HOL0010.1177/0959683616632894The HoloceneStegner 632894research-article2016

Research paper

The Holocene 15–­1 Stasis and change in Holocene small © The Author(s) 2016 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav mammal diversity during a period of DOI: 10.1177/0959683616632894 aridification in southeastern Utah hol.sagepub.com

Mary Allison Stegner

Abstract Biological conservation depends on understanding and disentangling the effects of decadal- to centennial-scale dynamics from the millennial-scale dynamics documented in the fossil record. The American Southwest is expected to become increasingly arid over the next few decades and will continue to experience large-scale human land use in various forms. A primary question is whether the ecological fluctuations recorded over the past few decades fall outside the range of variation expected in the absence of recent land use and management. I excavated and quantified mammal diversity change in two fossil-bearing alcoves, East Canyon Rims 2 and Rone Bailey Alcove, located in San Juan County, Utah. AMS radiocarbon dates on 33 bone samples from these sites span ~4.4–0.5 kyr and shed light on pre-industrial faunal dynamics in the region over the course of environmental change. Localities with comparable small mammal diversity have not been reported from this region previously, so these deposits provide novel insight into Holocene mammal diversity in southeastern Utah. Taxa recorded in these sites include leporids, sciurids, perognathines, arvicolines, Onychomys, Cynomys, Dipodomys, Peromyscus, Neotoma, and Thomomys. Using temporal cross-correlation, I tested for a relationship between regional temperature and species richness, evenness, relative abundance, and rank abundance. I also tested for changes in the overall taxon abundance distribution and visualized faunal relationships among time bins using non-metric multidimensional scaling of relative abundance data. None of the measures of diversity tested here were correlated with temperature change through time except for relative abundance of leporids. Overall, these results suggest that climatic fluctuations of the magnitude preserved in these deposits did not significantly alter the small mammal community, nor is there evidence that the presence, then exodus, of Native Americans from the region significantly affected small mammals.

Keywords climate change, Colorado Plateau, conservation paleobiology, diversity dynamics, late Holocene, small mammals Received 29 September 2015; revised manuscript accepted 20 December 2015

Introduction The biodiversity of the Colorado Plateau (CP; southwestern that typify ecological dynamics that play out over millennia? United States) is threatened by both land use and climate change Such questions can only be answered by tracing ecological impacts. This region is 42% federally owned land (less than 14% dynamics through thousands of years, using the natural experi- is privately owned), and over 80% of that land is multi-use, ments preserved in the fossil record (Conservation Paleobiology including mining, grazing, energy development (fossil fuel and Workshop, 2012; Hadly and Barnosky, 2009; Kidwell, 2015). solar/wind), off-road vehicle travel, timber harvesting, hiking, Biological conservation on the CP depends on understanding fishing, hunting, and other activities (Ernst and Prior-Magee, and disentangling the effects of decadal- to centennial-scale 2008). Understanding ecological dynamics on these multi-use dynamics, such as grazing and other human impacts, from the lands is crucial for preservation of the ecosystem services and millennial-scale dynamics documented in the fossil record. The intangible benefits they provide (Fu et al., 2015; Lawler et al., American Southwest is expected to become increasingly arid over 2014; Nash, 1967). While much information is becoming avail- the next few decades: conditions analogous to previous multi-year able on ecological dynamics that operate over decadal and shorter droughts, including the Dust Bowl, are expected to become the time scales, we still know little about the over-lying long-term norm (Seager et al., 2007). Under the A2 (highest) greenhouse gas dynamics that are important in conservation efforts (Conservation emissions scenario, precipitation is estimated to decline by around Paleobiology Workshop, 2012; Dietl and Flessa, 2011; Dietl 66% by 2090 and temperatures are expected to increase ~1.5–3°C et al., 2015; Hadly and Barnosky, 2009; Kidwell, 2015; Rick and by 2041–2070 and ~3–5°C by 2070–2090 (Garfin et al., 2014). Lockwood, 2013; Swetnam et al., 1999; Willis and Birks, 2006). A primary question is whether the ecological fluctuations recorded Department of Integrative Biology, University of California, Berkeley, USA over the past few decades fall outside the range of variation expected in the absence of recent land use and management. For Corresponding author: example, to what extent do current relative abundances, distribu- Mary Allison Stegner, Department of Integrative Biology, University of tion, and associations of species reflect recent adjustments of spe- California, Berkeley, 123 Laidley St, San Francisco, CA 94131, USA. cies due to anthropogenic pressures versus natural fluctuations Email: [email protected]

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016  The Holocene 2

The effects of climate on biota can take many forms: mam- Mexico, and Arizona meet. It is an arid region with high topo- mals can shift their geographic range, population abundance, graphic relief – from ~900 to 4300 m – pocked with isolated lac- physiology, body size, phenology, and so on, or, if a species’ envi- colithic mountains and creased with deep canyons formed by the ronmental tolerances are greater than the amount of environmen- Colorado River and its tributaries (Barnes, 1993; Chronic, 1990). tal change expected, they may remain observably unaffected. The CP is flanked to the west by the Great Basin, to the east by the Assessing community-level fluctuations is one way to synthesize Rocky Mountains, and to the south by the Sonoran and Chihua- these many responses and compare the whole mammal commu- huan Deserts. The western border is formed by a series of moun- nity from one time period to another. Detailed paleoecological tain ranges – the Uinta Mountains, Wasatch Range, and Fishlake records from Quaternary deposits have been remarkably useful in and Aquarius Plateaus (Rowe, 2007) – while the Mogollon Rim in characterizing these millennial-scale ecological dynamics in Arizona and the Rio Grande Rift Valley in New Mexico form the other regions (Anderson, 1993; Anderson et al., 1999, 2000; Bar- southwestern and southeastern borders, respectively (Foos, 1999). nosky et al., 2004; Betancourt, 1984; Blois et al., 2010; Graham and Grimm, 1990; Grayson, 2011; Hadly, 1996). Past studies have detected a strong relationship between faunal composition Localities and climate: in Samwell Cave (California), Blois et al. (2010) RBA and ECR2 are large, horizontally shallow alcoves in the show a correlation between warming climate and declining spe- Slickrock Entrada Sandstone cliff face of Rone Bailey Mesa, cies evenness and richness from ~17 to 1.5 kyr; Terry et al. (2011) Canyon Rims Recreation Area (BLM), San Juan County, Utah found that cave deposits in the Great Basin record proportional (Figure 1a–c). ECR2 is 1830 m (±5 m) in elevation, southeast- increases in ‘southern affinity taxa’ when climate warmed and facing, and roughly ~50 m across, ~30 m high at the mouth, and dried over the last ~7.5 kyr; and the Porcupine Cave (Colorado) ~20 m deep (Figure 1c). RBA is 1905 m (±5 m) in elevation and pit sequence spanning 1 ma to ~600 kyr revealed little community faces roughly southwest. RBA (~10 m high by ~3 m wide by ~5 m change due to climate change except that small herbivores were deep) is contained within a much larger alcove, and an apron of less diverse during warm interglacial periods (Barnosky et al., sediment roughly 10–15 m high extends from a maximum height 2004). However, while small mammals are important indicators at the midden, spilling southwest to the floor of the larger alcove of climate and environment, they do not always track vegetation (Figure 1b). Quaternary eolian sediments have accumulated in (Graham and Grimm, 1990) or temperature and precipitation pre- both ECR2 and RBA and buried bone and plant material collected dictably. For example, Lamar Cave in Yellowstone records the by roosting raptors (as evidenced by raptor pellets on the surface presence of mesic-adapted taxa during time intervals that are con- of the sediments) and woodrats (genus Neotoma). Woodrats sidered warm/arid (Hadly, 1996). At Mescal Cave in the northern remain active at both sites today, and mammalian carnivores also Mojave Desert, Stegner (2015) found that although both Neotoma may have played a role in bone accumulation in these middens. cinerea and Marmota flaviventris are considered boreal-adapted, The vegetation surrounding these sites today is a patchy mix- N. cinerea disappeared at the end of the Pleistocene while M. fla- ture of sagebrush-dominated areas, open perennial grassland, and viventris was present at the site during several thousand years of pinyon–juniper woodland. Cooler and more mesic pockets harbor Holocene warming and aridification. gambel oak (Quercus gambelii), barberry (Mahonia fremontii), Few Quaternary fossil records that sample the small mammal squawbush (Rhus trilobata), Utah service berry (Amelanchier community of the CP have been studied to date (Carrasco et al., utahensis), and spruce (Picea englemannii). At ECR2, invasive 2005; Emslie, 1986; FAUNMAP Working Group, 1994; Mead, Russian thistle (tumble weed, Salsola spp.) dominates the middle 1981; Mead and Phillips, 1981; Tweet et al., 2012), particularly in of the drainage where a primitive road passes through and forms southeastern Utah, where only a handful of Quaternary vertebrate dense patches up to approximately a meter high in places. Russian localities have been published. Most of these contain fewer than thistle is present near RBA as well, but it is sparser and does not five taxa – primarily large-bodied species – and few specimens produce the monotypic thickets seen near ECR2. The area around (Carrasco et al., 2005; FAUNMAP Working Group, 1994, 1996). Rone Bailey Mesa has been grazed by cattle and horses for over a Archaeological and fossil plant records from the Quaternary CP, century (Lavender, 1964). in contrast, have been more-extensively studied and provide an important context for new vertebrate fossil data presented here. Holocene rockshelter deposits are common across the CP (Mead Climate and vegetation history of CP et al., 2003; Tweet et al., 2012) and contain abundant small mam- Today, the CP marks the geographic transition between summer- mal remains that can be used to track the communities through wet (summer monsoon) to the south and summer-dry to the north- long time spans. To this end, I excavated and quantified mammal west (Anderson et al., 2000). Southeastern Utah currently diversity change in two fossil-bearing alcoves located in San Juan experiences summer monsoons, but the monsoon boundary shifts County, Utah. These localities, East Canyon Rims 2 (ECR2) and through time in response to temperature, snow pack in the Rocky Rone Bailey Alcove (RBA), contribute to our understanding of Mountains, location of the Jetstream, and other factors (Anderson natural variation in this system by providing faunal data for a et al., 2000). Because the CP has high topographic variability – period of recent climate change – a transition from generally cool- deep canyons and high peaks – typical precipitation and tempera- wet to warm-dry that occurred within the last 5 kyr. Accelerator ture are spatially heterogeneous. However, most of the region Mass Spectrometry (AMS) radiocarbon dates on 33 bone samples receives between 4 and 12 inches of precipitation annually and from these sites span ~4400 cal. BP to present and shed light on experiences minimum January temperatures around −10°C to pre-industrial faunal dynamics in the region over the course of −4°C (13–25°F) and maximum July temperatures between environmental change, most notably aridification. Localities with roughly 29°C and 38°C (84–100°F) (30-year normals from 1981 comparable mammal diversity have not been reported from this to 2010) (PRISM Climate Group, 2015). region previously, so these sites provide novel insight into Holo- The CP was cooler and more mesic during the last glacial cene mammal diversity in southeastern Utah. period of the Pleistocene, before ~14 kyr, at which time it was characterized by juniper woodland and sagebrush between ~600 and 1500 m; pinyon, limber pine, and Douglass fir between 1500 Study region and 1800 m; Engelmann spruce/juniper forest around 2100 m; and The CP is a physiogeographic province in North America cen- spruce/pine forest above 2700 m (Anderson et al., 2000; Cole, tered on the Four Corners Region where Utah, Colorado, New 1990). Since the last glacial maximum, the dominant tree and

