Statistical analysis
Long-term data from the LUQ-LTER that comprise the
years 1997–2007 were used to estimate the biannual density
of marked and unmarked shrimp in each pool. To
determine variations in population density, two-way
ANOVAs were performed for each pool using the lowrainfall
(LRF) and high-rainfall (HRF) periods and the
time (years 1997–2007) as factors. Prior to the analysis,
all the density data were tested for their fit to normal
distribution and for homogeneity of variance and were
transformed using the natural log transformation for
population data [Ln (X + 1)]. Adult shrimp were sorted
into size classes based on carapace length: Atya lanipes
in five 5.0-mm-size classes and Xiphocaris elongata in six
3.0-mm-size classes. Percentages of breeding (gravid)
females of both species in the LRF and HRF months
were plotted as annual histograms per pool during the
10-year period. Two-way ANOVAs were performed to
determine whether there were significant differences in
the mean cephalothorax length for each species per pool
(lower, middle and upper altitudes) and year (1997–
2007), and to determine variation in shrimp densities
across time and rainfall periods (low rainfall – LRF; and
high rainfall – HRF) per pool. Linear regressions (Mean
CL versus Time) for each species were performed to
compare the growth rates of the shrimp among pools;
the slopes were compared using analyses of covariance
(ANCOVA).