Zooplankton biodiversity and temporal dynamics (2005–2015) in a coastal station in western Portugal (Northeastern Atlantic Ocean)

Statistical analysis

Average abundance and biomass values were tested for differences among months, seasons (hereafter referring to astronomical seasons) and years using one-factor fixed effects ANOVA models, when the homogeneity of variance was verified in the Levene test. For non-significant homogeneity of variance, Kruskal-Wallis analyses were conducted instead. Pairwise tests (Tukey or Mann–Whitney) were used to identify the factors contributing to the statistical differences. The grouping of the data intended to reflect the irregularity in the sampling effort, by using the most uniform points of comparison possible. For all the statistical analyses, taxa, when necessary, were considered by major groups (Mollusca, Diplostraca, Decapoda, etc.). To uncover the dominant taxa/species responsible for the observed differences in the community composition, a principal component analysis (PCA) was performed. Additional PCA analyses were performed with standardized data (zero mean and unit variance, i.e., dimensionless data) to examine correlations among the environmental (temperature, chlorophyll a concentration, upwelling index and precipitation) and biological (biomass, abundance of zooplankton and taxa groups) parameters. The abundance ratios of Holoplankton:Meroplankton, Gelatinous:Crustaceans and Cyclopoida:Calanoida were determined to assess potential shifts in the zooplankton communities. Diversity (Shannon-Wiener; Simpson diversity index, 1-D), species richness (Margalef), eveness (Pielou, J) and taxonomic (Menhinick, D) indices were computed. Significant relationships were further explored with generalized linear models (GLM). For monthly analyses, environmental missing values were replaced by 10-year averages. Lomb-Scargle periodograms and Mann-Kendall trend tests were used on the abundance data to detect periodicities and monotonic tendencies in the occurrence patterns of each taxon, respectively. These methods are useful to explore unevenly sampled data and were applied to values of averaged abundances, adjusted by season and month, and to the entire time-series (excluding non-sampling periods). The statistical analyses were performed using Statistica (StatSoft, Inc., http://www.statsoft.com), PAST (Paleontological statistics software package for education and data analysis) (Hammer, Harper & Ryan, 2001) and MatLab (http://www.mathworks.com/).

Complementarily, dynamic factor analysis (DFA) computations were applied to the abundance data of the most abundant taxa identified in the previous analyses, using the Brodgar (Highland Statistics Ltd) software. DFA is an adequate tool to explore patterns in time-series, especially those covering short temporal periods or composed of nonstationary data (Zuur et al., 2003). Although the temporal range of the present data is not the ideal, the missing gaps on the data were one of the main reasons for the application of this technique. Nevertheless, caution is needed when examining these patterns that represent only indications that need to be improved in the future. The method provides smoothed functions through time, trends and relationships between variables. The analyses were performed on transformed data (ln [x + 1]) and the missing data identified as such. The environmental factors were used as explanatory variables. The DFA model fit was applied to the entire time-series and specifically to several Copepoda taxa/species. One to three common trends (CTs) were tested, and the best solution was chosen according to the lowest Akaike information criterion (AIC).

Hydrographic conditions

The zooplankton community at the CCW was influenced by the seasonal conditions usually found in temperate regions: high sea surface temperatures during summer and early autumn (average values above 18.1 °C from June to November), and lower temperatures in late autumn and winter (minimum of 13.5 ± 1.2 °C, Fig. 2B). Spring and summer months were also characterized by high productivity (Fig. 2B), as well as intense upwelling, which was strongest during July and August (Fig. 2A). The lowest productivity was detected in winter months, associated with the decrease of the upwelling intensity. The recent years were characterized by slightly higher maximum sea surface temperatures and increased variability in the productivity pattern (Fig. 2B). From 2011 onward, extended dry seasons were evident from the precipitation patterns, contrasting with the years prior, characterized by rainy periods from October to March and dry seasons mainly limited to late spring and summer (Fig. 2A).