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016 Stegner 3

Figure 1. (a) Map of localities and geographic landmarks mentioned in the text; regional map is inset, the dashed line indicates the rough outline of the Colorado Plateau. (b) RBA from approximately west; arrow indicates location of woodrat midden deposit. (c) ECR2 from approximately south; arrow indicates location of woodrat midden deposit. ECR2: East Canyon Rims 2 (UCMP #V36221); RBA: Rone Bailey Alcove (UCMP #V36222); ARCH: Arches National Park; CANY: Canyonlands National Park; GRCA: Glen Canyon Recreation Area; NABR: Natural Bridges Nation Monument.

shrub species – including Abies concolor (white fir), Artemisia strengthened (Weng and Jackson, 1999); this is also when the mod- spp. (sagebrush), Atriplex confertifolia (shadscale saltbush), Cer- ern monsoon boundary was established (Betancourt, 1984). This cocarpus intricatus (mountain mahogany), Ephedra spp. (Mor- cool mesic period gave way to an arid and warm mid-Holocene, mon tea), Juniperus spp. (Juniper), Opuntia spp. (prickly pear), from about 8.5 to 6 kyr (Reheis et al., 2005; Weng and Jackson, Picea spp. (Spruce), Pinus ponderosa (Ponderosa pine), Yucca 1999). From ~6 to 3 kyr, cool-wet conditions returned (Reheis et al., angustissima (narrow-leaf yucca), and many others – have gener- 2005) and fossil plant evidence from the Abajos suggests higher ally moved 700–900 m higher in elevation and 400–700 km up effective moisture before 3 kyr (Betancourt, 1984). At the end of this river (Cole, 1990). At the end of the Pleistocene, middens from period, around 3120 cal. BP, Maize agriculture arrived in this area as the Abajos and Comb Ridge (respectively ~45 and ~85 km south indicated by pollen in the Abajos records (Betancourt and Davis, of Rone Bailey Mesa) record a mixture of xeric- and mesic- 1984). Analysis of eolian and alluvial deposition in Canyonlands adapted plants; the mesic components (e.g. subalpine conifers) of National Park suggests that from 2 kyr to the present drier conditions the flora disappeared from lower elevations, but modern domi- set in, as evidenced by greater mobility of dune sand (Reheis et al., nant trees – pinyon (Pinus edulus and Pinus monophylla) and 2005). In contrast, flood deposits from Utah and Arizona suggest ponderosa (P. ponderosa) – did not appear at those elevations drier periods from 3.6 to 2.2 kyr and 0.8 to 0.6 kyr, contrasted with until the mid-Holocene (~7–3 kyr) (Betancourt, 1984; Coats et al., peak flood magnitude and frequency from 1.1 to 0.9 kyr and again 2008). Pinyon has been expanding from south to north across the after 0.5 kyr (Ely, 1997). These peak floods are linked to increases in CP since the end of the glacial period (Coats et al., 2008). winter storms and dissipating tropical cyclones rather than summer With increasing elevation in this arid region, precipitation precipitation or monsoon (Ely, 1997). Tree ring data record a mesic increases and temperature decreases on average (Betancourt, period from ~900 to 820 BP, followed by a series of at least six muti- 1984). The fossil localities are between ~1830 and 1900 m in decadal (>20 years) droughts – so-called megadroughts – that took elevation, so the plant community likely progressed from pinyon, place from ~1100 to 350 BP (Benson and Berry, 2009). The first of Douglas fir, and other conifers at the end of the Pleistocene, to these megadroughts occurred between 1087 and 1066 BP (Benson juniper woodland and sagebrush with smaller pockets of ponder- and Berry, 2009). Human settlements in the northern San Juan Basin osa and spruce over the course of the Holocene, to an assemblage (southeastern Utah and southwestern Colorado) were largely aban- dominated by sagebrush and grasslands with large pockets of doned by around 650 BP (Benson and Berry, 2009). Whether this pinyon–juniper today. Preliminary analysis of pollen from RBA exodus is attributable directly (e.g. through crop failure) or indi- has identified spruce in the deepest excavation level, around 4400 rectly (e.g. through social collapse and warfare possibly due to food cal. BP, and juniper throughout the deposit. shortage) to climatic changes is still debated, but climate-induced Climatically, the early Holocene was cooler than today, but more stress is a predominant theory today (Benson and Berry, 2009; mesic than the late Glacial because the summer monsoon was Stahle and Dean, 2010).

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016  The Holocene 4

In the last two centuries, large-scale livestock grazing on the Table 1. AMS radiocarbon and calibrated ages of bone collagen CP has transformed aspects of this region – among many impacts, samples from ECR2 and RBA. livestock break up soil crusts and destabilize soils, leading to Site Level UCMP# CAMS# 14C yr BP Cal. yr BP increased dust, soil loss over time, and exotic plant species inva- sion (Belnap, 2003; Belnap et al., 2009; Miller et al., 2011); graz- ECR2 ing also changes the plant structure including dominance by 1 233255 169707 1110 ± 30 937–1072 weedy invasive species (Belnap et al., 2009), increases wild fire 1 233180 167400 295 ± 30 291–460 potential (Belnap et al., 2009), reduces carbon storage and vegeta- 3 233188 169714 4335 ± 30 4844–4970 tion cover (Fernández et al., 2008; Jones, 2000; Miller et al., 3 233189a 169717 560 ± 30 587–644 2011), and leads to destruction of riparian habitats and waterways 3 233189a 170004 500 ± 30 501–553 as livestock congregate around water sources (Eldridge and Whit- 4 233193a 169712 905 ± 30 744–914 ford, 2009). While the midden data addressed here do not provide 4 233193a 170002 895 ± 35 735–911 clear records for faunal change between 250 BP and present, 4 233196 169719 895 ± 25 738–833 when livestock grazing began and then intensified, these localities 6 233207 167401 1065 ± 30 928–1054 elucidate what we might expect from the small mammal commu- 8 233219a 169709 1620 ± 30 1412–1569 nity in response to climate if grazing has no effect. 8 233219a 170042 1710 ± 35 1549–1704 8 233217 169711 1125 ± 30 959–1089 9 233221 170056 1185 ± 30 1050–1183 Methods 9 233222 170057 1355 ± 30 1240–1325 Excavation and identification of vertebrate fossils 10 233227a 169713 1245 ± 30 1171–1271 a Middens like RBA and ECR2 provide high-fidelity records of 10 233227 169718 1285 ± 30 1179–1285 diversity through time. These sites generally sample the small 10 233226 167402 1225 ± 30 1065–1260 vertebrate community in ~7 km radius around the site (Feranec RBA a et al., 2007; Hadly, 1999; Porder et al., 2003). Because ECR2 and 1 231186 169716 3330 ± 50 3451–3649 a RBA are ~4.7 km apart, thereby sampling an overlapping area, 1 231186 170041 3365 ± 30 3557–3693 a and they have similar taphonomic vectors, they can be treated as 1 231178 169720 >Modern 0 a representative of the same community and the stratigraphic levels 1 231178 170005 >Modern 0 a from the two sites can be combined to construct a chronological 1 231187 169721 >Modern 0 a sequence that extends from ~4.4 to 0.5 kyr. Taphonomic studies of 1 231187 170006 >Modern 0 woodrat- and owl-generated deposits reveal that they record rela- 1 231173 167403 >Modern 0 tive abundance and diversity with high accuracy (less than 1% 2 RBA02 170054 2545 ± 45 2489–2754 mismatch between modern middens and the living local commu- 2 231199 170055 1145 ± 35 972–1175 nity) (Terry, 2008, 2010a, 2010b). Deposits like these are also 4 231190 167404 2420 ± 35 2350–2700 spatially integrated, sampling small mammals from all microhab- 6 231224 169708 3455 ± 30 3640–3828 itats surrounding the deposits (Terry, 2010a). Time-averaging on 6 231230 169722 3395 ± 30 3569–3705 the scale of decades to centuries is actually an asset to studies of 6 231228 167405 3825 ± 50 4090–4410 diversity change through time because time-averaged middens AMS: Accelerator Mass Spectrometry; CAMS: Center for Accelerator sample rare species which are often missed by other survey tech- Mass Spectrometry; UCMP: University of California Museum of Paleon- niques (Barnosky et al., 2004; Hadly, 1999; Terry, 2010b), smooth tology; RBA: Rone Bailey Alcove. the effects of annual or decadal population booms and busts aSpecimens that were subsampled and dated twice. which are common in small mammals (Dickman et al., 2010; Terry, 2008; Whitford, 1976), and these sites accurately reflect the magnitude and direction of faunal baseline shifts (Terry, 2010b). Radiocarbon dating RBA and ECR2 were identified as promising deposits in 2012 I radiocarbon dated 33 bone collagen specimens (16 from ECR2 and 1 m2 test pits were excavated in each site in 2013. Excavation and 17 from RBA; Table 1) using AMS dating techniques at the levels were an arbitrary depth based on sediment volume (ranging Center for Accelerator Mass Spectrometry, Lawrence Livermore from 2.3 to 13.5 cm in thickness; Supplemental 1, available National Laboratory (Livermore, California, US). Seven of these online) because sediments were extremely unconsolidated sand specimens were subsampled and dated twice to confirm that mingled with roof fall (large sandstone blocks), and there was no results were consistent. Bone specimen preparation follows the apparent natural stratigraphy. Sediments were screened (1/2, 1/4, modified Longin method described by Brown et al. (1988) for and 1/16 inch2) and bone was picked from the 1/2- and 1/4-inch2 collagen extraction and follows methods outlined in Vogel et al. 2 screens in the field. All material collected on the 1/16-inch screen (1987) for converting CO2 into graphite for AMS analysis. AMS was bagged, taken to the University of California Museum of 14C results include a matrix-specific background correction and Paleontology (UCMP), and processed for this study. an estimate of the δ13C value of the material (Stuiver and Polach, Fossil material was separated from matrix by hand or with 1977). To convert radiocarbon years BP to calendar BP, I used the forceps. Material was initially sorted into rodent scats, insects, IntCal 13 calibration curve (Reimer et al., 2013) in OxCal Online, plant macrofossils, and vertebrate bone. Using morphological version 4.2 (Bronk Ramsey, 2009). criteria, bone was identified to class (Aves, Reptilia, Amphibia, Faunal data for excavation levels with overlapping radiocar- and Mammalia), and mammalian bone was identified to species bon age ranges (Figure 2) were combined, and I developed a chro- when possible (primarily craniodental remains), or to family or nology of six time bins with different degrees of time averaging, genus (in the case of postcranial or otherwise non-diagnostic dating from 4410 to 501 cal. BP. fossils), by direct comparison with modern skeletal specimens of known species held at the Museum of Vertebrate Zoology (MVZ) and UCMP or using published descriptions from pri- Sample standardization mary paleontological and mammalogical literature. All speci- Taxon abundance was quantified as number of individual speci- mens are curated at the UCMP. See Supplemental 2 (available mens (NISP) of craniodental material identified to family (Lepo- online) for Systematic Paleontology. ridae, Sciuridae), subfamily (Perognathinae, Arvicolinae), or

