Ontogenetic habitat associations of a demersal fish species, Pagrus auratus, identified using boosted regression trees

Compton, Tanya J.; Morrison, Mark A.; Leathwick, John R.; Carbines, Glen

doi:10.3354/meps09790

Cited by 39 publications

(28 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…BRT models were assessed using cross‐validated measures of model performance (Compton, Morrison, Leathwick, & Carbines, ; Elith et al, ). Model performance measures included the deviance explained and the area under the receiver operating characteristic curve (AUC).…”

Section: Methodsmentioning

confidence: 99%

“…Model performance measures included the deviance explained and the area under the receiver operating characteristic curve (AUC). The explained deviance provides a measure of the goodness‐of‐fit between the predicted and raw values (total deviance; Compton et al, ). AUC measures the model's ability to discriminate between presence and absence points.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Modelling the spatial distribution of cetaceans in New Zealand waters

Stephenson

Goetz

Sharp

et al. 2020

Diversity and Distributions

View full text Add to dashboard Cite

Aim Cetaceans are inherently difficult to study due to their elusive, pelagic and often highly migratory nature. New Zealand waters are home to 50% of the world's cetacean species, but their spatial distributions are poorly known. Here, we model distributions of 30 cetacean taxa using an extensive at‐sea sightings dataset (n > 14,000) and high‐resolution (1 km2) environmental data layers. Location New Zealand's Exclusive Economic Zone (EEZ). Methods Two models were used to predict probability of species occurrence based on available sightings records. For taxa with <50 sightings (n = 15), Relative Environmental Suitability (RES), and for taxa with ≥50 sightings (n = 15), Boosted Regression Tree (BRT) models were used. Independently collected presence/absence data were used for further model evaluation for a subset of taxa. Results RES models for rarely sighted species showed reasonable fits to available sightings and stranding data based on literature and expert knowledge on the species' autecology. BRT models showed high predictive power for commonly sighted species (AUC: 0.79–0.99). Important variables for predicting the occurrence of cetacean taxa were temperature residuals, bathymetry, distance to the 500 m isobath, mixed layer depth and water turbidity. Cetacean distribution patterns varied from highly localised, nearshore (e.g., Hector's dolphin), to more ubiquitous (e.g., common dolphin) to primarily offshore species (e.g., blue whale). Cetacean richness based on stacked species occurrence layers illustrated patterns of fewer inshore taxa with localised richness hotspots, and higher offshore richness especially in locales of the Macquarie Ridge, Bounty Trough and Chatham Rise. Main conclusions Predicted spatial distributions fill a major knowledge gap towards informing future assessments and conservation planning for cetaceans in New Zealand's extensive EEZ. While sightings datasets were not spatially comprehensive for any taxa, these two best available approaches allow for predictive modelling of both more common, and of rarely sighted, cetacean species with limited available information.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Modelling the spatial distribution of cetaceans in New Zealand waters

Stephenson

Goetz

Sharp

et al. 2020

Diversity and Distributions

View full text Add to dashboard Cite

show abstract

“…For any given species, as an individual grows its morphology and Manuscript received 2 May 2016; revised 9 February 2017; accepted 10 April 2017. These continuous changes in environmental niche requirements with ontogeny and patterns of population connectivity are a crucial link for understanding the processes that underpin community composition across a mosaic of adjoining benthic habitats (Compton et al 2012. Ward.…”

Section: Introductionmentioning

confidence: 99%

Characterizing ontogenetic habitat shifts in marine fishes: advancing nascent methods for marine spatial management

