Abstract
Relative to target species, priority conservation species occur rarely in fishery interactions, resulting in imbalanced, overdispersed data. We present Ensemble Random Forests (ERFs) as an intuitive extension of the Random Forest algorithm to handle rare event bias. Each Random Forest receives individual stratified randomly sampled training/test sets, then down-samples the majority class for each decision tree. Results are averaged across Random Forests to generate an ensemble prediction. Through simulation, we show that ERFs outperform Random Forest with and without down-sampling, as well as with the synthetic minority over-sampling technique, for highly class imbalanced to balanced datasets. Spatial covariance greatly impacts ERFs’ perceived performance, as shown through simulation and case studies. In case studies from the Hawaii deep-set longline fishery, giant manta ray Mobula birostris syn. Manta birostris and scalloped hammerhead Sphyrna lewini presence had high spatial covariance and high model test performance, while false killer whale Pseudorca crassidens had low spatial covariance and low model test performance. Overall, we find ERFs have 4 advantages: (1) reduced successive partitioning effects; (2) prediction uncertainty propagation; (3) better accounting for interacting covariates through balancing; and (4) minimization of false positives, as the majority of Random Forests within the ensemble vote correctly. As ERFs can readily mitigate rare event bias without requiring large presence sample sizes or imparting considerable balancing bias, they are likely to be a valuable tool in bycatch and species distribution modeling, as well as spatial conservation planning, especially for protected species where presence can be rare.
Highlights
Machine learning algorithms have proven to be a ubiquitous tool for modeling species distributions
The Ensemble Random Forests (ERFs) approach did as well or slightly better than Random Forest (RF)-DS and RF-synthetic minority oversampling technique (SMOTE) approaches for area under the curve (AUC), root mean squared error (RMSE), and true skill statistic (TSS) performance metrics across the range of detection probabilities for the external holdout dataset (Fig. 2A−C) and the full dataset (Fig. 2D−F)
The RF approach outperformed all other approaches for the RMSE performance metric for all detection probabilities that resulted in class imbalance (δ ≥ 0.25) (Fig. 2B,E)
Summary
Machine learning algorithms have proven to be a ubiquitous tool for modeling species distributions. A variety of methods have been utilized in ecology to deal with overdispersion (e.g. zero-inflated, hurdle, or delta− generalized linear models), most of which implement a mixture model in some fashion (Zuur et al 2009, Campbell 2015, Stock et al 2020) While these models can be well suited to modeling rare event data, large sample sizes are often needed to overcome the low proportion of presences and estimate model parameters with adequate precision (He & Garcia 2009, Rodriguez-Galiano et al 2012). Manta birostris (White et al 2018) (Threatened), scalloped hammerhead Sphyrna lewini (Endangered), and false killer whale Pseudorca crassidens (Endangered) The purpose of these simulation and case study objectives was to demonstrate the effects of sample size and sample covariation on model performance and the utility of ERFs as a bycatch and species distribution modeling tool.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
