodile Pallimnarchus pollens (26).
It is important to note that extinctions in Pleistocene Sahul were not restricted to the large-bodied species described above. Extinctions of small-bodied species, including frogs, bandicoots, dasyurids, and rodents (27), also occurred. Furthermore, a range of medium- and small-bodied species disappeared or underwent major geographic range shifts on time frames similar to those of larger taxa (28⇓⇓⇓⇓–33). Also notable is the fact that not all megafauna went extinct: Some underwent dwarfing, whereas others appear to have survived relatively unchanged, such as emus, cassowaries, wombats, salt-water crocodiles, and many species of large kangaroos, including reds (Macropus rufus), grays (Macropus giganteus and Macropus fuliginosus), and wallaroos (Macropus robustus, Macropus antilopinus, and Macropus bernardus). Taxa from the whole gamut of body size distributions, not just the largest, were affected by extrinsic factors during the Pleistocene. Only some of these factors led to extinctions. There is no reason to assume that all species lumped together as extinct megafauna were sufficiently similar ecologically or entangled (or contemporaneous, see below) to enable one simultaneous event to account for all these extinctions.
The last appearance dates in Sahul for the suite of taxa traditionally referred to as megafauna are consistent with a staggered extinction process that was in train well before the arrival of humans (3, 10, 12) (Fig. 3). Including new middle Pleistocene species recorded from south-central Australia (2), as many as 50 of the 88 known extinct megafaunal taxa are absent from fossil records postdating the Penultimate Glacial Maximum (PGM; MIS6) (3) at ∼130 ka (Fig. 3 and Table S1). Additional taxa disappeared at ∼85–80 ka (3, 6, 12). There is firm evidence for only ∼8–14 now-extinct megafaunal species overlapping with human presence on the continent. Nearly half of these late-surviving species are from New Guinea (3, 34), and most are identified from single occurrences.
Extinction chronology of Pleistocene megafaunal taxa in Sahul (88 species). For many species, terminal dates are debated, as is the definition of a “secure” date. If the selection criteria of Roberts et al. (9) are applied, as few as 8 species clearly survived to the time of human colonization of Sahul. Species reported to overlap with humans but lacking contextual information linking the dates to the taxa in question are illustrated in gray. The majority of taxa cannot be placed within the time frame of human arrival, and many lack reliable radiometric dates altogether, with some of the taxa claimed to have been the victims of human-mediated extinction known only from the Pliocene (e.g., Euryzygoma dunense). Of those taxa known to overlap with people, some persist well beyond a previously proposed 51- to 39-ka “extinction window” (9), whereas most of those falling within this window are not securely linked to the dates (six of nine taxa). Of those taxa potentially surviving until more recently than ∼50 ka, almost half are from New Guinea.
The uncertainty that currently exists around establishing accurate chronologies for megafaunal decline and extinction, particularly the paucity of securely dated fossil material (5), has been one of the major obstacles to resolving the question of what caused these extinctions. The absence of so many of the species in question following the MIS6 glacial maximum, a time of undoubtedly severe climate, with further attenuation through MIS5–MIS3 is consistent with a staggered extinction chronology. However, it has been suggested that this appearance of staggered loss could be a sampling artifact, referred to as the Signor–Lipps effect (13). Technically, this is conceivable, because the last occurrences in the fossil record are unlikely to document precisely the actual time of extinction. However, given the actual record and the fact that so many of the extinct taxa are not known to have existed within tens of thousands of years of human arrival, the most parsimonious explanation of the data are that the extinctions were indeed staggered over a period that began long before human arrival. A staggered extinction event is further supported by recent rigorous statistical tests of stratigraphically intact prehuman fossil sequences (Fig. 4) that show progressive losses of diversity over time (12), strongly suggesting that temporally progressive middle to late Pleistocene declines in diversity are not sampling artifacts.
Late Pleistocene temporal occurrences of megafauna from the intensively sampled Kings Creek Catchment, Darling Downs, southeastern Queensland (13), in relation to the hypothesized continental megafaunal extinction window and earliest evidence of human colonization. Rarefaction, bootstrapping, and associated statistical analyses demonstrate that the apparent temporally progressive loss of megafauna was a real phenomenon and cannot simply be explained as a result of taphonomic or sampling biases (14). Importantly, the data clearly point to prehuman climate changes as a driver of the losses and do not support the hypothesis that there was a local mass extinction of all megafauna at or around the hypothesized extinction window.
Extinctions of a comparable magnitude to those of megafauna have been observed for small-bodied species in a range of specific localized studies. For example, small-sized species (<3 kg) from Mount Etna in northeastern Australia (Fig. 2) experienced a 50% reduction in species richness (from 28 to 35 species to 16 to 12 species) over the course of the middle to late Pleistocene (35). These losses track progressive environmental change from rainforest to less mesic habitats. Likewise, the eastern Darling Downs records middle to late Pleistocene extinctions among both small- and large-bodied species concomitant with progressively drying environments (30⇓⇓–33).
