VERTICAL DISTRIBUTION OF FISHES RELATIVE TO PHYSICAL,
CHEMICAL AND BIOLOGICAL FEATURES IN
TWO CENTRAL ARIZONA RESERVOIRS
by
Peter Olof Bersell
A Thesis Presented in Partial Fulfillment of the Requirements for the Degree Master of Science
ARIZONA STATE UNIVERSITY
September 1973 ABSTRACT
Commencing in summer 1970, distribution of fishes in two central Arizona reservoirs was studied by vertical gill netting, stressing intra- and inter-lake variability. Transects were established and sampling was performed three times a year in an attempt to examine conditions during ecological as opposed to calendar seasons. Data on fish distributions, other biotic factors, and selected physico-chemical features were obtained. Multiple linear regression analysis was employed to determine which abiotic and biotic features were most important to vertical dispersion of fishes.
Despite high variability and often small sample size, data suggested dissolved oxygen, chlorophyll a, net plankton, and to a far lesser extent, temperature, as important factors in fish disper- sion patterns. Vertical light penetration (depth of the euphotic zone) seemed to act in a more subtle, indirect manner in its influence on fishes.
Species interactions (predator-prey) were noted and food- chain relationships seemed apparent in many instances. The majority of all fishes netted in the study (1970-71) were within the upper
10 m of the water column in both reservoirs. ACKNOWLEDGEMENTS
This study, one portion of an extensive survey of the Salt
River system, was funded through the National Marine Fisheries
Service (formerly the U.S. Bureau of Commercial Fisheries), under
P. L. 88-309 to Arizona Game and Fish Department and Arizona State
University.
The Salt River Project and the U.S. Forest Service were of assistance, as was the Arizona Game and Fish Department. Bill Inman and Bill Jones, of Roosevelt, Arizona, were particularly helpful in the field in terms of time, equipment storage, firsthand knowledge of the study area, and most importantly, hospitality.
Three persons worked on the over-all survey throughout the study, and contributed greatly to my portion. Donna Portz, whose tirelessness in the field, dedication in the laboratory, and effec- tiveness in dealing with computers, was truly amazing, and deserves special thanks. Sandra E. Willoughby was helpful in key-punching, laboratory work, and more importantly, in persistence. And most of all, for what he taught me about field work, and the dedication necessary to complete a project of this sort under all conditions,
I thank Jack Rinne, who never seems to tire when working to acquire new knowledge.
Iv In addition, Robert Cornelius, Mike Busdosh, Randy McNatt,
Chuck Minckley, "Skeets" Hickerson, Joe Cameron, Jon Scofield, Mike
McSorley, Bill Barfoot, and students of Dr. J. E. Deacon's Aquatic
Ecology class, University of Nevada, Las Vegas, provided discussion and physical assistance.
Dr. M. B. Sommerfeld and his students contributed time and data.
My committee members, Dr. G. A. Cole and Dr. Robert Ohmart, deserve thanks for reading and criticizing this manuscript, and for ideas and comments pertaining to the project. Dr. Cole was helpful also by loaning equipment and providing a sense of humor. My committee chairman, Dr. W. L. Minckley, deserves my gratitude for providing the opportunity to become involved in aquatic biology, and for discussion and guidance in this project.
My wife, Jill, deserves more appreciation and gratitude than can be expressed for help at home and in the field, and more than that, for helping me through the bad times and making the good times better. TABLE OF CONTENTS
Page
LIST OF TABLES ... vii
LIST OF FIGURES ...... viii
INTRODUCTION ..... 1
DESCRIPTION OF STUDY SITE ..... 2
METHODS AND MATERIALS ..... 7
RESULTS AND DISCUSSION .... 16
Physical and Chemical Limnology .... 16
Temperature .... 16
Dissolved Oxygen .... 19
Other Physical-Chemical Factors .... 22
Fishes .... 24
SUMMARY AND CONCLUSIONS .....56
VI LIST OF TABLES
Table Page
1. Selected morphologic parameters for Roosevelt and Apache reservoirs, at full-pool ...... 5
2. Approximate lake levels (maximum depth) during each field trip ...... 6
3. Variables used in Multiple Linear Regression Program . . 15
14 Common and scientific names of fishes used in text . . 23
5. Results of MLRP. Dependent variable = shad ...... 37
6. Results of MLRP. Dependent variable = bass ...... 41
7. Results of MLRP. Dependent variable = carp ...... 44
8. Results of MLRP. Dependent variable = catfish ...... 48
9. Results of MLRP. Dependent variable = sunfish ...... 52
10. Results of MLRP. Dependent variable = black crappie . . 55
vii LIST OF FIGURES
Figure Page
1. Map of study site ...... 4
2. Diagrammatic representation of a net set ...... 10
3. Representative temperature profiles for Roosevelt and Apache reservoirs ...... 18
4. Representative oxygen profiles for Roosevelt and Apache reservoirs ...... 21
5. Depth distribution of fishes during run #1 ...... 27
6. Depth distribution of fishes during run #2 ...... 29
7. Depth distribution of fishes during run #3 ...... 31
8. Depth distribution of fishes during run #4 ...... 33
9. Depth distribution of fishes during run #5 in Roosevelt reservoir ...... 35
VIII INTRODUCTION
A series of four contiguous reservoirs impound approximately
100 linear kilometers (km) of the Salt River, central Arizona, north- east of metropolitan Phoenix. These reservoirs are, from upstream to downstream, Roosevelt, Apache, Canyon, and Saguaro. They were built in the early 1900's to provide storage of water for irrigation, industrial and domestic water supplies for the arid, lower Sonoran
Desert region into which the Salt River originally flowed. With the vast population growth of the Phoenix metropolitan area, they soon became heavily utilized for recreation--fishing, boating, camping, and water skiing--yet not until a few years ago was much known about their potentials for multiple use of this type, especially from the biological standpoint.
