Mountain uplift explains differences in Palaeogene patterns of mammalian evolution and extinction between North America and Europe

Abstract

Patterns of late Palaeogene mammalian evolution appear to be very different between Eurasia and North America. Around the Eocene–Oligocene (EO) transition global temperatures in the Northern Hemisphere plummet: following this, European mammal faunas undergo a profound extinction event (the Grande Coupure), while in North America they appear to pass through this temperature event unscathed. Here, we investigate the role of surface uplift to environmental change and mammalian evolution through the Palaeogene (66–23 Ma). Palaeogene regional surface uplift in North America caused large-scale reorganization of precipitation patterns, particularly in the continental interior, in accord with our combined stable isotope and ecometric data. Changes in mammalian faunas reflect that these were dry and high-elevation palaeoenvironments. The scenario of Middle to Late Eocene (50–37 Ma) surface uplift, together with decreasing precipitation in higher-altitude regions of western North America, explains the enigma of the apparent lack of the large-scale mammal faunal change around the EO transition that characterized western Europe. We suggest that North American mammalian faunas were already pre-adapted to cooler and drier conditions preceding the EO boundary, resulting from the effects of a protracted history of surface uplift.

1. Introduction

Multiple studies have investigated the Palaeogene (66–23 Ma) evolution of mammalian faunas (e.g. [1–10]). Usually, these continental- or global-scale investigations employ evidence from long-term marine climate records (e.g. [11,12]) to explain patterns and trends in terrestrial faunal data, or to identify periods of rapid and profound terrestrial climate change. However, terrestrial studies that cover greater spatial scales and that integrate both faunal response and geologic data are not common. For example, the Eocene–Oligocene (EO) boundary at 33.9 Ma represents one of the most dramatic intervals of Cenozoic climatic change (e.g. [11]). However, there are relatively few terrestrial EO records from the mid-to-high latitudes, and terrestrial palaeoclimate data indicate spatially distinct climate responses (e.g. [13] and references therein).

Here, we use fossil data for large herbivorous mammals to spatially analyse the EO (56–23 Ma) development of habitats and palaeoenvironments in western North America. We complement the fossil record with a compilation of pedogenic and lacustrine oxygen isotope (δ¹⁸O) data for the Eocene and Oligocene of western North America [14,15] that track the long-term precipitation patterns of western North America. We then combine δ¹⁸O-based reconstructions of regional precipitation patterns with ecometric information from herbivorous mammals using a proxy method to estimate past precipitation [16]. Although we show the precipitation estimates in mm, we treat these in relative sense in this paper, for comparison between or within maps. We compare our results to published mammalian and terrestrial palaeoclimate data from Europe to investigate the effects of surface uplift on western North American palaeoclimate and how the resulting mammalian diversity pattern differs from that of European counterparts.

We postulate that Eocene regional surface uplift in western North America, starting at latest between 46 and 37 Ma, enhanced rather cool and (sub-)arid conditions in the continental interior, thus pre-adapting the herbivorous mammalian fauna to the climatic regimes of the later Oligocene. (Note that by ‘pre-adapting’ the fauna, we are not implying uniform change in all lineages; rather community changes by a combination of changes within lineages, extinction and the radiation of new taxa, as later discussed.) We provide data that palaeoenvironmental changes to which mammalian communities adapted in the Late Eocene–Early Oligocene were more protracted in North America and took place earlier than in Europe. This explains the apparent lack of response of mammalian faunas at the EO transition in North America: the changes that took place at that point in time in Europe had already happened in North America.

(a). Environmental and faunal context

Global temperatures were typically high during the Early Eocene (56–50 Ma; e.g. [12]), with moderate equator–pole temperature gradients. Tropical evergreen floral systems dominated the Northern Hemisphere (e.g. [5,17–19]). Precipitation levels were also high. Quantitative precipitation estimates are typically scarce, and in North America concentrate on foreland basins in Montana, Wyoming and Colorado. In Wyoming, the mean annual precipitation (MAP) estimates exceeded 1200 mm yr^–1 [18] (based on [20]), and climate modelling suggests that there were already east–west precipitation gradients developing [21]. This time interval, 56–50 Ma, also marks the highest temperatures of the Cenozoic, culminating in the Early Eocene Climatic Optimum (EECO; approx. 53–50 Ma). At the start of the Middle Eocene, the global mean temperature started to decrease (e.g. [12]). In North America, estimated MAP (based on leaf assemblage data) plummeted to around 750 mm yr^–1 [17,20–22] in Wyoming soon after 50 Ma, with the widespread ‘Bridgerian Crash’ in the mammalian faunas [18] from 50 to 47 Ma.

