Species distribution modelling of the Southern Ocean benthos: a review on methods, cautions and solutions

Abstract

Species distribution modelling studies the relationship between species occurrence records and their environmental setting, providing a valuable approach to predicting species distribution in the Southern Ocean (SO), a challenging region to investigate due to its remoteness and extreme weather and sea conditions. The specificity of SO studies, including restricted field access and sampling, the paucity of observations and difficulties in conducting biological experiments, limit the performance of species distribution models. In this review, we discuss some issues that may influence model performance when preparing datasets and calibrating models, namely the selection and quality of environmental descriptors, the spatial and temporal biases that may affect the quality of occurrence data, the choice of modelling algorithms and the spatial scale and limits of the projection area. We stress the importance of evaluating and communicating model uncertainties, and the most common evaluation metrics are reviewed and discussed accordingly. Based on a selection of case studies on SO benthic invertebrates, we highlight important cautions to take and pitfalls to avoid when modelling the distribution of SO species, and we provide some guidelines along with potential methods and original solutions that can be used for improving model performance.

Classic or hybrid? The performance of next generation ecological models to study the response of Southern Ocean species to changing environmental conditions

“…However, the relevance of niche estimation often constitutes the main limitation to ‘classic SDMs’, because their predictive performance strongly relies on sampling completeness (Araújo et al, 2005; Broennimann et al, 2007; Holt, 2009; Loehle & Leblanc, 1996; Randin et al, 2006; Vaughan & Ormerod, 2003). The heterogeneity of presence sampling induces statistical artefacts that can bias model predictions (Bahn & McGill, 2007; Currie, 2007), a substantial limitation that has already been stressed in former works on the Southern Ocean (Guillaumot et al, 2020, 2021; Guillaumot, Martin, et al, 2018).…”

“…(3) Finally, there is a lack of presence data to correctly calibrate the model, to validate it and to accurately consider model predictions as accurate likelihood of species distribution. Generating ecological models with small datasets was indeed shown to reduce modelling performances (Liu et al, 2019; Stockwell & Peterson, 2002) as it truncates predicted distribution and niche definition (El‐Gabbas & Dormann, 2018; Hortal et al, 2008), and may lead to a reduction in model accuracy because the presence and background datasets would not differ markedly (Luoto et al, 2005) and constrain the evaluation process (Pearson et al, 2007) (reviewed in Guillaumot et al, 2021). Therefore, common validation approaches such as the cross‐validation method (that uses a part of the dataset to train the model and another part to test it independently, Hijmans, 2012; Guillaumot et al, 2019) could not have been used for our study, which limited the power of our model evaluation.…”

“…Due to remoteness and harsh weather conditions, above all in winter, access to the field and data collection in the Southern Ocean are strongly limited (De Broyer et al, 2014), resulting in missing data, spatial and temporal aggregations of observations and difficulties to conduct biological experiments (see Guillaumot et al, 2021 as a review). However, research on marine life of the Southern Ocean has recently benefited from a significant coordinated and international effort with the emergence of oceanographic campaigns and international scientific programs such as the International Polar Year (IPY 2007–2008), the Census of Antarctic Marine Life (CAML 2005–2010) or the Scientific Committee on Antarctic Research, Evolution and Biodiversity in Antarctica (SCAR‐EBA 2006–2013) (nonexhaustive list) (De Broyer et al, 2014; Schiaparelli et al, 2013).…”