Galaiduk

Radford

Saunders

et al. 2017

Ecological Applications

View full text Add to dashboard Cite

Niche requirements and habitat resource partitioning by conspecific fishes of different sizes are significant knowledge gaps in the species distribution modelling domain. Management actions and operations are typically concentrated on static habitats, or specific areas of interest, without considering movement patterns of species associated with ontogenetic shifts in habitat usage. Generalized additive models were used to model the body-length-habitat relationships of six fish species. These models were used to identify subsets of environmental parameters that drive and explain the continuous length-habitat relationships for each of the study species, which vary in their degree of ecological and/or commercial importance. Continuous predictive maps of the length distributions for each of the six study species across approximately 200 km of the study area were created from these models. The spatial patterns in habitat partitioning by individuals of different body lengths for all six study species provide strong evidence for ontogenetic shifts. This highlights the importance of considering ontogenetic processes for marine spatial management. Importantly, predictive hotspot maps were created that identify potential areas that accumulate individuals of similar life stages of multiple species (e.g., multispecies nursery areas). In circumstances where limited resources are available for monitoring and management of fish resources, predictive modelling is a valuable tool for studying previously overlooked processes such as ontogenetic habitat shifts. Predictive modelling provides crucial information that elucidates spatial patterns in community composition across mosaics of benthic habitats. This novel technique can contribute to the spatial management of coastal fish and fisheries by identifying areas that are important for different life history stages of multiple fish species.

show abstract

“…Most available predictive applications of ML in marine fish ecology refer mainly to coral reef-associated assemblages in shallow waters (Pittman et al, 2007;Moore et al, 2009;Knudby et al, 2010a;Pittman and Brown, 2011). Such works on demersal fish in other ecosystem types are very scarce (Leathwick et al, 2006a;Monk et al, 2010;Compton et al, 2012;Bergström et al, 2013;Bucas et al, 2013).…”

Section: Introductionmentioning

confidence: 99%

Spatial prediction of demersal fish diversity in the Baltic Sea: comparison of machine learning and regression-based techniques

Smoliński

Radtke

2016

ICES Journal of Marine Science

View full text Add to dashboard Cite

Marine spatial planning (MSP) is considered a valuable tool in the ecosystem-based management of marine areas. Predictive modelling may be applied in the MSP framework to obtain spatially explicit information about biodiversity patterns. The growing number of statistical approaches used for this purpose implies the urgent need for comparisons between different predictive techniques. In this study, we evaluated the performance of selected machine learning and regression-based methods that were applied for modelling fish community indices. We hypothesized that habitat features can influence fish assemblage and investigated the effect of environmental gradients on demersal fish diversity (species richness and Shannon-Weaver Index). We used fish data from the Baltic International Trawl Surveys (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014) and maps of six potential predictors: bottom salinity, depth, seabed slope, growth season bottom temperature, seabed sediments and annual mean bottom current velocity. We compared the performance of six alternative modelling approaches: generalized linear models, generalized additive models, multivariate adaptive regression splines, support vector machines, boosted regression trees and random forests. We applied repeated 10-fold cross-validation, using accuracy as the measure of model quality. Finally, we selected random forest as the best performing algorithm and implemented it for the spatial prediction of fish diversity from the Baltic Proper to the Kattegat. To obtain information on the data reliability and confidence of the developed models, which are essential for MSP, we estimated the uncertainty of predictions with standard deviation of predictions obtained from all the trees in the ensemble random forest method. We showed how state-of-the-art predictive techniques, based on easily available data and simple Geographic Information System tools, can be used to obtain reliable spatial information about fish diversity. Our comparative work highlighted the potential of machine learning method to reduce prediction error in modelling of demersal fish diversity in the framework of MSP.

show abstract

Ontogenetic habitat associations of a demersal fish species, Pagrus auratus, identified using boosted regression trees

Cited by 39 publications

References 51 publications

Modelling the spatial distribution of cetaceans in New Zealand waters

Modelling the spatial distribution of cetaceans in New Zealand waters

Characterizing ontogenetic habitat shifts in marine fishes: advancing nascent methods for marine spatial management

Spatial prediction of demersal fish diversity in the Baltic Sea: comparison of machine learning and regression-based techniques

Contact Info

Product

Resources

About