Exploring the post-harvest ‘window of vulnerability’ to landslides in New Zealand steepland plantation forests

Abstract

Nature-based solutions (NbS) are actions to address societal problems. They are similar to soil and water bioengineering (SWB), where trees and forests are used to mitigate natural hazards. However, NbS can have unintended consequences.Forest-based NbS may involve the enhancement, rehabilitation or restoration of natural forests or planting of trees and forests to provide a range of services including the production of timber and wood. In New Zealand, planted steepland forests have been widely used as NbS for erosion control. While intact, these forests provide various beneficial ecosystem services. However, if these forests are harvested, there is a period of up to 6–‍8 years following clear-fell harvesting, known as the ‘window of vulnerability' (WoV) when the landscape is susceptible to rainfall-induced landslides. During this time, the combination of declining root strength and changes in soil hydrology can lead to shallow landslides, especially during heavy storms.This study focuses on three questions: determining whether there is a time within the WoV when susceptibility to rainfall-induced landslides, expressed as landslide density, reaches a maximum; are such landslides related to forestry infrastructure; and are those landslides connected to the stream network. We examined three areas in New Zealand (Tolaga Bay, Marlborough, and Tasman) where exceptional rain events triggered thousands of landslides on forest land harvested in the years immediately preceding those events. Using a range of high-resolution satellite imagery, we manually mapped rainfall-induced landslides and identified those due to the rain events.The maximum landslide number and density occurred on land harvested 1–4 years (and on average 2–3 years) before the event and varied slightly for each study area. Landslides also occurred in areas with trees up to harvest age of about 30 years and on areas with different vegetation covers, i.e., mature indigenous forests, pasture, scrub, etc. There were fewer landslides associated with forest infrastructure such as roads and landings than triggered on clearcut slopes. On average across the three study regions, about half the landslides were connected to streamlines, and so were able to deliver sediment and woody debris.Better information on susceptibility to rainfall-induced landslides following forest removal may help forest managers and regulators understand this hazard and what can (and cannot) be done to mitigate events which often result in ‘disastrous’ off-forest impacts as observed in New Zealand in recent years.