Are plant species able to keep pace with the rapidly changing climate?

Abstract

Future climate change is predicted to advance faster than the postglacial warming. Migration may therefore become a key driver for future development of biodiversity and ecosystem functioning. For 140 European plant species we computed past range shifts since the last glacial maximum and future range shifts for a variety of Intergovernmental Panel on Climate Change (IPCC) scenarios and global circulation models (GCMs). Range shift rates were estimated by means of species distribution modelling (SDM). With process-based seed dispersal models we estimated species-specific migration rates for 27 dispersal modes addressing dispersal by wind (anemochory) for different wind conditions, as well as dispersal by mammals (dispersal on animal's coat - epizoochory and dispersal by animals after feeding and digestion - endozoochory) considering different animal species. Our process-based modelled migration rates generally exceeded the postglacial range shift rates indicating that the process-based models we used are capable of predicting migration rates that are in accordance with realized past migration. For most of the considered species, the modelled migration rates were considerably lower than the expected future climate change induced range shift rates. This implies that most plant species will not entirely be able to follow future climate-change-induced range shifts due to dispersal limitation. Animals with large day- and home-ranges are highly important for achieving high migration rates for many plant species, whereas anemochory is relevant for only few species.

Figure 1. General work flow of the study: We compared the modelled potential future range shift rates and the modelled migration rates.

Future range shift rates can be seen as a measure of the distances that are required to be covered per year and the migration rates as a measure for the distances that can be covered by migration per year by plant species. The future range shifts were modelled by means of species distribution modelling (SDM), considering nine different environmental models for 2080. The migration rates were modelled by means of process-based models considering 27 different dispersal modes. For a coarse plausibility check, we tested if the modelled migration rates (maximum level estimation) can explain the modelled postglacial range shifts (minimum level estimation). The postglacial range shifts were also modelled by means of SDM. The comparison of the modelled potential future range shifts and the migration rates was carried out in a direct comparison of the annual rates as well as in a spatial explicit comparison of the potential distributions assuming no migration, full migration and „realistic“ migration (based on the modelled migration rates. We calculated the percentage of the predicted future range that is reached assuming the modelled migration rates for different dispersal modes (range filling).

Figure 2. Predicted future range shifts (annual averages) according to the nine environmental models for 2080.

Predicted shifts of the centroids (A) and of the range margins (B). Each boxplot represents N = 140 plant species.

Figure 3. Predicted migration rates for A) dispersal by wind (anemochory) for nine different meteorological scenarios B) dispersal by animals (endozoochory) for nine different animal species C) dispersal by animals (epizoochory) for nine different animal species.

Each box represents N = 140 plant species.

Figure 4. Potential dispersal limitation on the example of Geum urbanum.

A) Comparison of the potential future range shift rates according to the nine environmental models and the process-based modelled migration rates according to the 27 dispersal modes for Geum urbanum. The potential future range shift rates can be considered an estimator for the migration rates required in order to fulfil the potential future range completely. They are displayed as dots (black: centroid method and grey: margins method). The process-based modelled migration rates are displayed as black crosses. The values for the dispersal mode and the environmental model used in the map in fig. 4B (epizoochorous dispersal by Cervus elaphus and the A1 CCCMA environmental model for 2080) are marked by red circles. B) Potential range shift and dispersal limitation on the example of Geum urbanum. The map is based on a realized migration rate of 1.12 km/a corresponding to epizoochorous dispersal by Cervus elaphus. The predicted future range is according to the A1 CCCMA environmental model for 2080. Projection: Europe Albers Equal Area Conic.

Figure 5. Percentage of the predicted future range that is reached assuming dispersal by wind (A), endozoochory (B) and epizoochory (C) respectively.

The potential future range was estimated according to the A1 CCCMA environmental model for 2080. Each boxplot represents N = 140 plant species (see also fig. S6 and table S8).

Figure 6. Biodiversity loss due to dispersal limitation in terms of the considered 140 plant species.

A) Difference between predicted future distributions (2080) assuming full dispersal and “realistic” dispersal (according to our modelled migration rates taking 27 dispersal modes for migration into account): The differences were calculated for each of the nine environmental models and then averaged. In grey: areas where very few of the 140 species are predicted to occur in 2080 (<10% of the 140 species). B) Uncertainty of the model predictions: Standard deviation of the difference between full dispersal and “realistic” dispersal over the results for the nine environmental models. Projection: Europe Albers Equal Area Conic.