Quantile regression for estimating Douglas-fir natural regeneration potential using the R package quaxnat: Advanced ecological modeling for the management of nature conservation and silviculture

Abstract

Recent extreme weather conditions in Europe have led to widespread destruction of Norway spruce by storms and bark beetles, creating large clearings that need replanting. The shortage of planting material has shifted focus to natural regeneration processes, with Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco) emerging as a potential substitute due to its growth performance and drought tolerance. This study introduces and applies methods for investigating the regeneration ecology of Douglas-fir, focusing on the potential density of established regeneration and its dependence on the distance to the nearest seed source.

This dependence is modelled with various classical spatial dispersal kernels, the parameters of which are estimated with a quantile regression approach implemented in a new R package quaxnat. Regeneration data from 44,257 sample plots in the state forest of Lower Saxony, Germany, are combined with remote sensing-based positions of potential seed trees to illustrate these methods. Among the standard dispersal kernels provided by quaxnat, the spatial t distribution proves to be the most suitable. Here, for the .999th quantile, the estimated potential regeneration density reaches almost 11,000 trees per hectare in the immediate vicinity of the seed trees and decreases sharply with increasing distance.

A simple simulation model that takes dispersal and establishment into account illustrates how these results can be linked to management scenarios. The study provides valuable information for nature conservation and silviculture, suggesting buffer zones around sensitive habitats and guiding forest management decisions regarding natural regeneration options.