Pleistocene Megafaunal Collapse, Novel Plant Communities, and Enhanced Fire Regimes in North America

See allHide authors and affiliations

Science  20 Nov 2009:
Vol. 326, Issue 5956, pp. 1100-1103
DOI: 10.1126/science.1179504


Although the North American megafaunal extinctions and the formation of novel plant communities are well-known features of the last deglaciation, the causal relationships between these phenomena are unclear. Using the dung fungus Sporormiella and other paleoecological proxies from Appleman Lake, Indiana, and several New York sites, we established that the megafaunal decline closely preceded enhanced fire regimes and the development of plant communities that have no modern analogs. The loss of keystone megaherbivores may thus have altered ecosystem structure and function by the release of palatable hardwoods from herbivory pressure and by fuel accumulation. Megafaunal populations collapsed from 14,800 to 13,700 years ago, well before the final extinctions and during the Bølling-Allerød warm period. Human impacts remain plausible, but the decline predates Younger Dryas cooling and the extraterrestrial impact event proposed to have occurred 12,900 years ago.

In North America, Pleistocene-Holocene deglaciation [18 to 6 thousand years ago (ka); 1 ka = 1000 calendar years ago] was marked by massive biotic upheaval, including the extinction of 34 megafaunal genera (1), species migration and reorganization of terrestrial communities (2), the rise and decline of plant communities without modern analogs (3), and increased biomass burning (4). Individualistic plant species’ responses to climate change transformed the composition and distribution of vegetation formations, with rates of change highest between 13 and 10 ka (all ages are reported as calendar years before the present) (2). Many North American fossil pollen assemblages between 17 and 9 ka lack modern analogs, suggesting parent vegetation formations that were compositionally unlike any today (5). In the upper Midwest of the United States, “no-analog” pollen assemblages contained high percentages of temperate broadleaved trees, particularly Fraxinus (ash), Ostrya/Carpinus (hophornbeam/ironwood), and Ulmus (elm), coexisting with boreal conifers such as Picea (spruce) and Larix (larch) (2, 3). These no-analog communities apparently formed in response to higher-than-present insolation and temperature seasonality, but it has been suggested that they may have been linked to the Pleistocene megafauna (3).

Deglaciation and vegetation turnover coincided with the end-Pleistocene megafaunal extinctions in North America, which was part of a global time-transgressive extinction wave that was taxonomically selective and more severe for species of large body size (6, 7). In North America, >50% of all mammal species >32 kg and all species >1000 kg were extirpated (1). Hypothesized extinction drivers include climate change, human hunting, or a combination of the two; the relative importance of these mechanisms is debated (1, 8). An extraterrestrial impact at 12.9 ka has also been proposed (9) but is disputed (4, 10).

The apparent coincidence between megafaunal extinction, peak rates of vegetation change, and the rise of the no-analog communities (3) in eastern North America suggests causal relationships, but the direction of causation remains unclear. Hypothesized extinction mechanisms linked to vegetation change include habitat loss and fragmentation, disruption of coevolved plant and animal communities, and loss of vegetation mosaics (1, 8). Conversely, the removal of keystone megaherbivores might have triggered trophic effects (11). Contemporary exclusion experiments have highlighted megafaunal influences on plant community composition and structure though edaphic disturbance, selective herbivory, and seed dispersal and propagation (12). In African savannas, megaherbivores can suppress fire by reducing fuel loads and facilitating the growth of less-flammable species (13).

Testing causal hypotheses has been hampered by difficulties in establishing the precise temporal sequence of events, due to fossil scarcity and uncertainties in dating and cross-site correlations. Megafaunal remains are rare, particularly in the lakes and mires that archive late Quaternary pollen records, and changes in population densities cannot be inferred from most megafaunal fossil records. Many bone radiocarbon dates are erroneously young because of contamination by humic acids, although dating of purified bone collagen can remove this effect (14). The last-dated fossil for a species is unlikely to be from the last surviving individual.

We precisely established local lead-lag relationships among the no-analog plant communities, changes in fire regimes, and megafaunal population declines through a multiproxy sedimentary study of Sporormiella dung fungal spores, fossil pollen, and charcoal from a lake in mid-continental North America. Sporormiella is a genus of coprophilous fungi in the family Sporormiaceae that requires herbivore digestion to complete its life cycle, producing spores on the dung of mammals and some birds (15). Sporormiella spores have been found in mammoth (Mammuthus spp.) gut contents and coprolites (16, 17). Sporormiella spores track the end-Pleistocene megafaunal population decline: They are abundant in late-glacial sediments, scarce through the Holocene, and return to high abundances after the historic introduction of domestic grazers (16, 18). The spores are transported to lakes by slopewash, so Sporormiella abundances in lake sediments reflect both dung loadings in the watershed and distance to their source (19). Thus, Sporormiella abundances cannot be simply converted to dung volume and herbivore biomass, but Sporormiella abundances <2% of the arboreal pollen sum consistently indicate the extinction of megafauna at sites in Pleistocene North America (16) and in late Holocene Madagascar (20).