Since the detailed faunal history of North American Palaeogene has been summarized elsewhere (e.g. [7]), we only provide a brief overview of the main trends, and discuss these in relation to our results later.

The onset of the Eocene (56 Ma) marked the first appearance of many modern mammalian orders, such as the perissodactyl (odd-toed) and artiodactyl (even-toed) ungulates, and the true primates (euprimates). The ungulates are of particular interest in the determination of palaeoenvironmental conditions, and the Early Eocene forms were mainly small and generalized omnivores or herbivores. During the Early Eocene artiodactyls and perissodactyls diversified rapidly, and forms with dentitions specialized for folivory (i.e. leaf eating; primates as well as ungulates) appeared. The declining temperatures through the Palaeogene resulted in the extinction of smaller, generalized omnivorous forms among all ungulate groups. During the Middle Eocene (approx. 47–37 Ma), the generic diversity of modern types of ungulate genera increased to an Eocene maximum, while most of the archaic forms (i.e. ‘condylarths’) became extinct. The dental morphology of the Late Eocene ungulates shows a general shift from omnivory to folivory. The first artiodactyls with high-crowned teeth (teeth specialized to withstand an abrasive diet) appeared during the Oligocene. By the Late Oligocene, the mammalian diversity had declined to Early Eocene levels, but many of the taxa were now much larger (e.g. [3,17,18]). Changes were also apparent in mammalian dentition, indicative of changes in diet (and hence in environmental condition): by the Late Eocene (approx. 37–33 Ma) most ungulates had teeth where the originally blunt cusps of the bunodont omnivores now extended into cutting ridges (i.e. lophs), indicative of greater specialization for folivory. In perissodactyls, this was achieved through lophodont molars, and in ruminant and camelid artiodactyls with selenodont molar forms (see [3,23] and electronic supplementary material for more details).

2. Material and methods

The North American large herbivore mammal data derive from published fossil mammal compilations [23,24]. ‘Large’ mammals are those that exceed around 1–3 kg in body mass (the median body size of extant mammals; e.g. [25]): here we base this decision on taxonomic affinity, including the ungulate (or ungulate-like) orders Artiodactyla, Condylarthra†, Dinocerata†, Pantodonta†, Perissodactyla, Taeniodonta† and Tillodontia†, but excluding Rodentia and Lagomorpha. For each taxon, we scored the molar crown height (hypsodonty [26,27]) and the number of longitudinal cutting edges in antero-posterior direction (lophs) on molar crowns [2,16]. Hypsodonty categories are based on the ratio of height-to-length of the second molar (upper or lower). Brachydont teeth (i.e. category 1) have a ratio of less than 0.8, mesodont teeth (category 2) range from 0.8 to 1.2, and hypsodont teeth (category 3) are more than 1.2. Loph count scores vary from 0 to 3. We refer the reader to the electronic supplementary material for more details.

We used only those localities for which we could score at least two values for each metric. The entire dataset contains 357 localities and 4820 taxon occurrences. The entire Tertiary Mammals of North America dataset [23,24] will be made publicly available through the NOW database (http://www.helsinki.fi/science/now) in 2015.

(a). Precipitation estimates

To estimate past precipitation levels, we used a method based on present-day data from large herbivorous mammal communities and their relationship to MAP [16]. This method uses longitudinal loph count and hypsodonty to estimate MAP using the empirically determined relationship

where HYP is the mean hypsodonty of the mammal assemblage, LOP is the mean longitudinal loph count of the mammal assemblage and MAP is in millimetre per annum. The s.e. is ±412 mm, which is similar to other palaeoprecipitation proxy methods [27,28]. Despite the large uncertainties in estimating MAP, this study relies on relative differences in MAP through time, thus rendering our approach sensitive to changes in regional rainfall patterns over time. Our estimates fall in line with previous results (e.g. [17,18,20,21]), but we do not want to draw attention to exact values due to large standard errors and other sources of uncertainty. In the following text, we discuss the results in relative terms (humid, less humid, arid, etc.) because we are mapping out the development of environments from one end-member (tropical humid) to the other (temperate seasonal).

