Search for: All records

Ensemble-based change detection can improve map accuracies by combining information from multiple datasets. There is a growing literature investigating ensemble inputs and applications for forest disturbance detection and mapping. However, few studies have evaluated ensemble methods other than Random Forest classifiers, which rely on uninterpretable “black box” algorithms with hundreds of parameters. Additionally, most ensemble-based disturbance maps do not utilize independently and systematically collected field-based forest inventory measurements. Here, we compared three approaches for combining change detection results generated from multi-spectral Landsat time series with forest inventory measurements to map forest harvest events at an annual time step. We found that seven-parameter degenerate decision tree ensembles performed at least as well as 500-tree Random Forest ensembles trained and tested on the same LandTrendr segmentation results and both supervised decision tree methods consistently outperformed the top-performing voting approach (majority). Comparisons with an existing national forest disturbance dataset indicated notable improvements in accuracy that demonstrate the value of developing locally calibrated, process-specific disturbance datasets like the harvest event maps developed in this study. Furthermore, by using multi-date forest inventory measurements, we are able to establish a lower bound of 30% basal area removal on detectable harvests, providing biophysical context for our harvest event maps. Our results suggest that simple interpretable decision trees applied to multi-spectral temporal segmentation outputs can be as effective as more complex machine learning approaches for characterizing forest harvest events ranging from partial clearing to clear cuts, with important implications for locally accurate mapping of forest harvests and other types of disturbances. 

Abstract Understanding connections between ecosystem nitrogen (N) cycling and invasive insect defoliation could facilitate the prediction of disturbance impacts across a range of spatial scales. In this study we investigated relationships between ecosystem N cycling and tree defoliation during a recent 2015–18 irruption of invasive gypsy moth caterpillars (Lymantria dispar), which can cause tree stress and sometimes mortality following multiple years of defoliation. Nitrogen is a critical nutrient that limits the growth of caterpillars and plants in temperate forests. In this study, we assessed the associations among N concentrations, soil solution N availability and defoliation intensity by L. dispar at the scale of individual trees and forest plots. We measured leaf and soil N concentrations and soil solution inorganic N availability among individual red oak trees (Quercus rubra) in Amherst, MA and across a network of forest plots in Central Massachusetts. We combined these field data with estimated defoliation severity derived from Landsat imagery to assess relationships between plot-scale defoliation and ecosystem N cycling. We found that trees in soil with lower N concentrations experienced more herbivory than trees in soil with higher N concentrations. Additionally, forest plots with lower N soil were correlated with more severe L. dispar defoliation, which matched the tree-level relationship. The amount of inorganic N in soil solution was strongly positively correlated with defoliation intensity and the number of sequential years of defoliation. These results suggested that higher ecosystem N pools might promote the resistance of oak trees to L. dispar defoliation and that defoliation severity across multiple years is associated with a linear increase in soil solution inorganic N. 

Carbon starvation posits that defoliation‐ and drought‐induced mortality results from drawing down stored non‐structural carbohydrates (NSCs), but evidence is mixed, and few studies evaluate mortality directly. We tested the relationships among defoliation severity, NSC drawdown and tree mortality by measuring NSCs in mature oak trees defoliated by an invasive insect,Lymantria dispar, across a natural gradient of defoliation severity.

We collected stem and root samples from mature oaks (Quercus rubraandQ.alba) in interior forests (n = 34) and forest edges (n = 47) in central Massachusetts, USA. Total NSC (TNC; sugar + starch) stores were analysed with respect to tree size, species and defoliation severity, which ranged between 5% and 100%.

TNC stores declined significantly with increasingly severe defoliation. Forest edge trees had higher TNC stores that were less sensitive to defoliation than interior forest trees, although this may be a result of differing defoliation history. Furthermore, we observed a mortality threshold of 1.5% dry weight TNC.

Our study draws a direct link between insect defoliation and TNC reserves and defines a TNC threshold below which mortality is highly likely. These findings advance understanding and improve model parametrization of tree response to insect outbreaks, an increasing threat with globalization and climate change.

A freePlain Language Summarycan be found within the Supporting Information of this article.

 

Remote sensing imagery can provide critical information on the magnitude and extent of damage caused by forest pests and pathogens. However, monitoring short‐term changes in deciduous forest condition caused by defoliating insects is challenging and requires approaches that directly account for seasonal vegetation dynamics. We implemented a previously published harmonic modeling approach for forest condition monitoring in Google Earth Engine and systematically assessed the relative ability of condition change products generated using various model parameterizations for predicting pest abundances and defoliation during the 2016–2018 gypsy moth (Lymantria dispar) outbreak in southern New England. Our comparisons revealed that most models made reasonable predictions of changes in canopy condition and egg and larval abundances ofL. dispar, indicating a strong correlation between our harmonic‐based estimates of condition change and defoliator activity. The greatest differences in predictive ability were in the spectral domain, with assessments based on Tasseled Cap Greenness, Simple Ratio, and the Enhanced Vegetation Index ranking among the top models, and the commonly used Normalized Difference Vegetation Index consistently exhibiting poorer performance. We also observed notable differences in the magnitude of scores for different baseline periods. Additionally, we found that Landsat‐based condition scores better explained larval abundance than egg mass counts, which have historically been used as a proxy for later‐season larval abundance, indicating that our remote sensing approach may be more accurate and cost‐effective for generating consistent retrospective assessments ofL. disparpopulation abundance in addition to estimates of canopy damage. These findings provide important linkages between spectral changes detected using a harmonic modeling approach and biophysical aspects of defoliator activity, with potential to extend monitoring and prediction to regional or even continental scales.