The Aftermath of Megafaunal Extinction: Ecosystem Transformation in Pleistocene Australia

See allHide authors and affiliations

Science  23 Mar 2012:
Vol. 335, Issue 6075, pp. 1483-1486
DOI: 10.1126/science.1214261

This article has a correction. Please see:


Giant vertebrates dominated many Pleistocene ecosystems. Many were herbivores, and their sudden extinction in prehistory could have had large ecological impacts. We used a high-resolution 130,000-year environmental record to help resolve the cause and reconstruct the ecological consequences of extinction of Australia’s megafauna. Our results suggest that human arrival rather than climate caused megafaunal extinction, which then triggered replacement of mixed rainforest by sclerophyll vegetation through a combination of direct effects on vegetation of relaxed herbivore pressure and increased fire in the landscape. This ecosystem shift was as large as any effect of climate change over the last glacial cycle, and indicates the magnitude of changes that may have followed megafaunal extinction elsewhere in the world.

The disappearance of megafauna from most of the world’s ecosystems was a major event in recent Earth history. Most research into these extinctions has focused on possible causes, which include climate and human colonization (1), and less attention has been paid to the consequences of the extinctions (2). Large herbivores have strong effects on ecosystems, by maintaining vegetation openness and patchiness, removing material that would otherwise fuel landscape fire, dispersing seeds, and physically disturbing soil and recycling nutrients (3). Therefore, megafaunal extinction might have caused major changes to vegetation and the functioning of ecosystems. We know little about this because we lack detailed ecological reconstructions able to distinguish the effects of large-herbivore extinctions from environmental changes that could have caused them. Recent studies from North America show that megafaunal decline was followed by vegetation change and increased fire (4, 5). However, these events happened in the latest Pleistocene during a time of rapid climate change, so it is difficult to resolve the contributions to them of megafaunal extinction versus climate.

Australia’s megafauna included 20 or more genera of giant marsupials, monotremes, birds, and reptiles (6), which were extinct by 40 thousand years ago (ka) (710), soon after people colonized Australia (11), suggesting that people caused the extinction. Proposed mechanisms are overhunting (6, 12), vegetation change due to landscape burning by people (9), or a combination thereof. Indications of increased fire and changed vegetation around the time of human arrival support a role for landscape burning (9, 13), but it could be that fire increased as a consequence of the loss of large browsers and grazers, and fire then caused vegetation change (14). An alternative view is that Australia’s megafauna was in long-term decline because of climate drying, and the human contribution to its ultimate disappearance was small (15).

We resolved relations among megafaunal extinction, fire, climate, and vegetation at the Lynch’s Crater palaeolake/swamp in northeast Australia. Previous work at this site has documented an increase in charcoal and a shift from rainforest to sclerophyll vegetation beginning ~45 ka that is generally interpreted as a result of landscape burning by people (13, 16). There is no archaeological record at the site, but regional archeology confirms that occupation may have begun as early as 49 ka and was widespread by 40 ka (1719). We used pollen and charcoal to reconstruct vegetation, fire, and climate changes and spores of the fungus Sporormiella to indicate large-herbivore activity (20). Sporormiella depends on ingestion by herbivores to complete its life cycle; it sporulates in their dung. Sporormiella can be found in dung of herbivores across a wide range of body sizes, but spore counts are most strongly affected by activity of large herbivores (21, 22) and thus provide a proxy for large-herbivore biomass (4, 5, 2124).

We analyzed two cores, a long core (“core 1”) spanning 130 to 24 ka (Fig. 1A) and another ranging from ~54 to 3 ka (“core 2”), which was sampled at ~100-year intervals between 54 and 24 ka (Fig. 1B). Core 1 encompasses two major climate transitions during the 80 thousand years (ky) preceding human arrival: (i) a shift from warm and humid to cool and dry conditions around 120 ka (termination of the last interglacial) and (ii) a further cooling and drying ~75 ka [onset of marine isotope stage 4 (MIS 4)]. Both events are marked by stepwise decline of rainforest angiosperms in favor of sclerophyll taxa and rainforest gymnosperms. Between 60 and 55 ka, high representation of rainforest gymnosperms indicates a return to somewhat warmer and more humid conditions, and after 55 ka, an increase in mid-storey plants and swamp/aquatic vegetation suggests a change from shallow lake to peat-swamp conditions. Most likely, the landscape at that time included a mix of vegetation types, with patches of grassland and sedges interspersed with mixed forest and woodland. High Sporormiella counts suggest that large herbivores ranged very close to or over the swamp surface.

