SUPPLEMENTARY INFORMATION

Transcription

1 SUPPLEMENTARY TABLES Supplementary Table 1. AMS radiocarbon dates on bone collagen from specimens distributed throughout the Samwell Cave Popcorn Dome deposit. CAMS UCMP Species fraction 14 C 2-σ age range Number ID ID Element Level δ 13 C modern D14C age (cal yr BP) Mammalia ulna e3a ,815 2,790-3,070 ± ± 3.27 ± Mammalia unknown e4a ,900 7,620-7,850 ± ± 2.32 ± Mammalia humerus e5a ,765 9,550-10,120 ± ± 1.92 ± Mammalia R humerus e6a ,940 5,330-5,930 ± ± 7.18 ± Mammalia R humerus e6b ,320 12,970-13,390 ± ± 3.31 ± Mammalia L humerus e7b ,440 17,680-19,500 ± ± 7.08 ±

2 Supplementary Table 2. Non-standardized abundance data for the e-levels of Samwell Cave Popcorn Dome deposit. Artiodactyla Carnivora 1 (1) Lagomorpha e1 e2 e3a e3b e4a NISP (MNI) NISP (MNI) NISP (MNI) NISP (MNI) NISP (MNI) Leporidae 3 (1) 2 (1) 1 (1) Rodentia Aplodontidae Aplodontia rufa Geomyidae Thomomys spp Thomomys bottae 5 (3) 9 (5) 2 (1) 6 (3) 6 (3) Thomomys cf. mazama 2 (2) Cricetidae Neotominae Neotoma spp. 32 (5) 44 (9) 28 (4) 35 (6) 28 (4) Peromyscus spp. 151 (24) 171 (33) 136 (24) 220 (38) 203 (34) Reithrodontomys cf. megalotis 1 (1) 1 (1) 1 (1) Arvicolinae Microtus spp. 38 (7) 30 (4) 15 (3) 51 (6) 40 (6) Myodes cf. californicus cf. Arborimus albipes 1 (1) Sciuridae Glaucomys cf. sabrinus 2 (1) 1 (1) Sciurus griseus 1 (1) 1 (1) 3 (2) 1 (1) Spermophilus beecheyi 1 (1) 1 (1) 2 (1) 4 (2) 4 (1) Spermophilus lateralis 2 (1) 1 (1) Tamias spp. 1 (1) 1 (1) 1 (1) Tamiasciurus douglasii 1 (1) 1 (1) 1 (1) 2 (2) 1 (1) Soricomorpha Soricidae Sorex spp Talpidae cf. Scapanus latimanus Chiroptera Total NISP NISP: number of identified specimens; MNI: minimum number of individuals. Total NISP for each level is noted at the bottom of the column. 2

3 Supplementary Table 2 continued. Artiodactyla 2 (1) Carnivora Lagomorpha e4b e5a e5b e6a e6b NISP (MNI) NISP (MNI) NISP (MNI) NISP (MNI) NISP (MNI) Leporidae 2 (1) 2 (1) 1 (1) 0 (1) Rodentia Aplodontidae Aplodontia rufa Geomyidae Thomomys spp Thomomys bottae 1 (1) 1 (1) 3 (2) 1 (1) 1 (1) Thomomys cf. mazama 3 (2) 1 (1) 5 (3) Cricetidae Neotominae Neotoma spp. 20 (4) 39 (7) 14 (4) 28 (5) 24 (5) Peromyscus spp. 112 (20) 184 (32) 77 (14) 132 (22) 114 (25) Reithrodontomys cf. megalotis 1 (1) Arvicolinae Microtus spp. 31 (5) 32 (5) 13 (2) 15 (2) 19 (3) Myodes cf. californicus 2 (1) cf. Arborimus albipes 1 (1) 1 (1) 1 (1) 1 (1) Sciuridae Glaucomys cf. sabrinus 1 (1) Sciurus griseus 1 (1) 2 (1) 3 (1) Spermophilus beecheyi 1 (1) 5 (2) 3 (1) 4 (1) 2 (2) Spermophilus lateralis 1 (1) 1 (1) 2 (1) 1 (1) Tamias spp. 1 (1) 1 (1) 3 (1) Tamiasciurus douglasii 2 (1) 1 (1) 3 (1) 2 (2) Soricomorpha Soricidae Sorex spp Talpidae cf. Scapanus latimanus Chiroptera Total NISP NISP: number of identified specimens; MNI: minimum number of individuals. Total NISP for each level is noted at the bottom of the column. 3

