The set of environmental conditions under which a taxon can survive and maintain viable populations, known as the ecological niche, is a fundamental determinant of a taxon's distribution. Because of the central importance of ecological niches, they have been assumed to remain relatively stable during intervals of morphological stasis. However, the assumption of niche stability has rarely been tested directly with fossil data spanning multiple temporal intervals. Thus, the conditions under which this assumption is likely to be accurate are not well understood. In this study, we use ecological niche modeling (ENM) to reconstruct the ecological niche for 11 genera of marine benthos (crinoids, trilobites, molluscs, bryozoans, and corals) from the Type Cincinnatian Series (Late Ordovician, Katian Stage) across nine temporal intervals spanning approximately three million years. This interval includes both abiotic environmental change (gradual sea-level fall) and biotic change (rapid pulses of the Richmondian Invasion), thus allowing the relative effect of different environmental perturbations to be constrained. A previous symmetrical analysis of niche stability of brachiopod species recovered an increase in niche evolution following the Richmondian Invasion. Herein we test the generality of the brachiopod pattern within the community. Niche stability was evaluated in geographic space, ecological space, and niche parameter space. Niche stability varied through time; during the Pre-Invasion interval, taxa exhibited niche stability during gradual shallowing of sea level in the basin, whereas niche evolution became more common during the Richmondian Invasion. Taxa adjusted to the increased competition by altering aspects of their niche. Notably, surviving taxa contracted their niche into a subset of their previous niche parameters. This represents an adaptive response to increased competition for resources with the newly established invader taxa, and it was employed most successfully by generalist taxa. Patterns of niche evolution were congruent between clades, among feeding styles, and across taxonomic levels.
Supplemental Text
Calculation of percent geographic overlap between models.
Supplemental Figure 1
Maxent output example for Flexicalymene for the first time-slice of
the C5: (A) Original Distribution Model; (B) Forward Projection Distribution Model generated by projecting the niche modeled developed for the first time-slice onto the environmental parameters of second time-slice; (C) Omission and Predicted Area; (D) Sensitivity vs. 1-Specificity including AUC values; (E) Jackknife of regularized training gain; (F) Jackknife of test gain; (G) Response curves for each variable.
SupplementalFigure1_MaxentOutput.pdf
Supplemental Table 2
Occurrence data by time-slice.
SupplementalTable2_OccurrenceData.pdf
Supplemental Table 1
List of localities from which occurrence data were collected.
SupplementalTable1_CollectingLocalities_HB_Revised.pdf
Supplemental Table 3
Environmental Parameters (Modified from Malizia & Stigall, 2011).
SupplementalTable3_EnvironmentalCoding.pdf
Supplemental Table 4
Area Under the Curve (AUC) values for GARP and Maxent Models
SupplementalTable4_ModelAUCvalues.pdf
Supplemental Table 5
Weighted percent of geographic overlap between models generated by
GARP and Maxent per time slice.
SupplementalTable5_MaxentGARPoverlapALS.pdf
Supplemental Table 6
Overlap calculations for percent geographic overlap between Maxent models. The weighting scheme and percent overlap calculations are described in Supplemental Text.
SupplementalTable6_Geographicoverlap.pdf
Supplemental Table 7
ENMtools I and D values relative to the 95% confidence interval. Both of these metrics are measured from 0 (complete dissimilarity) to 1 (identity). I/D values that fall within the 95% C.I. (“in”) indicate models that are no more/less similar than random and are interpreted as evidence of niche conservatism. I/D values that fall above the 95% C.I. (“out, high”) and indicate statistical niche conservatism. I/D values that fall below the 95% C.I. (“out, low”) are interpreted as evidence of niche evolution.
SupplementalTable7_IDvalues.pdf
Supplemental Table 8
Percent contribution per environmental variable for each Maxent model.
SupplementalTable8_EnvPercentCont.pdf
Supplemental Appendix 1
List of references for localities in Supplemental Tables 1 and 2
Supplemental_Appendix1references_HB_Revised.pdf
Supplemental Appendix 2
Legend for models and analyses in Supplemental Appendices 3 and 4.
SupplementalAppendix2_Legend.pdf
Supplemental Appendix 3
Reclassified GARP models. Each figure shows the reclassified GARP models for each genus in each time-slice of each sequence. Note that GARP values indicate suitability and range from 0 to 100. See Supplemental Appendix 2 for explanation of the color values.
SupplementalAppendix3_GARPoutput.pdf
Supplemental Appendix 4
Original Maxent Models, Reclassified Maxent Models, and Geographic overlap analyses. Each figure shows the original Maxent model outputs in the left column, and the reclassification of that model to the right. The far right images are the final reclassified overlap analysis of the forward projected model and the original model of the next consecutive time slice. Model output values range from 0.0 to 1.0 and indicate probability. Final overlap calculation values indicate agreement, and range from 1 to 4. See Supplemental Appendix 2 for explanation of symbology.
SupplementalAppendix4_Maxent Models.pdf