Frozen Fauna of the Mammoth Steppe: The Story of Blue Babe

by Guthrie, R. Dale

  First, however, let me explain why this question and our shared efforts to understand Pleistocene environments in the far north are important. The area in question is immense, truly circumpolar, but this debate about late Pleistocene events is not simply a question of how to paint a large panel of stage scenery. Our reconstructions of this past critically shape our present ecological understanding. The pulls and tugs of ideas in this debate involve fundamental questions in ecological theory, for example, how species and communities react to different climatic changes, as well as more specific issues, such as the ecology of Blue Babe. Finally, the character of the late Pleistocene north is closely linked to the question of when early peoples colonized the New World.

  Palynologists have argued against the term and concept of a Mammoth Steppe primarily on the basis of the meager pollen influx (the comparative numbers of pollen in each cubic centimeter of sediment) during peak glacials. They concluded from these small amounts of pollen that Alaska was a barren landscape, or polar desert, uninhabitable or nearly so by the “mammoth fauna” we find as fossils. Rather, they propose that these animals must have been in Alaska mainly during the wetter and warmer interstades and inter-glacials.

  Ritchie, Cwynar, and Colinvaux state or imply several propositions which I review in detail. I think sufficient data exist to test them. The propositions are as follows:

  A. The amount of pollen produced (the pollen influx in lake-bottom cores) was so low during the glacials that it can only be matched by a present-day high Arctic polar desert or barren tundra.

  B. Thus, the glacial environment, as reconstructed from pollen data, was so harsh it could not possibly have supported the “mammoth fauna”; this is said to be borne out by the lack of dated large mammal fossils from full glacial times.

  C. The taxa of glacial (Duvany Yar) pollen show that species from southern steppes were not present; glacial taxa were restricted to tundra species found throughout the north today. Thus, glacial vegetation was not similar to a steppe but was barren tundra like that now existing in the very far north.

  After examining these propositions I review arguments supporting the existence of an arid grassy steppe during both the glacials and their interstades (Blue Babe’s time). My argument for this steppe takes the following outline:

  1. Contrary to some palynologists’ contentions, pollen cores that meet minimum acceptable standards for influx analysis do show herb pollen production peaking during the glacial maximum (Duvany Yar) and declining dramatically from the glacial maximum to late Holocene. Herb pollen influx during the peak Duvany Yar is not definitive of barren tundra but is in fact comparable to that of some arid grasslands today.

  2. There are sufficient numbers of dated fossils to show that the dominant members of the “mammoth fauna” (bison, horse, and mammoth) were present continuously in the far north during the late Pleistocene (at least to radiocarbon-dating limits), a span of time covering part of the last interstadial (Boutillier Interval), the full glacial peak (Duvany Yar), and the late glacial (Birch Period). Additionally, most of the less common Beringian species in the fossil record have radiocarbon dates associating them with the full glacial.

  3. There are, in the Pleistocene fossil record, flora and fauna that we associate with more xeric environments farther south. These taxa are now generally absent in the area once occupied by the Mammoth Steppe. Those that do occur in the far north are now found in rare, special habitats.

  4. The paleoenvironments of Alaska and Beringia should not be analyzed totally separately from those of northern Eurasia. Pleistocene large mammal communities in Alaska and the Yukon Territory were the easternmost part of mammalian communities, extending across Eurasia to the shores of the Atlantic. Evidence from that faunal complex suggests a generally similar cold and dry environment where graminoids predominated, although with locally diverse facies.

  5. The mammals (particularly large mammals) taken individually and as a community were grazing specialists.

  6. Teeth and stomach contents of frozen mummies of these same species show they were indeed mainly grazers.

  7. Details of foot morphology indicate they were adapted to a firm substrate, quite unlike the moist and yielding soils over much of the region today, apart from mountainous terrain.

  8. Their body size indicates that they were not on marginal tundra or polar desert summer range but were able to acquire large quantities of quality forage during the summer growth season.

  9. There is a continuity and coherence to the Mammoth Steppe fauna. The fossils of most members of this large-mammal grazing community are ubiquitous, distributed throughout the unglaciated north from high areas adjacent to alpine glaciers to the broad lowlands, quite distant from hill country.

