Oligo-Miocene climate change and mammal body-size evolution in thenorthwest United States: a test of Bergmann’s Rule

John D. Orcutt and Samantha S. B. Hopkins

Abstract.—Whether or not climate plays a causal role in mammal body-size evolution is one of thelongest-standing debates in ecology. Bergmann’s Rule, the longest-standing modeladdressing thistopic, posits that geographic body-mass patterns are driven by temperature, whereas subsequentresearch has suggested that other ecological variables, particularly precipitation and seasonality, maybe the major drivers of body-size evolution. While paleoecological data provide a unique and crucialperspective on this debate, paleontological tests of Bergmann’s rule and its corollaries have been scarce.We present a study of body-size evolution in three ecologically distinct families of mammal (equids,canids, and sciurids) during the Oligo-Miocene of the northwest United States, an ideal naturallaboratory for such studies because of its rich fossil and paleoclimatic records. Body-size trends aredifferent in all three groups, and in no case is a significant relationship observed between body size andany climatic variable, counter to what has been observed in modern ecosystems. We suggest that formost of the Cenozoic, at least in the Northwest, body mass has not been driven by any one climaticfactor but instead has been the product of complex interactions between organisms and theirenvironments, though the nature of these interactions varies from taxon to taxon. The relationship thatexists between climate and body size in many groups of modern mammals, therefore, is the exceptionto the rule and may be the product of an exceptionally cool and volatile global climate. Asanthropogenic global warming continues and ushers in climatic conditions more comparable to earlierintervals of the Cenozoic than to the modern day, models of corresponding biotic variables such asbody size may lose predictive power if they do not incorporate paleoecological data.

John D. Orcutt and Samantha S. B. Hopkins. Department of Geological Sciences, University of Oregon, Eugene,Oregon 97403, U.S.A. E-mail: jorcutt@uoregon.edu

Accepted: 24 May 2013Published online: 5 September 2013Supplemental materials deposited at Dryad: doi 10.5061/dryad.n15r5

Introduction

One of the longest-standing ecologicalquestions is how—or, indeed, whether—evo-lutionary patterns are shaped by climate. Infact, this question is as old as the science ofecology itself, as the influence of climate onplant biogeography was addressed by Hum-boldt and Bonpland (1807) in one of the field’sfoundational papers. In the subsequent twocenturies, countless researchers have investi-gated the relationship between climate andbiotic change, especially in recent years as theadvent of anthropogenic global warming hasmade understanding such relationships all themore crucial. Despite their importance inpredicting the effects of future climate change,models positing a relationship between cli-matic and biotic variables often remain subjectto a great deal of debate. Even one of theearliest climatic hypotheses to be proposed isstill undergoing significant revision; this hy-

pothesis was first proposed by Bergmann(1847) as an explanation for latitudinal gradi-ents in mammal body size and has borne hisname ever since. In its original formulation,Bergmann’s Rule holds that body size isdriven by temperature: large taxa are favoredin cold climates, such as those seen toward thepoles, because their small surface area-to-volume ratios make them more effective atretaining heat, whereas smaller taxa arefavored in warm climates, such as those seentoward the equator, because they are moreeffective at dispersing heat.

Bergmann’s Rule has been tested numeroustimes in the century and a half since itsoriginal formulation. The latitudinal gradientsobserved by Bergmann have been confirmedin many mammal taxa (Ashton et al. 2000;Meiri and Dayan 2003; Blackburn and Haw-kins 2004; Rodrıguez et al. 2008) and have alsobeen reported in birds (James 1970; Blackburn

� 2013 The Paleontological Society. All rights reserved. 0094-8373/13/3904-0009/$1.00

Paleobiology, 39(4), 2013, pp. 648–661DOI: 10.1666/13006

and Gaston 1996; Ashton 2002b; Meiri andDayan 2003) and several taxa of poikilother-mic animals (Ray 1960; Lindsey 1966; Ashton2002a; Ashton and Feldman 2003); however,multiple alternative mechanisms have beenproposed that may shape these patterns (Table1). Many of these mechanisms, includingBergmann’s Rule sensu stricto, hold that thephysical environment, and in particular cli-mate, plays a major role in driving body-sizeevolution. Besides temperature, the two cli-matic variables most frequently associatedwith body mass are precipitation and season-ality. Like temperature, precipitation is gener-ally higher in low-latitude ecosystems and hasbeen suggested to show a negative relation-ship with body size, as large animals arethought to be able to store larger reserves ofwater, allowing them to survive periods ofdrought in dry environments (James 1970).Similarly, it has been suggested that largeanimals have greater capacity for energyreserves, allowing them to survive periods offamine in seasonal climates, which tend topredominate toward the poles, leading to apositive relationship between seasonality andbody size (Millar and Hickling 1990).

Another group of hypotheses posit theimportance of biotic mechanisms. Agonisticinteractions are among the most frequentlycited potential drivers of body-size evolution.Large size may confer an advantage to preytaxa by providing some protection againstpredators, and therefore taxa in environmentswith high predation pressure might be expect-ed to be larger than those that are preyedupon relatively infrequently (Korpimaki andNorrdahl 1989). The relationship betweenbody size and competition is somewhat more

complicated: Damuth (1993) found that islandtaxa, freed from competition, tended to evolveto an energetically efficient, medium-rangebody size, suggesting that increased competi-tion may select for either increased or de-creased body mass depending on the taxon inquestion. Although both of these variables canbe difficult to measure, it has long beenrecognized that richness increases toward theequator (Simpson 1964), and the increasedabundance of potential predators and com-petitors in subtropical and tropical environ-ments could lead to the appearance of body-size gradients. Food availability is also hy-pothesized to have an influence on body size.For predatory taxa, the presence of large preyanimals may select for larger individuals(Erlinge 1987), whereas the body size ofherbivores often shows a positive relationshipwith primary productivity (Rosenzweig 1968).However, primary productivity does notfollow a simple latitudinal gradient in modernecosystems, and would not be expected tolead to the formation of monotonic latitudinalbody-size gradients.

