Recovery of Gila Trout Descended from South Diamond Creek from Recently Hybridized Populations in the Mogollon Creek Drainage

Robb F. Leary

Fred W. Allendorf

and

Naohisa Kanda

Wild Trout and Salmon Genetics Laboratory Report 99/1 Division of Biological Sciences The University of Montana Missoula, Montana 59812 Abstract: Gila trout, Oncorhynchus gilae, were extirpated from South Diamond Creek in 1995 by a fire and rain event. Prior to this, Gila trout from South Diamond Creek were used to establish populations in the Mogollon Creek drainage, but in 1995 it was discovered these populations were composed of Gila trout and descendants of first generation hybrids between Gila and rainbow trout, 0. mykiss. Since South Diamond Creek represents one of only four native Gila trout populations remaining, it was decided to attempt to recover Gila trout from the Mogollon Creek drainage populations.

Fish from Mogollon Creek and its tributary Trail Canyon were transported to the Mescalaro National Fish Hatchery in 1996. In 1997, pair matings (one male and one female) were performed using the sexually mature fish from Mogollon Creek and Trail Canyon. After spawning the fish were sacrificed, and electrophoretic analysis of the products of seven protein coding loci known to distinguish Gila and rainbow trout was used to determine the genetic characteristics of the fish spawned and the likelihood that an individual was a Gila trout. This procedure was repeated using the sexually mature fish in 1998. In 1997, the electrophoretic data indicated that the fish spawned from both streams were mainly a mixture of Gila trout and first generation backcrosses to Gila trout. Thus, progeny produced this year from Gila trout by Gila trout matings represent a valuable resource for conservation and restoration efforts.

Protein data from the fish spawned in 1998 indicated the extent of hybridization had increased. There was evidence of second generation backcrosses to Gila trout among the fish from both streams. Because of this, there was a good chance that some fish that electrophoretically appeared to be Gila trout actually were of hybrid origin. Fish produced this year, therefore, should be excluded from conservation and restoration efforts. Alternatively, each individual's genotype at 40 or more nuclear DNA regions that distinguish Gila and rainbow could be determined to ensure only Gila trout are used.

1 INTRODUCTION

Gila trout, Oncorhynchus gilae, are native to the Gila River drainage in New Mexico and Arizona and the Verde River drainage, Arizona (Miller 1950; Minckley 1973; Behnke 1992). The species is currently listed as endangered under the Endangered Species Act (United States Fish and Wildlife Service 1989). It is believed to be extinct from the Verde River drainage (United States Fish and Wildlife Service 1993) and analysis of samples collected from 1995 through 1997 indicated only four native populations exist in the Gila River drainage: Main Diamond Creek and its replicate McKnight Creek in the Mimbres River drainage, a replicate of South Diamond Creek in the Mogollon Creek drainage, Whiskey Creek, and Spruce Creek and its replicate in Big Dry Creek (Leary and Allendorf 1998).

Reasons for the large reduction in abundance of Gila trout, in numbers of individuals and populations, are due to human alteration of the environment and the introduction of non-native fishes (Miller 1961; United States Fish and Wildlife Service 1993). Human disturbances have resulted in a reduction of vegetation, increased sedimentation and water temperature in streams, and reduced stream flows. Non- native salmonids such as brown trout, SaImo trutta, are thought to result in the elimination of Gila trout populations through predation and competition (Dowling and Childs 1992; United States Fish and Wildlife Service 1993). Introduced rainbow trout, 0. mykiss, have eliminated many Gila trout populations through introgression and the formation of random mating hybrid swarms (Loudenslager et al. 1986; United States Fish and Wildlife Service 1993; Leary and Allendorf 1998; Riddle et al. 1998). That is, populations in which essentially all fish are of hybrid origin.

The above factors have isolated the remaining native Gila trout populations from each other in small headwater streams. This fragmentation jeopardizes their continued existence. The populations can be severely reduced in size or eliminated by natural events such as drought, fire, or flood.