In 1964, intensive biological work was begun on these reservoirs by personnel from Arizona State University under the auspices of the Arizona Game and Fish Department. These studies have been generally summarized by Rinne (1973). In 1970-71, I studied vertical distribution of fishes in two of the lakes, Roosevelt and Apache, relative to selected physico-chemical and biotic factors, in an attempt to further define the ecology of some of the more important commercial and game species. DESCRIPTION OF STUDY SITES
Roosevelt Lake, impounded by Roosevelt Dam, the largest and uppermost of the Salt River reservoirs, was completed in 1911. This dam is 220.4 meters (m) in length at its crest and 85.4 m in maximum height. The resulting lake inundates a relatively broad valley located at the confluence of the Salt River and Tonto Creek (Fig. 1).
Apache Lake lies in a narrow, steep-walled canyon, immediately down- stream from Roosevelt Lake. When Apache Lake is full, its headwaters begin at the base of Roosevelt Dam. Horse Mesa Dam which impounds
Apache Lake is 201.2 m long, 91.4 in high, and was completed in 1927.
Table 1 lists full-pool morphologic features which emphasize some of the differences between the broader, more-open Roosevelt and the narrow, fjord-like Apache Lake. Three parameters are particularly striking in this respect. Mean width of Roosevelt is approximately five times that of Apache, volume nearly six times Apache, and slope of basin is 32.0 percent (%) in Apache as compared to only 7.0% for
Roosevelt. During the course of study, Roosevelt Lake was never completely full and Apache Lake varied from completely full to essentially empty (Table 2). 3
Figure I. Map of study site, Roosevelt and Apache reservoirs,
central Arizona; see text for explanation of
transects. Tonto Arm
Roosevelt Lake
W3 --- /JRooseve It Dam W2 Salt Arm R-1
Apache Lake
Horse Mesa Dam
/4-4 Phoenix (arprox. 65.0 km) 5
Table I. Selected morphologic parameters for Roosevelt and
Apache reservoirs, at full pool; from Rinne (1973).
Parameter Roosevelt Apache
Maximum depth (m) 62.5 76.2
Mean depth (m) 24.3 28.7
Maximum width ( km) 3.9 1.0
Mean width ( km) 2.1 1 0.4
Maximum length ( km) 33.7 28.8
Shoreline length ( km) 158.8 88.5
2 Surface area ( km ) 70.1 10.8
3 7 Volume (m X 10 ) 170.4 30.2
Volume development index 1.2 1.1
Shoreline development i ndex 5.3 7.6
Slope of basin (%) 7.0 32.0 6
Table 2. Approximate lake levels (maximum depth) during each
field trip. Full pool values are in parentheses.
Run number Roosevelt Reservoir Apache Reservoir (62.5 m) (76.2 m)
54.6 m 10.0 m
2 56.6 m 7:).0 m
52.1 m /6.0 in
4 50.9 in 38.4 m
50./ m METHODS AND MATERIALS
Vertical gill nets were used to gather data upon fish distri- bution. This technique has been employed by others (Hartman, 1962;
Horak and Tanner, 196)4; Lackey, 1968; Grinstead, 1969), but gill nets, in general, may be highly selective (McCombie and Fry, 1960; Berst,
1961; Heard, 1962; McCombie and Berst, 1969; Johnson, et al., 1970; and references cited). Lackey (1968) provided a description of construction of vertical gill nets and modifications of boats needed to operate such equipment. Net design and accompanying equipment used in my study most closely resembled those employed by Lackey.
White, nylon gill nets were used exclusively. Three nets, each of a single, different mesh size, were positioned at a net-site and constituted what will be subsequently referred to as a "gang."
It was previously determined (Johnson, et al., 1970) that 1.27 centimeter (cm), 3.81 cm, and 7.62 cm (bar measure) were most effective in sampling fish populations of these same reservoirs.
Nets were supported with #120 nylon lines and were approximately three meters in width. Nets were dyed at meter intervals to produce a band of color several cm in width to facilitate accurate recording of data. The first 10 m were marked with red, the next with orange, then yellow, green, blue, and purple, respectively, depending on net length. Two 20-m gangs, and one each of 10, 30, 4o, 50, and 60 m 8 were constructed to adequately cover all net sites at existing or maximum lake levels. The bottom of each net was fastened to 1.91-cm thin-walled conduit, 3.2 m in length, which served both to weight and spread the bottom of the net. In addition, conduits were attached at 10 m intervals while nets were fishing. The top (surface) end was attached to a similar section of conduit fitted with styrofoam floats. This provided flotation and spreading and also functioned as a roller around which nets could be wrapped for drying and storage.
Nets were normally set at each site for a minimum of three nights, however, in event of inclement weather or other unanticipated events, nets were occasionally fished for an additional night or nights.
Net placement is diagrammed in Figure 2, a representation of a20-meter gang set. Each net of the gang was attached with heavy metal rings to loops tied into the #120 line. A yellow, flasher light was attached to the top of one of the buoys to serve as a warning for watercraft (Fig. 2).
Rigging in the boat consisted of wooden uprights, crossbraced and fastened to a rail on the gunwale. Uprights were notched to receive and hold ends of the roller floats. In operating a gang, a single net was detached, raised from the water, and a crank attached to the conduit on the roller float. Netting was then wound onto the roller and fishes removed and their depths of capture recorded before resetting. 9
Figure 2. Diagrammatic drawing of net set: (A) buoys, one
with warning light; (B) roller floats; (C) conduit
spreaders; (D) anchor ropes. ===,== .
1.27 cm 3.81 cm 7.62 cm 11
Netting data were obtained over five periods of time: in
August 1970; November 1970; March-April 1971; August 1971; and in
November 1971. Hereafter these will be referred to as runs #1 through
#5, respectively. Data were obtained from both the upstream reservoirs during the first four runs, but only from Roosevelt Lake during run #5 since Apache Lake had been essentially drained in autumn 1971 for work on Horse Mesa Dam.
