Integrating variable traits and inheritance into the individual-based model LAVESI for evaluation of their importance for larch forest performance under future adverse conditions

Abstract:

With changing climate boreal forest are shifting further north and become threatened by droughts in the south. However, whether boreal forest species can adapt to novel situations and reduce their extinction risk is largely unknown but crucial to predict future performance of populations. Exploring variable traits and ultimately trait inheritance in an individual-based model could improve our understanding and help future projections.

We updated the vegetation model LAVESI with the possibility of variation in traits values that are normally distributed and the option of determining the trait values based on the parental values, thus allowing inheritance.

Using current climate data and future projections of climate we ran simulation experiments of Larix gmelinii stands in the two areas of interest, the northern tree line expanding due to increasing temperatures and the southern area threatened by drought. For Seed weight affecting migration further north and drought resistance protecting stands in the south a comparison of the model variants: uniform, variable and inherited traits is being performed. The results will be presented and will allow to disentangle how far migration rate and survival rate are influenced. In preliminary test it was already shown, that both the addition of variation and inheritance led to an increase in migration rate, with the latter being stronger. It is also expected that in changing temperatures leading to droughts the variation that allow for adaption would lead to better surviving populations.

We expect that variable traits ensure that if the environment changes necessary trait variants are available. Inheritance could let the populations adapt to environments and promote successful trait values and therefore lead to more optimised populations, that are able to spread faster and be as resilient as needed.

With this we show that implementing trait variation and inheritance may contribute to creating more accurately predicting models