TAXONOMIC ANALYSIS OF GILA AND GILA x RAINBOW

TROUT IN SOUTHWESTERN NEW MEXICO

BY

ROBERT EARL DAVID, B.S.

A Thesis submitted to the Graduate School

in partial fulfillment of the requirements

for the Degree

Master of Science

Major Subject: Wildlife (Fishery) Science

New Mexico State University

Las Cruces, New Mexico

May 1976 "Taxonomic Analysis of Gila and Gila x Rainbow in Southwestern

New Mexico," a thesis prepared by Robert Earl David in partial ful-

fillment of the requirements for the degree, Master of Science, has

been approved and accepted by the following:

Dean of the Graduate School

Chairman of the Examining Committee

Date

Committee in Charge:

Dr. Douglas B. Jester

Dr. G. Morris Southward

Dr. Bradley A. Blake

Mr. Paul R. Turner

11 ACKNOWLEDGEMENTS

Financial support for this study was contributed by the New Mexico

Department of Game and and the U. S. Fish and Wildlife Service.

I would like to express my appreciation to Dr. Douglas B. Jester,

Associate Professor Fishery Science, for continued support and en- couragement throughout this project.

Special thanks are extended to Mr. Paul Turner, Assistant Professor

Fishery Science, Kim Mello, Gregg Schmitt, Allen Campbell, Becky Campbell, and many undergraduates who assisted in collecting specimens.

Thanks are due Dr. Robert J. Behnke, Colorado State University, for continued advisement on taxonomic problems.

Thanks also go to Dr. John Hubbard and other employees of the New

Mexico Department of Game and Fish whose support made this project possible.

A deep appreciation goes to my wife Kathy for her devotion and encouragement.

111 VITA

September 20, 1950 - Born in Prescott, Arizona

May 1968 - Graduated from Sandia High School, Albuquerque, New Mexico

May 1972 - B.S., Wildlife Science, New Mexico State University, Las

Cruces, New Mexico

1972-1974 - U.S. Army

1974-1976 - Graduate Teaching Assistant, Department of Fishery and

Wildlife Sciences, New Mexico State University

PROFESSIONAL AND HONORARY SOCIETIES

The Wildlife Society

American Fisheries Society

IV' ABSTRACT

TAXONOMIC ANALYSIS OF GILA AND GILA x RAINBOW

TROUT IN SOUTHWESTERN NEW MEXICO

BY

ROBERT EARL DAVID, B.S.

Master of Science in Wildlife (Fishery) Science

New Mexico State University

Las Cruces, New Mexico, 1976

Doctor Douglas B. Jester, Chairman

Samples of trout from 19 streams in the Gila National Forest,

New Mexico, were collected and analyzed to determine taxonomic status.

Of 23 taximetric characters, ten were selected as key distinguishing characteristics.

Main Diamond Creek was selected to represent a pure population of

Gila trout, and Turkey Creek at Corral Canyon was used to represent a pure population of . Student's t-test with pooled variances was used to compare the means of 10 key taximetric characters of an un- known population with means of characteristics from the Main Diamond

Creek population.

Samples found to represent pure Gila trout were from populations in Main Diamond, South Diamond, McKenna, Iron, and Spruce Creeks.

Upper Sheep Corral and McKnight Creeks, containing introduced populations of Gila trout, were not analyzed. Evidence indicates that Spruce Creek, previously believed to be an introduced population of Gila trout, may represent a native population of the San Francisco drainage. Data suggests a hypothesis that the Spruce Creek population represents an intergrade between Gila and Arizona trout.

Tributaries of the San Francisco River that contain hybrid popu- lations are Whitewater, Dry, and Lipsey Creeks. Influence of either

Arizona or is seen in these populations.

Tributaries of the Gila River which were found to contain Gila x rainbow trout hybrids were Black Canyon, Sycamore, White, Langstroth,

Trail Canyon, upper Mogollon, upper Turkey, and West Fork of Mogollon

Creeks. Streams containing pure populations of rainbow trout were lower Sheep Corral Creek and Turkey Creek at Corral Canyon.

vi TABLE OF CONTENTS

Page ...... LIST OF TABLES viii ...... LIST OF FIGURES ix ...... INTRODUCTION 1 ...... ME'tHODS 3 ...... STUDY AREA 9 ...... RESULTS 11 ...... DISCUSSION 21 ...... South Diamond Creek 21 ...... McKenna Creek 23 ...... Iron Creek 24 ...... Spruce Creek 24 ...... Black Canyon Creek 28 ...... Sycamore Creek 28 ...... Whitewater, Lipsey, and Dry Creeks 29 ...... Langstroth, White, and Upper Mogollon Creeks 31

Trail Canyon, West Fork of Mogollon, and Upper Turkey Creeks ...... 32

Lower Sheep Corral Creek ...... • 32

SUMMARY ...... 33

LITERATURE CITED ...... 35

VII LIST OF TABLES

Table Page

1. Measurements of 23 characteristics of Gila trout (Main Diamond Creek) and rainbow trout (Turkey Creek at Corral Canyon), showing sample sizes, ranges, and means (Main Diamond Creek in parentheses) ...... 8

2. Regression and correlation showing proportional increases in length of morphometric characters with increase of standard length of Gila trout in Main Diamond Creek (top) and rainbow trout in Turkey Creek at Corral Canyon (bottom). Required r for Gila trout is 0.735. Required r for rainbow trout is 0.708 ...... 14

3. Means of 10 characteristics from each of 19 populations sampled. Populations exhibiting occasional basibranchial teeth are marked with an asterisk ...... 15

4. Computed t for means of 10 characteristics from each of 18 populations compared to Main Diamond Creek. Degrees of freedom are shown in parentheses. Computed t's which are not significantly different from Main Diamond Creek (Pr < 0.05) are marked with an asterisk . • 18

VIII LIST OF FIGURES

Figure Page

1. Electrofishing on Spruce Creek 4

2. Ten morphometric measurements used in taxonomic analysis ...... 5

3. Construction of sampling grid and location of sampling sites for use in spot density analysis ...... 7

4. Study area in southwestern New Mexico. Sampling sites are indicated by 0 ...... 10

5 Hyoid apparatus of four specimens of Gila trout from Spruce Creek showing location of basibranchial teeth (indicated by "x") ...... 16

6. Means and 95 percent confidence intervals for 10 characteristics from 19 populations of trout ...... 20

7 Gila trout from South Diamond (upper), Main Diamond (middle), and Spruce Creeks (lower), showing typical spotting patterns from each stream ...... 22

8. Trout from Lipsey (upper and middle) and Dry Creeks (lower), showing typical spotting patterns from each stream ...... 30

Ix INTRODUCTION

The Gila trout, gilae Miller 1950, is one of two native

salmonids of New Mexico. Present distribution is restricted to the

headwaters of the Gila and San Francisco Rivers in New Mexico and Gap

Creek, tributary of the Verde River, Arizona. The Arizona population

resulted from a 1974 transplant from Main Diamond Creek, Gila National

Forest, New Mexico.