Four transects, oriented in a north-south direction, were arbitrarily established on Roosevelt Lake. These were numbered from east to west (#1 through #4, Fig. 1). Transects R-1 and R-4 each had three, regularly-spaced net sets along the transect, and transects
R-2 and R-3, four each. The five transects on Apache Lake were numbered generally from east to west. A single net set was positioned in the narrow canyon at A-1 (Fig. 1) while three sets were made at each of the remaining 4 transects. Transects R-1 and R-4 at the upper ends of the two arms of Roosevelt were shallow, rarely exceeding 15 m in depth. During periods of low water it was necessary to move these transects several hundred meters down-lake in order to have sufficient depth of water to derive meaningful profile data. Transects R-2 and
R-3 included three net sites in deeper water (25.0 to 40.0 m), and one in shallower water over a shelf (ca. 8.0 to 12.0 m) at the north end of either transect.
In Apache, the single gang at A-1 was set in mid-channel in
10 m or less of water for runs #1 through #3; this transect was dried 12 on run #4, as were all transects in that lake on run #5. At A-2, one gang was set over a shelf at the east end (12.0 to 15.0 m), one in the old river channel (ca. 25.0 m), and the remaining set was in shallower water (17.0 m) to the west. At A-3, again, one set was positioned in the channel to the southeast (ca. 30.0 m), and two
(12.0 and 20.0 m) were made over a long, gently-sloping shelf on the northwest end of the transect. A-4 had three gangs evenly spaced; two in deeper water (4o.o and 50.0 m), and one on a shelf (ca. 15.0 to 20.0 m) near the south shore. The southeast gang (A-5-3) at A-5 was positioned in deepest water (ca. 60.0 m) while the other two were placed at successively shallower depths (4o.o and 20.0 m, respectively). It should be reiterated that run #4 on Apache Lake was during its drawdown, and therefore all depth data given above were reduced by about 4o in and many sets could not be used.
Temperature, dissolved oxygen, specific conductance, and pH were measured with a Hydrolab II-B (Hydrolab Corporation, Austin,
Texas) which compensates automatically for temperature variation when measuring conductance. Measurements were made at meter intervals for the top 20 m, and at 10 m intervals thereafter. Vertical light penetration was measured with a submarine photometer. A Hach colorimeter (Hach Chemical Company, Ames, Iowa) was used to determine turbidity which is expressed as Jackson Turbidity Units (JTU's).
Water samples were taken at meter intervals with a Kemmerer water sampler. 13
Phytoplankton was estimated through chlorophyll techniques
(Richards with Thompson, 1952; Creitz and Richards,1955). Water samples were filtered in the field and the filter, containing plankton, was frozen. Later, chlorophyll was extracted with 95% acetone and values determined with a Bausch and Lomb Spectronic 70 at appropriate wave-lengths (Portz,1973). Chlorophyll a (chi a), due to its almost universal presence in phytoplankton, was used as an indicator of places of nutrient input or availability (Brylinsky and Mann,1973), and indirectly as an indicator of food supply for fishes.
Net plankton was collected with a Clarke-Bumpus plankton sampler with a #10, nylon net. Tows were normally made at surface,
3.5, 7.0, and 14.0 m. Samples were split in the laboratory to 1/4 original volume, filtered, and weighed to obtain wet weights. For purposes of this study, net plankton was assumed to represent principally the animal components of the biomass. More detailed information concerning the above physico-chemical and biological factors, sampling techniques, and data analysis, other than those concerned with vertical distribution of fishes was provided by Rinne
(1973) and Portz (1973).
A GE 425 computer was used for data analysis. A program designed to delineate correlations between fishes and physico-chemical factors, and fishes and biological parameters, was employed. This program (Multiple Linear Regression Program [MLRP]) operated in the
following manner. Each species of fish was used in turn as the
dependent variable and those factors having sufficient data to be
meaningful for study of vertical distribution were utilized as
independent variables. The program first sorts out all independent
variables in order of contribution to multiple linear regression.
It then eliminates those variables which do not show significant correlation at the 95% confidence level (designated as level 2 in the program). If more than one factor is correlated at this level of confidence, the factors are listed in order of importance to regression. When no factors correlate at the 95% level, the program then drops to the 90% level (level 1), and follows a similar routine of deleting all variables not significantly correlated at that level.
In case no factors show a significant correlation at level 1, a third alternative, level 0, employs error sum of squares and ranks inde- pendent variables in order of their contribution to the variance.
The result is a complete listing of all independent factors employed in the particular analysis ranked from least to most importance according to contribution to the variance. Variables employed and respective analyses made with MLRP are in Table 3. 15
Table 3. Variables used in multiple linear regression analyses.
Variable Roosevelt Reservoir Apache Reservoir
Run #1 #2 #3 #4 #5 #2 #3 #4
Non-biological parameters:
Temperature X X X X X X X X
Diss. oxygen X X X X X X X X
Conductance X X X X X X X X
Turbidity X X X
pH X X X
Biological parameters:
Chl-a_ X X X X Net-plankton X X X X X X X X
Shad X X X X X X X X
Bass X X X X X X X
Carp X X X X X
Catfish X X X X X X X
Sunfish X X X X X
Black crappie X X X X X RESULTS AND DISCUSSION
Physical and Chemical Limnology
Temperature--Profiles vary seasonally in these reservoirs and were indicative of their basic thermal nature or behaviour (Fig. 3).
Similarities of profiles in an annual cycle between the two bodies of water were evident. Both are warm monomictic lakes--lakes of warmer 0 latitudes in which temperatures never fall below 4.o Celsius (C) at any depth, circulation occurs during winter and direct stratification in summer (Reid,1961). Apache Lake (run #4) displayed an isothermal condition at a time when thermal stratification was normal. This was attributable to its drainage through deep penstocks, by which most of the hypolimnion was removed leaving only warm epilimnion waters.
Dendy (1945) described a similar situation in Norris Reservoir,
Tennessee.