Although many refinements of and alterna-tives to Bergmann’s Rule sensu stricto havebeen proposed, no consensus exists within theecological community as to which hypothesisis most likely correct. This could be aconsequence of the inherent complexity ofnatural systems: body size may be a productof a combination of factors and different taxamay respond differently to the same stimuli.Alternatively, it could be a product of the datathat have been used to examine body-sizeevolution: neontological data are plentiful andrelatively easily controlled, but are not with-out their drawbacks. Modern ecological data

TABLE 1. Predictions of different body-mass hypotheses. Plus sign indicates a positive relationship with body mass,minus sign indicates a negative relationship with body mass, and question mark indicates a relationship that variesbased on size.

VariableExpected correlation

with body mass Explanation

Temperature � Large body size allows animals to retain heat more efficientlyPrecipitation � Large body size allows animals to harbor greater water reservesSeasonality þ Large size allows animals to harbor greater energy reservesPredation pressure þ Large size makes animals more resistant to predationCompetition ? Competition forces animals away from an optimal body sizePrey size þ Large predators are better equipped to hunt larger preyPrimary productivity þ Productive ecosystems can support larger animals

BODY SIZE IN OLIGO-MIOCENE MAMMALS 649

have been gathered from an interval thatrepresents a relatively narrow range of climat-ic conditions (Berteaux et al. 2006) duringwhich climatic variables tend to covary.Predicting biotic responses to climatic regimesdiffering from those of the previous fewdecades, therefore, requires extrapolating wellbeyond the range of observed conditions.Further, neontological data by definition canonly be used to address trends through timeacross short temporal scales. Both of theseconcerns can be addressed by substitutingtime for space through the use of paleonto-logical and paleoecological data. Given a well-sampled fossil record, robust estimates ofbody mass or well-established proxies forsize, and detailed paleoclimatic records, suchdata make it possible to observe bioticresponses to large-scale climate change, totease apart ecological variables that covary inmodern ecosystems, and to visualize trendsthrough time, providing an extra dimension toecological analyses.

The influence of climate on mammalianbiotic change through time has been the focusof several paleobiological studies, thoughthese have predominantly addressed diversityrather than body-mass evolution. The resultsof this research have been somewhat ambig-uous, with some researchers (e.g., Barnosky2001; Barnosky et al. 2003; Retallack 2007)finding a relationship between mean annualtemperature and diversity and others (e.g.,Prothero 2004; Alroy et al. 2000; Rose et al.2011) finding no predictable relationshipbetween the two variables. Body size inCenozoic mammals is also a rich area ofstudy, but until recently the field has primarilycentered on body-mass trends within groups,and in particular on Cope’s Rule, the hypoth-esis that taxa within lineages tend to increasein size with time (e.g., Alroy 1998; Gould andMacFadden 2004; Van Valkenburgh et al.2004). One of the first paleontological studiesto specifically address Bergmann’s Rule inpre-Pleistocene mammals was conducted byGingerich (2003), who observed brief butsignificant decreases in body size in thecondylarths Ectocion and Copecion as well asthe perissodactyl Hyracotherium during thePaleocene-Eocene Thermal Maximum (PETM)

in northwest Wyoming. While Gingerichacknowledged that climate likely played arole in this temporary dwarfing, he suggestedthat the magnitude of body-size change wastoo large to be explained by temperaturealone. More recently, Smith et al. (2010)analyzed maximum body mass in mammalsat a continental to global scale between theCretaceous and Recent. The authors comparedtheir body-size data with several abioticfactors, including climate, and found a signif-icant relationship between temperature andbody mass, though the authors caution thatthis does not necessarily imply a causalrelationship. A clearer picture of how climaticvariables play a role in shaping mammalbody-size evolution through time may beobtained by analyzing trends on a regionalscale. This allows large-scale trends in bodysize to be observed, while at the same timeconstraining climatic variables.

In order to carry out a localized paleoeco-logical test of Bergmann’s Rule and its climaticcorollaries, we have reconstructed body-sizetrends for three Oligo-Miocene mammal taxain the northwestern United States, a regionunusual for the high quality of both its fossiland paleoclimatic records. As climatic vari-ables fluctuate through time, Bergmann’s Ruleand its corollaries make specific predictionsabout how body size should respond. Berg-mann’s Rule sensu stricto predicts that bodysize should track temperature through time,with smaller-bodied forms predominatingduring periods of global warming and larg-er-bodied forms predominating in coolerconditions. If either of the other climatichypotheses is correct, however, larger taxashould be prevalent during periods of de-creased rainfall or increased seasonality. Thethree models discussed above predict not onlythe direction of the relationship between aclimatic variable and body size, but also therelative magnitude of change: a major changein temperature, precipitation, or seasonalityshould drive a comparably large change inbody mass, whereas taxa should not changesignificantly in size during periods of climaticstasis.

650 JOHN D. ORCUTT AND SAMANTHA S. B. HOPKINS

Methods

Scope of Project and Paleoclimatic Proxies.—The temporal scope of this project is theArikareean–Hemphillian North AmericanLand Mammal Ages (late Oligocene–earliestPliocene, 30–5 Ma [Tedford et al. 2004]). Thisinterval is an ideal one in which to study therelationship between climate and body mass,both because of its excellent mammal fossilrecord and because it encompasses consider-able fluctuation in global temperature (Zachoset al. 2001). This fluctuation includes bothintervals of global warming (e.g., late Oligo-cene warming, 26–24 Ma; Mid-Miocene Cli-matic Optimum, MMCO, 17–15 Ma) andintervals of global cooling (e.g., Oligo-Mioceneglaciation, 24 Ma; late Miocene cooling, 15–5Ma).

Previous analyses of mammal body sizethrough time have focused primarily onphylogenetic body-mass trends (e.g., Alroy1998, 2003; Gould and MacFadden 2004), andin consequence have been conducted at acontinental scale. Because this study addressesthe relationship between body size andclimate and because climate can vary consid-erably even within a fairly small area, it wasconducted within a single biogeographicregion: the Northwest of the United States(the Columbia Plateau faunal province ofTedford et al. 2004), which encompassesprimarily sites from eastern Oregon, but alsoincludes faunas from southern Washington,southeast Idaho, and northern Nevada (Fig.1). The region’s fossil record is both rich andnearly continuous over the course of theArikareean–Hemphillian interval. Many ofthese faunas, such as those from the Arikar-eean–Hemphillian beds of the John Day Basin,the Barstovian–Hemphillian formations ofOregon’s Owyhee Basin and Nevada’s VirginValley, and the Clarendonian–Hemphillianbeds along the Columbia River in Oregonand Washington, have been extensively col-lected and studied.

