Bayesian Inference of Ecological Interactions from Spatial Data

"Does a Respiratory Virus Have an Ecological Niche, and If So, Can It Be Mapped?" Yes and Yes

Although the utility of Ecological Niche Models (ENM) and Species Distribution Models (SDM) has been demonstrated in many ecological applications, their suitability for modelling epidemics or pandemics, such as SARS-Cov-2, has been questioned. In this paper, contrary to this viewpoint, we show that ENMs and SDMs can be created that can describe the evolution of pandemics, both in space and time. As an illustrative use case, we create models for predicting confirmed cases of COVID-19, viewed as our target "species", in Mexico through 2020 and 2021, showing that the models are predictive in both space and time. In order to achieve this, we extend a recently developed Bayesian framework for niche modelling, to include: (i) dynamic, non-equilibrium "species" distributions; (ii) a wider set of habitat variables, including behavioural, socio-economic and socio-demographic variables, as well as standard climatic variables; (iii) distinct models and associated niches for different species characteristics, showing how the niche, as deduced through presence-absence data, can differ from that deduced from abundance data. We show that the niche associated with those places with the highest abundance of cases has been highly conserved throughout the pandemic, while the inferred niche associated with presence of cases has been changing. Finally, we show how causal chains can be inferred and confounding identified by showing that behavioural and social factors are much more predictive than climate and that, further, the latter is confounded by the former.

Disentangling the contributions of biotic and abiotic predictors in the niche and the species distribution model of Trypanosoma cruzi, etiological agent of Chagas disease

The potential benefits of incorporating biotic, as well as abiotic, predictors in niche and species distribution models (SDMs), as well as how to achieve this, is still debated, with their interpretability and explanatory potential being particularly questioned. It is therefore important to stress test modelling methodologies that include biotic factors against use cases where there is ample knowledge of the potential biotic component of the niche. Relatively well studied and important vector-borne diseases offer just such an opportunity, where knowledge of the agents involved in the transmission cycle -vectors and hosts- can serve to calibrate and test the niche model and corresponding SDM. Here, we study the contributions of biotic -14 vectors, 459 potential hosts- and abiotic -258 climatic categories- predictors to the explanatory and predictive features of the niche and corresponding SDM for the etiological agent of Chagas disease, Trypanosoma cruzi, in Mexico. Using an established spatial data mining technique, we generate biotic, abiotic and biotic+abiotic niche and SDM models. We test our models by comparing predictions of the most important probable hosts of Chagas disease with a previously published list of confirmed hosts. We quantify, compare, and contrast the individual and total contributions of predictors to the niche and distribution of Chagas disease in Mexico. We assess the relative predictive potential of these variables to model performance, showing that models that include relevant biotic niche variables lead to more predictive, more ecologically realistic SDMs. Our research illustrates a useful general procedure for identifying and ranking potential biotic interactions and for assessing the relative importance of biotic and abiotic predictors. We conclude that the inclusion of both abiotic and biotic predictors in SDMs not only provides more predictive and accurate models but also models that are more understandable and explainable from an ecological niche perspective.

Modeling Mosquitoes and their Potential Odonate Predators Under Different Land Uses

To efficiently face the accelerated landscape transformation and its consequences in restructuring biotic communities and ecosystem services, one first question is which regional systems deserve prioritization for empirical assessments and interventive strategies. For the particular case of vector-borne disease control, we should consider generalist predators exhibiting differential responses to land-use change, as is the case of odonate insects. Thus, our aim was to infer land uses in Mexico where odonates (i.e., damselflies and dragonflies) might have some potential to predate mosquitoes of medical relevance. The study area included the hydrological basins of central Mexico. We modelled 167 species of odonates, four species of mosquitoes, and 51 land-use categories. Inferring spatial co-occurrence patterns from data mining and complex networks, we identified: (1) the ecological network of odonates and mosquitoes and (2) the land uses shared by these two groups. We inferred that 34% of odonate species co-occur with mosquitoes of medical relevance mainly in some preserved-mountain mesophyll cloud forest, high evergreen rainforest, and low tropical dry forest-but also in highly modified-human settlements, irrigation-based and pastures crop fields-land uses with strong human presence. Our findings highlight the relevance of community-regional studies for understanding the public health consequences of landscape change.