The major difference in the annual temperature cycles of these two reservoirs was depth of metalimnion formation. In
Roosevelt, the metalimnion was positioned between 10 and 15 in during maximum depth. This reservoir is relatively open and frequently stirred by high winds from local summer thunderstorms. Apache is more protected by surrounding terrain and maximum metalimnion depth occurs at roughly half that of Roosevelt (5.0 to 7.0 m). Temperatures within the metalimnion were very similar for both lakes and ranged between 21.0 and 23.4°C. 17
Figure 3. Representative temperature profiles for both
reservoirs during respective runs: (A) Roosevelt,
#I; (B) Roosevelt, #2; (C) Roosevelt #3; (D)
Roosevelt, #4; (E) Roosevelt, #5; (F) Apache,
#1; (G) Apache, #2; (H) Apache, #3; (I) Apache, #4. 5-
10_
15—
I
E 20 I I I 1 i f I i I 1 I I 10 15 20 25 30 10 15 PO 25 30 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30
0— DEPTH-METIERS 5...
10
15
t rr I I I i I 1 1 H 10 15 PC 25 30 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 c° DEGREES CENTIGRADE 19
The thermocline may be an important factor in determining vertical distribution of fishes. When present, it provides fishes a range of temperature regimes for habitation. The terms thermocline and metalimnion are often used interchangably and may cause misunder- standing and ambiguity. The thermocline is defined and employed in discussion in this paper as a plane lying halfway between those two readings with the greatest difference in the temperature profile
(Hutchinson1 1957). Thus defined, the thermocline lies within the metalimnion or area of rapid change in water temperatures.
Dissolved oxygen--Dissolved oxygen curves closely followed seasonal temperature profiles (Fig. 4). Hypolimnetic waters of both reservoirs become oxygen deficient, dropping to less than 1.0 milli- gram per liter (mg/1) during summer months. During winter homothermy, oxygen concentrations remained at higher levels throughout the water column. Abnormality in dissolved oxygen profiles occurred during run #4 on Apache Lake while drainage was occurring. Oxygen concen- tration gradually decreased in stepwise fashion with depth and not at any well-defined point. Under more normal conditions, the depth at which oxygen decreased most rapidly was more nearly similar between the two lakes than were temperature data, but Roosevelt had a slightly thicker zone of reduced oxygen concentration than Apache.
Eley, et al. (1967) netted a variety of fishes in Keystone
Reservoir, Tennessee, and compared dissolved oxygen and vertical fish distribution. Fishes netted were carp, channel catfish, black bullheads, white crappie, gizzard shad, largemouth bass, smallmouth 20
Figure 4. Representative oxygen profiles for both reservoirs
during respective runs. Alphabetical sequence,
locality and runs are the same as in Figure 4. I I I 6 8 10 0 2 4 6 8 10 2 4 6 8 10 2 10 12
1 0
15
20 1 1 1 I 1 1 1 1 0 2 4 6 8 0 12 2 4 6 8 10 12 2 4 6 8 10 12 2 4 6 8 10 12 DISSOLVED OXYGEN - MG/ L 22
1 buffalo, bluegills, and green sunfish. They reported oxygen concentrations lower than 2.0 mg/1, limiting to fish distribution.
Of the above listed species, only white crappie, black bullheads, and carp were caught in water with 2.0 mg/1 or less dissolved oxygen.
Adoption of 2.0 mg/1 as a lower limit contrasts with Ellis (1937) who suggested 5.0 mg/1 as a lower limit on the Ohio River drainage. Since the study of Eley, et al. (1967) was performed on a group of species similar to those caught by me (Table 4), and in a reservoir as opposed to a riverine habitat, their value of 2.0 mg/1 was adopted as a basis for discussion of distribution of fishes with respect to dissolved oxygen in this study.
Other Physical-Chemical Factors
The remaining factors were considered of less importance in vertical fish distribution and are condensed. Complete data on these, and the above two discussed parameters, may be found in Rinne (1973).
The great majority of fishes were netted during nocturnal hours in my study. The effect of subsurface light penetration on vertical fish distribution seemed, however, important in an indirect manner.
Phytoplankton, the primary producers in these aquatic ecosystems, are directly limited by light (solar radiation). This may in turn deter- mine over-all (day and night) depth distribution of net plankton, and indirectly, the fishes.
1 A list of scientific names of fishes caught by me, and other fishes mentioned in text is in Table 4. 23
Table 4. Common and scientific names of fishes mentioned in text
(followMg Bailey, et al., 1970). Those marked with an
asterisk (*) were netted in the present study; those with
a double asterisk (**) were represented by fewer than 10
individuals.
Common name Scientific name gizzard shad Dorosoma cepedianum (Lesueur) threadfin shad* Dorosoma petenense (Gunther) carp* Cyprinus carpio Linnaeus channel catfish* Ictalurus punctatus (Rafinesque) yellow bullhead** Ictalurus natalis (Lesueur) black bullhead Ictalurus melas (Rafinesque) smallmouth buffalofish** Ictiobus bubalus (Rafinesque) bigmcuth buffalofish* lctiobus cyprinellus (Valenciennes) black buffalofish** Ictiobus niger (Rafinesque) / largemouth bass* Micropterus salmoides (Ldcepede) green sunfish* Lepomis cyanellus Rafinesque bluegill* ,Lepomis macrochirus Rafinesque redear sunfish** Lepomis microlophus (Gunther) warmouth** Lepomis gulosis (Cuvier) black crappie* Pomoxis nigromaculatus (Lesueur) white crappie Pomoxis annularis Rafinesque 24
The two reservoirs displayed a marked difference in depths of euphotic zones (Rinne„ 1973). Apache, the clearer of the two, had a calculated mean depth of the euphotic zone of 9.2 m while Roosevelt was more turbid, with a mean depth of 5.4 m during my study period.
Specific conductance values ranged from 1,031 to 3,380 micromhos per cm. These conductance values for Apache and Roosevelt lakes are equivalent to 500 and 1,800 mg/1 total dissolved solids, and were variable spatially and temporally.