Electrophoretic analysis of proteins indicated that substantial genetic divergence exists among the native Gila trout populations (Leary and Allendorf 1998). This divergence is mainly due to the presence of variant alleles in only one population but at high frequency and the presence of variant alleles in multiple populations but at highly variable frequencies. Preserving the genetic variation represented by the native Gila trout populations, therefore, will require ensuring the continued existence of all of them.

In 1995, Gila trout were extirpated from South Diamond Creek by a fire and rain event. Furthermore, electrophoretic analysis of proteins indicated that the replicate populations of South Diamond Creek Gila trout in the Mogollon Creek drainage were now comprised of Gila trout and descendants of first generation hybrids between Gila and rainbow trout (Leary and Allendorf 1998). Without human intervention, therefore, it

2 was considered highly likely that within a few more generations the Mogollon Creek drainage populations would become hybrid swarms and the valuable genetic resource represented by South Diamond Creek Gila trout would be permanently lost. Thus, a plan was developed and implemented to recover Gila trout from the Mogollon Creek drainage. This paper presents the results of the recovery attempt.

METHODS

In 1996, a combined total of 600 trout were collected using electrofishing from the lower reaches of Mogollon Creek and its tributary Trail Canyon. The fish were transported by helicopter to hatchery trucks and then they were brought to the Mescalaro National Fish Hatchery where they were raised as separate Mogollon and Trail Canyon groups. After electrofishing, the streams were treated with antimycin A to remove all remaining fish for subsequent reintroduction of Gila trout.

Pair matings (one male and one female) were petformed in 1997 at the Mescalaro National Fish Hatchery using the sexually mature fish from the Mogollon and Trail Canyon groups. Each family egg lot was incubated separately and all parents were sacrificed for electrophoretic analysis. Labelled eye, liver, and muscle tissue from each parent was sent frozen to the Museum of Southwestern Biology, University of New Mexico, and then to the Wild Trout and Salmon Genetics Laboratory, University of Montana, for electrophoretic analysis. After electrophoretic analysis, only those egg lots that appeared to be produced from Gila trout by Gila trout matings were retained. The remaining egg lots were sacrificed. This procedure was repeated in 1998. Since parents were sacrificed after spawning, in terms of individuals the fish spawned in 1997 and 1998 are completely different groups.

Electrophoresis and data analysis

Horizontal starch gel electrophoresis was used to determine each fish's genetic characteristics (genotypes) at seven loci (genes) coding for proteins present in muscle, liver, or eye tissue that previous results indicated distinguish Gila and rainbow trout (Leary and Allendorf 1998): alcohol dehydrogenase (ADH*, liver), fumarate hydratase (FH-1*, liver), glyceraldehyde-3-phosphate dehydrogenase (GAPDH-4*, eye), lactate dehydrogenase (LDH-C*, eye), phosphoglucomutase (PGM-1*, muscle), phosphoglycerate kinase (PGK-2*, muscle), and tripeptide aminopeptidase (PEPB*, eye). Electorphoretic procedures were identical to those used by Leary and Allendorf (1998).

We used three methods to examine the extent of hybridization in the fish spawned. First, the proportion of Gila and rainbow trout genes in the Trail Canyon and Mogollon Creek fish, treated separately, was estimated by averaging the frequency of Gila and rainbow trout alleles over all seven loci. Contingency table chi-square analysis

3 was used to test for heterogeneity of the proportion of Gila and rainbow trout alleles between the fish spawned in 1997 and 1998 from the same stream.