The Gila trout is listed as an by the United

States Department of the Interior (1973) and the New Mexico Department

of Game and Fish (1975). An Interim Gila Trout Recovery Team was formed

in May, 1974, and is attempting to restore the Gila trout to selected native streams in the Gila National Forest, New Mexico.

Research has been done on both ecology (Regan 1964, 1966, Jester

1965, Hanson 1971, LaBounty and Minkley, 1972) and (Cope and

Yarrow 1875, Miller 1950, 1961, 1972, Huntington 1955, Needham and Gard

1959, Simon 1964, Behnke 1967, 1970a, 1970b, 1973, Legendre and Behnke

1972). The description of the species (Miller 1950) contains information which is no longer valid. The Arizona trout, Salmo apache Miller 1972,

had not been described and populations of Arizona trout were listed as

Gila trout.

Origin of the Gila trout is subject to speculation because an

adequate fossil record has not been discovered. The most popular

hypothesis (Miller 1950) correlates the origin with late Pliocene or early Pleistocene events. At that time, the ocean temperatures were

cold enough to allow a southern migration of Pacific salmonids to enter

the Gulf of California. Further movement of one of these salmonids led

1 it up the Gila River into the headwaters where effects of isolation

and inbreeding produced what is recognized now as the Gila trout.

Prior to this study, Gila trout populations were known to exist

in six streams of the Gila National Forest, New Mexico: Main Diamond

Creek, South Diamond Creek, McKenna Creek, Spruce Creek, and intro-

duced populations in McKnight Creek and Sheep Corral Creek. A seventh

population was discovered in 1975 in the headwaters of Iron Creek, a headwater tributary of the Middle Fork of the Gila River.

Objectives of this study were to determine purity of previously known and recently discovered Gila trout populations and identify streams

containing Gila x rainbow trout hybrids. This thesis is presented to

define more accurately the status of the Gila trout in order to facilitate

restoration or other management. METHODS

Population samples were taken by electrofishing (Fig. 1). Speci- mens were fixed in 10 percent formalin containing three to four ml of

Ionol (Shell Chemical Company) per gallon as a color-preserving anti- oxidant. Fixed specimens were stored in 40 percent isopropyl alcohol.

Each specimen was marked with a numbered opercular strap tag for in- dividual identification. All meristic and morphometric measurements were reviewed with Dr. Robert J. Behnke, Colorado State University, in order to conform with standard measuring techniques in trout taxonomy.

All specimens were analyzed by the author to avoid induced variation.

Therefore, any idiosyncrasies in measurement which may not have been eliminated are consistent for all samples.

Populations were analyzed using the morphometric characteristics shown in Figure 2 and meristic characteristics described below.

Measurements were done with double-pointed dividers and a rule grad- uated in mm. Basibranchial teeth and gill rakers were counted after a

24-hour application of alizarin red stain. Scale counts were aided by removal of the mucosa and application of a one percent solution of mala- chite green stain. Scale counts were done by standard methods and included a lateral series two rows above the lateral line. Scales above the lateral line were counted, starting at the dorsal origin and counting a transverse row ventrally to the first scale above the lateral line.

Pyloric caeca were counted by removal. Counts requiring magnification were done at 10X and 20X under a stereoscopic microscope. Vertebrae were counted from X-rays.

3 4

FIGURE 1. ELECTROFISHING ON SPRUCE CREEK

5

1. ORBIT LENGTH G. DORSAL FIN BASE LENGTH 2, UPPER JAW LENGTH 7, DORSAL FIN DEPRESSED LENGTH 3, HEAD LENGTH 8. ADIPOSE FIN LENGTH 4, BODY DEPTH 9. CAUDAL PEDUNCLE LENGTH 5. PREDORSAL LENGTH 10. CAUDAL PEDUNCLE DEPTH 11. STANDARD LENGTH

7

10

11

FIGURE 2, ELEVEN MORPHOMETRIC NEASURESMENTS USED IN TAXONOMIC ANALYSIS 6

Evaluation of spot density was made with a modification of a method

described by Miller (1972). An acetate overlay was constructed with a

5 x 10-mm window centered on the specimen using a graduated line to

place the window midway between the lateral line and the origin of the

dorsal fin. All spots within the window were counted including those

spots in contact with the top and left borders. Spots touching the

bottom and right hand borders of the window were not counted. A second

count was taken between the insertion of the dorsal fin and lateral

line. Adjoining spots were counted as one spot. All minute, yet definite, 2 "flecks" were counted. Spots from both counts represent spots per cm

for each specimen. Location of spot samples and construction of the

acetate overlay are shown in Figure 3.

All linear measurements were standardized as thousandths of the

standard length, as follows: character (mm) Standardized character - (1,000). standard length (mm) Twenty-three characters are expressed in terms of ranges and means

(Table 1). 7

FIGURE 3, CONSTRUCTION OF SAMPLING GRID AND LOCATION OF SAMPLING SITES FOR USE IN SPOT DENSITY ANALYSIS. 8

TABLE 1. MEASUREMENTS,OF TWENTY-THREE CHAWTERISTICS OF GiLA TROW (MAIN DIAMOND UREEK) AND RAINBOW TROUT (IURKEY UREEK AT LORRAL UANYCN), SHOWING SAMPLE SIZES, RANGES, AND MEANS (MAIN DIAMOND CREEK IN PARENTHESES).

SAMPLE MEASUREMENT SIZES RANGES MEANS

BODY DEPTH (11) 12 (239-311) 228-281 (276.3) 257.1

HEAD LENGTH (11) 12 (291-333) 280-318 (313.3) 292.7

ORBIT LENGTH (11) 12 ( 78- 90) 74- 92 ( 84.7) 84.2

UPPER JAW LENGTH (11) 12 (157-202) 145-177 (178.0) 160.1

PREDCRSAL LENGTH (11) 12 (515-568) 496-541 (545.7) 511.6

DORSAL FIN BASE LENGTH (11) 12 (151-164) 135-164 (154.1) 152.1

DORSAL FIN DEPRESSED LENGTH (11) 12 (238-262) 225-267 (251.7) 247.1

ADIPOSE FIND DEPRESSED LENGTH (11) 12 (101-123) 81-108 (110.0) 91.8

CAUDAL PEDUNCLE DEPTH (11) 12 (109-126) 81-119 (118.5) 106.4

CAUDAL PEDUNCLE LENGTH (11) 12 (138-157) 128-160 (146.6) 145.2

VERTEBRAE (12) 11 ( 60- 62) 70- 63 ( 60.5) 62.0

UPPER GILL RAKERS (11) 12 ( 7- 10) 7- 9 ( 8.2) 7.7

LOWER GILL RAKERS (11) 12 ( 11- 13) 11- 14 ( 11.6) 12.0

TOTAL GILL RAKERS (11) 12 ( 18- 22) 17- 22 ( 19.8) 19.6

RIGHT BRANCHIOSTEGAL RAYS (11) 12 ( 10- 11) 10- 12 ( 10.1) 10.6

LEFT BRANCHIOSTEGAL RAYS (11) 12 ( 10- 11) 10- 12 ( 10.7) 10.8 PECTORAL FIN RAYS (11) 12 ( 13- 14) 14- 15 ( 13.7) 14.1

PELVIC FIN RAYS (11) 12 ( 9- 10) 10 ( 9.6) 10

SCALES ABOVE LATERAL LINE ( 9) 11 ( 29- 34) 27- 32 ( 31.5) 29.4

LATERAL SERIES ( 9) 11 (143-160) 128-138 (149.4) 134.1

PYLORIC CAECA (11) 12 ( 29- 37) 42- 58 ( 32.5) 48.3

DENTITICN (11) 12 ( 0 ) 0 ( 0 ) 0

SPOTS PER SQUARE CENTIMETER (11) 12 ( 13- 29) 12- 20 ( 22.2) 15.6 STUDY AREA

Samples of trout from 19 streams in the Gila National Forest were used in this study (Figure 4). The Gila National Forest, located in southwestern New Mexico, is the largest National Forest in the con- tinental United States. It encompasses 3.3 million acres of forest and semi-arid rangeland. Within its boundaries is the , the world's first, designated as such in 1924. The Continental Divide winds through 170 miles of the forest which includes the Mogollon, Tularosa,