Hydrogen ion concentration (expressed as pH) ranged from 6.7 to 9.4 in these reservoirs. Doudoroff (1957) stated that adult freshwater fishes may live indefinitely in waters ranging from pH 5.0 to 9.0.
As noted before, turbidity values were generally higher in
Roosevelt. This was expected since the reservoir received direct inflow from the Salt River and Tonto Creek and therefore served as a settling basin for the system. In addition, relative surface areas and influence of wind (see Rinne,1973) were important causative agents.
Fishes
Minckley and Johnson (1968) recorded 40 species of fishes from the four reservoirs in the Salt River chain. I caught only 13 kinds
(Table 4), due, in part, to the type of gear employed. Threadfin shad comprised more than two-thirds of all fishes netted during my study--
1,490 were taken from Roosevelt and 802 from Apache. Seven hundred 25 thirty-three shad were netted during run #1 in Roosevelt; 102 in
Apache. A bimodal distribution (depth netted) pattern occurred in
Roosevelt with peaks at 3, and 7 to 9 in (Fig. 5). In Apache, a
single peak occurred at 7 to 9 m. Ninety-nine percent of the shad
in Roosevelt were netted above the thermocline compared to 30.0% in
Apache. Two hundred and 155 shad were caught in Roosevelt and Apache, respectively, during run #2. Unlike before, similar vertical disper-
sion patterns occurred in both reservoirs (Fig. 6), although more
fishes occupied deeper waters in Roosevelt at this time. During run
#3, 384 shad were netted from Roosevelt and 342 from Apache. Greater
numbers of shad occurred at 20 in in Apache, however, distribution was
quite similar to the previous sampling period (Fig. 7). Shad
displayed slightly deeper distributional patterns during run #4 in
Apache, but the majority of them occurred in the uppermost portion of the water column (Fig. 8). All shad netted in Roosevelt during this
run were above the thermocline. Most of the 82 shad netted during run #5 in Roosevelt Lake were in the upper few meters of water (Fig. 9).
Field observations of Ferguson (1958) suggested that gizzard
shad inhabited waters ranging in temperature from 22.5 to 23.0° C.
0 Threadfin shad occupied warmer waters (>28.0 C) in Roosevelt Lake
in summer months. By contrast, when a range of water temperatures
existed in Apache Lake the majority of fishes netted inhabited cooler water (<28.0° C), below the thermocline. 26
Figure 5. Depth distribution of fishes caught during run #I:
tfs = threadfin shad; lc = bigmouth buffalo; cat =
catfish; sun = sunfish; BC = black crappie. Top
figure is Roosevelt Reservoir; bottom, Apache
Reservoir. Depth (m) is shown on the ordinal axis.
27
le"
4 0
10 • = 5 FISH
20
30 32
Figure 8. Depth distribution of fishes I n Roosevelt (upper)
and Apache (lower) reservoirs during run #4.
31
4 44 45. 1. 4. • VP
10
20
30
fir *TOLIFIfI°
IIII= 5 FISH 30
Figure 7. Depth distribution of fishes in Roosevelt (upper)
and Apache (lower) reservoirs during run #3. Add-
itional abbreviations not previously used: lb =
smallmouth buffalo; In = black buffalo. A single
shad (42 m) and one carp (39 m) are ommitted from
the lower figure. 29
• , •• Ne• •• 4
44
5 FISH 28
Figure 6. Depth distribution of fishes in Roosevelt (upper)
and Apache (lower) reservoirs during run #2.
33
0
10
20
14° 0
117. 5 FISH 10
20 31+
Figure 9. Depth distribution of fishes in Roosevelt
Reservoir during run #5. 35
1
I m :-.. 5 FISH I
20 - 1 36
During summer de-oxygenation, far less than 1.0% (3 of 919) of the shad caught in Roosevelt and approximately 16.0% (32 of 205) of those netted in Apache were in water of less than 2.0 mg/1 dissolved oxygen. Dendy (1946) reported 20% (47 of 233) of gizzard shad netted in Norris Reservoir, Tennessee, in water of less than 3.0 mg/1 dissolved oxygen. He also recorded a mean depth of capture of 6 m for gizzard shad. By comparison, Bryanand Howell (1945) reported gizzard shad occupying primarily the top 3.5 m of the water column.
Mean depth of capture of threadfin shad was quite similar in Roosevelt
(4.2 m) and Apache (5.2 m) lakes. Furthermore, 73% (Apache) and 61%
(Roosevelt) of the shad netted were taken within what in daylight was the euphotic zone.
The results of MLRP analyses for shad are in Table 5.
Examination of non-biological comparisons suggests temperature and to a lesser extent oxygen as most frequently significantly correlated with depth distribution of shad. Variance explained by multiple regression, however, was quite low. For example, temperature (run #1,
Roosevelt) displayed a negative correlation with shad, while during run #3 (Roosevelt) a more readily explainable, positive correlation was present.
The biological parameters compared with shad appear more instructive. Partial correlations show food-chain relationships, either with a predator species or with phyto- and/or net plankton.
In addition, those cases in which a high percentage of variability is explained (i.e., runs #3 on Roosevelt and #2 and #4 on Apache) the highest partial correlation is with a predator species. 37
Table 5. Results of MLRP employing threadfin shad as the dependent
variable.
Run DF Variability Multiple Confidence Partial
explained correlation level correlation
Non-biological parameters:
Roosevelt Reservoir
#1 192 21% 0.46 2 conductance +
temperature _
dis. oxygen +
#2 202 20% 0.45 2 dis. oxygen +
#3 150 24% 0.49 2 temperature +
#4 85 6% 0.26 2 temperature +
#5 141 36% 0.61 2 conductance -
temperature +
turbidity +
pH +
Apache Reservoir
#2 144 11% 0.34 2 dis. oxygen 4
#3 170 1% 0.13 0 temperature +
dis. oxygen
conductance
#4 26 34% 0.59 2 temperature + 38
Table 5. concluded.
