Projected climate change threatens significant range contraction of Cochemiea halei (Cactaceae), an island endemic, serpentine-adapted plant species at risk of extinction

Abstract

Aim: Threats faced by narrowly distributed endemic plant species in the face of the Earth's sixth mass extinction and climate change exposure are especially severe for taxa on islands. We investigated the current and projected distribution and range changes of Cochemiea halei, an endemic island cactus. This taxon is of conservation concern, currently listed as vulnerable on the International Union for the Conservation of Nature Red List and as a species of special concern under Mexican federal law. The goals of this study are to (a) identify the correlations between climate variables and current suitable habitat for C. halei; (b) determine whether the species is a serpentine endemic or has a facultative relationship with ultramafic soils; and (c) predict range changes of the species based on climate change scenarios.

Location: The island archipelago in Bahía Magdalena on the Pacific coast, Baja California Sur, Mexico.

Methods: We used temperature and precipitation variables at 30-arc second resolution and soil type, employing multiple species distribution modeling methods, to identify important climate and soil conditions driving current habitat suitability. The best model of current suitability is used to predict possible effects of four climate change scenarios based on best-case to worst-case representative concentration pathways, with projected climate data from two general circulation models, over two time periods.

Main conclusions: The occurrence of the species is found to be strongly correlated with ultramafic soils. The most important climate predictor for habitat suitability is annual temperature range. The species is predicted to undergo range contractions from 21% to 53%, depending on the severity and duration of exposure to climate change. The broader implications for a wide range of narrowly adapted, threatened, and endemic plant species indicate an urgent need for threat assessment based on habitat suitability and climate change modeling.

Keywords: Cactaceae; biodiversity; biogeography; climate change; endangered species; island endemism; range shifts; serpentine adaptation; species distribution modeling.

Figure 1

The Bahía Magdalena region and islands, the known area of distribution of Cochemiea halei. The islands and named land masses in Bahía Magdalena: (a) Cabo San Lazaro, the most northwesterly land mass of the islands. (b) The main land mass of Isla Magdalena. (c) Isla Margarita. Map created using ArcGIS® software by Esri. ArcGIS^® and ArcMap™ are the intellectual property of Esri and are used herein under license. Copyright^© Esri. All rights reserved

Figure 2

Cochemiea halei in habitat on Isla Magdalena, growing in pure ultramafic rock

Figure 3

Predictions of suitable habitat for Cochemiea halei. The map shows predictions of habitat suitability, on a probability scale of zero (transparent) to 1 (dark green). The model predictions derive from a BRT method, using WorldClim V. 2.0 data, at 30‐arc sec resolution. 44 presences and 207 pseudoabsences were used. The following variables were used: annual temperature range, the mean temperature of the warmest quarter (July–September), precipitation of the warmest quarter (July–September), precipitation of the coldest quarter (December–February), and soil type. The model fitted 11,125 trees, with a 10‐fold cross‐validated AUC of 0.96. The parameters used for the boosted regression tree analysis were a tree complexity of 2, a learning rate of 0.0007, bag fraction of 0.7, and a step size of 25

Figure 4

Partial response plots of climate variables used in species distribution modeling for Cochemiea halei. The plots show the marginal response of C. halei to each variable. The variables are (a) annual temperature range, “Trng,” (b) average temperature of the warmest quarter, “Awarm,” (c) precipitation of the warmest quarter, “Pwarm,” and (d) precipitation of the coldest quarter. The y‐axis for each plot is on a logit scale, showing the relative impact of values of the variable on the probability of occurrence. The x‐axis for the temperature variables, (a) and (b), is in degrees C. Precipitation variables, (c) and (d), are in mm. The x‐axis is marked with a decile rug plot

Figure 5

Percent contributions of climate predictors to model performance in estimating the current suitable habitat of Cochemiea halei. Percent influence: annual temperature range, 40.5%; average temperature of the warmest quarter, 21%; soil type, 16%; precipitation of the coldest quarter (December–February), 15%; precipitation of the warmest quarter (July–September), 7.2%

Figure 6

Percent changes in predicted current suitable habitat of Cochemiea halei, with soil type and without soil type. Values on the scale bar and their corresponding colors on the map show percent increase or percent decrease in predicted probability of occurrence when soil type is not used as a predictor. The majority of the map predicts an increase in the probability of occurrence (green colors), especially in areas where ultramafic soils are not predominant. Also, the model without soil type predicts less suitable habitat (red), in some areas that are predominantly ultramafic soils. The best model without soil type fitted 12,650 trees, with a 10‐fold cross‐validated AUC of 0.95. The parameters used for the boosted regression tree analysis were a tree complexity of 1, a learning rate of 0.0007, bag fraction of 0.7, and a step size of 15

Figure 7

Predictions of range expansion and contraction for Cochemiea halei, to the year 2070, using two representative concentration pathways. Light green areas represent projected areas of range expansion, and red areas represent projected areas of range contraction. (a) predicted expansion and contraction for a representative concentration pathway (RCP) of 2.6, representing relatively low climate change exposure, and (b) predicted expansion and contraction for RCP 8.5, representing a high degree of climate change exposure. The model used for projection was the best performing BRT model with soil type. Projections were done using the mean values from the same climate variables used in habitat suitability modeling, using the Hadley Center Global Environmental Model version 2‐ES (HadGEM2‐ES) and the Community Climate System Model v. 4 (CCSM4).

Figure 8

Influence of the two strongest climate predictors on habitat contraction, refuge, and expansion, for Cochemiea halei, for four representative concentration pathway scenarios, projected to 2070. Each box plot shows the range, 1st quartile, median, and 3rd quartile of annual temperature ranges and average temperature of the warmest quarter for each type of predicted future habitat prediction: contraction, refuge, expansion, and the present habitat. The y‐axes are degrees C. All data are from the projections to 2070.