4 Supplementary Table 2 continued. e7a e7b e8a e8b NISP (MNI) NISP (MNI) NISP (MNI) NISP (MNI) Artiodactyla 1 (1) 2 (1) Carnivora 1 (1) Lagomorpha Leporidae 1 (1) 3 (1) 2 (1) 1 (1) Rodentia Aplodontidae Aplodontia rufa 1 (1) 1 (1) 3 (1) Geomyidae Thomomys spp Thomomys bottae 1 (1) 1 (1) Thomomys cf. mazama 4 (2) 2 (1) 8 (5) 5 (4) Cricetidae Neotominae Neotoma spp. 37 (5) 35 (5) 73 (8) 54 (6) Peromyscus spp. 119 (21) 58 (16) 87 (19) 100 (18) Reithrodontomys cf. megalotis 1 (1) 1 (1) Arvicolinae Microtus spp. 24 (3) 19 (4) 35 (7) 30 (4) Myodes cf. californicus cf. Arborimus albipes 3 (2) 2 (1) 4 (2) 4 (1) Sciuridae Glaucomys cf. sabrinus 2 (1) 1 (1) Sciurus griseus 2 (1) 2 (1) Spermophilus beecheyi 5 (1) 5 (2) 6 (2) 7 (2) Spermophilus lateralis 4 (2) 2 (1) 12 (3) 2 (1) Tamias spp. 7 (2) 4 (1) 6 (1) 6 (1) Tamiasciurus douglasii 4 (1) 3 (1) 5 (2) 11 (2) Soricomorpha Soricidae Sorex spp Talpidae cf. Scapanus latimanus Chiroptera Total NISP NISP: number of identified specimens; MNI: minimum number of individuals. Total NISP for each level is noted at the bottom of the column. 4

5 Supplementary Table 3. Small-mammal (Lagomorpha, Rodentia, and Soricomorpha) standardized abundance values (NISP s ) used for diversity analysis. e1 e2 e3a e3b e4a e4b e5a e5b e6a e6b e7a e7b e8a e8b Leporidae Aplodontia rufa Thomomys bottae Thomomys cf. mazama Neotoma spp Peromyscus spp Reithrodontomys cf. megalotis Microtus spp Myodes cf. californicus cf. Arborimus albipes Glaucomys cf. sabrinus Sciurus griseus Spermophilus beecheyi Spermophilus lateralis Tamias spp Tamiasciurus douglasii Sorex spp cf. Scapanus latimanus Column headings refer to levels within the fossil deposit, from e1 (top) to e8b (bottom). Since we obtained ancient DNA verification of the identity of some Thomomys spp., we estimated the proportion of Thomomys spp. specimens likely belonging to each species (based on the proportion of T. bottae or T. cf. mazama in that level) and added that estimate to the NISP of each species. Thomomys NISP s values were based on these values rather than NISP given in Supplementary Table

6 Supplementary Table 4. Taxonomic richness (S) and evenness (PIE) for a, levels and b, periods. a b Level S (se) PIE (se) Period S (se) PIE (se) e (0.03) 0.59 (0.001) e (0.05) 0.59 (0.001) e3a 7.53 (0.04) 0.56 (0.001) e3b 8.52 (0.04) 0.61 (0.001) Holocene 8.01 (0.04) 0.59 (0.001) e4a 9 (0.04) 0.57 (0.002) e4b 6.74 (0.03) 0.6 (0.001) e5a 9.92 (0.05) 0.59 (0.001) e5b 7.07 (0.03) 0.61 (0.001) e6a (0.04) 0.64 (0.001) e6b 9 (0.03) 0.64 (0.001) Transition (0.04) 0.67 (0.001) e7a (0.04) 0.73 (0.001) e7b (0.03) 0.81 (0.001) e8a (0.04) 0.8 (0.001) Pleistocene (0.04) 0.80 (0.001) e8b (0.04) 0.79 (0.001) Values for each level are the mean statistic (± standard error [se]) of 1000 bootstrap samples at the standardized sample size of n=132. The period statistics were calculated based on the mean of the level means and standard errors for each period. 6