Run DE Variability Multiple Confidence Partial
explained correlation level correlation
Biological parameters:
Roosevelt Reservoir
#2 29 10% 0.32 I net-plankton
#3 31 53% 0.73 2 black crappie +
net-plankton -
#4 33 33% 0.58 2 bass +
chl-a_ i-
#5 38 14% 0.38 0 net-plankton -4-
chl-a
black crappie -
catfish
carp
bass
Apache Reservoir
#2 20 71% 0.84 2 catfish +
#3 42 8% 0.28 1 carp +
#4 12 92% 0.96 2 catfish +
carp
sunfish 39
These data suggest water temperature, dissolved oxygen and planktonic food organisms, plus predator-prey interactions as important factors in vertical distribution of threadfin shad in the two reservoirs.
Largemouth bass ranked second to shad in numbers of fishes netted. These were the most important game fish in the reservoirs at the time of my study. Over-all, 222 specimens were netted in
Roosevelt, at an average depth of 5.8 m. Only 37 were collected from
Apache Lake at an average capture depth of 6.7 m.
Forty-four bass were taken during run #1 in Roosevelt compared to only four in Apache. Ninety-eight percent (Roosevelt) and two of four specimens (Apache) netted were taken above the thermocline and most were in the upper 10 m. During run #2 fifty-one bass were collected from Roosevelt and eight from Apache. The majority of specimens collected from both reservoirs during runs #2 and #3 were also above 10 m. All bass caught during run #4 in Roosevelt (26 fish) were above the thermocline and were distributed throughout the upper
12 m of the water column. In Apache, of 10 bass netted during the same run, all were in the upper 7 m of water. All 33 bass collected in Roosevelt during run #5 were again collected at depths of less than
10 m.
When a choice of temperatures was available in both reservoirs, largemouth bass occupied mostly warmer waters in Roosevelt. By contrast, in Apache, about half were netted in water below the 14o thermocline. Laboratory studies by Ferguson (1958) indicated 30.0 to 0 32.0 C as the final point of temperature preference for largemouth bass. His summer field observations suggested 26.6 to 27.7° C as the preferred temperature range of this same species for lakes in Ontario.
In runs #1 and #4 10% (7 of 70) of the bass in Roosevelt and
20% (1 of 5) in Apache were netted from water that contained less than
2.0 mg/1 dissolved oxygen. Emig (1966) reviewed literature which suggested largemouth bass avoided concentrations of dissolved oxygen approaching 1.5 mg/1 and succumbed at lower concentrations. Mayhew
(1963) reported that 92.5% of all largemouth bass captured in Red Haw
Lake, Iowa, were above the point of total oxygen depletion.
With respect to vertical light, approximately half the bass netted in both reservoirs, 58.5% in Roosevelt and 54.1% in Apache, were above mean depth of the euphotic zone.
The non-biological MLRP analysis for bass (Table 6) suggests temperature the most consistently correlated factor. Variation explained by multiple regression is again low. In the biological comparisons, run #4 shows a close positive association of bass and shad in Roosevelt, and in runs #3 and #4, bass were closely allied with carp and catfish.
Over-all, 61 carp were caught in Roosevelt and 77 in Apache.
Average depth of capture was 7.7 and 11.6 m, respectively. All carp
caught during run #1 in Roosevelt were above the thermocline compared to only 21% in Apache. Similar vertical dispersion occurred in
Roosevelt during the following summer (run #4). 1 1
Table 6. Results of MLRP employing largemouth bass as the dependent
variable
Run DF Variablilty Multiple Confidence Partial
explained correlation level correlation
Non-biological parameters:
Roosevelt Reservoir
#1 193 8% 0.30 2 dis. oxygen +
temperature +
#2 201 9% 0.30 2 conductance +
temperature +
#3 150 11% 0.32 2 temperature +
#4 85 10% 0.32 2 dis. oxygen
#5 144 9% 0.31 2 conductance
Apache Reservoir
#3 170 0% 0.08 0 conductance +
dis. oxygen +
#4 26 21% 0.46 2 temperature +
Biological parameters:
Roosevelt reservoir
#2 25 2% 0.15 0 net-plankton -
carp
catfish 142
Table 6. concluded.
Run DF Variability Multiple Confidence Partial
explained correlation level correlation
shad
black crappie -
#3 28 8% 0.29 0 black crappie +
chl-a
net-plankton
shad
catfish +
#4 34 21% 0.47 2 shad +
#5 43 7% 0.27 1 net-plankton +
Apache Reservoir
#3 42 58% 0.76 2 carp +
#4 14 53% 0.73 2 catfish + 43
In general, carp were found scattered through the uppermost
15 to 20 m, especially in Apache (Figs. 5-8). In Roosevelt Lake, they tended to be nearer the surface. Ferguson (1958) listed 32.0° C as the preferred temperature for carp in the laboratory. This temperature was never recorded during my study, but carp were found almost exclusively in warmer water above the thermmcline in Roosevelt.
By contrast, most carp in Apache were in cooler waters below the thermocline.
In Roosevelt, during runs #1 and #4, approximately 30% (11 of 37) of the carp caught were netted in water with less than 2.0 mg/1 dissolved oxygen compared to 21% (4 of 19) in Apache. Carp are bottom feeders as indicated by the presence of clams (Corbicula maniliensis
[Phillipi]) in their stomachs, yet during the entire study period,
27.9% of the carp from Roosevelt, and 41.6% from Apache were collected within the euphotic zone, reflecting, in part, the greater clarity of the latter reservoir. However, carp was the only species which appeared to occur in greater numbers below the limit of the euphotic zone in both reservoirs. Dendy (1945, 1946) reported carp widely distributed at all depths in Norris Reservoir, Tennessee. My data also suggest carp more widely distributed with respect to depth than other species, particularly in Apache Lake.
The non-biological MLRP analysis employing carp as dependent yielded very low percentages of variation explained (Table 7). In instances where multiple correlation values were significant at level 2, the significance of the partial correlations varied. 44 Table 7. Results of MLRP employing carp as the dependent variable.