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016 Stegner 5

Figure 2. Calibrated age (cal. yr BP) of bone collagen samples from ECR2 and RBA. Diamonds indicate mean cal. yr BP, and vertical lines indicate the age range. Brackets indicate the excavation levels that were combined to produce time bins (letters). ECR2: East Canyon Rims 2; RBA: Rone Bailey Alcove. genus (Onychomys, Cynomys, Dipodomys, Peromyscus, Neo- product moment correlation = 0.017, p = 0.97; standardized rich- toma, and Thomomys). As compared with minimum number of ness: Pearson’s product moment correlation = –0.276, p = 0.60). individuals (MNI), NISP has been demonstrated to be a less- All analyses were performed using the R Project for Statistical biased indicator of relative importance of a taxon in the commu- Computing version 3.2.0 (R Core Team, 2015) and with the vegan nity (Blois et al., 2010; Grayson, 1978). Species-level package (Oksanen et al., 2015), DescTools package (Signorell identifications within these taxa are often only possible by com- et al., 2015), and SQS function (Alroy, 2010). parison of skeletal elements that are seldom well represented in the fossil record. Analysis at taxonomic levels above the species level allows incorporation of more of the fossil data represented Analysis of diversity change in these deposits, improving abundance estimations by increasing I assessed diversity change through time in several different ways. sample size. Furthermore, analysis at the genus, family, and sub- First, I used pairwise Fisher’s tests to compare the relative abun- family levels provides the requisite information needed to recog- dance distribution from one time bin to the next (with a Monte nize the relative changes in the small mammal taxa that fill Carlo p-value simulation, 1000 replicates, and both with and different functional roles in the ecosystem (Barnosky, 2004; Bar- without a Holm p-value correction). Second, I calculated Hurl- nosky et al., 2004; Blois et al., 2010; Grayson, 2000; Terry et al., bert’s probability of interspecific encounter (PIE) for each time 2011) and which can be used as appropriate metrics of the envi- bin and compared across time bins. PIE is a measure of species ronmental state (Badgley and Fox, 2000; Barnosky and Shabel, evenness and is independent of sample size, unlike many other 2005; Barnosky et al., 2004; Stegner and Holmes, 2013). commonly used diversity metrics, for example, Shannon and I calculated relative abundance as NISP of the taxon divided by Simpson indices (Blois et al., 2010; Gotelli and Ellison, 2013; the total number of specimens for each time bin. I then estimated Hurlbert, 1971). Because PIE is not biased by sample size, resam- 95% confidence intervals around each relative abundance estimate pling is not necessary for estimating the uncertainty in the metric. using Goodman’s (1965) simultaneous confidence interval calcu- Instead, I followed the equation presented in Davis (2005) for lations (Calede et al., 2011; McHorse et al., 2012), implemented in computing variance in PIE. Because time bins are likely to be the DescTools package in R (Signorell et al., 2015). serially autocorrelated and therefore non-independent, I com- Species richness is correlated with sample size, so I standard- pared the overlap in variance to determine whether PIE was sig- ized richness using shareholder quorum subsampling (SQS) at a nificantly different among time bins (Blois et al., 2010). coverage level of 0.6 (Alroy, 2010). While the goal of traditional I visualized faunal relationships among time bins using non- rarefaction is uniform sampling, the purpose of SQS is ‘fair’ sam- metric multidimensional scaling (NMDS) of relative abundance pling that is reflective of the true standing diversity. In SQS, sam- data. Unlike other ordination techniques (e.g. principal compo- ple size is not fixed; rather, SQS fixes coverage, the proportion of nents or principal coordinates analysis), NMDS does not preserve the entire frequency distribution represented by the species in the true distances between objects (time bins or taxa in this case) and resample. Rarefaction dampens the richness of very diverse sam- instead uses rank order of the distances between objects to arrange ples, so relative diversity estimates are not linear, whereas in SQS them in multidimensional space. This technique places more dif- a site that is twice as diverse as another has an SQS richness value ferent objects further apart and more similar objects closer that is twice as large (Alroy, 2010). Estimated richness values together, which is not always the case in other ordination tech- produced by SQS appear low relative to raw richness, but are lin- niques (Gotelli and Ellison, 2013). early proportional to one another. Rank-order abundance was scaled from 1 to 10, to account for Estimation of confidence intervals is one way to correct for time bins where not all species were represented or when some differences in time bin duration, but longer time bins inevitably taxa had the same number of specimens. I made pairwise com- have more opportunity to sample more species, and low sample parisons of rank-order abundance between time bins using Spear- size, as in bins F and A, can lead to a randomly biased NISP and man rank correlation tests with and without a Holm p-value abundance distribution structure. I tested for a correlation between correction. The Spearman coefficient ranges from −1, indicating duration and both raw and standardized richness per time bin and an inverse relationship, to 1, indicating perfect correspondence found no relationship between the two (raw richness: Pearson’s (Terry, 2010a). To determine whether particular taxa were

