The role of bioregionalization in improving our understanding of global biodiversity
With the recent boom in ‘big data’, macroecologists and biogeographers have focused on creating robust, reproducible bioregionalizations that either outperform or validate the most widely accepted biogeographic frameworks. Despite the great contribution made by these frameworks to our understanding of global biodiversity, a higher-order research goal is to define a system of small biogeographic units that nest within larger ones, under the hypothesis that this will better reflect biodiversity patterns across multiple taxa. Using a hierarchical biogeographic template that crosses from a regional- (bioregions within a continent, such as the Interim Biogeographic Regionalization for Australia; IBRA) to global-scale (ecoregions and biomes of the World Wildlife Fund; WWF), I aim in this thesis to investigate how we can improve the contribution of bioregionalizations to our understanding and conservation of biodiversity at large scales.
First (Chapter 2), I used ‘systematic mapping’—an evidence synthesis technique—to quantitatively summarize where, what, and how biodiversity has been investigated within the context of IBRA bioregions for almost 25 years. My synthesis showed that i) research effort is distributed unevenly among and within bioregions; ii) there are marked methodological preferences in bioregional studies (e.g., using inference over predictive and mechanistic approaches); and iii) there exist disparities in biodiversity foci, with species as the preferred level of biological organization, and mammals and birds as the most studied taxa. This synthesis collectively reveals that we have a limited understanding of the relationship between biodiversity and IBRA bioregions, even for well-studied taxa, and that incomplete knowledge of biodiversity within the IBRA bioregions’ context will be the norm for quite some time.
To ameliorate the confounding effect that inadequate information on species’ distributions might have in the definition of a hierarchical system of biogeographic units, I next (Chapter 3) developed a reproducible, data-driven alternative to expert-generated range maps for estimating the extent of occurrence (EOO) of hundreds—or even thousands—of species, by combining presence-only data and the finer-resolution of the IBRA framework (subregions) within a consistent, objective quantitative framework. For multiple example-sets of Australian terrestrial species, I evaluated alternative approaches, using visual checks of data-driven EOO maps and/or their comparison with expert-generated range maps in terms of spatial concordance, and species richness at subregional, bioregional, and ecoregional scales. I showed that this novel data-driven approach can reliably map EOOs of species whose distributions aligns with three broad types of geographic patterns (wide[1]range, habitat-specialists, and range-restricted) for vertebrate taxa and vascular plants.
Using these data-driven EOO maps as the primary source of information on species’ distributions, I then (Chapter 4) chose to leverage on the fact that the IBRA framework is a more detailed spatial classification of WWF’s global bioregionalization (ecoregions nested within biomes). Specifically, I examined whether a hierarchical system that is more directly relevant to biodiversity can be derived using information on species richness and composition within IBRA bioregions. Within an information-analytical framework, I demonstrated quantitatively that an algorithmic, data-driven model of diversity patterns within IBRA bioregions outperformed an expert-based collection of bioregions into WWF’s ecoregions and biomes. The hierarchical system of spatial partitioning of bioregions is more ecologically interpretable and better suited to inform biodiversity research, because of its superior depiction of multiple taxa’ distributional patterns.
With an incomplete knowledge of the study system, IBRA bioregions were defined with the goal of mapping and describing major ecosystems. These, by definition, include interacting species. As the distribution of mammals and their interactions are well documented in the Australian island state of Tasmania, in Chapter 5 I used some of its species to parameterize and validate a spatially explicit mechanistic model for a simple but ecologically sensible community of eight ‘virtual species’ that were suitable for in silico biogeographic experiments. Comparison between the virtual community and that from the real world (camera-trap sites with long-term monitoring), in terms of distributional and community patterns (alpha, and beta diversity), shows that it is possible to model a stable ecological network that also concurrently mimics multiple species’ actual distributional and community patterns. This was achieved by grounding the simulation of virtual realized niches in a food-web-oriented approach to species interactions.
Finally, in Chapter 6 I used the stable Tasmanian virtual mammal community derived in Chapter 5 to investigate the geographical and ecological features of bioregions that are most useful for predicting species’ distributions, and community patterns across such units. I iteratively left out one of the nine IBRA bioregions from Tasmania, to predict ‘out-of-sample’ the species’ probability of occurrence, using stacked and joint species distribution modeling (SSDM and JSDM, respectively). Contrasting the predicted bioregional species- and community-level patterns to the ‘actual’ ones in the virtual community shows that modeling distributions of bioregional communities by assuming that species’ occurrences are correlated (JSDM—for which a species association parameter is estimated) rather than by treating them as independent (SSDM) resulted in more accurate predictions of species’ realized distributions; and thus, a better representation of alpha and beta diversity. This implies that both biophysical covariates and ecological features—herein species interactions—found across bioregions, can be used successfully within a unifying modeling approach to make generalized predictions on community patterns.
In conclusion, my research for this thesis revealed that a systematic, quantitative-based examination of an existing bioregionalization within its spatial structure is indeed a useful approach to refining our understanding and conservation of biodiversity at large scales. I showed that an increasing availability of information on both species’ distributions and other facets of biodiversity (e.g., spatial structuring of communities), when coupled to high computational power, advances in aggregative and comparative techniques and grounded in ecological theory, make this data- and algorithm-based approach to bioregionalization assessment a possible and demonstrably successful tool for applied ecology. This is an important outcome, considering that in many cases, such biogeographic templates are the basis for biodiversity research, policy, and conservation.
History
Sub-type
- PhD Thesis