Diablo, and Black Range Mountains. Elevations range from 4,500 feet to almost 11,000 feet MSL, with the highest point occurring on Whitewater

Baldy peak. Six of Merriam's seven Life Zones from Upper Sonoran to

Arctic-alpine are represented in this region (Bailey 1913). Three major rivers drain the area. The San Francisco and Gila Rivers are west of the Continental divide and the Mimbres River drains into a closed basin to the east (U.S. Forest Service 1974).

9 ••LACK _ r•

Will WATER WILLOW • REEK W1.14jirtj PEAK GROUSEPEAK • • ENTER RAINYPEAK Tk r VIA DIA ELKND SAN F RI ERNCISCO GILA LAVW RIVER

DRY CREEK

JCOPERAS NEW MEXICO ( PEAK

••■•• LAKE ROBERTS L _ _ _ GILA • 1 CORALS EP 4111 WILDERNESS EK RIVERILA 0 1 2 3 4 5 MILES GILA WILDERNESS BOUNDRY

FIGUT Lk STUDY AFEA IN SWIMS-FERN NEW NEXICO, SANPLING SITES APE INDICATED BYO, RESULTS

When attempting to distinguish Gila trout from introduced rainbow trout and Gila x rainbow hybrids, standards must be set to describe characteristics representing these taxa. When considering characteristics of the Gila trout, it is necessary to accept the Main Diamond Creek popu- lation as a pure strain because the holotypic series from which the species was described is from that location and no subsequent data have invalidated the assumption of purity. Therefore, measurements repre- senting pure Gila trout are from a collection taken November 1, 1974, from Main Diamond Creek (Table 1).

Selection of a population to represent pure rainbow trout was more difficult. Mass production, selective breeding, and mixing of strains in hatcheries surely have modified characteristics of this species to some degree. It is probable that more than one strain of rainbow trout was introduced into waters in the Gila National Forest in stocking programs, and it is impossible to determine sources of rainbow trout which produced the hybrid populations present today. It is also possible that characteristics of a species may be altered by environmental factors after introduction into an area from hatchery conditions. A reproducing population of rainbow trout from Turkey Creek at the confluence with

Corral Canyon (Figure 4) was selected to represent a pure population be- cause of general rainbow appearance and absence of characteristics which might suggest Gila or cutthroat trout influence. The first recorded 1 stocking of rainbow trout into Turkey Creek was in 1934 . Twelve

'Stocking records of New Mexico Dept. Game and Fish, compiled in 1972.

11 12 specimens were collected from the Turkey Creek population at Corral Canyon, two of which exhibited latent fin abrasion scars on dorsal and pelvic fins.

Fin abrasion and regeneration indicate that these specimens probably were hatchery stock. Measurements from this population are very similar to those listed by Behnke (1970) and Miller (1972) in describing typical rain- bow trout, confirming our judgement in selecting this population for a standard. Ranges and means for characteristics of population samples representing Gila trout from Main Diamond Creek and rainbow trout from

Turkey Creek at Corral Canyon are shown in Table 1.

Of the 23 characters analyzed for each population, 10 characters appeared to differ adequately for use in further analysis. These were:

head length vertebrae counts

upper jaw length scales above the lateral line

predorsal length scales two rows above the lateral line

adipose fin length (lateral series)

caudal peduncle length pyloric caeca counts

Several criteria were used in selection of these characters. Behnke

(1970) lists spotting pattern, vertebrae counts, scale counts, pyloric caeca counts, and adipose fin length as valuable in separating Gila and rainbow trout. Pyloric caeca counts are completely under genetic control, and not influenced by environmental factors (Behnke, op cit.).

Koster (1957) stated that the upper jaw is relatively long and usually extends posteriorly beyond the orbit. Miller (1950) listed the following characteristics in describing the Gila trout; extremely fine and profuse spotting on dorsal and caudal fins, body spots mostly above the lateral line, adipose fin unusually large and well spotted, head 13 long and conical, maxillary long and extending far behind the eye, 133 to

151 scales along the side just above the lateral line, and 31 to 36 pyloric caeca. Hyoid or basibranchial teeth absent.

The use of morphometric characters to distinguish species is accomplished by using ratios of each measurement to standard length. Validity of ratios depends upon proportional growth between standard length and each character or upon the fish retaining its body form as it grows. Regression and correlation were used to confirm proportional growth of morphometric characters and standard length. In the model Y = a + bX, Y = taxonomic character and X = standard length. Samples consisted of 11 Gila trout and

12 rainbow trout. Observed r, was equal to or greater than values required for correlation at 99.5% for all morphometric characters. Therefore, ratios of morphometric characters used in population comparisons are valid

Table 2).

Ten characteristics selected for further analysis (Table 3) were morphometric and meristic characters whose means appear to differ significantly between Gila and rainbow trout (Table 1). An additional characteristic considered is the presence of residual basibranchial teeth in trout from Spruce, Whitewater, Lipsey, and Dry Creeks, all tributaries of the San Francisco River. Only one specimen from samples of 9 to 11 fish from Whitewater, Lipsey, and Dry Creeks contained basi- branchial teeth. Four of 12 specimens from Spruce Creek contained basibranchial teeth; specimen number 2247 had one strong tooth, specimen number 2270 had a single weak tooth, specimen number 2272 had one strong tooth, and specimen number 2263 had one strong and one weak tooth (Fig. 5).

Two methods were used to compare characteristics of populations to determine taxonomic status. The first, Student's t-test with pooled 14 TABLE 2 . REGRESSION AND CORRELATION SHOWING PROPORTIONAL INCREASES IN LENGTH OF MORPHOMETRIC CHARACTERS WITH INCREASE OF STANDARD LENGTH OF GILA TROUT IN rAIN DIAMND CREEK (TOP) AND RAINBOW TROUT IN TURKEY CREEK AT CORRAL CANYON (BOTTOM). REQUIRED R FOR GI LA TROUT IS .735. REQUIRED R FOR RAINBOW TROUT IS ./08.

