Mapping canopy nitrogen-scapes to assess foraging habitat for a vulnerable arboreal folivore in mixed-species Eucalyptus forests.
Benjamin WagnerPatrick J BakerBen D MooreCraig R NitschkePublished in: Ecology and evolution (2021)
Herbivore foraging decisions are closely related to plant nutritional quality. For arboreal folivores with specialized diets, such as the vulnerable greater glider ( Petauroides volans ), the abundance of suitable forage trees can influence habitat suitability and species occurrence. The ability to model and map foliar nitrogen would therefore enhance our understanding of folivore habitat use at finer scales. We tested whether high-resolution multispectral imagery, collected by a lightweight and low-cost commercial unoccupied aerial vehicle (UAV), could be used to predict total and digestible foliar nitrogen (N and digN) at the tree canopy level and forest stand-scale from leaf-scale chemistry measurements across a gradient of mixed-species Eucalyptus forests in southeastern Australia. We surveyed temperate Eucalyptus forests across an elevational and topographic gradient from sea level to high elevation (50-1200 m a.s.l.) for forest structure, leaf chemistry, and greater glider occurrence. Using measures of multispectral leaf reflectance and spectral indices, we estimated N and digN and mapped N and favorable feeding habitat using machine learning algorithms. Our surveys covered 17 Eucalyptus species ranging in foliar N from 0.63% to 1.92% dry matter (DM) and digN from 0.45% to 1.73% DM. Both multispectral leaf reflectance and spectral indices were strong predictors for N and digN in model cross-validation. At the tree level, 79% of variability between observed and predicted measures of nitrogen was explained. A spatial supervised classification model correctly identified 80% of canopy pixels associated with high N concentrations (≥1% DM). We developed a successful method for estimating foliar nitrogen of a range of temperate Eucalyptus species using UAV multispectral imagery at the tree canopy level and stand scale. The ability to spatially quantify feeding habitat using UAV imagery allows remote assessments of greater glider habitat at a scale relevant to support ground surveys, management, and conservation for the vulnerable greater glider across southeastern Australia.
Keyphrases
- climate change
- high resolution
- machine learning
- fluorescence imaging
- low cost
- risk assessment
- deep learning
- genetic diversity
- optical coherence tomography
- cross sectional
- type diabetes
- palliative care
- mass spectrometry
- metabolic syndrome
- weight loss
- magnetic resonance imaging
- insulin resistance
- glycemic control
- essential oil
- amino acid
- antibiotic resistance genes
- drug induced