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016  The Holocene 6 consistently higher or lower in rank than the others, I compared level 3 were not dated and were also excluded because it is not the ranks of taxa using pairwise Mann–Whitney U-tests with and known whether they are highly mixed, and it appears that younger without Holm p-value corrections. levels may be more mixed than older levels. Bioturbation by peo- ple, cattle, and rodents is one possible explanation for the presence of older bones in top excavation levels. Second, while there is some Climate correlation mixing in these sites, dates generally get older with increasing To evaluate the role of climate across all six time bins, I used recon- depth, suggesting it is reasonable to assume that the majority of the structed mean July temperature (MJT) for the Southwestern United material in each level has not moved dramatically up or down in the States from Viau et al. (2006), downloaded from the National Cli- deposit. Combining neighboring excavation levels into single anal- matic Data Center (http://www.ncdc.noaa.gov). These reconstruc- ysis units helps to strengthen this assumption. However, interpreta- tions are based on 752 fossil and 4590 modern pollen records from tion of the age of individual specimens is not possible without a across the continent and estimate MJT in 100-year intervals from date on the specimen itself. 14 kyr to the present (Viau et al., 2006). Following the methods described in Terry and Rowe (2015) and Rowe and Terry (2014), I fit the climate data with a spline smoothing function (R function Relative and rank-order abundance spline: stats, method: natural) and then analytically averaged the cli- Abundance of taxa fluctuates across the record (Table 2, Figure 3), mate data over the period of time encompassed by each time bin. I but 95% confidence intervals demonstrate that there is no signifi- then used cross-correlation to test for a relationship between climate cant difference in relative abundance through time within each and species richness, evenness, and relative and rank abundance of taxon and few differences among taxa (Figure 3). In time bin A, each taxon. The significance of the correlations was tested using observed relative abundances of arvicolines and Thomomys were at permutation tests (500 iterations). To evaluate lagged correlations, I their highest, while leporids, perognathines, Cynomys, and Ony- shifted the climate data by 10-year increments, up to 200 years, each chomys were not sampled at all. During B, leporids, perognathines, time re-averaging the fitted climate data and recalculating the cor- Cynomys, and Onychomys first appeared, and perognathines were relation (Terry and Rowe, 2015). at their most abundant across the record. Dipodomys also increased in abundance, while sciurids, arvicolines, Peromyscus, Neotoma, and Thomomys all declined. In bin C, leporids, Cynomys, Neotoma, Results Onychomys, Peromyscus, and arvicolines all increased, while sci- Radiocarbon chronology urids, Thomomys, Dipodomys, and perognathines declined. Lepo- rids and Cynomys continued to increase in bin D, along with slight Radiocarbon chronologies for each site show that most dates were increases in Neotoma, Onychomys, Peromyscus, and sciurids. stratigraphically consistent, with levels ranging from 143 (ECR2 Dipodomys and Thomomys continued to decline, Thomomys reach- level 3) to 841 (RBA level 6) years of time averaging (Table 1, ing its lowest abundance across the record, and arvicolines also Figure 2). Dates presented in Figure 2 are derived from the most decreased. Cynomys was again absent during bin E; Neotoma likely age for each specimen. However, there is a range of possible declined while Onychomys and Peromyscus slightly increased. 14 calibrated cal. BP for a given C date (Figure 2); I used the proba- After relatively low abundance in the preceding time bin, Thomo- bility distribution associated with each date and the overlap in pos- mys sharply increased in E. Perognathines, Dipodomys, and sci- sible calibrated ages to determine which excavation levels could urids also increased while leporids slightly declined. Bin F is reasonably be considered non-overlapping in age. The three older characterized by extremes: leporids, Thomomys, and Neotoma were time bins pertain to RBA, while the younger are from ECR2. Time at their highest abundance, but Onychomys, Cynomys, and arvico- bin A, the oldest, represents excavation level 6 from RBA and has lines were entirely absent. Dipodomys declined slightly, and Pero- peak age probabilities at 3699.5, 3639.5, and 4184.5 cal. BP. RBA myscus, sciurids, and perognathines maintained low abundance. level 5 was not dated, but dates for levels 6 and 4 are consistent and The rank-order abundances are all positively correlated across non-overlapping, so here I assume that stratigraphic mixing in these time bins (Figure 4, Table 3). Without the Holm p-value correc- older layers is either not present or very minimal and that level 5 tion for multiple statistical tests, the rank-order abundances of B, can be treated as a separate faunal unit (time bin B). However, addi- E, and F are significantly correlated with one another; addition- tional radiocarbon dates are necessary to confirm this assumption. ally, C is correlated with A and E, and D is correlated with F. All Level 4 encompasses time bin C with peak age probability at of these comparisons have correlation coefficients over 0.60. 2424.5 cal. BP. Time bin D represents ECR2 excavation levels With a Holm correction, only E versus C and F versus E are sig- 8–10, with peaks at 1024.5, 1169.5, 1179.5, 1194.5, 1214.5, 1289.5, nificantly correlated. The Mann–Whitney U-tests (Table 4, Figure 4) 1529.5, and 1609.5 cal. BP. Time bin E represents ECR2 levels 4–6 show that, without a p-value correction, Dipodomys and Thomo- and peaks at 794.5, 789.5, and 959.5 cal. BP. Time bin F is repre- mys are significantly more rank abundant (more common taxa sented by ECR2 level 3. The oldest date from ECR2, 4970–4844 have lower rank abundance values; that is, rank of 1 is the first cal. BP, is from excavation level 3 and is ~2.6 kyr older than the most common taxon) than all taxa except for leporidae and Neo- next oldest date (from level 8). If this is excluded, bin F/level 3 has toma (only Dipodomys is significantly higher in rank than Neo- peaks at 524.5 and 544.5 cal. BP. While there is no evidence that toma). Arvicolines are significantly less rank abundant than all this particularly old date is inaccurate, either as a result of contami- other taxa except for Onychomys, perognathines, and sciurids; nation in the lab or in the deposit, these kinds of contamination are Cynomys is significantly less rank abundant than all taxa except extremely difficult to diagnose. However, all other dates for ECR2 Onychomys, perognathines, and arvicolines. No comparisons are are consistently younger than 1704 cal. BP, and so this date of significant when a Holm p-value correction is used. nearly 5 kyr in the middle of the deposit has a few important impli- cations. First, it suggests that stratigraphic mixing is present to a certain degree in this site, but that it occurs rarely because dates Diversity dynamics from other levels tend to be consistent and arranged in a predictable The abundance distribution of bin D was significantly different chronological order. ECR2 level 1 and RBA levels 1 and 2 both (Fisher’s test without p-value correction) from all time bins have wider age ranges and show more mixing than the deeper/older except E (Table 3). Bins B and F were also significantly different excavation levels. Those levels have not been included in these from one another. NMDS (Figure 5) illustrates the relative degree analyses because the magnitudes of time averaging they encompass of compositional similarity among time bins, although no time entirely overlap deeper excavation levels. ERC2 level 2 and RBA bins form tight groups with one another. Sample size is low in

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016 Stegner 7 only; the same order applies for all taxa. for applies the same order only; Dipodomys of time bins is illustrated for The order 95% confidence intervals. simultaneous Goodman’s bars are Vertical time. e abundance of each taxon through Relativ

Figure 3. Figure

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016  The Holocene 8

Table 2. Number of individual specimens (NISP) (above) with relative abundance (taxon NISP/total NISP) and 95% Goodman’s simultaneous confidence intervals (italics) (below); uncorrected species richness and standardized (SQS) richness.

Taxon A B C D E F

Arvicolinae 1 1 4 1 2 0 0.08 (0–0.63) 0.01 (0–0.2) 0.05 (0.01–0.24) 0.02 (0–0.28) 0.01 (0–0.13) 0 (0–0.3) Cynomys 0 1 2 3 0 0 0 (0–0.57) 0.01 (0–0.2) 0.02 (0–0.2) 0.06 (0.01–0.33) 0 (0–0.1) 0 (0–0.3) Dipodomys 4 35 30 11 32 6 0.31 (0.05–0.78) 0.45 (0.25–0.67) 0.35 (0.18–0.57) 0.22 (0.07–0.5) 0.22 (0.11–0.38) 0.15 (0.03–0.48) Leporidae 0 3 4 12 22 11 0 (0–0.57) 0.04 (0.01–0.24) 0.05 (0.01–0.24) 0.24 (0.08–0.52) 0.15 (0.07–0.31) 0.28 (0.09–0.6) Neotoma 1 4 10 6 11 5 0.08 (0–0.63) 0.05 (0.01–0.25) 0.12 (0.04–0.33) 0.12 (0.03–0.4) 0.08 (0.02–0.22) 0.13 (0.03–0.46) Onychomys 0 2 7 2 9 0 0 (0–0.57) 0.03 (0–0.22) 0.08 (0.02–0.28) 0.04 (0–0.3) 0.06 (0.02–0.2) 0 (0–0.3) Perognathinae 0 5 3 0 5 1 0 (0–0.57) 0.06 (0.01–0.27) 0.04 (0–0.22) 0 (0–0.25) 0.03 (0.01–0.16) 0.03 (0–0.34) Peromyscus 1 3 7 4 20 1 0.08 (0–0.63) 0.04 (0.01–0.24) 0.08 (0.02–0.28) 0.08 (0.01–0.35) 0.14 (0.06–0.29) 0.03 (0–0.34) Sciuridae 1 6 2 8 3 3 0.08 (0–0.63) 0.08 (0.02–0.29) 0.02 (0–0.2) 0.16 (0.04–0.44) 0.02 (0–0.14) 0.08 (0.01–0.4) Thomomys 5 18 16 4 42 12 0.38 (0.08–0.82) 0.23 (0.09–0.46) 0.19 (0.07–0.41) 0.08 (0.01–0.35) 0.29 (0.16–0.46) 0.31 (0.11–0.63) Uncorrected richness 6 10 10 9 9 7 SQS richness 2.567 2.261 3.183 3.97 3.972 2.712

SQS: shareholder quorum subsampling.

Figure 4. Boxplot of rank-order abundance ranks across time bins for each taxon. bins A (n = 13) and F (n = 39), which could have a meaningful (r = 0.83, p = 0.026). Because sample sizes are low, particularly in impact on the results of the Fisher’s tests and on the arrangement bins A and B, there is high degree of uncertainty around the relative of time bins and taxa in the NMDS. abundance of each taxon, illustrated in Figure 3. Additional speci- PIE (Figure 6) was relatively high in time bins A (4410–3569 mens could change these results by changing the relative abundance cal. BP), C (estimated 2700–2350 yr. BP), E (1054–735 cal. BP), distribution for each time bin. Additional specimens are more likely and F (644–501 cal. BP). It reached a peak during bin D (1704– to change the rank abundance for less common taxa (ranks above 3) 959 cal. BP), but was significantly lower than other times during because the magnitude of difference in NISP is smaller among the bin B (3569–2700 cal. BP). Variance is non-overlapping for all less common taxa than among the most common taxa. In short, time bins, except A and F overlap. larger sample size is unlikely to change the fact that Dipodomys, Thomomys, and leporids are generally the most common taxa. Climate correlations There were no significant correlations between MJT and diversity Discussion using temporal cross-correlation except for relative abundance of Overall, neither diversity nor abundance in ECR2 and RBA cor- leporids, which had the highest correlation coefficient at lag = 0 relates with the climate record assessed here. There were no

