Climate change and management of forest biodiversity: predicting the impacts of climate matching strategies

Introducing warm-adapted tree provenances (climate matching) has been suggested as an adaptive strategic response of forestry to future climate change. This project set out to measure impacts of introduced provenances of sessile oak, Quercus petraea, on the abundance patterns of herbivorous insects and, in case of gall-inducers, also their parasitoid natural enemies. To do so we utilized one of the largest existing experimental provenance trials, managed by INRA at Petite Charnie in France.

Climate change predictions suggest that native UK forest trees may struggle to cope with future climates. One response to this problem is to plant in the UK trees from regions whose climates match predicted futures (climate matching). However, such introductions may have implications for native wildlife (such as herbivorous insects and their predators) dependent on these trees. We set out to find to what extend introducing such climate matched trees would impact native UK wildlife. Our results so far suggest that introductions from central france (which match moderate global warming predictions) are unlikely to have major impacts. However introductions from further south may be significantly out of step, in terms of the biological timing of events, with native wildlife - with negative effects likely

First, we established that tree phenotypic traits including budburst, leaf retention times, and forestry-relevant growth measures are indeed locally adapted, and differ between trees from different origins more than expected simply by genetic drift. Phenology and growth form selection regimes are well predicted by climates at the provenance origin, as used in climate matching models. However, growth patterns are better predicted by models including geographic gradients in both latitude and longitude. Having developed the statistical methods, we are now collaborating with colleagues at INRA on a larger scale analysis including multiple INRA trials with the same provenances, so allowing quantification of any gene by environment interactions in the tree phenotypes.
We amassed a unique dataset on herbivore abundance data across 20 provenances chosen to span observed diversity in phenotypic parameters. In surveys we searched 62,400 shoots for herbivores resulting in 133,590 records. Most comprise 3 feeding guilds: 60,820 gallers, 16,738 leaf-mining moths and beetles, and 16,738 external feeding caterpillars. Our choice of resolution of insect taxa has been guided by DNA barcoding of 4500 specimens of leaf miners and external feeders, covering >80% of the observed morpho-species. This approach has both linked previously unmatched lifecycle stages, and corrected errors morphology-based identification in multiple guilds.
Publications to date primarily concern methods and the initial molecular explorations of insect diversity. While draft versions exist for the publications listed above as in prep, the statistical analyses will continue for a short while. Multivariate analyses suggest that tree provenance is a strong predictor of the community structure of associated herbivores. Even models incorporating only variables used in climate matching reveal strong correlations, consistent with the idea that herbivores respond to the traits shown in our prior analyses to be influenced by local adaptation of oak provenances. Ongoing analyses will establish how individual insect species respond to tree phenotypic differences. So far, phenology variables have consistently influenced herbivore abundance although, intriguingly, not always in a consistent direction. This suggests that predictions for significant species, such as pests or keystone drivers of biodiversity, will require targeted studies, using the statistical tools developed here. Further, simple geographic distance is not a good predictor for phenotypic differences between tree provenances or for impacts on herbivore communities. Differences in tree selection regime perform better but have to be considerable to cause large effects on associated herbivores. This means that if provenances for climate matching were sourced from sites 2° south of target introduction sites (as recommended in IPCC low emission scenario for climate matching by Broadmeadow et al. 2005), then they would be unlikely to have a major impact on native herbivore communities. In contrast, introductions from 10° south of introduction sites (as recommended for IPCC high emission scenarios) would probably have a major impact. We communicated this conclusion to Mark Broadmeadow, the author of the climate matching strategy and now main advisor for climate change in the Forestry Commission, and others at our second project workshop.