Run DF Variability Multiple Confidence Partial
explained correlation level correlation
Non-biological parameters:
Roosevelt Reservoir
#2 202 3% 0.69 2 dis. oxygen +
#5 144 3% 0.17 2 temperature +
Apache Reservoir
#2 142 0.07 0 temperature
dis. oxygen
conductance
1/3 170 0.11 0 temperature
dis. oxygen
conductance
#4 26 16% 0.40 2 conductance
Biological parameters:
Roosevelt Reservoir
#2 29 48% 0.70 2 black crappie +
#5 43 20% 0.45 2 chl-a
Apache Reservoir
#3 41 62% 0.79 2 bass
net-plankton 145
Table 7. concluded.
Run DF Variability Multiple Confidence Partial
explained correlation level correlation
#4 12 63% 0.79 2 sunfish
catfish
shad 46
The MLRP comparisons of biological factors (Table 7) resulted in relatively high values of variability explained. Significant
partial correlations were present with black crappie and bass upon
single occasions and sunfish in three instances.
No statistical conclusions concerning depth distribution of
black and smallmouth buffalofishes were made because of insufficient
data. Only 27 bigmouth buffalofish were caught in Roosevelt during the study period and 14 in Apache. Absence of black and smallmouth
buffalofishes in Roosevelt is inexplicable, but most likely relates
to past periods of drought and desiccation of the reservoir (Johnson
and Minckley,1972). The average depth of capture for bigmouth
buffalo for the entire study was 6.2 and 7.8 in Roosevelt and Apache, respectively.
All buffalofish caught in Roosevelt were above thermocline
depth during runs #1 and #4. No bigmouth buffalofish was collected
in water with less than 2.0 mg/1 dissolved oxygen during run #1.
During run #4, 50% (1 of 2) and 20% (1 of 5) in Roosevelt and
Apache, respectively, were collected in water with less than 2.0 mg/1
dissolved oxygen. Approximately 33% of the bigmouth in Roosevelt and
64.3% in Apache were collected within the euphotic zone.
Forty-six catfish, including only one yellow bullhead, were
netted from Roosevelt at an average depth of 7.7 m. Thirty-five,
including three yellow bullheads, were collected from Apache at a
calculated mean depth of 6.7 m. All catfish caught in Roosevelt
during runs #1 and #4 were above the thermocline. 47
Approximately 20% (3 of 14) of the catfish from Roosevelt were collected in water with less than 2.0 mg/1 dissolved oxygen.
In Apache, no catfish was caught in poorly oxygenated (less than
2.0 mg/1) water. In Roosevelt, 60.9% of the catfish were netted in the euphotic zone compared to 57.1% in Apache. Mayhew (1963) found most catfish at approximately 25.0° C with a few specimens in water containing no detectable dissolved oxygen. Miller (1966) suggested lethal levels of dissolved oxygen for channel catfish of 0.76 and
0.89 mg/1 at 25.0 and 30.0° C, respectively. Borges (1950) recorded channel catfish distributed at all depths in the Niangua Arm of the
Lake of the Ozarks and suggested this species appeared more able to endure low dissolved oxygen than other fishes.
The non-biological MLRP analysis for catfish indicated that in one instance (run #4, Apache) variability explained by multiple regression was high. In this case, temperature was the only parameter correlated at level 2. Biological comparisons (Table 8) suggested high positive correlation with shad in runs #2 and #4 in Apache Lake.
Bluegill were the dominate sunfish in both lakes, but because of the small sample sizes all sunfishes of the genus Lepomis were placed together for statistical analysis. Twenty-three sunfish were netted from Roosevelt at an average depth of 4.8 m. Seventy-three were collected from Apache at an average depth of 5.4 m. Nineteen of the 20 specimens netted from Roosevelt during runs #1 and #4 were above the thermocline. Four of the sunfish caught from Roosevelt 48
Table 8. Results of MLRP employing catfish as the dependent variable.
Run OF Variability Multiple Confidence Partial
explained correlation level correlation
Non-biological parameters:
Roosevelt Reservoir
#2 200 0% 0.07 0 conductance -
dis. oxygen +
temperature -
#3 148 l% 0.12 0 dis. oxygen -
temperature +
conductance
#4 84 7% 0.26 I dis. oxygen
temperature +
#5 143 9% 0.30 2 temperature +
conductance _
Apache Reservoir
#2 144 5% 0.23 2 dis. oxygen +
#3 170 1% 0.11 0 conductance +
temperature +
dis. oxygen _
Biological parameters:
Roosevelt Reservoir
#2 25 3% 0.18 0 net-plankton 49
Table 8. continued.
Run DF Variability Multiple Confidence Partial
explained correlation l evel correlation
bass
black crappie -
shad
carp
#3 31 16% 0.41 2 net-plankton +
chl-a — #5 38 6% 0.25 0 shad +
bass
carp
chl-a
black crappie -
net-plankton
Apache Reservoir
#2 20 71% 0.84 2 shad +
#3 38 2% 0.15 0 shad
net-plankton
sunfish
bass
carp + 50
Table 8. concluded.
Run OF Variability Multiple Confidence Partial
explained correlation level correlation
#4 12 91% 0.95 2 shad
carp
sunfish 51 during run #1 were the only specimens taken in any run from water less than 2.0 mg/1 dissolved oxygen. Sixty-four percent and 56.2% of the sunfish were taken from the euphotic zone in Roosevelt and
Apache, respectively. The non-biological MI,RP analysis (Table 9) for sunfish shows highest values of variability explained during run #4 in both reservoirs. Positive partial correlations with water temperature for Apache on runs #3 and #4 seemed significant.
Biologically, run #4 on Apache suggested a significant partial correlation with shad indicating that while shad are probably not a major part of the diet of sunfish according to stomach samples
examined by me, both shad and the smaller sunfishes are simply found
in the upper part of the water column.