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016 Stegner 9

Table 3. Correlation coefficients from pairwise Spearman rank Chapman and Willner, 1978). It is not generally possible to distin- correlation tests on rank-order abundance of taxa for each time bin guish Lepus from Sylvilagus using morphological criteria or size (above); Fisher’s test uncorrected p-values, for pairwise comparison of of loose teeth (Robinson and Matthee, 2005; Stegner, 2015), so it abundance distribution among time bins (below, in italics). is unclear whether the increase in leporids in RBA and ECR2 is ABC DE driven by one genus or whether both became more abundant through time. Climate was almost certainly becoming more arid B 0.62 at these sites after about 2 kyr (Cook and Krusic, 2004; Pederson 0.54 et al., 2011; Reheis et al., 2005) and local changes in the plant C 0.65* 0.46 community were likely taking place, which may have favored 0.41 0.08 leporids. Identification of the pollen from these sites would shed D 0.32 0.46 0.29 light on this question. 0.00* 0.01* 0.00** A second pattern is qualitatively evident in the relative abun- E 0.52 0.64* 0.81* 0.55 dance data (Table 2): when Thomomys have lower relative abun- 0.72 0.30 0.11 0.08 dance, Neotoma and Peromyscus are higher in abundance. Also, F 0.53 0.76* 0.51 0.73* 0.82** as Dipodomys declines through time, leporids increase. How- 0.63 0.05* 0.29 0.01* 0.49 ever, Dipodomys and Thomomys remain the most common (first and second most abundant, respectively) throughout the record, *Statistical significance without a Holm p-value correction. except bin D (leporids are first in rank abundance, Dipodomys **Significance with a Holm correction. are second, and Thomomys are fifth) and bin F (Thomomys are first, leporids are second, and Dipodomys are third). Terry et al. significant differences in abundance of taxa throughout the (2011) found that granivores tended to increase in abundance as record, from ~4.4 to 0.5 kyr. Most of the taxa present in these climate warmed: of the two granivores at ECR2 and RBA, deposits are tolerant or characteristic of xeric ecosystems, and Dipodomys actually decline in abundance as climate warmed, only one taxon present in the fossil record is not found in the and perognathines appear to fluctuate at low abundance indepen- immediate vicinity of the caves today (voles, genus Lemmiscus dently of climate. Abiotic affinities do not seem to be related to and Microtus). this relative abundance pattern: Dipodomys are physiologically Species of Microtus are typically found in grasslands and adapted to dry environments (Vimtrup and Schmidt-Nielsen, meadows, often associated with montane and/or riparian ecosys- 2005) and are found in a range of arid and semi-arid habitats tems (Kays and Wilson, 2009). In the fossil record, Microtus (Garrison and Best, 1990; Kays and Wilson, 2009); Dipodomys tends to be associated with cooler and more mesic times (Bar- ordii, the only species found in northern San Juan County today, nosky, 2004; Hadly, 1996; Hadly and Barnosky, 2009). In con- is almost always associated with sandy soils (Garrison and Best, trast, the sagebrush vole, Lemmiscus curtatus, is found in more 1990). The two species of Thomomys present in Utah today are arid brushy environments – usually sagebrush or rabbit brush Thomomys talpoides, found generally in the mountains and high (Carroll and Genoways, 1980). L. curtatus is restricted to the valleys of Utah (Durrant, 1952), and Thomomys bottae, generally Great Basin today but is found in late Pleistocene records from found in agricultural fields (such as alfalfa), lower valleys, and across the CP (Mead et al., 2003; Murray et al., 2005). Teeth of L. desert mountains (Durrant, 1952; Verts and Carraway, 1999). curtatus are present in ECR2 level 4 (time bin E, 1054–735 cal. However, both have extremely wide geographic, elevational, and BP) and RBA level 4 (time bin C, estimated 2700–2350 ybp). To ecological ranges (Durrant, 1952; Jones and Baxter, 2004; Verts my knowledge, these are the youngest records of L. curtatus on and Carraway, 1999), and so their present-day distribution in the CP (Bell and Glennon, 2003; Carrasco et al., 2005; FAUN- Utah is not necessarily a good indicator of their fossil distribu- MAP Working Group, 1994, 1996; Mead et al., 2003; Neotoma tion or habitat preferences. Paleoecological Database, 2013): other occurrences of this spe- Peromyscus and Neotoma are also fairly generalist in their cli- cies on the CP in the last 40 kyr are Sheep Camp Shelter (late matic requirements and are dietary generalists as well (Blois Glacial/Holocene; Gillespie, 1985), Screaming Neotoma Cave et al., 2010; Sorensen et al., 2004). Certain species within these (29–25 kyr; Bell and Glennon, 2003; Mead et al., 2003), and Isleta genera are specialists, however – for example, Peromyscus truei Cave No. 2 (late Glacial/Holocene; Harris and Findley, 1964). (pinyon mouse) is found associated with pinyon (Hoffmeister, As MJT increased, leporid relative abundance also increased. 1981), and Neotoma stephensi is associated with juniper or, occa- However, the 95% confidence intervals around relative abun- sionally, other conifers (Sorensen et al., 2004). Finer taxonomic dance of leporids overlap for all time bins, so it is possible that the resolution of the fossil data would certainly aid in interpretation observed increase is a sampling artifact. Sylvilagus and Lepus are of abundance dynamics, for example, by revealing intra-generic both ubiquitous across the CP and in western North America gen- turnover, but isolated teeth of Peromyscus and Neotoma cannot be erally and, as genera, are tolerant of a wide range of habitats identified to species using standard morphological criteria (Gray- (Kays and Wilson, 2009). In a study of Chihuahuan Desert Lepus son, 1988; Hooper, 1957; Mead and Phillips, 1981; Mead et al., californicus, Hernández et al. (2011) found a correlation between 2003; Repenning, 2004; Stegner, 2015). Using the dental mor- leporid abundance and both precipitation and plant productivity, phological criteria developed by Hooper (1957), I attempted to and conclude that food supply has a strong effect on L. californi- identify Peromyscus teeth from ECR2 and RBA to species. How- cus abundance. In contrast, Lightfoot et al. (2010) also assessed ever, almost all teeth were morphologically aligned with at least leporid abundance dynamics in the Chihuahuan Desert but found two of four species present in this region today: Peromyscus man- no correlation between Sylvilagus audubonii or L. californicus iculatus, Peromyscus crinitus, Peromyscus boylei, and P. truei abundance and precipitation or productivity. Rather, Lightfoot (Supplemental 3, available online). et al. (2010) observed higher diversity of both species in black The few sciurid specimens from ECR2 and RBA are almost gramma grassland as compared with creosote bush shrubland. L. exclusively loose cheek teeth, and the high degree of wear on californicus favors areas with sparse grass and vegetation cover, these teeth makes identification to genus challenging, with the including heavily grazed areas, but avoids tall grasses and forests exception of Cynomys which is more hypsodont than other sci- (Best, 1996). In contrast, the species of Sylvilgaus present in urids and larger than all other sciurids except Marmota. Today, southeastern Utah today (S. audubonii and Sylvilagus nuttali) are the sciurids present in southeastern Utah include the burrowing more commonly associated with brushy areas (Chapman, 1975; Cynomys gunnisoni, Ammospermophilus leucurus (white-tailed

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016  The Holocene 10

Table 4. The p-values from pairwise Mann–Whitney tests.

Dipodomys Thomomys Neotoma Leporidae Peromyscus Sciuridae Onychomys Perognathinae Arvicolinae

Dipodomys (1.83) Thomomys (2.08) 0.796 Neotoma (4.25) 0.006* 0.062 Leporidae (4.58) 0.117 0.100 1.000 Peromyscus (5.25) 0.005* 0.023* 0.166 0.806 Sciuridae (5.50) 0.007* 0.029* 0.625 0.572 0.936 Onychomys (7.33) 0.005* 0.008* 0.016 0.147 0.063 0.257 Perognathinae (7.33) 0.005* 0.008* 0.029 0.106 0.075 0.261 1.000 Arvicolinae (7.92) 0.004* 0.007* 0.015* 0.042* 0.041* 0.168 0.288 0.518 Cynomys (8.92) 0.005* 0.005* 0.005* 0.010* 0.005* 0.043* 0.064 0.147 0.325

*Significance without a Holm correction. No pairwise comparisons were significant with a Holm correction. Mean rank for each taxon is in parentheses.

Figure 5. Plot of non-metric multidimensional scaling of relative abundance. Bold letters indicate time bins. antelope squirrel), Otospermophilus variegatus (rock squirrel), a subgenus-level, is unlikely to be a good indicator of climate: and Xerospermophilus spilosoma (spotted ground squirrel); prairie dogs can persist in semi-desert (C. (L.) leucurus), grass- largely ground-dwelling Tamias mimimus (least chipmunk), lands and prairies (C. (L.) parvidens, C. (C.) ludovicianus), as Tamias quadrivitattus (Colorado chipmunk), and Tamias rufus well as montane meadows (C. (L.) gunnisoni, C. (L.) leucurus) (Hopi chipmunk); and tree-dwelling Sciurus aberti (Abert’s (Kays and Wilson, 2009). squirrel). All of these species, with the exception of S. aberti, are Two additional arid-adapted taxa are present throughout the found in arid desert and scrubland habitats today (Belk and Smith, deposits in relatively low abundance. Today, perognathines – 1991; Best et al., 1994; Burt and Best, 1994; Kays and Wilson, genus Perognathus and Chaetodipus – are generally found in arid 2009; Nash and Seaman, 1977; Oaks et al., 1987; Streubel and or semi-arid grasslands, desert scrub, or shrub steppe, but several Fitzgerald, 1978; Verts and Carraway, 2001). I have trapped both species also occupy woodlands and chaparral (Kays and Wilson, Tamias and Ammospermophilus in patches of pinyon–juniper 2009). Only Perognathus is present on the CP today. Onychomys woodland near these sites. S. aberti is present in the Abajos today also occupies arid and semi-arid regions of North America: Ony- (Schaefer, 1991) but not outside montane, forested regions (Kays chomys leucogaster, the only species currently found on the CP, and Wilson, 2009) – the presence of S. aberti in the fossil record prefers shrub steppes and grasslands (McCarty, 1975, 1978). O. of ECR2 and RBA might indicate cool/mesic climatic conditions. leucogaster can be distinguished from its only sister species, Ony- The sciurid teeth recovered from ECR2 and RBA are likely too chomys torridus – found in the Sonoran and part of the Chihua- small to be Sciurus, but further research will be necessary to con- huan Deserts – using length and width dimension of the m1 and firm. The Cynomys recovered from RBA and ECR2 are white- m3 (Carleton and Eshelman, 1979). However, because there is tailed prairie dogs, subgenus Leucocrossuromys, which still overlap in the measurements for these species (Carlton and Eshel- persist less than 1 km southeast of ECR2 today. Cynomys, even on man, 1979), a sample size larger than what is available at ECR2