Besides a well-studied and fairly completefossil record, the Northwest is also remarkablefor its detailed record of regional climate.Although other regions in North America,most notably the Great Plains, may provide a

greater density of fossils, the robust paleocli-matic framework of the Northwest makespossible the direct comparison of body massand climate, allowing a more rigorous test ofBergmann’s Rule than would be possibleelsewhere. This framework is derived primar-ily from climofunctions based on the physicaland chemical characteristics of paleosols.Sheldon et al. (2002) established relationshipsbetween the chemical index of alterationwithout potassium and mean annual temper-ature and between the molar ratio of bases toalumina and mean annual precipitation, andRetallack (2005) demonstrated that meanannual range of precipitation (a measurementof seasonality) can be estimated from thethickness of carbonate nodule-bearing hori-zons of soils. Retallack (2007) collected datafrom Cenozoic paleosols from sites through-out eastern Oregon to reconstruct regionaltrends in temperature, precipitation, andseasonality. These data were primarily gath-ered from sites in the Columbia, John Day, andOwyhee Basins, which are also the richestfossil-bearing basins in the region. Not only isthere a large amount of overlap between thefossil and paleoclimatic records of the regionbut also the climatic signal from coeval sites isbroadly similar, making it reasonable to usedata gathered from them as proxies for climateacross the Northwest. The robustness ofpaleopedological climofunctions has frequent-ly been called into question (e.g., Royer 1999)

FIGURE 1. Map of study area and approximate location ofsites included. A, Arikareean NALMA; Hf, Hemingfor-dian NALMA; B, Barstovian NALMA; C, ClarendonianNALMA; Hh, Hemphillian NALMA.

BODY SIZE IN OLIGO-MIOCENE MAMMALS 651

and similar studies often rely on isotopicproxies for climate (e.g., Rose et al. 2011).However, two aspects of the paleopedologicalrecord have led to its inclusion here. First, thechemical assays developed by Sheldon et al.(2002) have a more robust relationship withtemperature and precipitation than do manypaleopedological proxies for climate. Second,the paleosol record of eastern Oregon isextremely well sampled; the database usedin this study encompasses 113 samples. Manyof these samples were collected at sites thathave also yielded extensive mammalian fossilmaterial, allowing site-specific trends in cli-mate and body mass to be observed.

Body-Mass Estimation.—The size of theNorthwest fossil record is sufficiently largeto make an analysis of body-size trends in allmammals impractical. We have chosen tofocus on three well-represented families ofmammals: equids, canids, and sciurids. Eachgroup is distinct from the other two in termsof size, diet, and ecology; in fact, eachrepresents one of the poles of Schad’s (1977;modified by Retallack 2004) ternary model ofanimal lifestyles (equids are large herbivores,canids are mid-sized carnivores, and sciuridsare small granivores). Because these familieseach exhibit a different ecology, any body-sizetrend apparent in all three may reasonably bepresumed to represent an ecosystem-widesignal. Further, robust taxonomic frameworksexist for all three families (MacFadden 1992;Goodwin 2008; Wang and Tedford 2008),making observations of trends at manytaxonomic levels possible. This is importantbecause Bergmann’s (1847) original study didnot focus only on species, as much subsequentresearch has, but on trends at the genus andfamily level as well (James 1970). In fact, itwas at these higher taxonomic levels that heobserved the clearest latitudinal gradients.

A well-established relationship betweendental dimensions and body mass has beenestablished for each of the three familiesstudied. Sciurid body mass can be estimatedfrom the length of the lower tooth row(Hopkins 2007); however, there are relativelyfew sciurid specimens that preserve the entirelower tooth row, so first lower molar length isused as a body-mass proxy instead, because

size of first lower molar has been shown to bea good proxy for body size across allmammals, provided that the study is taxo-nomically constrained (Creighton 1980). Canidbody mass is related to the antero-posteriorlength of the first lower molar (the lowercarnassial; Van Valkenburgh 1990). Becausefirst lower molar length was the only body-size proxy used in these two families, dentalmeasurements were compared directly withone another rather than transforming the databy calculating body mass, thus reducing theamount of error in the analysis. Equid bodysize is based on the regressions for perisso-dactyls and hyracoids developed by Janis(1990), which used the length of any lowercheek tooth or second upper molars. Becausemultiple body-mass regressions exist forequids, we calculated body mass in order toinclude the largest number of specimenspossible.

Only specimens already collected and cata-logued in museum and university collectionswere measured. Data are derived from sur-veys of the collections at the AmericanMuseum of Natural History (AMNH), IdahoMuseum of Natural History (IMNH), JohnDay Fossil Beds National Monument (JODA),Natural History Museum of Los AngelesCounty (LACM), Sierra College Natural His-tory Museum (SCNHM), South Dakota Schoolof Mines and Technology (SDSMT), Universityof California Museum of Paleontology(UCMP), University of Oregon Museum ofNatural and Cultural History (UOMNH), andUniversity of Washington Burke Museum(UWBM). Some intervals (e.g., the Hemingfor-dian) were underrepresented in the collectionsvisited; we therefore augmented these areas ofthe data set by including published measure-ments of specimens from collections that werenot visited. Measurements were made usingMitutoyo Absolute Digimatic Calipers. Intotal, 67 sciurid specimens, 66 canid speci-mens, and 230 equid specimens were includedin the database; more precise details about thespecimens used are available in the supple-mentary material.