Using Data Mining and Network Analysis to Infer Arboviral Dynamics: The Case of Mosquito-Borne Flaviviruses Reported in Mexico

Given the significant impact of mosquito-borne flaviviruses (MBFVs) on both human and animal health, predicting their dynamics and understanding their transmission cycle is of the utmost importance. Usually, predictions about the distribution of priority pathogens, such as Dengue, Yellow fever, West Nile Virus and St. Louis encephalitis, relate abiotic elements to simple biotic components, such as a single causal agent. Furthermore, focusing on single pathogens neglects the possibility of interactions and the existence of common elements in the transmission cycles of multiple pathogens. A necessary, but not sufficient, condition that a mosquito be a vector of a MBFV is that it co-occurs with hosts of the pathogen. We therefore use a recently developed modeling framework, based on co-occurrence data, to infer potential biotic interactions between those mosquito and mammal species which have previously been identified as vectors or confirmed positives of at least one of the considered MBFVs. We thus create models for predicting the relative importance of mosquito species as potential vectors for each pathogen, and also for all pathogens together, using the known vectors to validate the models. We infer that various mosquito species are likely to be significant vectors, even though they have not currently been identified as such, and are likely to harbor multiple pathogens, again validating the predictions with known results. Besides the above "niche-based" viewpoint we also consider an assemblage-based analysis, wherein we use a community-identification algorithm to identify those mosquito and/or mammal species that form assemblages by dint of their significant degree of co-occurrence. The most cohesive assemblage includes important primary vectors, such as A. aegypti, A. albopictus, C. quinquefasciatus, C. pipiens and mammals with abundant populations that are well-adapted to human environments, such as the white-tailed deer (Odocoileus virginianus), peccary (Tayassu pecari), opossum (Didelphis marsupialis) and bats (Artibeus lituratus and Sturnira lilium). Our results suggest that this assemblage has an important role in the transmission dynamics of this viral group viewed as a complex multi-pathogen-vector-host system. By including biotic risk factors our approach also modifies the geographical risk profiles of the spatial distribution of MBFVs in Mexico relative to a consideration of only abiotic niche variables.

SPECIES: A platform for the exploration of ecological data

The modeling of ecological data that include both abiotic and biotic factors is fundamental to our understanding of ecosystems. Repositories of biodiversity data, such as GBIF, iDigBio, Atlas of Living Australia, and SNIB (Mexico's National System of Biodiversity Information), contain a great deal of information that can lead to knowledge discovery about ecosystems. However, there is a lack of tools with which to efficiently extract such knowledge. In this paper, we present SPECIES, an open, web-based platform designed to extract implicit information contained in large scale sets of ecological data. SPECIES is based on a tested methodology, wherein the correlations of variables of arbitrary type and spatial resolution, both biotic and abiotic, discrete and continuous, may be explored from both niche and network perspectives. In distinction to other modeling systems, SPECIES is a full stack exploratory tool that integrates the three basic components: data (which is incrementally growing), a statistical modeling and analysis engine, and an interactive visualization front end. Combined, these components provide a powerful tool that may guide ecologists toward new insights. SPECIES is optimized to support fast hypothesis prototyping and testing, analyzing thousands of biotic and abiotic variables, and presenting descriptive results to the user at different levels of detail. SPECIES is an open-access platform available online (http://species.conabio.gob.mx), that is, powerful, flexible, and easy to use. It allows for the exploration and incorporation of ecological data and its subsequent integration into predictive models for both potential ecological niche and geographic distribution. It also provides an ecosystemic, network-based analysis that may guide the researcher in identifying relations between different biota, such as the relation between disease vectors and potential disease hosts.

Species co-occurrence networks: Can they reveal trophic and non-trophic interactions in ecological communities?

Co-occurrence methods are increasingly utilized in ecology to infer networks of species interactions where detailed knowledge based on empirical studies is difficult to obtain. Their use is particularly common, but not restricted to, microbial networks constructed from metagenomic analyses. In this study, we test the efficacy of this procedure by comparing an inferred network constructed using spatially intensive co-occurrence data from the rocky intertidal zone in central Chile to a well-resolved, empirically based, species interaction network from the same region. We evaluated the overlap in the information provided by each network and the extent to which there is a bias for co-occurrence data to better detect known trophic or non-trophic, positive or negative interactions. We found a poor correspondence between the co-occurrence network and the known species interactions with overall sensitivity (probability of true link detection) equal to 0.469, and specificity (true non-interaction) equal to 0.527. The ability to detect interactions varied with interaction type. Positive non-trophic interactions such as commensalism and facilitation were detected at the highest rates. These results demonstrate that co-occurrence networks do not represent classical ecological networks in which interactions are defined by direct observations or experimental manipulations. Co-occurrence networks provide information about the joint spatial effects of environmental conditions, recruitment, and, to some extent, biotic interactions, and among the latter, they tend to better detect niche-expanding positive non-trophic interactions. Detection of links (sensitivity or specificity) was not higher for well-known intertidal keystone species than for the rest of consumers in the community. Thus, as observed in previous empirical and theoretical studies, patterns of interactions in co-occurrence networks must be interpreted with caution, especially when extending interaction-based ecological theory to interpret network variability and stability. Co-occurrence networks may be particularly valuable for analysis of community dynamics that blends interactions and environment, rather than pairwise interactions alone.