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016 Stegner 11

Figure 6. PIE index for each time bin; vertical black bars and numbers are the variance associated with each PIE value. Letters indicate time bin and horizontal black bars are the age range (cal. yr BP). Time bin B was not dated, so the age range is unknown. The gray line and points are reconstructed mean July temperature anomaly from Viau et al. (2006). PIE: probability of interspecific encounter. and RBA is required. Furthermore, many mammal species change 3569–2700 BP; cool and possibly more wet). At that time, relative their body size in response to climate (Smith and Betancourt, abundance of Dipodomys was extremely high, and Neotoma and 1998), so morphological criteria are preferable to size for making Peromyscus were at their lowest abundance. Evenness reached a identifications (Bell et al., 2010). peak in time bin D, when the three most common taxa (Dipodo- Taxa with similar relative abundance dynamics group closer mys, Thomomys, and leporids) all had average relative abundance. together in the NMDS results (Figure 5). Two close taxonomic Although the uncertainly around PIE does not overlap for most groups are evident: (1) Dipodomys and Thomomys (highest rank- time bins, the magnitude of evenness change is low – from a maxi- order abundance across the record, higher relative abundance in the mum PIE of 0.85 in D to a minimum of 0.73 in B, a change of only beginning and end of the record, lower in the middle), and (2) Neo- 0.12 – supporting the conclusion that small mammal abundance at toma, Peromyscus, and sciurids (higher in the beginning, then an these sites was not detectably altered by the changes in climate that early dip in abundance followed by an increase, then yet another occurred during the interval of deposition. dip). These groupings do not seem to be connected to climatic affin- Other factors not tested in this study, like Ancestral Puebloan ity because a mixture of taxa with xeric, mesic, or neutral affinity is population size and land use, may influence small mammal diver- found in each grouping and abundance does not correlate with MJT. sity dynamics. However, Fisher’s tests reveal that only time bin D Desert taxa might be expected to respond more to aridity than to was significantly different from the others in terms of the relative temperature (Stapp, 2010). Several lines of evidence indicate that abundance distribution across the record. Furthermore, the taxo- conditions were relatively cool and wet during time bins A–C and nomic and functional components of this community have not then more arid and warm during D–F (Anderson et al., 2000; changed markedly through the environmental changes repre- Betancourt, 1984; Betancourt and Davis, 1984; Coats et al., 2008; sented by the time sampled by the deposits, suggesting that these Reheis et al., 2005). Bins E and F encompass a period of time when groups persist despite natural variation and/or external perturba- at least six ‘megadroughts’ took place, including the droughts that tion in this system. However, finer taxonomic resolution might are thought to have dispersed the Ancestral Puebloans at Chaco, reveal a signature of climate change: for example, Thomomys Mesa Verde, Kayenta, and other large CP settlements (Benson and subgenus Megascapheus is present in ECR2 and RBA throughout Berry, 2009; Benson et al., 2007). However, time bins with more the time period encompassed by this study, but the only specimen similar climate (A–C vs D–F) do not group together on the NMDS, of Thomomys subgenus Thomomys is from ECR2 level 3 (time supporting the conclusion that relative and rank-order abundance bin F, 644–501 cal. BP). T. subgenus Thomomys is better able to dynamics are disconnected from climate, at least with regard to the burrow in harder soils than T. subgenus Megascapheus, and soils broad-scale trends discussed here. One possible explanation is that tend to get harder as climate dries (Marcy et al., 2013). While time-averaged stratigraphic layers capture short cool or mesic peri- Thomomys abundance has remained high relative to other taxa for ods within an overall aridifying and warming trend (Hadly, 1996). thousands of years, species and subspecies of Thomomys natu- However, sample sizes for these cool/mesic taxa should be lower rally replace one another as climate changes (Hadly and Bar- than for warm/xeric taxa if taphonomic processes are the same dur- nosky, 2009). ing wet and dry times. Greater sample sizes at RBA and ECR2, particularly at the Evenness, as measured using the PIE index, is not significantly beginning and end of the record (time bins A and F), could alter the correlated with climate. PIE is notably low during bin B (estimated observed abundance patterns and might reveal a correlation

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016  The Holocene 12 between diversity and climate. However, the results reported here Funding are in keeping with other studies of small mammal dynamics I am grateful to the following sources for funding this research: through the Holocene in western North America. For example, at the Canyonlands Natural History Association, UC Berkeley Samwell Cave in northern California, Blois et al. (2010) found that Department of Integrative Biology, Sigma Xi, Sigma Xi-Berkeley small mammal evenness and richness dropped during the Pleisto- Chapter, the Paleontological Society, the Geological Society of cene–Holocene transition, but during the last ~5 kyr both measures America, the UC Museum of Paleontology, Kappy Wells and the of diversity were fairly stable. At Homestead Cave in northern TIDES Foundation, and NSF GRFP Grant DGE-1106400. Utah, Terry and Rowe (2015) showed that energy dynamics, in terms of which small mammal guilds are dominant, changed dra- matically at the end of the Pleistocene and again in the last century, References but were stable throughout the Holocene. In contrast, Rowe and Alroy J (2010) Fair sampling of taxonomic richness and unbiased Terry (2014) compared small mammal diversity dynamics in estimation of origination and extinction rates. In: Alroy J and Homestead Cave and Two Ledges Chamber, a site in northern Hunt G (eds) Quantitative Methods in Paleobiology (Paleon- Nevada, and demonstrated that during the same time interval tological society papers 16). Boulder, CO: Paleontological (8 kyr to the present), species that were shared between sites had Society, pp. 55–80. entirely different relative abundance dynamics. This suggests that Anderson RS (1993) A 35,000 year vegetation and climate his- local conditions may have a larger influence on relative abun- tory from Potato Lake, Mogollon Rim, Arizona. Quaternary dances at a site than does regional climate (Stegner, 2015). Research 40: 351–359. Anderson RS, Betancourt JL, Mead JI et al. (2000) Middle- and late-Wisconsin paleobotanic and paleoclimatic records from Conclusion the southern Colorado Plateau, USA. Palaeogeography, Pal- Two newly excavated late Holocene fossil midden localities, aeoclimatology, Palaeoecology 155: 31–57. ECR2 and RBA, with 33 AMS 14C dates provide fine-resolution Anderson RS, Hasbargen J, Koehler PA et al. (1999) Late Wis- information on small mammal diversity dynamics from ~4.4 to consin and Holocene subalpine forests of the Markagunt Pla- 0.5 kyr in southeastern Utah. These sites comprise a chronology teau of Utah, southwestern Colorado Plateau, U.S.A. Arctic, of six time bins. Regional temperature increased across the record, Antarctic, and Alpine Research 31: 366–378. and ‘megadroughts’ marked the time period from ~1100 to 350 Badgley C and Fox DL (2000) Ecological biogeography of North BP. I tested for an effect of temperature on community evenness American mammals: Species density and ecological structure and abundance of 10 small mammal taxa: leporids, perognathines, in relation to environmental gradients. Journal of Biogeogra- small sciurids, arvicolines, Cynomys, Neotoma, Dipodomys, Ony- phy 27: 1437–1467. chomys, Peromyscus, and Thomomys. Despite increases in tem- Barnes FA (1993) Geology of the Moab Area. Moab, UT: Canyon perature, measured as reconstructed MJT anomaly, neither is Country Publications. significantly correlated with relative abundance, rank abundance, Barnosky AD (2004) Faunal dynamics of small mammals through species richness, or the PIE evenness index. Although abundance the Pit sequence. In: Barnosky AD (ed.) Biodiversity Response does not correlate with climate, certain groups of species have to Climate Change in the Middle Pleistocene. Berkeley, CA: similar diversity trends through time. Identifying what drives University of California Press, pp. 318–326. these cross-taxa abundance patterns will be important for predict- Barnosky AD and Shabel AB (2005) Comparison of mammalian ing future abundance changes. species richness and community structure in historic and mid- Overall, these results indicate that climatic fluctuations of the Pleistocene times in the Colorado Rocky Mountains. Pro- magnitude preserved in these deposits do not significantly alter ceedings of the California Academy of Sciences 56: 50–61. the small mammal community at least on the taxonomic level Barnosky AD, Bell CJ, Emslie SD et al. (2004) Exceptional assessed here. Neither is there evidence that the presence, then record of mid-Pleistocene vertebrates helps differentiate cli- exodus, of Native Americans in the region significantly affected matic from anthropogenic ecosystem perturbations. Proceed- small mammals. This lends support to the findings of other stud- ings of the National Academy of Sciences of the United States ies, like Terry and Rowe (2015), that small mammal communities of America 101: 9297–9302. of western North America were stable throughout the Holocene, Belk MC and Smith HD (1991) Ammospermophilus leucurus. but have undergone dramatic changes in structure in the last Mammalian Species 368: 1–8. 100 years as a result of human activities leading to large-scale Bell CJ and Glennon J (2003) Arvicoline rodents from Screaming land cover changes, like invasive annual grasses. The metrics of Neotoma Cave, Southern Colorado Plateau, Apache County, community structure assessed here can be monitored in the mod- Arizona, with comments on the Pleistocene biogeography of ern small mammal community of this region: if these metrics are Lemmiscus curtatus. In: Schubert BW, Mead JI and Graham significantly different today or in the future from those seen in the RW (eds) Ice Age Cave Faunas of North America. Blooming- fossil record, it may indicate anthropogenic influences outside the ton, IN: Indiana University Press, pp. 34–63. range of natural variability. Bell CJ, Gauthier JA and Bever GS (2010) Covert biases, circu- larity, and apomorphies: A critical look at the North American Acknowledgements Quaternary Herpetofaunal Stability Hypothesis. Quaternary Thanks to AD Barnosky, AR Byrne, EB Davis, EA Hadly, DR International 271: 30–36. Lindberg, and JL Patton for helpful discussion and training. I Belnap J (2003) Biological soil crusts in deserts: A short review thank D Erwin, M Goodwin, P Holroyd, and C Marshall of the of their role in soil fertility, stabilization, and water relations. University of California Museum of Paleontology (UCMP) for Algological Studies 109: 113–126. curatorial and museum support; T Guilderson and P Zermeno for Belnap J, Reynolds RL, Reheis MC et al. (2009) Sediment losses radiocarbon dating training and assistance; undergraduate assis- and gains across a gradient of livestock grazing and plant tants M Hardesty-Moore, P Garcialuna, H Bhullar, and P Uppal invasion in a cool, semi-arid grassland, Colorado Plateau, for processing midden matrix; R Hunt-Foster for assistance with USA. Aeolian Research 1: 27–43. permitting; and the Canyonlands Research Center, Nature Con- Benson L and Berry MS (2009) Climate change and cultural servancy of Utah, and Redd Family of the Dugout Ranch for sup- response in the prehistoric American Southwest. KIVA: The port in the field. Journal of Southwestern Anthropology and History 75: 89–119.

