|
<br />;.
<br />
<br />/~
<br />
<br />DOUGLAS AND MARSH-GRAND CANYON CATOSTOMIDS
<br />
<br />contained fish captured by ASU personnel but
<br />already PIT-tagged; PIT-tag implanted by any of
<br />several agencies or research groups, including
<br />ASU. The third contained fish tagged previously
<br />by other researchers with either a Carlin or floy-
<br />tag, and subsequently PIT-tagged by ASU per-
<br />sonnel; old tags were removed and retained. For
<br />purposes of this report, fishes in all three cap-
<br />ture groups start their capture histories when
<br />first handled by ASU personnel. Therefore, fish
<br />previously PIT-tagged by other groups or agen-
<br />cies ("recaptured") were considered "tagged"
<br />by ASU personnel at recapture.
<br />All three categories were merged and sorted
<br />for a given species by PIT-tag number. Individ-
<br />uals then were condensed into a capture-history
<br />(CH) matrix (Burnham et aI., 1987), where
<br />each individual (i.e., each unique PIT-tag) com-
<br />prised a single row and each of the 49 sampling
<br />periods a column. If an individual was captured
<br />(or recaptured) during a given sampling peri-
<br />od, that respective column was scored "1," oth-
<br />erwise "0." Thus, an individual's capture and
<br />all subsequent recaptures were represented as a
<br />row vector in the CH-matrix.
<br />The CH-matrix was sorted two different ways
<br />for analysis. Capture histories were first com-
<br />piled by season and year (where winter = Dec.,
<br />Jan., Feb.; spring = March, April, May; summer
<br />= June, July, Aug.; and autumn = Sept., Oct.,
<br />and Nov.). In all, 16 seasons were represented
<br />(four each over four years). Individual C. lati-
<br />pinnis (~ 150 mm TL) were also compiled into
<br />nine 50 mm (TL) size classes.
<br />
<br />Population estimates.-Cormack:Jolly-Seber (CJS)
<br />population estimates were generated by trip (n
<br />= 49), season (n = 16; n = 4), and year (n =
<br />4). Trips were occasionally combined due to
<br />lack of recaptures. These were grouped with the
<br />trip immediately preceeding or following so as
<br />to maintain continuity by season; trips that were
<br />grouped received the same estimate. For X. tex-
<br />anus and hybrids, CJS estimates were generated
<br />for 26 of 49 trips, due to zero sample sizes for
<br />the remaining periods (Appendix). For C. lati-
<br />Pinnis, only four trips lacked recaptures.
<br />Sampling effort and population estimates
<br />were transformed to common logarithms. AN-
<br />CaVA was used to test for significant differ-
<br />ences among and between seasons, and be-
<br />tween years, with effort as covariate. Because of
<br />the manner in which trips were partitioned into
<br />seasons, and the staggered initiation and com-
<br />pletion of the project (i.e., July 1991-June
<br />1995), only years 1992, 1993, and 1994 could be
<br />tested in pairwise comparisons. The rationale
<br />and prerequisites for ANCOVA were discussed
<br />
<br />917
<br />
<br />in Douglas and Marsh (1996). Population esti-
<br />mates, standard deviations, and 95% confidence
<br />limits were generated for the entire LCR rather
<br />than by river reach (as defined in Douglas and
<br />Marsh, 1996). Open estimates (via POPAN-4 for
<br />Windows, A. N. Amason, L. Baniuk, C. ].
<br />Schwarz, and G. Boyer, Dept. Computer Sci.,
<br />Univ. Manitoba, Canada, 1995, unpubI.) were
<br />used exclusively, in that demographic closure
<br />was precluded due to the temporal span over
<br />which sampling was conducted.
<br />Annual survival probabilities (adjusted for ef-
<br />fort) were calculated by size class for C. latiPin-
<br />nis, again using POPAN-4. Probabilities could
<br />not be calculated for 1995 in that capture/re-
<br />capture data from 1996 were required. Survival
<br />probabilities were not calculated for X. texanus
<br />and hybrids due to low sample sizes.
<br />
<br />Hybrid identification.- The presence of a nuchal
<br />razor was used as a criterion for designating an
<br />individual as hybrid. Of the 41 individuals so
<br />identified, 12 were examined genetically via re-
<br />striction endonuclease analysis of mtDNA
<br />(Dowling et aI., 1996a:545). Nine of these were
<br />also assayed electrophoretically at five diagnos-
<br />able loci scorable from muscle tissue (D. G.
<br />Buth, pers. comm.).
<br />
<br />RESULTS
<br />
<br />Catostomus latipinnis.-We captured and PIT-
<br />tagged 2179 C. latipinnis, recaptured 1550, and
<br />replaced carlin/floy tags with PIT-tags on anoth-
<br />er 10. These 3739 individuals were collapsed
<br />into a CH-matrix of 2578 entries (where each
<br />unique PIT-tag was represented but once). In-
<br />dividuals often were recaptured several times
<br />during the study (maximum = 12). Cormack-
<br />Jolly-Seber population estimates (adjusted for
<br />effort over all 49 trips; Appendix) indicated
<br />greatest abundance in early summer 1994 [5214
<br />individuals, trips 36 (::t 587)and 37 (::t 575)].
<br />Smallest estimates occurred in winter 1992/
<br />spring 1993 (1591 individuals, trips 20 (::t 260)
<br />and 21 (::t 265); Fig. 1].
<br />Captures varied by season (Appendix; Table
<br />1), by size class (Table 2), and by effort [total
<br />effort = 401,367 net-hours; trip mean = 8191
<br />net-hours (::t 500)]. Interaction between popu-
<br />lation estimates and sampling effort was not sig-
<br />nificant across seasons. A standard ANCOVA in-
<br />dicated population size of C. latipinnis varied
<br />significantly over the 16 seasons [F = 2.44; df =
<br />16; P < 0.01; Proc. GLM ANCOVA, Statistical
<br />Analysis Systems (SAS, verso 6.08), Cary, NC,
<br />1989, unpubI.; Fig. 2].
<br />Pairwise linear contrasts of population esti-
<br />
|