Assessing the need and potential of assisted migration using species distribution models

Highlights

• Species distribution models provide useful insights for assisted migration (AM).

• Model outputs can be converted into metrics describing the need and potential for AM.

• Metrics combine losses and gains of suitable areas for a species into one index.

• Several dispersal-limited species may benefit from AM under strong climate change.

• Choice of climate variables has a larger effect on metrics than choice of modeling method.

Abstract

Assisted migration (AM) has been suggested as a management strategy for aiding species in reaching newly suitable locations as climate changes. Species distribution models (SDMs) can provide important insights for decisions on whether to assist a species in its migration; however, their application includes uncertainties. In this study, we use consensus SDMs to model the future suitable areas for 13 vascular plant species with poor dispersal capacity. Based on the outputs of SDMs under different climate change scenarios and future times, we quantify the predicted changes in suitable area by calculating metrics that describe the need and potential for migration. We find that, by the end of the 21st century, one of the species would benefit from AM under mild climate change, seven under moderate change, and for 12 out of 13 species studied AM appears to be a relevant conservation method under strong climate change. We also test the effect of different modeling attributes on the metrics and find little variation between SDMs constructed using different combinations of modeling methods and variable sets. However, the choice of climate variables had a larger influence on the level of the metrics than did the modeling method. We therefore suggest that the choice of climate variables should receive ample attention when measuring climate change threat using SDMs and that experiments aiming to uncover critical environmental factors for individual species should be extensively conducted. This study illustrates that dispersal assistance may be needed for many species under a wide range of possible future climates.

Introduction

Climate change is projected to cause accelerating changes in species' distributions and abundances, resulting, at worse, to increased risk of extinction (Dawson et al., 2011, Urban, 2015). Species may adjust to climatic changes by adapting in situ or by dispersing to areas where climate is becoming favorable (Dawson et al., 2011, Moritz and Agudo, 2013). However, the ability of species to adapt via evolutionary changes may be limited (Settele et al., 2014), and poor mobility and dispersal barriers may hamper their dispersal to new areas (Schloss et al., 2012, Corlett and Westcott, 2013). Under such circumstances, human intervention in the form of conservation and adaptation actions may allow persistence of species. Assisted migration (AM; also referred to as managed relocation or assisted colonization; Richardson et al., 2009, Seddon, 2010, Hewitt et al., 2011, Hällfors et al., 2014) has been suggested as one of the several management strategies for aiding species in adjusting to rapid climate change (Chauvenet et al., 2013) and approaches for recognizing species that would benefit from AM have been presented (e.g., Hoegh-Guldberg et al., 2008, Richardson et al., 2009, Pérez et al., 2012).

Deciding on AM for particular species involves various questions that range from ecological evaluations to assessments of invasion potential, legal constraints, and societal acceptability (Richardson et al., 2009, Schwartz, 2012). Nonetheless, AM frameworks often begin with an assessment of the ecological threats stemming from climate change (e.g. Hoegh-Guldberg et al., 2008, Richardson et al., 2009, Rout et al., 2013). Climate change vulnerability assessments (Pacifici et al., 2015) provide useful general insights for recognizing candidate species but they cannot specify whether AM is an effective and necessary method to enable persistence of a certain species, as they only characterize climate change sensitivity. In comparison, developing predictions of spatial changes in suitable area using, e.g., species distribution models (SDMs) and climate scenario projections from General Circulation Models (GCMs) can generate information on climate change exposure and be used in guiding conservation efforts (Schwartz, 2012, Guisan et al., 2013). In the context of AM, SDMs would typically be used to provide ‘first-filter’ assessments of climatically suitable areas for species translocation. Here we use distribution data on dispersal-limited European vascular plant species to model range change and quantify their need and potential for assisted species' dispersal. The theoretical basis for this approach is described in Hällfors et al. (manuscript), and this study provides the first empirical application of the approach.

The ultimate reason for using AM would be to avoid extinction of the focal species. Model-generated projected spatial changes in suitable area can be categorized as loss and gain of area of potential occupancy (Thomas et al., 2011). For AM to be relevant, a species needs to 1) have a need for migration, i.e., experience a considerable loss of suitable area, 2) have potential for migration, i.e., experience a considerable gain of new suitable area, and 3) have an inability to disperse, i.e., it cannot make use of the gained area (Hällfors et al. manuscript).

Here, we apply SDMs and use the results to quantify AM suitability for 13 vascular plant species, selected for poor dispersal ability, under various climate change scenarios and time periods. We first assess the migration need and migration potential for our study species by comparing the current range of the species and SDM projections of future suitable area. Next we combine these to arrive at an index that can be categorized based on thresholds to describe potential AM benefit (for details see Hällfors et al. manuscript). High values of the index indicate that the species has both the need of AM and high potential for AM to be effective.

We also address various sources of uncertainty in SDMs (Heikkinen et al., 2006, Synes and Osborne, 2011, Schwartz, 2012) as they are likely to affect any quantitative metrics based on SDM outputs. We develop multiple SDMs for our study species, using different modeling methods and sets of climatic predictor variables. We also develop projections of suitable areas under different climate change scenarios and for different time periods under the 21st century. This allows us to distinguish scenario-specific and temporal trajectories for the metrics as each species receives a value describing migration need and potential for each scenario and time period that the model is projected onto. We report the calculated metrics based on a consensus approach (Araújo and New, 2007) of the SDMs, i.e., predictions supported by the majority of the SDMs. However, to gain insight into the degree of variation in the results that the choice of modeling method and climate variable set can bring about, we also test the effect of these choices on the metrics.