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016 Stegner 13

Benson L, Peterson K and Stein J (2007) Anasazi (pre-Columbian Dietl GP, Kidwell SM, Brenner M et al. (2015) Conservation Native-American) migrations during the middle-12th and late- paleobiology: Leveraging knowledge of the past to inform 13th centuries – Were they drought induced? Climate Change conservation and restoration. Annual Review of Earth and 83: 187–213. Planetary Sciences 43: 79–103. Best TL (1996) Lepus californicus. Mammalian Species 530: Durrant SD (1952) Mammals of Utah: Taxonomy and Distribu- 1–10. tions, vol. 6. Lawrence, KS: University of Kansas Publica- Best TL, Burt SL and Bartig JL (1994) Tamias quadrivittatus. tions, Museum of Natural History. Mammalian Species 466: 1–7. Eldridge DJ and Whitford WG (2009) Soil disturbance by native Betancourt JL (1984) Late Quaternary plant zonation and climate animals along grazing gradients in an arid grassland. Journal in southeastern Utah. Great Basin Naturalist 44: 1–35. of Arid Environments 73: 1144–1148. Betancourt JL and Davis OK (1984) Packrat middens from Can- Ely L (1997) Response of extreme floods in the southwestern yon de Chelly, Northeastern Arizona: Paleoecological and United States to climate. Geomorphology 19: 175–201. archaeological implications. Quaternary Research 21: 56–64. Emslie SD (1986) Late Pleistocene vertebrates from Gunnison Blois JL, McGuire JL and Hadly EA (2010) Small mammal diver- County, Colorado. Journal of Vertebrate Paleontology 60: sity loss in response to Pleistocene climate change. Nature 170–176. 465: 773–775. Ernst AE and Prior-Magee JS (2008) Conservation status of the Bronk Ramsey C (2009) Bayesian analysis of radiocarbon dates. Colorado Plateau using Southwest Regional Gap Analysis Radiocarbon 51: 337–360. stewardship data. In: Van Riper C III and Sogge MK (eds) The Brown TA, Nelson DE, Vogel JS et al. (1988) Improved collagen Colorado Plateau III: Integrating Research and Resources extraction by modified Longin method. Radiocarbon 30: 171–177. Management for Effective Conservation. Tucson, AZ: Uni- Burt SL and Best TL (1994) Tamias rufus. Mammalian Species versity of Arizona Press, pp. 43–54. 460: 1–6. FAUNMAP Working Group (1994) FAUNMAP: A database doc- Calede JJM, Hopkins SSB and Davis EB (2011) Turnover in bur- umenting late Quaternary distributions of mammal species in rowing rodents: The roles of competition and habitat change. the United States. Illinois State Museum Scientific Papers 25: Palaeogeography, Palaeoclimatology, Palaeoecology 311: 1–690. 242–255. FAUNMAP Working Group (1996) Spatial responses of mam- Carleton MD and Eshelman RE (1979) A synopsis of fossil grass- mals to Late Quaternary environmental fluctuations. Science hopper mice, genus Onychomys, and their relationships to 272: 1601–1606. recent species. The University of Michigan Museum of Pale- Feranec RS, Hadly EA, Blois JL et al. (2007) Radiocarbon dates ontology Papers on Paleontology 21: 1–63. from the Pleistocene fossil deposits of Samwel Cave, Shasta Carrasco MA, Kraatz BP, Davis EB et al. (2005) Miocene Mam- County, California, USA. Radiocarbon 49: 117–121. mal Mapping Project (MIOMAP). University of California Fernández DP, Neff JC and Reynolds RL (2008) Biogeochemi- Museum of Paleontology. Available at: http://www.ucmp. cal and ecological impacts of livestock grazing in semi-arid berkeley.edu/miomap/. southeastern Utah, USA. Journal of Arid Environments 72: Carroll LE and Genoways HH (1980) Lagurus curtatus. Mam- 777–791. malian Species 124: 1–6. Foos A (1999) Geology of the Colorado Plateau. Report, National Chapman JA (1975) Sylvilagus nuttali. Mammalian Species 56: Parks Service. Available at: http://www.nature.nps.gov/geol- 1–3. ogy/education/foos/plateau.pdf. Chapman JA and Willner GR (1978) Sylvilagus audubonii. Mam- Fu B, Zhang L, Xu Z et al. (2015) Ecosystem services in changing malian Species 106: 1–4. land use. Journal of Soils and Sediments 15: 833–843. Chronic H (1990) Roadside Geology of Utah. Missoula, MT: Garfin G, Franco G, Blanco H et al. (2014) Southwest. In: Melillo Mountain Press Publishing Company. J, Richmond TC and Yohe GW (eds) Climate Change Impacts Coats LL, Cole KL and Mead JI (2008) 50,000 years of vegeta- in the United States: The Third National Climate Assessment. tion and climate history on the Colorado Plateau, Utah and Washington, DC: US Global Change Research Program, pp. Arizona, USA. Quaternary Research 70: 322–338. 462–486. Cole KL (1990) Reconstruction of past desert vegetation along Garrison TE and Best TL (1990) Dipodomys ordii. Mammalian the Colorado River using packrat middens. Palaeogeography, Species 353: 1–10. Palaeoclimatology, Palaeoecology 76: 349–366. Gillespie WB (1985) Holocene climate and environment of Chaco Conservation Paleobiology Workshop (2012) Conservation Canyon. In: Mathien FJ (ed.) Environment and Subsistence paleobiology: Opportunities for the earth sciences. Report, of Chaco Canyon, New Mexico (Publications in archaeology Division of Earth Sciences, National Science Foundation, 18E). Albuquerque, NM: United States Department of the Inte- Paleontological Research Institution, Ithaca, NY. rior, National Park Service, Chaco Canyon Studies, pp. 13–45. Cook ER and Krusic PJ (2004) The North American Drought Goodman LA (1965) On simultaneous confidence intervals for Atlas. Lamont-Doherty Earth Observatory and the National multinomial proportions. Technometrics 7: 247–254. Science Foundation. Available at: http://iridl.ldeo.columbia. Gotelli NJ and Ellison AM (2013) A Primer of Ecological Statis- edu/SOURCES/.LDEO/.TRL/.NADA2004/.pdsi-atlas.html. tics. 2nd Edition. Sunderland: Sinauer Associates. Davis EB (2005) Mammalian beta diversity in the Great Basin, Graham RW and Grimm EC (1990) Effects of global climate western USA: Palaeontological data suggest deep origin of change on the patterns of terrestrial biological communities. modern macroecological structure. Global Ecology and Bio- Trends in Ecology & Evolution 5: 289–292. geography 14: 479–490. Grayson DK (1978) Minimum numbers and sample size in verte- Dickman CR, Greenville AC, Beh C-L et al. (2010) Social orga- brate faunal analysis. American Antiquity 43: 53–65. nization and movements of desert rodents during population Grayson DK (1988) Danger Cave, Last Supper Cave, and Hang- ‘booms’ and ‘busts’ in central Australia. Journal of Mammal- ing Rock Shelter: The faunas. Anthropological Papers of the ogy 91: 798–810. American Museum of Natural History 66: 1–130. Dietl GP and Flessa KW (2011) Conservation paleobiology: Put- Grayson DK (2000) Mammalian response to the Middle Holo- ting the dead to work. Trends in Ecology & Evolution 26: cene climatic change in the Great Basin of the western United 30–37. States. Journal of Biogeography 27: 181–192.