Appleman Lake (LaGrange County, Indiana; 41.6237°N, 85.2136°W), a 21-ha kettle pond situated on glacial till and outwash from the Lake Michigan Lobe, is centrally located within the no-analog vegetation formations (Fig. 1). An 11.5-m sediment core was extracted in 2005 (21). Thirteen wood, pollen, and charcoal samples were submitted for radiocarbon analysis (table S1), four anomalous dates were rejected, and a linear age model was constructed from the remaining dates (fig. S1). No-analog communities were identified by plotting the minimum squared chord dissimilarities (SCDs) between Appleman pollen assemblages and their closest matches from the North American Modern Pollen Database; fossil samples with minimum dissimilarities >0.3 were considered to have no modern analog.

Fig. 1

Position of Appleman Lake, Indiana, and New York sites (blue circles) relative to (A) minimum squared chord dissimilarity for fossil pollen assemblages in eastern North America, (B) Picea pollen percentages, and (C) the distributions of Fraxinus and Ostrya/Carpinus pollen, all mapped for 14 ka. The mapped SCDs are based on a smaller list of taxa (25 pollen types) than used elsewhere in this paper; therefore, values >0.15 indicate fossil samples with no modern analog.

At Appleman Lake, Sporormiella was initially abundant, began declining at ~14.8 ka, fell below 2% at 13.7 ka (Fig. 2G), and remained <2% thereafter. The youngest-dated bones of most North American megafauna cluster between 13 and 11.5 ka (22), so the Sporormiella decline apparently indicates local population collapse and functional species extinction, not final extinction. The Sporormiella decline roughly coincided with the initiation of the Bølling-Allerød warm period, dated in Greenland at 14.69 ka (23). No-analog plant communities developed at 13.7 ka (Fig. 2H), after Sporormiella declined below 2%, and were marked by rises in Fraxinus nigra–type (black ash) and Ostrya/Carpinus (hophornbeam/ironwood) pollen abundances while Picea (spruce) abundances remained high (Fig. 2, A, E, and F). No-analog communities persisted until 11.9 ka (Fig. 2H) and were followed by a period of high Pinus (pine) abundances (Fig. 2B), then high abundances of Quercus (oak), marking the early Holocene establishment of deciduous forests in northern Indiana.

Fig. 2

Appleman Lake time series for (A to F) percent pollen abundances of selected taxa (NAP, nonarboreal pollen), (G) Sporormiella, (H) minimum squared chord dissimilarity, and (I) charcoal counts. Pollen abundances are expressed as a percentage of the upland pollen sum.

Fire regimes at Appleman had three stages. Before 14.3 ka, charcoal was uncommon, indicating few or no local fires (Fig. 2I) or insufficient biomass to produce abundant charcoal. After 14.3 ka, charcoal peaks were intermittent against low background rates of charcoal deposition. Charcoal first peaked at 14.1 ka, during the Sporormiella decline and before the increase in hardwood pollen abundances. This stage persisted until 10.7 ka, when rates of charcoal deposition and Quercus pollen abundances increased rapidly.

The close linkage between Sporormiella and changes in vegetation and fire at Appleman Lake is consistent with evidence from several New York sites (Fig. 3) (18). There, the onset of the Sporormiella decline ranged from <13 ka to >14 ka; dating is less certain than at Appleman due to hard-water contamination of bulk sediment radiocarbon dates (24). The New York sites are on the periphery of the late-glacial no-analog communities (Fig. 1) (3), and minimum SCDs are lower than at Appleman Lake (Fig. 3). Nevertheless, as at Appleman, hardwood abundances increased immediately after the Sporormiella decline, contributing to an increase in vegetation dissimilarity, and charcoal abundances peaked during or after the Sporormiella decline (Fig. 3).

Fig. 3

Comparison of the Appleman record with two New York sites reinforces the close connection among the Sporormiella decline (blue histogram), the first large pulse of sedimentary charcoal (red histogram, inverted axis), and increased vegetation dissimilarity from that of the present (black line). Black triangles indicate locations of radiocarbon dates from Appleman.

