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

over the space according to Moran’s I statistics. This is shown by regression residuals which exhibited significant spatial autocorrelation, i.e., there are spatial-dependent elements in the regression residuals (I = 0.012, p-value = 0.015). Therefore, the residuals autocovariate (RAC) was added as a new variable for the model selection, after which the spatial autocorrelation according to Moran’s I statistics was not significant (I = 0.003, p = 0.091). When looking for environmental predictors to explain the distribution of species richness, the best GAM explaining species richness included Chla, S, Tmean and RAC (Table 2). The adjusted R2 value revealed that 60% of the variability is explained by this model. The predicted values of alfa diversity were concordant with the observed data (r = 0.77; S4 Fig). Regarding species composition within the 0–200 m range, a significant spatial correlation according to Moran’s I statistics was observed, as regression residuals exhibited significant spa- tial autocorrelation, (I = 0.018, p-value = 0.005). After we added RAC as a new variable for the model selection, spatial autocorrelation did not occur based on Moran’s I statistics (I = -0.006, p-value = 0.726). The best GAM model explaining the species composition of copepods in the SPO included Chla, O2, S, Tmean and RAC (Table 3). The adjusted R2 value revealed that 70% of the variabil- ity is explained by this model. The predicted values of beta diversity were concordant with the observed data (r = 0.84; S5 Fig). Discussion Spatial patterns of biodiversity and their predictors

Hot and cold spots patterns of copepod distribution in the SPO do not seem to follow the pre- viously reported latitudinal trends of global patterns of marine species with peaks of diversity Table 2. GAM+RAC models for species richness. Statistics acronyms are BIC = Bayesian information criterion, ΔBIC = delta BIC (i.e., the difference in BIC score between the best model and the model being compared). Predictors’ acronyms are: Tmean = mean temperature, S = salinity, Chla = chlorophyll-a concentration, MLD = mixed layer depth, autocovariate = residuals autocovariate. The best model is highlighted in bold. Model BIC ΔBIC 1 Species richness ~ Chla, S, Tmean + autocovariate 2364.54 0 2 Species richness ~ Chla, MLD, S, Tmean + autocovariate 2365.47 0.93 https://doi.org/10.1371/journal.pone.0306440.t002 Table 1. Summary of spatial characteristics and cells sampled per each main surface current system of the SPO. CHC = Cape Horn Current; EAC = East Australian Current; HCS = Humboldt Current System; SEC = South Equatorial Current; SPSG = South Pacific Subtropical Gyre; WWD = West Wind Drift; n.a. = not applicable. Current Area (Km2) Total number of cells Number of sampled cells Total number of observations Number of species observed per current Sampled area per current (Km2) % of the current’s area sampled % of the area of the SPO sampled CHC 2.65 *106 152 5 55 3 6.16 *104 2.32 0.05 EAC 1.93 *107 1468 270 18207 148 3.33 *106 17.28 2.76 HCS 1.59 *107 1230 16 2527 91 1.97 *105 1.24 0.16 SEC 6.74 *106 685 12 510 94 1.48 *105 2.19 0.12 SPSG 6.14

*107 4436 67 899 24 8.26 *105 1.34 0.68 WWD 1.46 *107 804 23 147 10 2.83 *105 1.94 0.24 Total 1.21 *108 8775 393 22345 n.a. 4.84 *106 n.a. 4.02 https://doi.org/10.1371/journal.pone.0306440.t001 PLOS ONE Biodiversity of copepods in the South Pacific PLOS ONE | https://doi.org/10.1371/journal.pone.0306440 July 11, 2024 15 / 28