The argument for a primary human role in the extinction of megafauna has been based on the presumption that the arrival of humans on Sahul was synchronous with the disappearance of all now-extinct megafauna (36). A link between the two seemed obvious to advocates of a human-driven process (37), and a role for climate change was subsequently discounted. The case for a human role was buttressed by claims that MIS3 (60–28 ka) was a time of relatively stable climatic conditions (13). As such, human activities were therefore the only credible explanation for the extinctions (2, 11, 38).
Proponents of early human-mediated extinctions in Sahul have based their arguments and conclusions on a number of assumptions. Key among these are that all or most now-extinct megafauna survived the PGM and the ensuing 80,000 y to be present when people arrived, that the extinctions could largely or wholly be attributed to a single cause within a relatively narrow time frame, and that climate changes within the last two glacial cycles were unremarkable relative to those of previous cycles (3, 9).
The case for a human role was bolstered following the observation of a broad overlap between human arrival and terminal dates for a small number of late-surviving megafaunal species. A hypothetical “extinction window” at 51–39 ka was proposed during which, it was argued, all or most species of now-extinct megafauna disappeared (6). Modeling studies based on the same dataset suggested that human activities could have accounted for the decline and disappearance of all megafauna within 600 y (39). Various publications before and since have offered proxy data to support the notion of a primary human role (8, 40).
Human activities as a primary extinction driver are not, however, supported by the paleontology or the archeology. The complete lack of evidence for predation on, or even consumption of, megafauna by people aside, the extinction window noted above is based on a statistical analysis of just seven sites (7). Four of these have no published data and cannot be scrutinized. A fifth site, Menindee Lake (Fig. 2), is 10,000 y younger than originally proposed but still 10,000 y older than the archeological materials from the same site (7). The remaining two localities are >100 ka in age (7), and thus earlier than the arrival of people by at least 50,000 y. Only two sites in Sahul have secure excavated contexts with co-occurrence of extinct megafauna and people: Cuddie Springs in southeastern Australia and Nombe Rockshelter in the New Guinea highlands (7, 34) (Fig. 2). Two other sites have yielded evidence for single instances of megafauna persisting after the LGM. These fauna were identified in archeological sites but were recovered from noncultural (prehuman) horizons (Cloggs Cave and Seton Rockshelter) (7). No kill-sites are known (3, 7). Although megafauna are known from Tasmania, most extinct species had vanished before humans arrived. There is a single example of a short temporal overlap between people and Protemnodon anak, but the bones of P. anak have never been identified in the rich faunal assemblages recovered from the numerous Pleistocene archeological sites investigated (41).
Middle to late Pleistocene faunal extinctions and human arrival in Sahul occurred against a backdrop of significant climatic change. Until recently, our capacity to unravel the potential impacts of climatic flux on plants and animals, including humans, has been limited by the relatively coarse resolution of the fossil records. Numerous independent studies, including analyses based on high-resolution Antarctic ice core data, now allow us to reevaluate the climatic record, the long-term trends spanning some 800 ka, and the more detailed shorter term fluctuations at critical periods through time.
There is a growing consensus that Sahul was subject to stepwise and progressive drying beginning as early as 700 ka (42), and notably within the last 350–400 ka (43⇓⇓–46). In recent years, Antarctic ice core records (47⇓–49) have revealed a distinct change in glacial-interglacial cycles after ∼450 ka. The last five interglacial stages (MIS11.3, MIS9.3, MIS7.5, MIS5.5, and MIS1) were, on average, warmer than preceding interglacial stages, as indicated by Deuterium (δD), a proxy for temperature, in the European Project for Ice Coring in Antarctica (EPICA) Dome C (EDC) record. Concurrently, from ∼450 ka, the isotopic minima, representing the coldest or full glacial stages in the EDC δD record, show an obvious downward trend. These trends mean that isotopic changes recorded for the last glacial cycle (from MIS5.5–MIS2) are almost double the amplitude of the interglacial-glacial maximum cycles between 800 and 500 ka (Fig. 5). In addition, δD (hence, temperature) is conspicuously low in the EDC ice core data from ∼70–63 ka and during MIS2. Importantly, the occurrence of two extended and severe cold periods in the one glacial cycle is unprecedented in the 800,000 y Antarctic ice core record (48).
The EDC δD (‰) record over the last 900 ka (45⇓–47) illustrates the trend toward increasingly warm interglacials (red line) and increasingly cold glacial maxima (blue line) beginning at around 400 ka. (Upper) Graphic depicts the number of species that are not represented thereafter in the fossil record in the following temporal bins: >400 ka or undated, 400–126 ka, 126–51 ka, 51–39 ka, and <39 ka. Species with last appearance dates before human arrival (50–45 ka) are highlighted in green, and those with last appearance dates after this are highlighted in blue. There is little evidence for any extinction pulse associated with human arrival.
The high-resolution EDC record also documents considerable millennial-scale variability. There are four distinct Antarctic Isotopic Maxima between 60 ka and 35 ka, and it is now clear that there was marked aridification from ∼50–45 ka (50, 51), the interval during which the first humans arrived in Sahul.