Zooplankton composition and temporal distribution

Despite the seasonal patterns perceived in the environmental characterization, no significant differences, monotonic tendencies, or periodicities were detected by month or season (Tables SI and SII) in the patterns of the zooplankton abundance, biomass, and diversity at the CCW station, represented in Fig. 3 and Fig. S1. The only exceptions were the significant differences found in the zooplankton abundance for the summer and autumn samples, for the first years of sampling (Table SII). Accordingly, the monthly and seasonal patterns were relevant, with high zooplankton abundance and biomass during summer and early autumn months (Fig. 3), especially in July and October (monthly averages of 8,956 ± 11,981 ind.m⁻³, 213 ± 193.4 mg m⁻³ and 12,321 ± 16,283 ind.m⁻³, 86.2 ± 110.4 mg m⁻³, respectively). Therefore, slightly higher biodiversity was detected in summer months (Fig. S1), although not statistically significant (Table SII). The zooplankton biomass also peaked in spring (average of 77.8 ± 45 mg m⁻³, Fig. 3A), particularly in April (average of 107 ± 96.4 mg m⁻³; Fig. 3B). Winter months corresponded to the lowest zooplankton abundances (average of 4,116 ± 2,769 ind.m⁻³) and biomass (56 ± 46 mg m⁻³; 10.2 mg m⁻³in January) (Fig. 3A).

Among the 86 taxa found, 43 were identified to family or higher level, 25 to genus and 18 to species (Table SIV). The monthly, seasonal and interannual composition of the major taxa found in the samples is presented in Fig. 4.

Copepoda dominated the community (71%–96% of the total), presenting two abundance peaks in July and October. The Copepoda proportion decreased in summer (Figs. 4 and 5B), following the increased contribution of Mollusca (up to circa 25–58% of the samples), Diplostraca (Cladocera) and Cnidaria (Figs. 4 and 5). Cirripedia, Appendicularia and Decapoda were a relevant component of the winter and spring samples (circa 20%).

From the PCA analyses we found that the major contributors for the observed variability in the community composition were Acartia spp., Calanus spp., Oncaea spp., Bivalvia, Oithona spp., Evadne spp., Penilia avirostris and Centropages spp. (Fig. S5, Table SIV). Bivalvia, Acartia spp. and Evadne spp. were associated with the summer samples, while Penilia avirostris, Calanus spp., Oithona spp., Oncaea spp. and Centropages spp. contributed to the distinction of the autumn samples (Fig. S5, Table SIV).

Zooplankton interannual variability

The interannual data, showed a relative increase of the average zooplankton abundance (Fig. 5A). When comparing the years of 2005–2009 with the latter ones, a lower proportion of Copepoda, Decapoda and Cirripedia, and a higher proportion of Mollusca (mostly Bivalvia, see Table SIV) were suggested in the relative abundance (Fig. 4B). The DFA analysis also hinted at an increasing tendency of Mollusca abundance from 2011 onwards (Fig. S2A), similar abundances through the years for Copepoda, despite the decreases in 2009 and 2014, and decreases of Cirripedia and Decapoda larvae (Fig. S2).

The average meroplankton/holoplankton ratio has increased since 2009 (Fig. S4), reflecting the increase in the proportion of Bivalvia larvae. Regarding the gelatinous/Crustacea ratio, the Crustacea had a higher contribution to the samples than gelatinous zooplankton (Fig. S4), despite the enhanced abundance of the latter in spring months since 2010 (Table SIV, Fig. 5G, Figs. S2–S3). Lastly, the Cyclopoida/Calanoida ratio shows a different pattern, with low values registered in 2013–2015 (Fig. S4).

In the DFA model fit (Figs. S2 and S3), the AIC values obtained for one, two and three common trends (CTs) were, respectively: 1596.6, 1535.3 and 1509.4, for the entire time-series; and 1500.4, 1364.5 and 1317.8, for Copepoda (models with three CTs chosen as the best fit). The models generally depicted increasing trends in the abundance of Mollusca, Cnidaria and Diplostraca, but also of fish larvae/eggs and the total zooplankton abundance.

For the Copepoda, no significant monotonic tendencies were detected (Table SI), despite an apparent interannual decrease (Fig. 5B). The statistical analyses showed significant differences in copepod abundance for the period 2005–2009 (Table SII). The copepod abundance was positively correlated with the SST and the upwelling index in the PCA analyses (Figs. 6A and 6B).