  10. The inability of most of these grazing large mammals to tolerate present snow depths means that, due to low snowfall, wind redistribution of snow, or both, they had widespread access to adequate winter range. Snow depths are now limiting to all northern large mammals; a low snow cover would have enabled a larger diversity and density (standing biomass) of large mammals than exists today.

  I shall include the first three theses above with my discussion of the palynological propositions A, B, and C because they are related. The remaining points are then discussed under their appropriate headings.

  Proposition A

  The amount of pollen produced (the pollen influx in lake bottom cores) is so low during the glacials that it can only be matched by present-day high Arctic polar desert.

  Ecological inferences from mammals points unwaveringly toward a Pleistocene grassland environment throughout the north and yet the pollen evidence is in apparent disaccord. The problem lies not with the fauna, but with certain interpretations of pollen evidence.

  One fundamental difficulty is that pollen can be well identified among arboreal species, but identification is very poor for monocots. In fact, graminoids can only be separated into grasses and sedges with no greater taxonomic refinement. This has produced an understandable preoccupation among palynologists with arboreal species; I would argue, however, that this orientation has led to a misinterpretation of Beringian pollen, exacerbated by inappropriate notions about biome integrity through time.

  Palynologists have asked how one can propose the presence of productive grassland during the full glacials when today’s unproductive northern landscape produces much more pollen. But this question, which fails to discriminate between arboreal and nonarboreal pollen, reflects a misreading of their own data. The more accurate question is why several times more herb pollen was produced during the full glacials than today.

  Simply, I propose that pollen data do not support certain conclusions regarding pollen influx during the glacials. Pollen influx is rather abundant in the Holocene sections of pollen cores (in the thousands per cm3) and sparse (in the hundreds) during the peak of the last glacial, 17,000–22,000 years ago. The relative scarcity of pollen during the last peak glacial (Duvany Yar) is dramatic. But if we compare Holocene and Duvany Yar pollen influx with two necessary corrections—(1) excluding from our analysis two pollen cores that exhibit dramatically unequal sedimentation rates, and (2) looking at pollen influx of herbs only, particularly graminoids and sage, Artemesia—then the pollen record offers surprisingly strong support for the concept of a Mammoth Steppe.

  Two pollen profiles constitute the main evidence on which Ritchie, Cywnar, and Colinvaux base their objections to a Pleistocene Beringian vegetation that could support a complex community of large mammals during the glacials. One is from Hanging Lake, in northern Yukon Territory (Cwynar 1982) and the other is from St. Paul Island, one of the Pribilofs in the Bering Sea (Colinvaux 1981). Profiles of pollen influx are most credible when the sediments are similar and sedimentation rates are comparable throughout the core. Both cores get poor marks in each category. The Hanging Lake core has a long Holocene section, but the earlier 13,000–30,000 years are crammed into a short lower segment. The sharp break in deposition rate oc
curs at about 270 cm, dated at 12,800 + − 320 (GSC 2,846). Below that point there is a little more than half a meter of core; immediately beneath that point Cwynar begins to identify significant amounts of Tertiary pollen, presumably weathered from bedrock. There are also bedded aquatic peat zones beneath that break. Organic content is virtually nonexistent, and it is almost sterile of pollen. These characters suggest that during the Pleistocene the pond was dry and thus experienced mostly aerobic conditions or, at most, held intermittent annual or seasonal standing water during glacial times. Whatever its degree of sedimentary continuity, the Hanging Lake site does not seem an optimum core to use for a general model of pre-Holocene vegetation reconstruction in Beringia. No other core in the north shows such a dramatic discrepancy in herb influx numbers during Holocene and peak glacial, with one exception—the core taken from St. Paul Island (Colinvaux 1964).