In short, the temperature swings of the last few glacial cycles were without precedent and the last glacial cycle exhibits unusual variability. The climate patterns in the 100,000 y preceding the LGM were anything but benign. Rather than MIS3 being a mild and perhaps humid interstadial in Sahul (e.g., ref. 13), paleoenvironmental records demonstrate significant climatic inconstancy. For example, it has recently been shown that climatic variability was a key factor in the extinction of the large flightless bird G. newtoni near Lake Eyre by ∼50 ka (51), and the small mammals from Tight Entrance Cave show a marked increase in turnover at this time (10). With an increasing number of reports presenting high-resolution data through critical periods, further correlations between the disappearance of other megafauna and climate change seem likely to emerge.
Further support for a climatically driven process comes from analyses demonstrating that small-bodied species were susceptible to significant impacts resulting from climate change during the middle to late Pleistocene, before human arrival. Bottom-up, ecosystem-changing processes triggered by the disappearance of particular small-bodied species are increasingly being described for North American and Eurasian ecosystems (52, 53). Given trophic interdependencies, ecosystem changes of this kind in small keystone species could have reduced the resilience of larger-bodied species in those same ecosystems.
Paleoenvironmental records have also been used as important proxies for human activity in arguments for a human role in the extinctions. For example, increases in fire activity reconstructed from charcoal analyses of sediment samples recovered from Lynch’s Crater (northeast Queensland) have been assumed to be a proxy for human activities in particular (54). However, other records show increased burning long before people arrived (10). A recent comprehensive overview of charcoal work in Australasia (55) determined that fire histories more strongly correlate with climate than human activities.
Other proxies include dung-fungus (Sporormiella). In a recent study, also located at Lynch’s Crater, declining Sporormiella spores (a proxy for megaherbivores) at ∼41 ka were interpreted as evidence for human-mediated extinction of megafauna (40). Subsequent peaks in Sporormiella were attributed to extant kangaroos. The authors did not explore other equally plausible scenarios. Variation in Sporormiella peaks could as easily be attributed to changes in spore abundance correlated with changes in climate, especially moisture availability. Furthermore, significant contributions to all the Sporormiella peaks, not just those after 41 ka, could have been made by still-extant kangaroos or cassowaries.
Paleontological and isotopic data from middle Pleistocene fossil deposits in southern central Australia (2) have demonstrated arid adaptations in some now-extinct fauna. The implication is that subsequent increased aridity was unlikely to have led to late Pleistocene extinctions (2). A human role was inferred based on the unproven assumption that the extinctions occurred after people arrived in the region at ∼40 ka. However, as already noted, there is no evidence that many of these species survived the middle Pleistocene. Paleoenvironmental data show that the middle to late Pleistocene was particularly arid and variable and, as noted above, that climate deteriorated markedly from ∼50–45 ka (42⇓⇓–45, 48, 50, 51). It is clear that at least some arid-adapted species can be insulated against aridity. They are advantaged in times of water stress by increased mobility and water conservation strategies. However, there is a tradeoff. Large species must exist at lower densities and, of necessity, require greater home ranges to maintain viable populations. Further, adaptation to aridity does not insulate against hyperaridity (5). Even the surviving red kangaroo (M. rufus), superbly adapted as it is to arid conditions, has suffered massive population crashes during relatively benign climate intervals within historic times (23).
Many questions remain. As we work toward a better understanding of the biology and distribution of extinct species, a complex pattern of faunal extinctions across time and Sahul’s diverse landscape is emerging. Tracking individual histories of faunal species (56), although possible in many northern hemisphere contexts, is severely constrained in Sahul by the patchy fossil record (5), yet it is one area that would certainly help to resolve the nature of faunal turnover through time. Advances in our knowledge of paleoclimates are steadily improving our understanding of the complexities of the extinction processes. Mounting evidence increasingly points to climate change as the primary driver of Pleistocene faunal extinctions. Many species of megafauna did not persist into the late Pleistocene, and other subsequent extinctions postdated the PGM but were completed before humans arrived. Importantly, a role for humans in the disappearance of any surviving taxa, although possible, is yet to be demonstrated. Extinct species that persisted until or beyond ∼50–45 ka finally disappeared within the context of a rapidly deteriorating climate.
We thank Jim Allen for incisive and constructive comments on an early draft of the manuscript. This work was funded by the University of New South Wales and the Australian Research Council: Grants DP0666374 and DP0987985 (to S.W.), Grant DP0557923 (to J.T.F.), Grants DE120101533 and DP120101752 (to G.J.P.), and Grants LP100200486 and LP0989969 (to M.A.).
Author contributions: S.W. and J.H.F. designed research; S.W., J.H.F., G.J.P., and J.L. performed research; G.J.P., J.L., J.T.F., and S.D.M. analyzed data; and S.W., J.H.F., M.A., D.K.G., J.T.F., G.E.W., I.D., and S.D.M. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1302698110/-/DCSupplemental.