The procedure of Cockerham and Weir (1977) was used to test for a positive association between the presence of a Gila trout allele at one locus and another (gametic phase disequlibrium) for all possible pairs of loci in the fish spawned treated separately by stream and year. The amount of gametic phase disequilibrium would provide an insight into how recently in the past hybridization was initiated, and thus the likelihood that some fish spawned may be Gila trout. In Gila trout, rainbow trout, and

first generation hybrids (Gila x rainbow trout, F1), gametic phase disequilibrium relative to its maximum value will have a value of one. With random mating, it will decrease to

0.5 in second generation hybrids (F1 x F1 = F2) and first generation backcrosses (Gila trout x F : rainbow trout x F Each additional generation of hybridization 1 = F1BG 1 = F1BR)• will decrease gametic phase disequilibrium by one-half its former value. Thus, in third

generation hybrids (F2 x F2 = F3) and second generation. backcrosses (Gila trout x F18G = F2BG rainbow trout x FiBR = F28R) it will be 0.25, in fourth generation hybrids (F3 x F3 = F4) and third generation backcrosses (Gila trout x F28G = F_ 38G:, rainbow_ _ trouttrout_ _ x F 28R

F3BR) 0.125, etc.

A hybrid index was calculated for each fish spawned. At each locus analyzed, a Gila trout allele was given a value of zero and a rainbow trout allele a value of one. These values summed over the two alleles possessed by an individual at a locus yields the hybrid index for the locus. These values summed over all seven loci yields the individual's hybrid index.

The distribution of hybrid index scores in the fish spawned treated separately by stream and year would provide insight into how broadly rainbow trout genes were distributed among the fish and how recently in the past hybridization was initiated. This information would allow some assessment of the likelihood a particular fish was a Gila trout. Gila trout will have a hybrid index of zero, rainbow trout a hybrid index of 14, and

Fl., a hybrid index of seven. F2., will have a hybrid index binomially distributed about a value of seven, F18G., values binomially distributed about 3.5 (in reality 3 and 4 since individuals have integer values), and F2BG.s values binomially distributed about 1.75 (2). Thus, the distribution of hybrid index values for a group of fish composed of Gila trout,

Fl., and F18G.s may be trimodal with modes at zero, 3 or 4, and seven. In contrast, the distribution of hybrid index values for a group of fish from a hybrid swarm will be binomially distributed around an integer approximately equal to 14 times the proportion of rainbow trout genes in the population. Multimodal hybrid index distributions, therefore, would be indicative of fairly recent hybridization. In such situations, a mode at zero would indicate the population still contained Gila trout.

RESULTS

4 Comparisons between years within streams

The first issue to address is whether there is any evidence of differences in the extent of hybridization between the fish spawned in 1997 and 1998 from the same stream. If so, it would not be appropriate to combine the two groups into one for further analysis.

Contingency table chi-square analysis indicates that there is a highly significant 2 difference (X1 = 17.742; P <0.001) between the average proportion of rainbow trout alleles and Gila trout alleles between the 1997 and 1998 Trail Canyon parents. The fish spawned in 1998 possessed, on the average, about twice the proportion of rainbow trout alleles than those spawned in 1997 (Table 1).

There did not appear to be much gametic phase disequilibrium in either the 1997 or 1998 Trail Canyon parents. Only four of 21 comparisons were statistically significant in 1997 and only one comparison was significant in 1998 (Table 2). This suggests that there was less gametic phase disequilibrium in the 1998 than 1997 parents. This conclusion, however, is tentative based solely on these data because with the small number of parents each year only very strong gametic phase disequilibrium would be statistically detectable.

Stronger evidence for less gametic phase disequilibrium in the 1998 parents comes from a comparison of values between the years. The average value for all 21 comparisons in 1997 was 0.416 and only 0.210 in 1998. Furthermore, the sign of the difference between the 1997 and 1998 values for each pair of loci is positive in 15 of the 21 cases (sign test, P = 0.040). Thus, we conclude there was less gametic phase disequilibrium in the 1998 parents indicating, on the average, a longer history of hybridization.