Analysis.—The most straightforward quali-tative method for comparing body-masstrends with climatic fluctuation is to plot

652 JOHN D. ORCUTT AND SAMANTHA S. B. HOPKINS

visual representations of both variablesthrough time. In order to do so, we binnedbody size and paleoclimate data into NorthAmerican Land Mammal Age (NALMA)subdivisions (Tedford et al. 2004) and plottedthem in Microsoft Excel. Although thesesubdivisions vary in duration from 0.5 to 5Myr (with an average duration of 1.67 Myr),we opted to use them as bins in preference to1-Myr intervals because the dating of manysites is imprecise. Only 68 of the 410 publishedmammal-bearing localities in Oregon (17%)have been dated with radioisotopic methods,and localities in surrounding states are evenmore dependent on biostratigraphic dates(Carrasco et al. 2005). It therefore is currentlyimpossible to assign the overwhelming ma-jority of the sites included in this study to finerbins than NALMA subdivisions. Body-sizeprofiles (plots of mean body size throughtime) were constructed for all families, sub-families, tribes, subtribes, genera, and speciesthat were present in more than one temporalbin.

We tested whether body mass showed asignificant relationship with any of the threeclimatic variables by running two sets ofmultiple linear regressions in JMP statistical

discovery software. In the first, both bodymass and paleoclimatic data were binned intoland mammal age subdivisions. Only the 19taxa that were present in three or more binscould be included in this portion of the study.These taxa included all three families as wellas five subfamilies, five tribes, five genera, andone species (Table 2). In order to address theconcern that binning body mass and climateinto temporal intervals might dampen anyfine-scale ecological signal, we ran a secondset of regressions comparing site-specificvariables. This reduced the sample size toseven sites (Bone Creek, Mascall, Black Butte,Ironside, McKay, Rattlesnake, and Rome) andsix taxa (three families, one subfamily, andtwo tribes; Table 3), but allows a directcomparison between body size and climaticvariables within a temporally and geograph-ically constrained paleoecosystem. We deter-mined whether or not body-size trends withintaxa conformed to the predictions made byBergmann’s Rule and its corollaries by using asign test on the sign of the slope of eachregression.

Hypotheses positing a connection betweenclimate and body size predict not only thedirectionality of body-size change, but the

TABLE 2. Slope and significance of relationships between climate and mammal body mass at a regional scale (databinned into NALMA subdivisions). Columns indicate slope and significance (p-value) of the relationship. Plus andminus signs in parentheses after climatic variables indicate directionality predicted by Bergmann’s Rule and itscorollaries.

Temperature (�) Precipitation (�) Seasonality (þ)

Slope Significance Slope Significance Slope Significance

Sciuridae 0.09 0.36 �0.0006 0.68 0.003 0.96Sciurinae 0.08 0.38 �0.0005 0.72 –0.0004 0.99Marmotini 0.04 0.78 �0.001 0.22 0.003 0.97Protospermophilus 0.09 0.37 0.002 0.29 �0.06 0.01*Spermophilus �0.14 0.14 –0.00003 0.99 0.07 0.63S. wilsoni �0.34 0.16 0.0006 0.81 �0.20 0.42Canidae 1.47 0.18 0.03 0.12 �0.64 0.29Borophaginae 1.61 0.60 0.004 0.93 �0.05 0.98Borophagini 1.56 0.61 0.004 0.93 �0.06 0.97Caninae 3.07 0.05 �0.08 0.27 3.39 0.03*Canini 3.22 0.12 �0.11 0.16 3.60 0.09Equidae 28.61 0.37 �0.02 0.97 �7.95 0.63Anchitheriinae 19.65 0.03* 0.35 0.05 �6.07 0.30Archaeohippus 4.87 0.09 0.10 0.03* �3.25 0.12Parahippus 14.60 0.35 0.23 0.43 �4.16 0.51Equinae �61.30 0.14 �0.72 0.07 32.09 0.13Hipparionini �48.69 0.12 �0.24 0.60 15.78 0.74Equini �100.05 0.36 �1.20 0.22 109.18 0.08Pliohippus �91.91 0.39 �1.12 0.24 102.97 0.06

* Significant at p , 0.05.

BODY SIZE IN OLIGO-MIOCENE MAMMALS 653

magnitude of that change as well, as largechanges in climate should be associated withlarge changes in body mass. In order to testthese predictions, the magnitude of body-sizechange between NALMA subdivisions wascalculated by computing first differences inmean body mass for 12 family- to genus-leveltaxa between bins (Table 4). Although firstdifferences were always calculated for twosequential bins, not all units contain data, so insome cases these bins were separated by a gapof three or fewer NALMA subdivisions. Firstdifferences between the same bins were alsocalculated for the three climatic variables. Themagnitude of climate change was then re-gressed against the magnitude of body-sizechange for the 12 different taxa.

Results

Body-size trends vary widely between thethree families included in this study (Figs. 2–4). Mean sciurid body mass remains roughly

constant throughout the Oligo-Miocene, witha spike in the earliest Hemphillian. Equidbody mass is static through the Arikareean butbegins steadily increasing in the mid-Miocene.Patterns in canid body-mass evolution aremore complex, with periods of both increasingand decreasing size. In no case does bodymass closely track any climatic variable.

Only four taxa were found to have astatistically significant relationship with cli-mate at a regional level: significant positiverelationships were found to exist betweentemperature and body mass in ‘‘anchither-iine’’ horses and between precipitation andbody mass in the equid Archaeohippus, where-as positive relationships exist between sea-sonality and body mass in Protospermophilusand Caninae (Table 2). The site-specific datayielded only two significant results: positiverelationships between sciurid body mass andtemperature and between canid body massand precipitation (Table 3). In five of the six

TABLE 3. Site-specific slope and significance of relationships between climate and mammal body mass. Columnsindicate slope and significance (p-value) of the relationship. Plus and minus signs in parentheses after climatic variablesindicate directionality predicted by Bergmann’s Rule and its corollaries. ND, no data.

Temperature (�) Precipitation (�) Seasonality (þ)

Slope p-value Slope p-value Slope p-value

Sciuridae 0.47 0.05* 0.0003 0.94 �0.08 0.77Marmotini 0.88 0.12 �0.003 0.53 0.18 0.67Canidae �6.19 0.64 0.06 0.007* �4.48 0.15Equidae �23.49 0.73 �0.26 0.73 2.59 0.90Equinae �53.46 0.41 �0.73 0.30 37.08 0.05Hipparionini ND ND ND ND 50.63 0.10

* Significant at p , 0.05.