PARAMETER Y = A + BX OBSERVED R

ORBIT LENGTH 3.00355 + 0.06331X 0.893

UPPER JAW LENGTH -15.68758 + 0.28996X 0.939

HEAD LENGTH -11.09183 + 0.39220X 0.955

BODY DEPTH -19.85923 + 0.41744X 0.948

PREDORSAL LENGTH -12.29027 + 0.63288X 0.981

DORSAL FIN BASE LENGTH - 1.63010 + 0.16566X 0.911

DORSAL FIN DEPRESSED LENGTH - 4.82565 + 0.28604X 0.960

ADIPOSE FIN LENGTH - 0.41210 + 0.11295X 0.846

CAUDAL PEDUNCLE LENGTH - 0.13338 + 0.14557X 0.953

CAUDAL PEDUNCLE DEPTH - 6.09389 + 0.16189X 0.981

PARAMETER Y = A + BX OBSERVED R

DRBITIENGTH 3.85671 + 0.05750X 0.910

UPPER JAW LENGTH - 2.95106 + 0.18056X 0.926

HEAD LENGTH - 0.74165 + 0.29785X 0.964

BODY DEPTH 3.22839 + 0.23474X 0.895

PREDORSAL LENGTH - 0.47140 + 0.51479X 0.989

DORSAL FIN BASE LENGTH 0.32472 + 0.14986X 0.935

DORSAL FIN DEPRESSED LENGTH 0.86606 + 0.18712X 0.946

ADIPOSE FIN LENGTH 5.06596 + 0.05665X 0.758

CAUDAL PEDUNCLE LENGTH 4.50363 + 0.11416X 0.882

CAUDAL PEDUNCLE DEPTH 5.03270 + 0.07157X 0.708 TABLE 3. ItANS OF 10 CHARACTERISTICS FROM EACH OF 19 POPULATIONS SAMPLED. POPULATICNS EXHIBITING OCCASIONAL BASIBRANCHIAL TEETH ARE MARKED WITH AN ASTERIK.

SCALES UPPER ADIPOSE CAUDAL ABOVE SPCTS PER SAMPLE HEAD JAW PREDORSAL FIN P_DUNCLE VERTEBRAE LATERAL LATERAL PYLORIC SQUARE STREAM SIZE LENGTH LENGTH LENGTH LENGTH _ENGTH COUNT LINE SERIES CAECA CENTIMETER MAIN DIAMOND (N=11) 313.3 178.0 545.7 110.0 118.5 60.5 31.5 149.4 32.5 22.2 SOUTH DIAMOND (N= 8) 300.5 173.1 549.9 108.0 125.0 60.8 32.8 150.1 33.8 31.0 UPPER MCKENNA (N=12) 305.8 172.2 530.0 102.5 115.7 61.2 30.9 148.3 33.0 22.1 LOWER McXENNA (N= 5) 301.6 176.8 533.4 105.2 114.0 60.4 29.5 150.7 34.4 11.4 UPPER IRON (N=10) 298.0 165.5 525.4 109.4 124.4 60.9 31.1 152.5 34.7 25.8 SPRUCE* (N=12) 310.4 178.3 523.3 123.7 123.2 59.3 34.3 154.4 48.2 11.9 BLACK CANYON (N=11) 296.4 165.5 539.2 93.6 114.8 62.0 27.7 142.6 37.7 15.1 SYCAMORE (N= 6) 305.5 160.8 538.7 100.3 116.7 61.8 28.4 143.2 37.2 17.2 WHITEWATER* (N=11) 287.2 153.9 513.3 97.6 114.6 62.3 29.0 139.0 38.6 18.0 LANGSTROTH (N= 7) 313.3 180.3 519.0 102.7 107.7 61.3 29.3 137.9 38.4 17.3 LIPSEY * (N= 9) 296.2 160.5 524.0 88.7 124.1 63.2 27.8 133.0 46.3 11.9 UPPER TURKEY (N= 8) 291.4 157.9 523.0 100.5 113.0 - 28.9 133.1 40.3 16.3 WHITE (N= 8) 302.8 165.3 539.8 93.6 108.5 62.5 27.8 127.9 45.0 15.5 UPPER MCGOLLON (N= 7) 297.4 170.1 508.4 96.7 112.7 61.9 27.6 133.6 38.4 11.3 DRY* (N=11) 285.6 158.5 509.9 94.0 122.4 64.2 29.5 143.5 49.2 11.0 WEST FORK OF MOGOLLON (N= 6) 271.3 144.0 510.5 77.5 120.3 63.0 29.2 127.7 38.3 14.7 TRAIL CANYON (N= 5) 295.2 164.4 520.2 87.6 114.2 63.1 27.0 130.6 50.8 12.0 LOWER SHEEP CORRAL (N=10) 280.8 147.3 531.5 87.4 112.1 62.7 28.2 137.2 43.1 12.5 TURKEY AT CORRAL CANYON (N=12) 292.7 160.1 511.6 91.8 106.4 62.0 29.4 134.1 48.3 15.6

16

SPECIMEN #2247 SPECIMEN #2270

SPECIMEN #2272 SPECIMEN #2263

FIVE 5. HYOID APPARATUS OF FOUR SPECIMENS OF gLA TROUT FROMBSNUCE (REEK SHOWING LOCATION OF BASIBRANCHIAL TEE TH(INDICATED BY -X.).

17 variances, compares characteristics of unknown trout (possible Gila x rainbow hybrids) with the topotypic series of Gila trout from Main

Diamond Creek. The null hypothesis implies that the mean of a standardized characteristic in the unknown population is equal to the mean of the characteristic in the topotype population. Failure to reject the null hypothesis implies that the difference between the characteristic of the unknown sample and the known sample of Gila trout was small in relation to the variability in the data. Compu- tations of t-values were made with the model:

-R t = 1 2

SpV1/n1 + 1/n2

When:

X = mean of unknown population 1

X = mean of topotypic populations (Main Diamond Creek) 2

n = sample size of unknown population 1

n = sample size of topotypic population 2

Sp = pooled standard deviation 2 2 (EX ) (EX ) 2 1 2 _ + Ex 2 n 2 n ni + n2 -2

Values of t for means of ten characteristics of trout from 18 popu- lations which are and are not significantly different from those of Main

Diamond Creek are shown in Table 4.