The distribution of hybrid index scores in the two groups of parents supports the above conclusions. The 1998 parents had a significantly (Wilcoxon two-sample test, P <0.001) higher mean hybrid index (4.00) than the 1997 parents (1.69). This is partially a consequence of the 1998 parents possessing a higher average proportion of rainbow trout alleles. It also reflects the fact that the rainbow trout alleles were possessed by a higher proportion of the fish spawned in 1998 than in 1997 (Fig. 1) suggesting again a longer history of hybridization for the 1998 parents.

Similar results were obtained for the two groups of Mogollon Creek parents. The 1998 parents possessed, on the average, about twice the proportion of rainbow trout alleles as the 1997 parents (Table 3; X ,2 = 48.901, P < 0.001). There was strong gametic phase disequilibrium in both groups of parents, but it again appeared to be stronger in the 1997 parents. All comparisons were statistically significant in the 1997 parents and 19 of the 21 were significant in the 1998 parents (Table 4). The average

5 value for all 1997 comparisons (0.676) was higher than that for the 1998 comparisons (0.395) and the difference between the 1997 and 1998 values for each pair of loci is positive in 19 of the 21 cases (p = 0.0001). The higher proportion of rainbow trout alleles and the longer history of hybridization in the 1998 parents is again reflected by a higher mean hybrid index (1997 = 0.66, 1998 = 1.69; P <0.001) and a higher proportion of fish possessing rainbow trout alleles (Fig. 2).

Likelihood of individuals being Gila trout

The next issue to address is to qualitatively assess, using the distribution of hybrid index values, the likelihood that individuals with a hybrid index of zero are Gila trout. Since the above results clearly indicate the extent of hybridization was greater in the 1998 than 1997 parents from both streams, this was done for all four groups of parents.

The distribution of hybrid index values was markedly bimodal among the 1997 Trail Canyon parents (Fig. 1). About half the fish had values of zero, characteristic of Gila trout, and the other fish had positive values centered around four, characteristic of first generation backcrosses to Gila trout. All the fish in the latter group were homozygous at some loci for Gila trout alleles and heterozygous at other loci, again characteristic of first generation backcrosses to Gila trout. Thus, the 1997 Trail Canyon parents appear to largely have been a mixture of Gila trout and first generation backcrosses to Gila trout. The chances that an individual in the latter group would have a hybrid index of zero is (0.5)7 = 0.008. The chances that none out of seven (one more than the number of fish definitely of hybrid origin detected) first generation backcrosses would have a hybrid index of zero, therefore, is (1 - 0.008)7 = 0.945. Thus, it is very likely that all 1997 Trail Canyon parents with a zero hybrid index were Gila trout.

The hybrid index values also divided the 1998 Trail Canyon parents into two groups (Fig. 1). The most abundant group had a modal value of three and was skewed to seven. All but three fish in this group were homozygous at some loci for Gila trout alleles and heterozygous at the other loci indicating they were at least first generation backcrosses to Gila trout. Two of the other three fish in this group were homozygous for Gila trout alleles at some loci, homozygous for rainbow trout alleles at other loci, and heterozygous at the remaining loci indicting them to be at least second generation hybrids. The final fish was heterozygous at all the loci. It is not likely (0.008) that such a fish would be a first generation backcross to Gila trout or a second generation hybrid. Thus, this fish was most likely a first generation hybrid.

The second group of 1998 Trail Canyon parents was composed of only two fish. One fish had a hybrid index characteristic of Gila trout and the other a value most characteristic of at least a second generation backcross to Gila trout. In this situation, whether or not an individual with a zero hybrid index is a Gila trout becomes more

6 problematic. The chances of a second generation backcross to Gila trout having a zero value is 0.133 and thus two such fish a nonzero value (1 - 0.133)2 or 0.751. Thus, there is a reasonable chance that none of the 1998 Trail Canyon parents were Gila trout.

