|
<br />I:
<br />
<br />MARKING TECHNIQUES FOR COLORADO SQUA WFlSH
<br />
<br />~
<br />
<br />survival was estimated by keeping marked fish
<br />overnight (18-19 h) in the net-pen.
<br />To determine the extent of movement of marked
<br />fish out of the study area, we sampled 11.3 and
<br />16.1 km downstream of the study reach 20 October
<br />1992 and 20-21 April 1993, respectively.
<br />Parameter estimates.-Movement was mea-
<br />sured as the probability that a fish will move from
<br />a 1.6-km stream section to an adjacent section per
<br />day, determined by the method outlined in Hilborn
<br />(1990). Briefly, this method uses space and time
<br />strata of tag releases, recaptures, and seining effort
<br />to make maximum-likelihood estimates of several
<br />parameters, including probability of movement
<br />from one stream section to another. We selected
<br />the closed mark-recapture models of Otis et al.
<br />(1978) as the most appropriate for our study. These
<br />models give maximum-likelihood estimates of
<br />population size and probability of capture. The
<br />computer program CAPTURE (White et al. 1982)
<br />was used to make the population estimates. Catch
<br />per unit effort was defined as the catch of young
<br />Colorado squawfish per seine haul; the backwater
<br />was the primary sampling unit. Survival variance
<br />was calculated by the delta method for propagation
<br />of errors (Robson and Spangler 1978).
<br />Simulations.-Monte Carlo simulations were
<br />used to examine the effects that alternative sam-
<br />pling regimes, small sample bias, and model as-
<br />sumption failures had on population estimates. The
<br />simulation model was modified from one used by
<br />Hilborn (1990) for studying fish movement pat-
<br />terns from tag recoveries. Our population dynam-
<br />ics model for marked and unmarked fish was
<br />
<br />Nm,t+l = Nm,t + Ru,t - Em,t
<br />
<br />and
<br />
<br />Nu,t+I = Nu,t - Ru,t + It - Eu,t,
<br />
<br />and our observation model was
<br />Rm,t = Nm.t'pt
<br />
<br />and
<br />
<br />Ru.t = Nu,t'pt;
<br />
<br />N m, t number of marked fish in the popu-
<br />lation on day t,
<br />Nu,t number of unmarked fish in, the pop-
<br />ulation on day t,
<br />Rm, t number of marked fish observed on
<br />day t,
<br />Ru. t number of unmarked fish observed on
<br />day t,
<br />
<br />907
<br />
<br />Rm, t predicted number of marked fish ob-
<br />served on day t,
<br />Ru, t predicted number of unmarked fish
<br />observed on day t,
<br />Em, t number of marked fish emigrating on
<br />day t,
<br />Eu, t number of unmarked fish emigrating
<br />on day t,
<br />Em, t mean number of marked fish emi-
<br />grating on day t,
<br />Eu. t mean number of unmarked fish emi-
<br />grating on day t,
<br />I t number of unmarked immigrants on
<br />day t, and
<br />P t probability of capture on day t.
<br />
<br />The model assumed that unmarked fish collected
<br />on day t (Ru,t) were then all marked and released
<br />on that day and that all immigrants (II) Were un-
<br />marked fish that drifted downstream into the sam-
<br />ple section. Daily immigration and emigration (Et)
<br />were generated from uniform distributions 0 < It
<br />< 2. j and 0 < Et < 2. E. The number of emigrants
<br />was divided between marked (Em.t) and unmarked
<br />(Eu,t) in the proportion they occurred in th~ pop-
<br />ulation.
<br />The numbers of observed marked (Rm,t) and un-
<br />marked fish (Ru,t) were generated from Poisson
<br />distributions with means Rm,t and Ru,t . The prob-
<br />ability of capture on sample day t (Pt) was gen-
<br />erated from a uniform distribution a < PI < b,
<br />where a and b varied among simulations. Under-
<br />lying values for population size (Nm,t), number of
<br />sample passes, range of probability of capture (a,
<br />b), and rates of immigration (1) and emigration (E)
<br />were specified for each simulation run.
<br />
<br />Results
<br />
<br />Mark Longevity
<br />
<br />All three marking procedures had mark retention
<br />rates above 97% for 21 d after marking (Table 1).
<br />However, damage to spinal columns and internal
<br />organs during inoculation caused higher mortality
<br />for fish marked with tattoo ink than fish marked
<br />with the other two procedures. After 142d, the
<br />elastic polymer had significantly higher ret~ntion.
<br />Fish marked by the three methods showetl sim-
<br />ilar growth (50-86 mm TL) and mortality (range,
<br />12-22%) after 142 d. Much of the mortality was
<br />attributed to the protozoan Ichthyopthirius multi-
<br />flUs; an outbreak began about day 70 and ended
<br />about day 100.
<br />Fish with recognizable elastic polymer marks (N
<br />= 107) were kept an additional 103 d (22 April-
<br />
|