Models to predict the distribution and abundance of breeding ducks in Canada

Abstract

ABSTRACT. Detailed knowledge of waterfowl abundance and distribution across Canada is lacking, which limits our ability to effectively conserve and manage their populations. We used 15 years of data from an aerial transect survey to model the abundance of 17 species or species groups of ducks within southern and boreal Canada. We included 78 climatic, hydrological, and landscape variables in Boosted Regression Tree models, allowing flexible response curves and multiway interactions among variables. We assessed predictive performance of the models using four metrics and calculated uncertainty as the coefficient of variation of predictions across 20 replicate models. Maps of predicted relative abundance were generated from resulting models, and they largely match spatial patterns evident in the transect data. We observed two main distribution patterns: a concentrated prairie-parkland distribution and a more dispersed pan-Canadian distribution. These patterns were congruent with the relative importance of predictor variables and model evaluation statistics among the two groups of distributions. Most species had a hydrological variable as the most important predictor, although the specific hydrological variable differed somewhat among species. In some cases, important variables had clear ecological interpretations, but in some instances, e.g., topographic roughness, they may simply reflect chance correlations between species distributions and environmental variables identified by the model-building process. Given the performance of our models, we suggest that the resulting prediction maps can be used in future research and to guide conservation activities, particularly within the bounds of the survey area. Des modèles pour prédire la répartition et l'abondance des canards nichant au CanadaRÉSUMÉ. Le manque de connaissances détaillées sur l'abondance et la répartition des espèces de sauvagine dans l'ensemble du Canada limite notre capacité à conserver et à gérer leurs populations de façon efficace. Au moyen de 15 années de données provenant d'inventaires aériens réalisés par transects, nous avons modélisé l'abondance de 17 espèces ou groupes d'espèces de canards dans les parties méridionale et boréale du Canada. Nous avons inclus 78 variables relatives au climat, à l'hydrologie et au paysage dans des modèles amplifiés d'arbres de régression, pour lesquels des courbes de réponse flexibles et les interactions multidimensionnelles entre les variables ont pu être examinées. Nous avons évalué la capacité de prédiction des modèles à l'aide de quatre paramètres et avons estimé l'incertitude en calculant le coefficient de variation des prédictions d'un jeu de 20 répétitions de modèles. Des cartes de l'abondance relative prédite ont été produites à partir des modèles et elles correspondent bien au profil spatial des données récoltées dans les transects. Nous avons observé deux tendances principales dans la répartition : l'une centrée sur les Prairies et les forêts-parcs, l'autre plus dispersée à la grandeur du Canada. Ces ten...

Modeling distribution and abundance of multiple species: Different pooling strategies produce similar results

“…We might have expected greater differences between modeling strategies had we created finely tuned species-specific models with reduced predictor sets. Because BRTs show higher variability in extrapolated predictions compared to interpolated predictions (Kusiak et al 2010;Barker et al 2014), we speculate that differences between modeling strategies might be greater when predicting outside the study area or outside species ranges. The particular result would depend on individual species' ranges and data quality, the environmental conditions within and outside the study area.…”

“…We therefore reduced the variable set by calculating seasonal means for precipitation and for minimum and maximum temperature. We did not execute further dimension reduction techniques because we wished to maintain ecological interpretability of each variable, which we discuss elsewhere (Barker et al 2014) To limit the source of variation among model results to modeling strategy, we used the same set of predictor variables for all models instead of finely tuning models with speciesspecific variables.…”

La abundancia de aves acuáticas (anseriformes) en relación con la complejidad del paisaje en un sitio RAMSAR del norte de México

“…Ante la falta de datos histó-ricos para estas variables, se recomienda que los trabajos de monitoreo colecten esta información periódicamente. Aunque el uso de imágenes de satélites podría mejorar para análisis espaciales de gran escala (Barker et al, 2014a), en nuestro caso fueron limitados por la falta de correspondencia temporal en la fenología de los cultivos con la presencia de las aves.…”

Quantifying avian relative abundance and ecosystem service value to identify conservation opportunities in the Midwestern U.S.

“…We modeled avian mean relative abundance as the performanceweighted average of 20 BRT models generated for each species (as in Barker et al 2014). We assessed the performance of each model by cross-validation using spatially stratified subsampling.…”

“…We generated 20 training and test datasets for each modeled species in order to accurately assess model performance and to generate ensemble estimates of abundance robust to the data partitioning process (Barker et al 2014). Training and test datasets were generated using a masked geographically structured (Radosavljevic and Anderson 2014) sub-setting routine that randomly assigns points to training and test datasets at a ratio of 5:1 based on a 5-km grid overlaid on the study area.…”