Acartia spp. (20% of all samples), Calanus spp. (14%), Oncaea spp. (12%) and Oithona spp. (6%) dominated the 2005–2009 samples (from 51 to 96%), decreasing in proportion from 2010 onwards (Fig. 7, Table SIV). In the ANOVA and Kruskal-Wallis analyses, these interannual differences were only significant for Acartia spp. (Table SII), although a significant monotonic decreasing tendency was detected for Calanus spp. in the Mann-Kendall tests (Table SI) and represented in the DFA analyses (Figs. S2 and S3). Higher proportions of Calanoida were detected in recent years, with increasing tendencies of Clausocalanus spp., Paracalanus spp. (up to 3% of the total, only significant for the former, per year; Tables SI, SII and SIV), and Centropages spp. (Fig. 7), also generated by the DFA analyses (Figs. S2 and S3). In the PCA analyses, Calanus spp., Acartia spp. and Paracalanus spp. showed positive correlations with the SST (Fig. 6C).

The observed decrease in the Cyclopoida: Calanoida ratio, with lower proportions of Cyclopoida in the 2013–2014 samples (Table SIV, Fig. S4), followed the decreasing trend of Oithona spp. shown in the DFA models (Figs. S2B, S2E, S2F and S3B). On the other hand, the lower proportion of Oithona spp. was positively correlated with precipitation, as shown in the PCA analyses (Fig. 6C). A positive correlation with precipitation was also obtained for Oncaea spp. (Fig. 6C). Additionally, Oithona spp. was positively correlated with the upwelling index in the GLM analyses (Table SV).

The increasing monotonic tendency of Mollusca (circa 10% in the total samples; Table SI), was significantly different for late spring and summer months (Fig. 5C), with positive correlations with year, upwelling intensity, and SST in the PCAs (Figs. 6A and 6B). Bivalvia, the most represented (average of 97% of the samples; Table SIV), showed a monotonic increasing tendency per season and month (MK p < 0.02 in all cases; not shown), and a significant correlation with SST (Table SV). For Mollusca, significant differences were found between the first and later years of sampling (Table SII).

Abundance peaks of Diplostraca (Cladocera) were detected in July and October (Fig. 4A). No monotonic tendencies were observed (Table SI), despite the apparent autumn and summer interannual decrease (Fig. 5D). The slight increase for spring periods (Figs. 4B and 5D) was also hinted at in the DFA analyses (Figs. S2 and S3). Positive PCA-based correlations were obtained with SST and the upwelling intensity (Fig. 6A). Evadne spp. and Penilia avirostris were the most abundant taxa (48 and 43% of the total, respectively) (Table SIV) and the latter showed significant correlations with SST and the upwelling intensity (Table SV).

Cirripedia registered the highest abundances in late winter and spring (Fig. 4A), with monotonic decreasing tendencies (Table SI, Fig. 5F), that were not entirely represented in the DFA analyses (Figs. S2 and S3). The PCA analyses revealed negative correlations with SST and chlorophyll a (Fig. 6A).

For Cnidaria, mainly siphonophores (circa 60%) and hydromedusae (39%) were identified (see Table SIV). July registered the highest abundances (Fig. 4A) and, the apparent spring/autumn interannual increase (Figs. 4B and 5G), depicted in the DFA analyses (Figs. S2 and S3), did not manifest in monotonic tendencies (Table SI). Positive PCA-based correlations were obtained with year, SST and upwelling intensity (Figs. 6A and 6B).

Decapoda showed decreasing monotonic tendencies (Table SI), and significant differences for winter samples and the first year of sampling (Table SII). Abundance peaks were detected in winter and spring (Fig. 4B), as well as March (Fig. 4A), showing a significant correlation with SST in the GLM analyses (Table SV).

General patterns of zooplankton biomass, abundance and composition

High year-round productivity was observed for the study area, without major significant trends or periodicities detected in the zooplankton abundance, biomass, and diversity, even despite the relatively lower winter values. The high chlorophyll levels during the entire year were undeniable, when compared with values obtained in other works (e.g., Bode et al., 2009; Eloire et al., 2010; Bresnan et al., 2015; Yebra et al., 2020), namely for Galician waters (Buttay et al., 2016). The average zooplankton biomass followed the values reported by Domínguez et al. (2017) for the northern Portuguese coast. Nevertheless, the present study detected spring/summer and autumn peaks in zooplankton abundance, complying with the common seasonal shifts reported for zooplankton communities of temperate latitudes and for the region (e.g., Valdés & Moral, 1998).

