Fig 1.
Southern Caribbean Sea upwelling and region of interest in the southeastern Caribbean Sea.
(a) Caribbean Sea satellite Sea Surface Temperature (SST) climatology for February (1994–2009) showing the peak of the seasonal upwelling in the southern Caribbean; the region of interest off northeastern Venezuela is marked with a red rectangle. (b) Known distribution of Sardinella aurita (hashed lines) and main fishing areas (black) off northeastern Venezuela. Ground-truthing of the acoustic backscattering echo sound surveys was done for every VECEP survey by exploratory fishing within the three areas delimited in red: 1) Gulf of Cariaco, 2) South Area, and 3) North Area.
Fig 2.
Sea surface temperature and chlorophyll-a conditions in the study area.
The study area (10°N to 12°N and 61.5°W to 65°W) is a region affected by strong upwelling during the first half of the year, and also by the Orinoco River discharge plume to the east. (a) SST climatology for February (1994–2009) shows strong upwelling conditions, including the upwelling plumes (blue and green colors), the upwelling foci (circles, from Rueda-Roa and Muller-Karger [19]), and warmer SSTs to the north in the open Caribbean Sea and to the east in the region of the Orinoco river plume. (b) Chlorophyll concentration (Chl) climatology for February (1998–2009) showing high Chl within the upwelled waters. The eastern region shows false higher Chl due to Colored Dissolved Organic Matter (CDOM) in the Orinoco River plume (delimited by the broken black line). (c) Correlation coefficient between weekly time series of SST and logChl (1998–2009). Spatial averages for this study were computed only for the area to the west delimited with the red polygon to avoid inclusion of the Orinoco plume, i.e. the area with positive SST-Chl correlations or not significant (p>0.05) negative correlations.
Fig 3.
Spanish sardine relative abundance index (sardine-NASC) and SST during each VECEP survey period.
Sardine-NASC is proportional to circle size along the tracks of each VECEP acoustic backscattering echo sound survey. SST was averaged within the days of the survey indicated in each panel. The order of the panels is according to the month when the survey was conducted.
Fig 4.
Spatially-averaged SST during each VECEP cruise compared with the SST climatology.
Spatially-averaged SST during each VECEP acoustic cruise (V1 to V8) compared to the weekly SST climatology (1994–2009). Spatial averages and standard deviations (SD) were calculated for the area defined in Fig 2C, which excludes the Orinoco plume. An envelope of ±1 SD are shown for the climatology (dotted line) and the Vecep averages (bars).
Fig 5.
Averages of sardine-NASC and schools for SST intervals.
SST histograms calculated from SST extracted along each cruise track. For each SST-bin of 0.3°C, averages of log(sardine-NASC) (red line) and of the number of small pelagic schools (blue line) were calculated and scaled to percent. Stars indicate the SST-bin where 50% of the average sardine-NASC and school number occurred.
Fig 6.
Boxplots and Generalized Additive Model (GAM) analysis of sardine-NASC and Schools for each Vecep cruise.
Boxplots of (a) sardine-NASC and (b) Schools for each Vecep cruise, and GAM analysis of (c) log(sardine-NASC) and (d) number of Schools vs. hour. Regardless of the season, the first four surveys showed higher averages of sardine-NASC than the last four surveys; however, the number of schools didn’t show differences between surveys. The GAM analyses show the typical diel patterns of aggregation of sardines. The backscattering strength of sardine is weaker during day time, because they are concentrated into school of fish; the opposite occurs during night time.
Fig 7.
GAM analysis of sardine-NASC vs. SST, distance from upwelling foci, and location for different seasons.
Univariate GAM analysis of log(sardine-NASC) vs. SST, distance from upwelling foci, and Location (longitude, latitude) for: Cool (top a, b and c, VECEP 2, 3, 5, 6 and 8), Transition (middle d, e and f, VECEP 4), and “Warm” (bottom g, h and i, VECEP 1 and 7) conditions. The y-axes (and the isolines in the map) show relative changes of the environmental variable on sardine-NASC, f(x), so that a value of zero represents the mean effect. Positive isolines in Location are highlighted with hashed lines. Each parameter has similar scales to contrast its importance between different upwelling conditions. For SST and foci distance the values where the data lay is marked on the x-axis (rugplot). The 95% confidence intervals are grey-shaded for SST and foci distance. All terms were significant (p<0.01) and the sardine-NASC ‘percentage of deviance explained’ (analogous to variance in a linear regression) are shown for each term.
Fig 8.
Chlorophyll-a conditions and GAM analysis between strong and weak upwelling conditions.
Comparison of Chlorophyll-a conditions between strong upwelling (left column) and weak upwelling (right column) conditions for: (a) Satellite Chlorophyll-a (Chl); GAM analysis of (b) log(sardine-NASC) vs. logChl, (c) logChl vs. distance from upwelling foci, and (d) logChl vs. SST. GAM 95% confidence intervals are grey shaded. All terms were significant (p<0.01) and the percentage of deviance explained are shown.
Table 1.
GAM’s percentage of deviance explained of sardine-NASC and Schools with the different environmental predictors for the three seasons of the upwelling cycle.
Fig 9.
GAM analysis of schools vs. SST, distance from upwelling foci, and location for different seasons.
Univariate GAM analysis of the number of small pelagics schools vs. SST, distance from upwelling foci, and location (longitude, latitude) for: Cool (top a, b and c, VECEP 2, 3, 5, 6 and 8), Transition (middle d, e and f, VECEP 4) and weak or “Warm” (bottom g, h and i, VECEP 1 and 7) upwelling conditions. GAM 95% confidence intervals are grey shaded; Lat-Lon positive isolines are highlighted with hashed lines. Percentage of deviance explained is shown for each term. All terms were significant (p<0.01), except for 9e and 9g, which were no-significant (n.s.).
Fig 10.
Relationship between the pelagic fish density and the number of pelagic school averaged for each VECEP cruise.
Pelagic fish density (pelagic-NASC) vs the number of pelagic schools averaged for all the elementary sampling distance unit (ESDU) at each cruise (All), and for the ESDUs within 10 km of the upwelling foci (<10km) and the ESDUs farther offshore (>10km). Correlations were significant (p<0.05) and with similar slope, but data closer to the upwelling foci had a higher intercept, due to the larger average of school number closer to the coast in comparison with the rest of the ESDUs.
Fig 11.
Comparison of annual catch of Sardinella aurita in northeastern Venezuela with annual spatial averages of SST and Chl.
SST and Chl were spatially-averaged for the area demarked in Fig 2C. Long term means of sardine catch (1990–2010), SST (1994–2009) and Chl (1998–2009) are shown for reference (horizontal dotted lines). Annual catch of Spanish sardine was obtained from the Global Capture Production Dataset from FAO [4]. Up to 2014 the annual sardine catch had remained around the 2010 catch of 50,000 tons [4].