Molecular assessment of natural disturbance regime in a temperate swamp forest

Abstract

Advanced knowledge of forest disturbance regimes (the temporal and spatial patterns of periodic disturbances) has key relevance to implementation of effective managements for forest communities. For many forest ecosystems such knowledge is inadequate or is lacking. Spatially explicit simulations, testing effects of alternative disturbance regimes on fine-scale spatial genetic structure (FSGS) may provide information on key components of forest disturbance regimes. Here, we compare FSGSs assessed using 18 microsatellite markers with spatially explicit simulations, to characterise a natural disturbance regime in a European temperate swamp forest dominated by Alnus glutinosa. Individual-based population genetic simulations were carried out for three previously introduced models of disturbance regime, to determine which best fits the FSGS assessed by the kinship coefficient in a natural old-growth forest. We hypothesised a low proportion of trees removed (the gap dynamic model) accentuates high-relatedness among individuals over short distances (i.e. it accentuates FSGS), while a high proportion of trees eliminated by a disturbance (the cyclic succession model and the stand-replacing dynamic model) leads either to random FSGS (i.e. a point pattern of genotypes that cannot be differentiated from a complete spatial randomness) or a high spatial clustering of genotypes. Spatial genetic analyses found significant positive autocorrelations of genotypes at < 30 m in the natural old-growth forest, which indicates A. glutinosa individuals were more highly clustered than for a null model with complete spatial randomness. However, the strength of FSGS was low as indicated by a kinship coefficient < 0.02. Comparing the FSGSs of simulated disturbance regimes with the FSGS recorded in the old-growth forest, the gap dynamic model was the most probable scenario, with small canopy gaps occurring continually through time. The cyclic succession model and the stand-replacing dynamic model also led to non-random FSGSs but in both these models, the FSGS was stronger and more coarse-grained than that observed. The approach of comparing the FSGS recorded in a natural forest with simulations of disturbance regimes allows support to be found for (or against) previously hypothesised disturbance patterns. This approach should be considered alongside the well-established dendrochronological and palaeoecological methods.