Fig. 1

Pollen, charcoal and Sporormiella diagrams for Lynch’s Crater, showing the (A) core 1 and (B) core 2 records. The interval during which Sporormiella declined and charcoal first increased is shaded gray and aligned on the two cores.

These conditions ended around 41 ka, when Sporormiella counts dropped almost to zero. The timing of this decline agrees closely with estimates of the date of megafaunal extinction from fossil evidence elsewhere in Australia (7, 10). At Lynch’s Crater, it was associated with increased charcoal and grasses with a further rise in sclerophyll vegetation at the expense of rainforest angiosperms and (especially) gymnosperms. Unlike the two previous advances of sclerophyll vegetation, this shift has no climatic explanation. It preceded climate drying at the MIS 3/2 boundary by ~10 ky, and although sediment moisture conditions at Lynch’s Crater show millennial scale fluctuations during the interval from 45 to ~25 ka (25), there is no discernible relation to the trend in Sporormiella. The transformation appears in both cores, with overlapping radiocarbon dates, and can be used to map core 2 onto core 1 (Fig. 1).

We analyzed this transformation in detail by focusing on the interval from 694 to 604 cm in core 2, which has an estimated age range from ~43 to 38 ka and encompasses the Sporormiella decline and charcoal rise (Fig. 2A). Sporormiella decline and charcoal increase were negatively correlated through this interval [correlation coefficient r = –0.54; 95% confidence interval (CI): –0.72 to –0.30], but neither was directly correlated with vegetation change [correlations with percent sclerophyll were 0.10 (95% CI: –0.20 to 0.38) for Sporormiella and 0.05 (95% CI: –0.33 to 0.45) for charcoal]. The reason that vegetation change was not directly correlated with Sporormiella and charcoal was that shifts in vegetation lagged behind changes in Sporormiella and charcoal (see below). This rules out fire as a cause of megafaunal extinction, because that hypothesis depends on fire causing vegetation change, which would then cause megafaunal decline. In that case, vegetation change should have led megafaunal decline, rather than following it.

Fig. 2

Changes in Sporormiella, charcoal, and vegetation during the interval in the period around 41 ka, in core 2. (A) Pollen, charcoal, and Sporormiella counts, measured as percent of the pollen sum, except for charcoal (particles per cubic centimeter) and total pollen influx from forest trees (grains per square centimeter per year). (B) Moving regression analysis, showing the coefficient of the regression of each variable on time, recalculated for each point through the series. Dashed arrows mark the onset of the largest changes in each variable.

The vegetation reorganization consisted of a succession of changes that unfolded over a range of time scales. There was a rapid rise in grass and a more gradual increase in sclerophyll plants as a proportion of the forest vegetation; these could represent phenological and successional responses in those plant groups to reduced herbivory or increased fire. Later, there was a large increase in total pollen influx from forest trees, which we interpret as a longer-term development of uniform sclerophyll forest (with a grassy understorey) in place of a previously more patchy forest structure (Fig. 2A). To clarify the timing of these events, we calculated a moving regression of each variable on depth, using nine depth intervals as the frame for each regression (equivalent to ~1000 years, according to our age model; see the supporting online material). This frame was shifted in steps of one depth interval (just over 100 years) for each successive regression calculation. Peaks and troughs in the plot of regression coefficients mark the onset of the steepest millennium-scale rises and falls in each variable (Fig. 2B). This analysis showed that the major increase of charcoal lagged Sporormiella decline by ~100 years, grass followed Sporormiella decline by ~300 years, the rise in sclerophyll vegetation lagged Sporormiella decline by ~400 years, and total pollen influx from forest trees increased after 1600 years.