The second method used to compare characteristics of populations is visual comparison of means and 95 percent confidence intervals. Confidence TABLE 4, COMPUTED T FOR MEANS Of TEN QjPRACTERISTICS FROM EACH OF EIGH7EEN POPULATION§ COMeARED TO MAIN 14,14MCND CRIEK. DEGREES OF FREEDOM (N, + N2 - 2) ARE SHC*I IN PARENTHESES. 0 05 COMPUTED 'S WHICH ARE NOT SIGNIFICANTLY DIFFERENT FROM MAIN_UIAMCND CREEK R . ARE MARKED WITH AN ASTERIR,

UPPER ADIPOSE CAUDAL SPOTS PER HEAD Jaw PREDORSAL F'IN VALiE VtgrAE LATERAL LATERAL SQUARE STREAM LENGTH LENGTH LENGTH LENGTH LINE SERIES PE: CENTIMETER SOUTH DIAMCND (17) 1.5821. (17) 0.6543. (17) 1.1488. (17) 0.5380. (17) -2.4908 (19) -0.7842* (13) -1.6805. (13) -0.2749* (17) -0.9439w (17) -5.1412 UPPER MISENNa (21) 0.9192* (21) 0.8213* (21) 1.7867 (21) 2.6131 (21) 1.2308* (26) -2.0442 (18) 0.8073. (16) 0.3334* (21) -0.3102* (21) 0.0534. LO4ER MCKENNA (14) 1.3832. (14) 0.1541. (14) 1.4007 (14) 1.0743. (14) 1.3560. (20) 0.0573* (11) 2.4627 (10) -0.4029* (14) -1.0303. (14) 4.7607 UPPER IRCN (19) 2.5507* (19) 2.2819 (19) 3.3852 (19) 0.1876. (19) -1.8966 (21) 1.3715* (17) 0.4539* (16) -1.0088* (19) -0.9643* (17) -1.1036. SPRUCE (21) 0.4615. (21) -0.0475* (21) (21) -4.6243 (21) -1.6963* (24) 4.2938 (19) -4.2941 (17) 1.9828 (21)-10.2782 (21) 6.7534 BLACK CANYON (20) 2.7278 (20) 1.8701. (20) 0.7168w (19) 3.2072 (30) 1.4793* (20) -4.3620 (14) 5.6312 (12) 2.4876 (20) -1.5643* (20) 3.4068 SYCAMORE (15) 1.0182. (16) 2.3739 (15) 0.8698* (15) 2.3817 (15) 0.6216* (20) -3.7963 (12) 2.9741 (11) 1.4260. (15) -2.0942 (15) 2.3992 WHITEWATER (20) 3.5257 (20) 4.0759 (20) 3.0609 (20) 2.8226 (20) 1.1684. (22) -5.6621 (12) 2.6371 (10) 2.3868 (20) -2.8763 (20) 2.4870 LANGSTROTH (16) -0.0018* (16) 0.3282* (16) 3.6479 (16) 2.3193 (16) 3.7331 (19) -2.1436 (14) 3.2049 (18) 3.6903 (16) -2.9064 (15) 2.0462 LIPSEY (18) 3.0363 (18) 2.7985 (17) 2.9692 (18) 5.4489 (18) -2.1038 (20) -9.6052 (15) 5.6356 (15) 7.3910 (18) -7.1136 (18) 5.9291 UPPER TURKEY (17) 2.4430 (17) 2.7211 (17) 2.4100 (17) 2.4134 (17) 1.5434. (14) 3.1259 (13) 5.7652 (17) -4.5972 (16) 2.6890 WHITE (17) 1.4350* (17) 1.8501 (17) 0.7417* (17) 4.7496 (17) 4.2422 (19) -4.7636 (15) 4.8127 (14) 9.0977 (17) -6.9244 (17) 2.6578 UPPER MCGOLLON (16) 1.9153 (16) 1.0639* (16) 4.3252 (16) 3.3642 (16) 1.7637 (18) -3.1650 (14) 5.2344 (13) 3.5803 (16) -2.2682 (16) 5.2126 DRY (20) 4.0387 (20) 2.8676 (20) 4.8365 (20) 3.6045 (19) -1.4087. (17) -6.9102 (18) 2.4713 (17) 1.2097* (20) -7.4107 (20) 6.2363 WEST FORK OF MOGOLLON (15) 5.5300 (15) 4.5498 (15) 4.1601 (15) 9.5995 (15) -0.6962* (18) -8.5431 (13) 2.8425 (12) 6.5844 (15) -3.5818 (15) 3.9864 TRAIL CANYCN (14) 2.1355 (14) 1.6567* (14) 2.9023 (14) 6.2347 (14) 1.3882* (19) -6.8791 (12) 6.0018 (11) 6.0396 (14) -7.5207 (14) 4.7413 LOWER SEEP CORRAL (19) 5.6075 (19) 5.2355 (19) 2.1469 (19) 7.8334 (19) 3.0950 (20) -5.2395 (16) 4.6018 (15) 6.0138 (18) -5.6474 (19) 6.0117 TURKEY AT CORRAL CANYON (21) 3.5112 (21) 3.114 (21) 5.5502 (21) 5.4403 (21) 3.1943 (22) -4.5134 (18) 3.2405 (17) 7.6199 (21) -7.4263 (21) 4.5948 19 intervals were computed with the model;

95% CI = ± S t R n-1; 1-a/2

When:

X = mean of characteristic in test population 1 S S_ = x - standard deviation of the mean

2 EX.2 - (EX ) 1 i S = x n-1

t = table-t n-1 n-1; 1-a/2 with degrees of freedom

Means and 95 percent confidence intervals are shown graphically in

Figure 6. UPPER JAW LENGTH PREDORSAL LENGTH CAUDAL PEDUNCLE LENGTH SCALES ABOVE LATERAL LINE PYLOR IC CAECA MAIN DI AMOND --I-- -1-._4_ -1- SOUTH DI AMOND --4--- —I--- SPRUCE I --1--- -I- -I- UPPER MCKENNA I --1--. --i--. LCWER MCKENNA I -1- --I----.--I--. IRON -4-- -I- .--I---. --- 4--. BLACK CANYON ...... 1 -.■ 1.■ LANGSTROTFI . -. .. -1-. ..■,..... WHITE --4-- -I- I —I— 71--- h1-1 TEWATER -4--- SYCAMORE I -.-I-.-___. TRAIL CANYON I I I —I— UPPER MOGOLLCN I I I -I-- DRY I -I- LIPSEY ---I --4-- -4-- --.-11-.1- -... UPPER TURKEY ---4------4---. WEST FORK OF MOGOLLCN -4- --4--. --I-- LOVER SHEEP CORRAL 4 --1__, 4 -- -I- - - - —1—.. TURKEY AT CORRAL CANYON --4-- -I-- —1---- 132 i4i 150 119 14. 1i6 115 194 491 501 511 521 521 541 511 561 99 103 ioa 113 1111 113 Al 132 24 26 28 30 32 34 36 38 27 32 37 42 47 52 57 612

HEAD LENGTH ADIPOSE FIN LENGTH VERTEBRAE LATERAL SERIES SPOTS PEL:f . MAIN DI AMCND ,-_4__. -4- -I-- SOUTH DIAMOND 1 --4-- --I--. --1.- SPRUCE I --I- -4- -I- -I-- UPPER MCKENNA -4- __4_ LOWER MCKENNA --4---"--1- IRON --t--- --1-- -4-- .- -I-. BLACK CANYCN --i--- I .—f--. -I-- LANGSTROTH ---,-- -I--. .--__)__. --I--. WHITE I --1.— -1- WH I TEWATER --1-. ...--I----, SYCAMORE I I --4- TRAIL CANYON ---4---- .---1-- --4-- .---4--- UPPER MOGOLLON --4-- DRY --4___- --,-- LI -- PSEY -4--. -4.--. -4- -4- 1---. UPPER TURKEY I ---,-- ---1-- --1---- WEST FORK OF MOGOLLON ---I--- --I-- LOWER SHEEP CORRAL -t- -4-- -1- .--f... —I---- -1—... TURKEY AT CORRAL CANYCN -I-.- ..-1•••• .--4.-• 260 269 2'7S 257 296 305 314 323 72 80 n 96 104112 120 129 99 60 61 62 63 64 65 611 in lie tis la2 Ci9 146 Ti3 160 6 10 14 18 22 2111 30 34