  The deep core from St. Paul Island (Cagaloq Lake) also has a well-dated Holocene section down to about 10,000 years ago. But farther down, the core is mainly sand, suggesting, as Colinvaux proposes, that a permanent pond was only established at the beginning of the Holocene, with the onset of the oceanic climate of the flooded land-bridge. Like the Hanging Lake core, most of the St. Paul core is Holocene; the condensed earlier portion of sand has virtually no pollen, as one would predict from aerobically exposed, wind distributed sand. Although there are intermittent layers of silt in the sand it is not possible to determine influx in these bands because the sedimentary rates during these times are unknown. But we do not have to depend on these two cores for paleoreconstruction. There are pollen profiles from other Alaskan lakes which better meet sedimentary constraints for pollen influx studies.

  Before turning to these other pollen profiles, let us discuss the lumping of arboreal and nonarboreal pollen to produce total pollen concentrations or influx. One characteristic of northern trees is their ability to produce large volumes of pollen. Hiking through an alder thicket in the early spring, one can literally turn golden with pollen dust. I would argue that to compare total pollen influx of Holocene forests with glacial-aged (Duvany Yar) herb communities is not very meaningful. Ostensibly, influx studies allow one to separate chronological patterns of different taxa; this cannot be done in percentage diagrams. Thus, I use these influx studies where arboreal and nonarboreal pollen influx are listed separately, and set aside the arboreal pollen.

  Most herbs in the north are not big pollen producers. Many herb families that produce significant quantities of pollen at temperate latitudes are rare or minor elements in northern vegetation. These include ragweeds, Ambrosieae, goosefoots, Chenopodiaceae, and composites (except for Artemesia), Compositae. Furthermore, the ability of graminoids to shift from sexual to vegetative reproduction, in response to grazing pressure, means grasses can be silent in the pollen record.

  Despite these problems, influx profiles, within taxa, allow us to compare herb pollen from glacial times with that of the late Holocene. We know there is a large herb biomass in many parts of Alaska today. In fact, herbs probably constitute as much of the ground cover as trees, if we count understory cover, muskegs, lowland tundra, alpine tundra, and so forth. Thus we must ask if there was more or less herb biomass during the glacials. Comparing pollen accumulation in that way we could formulate a more meaningful influx comparison of herb biomass between today’s herb vegetation and the unknown herb community of the peak glacial.

  In figures 9.1–9.4 I have compared the four published cores extending back beyond 24,000 B.P. and the beginning of the Duvanny Yar Interval, the last stage of the last glacial (isotope stage 2), in which influx data were calculated. Other published cores with influx data either do not go beyond the Birch period (14,000) or go only sightly beyond it at the very bottom of the core. Judging from these cores one might conclude that there were few permanent lakes during the full glacials.

  Following pollen influx down the core, the amount of herb pollen increases by a factor of three once past the Holocene and into the glacial. If the implicit assumption that pollen influx correlates directly and positively with volume of herb vegetation were correct, then one might conclude that full glacial herbaceous vegetation was much greater in volume than that of today.

  Fig. 9.1. Imuruk Lake herbaceous pollen. Colinvaux’s (1964) deep core from Imuruk Lake was the first core to reach into the interstade (Boutellier Interval). I have used only the portion of the core that is within the radiocarbon range and have calculated influx from his pollen concentration figures. The greatest influx of herb pollen is during the Duvanny Yar Interval, or the last glacial maximum. This contradicts the idea that pollen influx was almost nonexistent during full glacial as it is in a polar desert.

  Dating in these three cores is relatively good, although there are a few inverted radiocarbon dates. Fortunately, we know enough about patterns of postglacial arboreal recolonization to use that as a cross-check on the carbon dates. All three cores show a taxonomic profile consistent with the well-dated chronology of other Holocene and late glacial-aged cores. Squirrel River (Anderson 1982, 1985) dates back beyond 23,000, and the base of the Kaiyak Lake core is beyond the limits of radiocarbon range (Anderson 1982, 1985); Joe Lake is at 28,000 (Anderson 1988), while Colinvaux’s (1964) Imuruk Lake core extends back at least to the last interglacial (but I have only used the portion pertaining to the late glacial and Holocene). Dates in this core are unclear. I have assumed, as did Colinvaux, that the penultimate mode of arboreal pollen represents the last interstadial or Boutillier (oxygen isotope stage 3). Colinvaux gave percentage and total pollen concentration, so I combined his figures to calculate the influx of individual taxa and herbs (120 cm per 15,000 years for the full glacial; the top of the core represents wave-disturbed sediments). In all cases, I used the author’s chronological assessments. All three cores were taken from the central part of Beringia, with Anderson’s cores closer to the Brooks Range. We can assume that they represent a general case for much of Beringia.