(b). Spatial analysis

We subdivided our data into time intervals following the North American Land Mammal (NALMA) ages (based on [29]) covering the time interval from 56 to 23 Ma. We plotted the mammal assemblage data using MapInfo v. 11.5. In figure 1, we show three distinct phases of the western North American uplift following the SWEEP (Southward Encroachment of an Eocene North American Plateau) model [14,15], and the modern US state names to help the orientation. The SWEEP model posits that surface uplift in western North America was largely diachronous, starting prior to 50 Ma in British Columbia and Montana, and progressing south over time. This north-to-south development of similar-to-modern topography coincided with magmatic and tectonic processes that affected large regions of the North American continental lithosphere [15]. In addition, to quantify regional estimates, we pooled the data and divided the dataset to eastern and western regions (with 115°W longitude as a rough separation line between the present-day Rocky Mountains and the Great Basin to the west, which has undergone major Neogene extension).

Figure 1.

The modern US state names in western North America. Overlaid on this map are the three phases of regional surface uplift from Mix et al. [14]. I, phase I; II, phase II; III, phase III.

3. Results

During the first 10 Myr of the Cenozoic (Palaeocene, approx. 66–56 Ma) the mammalian communities depict generally uniform environments over the entire area of western North America. We therefore focus on the time interval 56–23 Ma, when major reorganization of topography would have affected western North American habitats, as well as atmospheric circulation and precipitation patterns (figure 2; see also [15,21,30]).

Figure 2.

Mean estimated precipitation in millimetres per year for individual Palaeocene–Oligocene time intervals. The black stars are fossil localities. The precipitation estimates are based on Liu et al.'s method [16] and have standard error of 412 mm yr^–1. The three phases of regional surface uplift are shown in grey raster. (a) Phase I, (b,c) phase II, (d–h) phase III. (a) Wasatchian (55–50 Ma), (b) Bridgerian (50–46.3 Ma), (c) Uintan (46.3–40 Ma), (d) Duschesnean (40–37 Ma), (e) Chadronian (37–33.7 Ma), (f) Orellan (33.7–32 Ma), (g) Whitneyan (32–30 Ma), (h) Early Arikareean (30–23 Ma). The black dashed line in (a) shows the 115° W separation line.

Our data indicate that during the Wasatchian (approx. 56–50 Ma) extensive regions of western North America were still fairly humid (blue and green colour code in figure 2); the first indication of less humid environments occurs in the southwest (figure 2a; see also table 1). During this time, high elevation (more than 3000 m, most probably accompanied by significant topographic relief) already characterized central British Columbia [14,15,31,32], the Colorado Plateau region [33,34] and the Sierra Nevada (e.g. [35]), while the orogenic front was active in Idaho–Utah–Arizona (e.g. [36,37]).

Table 1.

Mean precipitation values for eastern and western areas (border 115° W), s.e. is ±412 mm.

NALMA	mean MAP (mm yr−1) for Eastern Part	mean MAP (mm yr−1) for Western Part
Puercan 65–63.9 Ma	2311	2316
Torrejonian 63.9–60.9 Ma	2183	n.a.
Tiffanian 60.9–56 Ma	1981	1170
Clarforkian 56–55 Ma	2034	1598
Wasatchian 55–50 Ma	1895	1672
Bridgerian 50–46.3 Ma	1754	1456
Uintan 46.3–40 Ma	1645	n.a.
Duchesnean 40–37 Ma	1481	1599
Chadronian 37–33.7 Ma	1178	1391
Orellan 33.7–32 Ma	1115	899
Whitneyan 32–30 Ma	1023	n.a.
Early Arikareean 30–23 Ma	661	n.a.

