Data and code from: Indigenous and European-American land-use legacies in forest composition, Fort Drum, northern New York State: What do species distribution models detect across a local extent?

This study aimed to evaluate how past land use shapes forest compositional
patterns using fine-resolution, local-extent species distribution models
(SDMs), examining whether legacies of Indigenous and European-American
land-use regimes interact to influence past and present forest composition
and whether proxies of past land use improve SDMs. The study was conducted
at Fort Drum, northern New York State, US (434 km²), focusing on 27 tree
taxa across 18 genera, including Acer, Amelanchier, Betula, Carpinus,
Carya, Fagus, Fraxinus, Juglans, Ostrya, Pinus, Populus, Prunus, Quercus,
Salix, Thuja, Tilia, Tsuga, and Ulmus. Species distribution models were
developed for tree taxa circa 2000 using a forest inventory of 10,043
plots, and analyses incorporated original land survey records circa 1800
alongside archaeological data. Results indicated that “distance to nearest
circa 1950 forest” was the most important predictor of circa 2000 tree
taxon distributions, surpassing soil conditions such as moisture and pH,
and its inclusion increased SDM predictive performance (mean increase in
AUC = 0.025). SDMs did not strongly indicate widespread Indigenous
land-use legacies in taxon distributions but revealed some relationships
between archaeological sensitivity and oak and pine distributions circa
1800 and 2000. Overall, measures of past land use or land cover improved
the predictive power of fine-resolution SDMs, suggesting that interacting
legacies of differing land-use regimes may influence current species
distributions.