Biodiversity patterns of epipelagic copepods in the South Pacific Ocean: Strengths and limitations of current data bases

from the selected environmental variables. We employed Generalized Additive Models (GAMs) to evaluate the relationship between copepod alpha and beta diversity in the 0–200 m layer of the South Pacific Ocean and the environmental predictors while considering spatial autocorre- lation in our analyses. Spatial correlation is a pattern in which observations are related to one another by the geographic distance between the observation, with locations close to each other exhibiting more similar values than those further apart, thus violating the assumption of inde- pendent and identically distributed errors, which can inflate type I errors [67]. This can lead to the selection of unimportant explanatory variables and poorly estimated parameters, moreover in ecology, as autocorrelation is a general property of ecological variables measured over geo- graphic space [67]. Therefore, to take spatial autocorrelation into account, we used the residuals autocovariate (RAC) approximation [67] by including autocorrelation in the GAMs by adding another term to it that represents the influence of neighboring observations on the dependent variable at a particular location. Subsequently, the residuals of the auto-covariables were used as a new predictive variable in the models. The calculations of GAM+RAC were performed with the spdep package [68, 69] implemented in R software (R Core Team, 2021). The model selection was done through Bayesian Information Criterion (BIC), that is based on information theory but within a Bayesian context and imposes a greater penalty for the

number of parameters compared to the Akaike information criterion [70]. The best model is the one that provides the minimum BIC, denoted by BIC*, and delta BIC can be computed as: ΔBIC = BICm−BIC*. This implies that for given M models, the magnitude of the delta BIC can be interpreted as evidence against a candidate model being the best model [70]. As a rule, val- ues of ΔBIC less than 2 are not worth more than a mention, values between 2 and 6 indicate that evidence against the candidate model is positive, values between 6 and 10 indicate that evi- dence against the candidate model is strong, and values greater than 10 indicate that evidence against the candidate model is very strong [70]. The analyses were conducted through the package MuMIn [71], in R software (R Core Team, 2021). Beta diversity was evaluated at each side of the SPO using Generalized Dissimilarity Model- ling (GDM)-based spatial analysis technique, which involves mapping the first three axes derived from a principal component analysis (PCA) performed on the set of transformed pre- dictor layers [72]. It first formulates the relationship between biological dissimilarity and spa- tial distance by transforming each predictor variable through l-spline basis functions considering a term called ‘predicted ecological distance’ in a negative exponential function, with the assumption that dissimilarity between site-pairs increases monotonically from 0 to 1 and in a saturating manner with predicted ecological distance. Based on the maximum height

of the spline function (and hence, the maximum value of the transformed predictor), which shows the importance of the predictor in explaining dissimilarities, only the predictors with maximum height over zero were selected for the generation and plotting of the spatial layers. A RGB (RedGreenBlue) color palette dimension was assigned to each of the three axes of the PCA, and they were mapped by combining them, finally enabling the visualization of areas predicted to have more similar species composition (shown as similar colors) or less similar species composition (shown as different colors). The GDM analysis was performed using the package gdm [73] in R software (R Core Team, 2021) and its plotting was done through Kri- ging interpolation using ArcMap 10.4.1 (ESRI 2016) software. Results Diversity patterns For the South Pacific Ocean, a total of 178 species were reported at the upper 200 m water col- umn of its basin. Regarding the main species present at each main surface current system, we PLOS ONE Biodiversity of copepods in the South Pacific PLOS ONE | https://doi.org/10.1371/journal.pone.0306440 July 11, 2024 7 / 28