During the Bridgerian (approx. 50–46 Ma), the SWEEP model suggests high-elevation/high-relief areas spreading southwards from British Columbia towards Idaho–Montana–Wyoming [14,15] (figure 2b). The few mammal localities located to the west of the tectonic active region now show less humid environments, while the localities in the tectonically active region (Idaho–Utah–Arizona) show higher precipitation estimates. During the Uintan (middle Middle Eocene, approx. 46–40 Ma; figure 2c), this pattern remained essentially identical, with the tectonically active region becoming less humid, and a few more humid localities along the Pacific coast.

During the Duchesnean (late Middle Eocene, approx. 40–37 Ma; figure 2d), regions characterized by precipitation patterns similar to the present day (humid western part, more arid eastern part east of the Rocky Mountains) reached more southern latitudes, perhaps as far as southern Nevada (based on stable isotope data [14,38]). The Duchesnean fossil mammal data depict uniformly less humid environments than earlier in the Eocene (change from blue to green in precipitation maps), an environmental regime that greatly intensifies during the Chadronian (Late Eocene, approx. 37–34 Ma), in particular in the Wyoming–Utah–Montana area (figure 2e). The only areas suggestive of more humid environments in the Chadronian were in New Mexico, although the southernmost localities (Texas–Mexico) were more arid. Unfortunately, there is a lack of Chadronian Pacific coast mammal fossil localities.

During the Orellan (early Early Oligocene, approx. 34–32 Ma; figure 2f), mammal localities in and to the east of the tectonically active region continued to show relatively less humid conditions. The sole west coast locality indicates more humid conditions, and another isolated locality in the Mississippi region is also consistent with more humid conditions (green colour), perhaps outlining the effects of orographic rainfall on either side of the high-elevation region with precipitation delivered by the Pacific westerlies (in the west) or sourced in the Gulf of Mexico (in the east). Whitneyan (late Early Oligocene, approx. 32–30 Ma; figure 2g) conditions remained similar to the Orellan. By contrast, the Early Arikareean (early Late Oligocene, approx. 30–23 Ma; figure 2h) mammalian faunas show continuous decrease in humidity overall.

On a continental scale, mean MAP estimates decrease throughout the entire Palaeogene (figure 3a). If the data are split into eastern and western segments (at longitude 115° W; figure 3b,c), the results become even more evident. While the eastern part, reflecting the continental interior, is similar to our continental compilation (figure 3b), the western part (figure 3c) is markedly different, suggesting the latter was more humid than the continental interior (despite the limited sampling density).

Figure 3.

(a) Continental summary curves for MAP estimates (solid line) and mean hypsodonty (hatched line) values. x-axis: land mammal ages (age in million years. Grey solid vertical bar marks the EO boundary. (b,c) Summary curves for MAP estimates (solid line) and Mean hypsodonty (hatched line) values for (b) eastern and (c) western parts of the investigated area. We used 115° W as a separation line between eastern and western regions. (Online version in colour.)

4. Discussion

Considerable controversy exists about the climatic and environmental changes at or prior to the EO transition (e.g. [13,39–45]), especially as to the effect of global cooling at the EO transition on mammalian evolution and community structure (see e.g. [39,40] and references therein). Recent studies show that climatic changes accompanying the EO transition were of different magnitude in different parts of the world (see e.g. [46], Spain; [41], South America; [13,42], UK; [43,47], UK and USA). In this context, the topographic history of North America, and its influence on Northern Hemisphere environmental change, have received little attention. Stable isotope palaeoaltimetry and palaeobotanical data indicate that the western North American highlands (separating the continental interiors from the coastal regions) were extensive (e.g. [48]) and occupied regions as far south as northern Nevada around 40–37 Ma (e.g. [14,15,32]) during the late Midde Eocene (Duchesnean NALMA), causing increased aridity in the western continental interior from 37 Ma onwards (figure 2e–h). Increased rainfall along the western and eastern flanks (steep orographic rainfall gradients), with rather high δ¹⁸O values, contrasts with a relatively dry continental interior (west of 115°) characterized by low δ¹⁸O values (figure 4a,b). The changes in both δ¹⁸O values of precipitation and North American mammalian community structure occur prior to the EO transition (34–33 Ma), in concert with the development of high surface elevation in western North America.

Figure 4.