The composition of the zooplankton community at CCW was alike to what has been reported for the Iberian coast (e.g., Valdes et al., 1990; Valdés & Moral, 1998; Valdés et al., 2007; Bode et al., 2009; Sobrinho-Gonçalves et al., 2013; Domínguez et al., 2017). The dominance of Copepoda also followed the results of works for other areas of the northeastern Atlantic and adjacent seas (e.g., Eloire et al., 2010; Bresnan et al., 2015; Valdés et al., 2021). The spring/summer decrease observed in the Copepoda proportion was not accompanied by the taxa abundance, which was higher during these periods, largely due to the greater diversity of other zooplankton groups and the dominance of certain copepod taxa. Acartia spp., Paracalanus spp., Clausocalanus spp., and Oncaea spp., highly abundant in the samples, have also been described as widely distributed and frequent in the northwestern Iberian coast (e.g., Bode et al., 2012), being some of the most common taxa recorded in the northeastern Atlantic (e.g., Valdés et al., 2007; Sobrinho-Gonçalves et al., 2013; Domínguez et al., 2017). The importance of Copepoda as intermediaries between distinct trophic levels across the marine food web is undeniable, namely for planktivorous fish (e.g., Garrido et al., 2015) that usually decrease in abundance following the lower availability of Copepoda (e.g., Heneghan et al., 2020).

Interannual shifts in the zooplankton community

The spring/summer and autumn zooplankton abundance peaks followed the seasonal variation of the environmental variables, namely the sea surface temperature and upwelling index, which had relevant impacts in some of the taxa (see below). The link between the zooplankton abundance and the upwelling intensity is not surprising, given its well-known importance for zooplankton productivity (e.g., Pires & Dos Santos, 2020). In the Portuguese coast, wind-driven upwelling of colder and richer deep waters is frequent in spring/summer months, when temperature increases and north/northwestern winds and southward currents dominate, contrasting with the winter prevalence of southwestern winds and northward slope flows (e.g., Relvas et al., 2007). The upwelling regimes, together with the influence of the Tagus plume, especially when coupled with high precipitation (e.g., Vaz et al., 2009), enhance chlorophyll concentration and phytoplankton production at surface coastal waters, leading to increased zooplankton abundance. The local sheltered dynamics, with slower currents and recirculation features, result in the coastal retention of plankton (e.g., Moita et al., 2003) thus contributing to the high productivity and biodiversity observed herein. Indeed, previous works pointed to distinct inshore and offshore communities of zooplankton in the Portuguese coast (e.g., Bartilotti et al., 2014; Domínguez et al., 2017), stressing the importance of the coastal environmental conditions in shaping zooplankton communities. This may have important implications for the regional ecosystem yield capacity, since the frequent local generation of upwelling filaments (e.g., Haynes, Barton & Pilling, 1993) may contribute to the transfer of productivity towards offshore areas.

The slightly increasing interannual zooplankton abundances detected herein for spring periods, contrast with the general declining trends reported for the North Atlantic (e.g., Iriarte et al., 2022; McQuatters-Gollop et al., 2022). Nonetheless, our observations parallel the increasing zooplankton trends detected for the Benguela upwelling region (Verheye & Richardson, 1998). Inhabiting a transition area between temperate and tropical environments (e.g., Boaventura et al., 2002) that comprise the northern/southern limit of distribution of many species, the CCW zooplankton communities may experience less drastic effects than those reported for the North Atlantic. On the other hand, the upwelling regimes, as one of the drivers of long-term increase of crustacean zooplankton (Verheye & Richardson, 1998), may also explain the observed trend. In the long-term analyses for the western Atlantic by Morse et al. (2017), the spring community shifts detected in the composition and abundance of zooplankton were associated with environmental changes mediated by the North Atlantic Oscillation (NAO).

We report two main periods (2006 and 2009/2010) of lower abundance of zooplankton, particularly for Copepoda, fish larvae, Cnidaria, Diplostraca and Cirripedia, that apparently induced shifts in the community. The 2009/2010 period was characterized by unusual climatic variability throughout the North Atlantic, including warmer and more saline waters, driven by a strong negative NAO (e.g., Hughes, Holliday & Beszczynska-Möller, 2011). During this period, in Portugal, heatwaves were registered from spring to autumn, extending the annual dry season (IPMA, 2010; IPMA, 2011).