Modelos de distribución de especies aplicados a la gestión de especies exóticas invasoras y a la conservación de especies amenazadas

  1. Sara Santamarina García
Supervised by:
  1. Carmen Acedo Director
  2. Estrella Alfaro Sáiz Director
  3. María Reyes Tárrega García-Mares Tutor

Defence university: Universidad de León

Year of defence: 2023

  1. Juan Carlos Moreno Saiz Chair
  2. Victoria Ferrero Vaquero Secretary
  3. Gianluigi Bacchetta Committee member

Type: Thesis


In the Anthropocene, exploring solutions for the conservation of plants, the support of life on earth, in the face of global risks such as biological invasion, must be a priority. The growing demand for new modeling approaches for application in biodiversity management, conservation and risk assessment has increased in recent years. The science of modeling must advance in this direction, providing recommendations that can respond to different cases, based on real experience. The aim of this Thesis is to resolve important questions about this issue, using complementary approaches that can be extrapolated to other equivalent situations. As model species we studied Paraserianthes lophantha (Fabaceae), an invasive alien species, a close relative of the Australian acacias, Acacia spp. and Quercus lusitanica (Fagaceae), a threatened species of conservation concern. Specifically, we assessed the risk of the invasive potential of P. lophantha on Q. lusitanica holistically, using species distribution models as tools, complemented with experimental studies of the response of the seedlings of both species to drought, a critical stage in the plants development. The models were fitted with localized occurrence records from reliable confirmed citations of the species included in reference plant databases. In addition, all the records from Galicia were consulted in their original source, and located in the field for them to be more accurately georeferenced. The environmental variables come from databases related to climate (WORLDCLIM, Climatic Atlas of the Iberian Peninsula and CHELSA), land use (SIOSE), topography, pedology or hydrography. In some cases they were obtained using geographic information systems (ArcGis and Qgis). We explored (1) different species distribution modeling approaches (SDMs) applicable to small-area management, analyzing the predictive potential of models calibrated at different scales to direct management efforts in very small areas; (2) the potential of exotic species SDMs, calibrated with data from invaded ranges, comparing the predictive potential of SDMs of P. lophantha calibrated with native versus invaded range data - comparing in turn the predictions of models calibrated at regional versus local scales; (3) we introduce into the modeling variables related to the possible pathways of introduction and dispersal of exotic species - land cover data; (4) we evaluate the possible competitive interaction under drought conditions of P. lophantha versus Quercus lusitanica; (5) we study the niche dynamics of P. lophantha among various invaded ranges, for the choice of calibration areas for predictive models as well as for a better understanding of the species’ tolerance; (6) we evaluated whether the multisite study of niche dynamics between the native range and more than one invasion area can contribute to anticipate the expansion potential of invasive alien species; (7) we evaluate whether the results of a small-scale niche overlap analysis can be used for local conservation purposes; (8) we project the distributions of P. lophantha and Q. lusitanica in future climate scenarios; (9) we assessed the potential threat of P. lophantha on Q. lusitanica, obtaining risk maps, combining different SDMs generated for both; (10) we implemented the results in the risk analysis of P. lophantha at the European level. The most accurate species distribution models (SDMs), calibrated with suboptimal data, but with significant fine-grained variables (25m), which explain the distribution in those localities (e.g. land use, orientation...), and the selection of optimal model complexity, using a qualitative evaluation, allowed a more adequate spatial discrimination for Paraserianthes lophantha and Quercus lusitanica than the SDMs. The selection of the optimal model complexity, using a qualitative assessment, allowed a more adequate spatial discrimination for P. lophantha and Q. lusitanica than SDMs calibrated with full range data, calibrated with coarse-grained (1km) climatic variables, which reinforces the usefulness of combining local and regional approaches to promote local management actions. Furthermore, SDMs of P. lophantha, calibrated in its invaded range, were more useful than SDMs calibrated in its native range, demonstrating that their use to support management should be applied at least for exploratory purposes. This allows incorporating significant variables for the distribution of exotic species at small scales-for example, land uses-or predicting the distribution of exotic species suspected to have undergone niche changes. Likewise, we find that regionally calibrated SDMs, using high-resolution, locally focused variables, are suitable for improving predictions in local areas by including information on a larger number of occurrence data and reducing overfitting to local data. In addition, they provide a sharp discrimination of areas where conservation actions should be focused at a regional scale. All of these approaches overcame the limitations of assessing and managing the local risk of a biological invasion on a disjunct population. Our experimental study did not identify that either species would competitively outcompete the other under these conditions and stage of development. The effect of water stress was greater on Q. lusitanica, which experienced a large reduction in water potential and mortality when subjected to the drought treatment. Exploring the niche dynamics of P. lophantha through multi-site comparisons, considering different areas of invasion, improved the understanding of the invasion, avoided underestimating the potential for large- scale expansion of P. lophantha in Europe, and proved that using data from both native and invaded ranges is more adequate to predict the expansion potential of the species. Also, we studied the overlap of its niche with that of the small-scale Q. lusitanica population in Monte Pindo (A Coruña, Galicia, Spain), demonstrating that P. lophantha is a successful colonizer that can seriously threaten this natural area. This analysis also reinforced the results of previous predictive studies indicating that local management to prevent the expansion of P. lophantha should be a concern for the conservation of Q. lusitanica. Ensemble prediction models of the distribution of the two species on a large scale (Europe) corroborated that Paraserianthes lophantha has a remarkable potential for expansion over European coastal areas with a mild climate. However, the expansion in the climates of the Atlantic façade is expected to be more conflicting than on the coasts of the Mediterranean Sea. Regarding Quercus lusitanica, we anticipate that the effects of climate change will be much more important and could seriously compromise the conservation of the species. The combination of the SDMs calibrated with the different approaches in this thesis (regional and local scale) in risk models, corroborated the hypothesis that P. lophantha could pose a threat to the conservation of Q. lusitanica by identifying areas with a high risk of invasion, in which contact between both species could occur. Finally, the risk analysis was performed following the European analysis for Paraserianthes lophantha, using the results from the previous chapters. The species distribution modeling work provided new information to answer the questions of the model, providing scientific evidence regarding the species’ establishment and expansion capacity and impact the European territory, also incorporating aspects of climate change.