(a) Oxygen isotopic compositions of palaeoprecipitation (δ¹⁸Op) from 39 to 28 Ma (n = 264; based on Mix et al. [14]). δ¹⁸Op values are derived from stable isotope analysis of hydrated volcanic glass (diamonds), authigenic smectite clays (squares) and lacustrine or pedogenic carbonate (circles). The black horizontal line shows an east–west transect at 40° N. (b) δ¹⁸Op along the 40° N east–west transect shown in (a), highlighting (oxygen isotopic) rainfall gradients on either side of high topography. (Online version in colour.)

(a). Changes in mammalian communities in relation to mountain uplift

Our results show that regional changes in topography and regional relief exerted a major climatic impact in western North America by causing shifts in precipitation patterns, resulting in a drier interior of the continent. In accordance with our data, climate model results [21,30] delineate a humid, tropical to subtropical climate along the southern coastal plain (Texas) and in the southerly foreland basins to the east of the modern Rocky Mountains during the Early and Middle Eocene.

Our data indicate that the Duchesnean (late Middle Eocene) was uniformly less humid than previous time intervals, with considerably drier environments in the Chadronian (Late Eocene). During the Late Eocene (figure 2d–e), there is an almost complete loss of primates and archaic forms of ungulates, and a dramatic decrease in the number of taxa with bunodont cheek teeth (reflecting a generally omnivorous diet) in the herbivore fauna, from 30% in the Middle Eocene to around 10% in the Late Eocene [3]. This decrease represented the loss not only of archaic ‘condylarth’ ungulates, but also of the more primitive ‘dichobunid’ artiodactyls. The late Middle to Late Eocene (Uintan–Chadronian; figure 2c–e) marked the appearance of many new taxa, such as rhinocerotid perissodactyls, and families of suiform (tayassuids) and selenodont artiodactyls (oreodonts, anthracotheres, protoceratids, camelids and ruminants; e.g. [3]). Others have previously noted shifts in mammalian communities in the Middle to Late Eocene (e.g. [39,40,49,50]; figure 2c–e), but they have been unable to explain why these changes were much more evident in North America than in Europe.

(b). Contrast to Europe

In contrast to North America, European temperature and precipitation estimates remained high throughout the Eocene (e.g. [19,51]; see also [52]). Proxy data indicated a trend towards drier climates with marked seasonality during the Late Eocene, but continental cooling and drying was not as extreme as in North America [19,53,54], probably because Europe was an archipelago in the tropical Tethys Sea. During the Eocene, central Asia also started to become drier due to continentalization [55]. Just after the EO boundary, the Grande Coupure event took place, characterized by widespread extinction and dispersal-generated origination of taxa [6]. This event marked a sudden change from the archaic endemic European faunas to a new faunal assemblage with major components of Asian origin. Pre-Grande Coupure faunas were dominated by archaic ungulates and rodents, whereas post-Grande Coupure faunas included the ancestors of modern ungulates, such as true rhinos (rhinocerotids) and more modern types of artiodactyls (including anthracotheres, ruminants and suids). The origin and potential cause of the Grande Coupure are still largely elusive; hypotheses include climate cooling at the EO transition, dispersal from outside the main European landmasses, or a combination of climate change (cooling) and competition following dispersal into Europe (e.g. [6,56]). An understanding of the cause of the events in Europe is confounded by the minimal faunal turnover in the Early Oligocene (Orellan and Whitneyan NALMAs) in North America mammalian communities at the time of the earliest Oligocene global climatic deterioration, although it was preceded by a long-term decline in diversity over the course of the Late Eocene (e.g. [39,40,50]). The North American record also shows a pattern of episodic input of immigrant taxa from the Middle Eocene in [24].

(c). Mountain uplift explains the lack of change at the Eocene–Oligocene transition in North America