F1GUFE 6. WAS AND 95 PERCENT CalFILENCE INTERVALS FOR 10 CHARACTERISTICS FROM 19 PCPUIATICNS OF TROUT. DISCUSSION

In determining taxonomic status of a trout population, a combination of characters must be used because no single characteristic has been found that can separate pure Gila trout from hybrids. However, t-tests and visual comparisons of means and 95 percent confidence intervals reveal some characteristics which are more reliable than others. For example, t-values show that mean caudal peduncle lengths (Table 4) from seven Gila x rainbow populations are not significantly different from the mean caudal peduncle length of fish from Main Diamond Creek, and this character alone is not reliable. Conversely, t-values for means of adipose fin length, vertebrae counts, scalation, and pyloric caeca counts of hybrid populations are significantly different from Main Diamond Creek and are more reliable for determining taxonomic status. Means and 95 percent confidence inter- vals (Fig. 6) with considerable overlap between pure Gila trout (Main

Diamond Creek) and pure rainbow trout (Turkey Creek at Corral Canyon) also are of less taxonomic value. Taxonomic status,and reasons therefor, are discussed below for 17 populations.

South Diamond Creek

South Diamond Creek contains a pure population of Gila trout. These fish have 8 of 10 characteristics which are not significantly different from trout in Main Diamond Creek (Table 4). Considerable overlap occurs in

95 percent confidence intervals between this population and Main Diamond 2 Creek for all characteristics except spots per cm (Fig. 6). Spot density 2 for this population is illustrated by a mean of 31.0 spots per cm com- 2 pared to 22.2 per cm for Main Diamond Creek trout (Fig. 7). Spots are

21 22

FIGURE 7. GILA TROUT FRQM SOUTH DIAMCND (UPPER), MAIN DIAMOND (MIDDLE), AND SPRUCE (-REEKS (LOWER), SHOWING TYPICAL SPOTTING PATTERNS FROM EACH STREAM. 23 particularly fine and dense on caudal and dorsal fins, extending farther below the lateral line on specimens from South Diamond Creek. Although proximal to Main Diamond Creek, South Diamond Creek is completely isolated by dry stream beds and probably has been so for many centuries.

The difference in spot density is attributed to this isolation. Popu- lation densities are lower and specimens larger in this stream than in other Gila trout populations.

McKenna Creek

Two barriers in McKenna Creek prevent upstream migration of fish.

One, a 2.4-m vertical waterfall, is just upstream from the confluence of McKenna Creek and the West Fork of the Gila River. The other, approximately 1.2 km upstream from the first, was formed recently by a fallen tree and boulder jam, which blocks upstream movements of trout.

Samples were taken from populations above and below this barrier.

Characteristics of both populations are almost identical except for spot density. Specimens collected below the barrier have a mean of 11.4 2 spots per cm , while specimens above have 22.1 (Main Diamond Creek speci- 2 mens above have 22.2 spots per cm ). The phenotypic difference is attributed to periodic accumulations of volcanic rhyolite tuff below the second barrier. A small canyon entering McKenna Creek from the south, approximately 0.4 km upstream from the first barrier, is the source of the tuff, which appears during periods of heavy runoff. The tuff has a light color and sand-like consistency which covers the stream bottom and produces an almost-white background for trout.

When considering McKenna Creek populations together, they are very similar to the Main Diamond Creek population. Means of seven characteristics 24 of fish upstream from the barrier and eight characteristics of fish downstream from the barrier were not significantly different from those of fish from Main Diamond Creek (Table 4). No rainbow trout influence was detected and the McKenna Creek population consists of pure Gila trout (Fig. 6).

Iron Creek

Iron Creek contains the most-recently discovered population of Gila trout, located during June, 1975, in the extreme upper drainage. The population is protected from hybridization with rainbows by a combination of biological and geological barriers, and from competition and predation by , Salmo trutta, by the geological barrier. The geological barrier consists of intermittent flow through gravel streambed. The extent of the brown trout population separating Gila from rainbow trout is unknown. Although Iron Creek specimens exhibited the shortest head lengths, shortest upper jaw lengths, and highest pyloric caeca counts of all Gila trout populations, t-tests showed no significant differences in means of adipose fin lengths, vertebrae counts, scalation, pyloric 2 caeca counts, and spots per cm between Iron Creek and Main Diamond

Creek fish. Iron Creek contains a pure populations of Gila trout, with several characteristics affected by isolation. Several of the affected characteristics vary away from rainbow trout and others vary toward rainbows (Fig. 6).

Spruce Creek

In his discussion of Spruce Creek trout, Miller (1950) refers to testimony related to Mr. Edwin Shelly, former Superintendent of Jenks 25

Cabin and Glenwood state hatcheries (New Mexico), by an "old timer".

According to this testimony:

"the San Francisco River was originally without native

trout. In 1905, three men were hired by some prospectors

to pack fish from the Gila drainage to Big Dry Creek,

Little Dry Creek, Little Whitewater Creek, Whitewater

Creek, and Mineral Creek, tributaries of the San

Francisco River. These fish were transported success-

fully across the divide in beer kegs carried on mules."

Miller assumed that the Spruce Creek population resulted from these introductions since Spruce Creek is a tributary to Dry Creek.

During June, 1975, a complete physical survey of Spruce Creek was made. Spruce Creek, approximately 200 m upstream from the confluence with Dry Creek, contains a large population of brown trout. Upstream movement of browns is stopped by a waterfall approximately 2 m high.

Gila trout populations begin just above this barrier. Approximately

100 m upstream is a series of barriers from 10 to 15 m high, com- pletely eliminating any upstream movement of fish from Dry Creek, even during flood flows. The remainder of Spruce Creek upstream from the second series of barriers is also divided into at least three sections by barriers which prevent upstream movement of trout. Popu- lations of Gila trout are present upstream from each of these barriers.

For these fish to exist as an introduced population, they would have had to have been stocked upstream from the uppermost barrier in Spruce

Creek--an area devoid of trails and very rugged--or artificially trans- ferred above each barrier from a population introduced downstream. 26

Both of these possibilities are very unlikely and unsubstantiated by testimony offered by Miller.

If the Spruce Creek population was introduced, it should resemble the population from which it was taken. The closest headwaters of the

Gila River are Iron Creek and Rain Creek (Fig. 4). Rain Creek trout phenotypes collected in 1975 suggest pure or almost pure rainbow trout.

By comparing sample means and 95 percent confidence intervals of characteristics from Iron and Spruce Creeks, major differences are evident (Fig. 6).

The most-outstanding difference between Spruce Creek fish and other

Gila trout populations is the presence of basibranchial teeth in Spruce

Creek specimens (Fig. 5). Basibranchial teeth were found in 4 of the

12 specimens comprising the Spruce Creek sample. Basibranchial teeth have never been reported in Gila or rainbow trout except in three of four specimens referred to as Gila trout from Oak Creek, tributary to the Verde River, tributary to the Salt River, Arizona. These fish were collected in 1888-89 (Miller, 1972). Basibranchial teeth are also occasional in Arizona trout (Miller op. cit.).