  The picture presented by these three cores is fairly consistent. They show a pattern of relatively high herb influx during peak glacial, as compared to late Holocene segments. Grass pollen reaches its peak during full glacial, contrary to the conclusions of Ritchie and Cwynar (1982). Artemisia, or sage, also peaks during the full glacial; the same is true for the nongraminoid herbs. Influx of sedge pollen is less consistent. Sedge pollen is slightly more abundant during the wetter late glacial and Holocene segments of Anderson’s two pollen cores (sedge forms a common vegetation on the margins of lakes in Alaska today, although a number of species are found only in dry habitats). Herb pollen is least in all three cores during the late Holocene, despite the fact that herbaceous vegetation comprises a large part of the flora in those areas today. One can only conclude that today, and in the late Holocene, herbs produced relatively little pollen, compared to the full glacial.

  Fig. 9.2. Squirrel Lake herbaceous pollen. Three of Anderson’s deep cores show a pattern similar to Colinvaux’s core at Imuruk Lake. Squirrel Lake (Anderson 1982) shows almost twice the herb pollen influx during the full glacial (Duvanny Yar) as in the late Holocene.

  Fig. 9.3. Kaiyak Lake herbaceous pollen. The second of Anderson’s (1982) cores, at Kaiyak Lake, exhibits a herb pollen pattern in the Duvanny Yar which is similar to the core from Squirrel Lake.

  Pollen assemblage still does not tell which species of grass and sedge were present, but one can propose (on the basis of southern steppe grass phytoliths, discussed later in this chapter) that increased aridity during the glacials would have caused different composition and distribution of herb vegetation than one finds today.

  The Squirrel River, Kaiyak Lake, Joe Lake, and Imuruk Lake cores, which all reach past the most recent peak glacial, exhibit relatively large herb pollen influx during the early Holocene. In this, the deep cores are similar to more time-shallow cores (Cwynar 1982; Ager 1975; Ritchie 1984; Brubaker, Garfinkel, and Edwards 1983), but the latter do not reach beyond the sawtooth swings of herb pol
len that precede the higher influx during the full glacial.

  One thing is clear from these cores: herb pollen influx during full glacial times, as compared to the Holocene, was not reduced as Colinvaux, Ritchie, and Cwynar have proposed; rather, several times more herb pollen was produced during peak glacials than in the late Holocene sections of the cores. This allows us to reject the first part of the palynologists’ Proposition A.

  The comparatively low herb pollen influx during the late Holocene and the last interstadial may be due to the increase of sites with humic soils. As I discussed in the last chapter, the acidity of the humic mat decreases available nutrients and its mulch insulates the soil, decreasing summer thaw and, in turn, reducing access to soil nutrients. Reduced nutrient and mineral access would produce more conservative reproductive effort (Grime 1979). Contrary to one’s intuitive sense, the warm and wet conditions of the interstades-interglacials may have increased standing biomass but may not have greatly increased productivity, especially that useful to large mammals.

  Now let us look at the second part of Proposition A: the argument that low annual pollen influx rates during the peak glacial indicate barren tundra, or polar desert, when compared with modern vegetation. From diagrams in figures 9.1–9.4, one can see that annual influx rates are about 400 grains per cm2 per year during full glacial from central Beringia. Ritchie and Cwynar (1982) compare these low productions with a grassland farther south. They chose a previously published study of five different sites from southern Manitoba. Unfortunately, southern Manitoba is not the best choice for a comparison because it is not thinly vegetated, shortgrass country as one might find near, say, Medicine Hat, Alberta. Southern Manitoba is characterized by tall and mixed grasslands extending northward from the midwest United States. Biogeographic maps even show big bluestem (Andropogon gerardii) to be one of the native dominants in parts of southern Manitoba. Big bluestem is a signature species of tallgrass prairie which once occupied the Iowa-Illinois corn belt.