Premise: Ecogeographic isolation, or geographic isolation caused by ecological divergence, is thought to be of primary importance in speciation, yet is difficult to demonstrate and quantify. To determine whether distributions are limited by divergent adaptation or historical contingency, the gold standard is to reciprocally transplant taxa between their geographic ranges. Alternatively, ecogeographic isolation is inferred from species distribution models and niche divergence tests based on widely available environmental and occurrence data.

Methods: We test for ecogeographic isolation between two sister species of California annual wildflowers, Clarkia concinna and C. breweri, with a hybrid approach. We use niche models to predict water availability as the major axis of ecological divergence and then test that with a greenhouse experiment. Specifically, we manipulate water availability in field soils for two populations of each species and predict higher fitness in conditions representing home habitats to those representing the environment of each’s sister species.

Key Results: Water availability and soil representing C. concinna generally increased both species’ fitness. Thus, water and soil may indeed limit C. concinna from colonizing the range of C. breweri, but not vice versa. We suggest that the competitive environment and pollinator availability, which are not directly captured with either approach, may be key biotic factors correlated with climate that contribute to unexplained ecogeographic isolation for C. breweri.

Conclusions: Ours is a valuable approach to assessing ecogeographic isolation, in that it balances feasibility with model validation, and our results have implications for species distribution modeling efforts geared towards predicting climate change responses.

README File Manifest.txt lists and describes all the files.

File Name:

CCCB_Fitness.csv contains the results from our greenhouse reciprocal transplant experiment. The number of variables has been reduced to just what we used for our study.

CCCB_Fitness_metadata.csv contains explanations for each variable name in CCCB_Fitness.csv.

Greenhouse_GLM_Fig5_6.R is the R code used in creating the zero-inflated and conditional generalized linear mixed models, as well as the graphs for Figure 5, Figure 6 and Supplemental Figure XX.

Bioclim_Figure2b_WilcoxonTests.R is the R script used to create box plots of the values associated with each of the BioClim variables with >10% percent contributions to our species distribution models.

c.c_occurence_data_cleaned.csv is the occurrence data for Clarkia concinna used in conjunction with BioClim data.

c.b _occurence_data_cleaned.csv is the occurrence data for Clarkia breweri used in conjunction with the BioClim data.

SDM_PctContribution_Graph_Fig2a.R is the R code used to create Figure 2a.

Env_Var_CCCB_allbioclimplussoil_REDUCED.csv is the data used in SDM_PctContribution_Graph_Fig2a.R to create Figure 2a from our species distribution model percent contribution of each BioClim variable and Rock Type.

RockType_Chisquare_Graph_Figure3.R is the R code used to create Figure 3, showing the proportion of occurrences of each study species on primary parent rock types. This code also contains the chi-squared test for significance between the two species.

primary_proportion_soil_2.csv is the data used to create Figure 3, containing the proportion of occurrences of each study species on primary rock types.

CC_CB_Primary_Rock_Reduced.csv is the data used for the chi-squared test in RockType_Chisquare_Graph_Figure3.R.

RockType1Raster.tif is the raster of the primary parent rock type and was derived from the USGS California geologic map polygon shape file (Ludington et al. 2005)

RockType1Raster.tif is the raster of the secondary parent rock type and was derived from the USGS California geologic map polygon shape file (Ludington et al. 2005)

Rock Type Lookup - Rocktype1 index.csv is used to look up lithology from numeric rock ids in RockType1Raster.tif.

Rock Type Lookup - Rocktype2 index.csv is used to look up lithology from numeric rock ids in RockType2Raster.tif.

Prep_Occurrence_Data.R is the script we used to download GBIF data of the two Clarkia species and remove erroneous entries. 

Prep_Targetted_Background_Data.R is the script we used to download and prepare the GBIF data of all of the non-clarkia species we used as pseudo-absence data. 

Prep_Env_Rasters.R is the script we used to prep environmental data raster bricks that were used to create the max ent models

Run_MaxEnt_Models.R is the script we used to build MaxEnt models for the two species.

CompetitiveEnvironment.R is the script to compare the competitive environments of Clarkia concinna and C. breweri.

CoverData.csv, CoverDataTTests.csv, and CoverMetaData1.csv are input files for CompetitiveEnvironment.R and contain the competitive environment data and other site characteristics. CoverMetaData1.csv shows how the species (“sp”) and site codes (“sitecode”) in CoverDataTTests.csv correspond to the details in Appendix S4. “Observer” indicates who recorded the field data, and is either Kathleen Kay (“Kathleen”) or Shelley Sianta (“Shelley”).

BroennimannAnalysis.R is the script we used to measure niche equivalency, measure niche similarity of the Clarkia species, and plot their environmental niche in PCA space. This script was adapted from Broennimann et al. (2012).