Bioclimatic modeling of potential vegetation types as an alternative to species distribution models for projecting plant species shifts under changing climates

•Managers need more robust methods for predicting species presence and other ecological characteristics under climate change.•Bioclimatic modeling was used to predict current and future potential vegetation type classes.•Species cover was linked to potential vegetation type classes to create species...

Full description

Saved in:
Bibliographic Details
Published in:Forest ecology and management Vol. 477; p. 118498
Main Authors: Keane, Robert E., Holsinger, Lisa M., Loehman, Rachel
Format: Journal Article
Language:English
Published: Elsevier B.V 01.12.2020
Subjects:
algorithms
climate change
Climate offsets
Constancy
ecosystems
forest ecology
geographical distribution
Habitat types
land management
landscapes
Random forests
Statistical modeling
trees
WXFIRE
ECODATA
PIAL
BPS
DAYMET
ABLA
FSVeg
Climate offsets
PVT
NRCS
Habitat types
VASC
CanESM2
NEX-DCP30
BCMPVT
THPL
CLUN
PIPO
SDM
FIA
Constancy
SWCC
SDM-F
LANDFIRE
LIBO
LAOC
SDM-M
Random forests
CNRM-CM5
GCM
Statistical modeling
MEFE
RF
HadGEM2-ES
XETE
WXFIRE
PSME
ISSN:0378-1127, 1872-7042
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Managers need more robust methods for predicting species presence and other ecological characteristics under climate change.•Bioclimatic modeling was used to predict current and future potential vegetation type classes.•Species cover was linked to potential vegetation type classes to create species presence.•Accuracies were around 50% for PVTs but were often above 70% for species. Land managers need new tools for planning novel futures due to climate change. Species distribution modeling (SDM) has been used extensively to predict future distributions of species under different climates, but their map products are often too coarse for fine-scale operational use. In this study we developed a flexible, efficient, and robust method for mapping current and future distributions and abundances of vegetation species and communities at the fine spatial resolutions that are germane to land management. First, we mapped Potential Vegetation Types (PVTs) using conventional statistical modeling techniques (Random Forests) that used bioclimatic ecosystem process and climate variables as predictors. We obtained over 50% accuracy across 13 mapped PVTs for our study area. We then applied future climate projections as climate input to the Random Forest model to generate future PVT maps, and used field data describing the occurrence of tree and non-tree species in each PVT category to model and map species distribution for current and future climate. These maps were then compared to two previous SDM mapping efforts with over 80% agreement and equivalent accuracy. Because PVTs represent the biophysical potential of the landscape to support vegetation communities as opposed to the vegetation that currently exists, they can be readily linked to climate forecasts and correlated with other, climate-sensitive ecological processes significant in land management, such as fire regimes and site productivity.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0378-1127
1872-7042
DOI:10.1016/j.foreco.2020.118498