Beyond distance-invariant survival in inverse recruitment modeling: A case study in Siberian Pinus sylvestris forests

Recruitment represents the net effect of seed dispersal and survival (germination and establishment) processes. Both are known to be a function of distance from the seed trees. Seed densities usually decline with distance from the seed trees and the survival function can have different shapes. It may e.g. increase with distance (Janzen–Connell type effects) or decrease in the case of facilitation. In recruitment modeling the dispersal kernel is often directly fitted to the number of saplings instead of to the number of seeds, by indirectly assuming a distance-invariant survival function. This assumption may be violated in many cases as the survival function may attain different shapes. The interaction of distance-dependent seed deposition and survival becomes more complex in the case of several seed trees, since processes shaping the survival function may act on seeds from different seed trees. We develop a recruitment model disentangling (1) seed dispersal as derived by a summed seed shadow model and (2) survival as a function of distance of several nearby seed trees. Using Bayesian model inversion we parameterize our model for post fire Pinus sylvestris regeneration, which is known to suffer from strong root competition in the vicinity of seed trees. We show that accounting for distance-dependent survival substantially improves the model performance for three tested seed dispersal kernels. We show that ignoring distance-dependency of survival leads to ecologically unrealistic parameter estimates, questionable seed dispersal properties and biased sapling density predictions at the landscape level. Disentangling seed dispersal and survival processes is essential for modeling recruitment patterns in the case of distance-dependent survival. Our model helps to learn about the connection between seed and sapling patterns particularly when several seed trees are present. Our modeling approach derives separate seed dispersal kernels, survival functions, and the resulting recruitment-patterns, also when only sapling counts are available.

► Distance-dependent survival is often neglected in inverse recruitment modeling.
► We present an approach to disentangle distance-dependent dispersal and survival processes.
► Based on a case study in Central Siberia we demonstrate the superiority of our approach.
► We show that biased parameter values and model predictions result from assuming distance-invariant survival.