8142

• UCREFRP Catalog Number: 8142
• Author: Horn, M. J.
• Title: Nutritional Limitation of Recruitment in the Razorback Sucker ‎(
• USFW Year: 1996.
• Copyright Material: NO

I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

66

W=aLb.
The test of log-linearity was accomplished by fitting a second-order polynomial term to
residuals of the log-linear regression model (Y = a + bL + cL2, where Y are the residuals

and L is the Log-transformed length). A second-order coefficient, c, not significantly

different from zero indicates no departure from log-linearity. If significant departures are

detected, the more complex Gompertz model would be utilized (pepin 1995):

LogW = i + bLogLc + e.

Equality of slopes between treatments was tested using an analysis of covariance with

length as a covariate. Significant differences in slope between treatments indicated the

relationship between length and weight as larvae grew were on different trajectories for

each treatment.

For all analyses, pre-feeding (yolk-sac) larvae were excluded. Pre-feeding larvae

typically do not fit the log-linear growth curve, so a plot of the mean-variance ratio of

weight against length class was used to determine the minimum cutoff size (Murphy et al.

1990).

To determine the efficacy of using lipids, weight, and length to distinguish among

feeding levels, discriminant function analyses were conducted for laboratory larvae at

each temperature. Reclassification of data based on the discriminant function was used as

an index of how well variables distinguished among the different groups. In addition,

cross-comparisons were made where the discriminant function from one experiment was

used to classify individuals from another. Reclassification using the two data sets avoids
the circularity of testing a discriminant model with the same data used to build it. In a