Spruce Creek specimens showed significant differences between means from Main Diamond Creek specimens for 7 of the 10 characteristics analyzed by t-tests (Table 4). The Spruce Creek sample showed greater means for adipose fin lengths, upper jaw lengths, scalation, and lower mean vertebrae numbers than any other Gila trout population, with all of these characteristics indicating greater differences from rainbow trout rather than affinities with it (Fig. 6). The mean pyloric caeca count for Spruce Creek is 48.2, which is similar to typical rainbow 27 trout. However, pyloric caeca in Spruce Creek specimens differ in size from those of rainbow trout. Caeca attached to the duodenum of rainbow trout are typically short and thin, but increase in length as they progress onto the stomach. Caeca from Spruce Creek specimens are two to three times larger in length and diameter on the duodenum and only increase slightly in size as they progress onto the stomach.

Several characteristics of Spruce Creek specimens suggest that they may represent an intergrade between Gila and Arizona .

Miller (1972) lists the permillage predorsal length of 12 paratypes of Arizona trout as 511. The mean predorsal lengths of specimens from

Main Diamond and Spruce Creeks are 545.7 and 523.3, respectively.

Miller (op.cit.) gives the mean vertebrae count for 55 paratypes of

Arizona trout as 59.5. Mean vertebrae counts of specimens from Main

Diamond and Spruce Creeks are 60.5 and 59.3 respectively. Basibranchial teeth are occasional in Arizona trout (Miller, op. cit.), occasional in

Spruce Creek specimens, and absent in the Main Diamond Creek population.

Behnke (1970) reportedthat the spotting of the Arizona trout is similar to that of the Gila trout, but the spots are larger, resembling some of the cutthroat trout. The spotting pattern in Spruce Creek specimens

(Fig. 7) is similar to that described for the Arizona trout. Mean spots 2 per cm for Spruce Creek is 11.9, while the mean for Main Diamond Creek is 22.2 (Table 3). Spots on Spruce Creek trout, particularly those on the predorsal area, are surrounded by a light circle (Fig. 7), similar to the "halo-effect" described by Miller (op. cit.) for the Arizona trout.

Spruce Creek trout certainly are not typical of other Gila trout populations and may deserve subspecific recognition. 28

Black Canyon Creek

Upper Black Canyon Creek is probably the least hybridized of past

Gila trout populations. Spotting patterns and life colors exhibited by some specimens in this population are impossible to distinguish from

Main Diamond Creek specimens without further analysis. Other phenotypes within this population appear to be almost pure rainbow trout. Although definitely a hybrid population, Black Canyon Creek specimens had means of upper jaw lengths, predorsal lengths, and pyloric caeca counts which were not significantly different from sample means of the same characteristics from the Main Diamond Creek sample (Table 4). Means and 95 percent con- fidence intervals show that most characteristics are intermediate between

Gila and rainbow trout. No effective barrier is known to exist on Black

Canyon Creek to separate this hybrid population from further introductions of rainbow trout. If the Gila trout phenotype is to be retained in Black

Canyon Creek, future stocking of rainbow trout in this stream must be avoided.

Sycamore Creek

The Sycamore Creek sample also represents a Gila x rainbow trout population. However, the typical Gila trout spotting pattern is not evident in any specimens in the sample. Sample means of four characteristics are not significantly different from sample means of the Main Diamond Creek population (Table 4). Means and 95 percent confidence intervals for most characteristics of Sycamore specimens are intermediate between Gila trout populations and pure rainbow trout from Turkey Creek at Corral Canyon

(Fig. 6). 29

Whitewater, Lipsey, and Dry Creeks

These three streams are all tributaries of the San Francisco River.

Whitewater Creek specimens show only slight indications of influence by

Gila trout. The only characteristic of this population not significantly

different from Main Diamond Creek is caudal peduncle length, which has proved to be unreliable in determining taxonomic status (Table 4).

Strangely, life colors exhibited by certain specimens in this sample are similar to those of the golden trout, Salmo aguabonita (Behnke, R. J., personal communication). The presence of a single basibranchial tooth in one specimen of this sample suggests cutthroat or Arizona trout in- fluence.

Lipsey Creek, a tributary of Whitewater Creek, produced specimens with very unusual spotting patterns. Spots were very large and cutthroat- like, completely covering the body both above and below the lateral line.

A curious arrangement of spots was present on the ventral surface of some specimens, between and posterior to the pectoral fins. The lower jaw of most specimens was also spotted (Fig. 8). The means of all characters analyzed for Lipsey Creek were significantly different from those of the sample from'Main Diamond Creek. A single basibranchial tooth was also present in one sptcimen from this sample, again suggesting cutthroat or Arizona trout influence.

Dry Creek produced specimens which were obviously cutthroat x rainbow trout hybrids. Spots are large and decrease in density as they progress anteriorly. Predorsal regions were devoid of spots in some specimens (Fig. 8). A red cutthroat mark was also present in some specimens. The presence of cutthroat influence in this population may 30

FIGURE 8. TROUT FROM LIPSEY (UPPER AND MIDDLE) AND DRY CREEKS (LOWER), SHOWING TYPICAL SPOTTING PATTERNS FROM EACH STREAM. 31

be attributed to introductions of "spotted trout", Salmo clarki sp., 2 into Big Dry Creek as early as 1935 . Sample means of two characteristics from Dry Creek specimens were not significantly different from the means of specimens from Main Diamond Creek. These were caudal peduncle length, and lateral series scale count. The mean of lateral series scale counts from specimens of Dry Creek (143.5) is intermediate between that for typical rainbow trout (134.1) and most cutthroat trout (150-200) (Behnke

1970).

Langstroth, White, and Upper Mogollon Creeks

Behnke (1970) determined that Langstroth and White Creeks contain

Gila x rainbow trout hybrid populations. He presented data from 1970 collections on White and Langstroth Creeks, listing the mean pyloric caeca count and vertebrae count from Langstroth specimens as 40.5 and

61.6, respectively. Mean values for the White Creek sample was 46.2 and 62.4. The 1975 collection from Langstroth Creek has mean pyloric caeca and vertebrae counts of 38.4 and 61.6,respectively. White Creek speci- mens had means of 45.0 and 62.4. Comparison of means from the 1970 and

1975 collections shows that these characteristics of these populations did not change significantly in five years. Means of only two characteristics from both White and Langstroth Creeks were not significantly different from means of the Main Diamond Creek sample (Table 4). Most characteristics indicate a closer affinity with rainbow than with Gila trout (Fig. 6). •

Upper Mogollon Creek is similar to White and Langstroth Creeks in containing a hybrid population with only little Gila trout influence

2 Ibid. 32 detectable (Fig. 6). The sample mean for upper jaw length of specimens from upper Mogollon Creek is the only characteristic not significantly different from that of the Gila troutin Main Diamond Creek (Table 4).

