Temperature and fossil mammal dental trait shifts during the first 10 Myr of the Cenozoic.

(A) Dental topographic trait values (boxplots) and mean variance (red curves) during the first 10 Myr of the Cenozoic, signifying the time after the K-Pg mass extinctions and before the Paleocene-Eocene hyperthermal event. Global temperature curve based on Zachos et al. 61. Dental traits measured include crown complexity (OPCR, orientation patch count rotated), curvature (DNE, Dirichlet normal energy), height (RFI, relief index), and slope. (B) Mammal tooth size distributions represented by log 10 square root tooth area, in units of log10 millimeters. (C) Variance of compressive bite performance based on tooth crown finite element simulations, in units of squared Joules. (D) Variance of shear bite performance based on tooth crown finite element simulations, in units of squared Joules. Examples of endemic Asian fossil specimens analyzed: (E) Lateral and ventral views of early Paleocene Chinese endemic pantodont (CEP) Bemalambda nanhsiungensis IVPP (Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences) V4116. (F) Lateral and ventral views of middle Paleocene CEP Harpyodus decorus IVPP 5035.1. (G) Lateral and occlusal views of late Paleocene CEP Guichilambda zhaii IVPP V12037.2 (dentary) and V12037.3 (maxillary fragment). Firey asteroid symbols indicate the end-Cretaceous asteroid impact in the Yucatán Peninsula; thermometer symbols indicate the Paleocene-Eocene hyperthermal event. Subdivisions of the Paleocene approximately correspond to the Shanghuan, Nongshanian, and Gashatan Asian Land Mammal Ages, respectively (see supplemental text for competing age boundary scenarios).

Association of paleopalynological data from the Nanxiong Basin, south China, and late Paleocene niche expansion in endemic Asian fossil mammals.

(A) Proportion of environmental humidity indicator taxa from early versus late Paleocene paleobotanical localities, respectively, in the Nanxiong Basin; data based on 35,36(Data S12). (B) Boxplots of dental complexity (OPCR, orientation patch count rotated) in the Chinese endemic pantodont (CEP) data partition across the three Paleocene time intervals examined. Note the concomitant increase in CEP tooth complexity (OPCR) and increased proportion of drought-tolerant plant species in the Nanxiong Basin during the late Paleocene. (C) Principal component morphospace of all tooth data analyzed; convex hulls delineate overall morphospace occupation during each time interval. Eigenvectors of the four dental topographic traits are indicated in blue. Late Paleocene shift and expansion in dental topographic morphospace is statistically significant at the p = 0.05 level (Table 1). Pantodont silhouette by S. Traver from phylopic.org.

Pairwise t test of dental topographic trait and disparity differences across adjacent time bins.

Dental topographic trait differences are assessed across time intervals in all-data, Chinese endemic pantodont, and non-pantodont partitions. Dental trait disparity was estimated based on all principal component axes using the outputs of PCA. Tooth size variance differences were calculated from tooth area or square root of tooth area in all-data and no-outlier partitions to assess effect of outliers on statistical significance (see Data S6 for details). Bolded font indicates p values < 0.05.

Correlation plots of dental topographic and bite performance traits in endemic Asian Paleocene mammals.

(A) Hypothetical correlation scenarios used to interpret stasis, directional, verses decoupled change through time in specimen data. (B) Pairwise ranked correlation coefficient estimated using Kendall’s τ between early and middle Paleocene dental topographic and performance traits in the main dataset. (C) Correlation between middle and late Paleocene traits in the main dataset. (D) Correlation between early and middle Paleocene traits in the Chines endemic pantodont (CEP) data partition. (E) Correlation between middle and late Paleocene traits in the CEP data partition. Topography-performance correlations are marked in red boxes. Decoupled/reversed trait correlations are marked in gray boxes.

Maps of the fossil locality areas and proportions of teeth analyzed, related to Fig. 1.

The Paleocene sedimentary basins sampled are indicated on a modern satellite image (A) and on a paleogeographic map of the K-Pg transition (B). Satellite map from NASA (www.nasa.gov) and paleogeographic map from C. Scotese (www.scotese.com)113. C. Proportions of tooth positions present in data partitions. M_1-M_3, lower first to third molars; M1_-M3_, upper first to third molars; P_2-P_4, lower second to fourth premolars; P2_-P4_, upper second to further premolars; E, early, M, middle, L, late. The middle Paleocene partition has no P_2 sampled, but otherwise the proportions of tooth positions present in each time interval are generally similar to each other.