The distribution of hybrid index values is somewhat bimodal among the 1997 Mogollon Creek parents (Fig. 2). The mode at four suggests that most fish definitely of hybrid origin were probably first generation backcrosses to Gila trout. This is supported by the individuals multiple locus genotypes. All but two were homozygous for Gila trout alleles at some loci and heterozygous at the others. The other two fish were heterozygous at all loci suggesting they were most likely first generation hybrids. The chances of 13 first generation backcrosses all having nonzero hybrid index values is 0.909. Thus, the 1997 Mogollon Creek parents appear to have been a mixture of Gila trout, first generation hybrids, and first generation backcrosses to Gila trout.

The distribution of hybrid index values is trimodal among the 1998 Mogollon Creek parents (Fig. 2). The mode at four suggests again the presence of first generation backcrosses to Gila trout. The mode at two suggests the presence of second generation backcrosses to Gila trout. The mode at zero suggests the presence of Gila trout. The multiple locus genotypes of the fish definitely of hybrid origin supports the above statements. All fish but one were homozygous for Gila trout alleles at some loci and heterozygous at the others. The remaining fish appeared to be at least a second generation hybrid as it was homozygous for Gila trout alleles at three loci, homozygous for rainbow trout alleles at two loci, and heterozygous at two loci. Thus, the 1998 Mogollon Creek parents appeared to be a mixture of Gila trout, first and second generation backcrosses to Gila trout, and a small proportion of second generation hybrids. The existence of second generation backcrosses again makes the interpretation of hybrid index values of zero problematic. The distribution of hybrid index values suggests about an equal proportion of first and second generation backcrosses. The chances of having 53 second generation backcrosses all having nonzero hybrid index values is 0.0005. This strongly indicates that some of the 1998 Mogollon Creek parents with zero hybrid index values are actually of hybrid origin.

DISCUSSION

The extent of hybridization was markedly different between the 1997 and 1998 Mogollon Creek and Trail Canyon parents. The proportion of rainbow genes was about twice as great among the 1998 parents than the 1997 parents. The most likely explanation for this is an additional infusion of rainbow trout genes into both populations. This could be the result of multiple rainbow trout introductions into the streams, an average longer life span for rainbow trout and a higher proportion of multiple spawning fish, or both factors.

There were additional differences between the 1997 and 1998 parents. In 1997,

7 both groups of parents appeared to be mainly a mixture of Gila trout and first generation backcrosses to Gila trout. Thus, in both groups of parents there was a high probability that all fish with a hybrid index of zero were Gila trout (Mogollon = 0.91; Trail Canyon = 0.95). Progeny from crosses between two such fish, therefore, would have high conservation and restoration value.

In contrast, the lower amount of gametic phase disequilibrium, higher proportion of fish definitely of hybrid origin, higher mean hybrid index, and the distribution of hybrid index values indicated a longer history of hybridization for the 1998 parents. Some of the fish appeared to be at least second generation backcrosses to Gila trout. Thus, there was a reasonable chance (0.25) that the 1998 Trail Canyon fish with a zero hybrid index actually was of hybrid origin and a very good chance (0.9995) that some of the 1998 Mogollon Creek fish with a hybrid index value of zero were of hybrid origin. It is very likely, therefore, that some of the apparent Gila by Gila trout matings in 1998 involved some second generation backcrosses. The progeny of these matings would very likely be third generation backcrosses.

Since retaining progeny from the apparent Gila by Gila trout 1998 matings almost undoubtedly would entail retaining a small percentage of third generation backcrosses, the question becomes what to do with these fish. The simplest step would be to exclude them from use in conservation and restoration efforts. If this step is taken, however, it needs to be recognized that this will essentially constitute destroying many Gila trout.

An alternative would be to obtain a fin clip and individually mark a number of fish. The fin clip could then be used to deduce each fish's genotype at a large number of nuclear DNA regions that distinguish Gila and rainbow trout. Those fish that appear to be Gila trout could then be used for conservation and restoration efforts.

