TABLE 1. Linear regression analyses for Pinnipedimorpha and extant families’ diversity through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Diversity of each group is considered from the time bin of first occurrence (Chattian for Pinnipedimorpha, Burdigalian for extant families). Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each group are reported, for complete results see Appendix 2, Table S1.
Pinnipedimorpha  Phocidae  Otariidae  Odobenidae  
R^{ ²}  0.909  0.775  0.993  0.906 
AIC  29.199  22.052  14.018  1.180 
pvalue  0.008  0.087  0.001  0.016 
Best model  A+B+C+D  A+B+D  A+B+C+D  A+B+D 
TABLE 2. Linear regression analyses for Pinnipedimorpha habitat preference through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Only prevalent categories (with at least one time bin with a count > 5) were considered. Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each category are reported, for complete results see Appendix 2, Table S2.
Marine coastal  Marine benthic  Marine pelagic  
R^{²}  0.914  0.847  0.885 
AIC  28.844  0.387  7.244 
pvalue  0.004  0.004  0.036 
Best model  A+C+D  A+C  A+B+C+D 
TABLE 3. Linear regression analyses for Pinnipedimorpha prey type preference through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Only prevalent categories (with at least one time bin with a count > 5) were considered. Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each category are reported, for complete results see Appendix 2, Table S3.
Fish  Squid/soft bodied invertebrates  Benthic invertebrates  
R^{²}  0.896  0.982  0.969 
AIC  27.725  9.734  4.775 
pvalue  0.011  < 0.0001  < 0.0001 
Best model  A+B+C+D  A+D  A+B+D 
TABLE 4. Linear regression analyses for Cetacea and Odontoceti and Mysticeti diversity through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each category are reported, for complete results see Appendix 2, Table S4.
Cetacea  Odontoceti  Mysticeti  
R^{²}  0.674  0.667  0.706 
AIC  90.967  84.175  58.829 
pvalue  0.001  0.001  0.001 
Best model  A+C  A+C  A+C 
TABLE 5. Linear regression analyses for Cetacea habitat preference through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each category are reported, for complete results see Appendix 2, Table S5.
Riverine  Estuarine  Marine coastal  Marine benthic  Marine pelagic  
R^{²}  0.601  0.240  0.676  0.906  0.710 
AIC  7.083  40.692  79.988  14.579  89.199 
pvalue  0.004  0.193  0.001  < 0.0001  0.001 
Best model  A+D  A+D  A+C  A+B+C  A+C 
TABLE 6. Linear regression analyses for Cetacea prey type preference through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each category are reported, for complete results see Appendix 2, Table S6.
Fish  Squid/softbodied inv.  Zooplankton  Benthic inv.  Tetrapods  
R^{²}  0.663  0.743  0.698  0.837  0.036 
AIC  89.536  73.845  59.565  6.252  27.313 
pvalue  0.001  0.001  0.001  < 0.0001  0.500 
Best model  A+C  A+B+C  A+C  B+C  C 
TABLE 7. Linear regression analyses for Cetacea prey capture strategies through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each category are reported, for complete results see Appendix 2, Table S7.
Biting/crushing  Suction  Filter  
R^{²}  0.490  0.762  0.703 
AIC  72.148  74.033  59.146 
pvalue  0.018  0.001  0.001 
Best model  A+C  A+B+C  A+C 
TABLE 8. Linear regression analyses for Sirenia and Desmostylia diversity through time against climate and productivity variables: δ^{18}O (variable A), δ^{13}C (variable B), diatom species diversity (variable C), sea level changes (variable D). Best model, including one, two, three, or all four variables, is selected based on its AIC (Akaike’s Information Criterion) score. Diversity of Desmostylia is considered from the time bin of first occurrence to the last (from Rupelian to Tortonian). To account for their low diversity, these groups were also considered together (“Sirenia and Desmostylia combined”). Significant values (p < 0.05) are in bold. Only data (R^{2}, pvalue, AIC) for the best model for each group are reported, for complete results see Appendix 2, Table S8.
Sirenia  Desmostylia  Sirenia and Desmostylia combined 

R^{²}  0.243  0.603  0.460 
AIC  27.978  5.200  37.158 
pvalue  0.189  0.158  0.025 
Best model  A+C  B+D  A+D 
TABLE 9. Correlation test (Pearson’s coefficient) for Sirenia, Desmostylia, and “Sirenia and Desmostylia combined” diversity through time with climate and productivity variables: δ^{18}O, δ^{13}C, diatom species diversity, sea level changes. Significant values (p < 0.05) are in bold. Cells for autocorrelated variables are grayed out. Pvalues for the correlations are reported in Appendix 2, Table S9.
Sirenia  Desmostylia  Sirenia and Desmostylia combined 

Sirenia  0.040  
Desmostylia  0.040  
Sirenia & Desmostylia  
δ^{18}O  0.252  0.207  0.168 
δ^{13}C  0.353  0.068  0.128 
Diatom diversity  0.870  0.660  0.742 
Sea level  0.823  0.379  0.554 