Finite element modeling boundary conditions and convergence tests, related to Figs. 1 and 3.

(A) Depiction of compressive/crushing force imposed on a lower molar cusp in a specimen of Guichilambda zhaii (IVPP V12037) in lateral view. (B) Depiction of shearing motion during hemimandible eversion/inversion cycle in coronal view. (C) Depiction of shearing motion in lateral view. (D) Boundary conditions of compressive bite simulations on the lower first molar of Altilambda tenuis (IVPP V5231). (E) Boundary conditions of shearing bite simulations on A. tenuis. Nodal constraints are indicated by stars, compressive force by circle, and shearing forces by black arrows. (F) Three-dimensional deviation map of the first lower molar of Altilambda pactus (IVPP V5228) comparing geometric difference between models built form original versus cast-derived image data; positive (red) and negative (blue) deviation maxima +/− 0.58 mm. (G) Three-dimensional deviation map of the first lower molar of Altilambda tenuis (IVPP V5231), positive (red) and negative (blue) deviation maxima +/− 0.39 mm. (H) Convergence test of adjusted strain energy values (in Joules) of compression and shear bite simulations in IVPP V5228 and V5231. Percentage values indicate differences between adjacent model quantities (measured in number of three-noded triangular [tri3] elements). All model values at the lowest quantity models are considered converged based on a <=10% threshold criterion.

Sample variance (in squared units of each metric) of dental topographic metrics and dental performance traits for the overall dataset, related to Fig. 1.

Time bins refer to the early Paleocene (Bin A), middle Paleocene (Bin B), and late Paleocene (Bin C). Color bars represent different data partitions: all data (salmon), all molars (teal), all premolars (green), lower molars (blue), and upper molars (purple).

Boxplots of dental topographic and finite element simulated traits, related to Table 1.

Dark lines in panels A, D, and G denote sample variance per million years, calculated from 1,000 bootstrap samples of trait values pulled from uniform distributions within their estimated uncertainty ranges. The variance magnitudes are scaled to the duration of the Asian land mammal ages representing each time interval. (A) DTA (DNE, OPCR, Slope, RFI) and FEA (compressive bite strain energy, shear bite strain energy) boxplots representing the all tooth positions in the all-taxon partition (B) DTA metrics for the molar partition of all taxa. (C) DTA metrics for the premolar partition of all taxa. (D) DTA and FEA boxplots of all tooth position in the non-pantodont data partition. (E) DTA metrics for molars in the non-pantodont data partition. (F) DTA metrics for premolars in the non-pantodont data partition. (G) DTA and FEA boxplots of all tooth positions in the Chinese endemic pantodont (CEP) partition. (H) DTA metrics for molars in the CEP partition. (I) DTA metrics for premolars in the CEP partition.

Trends through time for DTA and FEA traits by tooth position, related to Fig. 1.

(A) Categories of major trends observed through the three Paleocene time intervals. (B) Proportion of individual tooth position analyses that support the trend observed in the overall dataset. The first set of values are per-tooth variance; the second set of values are from pooled sample and per-tooth range lengths. (C) Individual tooth position variance trends in DTA and FEA trait values. (D) Individual tooth position range length trends in DTA and FEA trait values.

Linear regression analysis of dental topographic and bite performance datasets different data partitions, related to Fig. 3.

(A) All-taxon all-teeth data partition; data points are colored according to early (red), middle (green), and late (blue) Paleocene age of the specimens. (B) All-taxon all-teeth data partition without time bine groups. (C) Chinese endemic pantodont (CEP) all-teeth data partition. (D) CPE molar data partition. (E) CEP premolar data partition. (F) Non-pantodont all-teeth partition. (G) Non-pantodont molar partition. (H) Non-pantodont premolar partition. Performance variables estimated using finite element analysis includes tooth crown strain energy under compressive (left column within each panel) and shear bite (right column within each panel) simulations, respectively. Rows within panels represent different DTA metrics (from top to bottom: DNE, OPCR, Slope, RFI). Fitted regression lines are shown in blue; 95% confidence envelopes are shown in gray shades.

