|
<br />204
<br />
<br />~.
<br />
<br />The Southwestern Naturalist
<br />
<br />voL 49, no. 2
<br />
<br />(Hawkins, 1992; Osmundson et a!., 1997). Fur-
<br />thermore, a comparison of ages estimated
<br />from scales with growth of recaptured Colora-
<br />do pikeminnow suggests that scales are not re-
<br />liable for estimating ages for fish older than
<br />age 10 (Osmundson et a!., 1997).
<br />We hypothesized that problems with prior
<br />age estimation in this species might be solved
<br />by using structures other than scales or by us-
<br />ing a combination of methods. Therefore, our
<br />study objective was to compare otoliths, scales,
<br />and vertebrae for estimating age of Colorado
<br />pikeminnow, with the goal of providing a bet-
<br />ter way to determine ages and growth rates of
<br />this endangered species and, perhaps, other
<br />long-lived fishes.
<br />
<br />METHODs-Ages of 48 Colorado pikeminnow were
<br />estimated from specimens captured in the Green
<br />River basin of Colorado and Utah. Half of the fish
<br />were captured and released, and the other half in-
<br />cluded fresh specimens that were unintentionally
<br />killed by sampling procedures and frozen specimens
<br />captured by others. Ages were estimated from count-
<br />ing putative annuli in scales, otoliths, and vertebral
<br />centra. Total length (average TL = 481 mm, Sf) =
<br />98.5, range = 275 to 745 mm) and one or more of
<br />these body structures were obtained from each of
<br />the fish. We took scale samples from all fish and re-
<br />trieved one or more otoliths (lapillus and sagitta; n
<br />= 55) and 2 or more vertebral centra (n = 24) from
<br />fish that were victims of sampling mortality.
<br />Several scales were removed from each individual
<br />at a location midway between the lateral line and the
<br />insertion of the dorsal fin. Scales were cleaned with
<br />soap and water to remove dirt and epidermis, and
<br />annuli were counted by at least 2 readers using a
<br />hand lens (lOX) and a variable power stereomicro-
<br />scope (12 to lOOX). The modal average age esti-
<br />mated from the sample of scales was used as the final
<br />age estimation. Two or more of the larger, more an-
<br />terior vertebral centra were retrieved, cleaned, and
<br />examined as above (methods reviewed by Cailliet et
<br />aI., 1986). When accessible, 2 otoliths were collected
<br />from each individual and ages estimated indepen-
<br />dently using 2 techniques. In the first technique, the
<br />whole otolith was mounted, and observed annuli
<br />were counted by 3 readers using methods described
<br />by McCarthy and Minckley (1987). In the second
<br />technique, the otolith was sectioned through its nu-
<br />cleus, and annuli visible in the thin cross-section
<br />were counted as described by Schultz and Taylor
<br />(l987). Age determination was "blind" (i.e., sizes of
<br />fishes from which otoliths were removed were not
<br />provided to readers).
<br />In some instances, annuli were so closely com-
<br />
<br />pacted at the edge of a structure that a last annulus
<br />could not be accurately distinguished. In these cases,
<br />age was estimated as greater than the apparent num-
<br />ber of annuli (i.e., >n) and the estimated age was
<br />recorded as n + 1.
<br />For age validation of first few years of growth, oto-
<br />liths were removed from 13 Colorado pikeminnow
<br />reared at Dexter National fish Hatchery, Dexter,
<br />New Mexico (age class 0, n = 3 fish; age 1, n = 3;
<br />age 2, n = 2; age 4, n = 4; and age 5, n = 1), and
<br />their known ages were compared with ages deter-
<br />mined from whole otoliths. All other Colorado pi-
<br />keminnow we studied were wild fish, and their exact
<br />ages were unknown. However, wild Colorado pike-
<br />minnow were captured in the upper Colorado River
<br />basin, where scale ages have been validated (after
<br />correction for missing annuli) for ages up to 10
<br />years (Hawkins, 1992).
<br />Fish TL is often used as an estimator for growth,
<br />and Colorado pikeminnow age is closely related to
<br />TL (Vanicek, 1967; Vanicek and Kramer, 1969; Haw-
<br />kins, 1992; Osmundson et aI., 1997). Thus, we as-
<br />sumed that a significant relationship would exist be-
<br />tween fish age and total length over a random sam-
<br />ple of individuals. We used linear regression to eval-
<br />uate how well fish ages estimated by each of 4 aging
<br />techniques explained the variance in the TL of in-
<br />dividuals. The more confidence given to a fitted re-
<br />gression line (interpreted from a higher Coefficient
<br />of Determination, ,2) indicated that the structure
<br />would more precisely estimate age. After obtaining
<br />the most precise aging structure based on our cri-
<br />teria, the other techniques were compared to it us-
<br />ing structures obtained from the same fish. The Sign
<br />Test (Conover, 1999) was used to evaluate whether
<br />2 structures tended to give equivalent average ages.
<br />
<br />RESULTs-Age estimates obtained from the 4
<br />structures were variously correlated with TL
<br />(Fig. 1). Age estimates from vertebrae had the
<br />highest r- value (0.7361; n = 24; average age
<br />= 7.5 years, SD = 4.70, range = 2 to 13). Aging
<br />by otolith sections produced an r- value of
<br />0.6102 (n = 21; average age = 6.95, SD = 2.82,
<br />range = 4 to 14), and aging by scales produced
<br />an r- value of 0.5848 (n = 48; average age =
<br />7.02, SD = 2.37, range = 3 to 13). Aging with
<br />whole otoliths produced the lowest r- value
<br />(0.329; n = 34; average age = 7.76, SD = 2.75:
<br />range = 4 to 16).
<br />Ages estimated from vertebrae were strongly
<br />correlated with sectioned otoliths estimates for
<br />21 fish (r- = 0.861), and the 2 structures pro-
<br />duced similar average ages (neither of the 2
<br />produced ages larger than the other; Sign Test,
<br />P < 0.05). However, there was less correlation
<br />
<br />.
<br />
<br />..
<br />
|