7 Supplementary Table 5. Specimens captured during live-trapping within various habitats in the region surrounding Samwell Cave. Locality Name Sex Collector (Trapping Dates) Zone Easting Northing Species M F unk Number(s) Samwell Cave Peromyscus sp Lower Transect Peromyscus maniculatus JLB 059, 063 (06/27-29/2008) Peromyscus boylii JLB 068 Sorex trowbridgii 2 JLB 061, 067 Microtus californicus 1 JLB 069 Neotoma fuscipes 1 JLB 070 Mephitis mephitis Fender's Ferry Spermophilus beecheyi 2 1 JLB 082, 083 (7/3/08) Samwell Cave Peromyscus boylii JLB 071, 072 Upper Transect Peromyscus maniculatus 1 - (7/1-4/2008) Microtus californicus 1 2 JLB 073, 074 Ellery Creek Peromyscus maniculatus JLB 086, 087 (7/7-9/2008) Peromyscus boylii Spermophilus beecheyi JLB 088, 093, 098 Sorex trowbridgii 1 2 JLB 090, 091 Hirz Mountain Peromyscus maniculatus 1 3 JLB 100 (7/11-14/2008) Peromyscus boylii JLB 099, 101 Samwell Cave Microtus californicus 1 JLB 102 (7/18, 7/23/2008) Peromyscus boylii Bassariscus astutus 1 - Girard # Peromyscus maniculatus 9 4 JLB 104 (7/24-25/2008) Peromyscus boylii 1 - Girard # Peromyscus maniculatus (7/24-25/2008) Peromyscus boylii 3 2 JLB 108 Tamias sp. 4 9 JLB 109, 113, 124, 130 Glaucomys sabrinus 1 JLB 114 Sorex trowbridgii 1 JLB 127 Sylvilagus bachmani 1 - M: male; F: female; unk: sex unknown 7

8 SUPPLEMENTARY FIGURES Depth below surface (cm) Level e1 e2 e3a e3b e4a e4b e5a e5b e6a e6b e7a e7b e8a e8b Level e1 e2 e3a e3b e4a e4b e5a e5b e6a e6b e7a e7b e8a e8b Calendar years before present (cal yr BP) Supplementary Figure 1. Calibrated AMS radiocarbon dates on bone collagen from samples collected in the e-levels. Age is given in calendar years before present (cal yr BP) 35. The probability distribution of the age of each sample is shown, with the overall 95% confidence interval shown at the bottom. The points indicate the midpoint of the 95% confidence interval and the line indicates the best fit through the modal calibrated age based on a linear model (Age= 251*Depth, Residual standard error (RSE) df=6 =2901, Adj. R 2 =0.92, P= , AIC= ). The horizontal dashed lines indicate the boundaries between excavated levels. The projected ages of each level are indicated along the x-axis in gray. 8

9 a Peromyscus spp. Neotoma spp. Reithrodontomys cf. megalotis b Thomomys cf. mazama Thomomys bottae Standardized abundance (NISPs) c e Microtus spp. cf. Arborimus albipes Myodes cf. californicus Tamiasciurus douglasii Glaucomys cf. sabrinus d f Spermophilus lateralis Spermophilus beecheyi Tamias spp. Sciurus griseus g cf. Scapanus latimanus Sorex spp h Leporidae Aplodontia rufa Calendar years before present (cal yr BP) Supplementary Figure 2. Standardized number of identified specimens (NISP s ) of all smallmammal taxa represented in SCPD through time. Taxa are plotted on the same graph for space purposes only, and are usually arranged with taxa in a similar taxonomic category. 9

