Correlation of Simulated Densities to Larval Catch Data


Correlations between simulated densities and the corresponding larval catch densities at all of the sites were almost always positive, often significant, but uniformly low, indicating that the simulated densities captured the general pattern of distributions but were poor predictors of larval catch densities in individual nets (table 2). The way that the correlations vary among gear types and species, however, suggests some gear-specific differences with regard to size class and species.


The smallest size class of fish caught in pop nets was most highly correlated with simulated densities (correlation coefficients ranged from 0.88 in 2006 to 0.27 in 2009), and the largest size class of fish caught in pop nets was least correlated with simulated densities (the sign of the correlation coefficients was mixed, and the correlation coefficient was positive and significant only in 2008, at 0.24). In contrast, the middle size class of fish caught in larval trawls was most highly correlated with simulated densities (correlation coefficients ranged from 0.49 in 2006 to 0.57 in 2009), and the smallest size class caught in larval trawls was least highly correlated with simulated densities (correlation coefficients ranged from 0.56 in 2007 to 0.20 in 2009). The largest size class of fish caught in larval trawls was not correlated (p<0.05) with the simulated densities, but the correlation coefficients were all positive and ranged from 0.29 in 2007 to 0.23 in 2009. The smallest size class of fish caught in plankton nets was significantly correlated with simulated densities in 2008 and 2009 (correlation coefficient 0.21 in 2008 and 0.25 in 2009, p<0.05), but the middle size class caught in plankton nets was correlated with simulated densities only in 2009; and in 2008, the only year when the largest size class of fish was caught in plankton nets, the catch was not correlated with simulated density (p<0.05). 


When the 2009 larval catches were separated by species and the two species were modeled separately, the plankton net data were significantly correlated with the simulated densities of SNS/KLS suckers (correlation coefficient 0.22, p<0.05), but not with simulated densities of LRS. The larval trawl data were correlated approximately equally with simulated densities of SNS/KLS suckers and with LRS suckers (correlation coefficients 0.48 and 0.42, respectively; p<0.05). The pop net data were more highly correlated with the simulated densities for LRS suckers than for SNS/KLS suckers, although both correlations were significant (correlation coefficients 0.34 and 0.20, respectively; p<0.05). 


Pop net correlations were somewhat higher for small sized fish when nighttime-only drift was assumed to occur only upstream of the Modoc Point Road boundary, but plankton net and larval trawl correlations in 2009 did not change much based on the assumed extent of nighttime-only drift.


First posted April 2, 2012