Rot or not? Uncovering the spatial patterns and drivers of Norway spruce root rot with harvester data

Abstract

Root rot is a major problem for forestry, leading to reduced timber quality, growth losses, and increased disturbance risks. Harvester data provides a promising source of information for improving the knowledge on the root rot distribution. Here, we used harvester data (1) to map the risk of spruce root rot in southern and central Finland, and (2) to understand the drivers of the spatial patterns in rot occurrence. First, we built a statistical model predicting the percentage of stems affected by root rot on stand-level. To train the model, we used an extensive set of harvester data, containing 10,402 clear-cut forest stands, where the presence of root rot was recorded for each cut tree using an algorithm based on bucking patterns (i.e., cutting of the stem into different log assortments) recorded by the harvester. The model consisted of two parts, a fixed component describing the effects of different drivers of root rot, and a spatial random component describing the spatial patterns not explained by the fixed part of the model. The fixed part included forest and site attributes, landscape characteristics and proxies of forest-use legacies. The model was then used to map root rot risk, by predicting the probability of root rot occurrence using spatial data sets of the variables in the fixed part of the model, and the known rot status of locations in the data set for the random part of the model. Finally, the map was tested with an independent validation data, verifying its ability to identify the high-risk areas. Proxies of forest-use legacies, tree size and site fertility were found to drive the percentage of rot-affected stems in stands. The results quantify the root rot risk in Finland in higher detail than before and demonstrate the large potential of harvester data in informing about the risk of root rot in boreal forests.