10 a Evenness (PIE) PIE: All taxa PIE: Peromyscus removed PIE: individual nonperomyscus taxa removed S: all taxa Richness (S) b BrayCurtis Index BrayCurtis: all taxa BrayCurtis: Peromyscus removed BrayCurtis: individual nonperomyscus taxa removed Jaccard: all taxa Jaccard index Calendar years before present (cal yr BP) Supplementary Figure 3. Sensitivity of abundance-based diversity measures to each species. a, Evenness (solid and dashed black lines, left axis) and richness (gray line and dot, right axis); b, Bray-Curtis (solid and dashed black lines, left axis) and Jaccard (gray line and dot, right axis) Indices. On both graphs, thick black and gray lines are the original diversity metric, whereas the dashed lines are the diversity value with one species removed. Values are plotted as lines for purposes of illustration only. The vertical dotted lines indicate the boundaries between levels within SCPD. 10

11 a 18 O ( ) b O ( ) Calendar years before present (cal yr BP) Supplementary Figure 4. Comparison of the original (solid gray line) and downscaled (black dashed lines) climate data based on the δ 18 O ice core record from Greenland 20. a, sampled data; b, averaged data. The vertical dotted lines indicate the boundaries between levels within SCPD. 11

12 SUPPLEMENTARY DISCUSSION Excavation and Radiocarbon Chronology The SCPD deposit was formed as a woodrat (Neotoma) midden, evidenced by the accumulation of Neotoma fecal pellets, sticks and other nest material on the deposit surface as well as the large proportion of Neotoma specimens in the deposit. Woodrat middens accumulate when terrestrial and aerial predators eat small animals and excrete the undigested bones and fur as a scat or pellet. These scats are collected by woodrats and brought back to their middens, where they break down and are incorporated into the sediment 44,45. Woodrat middens that accumulate in caves have high fossilization potential due to the generally excellent preservational environment within caves 46. The small-mammal community accumulated in middens is sampled from a fairly constrained region, generally within 5-10 km of the cave 47, and has high fidelity to the surrounding community 39,48 and its dynamics through time. The deposit was split roughly down the middle, with the western side of the deposit showing a large discontinuity between late-pleistocene and late-holocene deposits consistent with an unconformity 31, and the eastern side of the deposit showing a more continuous depositional environment ( e-level samples; Supplementary Fig. 1, Supplementary Table 1). All radiocarbon dates were of high quality and none were rejected. All but one radiocarbon date show older age with depth (Supplementary Table 1: UCMP # ). However, dates from the two sections of the excavated pit show different depositional modes and indicate greater stratigraphic layering closest to the rock wall to the east and under the lowest ceiling. Here, we 12

13 present data and results from the eastern portion of the deposit only (Supplementary Table 1, Supplementary Fig. 1). Individual taxa The abundance of individual taxa changed substantially and independently of other taxa through time (Supplementary Fig. 2), but overall the four most dominant taxa (Peromyscus spp., Microtus spp., Neotoma spp., and Thomomys spp.) were present and abundant in every level of the deposit. Community structure The deposit can be roughly divided into three distinct communities: Late Pleistocene (Levels e7b through e8b; to cal yr BP); a transition period (Levels e6a through e7a; to 14,500 cal yr BP); and Holocene (Levels e1 through e5b, 0 to cal yr BP). There is no overlap in the 95% confidence intervals between late-pleistocene and Holocene time periods. The small-mammal communities of the Pleistocene and Holocene are significantly different in terms of all diversity metrics (Supplementary Table 4). One interesting implication of the changes we see (few small-mammal extirpations, yet local decreases in richness) is that spatial heterogeneity in communities should be increasing through time. This is consistent with the hypothesis that vegetation change at the end of the Pleistocene created mosaics of novel vegetation communities, which led to less overlap in mammal geographic ranges and consequently lower diversity 49. Faunal turnover The Bray-Curtis and Jaccard turnover metrics provided complementary views of faunal turnover. Bray-Curtis incorporates the relative abundance of each taxon into the distance estimator, whereas Jaccard only considers the presence or absence of each taxon in the 13

