TY - JOUR
T1 - Evaluating ecological niche model accuracy in predicting biotic invasions using South Florida's exotic lizard community
AU - Mothes, Caitlin C.
AU - Stroud, James T.
AU - Clements, Stephanie L.
AU - Searcy, Christopher A.
N1 - Funding Information:
We thank Michelle Afkhami, Al Uy and Don DeAngelis for their helpful feedback during the development of this manuscript. We also thank Jessica Cothern and Shantel Catania for assistance in collecting field data, and Amber Wright for aiding with the R code. This work was conducted under Florida Fish and Wildlife permit LSSC-16-00013 and IACUC protocol 17-061. Funding for this project was provided by the University of Miami.
Funding Information:
We thank Michelle Afkhami, Al Uy and Don DeAngelis for their helpful feedback during the development of this manuscript. We also thank Jessica Cothern and Shantel Catania for assistance in collecting field data, and Amber Wright for aiding with the R code. This work was conducted under Florida Fish and Wildlife permit LSSC‐ 16‐00013 and IACUC protocol 17‐061. Funding for this project was provided by the University of Miami.
Publisher Copyright:
© 2019 John Wiley & Sons Ltd
PY - 2019/2/1
Y1 - 2019/2/1
N2 - Aim: Predicting environmentally suitable areas for non-native species is an important step in managing biotic invasions, and ecological niche models are commonly used to accomplish this task. Depending on these models to enact appropriate management plans assumes their accuracy, but most niche model studies do not provide validation for their model outputs. South Florida hosts the world's most globally diverse non-native lizard community, providing a unique opportunity to evaluate the predictive ability of niche models by comparing model predictions to observed patterns of distribution, abundance and physiology in established non-native populations. Location: Florida, USA. Taxon: Lizards. Methods: Using Maxent, we developed niche models for all 29 non-native lizard species with established populations in Miami-Dade County, Florida, using native range data to predict habitat suitability in the invaded range. We then used independently collected field data on abundance, geographical spread and thermal tolerances of the non-native populations to evaluate Maxent's ability to make predictions in both geographical and environmental space in the non-native range. Results: Maxent performed well in predicting across geographical space where these non-native lizards were most likely to occur, but within a given geographical extent was unable to predict which individual species would be the most abundant or widespread. Comparisons with physiological data also revealed an imperfect fit, but without any consistent biases. Main conclusions: We performed one of the most extensive field validations of Maxent's ability to predict where invasions are likely to occur, and our results support its continued use in this role. However, the program was unable to predict the relative abundance and geographical spread of established species, indicating limited utility for identifying which invasive species will be the greatest management concern. These results underscore the importance of other factors, such as time since introduction, dispersal ability and biotic interactions in determining the relative success of non-native species post-establishment.
AB - Aim: Predicting environmentally suitable areas for non-native species is an important step in managing biotic invasions, and ecological niche models are commonly used to accomplish this task. Depending on these models to enact appropriate management plans assumes their accuracy, but most niche model studies do not provide validation for their model outputs. South Florida hosts the world's most globally diverse non-native lizard community, providing a unique opportunity to evaluate the predictive ability of niche models by comparing model predictions to observed patterns of distribution, abundance and physiology in established non-native populations. Location: Florida, USA. Taxon: Lizards. Methods: Using Maxent, we developed niche models for all 29 non-native lizard species with established populations in Miami-Dade County, Florida, using native range data to predict habitat suitability in the invaded range. We then used independently collected field data on abundance, geographical spread and thermal tolerances of the non-native populations to evaluate Maxent's ability to make predictions in both geographical and environmental space in the non-native range. Results: Maxent performed well in predicting across geographical space where these non-native lizards were most likely to occur, but within a given geographical extent was unable to predict which individual species would be the most abundant or widespread. Comparisons with physiological data also revealed an imperfect fit, but without any consistent biases. Main conclusions: We performed one of the most extensive field validations of Maxent's ability to predict where invasions are likely to occur, and our results support its continued use in this role. However, the program was unable to predict the relative abundance and geographical spread of established species, indicating limited utility for identifying which invasive species will be the greatest management concern. These results underscore the importance of other factors, such as time since introduction, dispersal ability and biotic interactions in determining the relative success of non-native species post-establishment.
KW - Maxent
KW - biological invasions
KW - climate suitability
KW - ecological niche model
KW - lizards
KW - model validation
KW - non-native species
KW - thermal tolerance
UR - http://www.scopus.com/inward/record.url?scp=85060332868&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85060332868&partnerID=8YFLogxK
U2 - 10.1111/jbi.13511
DO - 10.1111/jbi.13511
M3 - Article
AN - SCOPUS:85060332868
VL - 46
SP - 432
EP - 441
JO - Journal of Biogeography
JF - Journal of Biogeography
SN - 0305-0270
IS - 2
ER -