TABLE 4. Slope and significance of relationships between climate and mammal body size first differences. Columnsindicate slope and significance (p-value) of the relationship. Note that no correlation is significant (p , 0.05). Plus andminus signs in parentheses after climatic variables indicate directionality predicted by Bergmann’s Rule and itscorollaries. ND, no data.

Temperature (�) Precipitation (�) Seasonality (þ)

Slope p-value Slope p-value Slope p-value

Sciuridae 0.13 0.37 �0.001 0.40 0.03 0.55Sciurinae 0.12 0.41 �0.001 0.38 0.03 0.56Marmotini 0.05 0.75 �0.002 0.26 �0.008 0.95Spermophilus �0.15 0.41 0.002 0.46 �0.23 0.24Canidae 0.41 0.73 0.03 0.25 �0.32 0.65Borophaginae �0.88 0.40 �0.01 0.36 �0.01 0.98Borophagini �0.86 0.42 �0.01 0.38 �0.03 0.95Equidae �13.40 0.54 �0.27 0.35 �6.32 0.45Anchitheriinae ND ND ND ND �0.84 0.86Parahippus ND ND ND ND 5.5 0.17Equinae �32.46 0.38 �0.53 0.30 �17.33 0.81Hipparionini �48.36 0.40 �0.33 0.74 325.79 0.73

654 JOHN D. ORCUTT AND SAMANTHA S. B. HOPKINS

cases, the direction of the relationship was theopposite of what would be predicted byBergmann’s Rule and its corollaries; onlyProtospermophilus and Caninae showed theexpected positive relationship with seasonal-ity. For the regional data, only 46% of theregressions yielded a result in keeping withthe predictions made by Bergmann’s Ruleand other climatic hypotheses; this number

rises to only 63% in the site-specific data. Inall cases the sign test fails to support asignificant departure from a random distri-bution. Regression of body mass againstclimatic first differences did not yield anysignificant relationships (Table 4), nor did thedistribution of the signs of the regressionsdiffer significantly from a random distribu-tion.

FIGURE 2. Mean equid body size in the northwest United States through the Arikareean–Hemphillian interval. From topleft to bottom right charts show body mass at family, subfamily, tribe, genus, and species levels. Error bars represent onestandard error. Shaded boxes along x-axes represent Arikareean (Ar), Hemingfordian (Hf), Barstovian (Ba),Clarendonian (Cl), and Hemphillian (Hh) NALMA subdivisions (after Tedford et al. 2004).

BODY SIZE IN OLIGO-MIOCENE MAMMALS 655

Discussion

The results presented here do not support

Bergmann’s Rule and its climatic corollaries

in Oligo-Miocene mammals from the North-

west. There is no evidence for a causal link

between any one climatic variable and

mammal body-size evolution within the

study area and interval. Some significant

relationships between climate and body mass

were found, but in almost all cases the

direction of the relationship was the oppositeof what had been predicted. A certainnumber of false positives are to be expectedwhen running numerous statistical tests(Gotelli and Ellison 2004), and it is likely thatthese relationships fall into this categoryrather than representing actual ecologicalpatterns, a suggestion supported by theresults of the Bonferroni correction for thenumber of tests run. After applying thecorrection, only the relationship between

FIGURE 3. Mean canid body size in the northwest United States through the Arikareean–Hemphillian interval. Details asin Figure 2.

656 JOHN D. ORCUTT AND SAMANTHA S. B. HOPKINS

site-specific canid mass and precipitationremains statistically significant. Even thisresult is unlikely to represent a causalrelationship, as the precipitation hypothesispredicts a negative relationship whereas therelationship found here is positive. Theprecipitation and seasonality hypotheses areborne out by the data more often than isBergmann’s Rule sensu stricto, but in no caseis the distribution of expected and unexpect-ed results a significant departure from ran-

dom; as such, these patterns do not imply acausal relationship between climate and bodymass.

The body-size profiles plotted for taxa fromeastern Oregon and surrounding states furtherrefute the suggestion that any one climaticvariable predictably drove Oligo-Miocenebody-size evolution in the region. Not onlydo none of these profiles closely track tem-perature, but in addition all show extremelydifferent patterns, reflecting the absence of an

FIGURE 4. Mean sciurid body size in the northwest United States through the Arikareean–Hemphillian interval. Detailsas in Figure 2.

BODY SIZE IN OLIGO-MIOCENE MAMMALS 657

ecosystem-wide signal. Some of the patternspresented here parallel those observed at thecontinental scale and may suggest alternativedrivers of body-size evolution. Fossil horsesare the classic example of Cope’s Rule, as theyshow an increase in body size during thecourse of the Cenozoic, though as Gould andMacFadden (2004) observe, this reflects theevolution of large body size in severaldifferent lineages rather than a family-widetendency toward larger taxa. This size in-crease, as well as a concurrent increase inbody-size variation, is visible in horses fromthe Northwest beginning in the mid-Miocene,but is preceded by a period of stasis: from theearliest Arikareean through the end of theHemingfordian, equids (which are represent-ed during that interval solely by ‘‘anchither-iines’’) do not undergo a significant change insize. The size increase observed beginning inthe Barstovian is delayed relative to horses inNorth America at a continental scale (Mac-Fadden 1986) and is roughly coincident withthe breakup of woodland habitats after theMid-Miocene Climatic Optimum (MMCO)(ca.16 Ma [Zachos et al. 2001]). Woodlandhabitats had begun to give way to grasslandsduring the early Arikareean in the Northwest,but rebounded more strongly in the regionthan elsewhere in North America during theMMCO, as indicated by plant macrofossils,paleosols, and the presence of browsingherbivores in Barstovian faunas in the region(Retallack 2007), possibly explaining the de-layed increase in equid body mass. Thissuggests that, instead of being driven primar-ily by one climatic variable, horse body-sizeevolution has been influenced by large-scaleenvironmental change, with more open habi-tats favoring the evolution of larger taxawithin several different lineages.

