Use this URL to cite or link to this record in EThOS:
Title: Predicting species' range shifts under global change : when can species distribution models be useful?
Author: Rapacciuolo, Giovanni
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
Predicting how species’ distributions will change in response to environmental changes is fundamental for many aspects of agriculture, ecosystem service provision, human and animal health, and wildlife conservation. Correlative species distribution models (SDMs) are the primary tool for making such predictions; however, assessing their predictive accuracy is notoriously difficult, since predicted events are yet to occur. In this thesis, I tested the temporal transferability of widely-used SDMs based on coarse climate and land cover variables by validating these against records of recently-observed distribution changes in British vascular plants, birds, and butterflies. When transferred to a new time period, SDMs were generally accurate at discriminating between presence and absence across large portions of species’ ranges that had remained unchanged through time; however, their discrimination ability over portions of species ranges which had been observed to change occupancy status in time was no better than random. When considering the probabilistic nature of predictions, this lack of discrimination ability over dynamic portions of species’ ranges translated into significant deviations between values of predicted and observed probabilities of species’ gain and loss at given sites. Nonetheless, species’ gains and losses were more likely to be observed at sites with higher predicted probabilities of gain and loss, respectively. In addition, there was considerable variation in the temporal transferability of models with some models being more transferable than others. Differences amongst species were a major determinant of this variation. In particular, models of species found across a distinct and narrow range of environments, which also occurred across most geographical sites satisfying these requirements, were more transferable over time than others. Other factors leading to significant variation in temporal transferability were the choice of modelling framework and the geographical area over which a model was projected. My findings suggest that SDMs are unlikely to generate projections of single species’ range shifts that are accurate enough to provide a solid basis for local-scale conservation and management. However, they may be useful in pinpointing species and areas that are particularly vulnerable to environmental change and should thus be monitored closely. Clearly, more processes will need to be explicitly accounted for in our predictive models if we want to significantly increase their biological realism in the face of environmental change.
Supervisor: Purvis, Andy Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available