Although attractive, it needs to be recognized the proposed alternative to discarding all the fish would be extremely costly. Third generation backcrosses will, on the average, be heterozygous at only one-eighth of the loci analyzed. Thus, to be 97 percent certain an individual is not a hybrid 40 loci per individual will have to be analyzed. With this number of loci, the probability that a third generation backcross will be homozygous for Gila trout alleles at all loci is (0.5)5 = 0.031. As the number of potential hybrids analyzed increases, however, the number of loci needed to be analyzed per individual to be reasonably certain all hybrids are identified also increases. For example, if five percent of the fish were hybrids and 1,000 fish are analyzed then one would have only a (1 - 0.03)5° = 0.218 chance of detecting all 50 hybrids. In contrast, if 72 loci per individual were analyzed, then one would have a 50 (1 - (0.5)9) = 0.907 chance of detecting all 50 hybrids.

Finally, there is an option to spawn more fish from Trail Canyon and Mogollon

8 Creek being held at the Mescalaro National Fish Hatchery in 1999. Given the differences observed between the 1997 and 1998 parents, however, we expect this would probably not be a very fruitful effort. It is very likely that the extent of hybridization in the potential 1999 parents would be equal to or greater than that in the 1998 parents. If it is deemed desirable, this possibility can be tested by obtaining fin clips and marking the fish. Again the fin clips could be used to determine each fish's genotype at a number of nuclear DNA regions that distinguish Gila and rainbow trout and estimate the extent of hybridization in the potential parents prior to the spawning season.

ACKNOWLEDGMENTS

Financial support for this work was provided by the United States Fish and Wildlife Service. Tissue samples were supplied by A.M. Snyder of the University of New Mexico.

LITERATURE CITED

Behnke, R.J. 1992. Native Trout of Western North America. American Fisheries Society Monograph 6, American Fisheries Society, Bethesda, Maryland.

Cockerham, C.C., and B.S. Weir. 1977. Digenic descent measures for finite populations. Genetical Research 30:121-127.

Dowling, T.E., and M.R. Childs. 1992. Impact of hybridization on a threatened trout of the southwestern United States. Conservation Biology 6:355-364.

Leary, R.F., and F.W. Allendorf. 1998. Genetic issues in the conservation and restoration of the endangered Gila trout. University of Montana Wild Trout and Salmon Genetics Laboratory Report 98/1.

Loudenslager, E.J., J.N. Rinne, G.A.E. Gall, and R.E. David. 1986. Biochemical genetic studies of native Arizona and New Mexico trout. The Southwestern Naturalist 31:221-234.

Miller, R.R. 1950. Notes on the cutthroat and rainbow trouts with the description of a new species from the Gila River, New Mexico. Occasional Papers of the Museum of Zoology, University of Michigan. University of Michigan Press, Ann Arbor, Michigan 529:143.

Miller, R.R. 1961. Man and the changing fish fauna of the American southwest. Papers of the Michigan Academy of Science, Arts and Letters 46:365-404.

9 Minckley, W.L. 1973. Fishes of Arizona. Arizona Game and Fish Department, Phoenix, Arizona.

Riddle, B.R., D.L. Propst, and T.L. Yates. 1998. Mitochondrial DNA variation in Gila trout, Oncorhynchus gilae: Implications for management of an endangered species. Copeia 1998:31-39.

United States Fish and Wildlife Service. 1989. Endangered and threatened wildlife and plants. Code of the Federal Register. 50, Title 17:1-69.

United States Fish and Wildlife Service. 1993. Gila trout recovery plan. United States Fish and Wildlife Service, New Mexico Ecological Services State Office, Albuquerque, New Mexico.

10 TABLE 1

Allele frequencies at the diagnostic loci between Gila trout and rainbow trout analyzed in fish spawned from Trail Canyon in 1997 and 1998. At each locus, the allele characteristic of Gila trout is listed first.