Canid body mass also follows patternssimilar to those observed at larger scales.The family-level body-size profile appearschaotic, but clearer trends become apparentat lower taxonomic levels. All three subfam-ilies of canids (Hesperocyoninae, Borophagi-nae, and Caninae) are present in theNorthwest at the beginning of the Arikar-eean–Hemphillian interval as members ofcentral Oregon’s John Day fauna. Hespero-

cyonines are represented by the large-bodiedEnhydrocyon, Paraenhydrcyon, and Mesocyon,whereas both the Borophaginae and Caninaeare represented by much smaller taxa. Hes-perocyonines disappear from the fossil recordof the Northwest after the mid-Arikareean, atwhich point borophagines expand in bothdiversity and size. Members of this subfamilyincrease in size during each land mammal agesubdivision in which they are present, mirror-ing the pattern observed by Van Valkenburghet al. (2004) on a continental scale. Boropha-gines were largely extinct in the Northwest bythe end of the Clarendonian; this extinctionappears to have allowed Caninae the oppor-tunity to diversify elsewhere in North Amer-ica (Tedford et al. 2009), though thisdiversification is not well documented in theNorthwest, owing to a scarcity of Pliocenesediments. These patterns suggest that canidbody mass may not be shaped by climate, butrather by biotic interactions, particularly com-petition with other canids, leading to a patternof taxonomic replacement and subsequent sizeincrease.

Sciurids present a novel pattern: squirrelbody mass does not change significantlythroughout the course of the Arikareean–Hemphillian interval. This is true at alltaxonomic levels from family to genus (Proto-spermophilus, for example, does not changesignificantly in size between the earliestArikareean and the early Barstovian, a periodof nearly 14 Myr). This pattern may not betypical of all squirrels: tree squirrels areknown from the Northwest only from theArikareean John Day fauna, chipmunks ap-pear only in Clarendonian and Hemphillianfaunas, and the giant marmot Paenemarmotaappears only in the Thousand Creek fauna,accounting for the spike in squirrel body sizeobserved in the earliest Hemphillian. All othersciurids included in this study representground squirrels that were morphological-ly—and likely ecologically—analogous to themodern Spermophilus (itself an importantmember of late Miocene faunas in the region).Burrowing may serve to buffer ground squir-rels and other fossorial rodents against anumber of the proposed drivers of body-sizeevolution (Hopkins 2007); these include not

658 JOHN D. ORCUTT AND SAMANTHA S. B. HOPKINS

only temperature and other climatic variables,but also biotic interactions, in particularpredation and competition. Further analysesof body-size evolution in other burrowingtaxa, particularly geomyids, will help establishwhether similar trends are common to allfossorial mammals.

The research described here focuses onbody-size trends within taxonomic groups.However, it is possible that examining trendswithin ecological categories might reveal verydifferent patterns. This is especially worthconsidering among canids, which run thegamut from hypo- to hypercarnivorous forms(Wang and Tedford 2008). Indeed, Van Val-kenburgh et al. (2004) suggest that a causalrelationship exists between body size andhypercarnivory in canids, implying that ananalysis based on ecological categories mightyield different results for different groups.However, Bergmann’s Rule sensu stricto isexplicitly systematic, stating that body sizepatterns of taxa within a higher taxon aredriven by climate (Bergmann 1847; James1970). As a test of Bergmann’s Rule, ouranalysis has likewise focused on trends withintaxonomic groups. We do suggest, though,that analyzing body-size patterns withinecological categories through time would bea fruitful area of study, potentially providinginvaluable information on the major drivers ofbody-size evolution in Cenozoic mammals.

The lack of a relationship between climateand body size in the Oligo-Miocene of theNorthwest is in sharp contrast to the patternsobserved in extant mammals, including manycanids and sciurids, which tend to shownegative relationships with temperature andprecipitation and positive relationships withseasonality (Ashton et al. 2000). It alsocontrasts with the results of previous studiesof climate and mammal body-size evolution,particularly those of Gingerich (2003) andSmith et al. (2010), both of which suggest arelationship between body mass and temper-ature. In both these cases, however, there arereasons to doubt that a causal relationshipexists between the two variables. Althoughthe condylarth and perissodactyl genera ex-amined by Gingerich do show a markeddecrease in size coincident with the PETM,

he suggests that the magnitude of body-sizechange is far too large to be attributable solelyto climate. Rising temperature may haveplayed some role in the dwarfing of severalmammal lineages, but Gingerich suggests thatother factors, particularly an increase inatmospheric CO2 and a corresponding de-crease in the nutritive value of plant leaves,are more likely responsible for driving body-size evolution in Bighorn Basin mammals. Theglobal-scale relationship between mammalbody mass and temperature posited by Smithet al. (2010) is unlikely to imply simplecausation for several reasons. As discussedabove, the scale of the study is too large todetect local or regional variation. Further, asthe authors observe, there are many reasons tobelieve that the relationship they find betweenbody mass and temperature may not becausal. Global temperature influences and isinfluenced by several other factors, includingatmospheric oxygen concentration and landarea. Finally, Smith et al. (2010) did notconsider other climatic variables in theiranalysis. It therefore is not possible to saywhether temperature controls body size at aglobal scale or if body mass and temperatureboth respond to some third biotic or abioticvariable.

A relationship between climate and bodysize is not the only macroecological patternobserved in modern ecosystems that appearsnot to be universally applicable during theCenozoic. Rose et al. (2011), for instance,observed that there is no evidence for latitudi-nal richness gradients in North AmericanPaleocene mammals. Any number of factorsmay account for such discrepancies betweenneontological and paleontological trends. Roseet al. suggest that the absence of strongPaleocene diversity gradients may be due tothe faunal composition of the ecosystemsstudied (a large percentage of Paleocenemammals are members of extinct taxa withno modern analogue and thus might not beexpected to respond to environmental variablesin the same way as extant mammals) or to aprolonged recovery from the end-Cretaceousmass extinction. Because all three familiesstudied here are represented by extant mem-bers and because the Arikareean–Hemphillian

BODY SIZE IN OLIGO-MIOCENE MAMMALS 659

interval is not preceded by a major extinctionevent, neither of these factors can explain theabsence of a relationship between Oligo-Mio-cene climate and body size. Instead, this maybe due to the vastly different climatic regimesof the Oligo-Miocene and the Pleistocene andHolocene. Recent climate is both aberrantlycool and volatile (Zachos et al. 2001) and whatmay appear in modern ecosystems to be strongrelationships between climate and ecologicalvariables such as body size may not hold underthe warmer conditions that have characterizedmost of the Cenozoic. Similar analyses per-formed in other biogeographic regions andalong geographic gradients during the Oligo-Miocene will help establish whether or not theNorthwest is atypical in its apparent disconnectbetween climate and body size. If it is, in fact,the case that Bergmann’s rule does not apply inthe Oligo-Miocene, the implications for creat-ing and testing ecological models would beprofound.