Our data show that the western North American continental interior became drier during the late Middle Eocene when compared with the earlier part of the Eocene. We propose that this Middle Eocene drying could explain why the major faunal response to the subsequent Palaeogene climatic changes in North America occurred well before the EO transition, and contrasts with contemporaneous faunal response in Eurasia where tectonics and associated surface uplift played only a subordinate role. Increasing seasonal drying affected the vegetation in the North American continental interior already in the Late Eocene. Evidence of drying is reflected in vegetational changes: tropical vegetation was replaced by more temperate habitats, such as mixed coniferous and deciduous woodlands [53,54,57], which would have favoured herbivores more specialized to feed on fibrous plant material (such as the ruminant and camelid artiodactyls, which formed a large component of the later Eocene diversification). The disappearance of most bunodont taxa (omnivores or frugivores), and of obligate arboreal mammals such as primates, is also consistent with this interpretation of environmental shift. Among the ungulates, extinctions occurred in lineages that retained bunodont or semi-lophed teeth, indicative of a more omnivorous/frugivorous diet: the archaic ‘condylarths’ disappeared, as did much of the early radiation of perissodactyls (tapiroids and hyracotheriine equids), and the generalized ‘dichobunid’ artiodactyls were replaced by early members of the modern suborders (either dedicated omnivores such as peccaries, or more specialized folivores such as early oreodonts, camelids and ruminants; see [3] for more details). Among the horses, the small bunolophodont hyracotheres were replaced by the larger anchitheres with fully lophed teeth [58]. These changes are remarkably similar to the changes that took place at Europe 10 Myr or so later, during the Grande Coupure.

5. Conclusion

Topographic reorganization of large parts of western North America during the Eocene, between 50 and 37 Ma, impacted mammalian dispersal and community structure. Regional tectonically driven surface uplift resulted in large-scale reorganization of precipitation patterns, and our data show that the mammalian faunas adapted to these changes. These changes involved the radiation of ungulates with dentitions specialized for folivory, matched by the terminal decline of primates—the so-called switch from the ‘primate-dominated world’ to the ‘ungulate-dominated world’ [59].

We suggest that the Late Eocene mammalian faunas of North America were already ‘pre-adapted’ to the colder and drier global conditions that followed the EO climatic cooling. This was due to two effects: drier environments caused by changes in atmospheric circulation patterns resulting from regional topographic uplift, and the associated establishment of cool high-elevation environments. The scenario of Middle to Late Eocene uplift between 50 and 37 Ma, causing decrease in precipitation, explains the apparent enigma of the lack of large-scale mammal faunal change around the EO transition in North America. Drying had already affected the North American continent for around 10 Myr prior to global EO climate change. Even though there were large shifts in terms of precipitation, temperature and seasonality at this time (see [43] and discussion in [47]), the mammals were already pre-adapted to these conditions. The faunal response to the transformations in vegetation (e.g. greater seasonality, with palatable food items such as buds and berries being unavailable for much of the year) had already taken place well before the EO transition. This observation explains the difference in patterns in mammalian evolution and extinction between North America and Europe at the EO transition. In Europe, the environmental conditions remained largely stable throughout the Eocene, partly due to the geographical boundary conditions (an archipelago surrounded by warm sea). Further climate change at the EO transition was combined with the impact of immigrants from Asia, which were already adapted to drier environmental conditions. This immigration event seems to have tipped the scales, resulting in the wave of extinctions among European taxa seen at this time (i.e. the Grande Coupure).

Our results highlight the importance of regional tectonic and surface uplift processes on the evolution of mammalian faunas (e.g. [60]; for Miocene, see [61,62]). In the light of recent data [55] (our results here), the dramatic effects of the Grande Coupure in Europe may reflect the response of mammals living in warmer and more humid habitats and environments that were still prevalent in Europe during that time. The mammals adapted to these earlier conditions went extinct rapidly when a tipping point was reached in terms of both climatic change and immigration.

Authors' contributions

J.T.E., C.M.J. and A.M. designed the research; J.T.E. carried out the data analysis; all authors contributed data; all authors wrote the manuscript, with J.T.E. leading. All authors gave final approval for publication.

Competing interests

The authors have no competing interests.

Funding

This study was supported by an A. v. Humboldt fellowship (to J.T.E.) and an A. v. Humboldt Research Award (to C.P.C.), EU Marie Curie Fellowship (to J.T.E. and A.M.), and NSF grant EAR-1019648 (to C.P.C. and A.M.). A.M. acknowledges support through the LOEWE funding program (Landes-Offensive zur Entwicklung wissenschaftlich-ökonomischer Exzellenz) of Hesse's Ministry of Higher Education, Research, and the Arts, and an A. Cox Visiting Professorship, Stanford University. C.M.J. acknowledges the Bushnell Foundation (Brown University).