Sample and allele frequencies

Locus Alleles 1997 1998

ADH* 25 0.923 0.781 -100 0.077 0.219

FH-1* 70 0.846 0.765 100 0.154 0.235

GAPDH-4* 70 0.923 0.618 100 0.077 0.382

LDH-C* 110 0.885 0.588 100 0.115 0.412

PEPB* 135 0.885 0.735 100 0.115 0.265

PGK-2* 90 0.846 0.735 100 0.154 0.265

PGM-1* 133 0.846 0.735 100 0.154 0.265

Average Gila 0.879 0.708

Average rainbow 0.121 0.292

11 TABLE 2

Percentage of maximum possible gametic phase disequilibrium in fish from Trail Canyon spawned at the Mescalaro National Fish Hatchery in 1997 (above diagnol) and 1998 (below diagnol). Statistically significant (P < 0.05) values are in bold.

Loci ADH* FH-1* GAPDH-4* LDH-C* PEPB* PGK-2* PGM-1*

ADH* 24.6 40.6 32.9 32.9 39.3 24.6

FH-1* 10.0 88.6 3.3 88.9 56.6 56.6

GAPDH-4* 18.5 2.5 28.2 32.9 24.6 88.6

LDH-C* 19.4 13.3 45.8 12.6 46.1 3.3

PEPB* 31.1 13.8 21.3 72.0 46.1 46.1

PGK-2* 40.1 13.8 2.3 28.3 12.3 56.6

PGM-1* 31.1 4.3 21.3 31.8 3.8 3.8

12 TABLE 3

Allele frequencies at the diagnostic loci between Gila trout and rainbow trout analyzed in fish spawned from Mogollon Creek in 1997 and 1998. At each locus, the allele characteristic of Gila trout is listed first.

Sample and allele frequencies

Locus Alleles 1997 1998

ADH* 25 0.944 0.893 -100 0.056 0.107

FH-1* 70 0.963 0.853 100 0.037 0.147

GAPDH-4* 70 0.944 0.844 100 0.056 0.156

LDH-C* 110 0.932 0.874 100 0.068 0.126

PEPB* 135 0.944 0.834 100 0.056 0.166

PGK-2* 90 0.969 0.927 100 0.031 0.073

PGM-1* 133 0.969 0.945 100 0.031 0.055

Average Gila 0.952 0.881

Average rainbow 0.048 0.119

13 TABLE 4

Percentage of maximum possible gametic phase disequilibrium in fish from Mogollon Creek spawned at the Mescalaro National Fish Hatchery in 1997 (above diagnol) and 1998 (below diagnol). Statistically significant values (P <0.05) are in bold.

Loci ADH* FH-1* GAPDH-4* LDH-C* PEPB* PGK-1* PGM-1*

ADH* 60.2 71.6 94.1 83.6 74.4 54.9

FH-1* 33.3 77.7 75.9 59.6 34.4 75.9

GAPDH-4* 43.2 37.0 81.2 70.9 52.6 73.4

LDH-C* 43.8 35.6 36.5 93.1 72.7 72.0

PEPB* 48.7 31.9 49.5 51.0 52.6 52.2

PGK-2* 3.3 30.2 49.5 28.6 53.6 37.0

PGM-1* 55.7 25.2 51.5 55.7 54.2 11.8

14 Figure 1. Distribution of hybrid index values among fish from Trail Canyon spawned at the Mescalaro National Fish Hatchery in 1997 and 1998.

Figure 2. Distribution of hybrid index values among fish from Mogollon Creek spawned at the Mescalaro National Fish Hatchery in 1997 and 1998.

15 Trail Canyon

60

50

40

0.- 20

10

0 1 2 4 5 6 7 0 1 2 3 4 5 6 7 Hybrid Index MOGOLLON

90 1997 (N=80) 1998 (N=209)

80 '

70

,A,C• 40

30

20

10

0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Hybrid Index