A concern that has begun to emerge fromprojects such as the one presented here andthat of Rose et al. (2011) is the possibility thatmodern ecosystems are not necessarily keys tothose of the past. If this is the case, it has manyimportant ramifications, chief among themthat the present may also not be the key to thefuture. As anthropogenic global warmingcontinues, so too will widespread environ-mental change. Well-calibrated ecologicalmodels are crucial to predicting and mitigat-ing the effects of this change. Ecologists haveprimarily used modern ecosystems and or-ganisms as the basis for these models, but asclimate change ushers in conditions with nohistorical precedent, neontologically basedmodels may begin to lose some of theirpredictive power. Augmenting environmentalmodels with paleoecological data allows bioticresponses to instances of large-magnitudeclimate change to be included and in so doingcan enhance predictive power.

Acknowledgments

This research would not have been possiblewithout the assistance of collections staff atnumerous repositories, particularly J. Galkin(AMNH), M. Thompson (IMNH), C. Schierup(JODA), S. McLeod, V. Rhue, G. Takeuchi

(LACM), G. Bromm, R. Hilton (SCNHM), S.Shelton (SDSMT), P. Holroyd, S. Tomiya(UCMP), R. Eng, C. Richards, and C. Sidor(UWBM). J. Samuels (JODA) shared measure-ments of recently identified sciurid teeth. Wewould also like to thank the other members ofthe Vertebrate Paleontology Group at theUniversity of Oregon, especially E. Davisand W. McLaughlin, for their feedback andencouragement on this manuscript. Twoanonymous reviewers provided insightfulcritiques that have been taken into accountin the final version of this paper. Orcutt wouldlike to thank his Ph.D. committee members G.Retallack, R. Dorsey, and S. Frost for theirguidance throughout this research, as well asT. Fremd for suggesting Oligo-Miocene paleo-ecology as a topic in the first place and J.Calede for his helpful analytical discussions.

Literature Cited

Alroy, J. 1998. Cope’s rule and the dynamics of body mass

evolution in North American fossil mammals. Science 280:731–

734.

———. 2003. Taxonomic inflation and body mass distributions in

North American fossil mammals. Journal of Mammalogy

84:431–443.

Alroy, J., P. L. Koch, and J. C. Zachos. 2000. Global climate change

and North American mammal evolution. Paleobiology 26:259–

288.

Ashton, K. G. 2002a. Do amphibians follow Bergmann’s Rule?

Canadian Journal of Zoology 80:708–716.

———. 2002b. Patterns of within-species body size variation of

birds: strong evidence for Bergmann’s Rule. Global Ecology and

Biogeography 11:505–523.

Ashton, K. G., and C. R. Feldman. 2003. Bergmann’s Rule in

nonavian reptiles: turtles follow it, lizards and snakes reverse it.

Evolution 57:1151–1163.

Ashton, K. G., M. C. Tracy, and A. de Quiroz. 2000. Is Bergmann’s

Rule valid for mammals? American Naturalist 156:390–415.

Barnosky, A. D. 2001. Distinguishing the effects of the Red Queen

and the Court Jester on Miocene mammal evolution in the

northern Rocky Mountains. Journal of Vertebrate Paleontology

21:172–185.

Barnosky, A. D., E. A. Hadly, and C. J. Bell. 2003. Mammalian

response to global warming on varied temporal scales. Journal

of Mammalogy 84:354–368.

Bergmann, C., 1847. Ueber die Verhaltnisse der Warmeokonomie

der Thiere zu ihrer Grosse. Gottinger Studien 3:595–708.

Berteaux, D., M. M. Humphries, C. J. Krebs, M. Lima, A. G.

McAdam, N. Pettorelli, D. Reale, T. Saitoh, E. Tkadlec, R. B.

Weladji, and N. C. Stenseth. 2006. Constraints to projecting the

effects of climate change on mammals. Climate Research 32:151–

158.

Blackburn, T. M., and K. J. Gaston, 1996. Spatial patterns in the

body sizes of bird species in the New World. Oikos 77:436–446.

Blackburn, T. M., and B. A. Hawkins. 2004. Bergmann’s Rule and

the mammal fauna of northern North America. Ecography

27:715–724.

660 JOHN D. ORCUTT AND SAMANTHA S. B. HOPKINS

Carrasco, M. A., B. P. Kraatz, E. B. Davis, and A. D. Barnosky.

2005. Miocene Mammal Mapping Project (MIOMAP). http://

www.ucmp.berkeley.edu/miomap/.

Creighton, G. K. 1980. Static allometry of mammalian teeth and the

correlation of tooth size and body size in contemporary mammals.

Journal of the Zoological Society of London 191:435–443.

Damuth, J. 1993. Cope’s rule, the island rule and the scaling of

mammalian population density. Nature 365:748–750.

Erlinge, S. 1987. Why do European stoats Mustela erminea not

follow Bergmann’s Rule? Holarctic Ecology 10:33–39.

Gingerich, P. D. 2003. Mammalian responses to climate change at

the Paleocene-Eocene boundary: Polecat Bench record in the

northern Bighorn Basin, Wyoming. In S. L. Wing, P. D.

Gingerich, B. Schmitz, and E. Thomas, eds. Causes and

consequences of globally warm climates in the early Paleogene.

Geological Society of America Special Paper 369:463–478.

Goodwin, H. T. 2008. Sciuridae. Pp. 355–376 in C. M. Janis, G. F.