14 community and not its abundance. These metrics thus detect different types of community change (Fig. 2c, Supplementary Fig. 3). The Bray-Curtis Index peaked between 15,000 to 14,000 cal yr BP (Fig. 2c). Our taxonomic sensitivity analysis demonstrated that this peak was recovered regardless of which species was dropped from the analysis (gray and thick dashed lines in Supplementary Fig. 3), indicating that changing abundances of all species are contributing to the turnover peak at this time. However, when Peromyscus is removed from the analysis (thick dashed line, Supplementary Fig. 3), the turnover pattern demonstrated by the Bray-Curtis Index is similar to that of the Jaccard Index (black dots, Supplementary Fig. 3). This indicates that the period of high turnover from 11,000 to 7,500 cal yr BP detected by the Jaccard Index is due to changes in the presence and absence of all taxa except for Peromyscus. In other words, in a community without Peromyscus, diversity declines are driven more by, or at least equal to, the presence/absence of taxa and not simply their abundances. Peromyscus Deer mice, Peromyscus spp., contribute most to the decline in evenness. Smallmammal communities without Peromyscus still show significant declines in evenness, but the magnitude of the decline is not as large as when Peromyscus is included in the community (Supplementary Fig. 3). For example, evenness declines from 0.80 to 0.75, versus 0.81 (LGM) to 0.60 (Holocene) for the entire small-mammal community. The removal of other taxa from the community act in the opposite manner as Peromyscus by enhancing the Holocene decline of evenness. Additionally, without Peromyscus in the community, the pattern of evenness change becomes more similar to the pattern of richness change (Supplementary Fig. 3). 14

15 We were unable to identify the particular species of Peromyscus represented by the fossil data, but modern trapping studies conducted in summer 2008 indicate that at least two species inhabit the region today: Peromyscus boylii and P. maniculatus. We trapped 11 different small-mammal species (Supplementary Table 5), and overall, the most abundant taxon was Peromyscus, accounting for 82% of the total captures. Both P. boylii and P. maniculatus are abundant in the region and can attain high rates of reproduction 50. Peromyscus maniculatus is the most widely distributed rodent in North America and is found in a wide variety of habitats, whereas P. boylii occupies arid and semi-arid habitats throughout the southern US and Mexico 51. Potential biases Differential taxonomic refinement Uneven levels of taxonomic identification (i.e., some taxa identified to species, some to genus) could potentially alter our conclusions. To address this, we recalculated diversity at the genus and subfamily or family levels. In all cases, richness and evenness declined similarly to the original dataset, though the decline in richness was progressively less at higher taxonomic levels. The decline in evenness was similar in magnitude and timing at each taxonomic level. Time averaging Larger samples from communities approximate true community diversity following standard sampling curves 52. However, greater spatial- or temporal-averaging can result in an increase of apparent diversity because averaged samples may pool multiple communities as one 53. For example, if stratigraphic deposits from the Pleistocene encompassed more time than levels from the Holocene, the increased faunal diversity in the Pleistocene might be due to inclusion of several community transitions. The SCPD deposits followed a linear 15

