A species distribution model of the giant kelp Macrocystis pyrifera: Worldwide changes and a focus on the Southeast Pacific

Abstract

Worldwide climate-driven shifts in the distribution of species is of special concern when it involves habitat-forming species. In the coastal environment, large Laminarian algae-kelps-form key coastal ecosystems that support complex and diverse food webs. Among kelps, Macrocystis pyrifera is the most widely distributed habitat-forming species and provides essential ecosystem services. This study aimed to establish the main drivers of future distributional changes on a global scale and use them to predict future habitat suitability. Using species distribution models (SDM), we examined the changes in global distribution of M. pyrifera under different emission scenarios with a focus on the Southeast Pacific shores. To constrain the drivers of our simulations to the most important factors controlling kelp forest distribution across spatial scales, we explored a suite of environmental variables and validated the predictions derived from the SDMs. Minimum sea surface temperature was the single most important variable explaining the global distribution of suitable habitat for M. pyrifera. Under different climate change scenarios, we always observed a decrease of suitable habitat at low latitudes, while an increase was detected in other regions, mostly at high latitudes. Along the Southeast Pacific, we observed an upper range contraction of -17.08° S of latitude for 2090-2100 under the RCP8.5 scenario, implying a loss of habitat suitability throughout the coast of Peru and poleward to -27.83° S in Chile. Along the area of Northern Chile where a complete habitat loss is predicted by our model, natural stands are under heavy exploitation. The loss of habitat suitability will take place worldwide: Significant impacts on marine biodiversity and ecosystem functioning are likely. Furthermore, changes in habitat suitability are a harbinger of massive impacts in the socio-ecological systems of the Southeast Pacific.

Keywords: Southeast Pacific; climate change; distribution model; habitat forming; kelp forests; projection.

FIGURE 1

Global occurrences of Macrocystis pyrifera since 1970 downloaded from GBIF and used in this study. The occurrences were clean and a subsample were selected by creating a grid of 9.2 km² and randomly sampled one occurrence per grid cell to reduce the spatial aggregation of records.

FIGURE 2

Predicted probability of occurrence of Macrocystis pyrifera of the variables of ${\mathrm{SST}}_{\min }$ (Sea Surface Temperature minimum), ${\mathrm{SST}}_{\text{mean}}$ (Sea Surface Temperature mean), ${\mathrm{SST}}_{\max }$ (Sea Surface Temperature maximum), and salinity for the present model (subset predictors). The mean of the 100 replicates is shown in red and the mean ± one standard deviation is shown in blue.

FIGURE 3

The suitable habitat modeled for Macrocystis pyrifera with the variables of ${\mathrm{SST}}_{\min }$, ${\mathrm{SST}}_{\text{mean}}$, ${\mathrm{SST}}_{\max }$, and salinity. Different parts of the world are represented: (a) North‐West Pacific (Alaska/Canada), (b) Southeast Indian and Southwest Pacific (Australia/New Zealand), (c) North‐West Pacific (EU/Mexico), and (d) Southeast Atlantic (South Africa). The figure compares the distributions obtained in this model (subset predictors) with the future scenario 8.5 of 2090–2100, where the conserved distribution of the suitable habitat is shown in blue, the lost in red, and the gained in green. The distribution of habitat suitability was enlarged in thickness for better visualization.

FIGURE 4

The suitable habitat modeled for Macrocystis pyrifera with the variables of ${\mathrm{SST}}_{\min }$, ${\mathrm{SST}}_{\text{mean}}$, ${\mathrm{SST}}_{\max }$ and salinity for the Southeast Pacific. The figure compares the distributions obtained in this model (subset predictors) with the future RCP8.5 scenario for 2090–2100, where the conserved distribution of the suitable habitat is shown in blue, the lost in red, and the gained in green. The habitat suitability distribution was enlarged in thickness for better visualization.