Tooth size boxplots and adjusted r2 distributions from DTA-FEA linear regression analyses, related to Fig. 3.

(A) Boxplots of tooth size by 2D area (left column within panel) or square root of 2D area (right column within panel). The top row within panel A includes sampling outliers (very large Chinese endemic pantodonts, CEPs), bottom row excludes outliers. The larger CEPs are removed in the bottom plots in order to more fully display the distribution of tooth sizes across the bimodal distribution of sizes in the main data cluster in each time bin, respectively.(B) Distribution of adjusted r2 values from linear regression analysis of dental topographic and dental performance datasets for all-taxon all-teeth data partition. (C) CEP all-teeth partition. (D) CEP molar partition. (E) CEP premolar partition. (F) Non-pantodont all-teeth partition. (G) Non-pantodont molar partition. (H) Non-pantodont premolar partition. Adjusted r2 values were generated from 1,000 bootstrap samples of DNE and FEA traits values for each specimen from uniform distributions that incorporate uncertainty in DNE and FEA values estimates. P-values calculated from t test of bootstrap sample p values against a hypothesis of p >= 0.05.

Tooth size disparity bar plots, related to Fig. 1.

(A) Variance of tooth size (sqrt of 2D tooth area) in different data partitions. (B) Range length of tooth size (sqrt of 2D tooth area) in different partitions.

Correlation through time plots for upper molar DTA and FEA data, related to Fig. 3.

The upper diagonal in the left column shows early Paleocene (EP) correlations, the lower diagonal in the left column shows middle Paleocene (MP) correlations. The upper diagonal in the right column shows middle Paleocene correlations, and the lower diagonal in the right column shows late Paleocene (LP) correlations. Red boxes indicate correlations between DTA and FEA traits. Dark gray boxes indicate decoupling of correlation directions across time intervals. Upper M1 patterns general reflect the trend recovered from analysis of the overall dataset, but M2 and M3 results display inconsistent DTA-FEA correlations, possibly due to small sample sizes.

Correlation through time plots for lower molar DTA and FEA data, related to Fig. 3.

The upper diagonal in the left column shows early Paleocene (EP) correlations, the lower diagonal in the left column shows middle Paleocene (MP) correlations. The upper diagonal in the right column shows middle Paleocene correlations, and the lower diagonal in the right column shows late Paleocene (LP) correlations. Red boxes indicate correlations between DTA and FEA traits. Dark gray boxes indicate decoupling of correlation directions across time intervals. Lower molar patterns generally replicate those recovered in the overall analyses, but lower M1 and M2 signals appear to be stronger than those for lower M3.

Correlation through time plots for premolar DTA and FEA data, related to Fig. 3.

The upper diagonal in the left column shows early Paleocene (EP) correlations, the lower diagonal in the left column shows middle Paleocene (MP) correlations. The upper diagonal in the right column shows middle Paleocene correlations, and the lower diagonal in the right column shows late Paleocene (LP) correlations. Red boxes indicate correlations between DTA and FEA traits. Dark gray boxes indicate decoupling of correlation directions across time intervals. Low sample sizes make premolar correlations unstable, with general pattern showing EP-MP strengthening then MP-LP stasis or weakening

Two-block partial least squares (PLS) r coefficients from bootstrapped analyses, related to Fig. 3.

1,000 bootstrap samples of DTA and FEA values were taken from uniform distributions of trait uncertainty ranges, and two-block partial east squared analysis conducted on each sample for early-middle Paleocene and middle-late Paleocene data partition pairings, respectively. All r values are statistically significantly different (p < 0.05) between adjacent time intervals, based on a one-sample t test of the distribution of 1,000 p values from the bootstrap two-block PLS samples against p < 0.05.

Relative percentages of fossil pollen found in the Nanxiong Basin, Related to Fig. 2 and based on 96.

Sample size, disparity, and mean tooth size by tooth position. Related to Fig. 1.

Mean disparity difference is measured by pair-wise variance tests (var.test); mean tooth size difference is measured by pairwise t tests (t.test). p values <=0.05 are shaded. Overall disparity trends are also observed in premolar and upper molar data partitions, whereas overall tooth size trends are observed mainly in the lower premolar 4 data partition. Decrease in mean tooth size is most consistently observe across multiple tooth partitions for the early Paleocene to late Paleocene comparison.