Improving predictions of coastal benthic invertebrate occurrence and density using a multi-scalar approach

Modelling spatial distributions of biogenic habitat‐forming taxa to inform marine spatial planning

“…a greater density of conservation target(s) (Johnston et al, 2015). Here, models that predict habitat suitability were used in place of abundance models, but caution must be placed on the interpretation of habitat suitability as it may not correspond directly with abundance (Rullens et al, 2021; Stephenson et al, 2022; Watson et al, 2022). In some cases, the main drivers of taxon density can be different to drivers of the probability of occurrence (Johnston et al, 2015; Lester et al, 2022; Tôrres et al, 2012).…”

“…While the modelled layers produced here are a significant improvement from some of these layers, there is still room for improvement. Three key ways that spatially explicit tools could be maximized to improve spatial management in the HGMP, and the seas around New Zealand are as follows: (1) new occurrence data in data‐poor areas (like the eastern edge of the HGMP) could be used to update or independently evaluate the modelled layers produced here; (2) systematically collected abundance data could be used to model the abundance of the taxa groups used here using a hurdle model approach where abundance is ‘conditional on presence’ (see Stephenson et al, 2021, and Watson et al, 2022, for examples of this approach); and (3) other evaluation techniques could be incorporated into the modelling process to weight predictions. For example, spatial cross‐validation and associated fit metrics could be used in the ensemble stage to weight models based on their performance across the spatial extent of the study area.…”

“…The usefulness of habitat suitability modelling depends on a variety of factors including (1) the spatial scale of predictions (Watson et al, 2022); (2) the type of model used (e.g. presence‐only or abundance); (3) the quality of data used (e.g.…”