Trail Canyon, West Fork of Mogollon, and Upper Turkey Creeks

Phenotypes of specimens from these creeks appear to be more typical of pure rainbow trout than of a hybrid population. Samples from West

Fork of Mogollon and upper Turkey Creeks had only a single characteristic, caudal peduncle length, which was not significantly different from the

Main Diamond Creek sample. Fish from Trail Canyon Creek had two sample means which were not significantly different from means of the Main Diamond

Creek sample (Table 4). Gila influence in these three streams apparently has been absorbed by stocking of rainbow trout (Fig. 6).

Lower Sheep Corral Creek

During the spring of 1972, approximately 75 Gila trout from Main

Diamond Creek were stocked in the upper fishless section of Sheep Corral

Creek. This section of stream is protected from upstream migration of trout by a 3-m natural rock barrier just above its confluence with

Sapillo Creek (Fig. 4). During the summer of 1975, a population census was conducted on Sheep Corral Creek which revealed 15 Gila trout up- stream from the barrier. A sample of 10 trout was taken downstream from the barrier to test the possibility of Gila trout being washed over the barrier and hybridizing with rainbow trout downstream. Results of t-tests revealed that means of all 10 characteristcs of the lower Sheep

Corral sample were significantly different from means of the Main Diamond

Creek sample (Table 4). The population appears to be pure rainbow trout

(Table 3). 33

SUMMARY

Objectives of this study were to determine purity of previously known and recently discovered Gila trout populations and identify streams containing Gila x rainbow trout hybrids.

Streams containing pure populations of Gila trout are Main Diamond,

South Diamond, McKenna, and upper Iron Creeks. McKnight and upper Sheep

Corral Creeks contain populations of Gila trout introduced from Main

Diamond Creek in 1971 and 1972, and were not analyzed in this study.

Spruce Creek, classified as containing a pure population of Gila trout, may deserve subspecific recognition as an intergrade between Gila and

Arizona trouts. Geographic and taximetric evidence supports a hypothesis that the Spruce Creek population represents a native population from the upper San Francisco River.

Streams containing Gila x rainbow hybrid trout populations are

Black Canyon, Sycamore, Langstroth, upper Turkey, White, upper Mogollon,

West Fork of Mogollon, and Trail Canyon Creeks. Lipsey, Dry, and White- water Creeks contain hybrid trout consisting of rainbow with influence of cutthroat, Gila, or both. Pure populations of rainbow trout occur in lower Sheep Corral Creek and Turkey Creek at Corral Canyon.

Black Canyon Creek contains hybrid trout with phenotypes similar to Gila trout from Main Diamond Creek. In order to preserve these phenotypes, future stocking of rainbow trout in upper Black Canyon

Creek should be avoided.

This study revealed considerable variation between populations of

Gila trout in spotting pattern (Fig. 7) and other meristic and morpho- metric characters (Table 3 and Fig. 6). To preserve the individuality 34 of these populations, future restorations should be made with Gila trout from another stream in the same watershed to avoid mixing.

Restoration of tributaries to the upper San Francisco River should be done only with Spruce Creek trout. Likewise, restoration of streams on the East Fork of the Gila River should utilize Main Diamond or

South Diamond Creek as a source of Gila trout. Restoration on the

Middle and West Forks of the Gila River should utilize Gila trout from

Iron and McKenna Creeks, respectively. 35

LITERATURE CITED

Bailey, V. 1913. Life zones and crop zones of New Mexico. No. Amer. Fauna No. 35, USDA, Washington, D. C.

Behnke, R. J. 1967. Rare and endangered species: The Gila and Apache trouts. Rept. Colo. Coop. Fish. Unit, Colorado State Univ., Fort Collins., 11 pp. Mimeo.

. 1970a Rare and endangered species report: New information on Gila trout, Salmo gilae. Rept. Colo. Coop. Fish. Unit, Colorado State Univ., Fort Collins, 12 pp. Mimeo.

. 1970h The applications of cytogenetics and biochemical systematics to phylogenetic problems in the family . Trans. Amer. Fish. Soc. 99(2):237-248.

. 1973. Gila trout, Salmo gilae. Endangered species report. U.S. Fish and Wldlf. Svc., Reg. II, Albuquerque, N. M. 5 pp. Mimeo.

Cope, E. D., and H. C. Yarrow. 1875. Report upon the collections of in portions of Nevada, Utah, California, New Mexico, and Arizona during the years of 1871, 1872, 1873, and 1874. Rept. Geog. and Geol. Explor. and Surv. W. 100th Mend. (Wheeler Survey), 5:635-703.

Hanson, J. N. 1971. Investigations on Gila trout, Salmo gilae Miller, in southwestern New Mexico. Unpub. M.S. thesis. New Mex. State Univ. Las Cruces. 44 pp.

Huntington, E. H. 1955. Fisheries survey of the Gila and Mimbres Rivers drainages. D-J F-1-R, New Mexico Dept. Game and Fish, Santa Fe. Multilithed.

Jester, D. B. 1965. Gila trout. New Mexico Wildlife, Mar.-Apr. :16-19.

Koster, W. J. 1957. Guide to the fishes of New Mexico. Univ. New Mexico Press. Albuquerque. 116 pp.

LaBounty, J. F., and W. L. Minckley. 1972. Native fishes of the upper Gila River system, New Mexico. Pp. 134-146 in Symposium of rare and endangered wildlife of the southwestern Unite States. New Mexico Dept. Game and Fish, Santa Fe. Multilithed.

Legendre, P., and R. J. Behnke. 1972. Taximetric analysis of selected groups of western North American Salmo with respect to phylogenetic divergences. Syst. Zool. 21(3):292-307. 36

Miller, R. R. 1950. Notes on the cutthroat and rainbow trouts with a description of a new species from the Gila River, New Mexico. Occ. Pap. 529 Mus. Zool., Univ. Michigan Press, Ann Arbor.

. 1961. Man and the changing fish fauna of the American Southwest. Pap. Mich. Acad. Sci., Arts, Lett. 46:365-404.

. 1972. Classification of the native trouts of Arizona with the description of a new species, Salmo apache. Copeia 1972 (3):401-422.

Needham, P. R., and R. Gard. 1959. Rainbow trout in California and Mexico with notes on the cutthroat series. Univ. California Publ. Zool. 67(1):1-124.

New Mexico, State of. 1975. Protection of endangered species and subspecies of New Mexico. Secs. 53-2-50 through 53-2-59, N. M. Statutes, Annotated.

Regan, D. 1964. Ecology of the Gila trout. Unpub. M.S. thesis. Colorado State Univ., Fort Collins. 57 pp.

. 1966. Ecology of the Gila trout in Main Diamond Creek in New Mexico. Tech. Pap. 5, Bur. Sport Fish. and Wldlf. USF and WS. 24 pp.

Simon, R. C. 1964. Cytogenetics, relationships and evolution in Salmonidae. Unpub. Ph.D. thesis, Univ. Washington, Seattle. 76 pp.

U.S. Department of the Interior. 1973. Threatened wildlife of the United States. Resource Pub. 114. Washington, D.C. p. 23.

U.S. Forest Service. 1974. Map: Gila National Forest, New Mexico. Region III, Albuquerque, N. M.