Our data thus rule out hypotheses that (i) climate-forced changes in vegetation drove the megafaunal decline, and (ii) no-analog plant communities were created by megaherbivory. The first hypothesis is rejected because the Sporormiella decline preceded the major palynological events (particularly the shift from Picea to Pinus and increased abundances of hardwood taxa). However, climate change might have directly forced the megafaunal population declines, given the similar timing between the Sporormiella decline and the onset of Bølling-Allerød warming in Greenland (23). If so, climatic forcing apparently did not operate through habitat change, which is the mechanism underlying most climate-based extinction hypotheses (1).

The second hypothesis is excluded because the no-analog communities arose after Sporormiella declined. Thus, the increase in hardwood taxa (Fraxinus- and Ostrya/Carpinus at Appleman and Ulmus, Acer, and others in New York) may represent both a response to warming and a release from herbivory pressure. Many extant megaherbivores prefer broadleaf forage because of its higher nutrient and water content (12, 25). For example, moose (Alces alces) dietary preferences have reduced tree density and promoted shrubs and needle-leaved trees (26). Tradeoffs between fire and megafaunal consumption of biomass are documented in modern African ecosystems (13). A switch from herbivory- to fire-dominated disturbance regimes may explain why the first post-Sporormiella charcoal peak was consistently the largest across sites (Fig. 3), if this first fire burned both live biomass and litter untouched by herbivores.

The possibility that late-glacial vegetation and fire regimes were influenced by herbivory does not rule out climatic drivers of late Quaternary landscape change, which are well established (27, 28). Rather, we suggest a hierarchy of controls on deglacial vegetation history, with climate driving changes in plant and megafaunal ranges and abundances, which engendered further herbivory- and fire-regulated biotic interactions. The rise in Fraxinus nigra–type and Ostrya/Carpinus pollen ~15.0 ka (Fig. 2, E and F) may indicate the establishment of small tree populations under newly favorable climates, whereas secondary increases at 13.7 ka may indicate the expansion of populations under decreased herbivory pressure and a new fire regime. Thus, the formation of no-analog plant communities may have been jointly controlled by novel climates (highly seasonal insolation and temperatures) (5) and release from herbivory.

Humans may have affected late Quaternary vegetation history by intensifying fire regimes and contributing to the megafaunal extinctions. Increased sedimentary charcoal is associated with human arrival and megafaunal extinctions on islands (20), and the globally time-transgressive wave of late Pleistocene extinctions closely tracks human colonization history. Butchered mammoth bones excavated in southeastern Wisconsin date regional human presence to between 14.8 and 14.1 ka (29), coeval with the Sporormiella decline at Appleman Lake.

At all sites, the Sporormiella decline substantially predated initiation of the Younger Dryas, ruling out abrupt climatic cooling and the hypothesized extraterrestrial impact at 12.9 ka as a cause (9). More generally, the megafaunal declines apparently progressed over several thousand years, given the 1000-year duration of the Sporormiella decline at Appleman and the difference in timing between the onset of the Sporormiella decline (14.8 ka) and the final extinctions (~11.5 ka) (1). This evidence excludes rapid-extinction hypotheses such as an extraterrestrial impact or a Paleo-Indian blitzkrieg (30).

Our work thus shows close connections among the late-glacial histories of fire, vegetation, and mammalian herbivores and suggests that the loss of a broad guild of consumers contributed to substantial restructuring of plant communities and an enhanced fire regime. The sequence of events at Appleman rules out several hypothesized causes and effects of the megafaunal extinction but does not conclusively resolve the debate over climatic versus human causation (or both) of the North American megafaunal extinctions. However, several promising avenues exist. One is to search for spatial and temporal patterns in the late Pleistocene Sporormiella decline (time-transgressive versus synchronous) and to further check its association with vegetation and fire history. Another is to analyze the Sporormiella record at sites spanning the penultimate deglaciation (when humans were absent from North America) and in sites near well-dated records of Paleo-Indian activity. Such analyses should be extended to other continents, to study the ecological effects of the end-Pleistocene extinctions under different contexts of human, climate, and vegetation history (31). By resolving the causes and consequences of the late Pleistocene megafaunal extinctions, such work would address concerns about trophic effects arising from the contemporary widespread declines, extinctions, and restorations of megaherbivores.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Tables S1 and S2


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank E. Grimm, J. Mason, and S. Hotchkiss for discussions and T. Minckley, S. Lucas, J. Marsicek, D. Alhambra, G. Schellinger, and S. Hernandez for field and laboratory assistance. Initial core analyses were performed at the National Lacustrine Core Facility at the University of Minnesota. This work was supported by NSF (grants DEB-0716471 and DEB-0716951) and the Graduate School and the Climate, People and Environment Program at the University of Wisconsin.
View Abstract

Stay Connected to Science

Navigate This Article