16 accumulation with time, thus time bins were of roughly equal duration and time-averaging did not account for changes in community structure. However, because the exponential depositional model was only slightly less well supported statistically, but would indicate greater amounts of time averaging during the Pleistocene, we also assessed the possible influence of time averaging on our patterns of community change using the exponential model. First, to investigate how robust our results were to increasing the amount of time represented, we combined adjacent levels into longer time bins so that the average duration of Holocene time periods was longer than Pleistocene time bins (3426 versus 2984 years, respectively). Second, we used a sliding window approach where either two or three adjacent levels were combined to create levels of approximately equal duration. In all cases, time averaging did not affect the patterns of community structure. Results from using the total sample size from each level and the standardized sample (n=132 per level) were also similar for all statistics. The main statistic affected by our sampling routine was richness. Richness was positively correlated with sample size as expected (r=0.5672, P= 0.03), but because sample size was not correlated with time (range of NISP = 132 to 374; r= , P=0.64), this did not influence our conclusions regarding diversity trends through time. Higher community abundance correlates with higher likelihood of occurring in more than one level and also higher relative abundances within levels 54. Overall, the Holocene decline in richness and evenness was not explained by choice of depositional model, time averaging, or sample size. Taphonomy Taphonomic biases should be similar within a continuous fossil deposit and we saw no indication of taphonomic changes through time within SCPD. For example, there were no abrupt changes in community structure of small mammals that would indicate changing aerial 16

17 or terrestrial predators. Further, both the SCPD and previously excavated Samwell Cave localities demonstrate similar diversity metrics 31. Within the SCPD, a large part of the evenness signal in particular is due the increasing abundance of Peromyscus. This increase is unlikely to be due to taphonomic or sampling effects because Peromyscus NISP did not vary with time during the Holocene, as might be expected if post-depositional attrition occurred. Additionally, other taxa with similarly-sized small teeth (e.g., Tamias spp., Thomomys cf. mazama) showed larger NISP during the Pleistocene than during the Holocene, which is counter to preservational bias if small teeth were preferentially lost with increasing deposition time. The NISP s of Peromyscus was significantly negatively correlated with NISP s of Neotoma (Supplementary Fig. 2a; r = , P <0.001), suggesting the possibility that the increase in Peromyscus is due to the local decline of Neotoma. However, this correlation is likely to be spurious rather than causative because Neotoma generally exhibits a positive effect on the presence of Peromyscus by increasing the availability of suitable nest sites for Peromyscus 55,56. For example, there was either no association (P. truei), or a positive association (P. californicus), between Peromyscus and Neotoma fuscipes in central California

18 SUPPLEMENTARY NOTES 44 Hadly, E. A. Influence of late-holocene climate on northern Rocky Mountain mammals. Quaternary Res. 46, (1996). 45 Terry, R. C. Owl pellet taphonomy: a preliminary study of the post-regurgitation taphonomic history of pellets in temperate forest. Palaios 19, (2004). 46 Andrews, P. Owls, caves, and fossils: predation, preservation, and accumulation of small mammal bones in caves, with an analysis of the Pleistocene cave faunas from Westburysub-Mendip, Somerset, UK (University of Chicago Press, 1990). 47 Porder, S., Paytan, A. & Hadly, E. A. Mapping the origin of faunal assemblages using strontium isotopes. Paleobiology 29, (2003). 48 Reed, D. in Hominin environments in the East African Pliocene: an assessment of the faunal evidence (eds R. Bobe, A Zeresenay, & A. K. Behrensmeyer) (Springer, 2007). 49 Guthrie, R. in Quaternary Extinctions: A Prehistoric Revolution (eds P. S. Martin & R.G. Klein) (University of Arizona Press 1984). 50 Jameson Jr., E. W. Reproduction of deer mice (Peromyscus maniculatus and P. boylei) in the Sierra Nevada, California. J. Mamm. 34, (1953). 51 Hall, E. R. & Kelson, K. R. The mammals of North America Vol. 1 (The Ronald Press Company, 1959). 52 Rosenzweig, M. L. Species diversity in space and time (Cambridge University Press, 1995). 53 Behrensmeyer, A. K., Kidwell, S. M. & Gastaldo, R. A. Taphonomy and paleobiology. Paleobiology 26, (2000). 18

19 54 Hadly, E. A. & Maurer, B. A. Spatial and temporal patterns of species diversity in montane mammal communities of western North America. Evol. Ecol. Res. 3, (2001). 55 Cranford, J. A. The effect of woodrat houses on population density of Peromyscus. J. Mamm. 63, (1982). 56 Merritt, J. F. Factors influencing the local distribution of Peromyscus californicus in northern California. J. Mamm. 55, (1974). 19