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016  The Holocene 14

Grayson DK (2011) The Great Basin: A Natural Prehistory. Schubert BW, Mead JI and Graham RW (eds) Ice Age Cave Berkeley, CA: University of California Press. Faunas of North America. Bloomington, IN: Indiana Univer- Hadly EA (1996) Influence of late-Holocene climate on north- sity Press, pp. 64–86. ern Rocky Mountain mammals. Quaternary Research 46: Miller ME, Belote RT, Bowker MA et al. (2011) Alternative states 298–310. of a semiarid grassland ecosystem: Implications for ecosys- Hadly EA (1999) Fidelity of terrestrial vertebrate fossils to a tem services. Ecosphere 2: 1–18. modern ecosystem. Palaeogeography, Palaeoclimatology, Murray LK, Bell CJ, Dolan MT et al. (2005) Late Pleistocene Palaeoecology 149: 389–409. fauna from the southern Colorado Plateau, Navajo County, Hadly EA and Barnosky AD (2009) Vertebrate fossils and the Arizona. Southwestern Naturalist 50: 363–374. future of conservation biology. In: Dietl GP and Flessa KW Nash DJ and Seaman RN (1977) Sciurus aberti. Mammalian Spe- (eds) Conservation Paleobiology: Using the Past to Manage cies 80: 1–5. for the Future. Boulder, CO: The Paleontological Society, pp. Nash RF (1967) Wilderness and the American Mind. 5th Edition. 39–59. New Haven, CT: Yale University Press. Harris AH and Findley AH (1964) Pleistocene-recent fauna of Neotoma Paleoecological Database (2013) Available at: http:// the Isleta Caves, Bernalillo County, New Mexico. American www.neotomadb.org (accessed 9 June 2015). Journal of Science 262: 114–120. Oaks EC, Young PJ, Kirkland GL, Jr et al. (1987) Spermophilus Hernández L, Laundré JW, Grajales KM et al. (2011) Plant pro- variegatus. Mammalian Species 272: 1–8. ductivity, predation, and the abundance of black-tailed jack- Oksanen J, Blanchet FG, Kindt R et al. (2015) vegan: Community rabbits in the Chihuahuan Desert of Mexico. Journal of Arid ecology package, version 2.3-0. Available at: http://CRAN.R- Environments 75: 1043–1049. project.org/package=vegan. Hoffmeister DF (1981) Peromyscus truei. Mammalian Species Pederson GT, Gray ST, Woodhouse CA et al. (2011) The unusual 161: 1–5. nature of recent snowpack declines in the North American Hooper ET (1957) Dental patterns in mice of the genus Peromys- Cordillera. Science 333(6040): 332–335. cus. Miscellaneous Publications Museum of Zoology, Univer- Porder S, Paytan A and Hadly EA (2003) Mapping the origin of sity of Michigan 99: 1–59. faunal assemblages using strontium isotopes. Paleobiology Hurlbert SH (1971) The nonconcept of species diversity: A cri- 29: 197–204. tique and alternative parameters. Ecology 52: 577–586. PRISM Climate Group (2015) PRISM climate data, updated Jones A (2000) Effects of cattle grazing on North American arid 2015. Oregon State University. Available at: http://prism. ecosystems: A quantitative review. Western North American oregonstate.edu/normals/ (accessed 25 June 2015). Naturalist 60: 155–164. R Core Team (2015) R: A language and environment for statis- Jones CA and Baxter CN (2004) Thomomys bottae. Mammalian tical computing, version 3.2.0. Available at: http://www.R- Species 742: 1–14. project.org. Kays RW and Wilson DE (2009) Mammals of North America. 2nd Reheis MC, Reynolds RL, Goldstein H et al. (2005) Late Qua- Edition. Princeton, NJ: Princeton University Press. ternary eolian and alluvial response to paleoclimate, Can- Kidwell SM (2015) Biology in the Anthropocene: Challenges yonlands, southeastern Utah. Geological Society of America and insights from young fossil records. Proceedings of the Bulletin 117: 1051–1069. National Academy of Sciences of the United States of America Reimer PJ, Bard E, Bayliss A et al. (2013) IntCal13 and Marine13 112: 4922–4929. radiocarbon age calibration curves 0–50,000 years cal BP. Lavender D (1964) One Man’s West. 3rd Edition. New York: Dou- Radiocarbon 55: 1869–1887. bleday and Company, Inc. Repenning CA (2004) Fossil wood rats of Porcupine Cave: Tec- Lawler JJ, Lewis DJ, Nelson E et al. (2014) Projected land-use tonic or climatic controls? In: Barnosky AD (ed.) Biodiver- change impacts on ecosystem services in the United States. sity Response to Climate Change in the Middle Pleistocene. Proceedings of the National Academy of Sciences of the Berkeley, CA: University of California Press, pp. 193–206. United States of America 111: 7492–7497. Rick TC and Lockwood R (2013) Integrating paleobiology, Lightfoot DC, Davidson AD, McGlone CM et al. (2010) Rabbit archeology, and history to inform biological conservation. abundance relative to rainfall and plant production in northern Conservation Biology 27: 45–54. Chihuahuan Desert grassland and shrubland habitats. Western Robinson TJ and Matthee CA (2005) Phylogeny and evolutionary North American Naturalist 70: 490–499. origins of the Leporidae: a review of cytogenetics, molecu- McCarty R (1975) Onychomys torridus. Mammalian Species 59: lar analyses and a supermatrix analysis. Mammal Review 35: 1–5. 231–247. McCarty R (1978) Onychomys leucogaster. Mammalian Species Rowe RJ (2007) Legacies of land use and recent climate change: 87: 1–6. The small mammal fauna in the Mountains of Utah. The McHorse BK, Orcutt JD and Davis EB (2012) The carnivoran American Naturalist 170: 242–257. fauna of Rancho La Brea: Average or aberrant? Palaeo- Rowe RJ and Terry RC (2014) Small mammal response to envi- geography, Palaeoclimatology, Palaeoecology 329–330: ronmental change: Integrating past and present dynamics. 118–123. Journal of Mammalogy 95: 1157–1174. Marcy AE, Fendorf S and Hadly EA (2013) Morphological Schaefer TS (1991) Mammals of the Abajo Mountains, an iso- adaptations for digging and climate-impacted soil properties lated mountain range in San Juan County, southeastern Utah. define pocket gopher (Thomomys spp.) distributions. PLoS Occasional Papers: Museum, Texas Tech University 137: ONE 8: 1–14. 1–15. Mead JI (1981) The last 30,000 years of faunal history within the Seager R, Ting M, Held I et al. (2007) Model projections of an Grand Canyon, Arizona. Quaternary Research 15: 311–326. imminent transition to a more arid climate in southwestern Mead JI and Phillips AM, III (1981) The late Pleistocene and North America. Science 316: 1181–1184. Holocene fauna and flora of Vulture Cave, Grand Canyon, Signorell A, Aho K, Anderegg N et al. (2015) DescTools: Tools Arizona. Southwestern Naturalist 26: 257–288. for descriptive statistics package, version 0.99.13. Available Mead JI, Coats LL and Schubert BW (2003) Late Pleistocene at: https://cran.r-project.org/web/packages/DescTools/Desc- faunas from caves in the eastern Grand Canyon, Arizona. In: Tools.pdf.

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016 Stegner 15

Smith FA and Betancourt JL (1998) Response of bushy-tailed anthropogenic environmental change. PNAS 112: 9656– woodrats (Neotoma cinerea) to Late Quaternary climatic 9661. change in the Colorado Plateau. Quaternary Research 50: Terry RC (2010b) The dead do not lie: Using skeletal remains 1–50. for rapid assessment of historical small-mammal community Sorensen JS, Turnbull CA and Dearing MD (2004) A specialist baselines. Proceedings of the Royal Society B: Biological Sci- herbivore (Neotoma stephensi) absorbs fewer plant toxins ences 277: 1193–1201. than does a generalist (Neotoma albigula). Physiological and Terry RC, Li CL and Hadly EA (2011) Predicting small-mammal Biochemical Zoology 77: 139–148. responses to climatic warming: Autoecology, geographic Stahle DW and Dean JS (2010) North American tree rings, cli- range, and the Holocene fossil record. Global Change Biol- matic extremes, and social disasters. In: Hughes MK, Swet- ogy 17: 3019–3303. nam TW and Diaz HF (eds) Dendroclimatology. Dordrecht: Tweet JS, Santucci VL and Hunt AP (2012) An inventory of pack- Springer Science, pp. 297–327. rat (Neotoma spp.) middens in National Park Service areas. Stapp P (2010) Long-term studies of small mammal communities In: Hunt AP, Milàn J, Lucas SG et al. (eds) Vertebrate Copro- in arid and semiarid environments. Journal of Mammalogy lites (Bulletin 57). Albuquerque, NM: New Mexico Museum 91: 773–775. of Natural History and Science, pp. 355–368. Stegner MA (2015) The Mescal Cave fauna (San Bernardino Verts BJ and Carraway LN (1999) Thomomys talpoides. Mam- County, California) and testing assumptions of habitat fidel- malian Species 618: 1–11. ity in the Quaternary fossil record. Quaternary Research 83: Verts BJ and Carraway LN (2001) Tamias minimus. Mammalian 582–587. Species 653: 1–10. Stegner MA and Holmes M (2013) Using palaeontological data Viau AE, Gajewski K, Sawafa MC et al. (2006) Millennial-scale to assess mammalian community structure: Potential aid in temperature variations in North America during the Holo- conservation planning. Palaeogeography, Palaeoclimatol- cene. Journal of Geophysical Research 3: 1–12. ogy, Palaeoecology 372: 138–146. Vimtrup BJ and Schmidt-Nielsen B (2005) The histology of the Streubel DP and Fitzgerald JP (1978) Spermophilus spilosoma. kidney of kangaroo rats. The Anatomical Record 114: 515–528. Mammalian Species 101: 1–4. Vogel JS, Southon JR and Nelson DE (1987) Catalyst and binder Stuiver M and Polach HA (1977) Reporting of 14C data. Radio- effects in the use of filamentous graphite for AMS. Nuclear carbon 19: 355–363. Instruments & Methods in Physics Research Section B: Beam Swetnam TW, Allen CD and Betancourt JL (1999) Applied his- Interactions with Materials and Atoms 29: 50–56. torical ecology: Using the past to manage for the future. Eco- Weng C and Jackson ST (1999) Late Glacial and Holocene veg- logical Applications 9: 1189–1206. etation history and paleoclimate of the Kaibab Plateau, Ari- Terry RC (2008) Modeling the effects of predation, prey cycling, zona. Palaeogeography, Palaeoclimatology, Palaeoecology and time averaging on relative abundance in raptor-generated 153: 179–201. small mammal death assemblages. Palaios 23: 402–410. Whitford WG (1976) Temporal fluctuations in density and diver- Terry RC (2010a) On raptors and rodents: Testing the ecological sity of desert rodent populations. Journal of Mammalogy 57: fidelity and spatiotemporal resolution of cave death assem- 351–369. blages. Paleobiology 36: 137–160. Willis KJ and Birks HJB (2006) What is natural? The need for Terry RC and Rowe RJ (2015) Energy flow and functional com- a long-term perspective in biodiversity conservation. Science pensation in Great Basin small mammals under natural and 314: 1261–1265.

Downloaded from hol.sagepub.com at Stanford University Libraries on March 12, 2016