Gunnell, and M. D. Uhen, eds. Evolution of Tertiary mammals

of North America, Vol. 2. Small mammals, xenarthrans, and

marine mammals. Cambridge University Press, Cambridge.

Gotelli, N. J., and A. M. Ellison. 2004. A primer of ecological

statistics. Sinauer, Sunderland, Mass.

Gould, G. C., and B. J. MacFadden. 2004. Gigantism, dwarfism,

and Cope’s rule: ‘‘nothing in evolution makes sense without a

phylogeny.’’ Bulletin of the American Museum of Natural

History 285:219–237.

Hopkins, S. S. B. 2007. Causes of lineage decline in the

Aplodontidae: testing for the influence of physical and

biological change. Palaeogeography, Palaeoclimatology, Palae-

oecology 246:331–353.

Humboldt, A., and A. Bonpland. 1807. Essay on the geography of

plants. Schoell, Paris.

James, F. C. 1970. Geographic size variation in birds and its

relationship to climate. Ecology 51:365–390.

Janis, C. M. 1990. Correlation of cranial and dental variables with

body size in ungulates and macropodoids Pp. 255–299 in J.

Damuth and B. J. MacFadden, eds. Body size in mammalian

paleobiology. Cambridge University Press, Cambridge.

Korpimaki, E., and K. Norrdahl. 1989. Avian predation on

mustelids in Europe 1: occurrence and effects on body size

variation and life traits. Oikos 55:205–215.

Lindsey, C. C. 1966. Body sizes of poikilotherm vertebrates at

different latitudes. Evolution 20:456–465.

MacFadden, B. J. 1986. Fossil horses from ‘‘Eohippus’’ (Hyracothe-

rium) to Equus: scaling, Cope’s Law, and the evolution of body

size. Paleobiology 12:355–369.

———. 1992. Fossil horses: systematics, paleobiology, and

evolution of the Family Equidae. Cambridge University Press,

Cambridge.

Meiri, S., and T. Dayan. 2003. On the validity of Bergmann’s Rule.

Journal of Biogeography 30:331–351.

Millar, J. S., and G. J. Hickling. 1990. Fasting endurance and the

evolution of mammalian body size. Functional Ecology 4:5–12.

Prothero, D. R. 2004. Did impacts, volcanic eruptions, or climate

change affect mammalian evolution? Palaeogeography, Palae-

oclimatology, Palaeoecology 214:283–294.

Ray, C. 1960. The application of Bergmann’s and Allen’s rules to

the poikilotherms. Journal of Morphology 106:85–108.

Retallack, G. J. 2004. Ecological polarities of mid-Cenozoic fossil

plants and animals from central Oregon. Paleobiology 30:561–

588.

———. 2005. Pedogenic carbonate proxies for amount and

seasonality of precipitation in paleosols. Geology 33:333–336.

———. 2007. Cenozoic paleoclimate on land in North America.

Journal of Geology 115:271–294.

Rodrıguez, M. A., M. A. Olalla-Tarraga, and B. A. Hawkins. 2008.

Bergmann’s Rule and the geography mammal body size in the

Western Hemisphere. Global Ecology and Biogeography

17:274–283.

Rose, P. J., D. L. Fox, J. Marcot, and C. Badgley. 2011. Flat

latitudinal gradient in Paleocene mammal richness suggests

decoupling of climate and biodiversity. Geology 39:163–166.

Rosenzweig, M. L. 1968. The strategy of body size in mammalian

carnivores. American Midland Naturalist 80:299–315.

Royer, D. L. 1999. Depth to pedogenic carbonate horizon as a

paleoprecipitation indicator? Geology 27:1123–1126.

Schad, W. 1977. Man and mammals: toward a biology of form.

Waldorf, New York.

Sheldon, N. D., G. J. Retallack, and S. Tanaka. 2002. Geochemical

climofunctions from North American soils and application to

paleosols across the Eocene-Oligocene boundary in Oregon.

Journal of Geology 110:687–696.

Simpson, G. G. 1964. Species density of North American recent

mammals. Systematic Biology 13:57–73.

Smith, F. A., A. G. Boyer, J. H. Brown, D. P. Costa, T. Dayan, S. K.

M. Ernest, A. R. Evans, M. Fortelius, J. L. Gittleman, M. J.

Hamilton, L. E. Harding, K. Lintulaakso, S. K. Lyons, C.

McCain, J. G. Okie, J. J. Saarinen, R. M. Sibly, P. R. Stephens, J.

Theodor, and M. D. Uhen. 2010. The evolution of maximum

body size of terrestrial mammals. Science 330:1216–1219.

Tedford, R. H., L. B. Albright III, A. D. Barnosky, I. Ferrusquia-

Villafranca, R. M. Hunt, Jr., J. E. Storer, C. C. Swisher III, M. R.

Voorhies, S. D. Webb, and D. P. Whistler. 2004. Mammalian

biochronology of the Arikareean through Hemphillian interval

(late Oligocene through early Pliocene Epochs). Pp. 169–231 in

M. O. Woodburne, ed. Late Cretaceous and Cenozoic mammals

of North America. Columbia University Press, New York.

Tedford, R. H., X. Wang, and B. E. Taylor. 2009. Phylogenetic

systematics of the North American fossil Caninae (Carnivora:

Canidae). Bulletin of the American Museum of Natural History

325:1–218.

Van Valkenburgh, B. 1990. Skeletal and dental predictors of body

mass in carnivores. Pp. 181–205 in J. Damuth and B. J.

MacFadden, eds. Body size in mammalian paleobiology. Cam-

bridge University Press, Cambridge.

Van Valkenburgh, B., X. Wang, and J. Damuth. 2004. Cope’s rule,

hypercarnivory, and extinction in North American canids.

Science 306:101–104.

Wang, X., and R. H. Tedford. 2008. Dogs: their fossil relatives and

evolutionary history. Columbia University Press, New York.

Zachos, J., M. Pagani, L. Sloan, E. Thomas, and K. Billups. 2001.

Trends, rhythms, and aberrations in global climate 65 Ma to

present. Science 292:686–693.

BODY SIZE IN OLIGO-MIOCENE MAMMALS 661