The Poisson-lognormal model as a versatile framework for the joint analysis of species abundances

Joint Species Abundance Models (JSDM) provide a general multivariate framework to study the joint abundances of all species from a community. JSDM account for both structuring factors (environmental characteristics or gradients, such as habitat type or nutrient availability) and potential interactions between the species (competition, mutualism, parasitism, etc.), which is instrumental in disentangling meaningful ecological interactions from mere statistical associations.Modeling the dependency between the species is challenging because of the count-valued nature of abundance data and most JSDM rely on Gaussian latent layer to encode the dependencies between species in a covariance matrix. The multivariate Poisson-lognormal (PLN) model is one such model, which can be viewed as a multivariate mixed Poisson regression model. The inference of such models raises both statistical and computational issues, many of which were solved in recent contributions using variational techniques and convex optimization.The PLN model turns out to be a versatile framework, within which a variety of analyses can be performed, including multivariate sample comparison, clustering of sites or samples, dimension reduction (ordination) for visualization purposes, or inference of interaction networks. This paper presents the general PLN framework and illustrates its use on a series a typical experimental datasets. All the models and methods are implemented in the R package PLNmodels, available from cran.r-project.org.

Towards combining occurrence and abundance distribution models of Great Bustard for conservation: A global research template from Bohai Bay?

Species distribution models (SDMs) have become important and essential tools in conservation and management. However, SDMs built with count data, commonly referred to as species abundance models (SAMs), are still less used so far. SDMs are increasingly used now in conservation decisions, whereas SAMs are still not widely employed. Species occurrence and abundance do not frequently display similar patterns, often they are not even well correlated. This leads to an insufficient or misleading conservation. How to combine information from SDMs and SAMs all together for unified conservation remains a challenge. In this study, we put forward for the first time a priority protection index (PI). The PI combines the prediction results of occurrence and abundance models. We used the best-available presence and count records for an endangered farmland species, Great Bustard (Otis tarda dybowskii) in Bohai Bay, China, as a case study. We then applied the advanced Random Forest algorithm (Salford Systems Ltd. implementation), a powerful machine learning method, with eleven predictor variables to forecast the spatial occurrence as well as the abundance distribution. The results show that the occurrence model had a decent performance (ROC: 0.77) and the abundance model had a RMSE 26.54. It is of note that environmental variables influenced bustard occurrence and abundance differently. We found that occurrence and abundance models display different spatial distribution patterns. Still, combining occurrence and abundance indices to produce a priority protection index (PI) used for conservation could guide the protection of the areas with high occurrence and high abundance (e.g. in Strategic Conservation Planning). Due to the widespread use of SDMs and the rel. easy subsequent employment of SAMs these findings have a wide relevance and applicability, worldwide. We promote and strongly encourage to further test, apply and update the priority protection index (PI) elsewhere in order to explore the generality of these findings and methods readily available now for researchers.

Towards combining occurrence and abundance distribution models of Great Bustard for conservation: A global research template from Bohai Bay?

Species distribution models (SDMs) have become important and essential tools in conservation and management. However, SDMs built with count data, commonly referred to as species abundance models (SAMs), are still less used so far. SDMs are increasingly used now in conservation decisions, whereas SAMs are still not widely employed. Species occurrence and abundance do not frequently display similar patterns, often they are not even well correlated. This leads to an insufficient or misleading conservation. How to combine information from SDMs and SAMs all together for unified conservation remains a challenge. In this study, we put forward for the first time a priority protection index (PI). The PI combines the prediction results of occurrence and abundance models. We used the best-available presence and count records for an endangered farmland species, Great Bustard (Otis tarda dybowskii) in Bohai Bay, China, as a case study. We then applied the advanced Random Forest algorithm (Salford Systems Ltd. implementation), a powerful machine learning method, with eleven predictor variables to forecast the spatial occurrence as well as the abundance distribution. The results show that the occurrence model had a decent performance (ROC: 0.77) and the abundance model had a RMSE 26.54. It is of note that environmental variables influenced bustard occurrence and abundance differently. We found that occurrence and abundance models display different spatial distribution patterns. Still, combining occurrence and abundance indices to produce a priority protection index (PI) used for conservation could guide the protection of the areas with high occurrence and high abundance (e.g. in Strategic Conservation Planning). Due to the widespread use of SDMs and the rel. easy subsequent employment of SAMs these findings have a wide relevance and applicability, worldwide. We promote and strongly encourage to further test, apply and update the priority protection index (PI) elsewhere in order to explore the generality of these findings and methods readily available now for researchers.