Species distribution models applied to plant species selection in forest restoration: are model predictions comparable to expert opinion?
2014
Gastón González, Aitor | García Viñas, Juan Ignacio | Bravo Fernandez, Jose Alfredo | López Leiva, César | Oliet Pala, Juan Antonio | Roig Gómez, Sonia | Serrada, Rafael
An expert on local flora usually is the best option for plant species selection in most ecological restoration projects; although species selection often needs to be dealt with swiftly as well as on a limited budget, and obtaining the opinion of a local expert may not always be an economically viable alternative. In such cases, species distribution models (SDM) may offer a faster and more cost effective alternative. We asked six experts to rank native tree species according to their suitability at 24 forest sites. The predictive performance of the suitability rankings was evaluated by assessing their ability to discriminate present from absent species in the observed tree assemblages at each evaluation site. We used the area under the receiver operating characteristic curve to calculate the probability that the estimated suitability for a species present at a particular evaluation site is greater than the estimate for an absent species (both picked at random). Suitability rankings were also obtained from the predictions of SDM and the same procedure was used to estimate the predictive performance of the set of models at each site. The experts offered concordant suitability rankings at almost every evaluation site. There were no significant differences in the predictive performance of the SDM and four of the experts, although the SDM performed slightly better than the other two experts. Our results point to the suitability of the proposed species distribution modeling approach to obtain fast and cost effective recommendations for species selection in forest restoration projects.
Показать больше [+] Меньше [-]Ключевые слова АГРОВОК
Библиографическая информация
Эту запись предоставил Universidad Politécnica de Madrid