The fire increase that followed megafaunal decline could have been anthropogenic, but the extended trajectory of the rise in charcoal and its close matching with falling Sporormiella suggest instead that relaxation of herbivory directly caused increased fire, presumably by allowing the accumulation of fine fuel. The subsequent vegetation transformation could be explained in two ways: (i) by direct effects of relaxed herbivore pressure on vegetation density and composition or (ii) by release of fire as an ecological force, causing destruction of fire-sensitive rainforest vegetation with replacement by fire-tolerant sclerophyll taxa and grasses. We compared the importance of these two mechanisms by measuring the effects of Sporormiella and charcoal in linear models predicting changes in percent sclerophyll (lagged by four depth intervals, or ~400 years) over the interval shown in Fig. 2. We controlled temporal autocorrelation by fitting generalized least-squares models with an exponential correlation structure (26). The standardized regression coefficients (SRCs = coeffcients/SE) were –2.81 for Sporormiella and 2.53 for charcoal when fitted as single-term models; in a two-term model, the respective SRCs were –1.42 and 1.95. The changes in SRC values suggest that (i) there were independent contributions of both falling Sporormiella and rising charcoal to the subsequent rise in sclerophyll vegetation and (ii) the effect of charcoal was ~35% stronger.

After its initial rise charcoal remained high, and around 29 to 31 ka there was a large increase in macrocharcoal indicating that, for the first time in its history, the swamp itself was extensively burnt (20); short-lived spikes in Sporormiella associated with this burning probably represent grazing over the swamp bed by extant herbivores (probably kangaroos). Charcoal rose further in the Holocene, in complete contrast to the absence of fire in the previous interglacial.

Finally, we compared the magnitude of the ecological changes that followed megafaunal decline around 41 ka with earlier climate-driven shifts from 74 and 120 ka, by calculating standardized estimates of the sizes of effects of each event on Sporormiella, charcoal, and percent sclerophyll (Fig. 3). There was no significant effect on Sporormiella from the two episodes of climate drying, suggesting that the megafaunal extinction was not the culmination of a long-term decline driven by an increasingly arid climate. Had that been true, the Lynch’s Crater record should have shown evidence of declines of megafaunal biomass at times when the climate of the region became substantially more arid. Instead, megafaunal biomass was insensitive to episodes of climate drying, before declining abruptly during a period of stable climate. The increase in charcoal counts and the compositional shift to sclerophyll vegetation that followed megafaunal extinction were as large or larger than changes in the same directions associated with the two major climate changes in the earlier part of the last glacial cycle.

Fig. 3

Magnitudes of change in Sporormiella and charcoal counts and vegetation composition associated with two major climate shifts and megafaunal extinction. Changes in Sporormiella and charcoal are expressed as ratios of mean abundance during the 10 ky preceding each change to the 10 ky following it, converted to natural logs. Zero indicates no change. Vegetation changes are indicated by differences in percent sclerophyll for the 10 ky after and before each event. Confidence limits (95%) were derived by bootstrapping.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Table S1

References (2739)

References and Notes

  1. See the supporting material available on Science Online.
  2. Acknowledgments: This work was supported by the Australian Research Council, the National Geographic Society, Monash Univ., The Royal Society, Australian Institute of Nuclear Science and Engineering, and Natural Environment Research Council. We thank D. Bowman and E. Cameron for comments on the manuscript. The study was initiated by C.N.J. and B.W.B. and was designed by all authors. S.R. was responsible for data collection; B.W.B., C.N.J., S.R., and A.P.K. analyzed the data; C.S.M.T. developed the age model; C.N.J. led the writing; and all authors contributed to interpretation of results. The authors declare no competing interests. Original data will be provided by the corresponding author on request.
View Abstract

Stay Connected to Science

Navigate This Article