Black crappie were abundant in Roosevelt, ranking third (after shad and bass) in numbers of specimens netted. One hundred ten were netted during the entire study at an average depth of 3.4 m. Only two were caught in Apache at a mean depth of 2 m, precluding any comparisons between the lakes. All crappie netted in Roosevelt
during runs #1 and #4 were above the thermocline. In my study, only
4 of 12 crappie taken, during runs #1 and #4 were in water with less than 2.0 mg/1 dissolved oxygen. Dendy (1945) noted black crappie tended to avoid dissolved oxygen concentrations lower than 1.5 mg/1,
but in later studies he caught 4 of 21 black crappie in water with
less than 3 mg/1 dissolved oxygen (Dent; 1946).
Approximately 77.0% of the crappie were collected within the
euphotic zone. 52
Table 9. Results of MLRP employing sunfish as the dependent
variable.
Run DF Variability Multiple Confidence Partial
explained correlation level correlation
Non-biological:
Roosevelt Reservoir
#1 194 3% 0.17 2 conductance +
#4 82 19% 0.43 2 pH -
dis. oxygen +
conductance
Apache Reservoir
#2 142 2% 0.14 0 dis. oxygen +
conductance +
temperature
#3 171 4% 0.21 2 temperature +
dis. oxygen
#4 26 20% 0.45 2 temperature +
Biological parameters:
Roosevelt Reservoir
#4 30 4% 0.19 0 net-plankton
bass
shad
black crappie -
chl-a + 53
Table 9. concluded.
Run OF Variability Multiple Confidence Partial
explained correlation l evel correlation
Apache Reservoir
#2 18 1% 0.12 0 shad
net-plankton
catfish
#3 38 1% 0.11 0 carp
net-plankton
catfish
shad
bass
#4 12 55% 0.74 2 shad
carp
catfish 54
Very little can be stated from non-biological MLRP analyses
(Table 10). Biological comparisons employing black crappie as
dependent showed significant correlation with threadfin shad in run
#3 (Roosevelt) at level 2. 55
Table 10. Results of MLRP employing black crappie as the
dependent variable.
Run OF Variability Multiple Confidence Partial
explained correlation level correlation
Non-biological parameters:
Roosevelt Reservoir
#1 194 8% 0.26 2 conductance +
fi2 200 1% 0.11 0 dis. oxygen +
conductance -
temperature +
#3 150 5% 0.22 2 temperature +
#4 85 5% 0.22 2 pH
#5 143 11% 0.33 2 temperature +
conductance
Biological parameters
#2 29 48% 0.70 2 carp +
#3 30 69% 0.83 2 shad +
chl-a_ + net-plankton +
#4 30 4% 0.21 0 chl-a
sunfish
shad
net-plankton -
bass
#5 43 27% 0.52 2 chl-a SUMMARY AND CONCLUSIONS
The two reservoirs studied exhibited a number of significant contrasts. First of all, basin morphology and surrounding terrain were important in their effects. One major point of contrast was in depth of euphotic zone, the average depth being approximately twice as great in Apache as in Roosevelt. This condition was the reverse of thermocline depth in the two bodies of water, the average depth of thermocline in Roosevelt being roughly twice (10 to 15 m) that of
Apache.
A number of parameters studied yielded similar data for the two reservoirs. Probably the most significant was the depth of critically or drastically reduced dissolved oxygen concentration during periods of thermal stratification. Although oxygen curves generally tended to follow temperature profiles, the differences between the two reservoirs was not so marked as differences in temperature profiles.
With respect to the difference in thermocline depth, the almost identical depth of capture of fishes was totally unexpected.
Depth distribution of fishes (Figs. 5 through 9) indicated that during all sampling periods the upper portion of the water column,
_i.e., _ the top 10 m, contained the majority of fishes. This was the case in both lakes. In order to further demonstrate this between the two impoundments, correlation coefficients for numbers of all 57 fishes caught at each meter were computed for each run. A total of
878 fishes were taken from Roosevelt and 129 from Apache during run
#1. Values for r (correlation coefficient) and r2 (coefficient of determination) were 0.36 and 0.13, respectively, indicating a difference in fish distribution between the two lakes. The second run yielded a total of 290 fishes from Roosevelt, and 185 from Apache, but r (0.75) and r2 (0.56) indicated similar dispersion patterns in both lakes. The third sampling period produced 391 fishes in
Roosevelt and 412 in Apache. Again, r (0.80) and r2 (0.64) indicated similarity of distributional patterns of fishes between the two lakes.
Run #4 (294 fishes from Roosevelt; 314 from Apache) displayed the highest correlation of all sampling periods between the two lakes
(r = 0.88; r2 = 0.78).
The most plausible reason for the low correlation of patterns of fish dispersion between Roosevelt and Apache during run #1 seems the difference in the manner in which shad were distributed. In
Roosevelt, more than 700 shad were caught in a markedly bimodal, vertical pattern, as compared to only about 100 shad from Apache, with a single peak in numbers by depth. Both peaks in number of fishes caught in Roosevelt, however, were in the top 10 m of water, above the thermocline. Furthermore, average depths of capture point out the fact that approximately the same portion of the water column was utilized by fishes in both reservoirs at all times of year, as indicated by seasonal sampling. In Roosevelt, all species had a 58 grand average depth of capture of less than 10 m. Black crappie inhabited shallowest waters as indicated by their mean depth of capture, followed by shad. Catfish occupied deepest waters, closely followed by carp. In Apache, all species of fishes, with the excep- tion of carp, likewise had an average depth of capture of less than
10 m. Shad inhabited shallowest waters and carp the deepest.
Certain parameters were ruled out as having a demonstrable effect on depth distribution of fishes. Conductivity, pH, and turbidity, although variable during different seasons, and even in different sectors of the same reservoirs during a given season, tended to vary more horizontally than vertically (Rinne,1973). The range of values found in this survey were never of a limiting nature to fish life in the upper 10 in of water in which they lived.
The surprising factor to be essentially ruled as important in determining depth distribution of fishes in these two bodies of water was depth of thermocline. Temperature profiles from both reservoirs during stratified periods seemed to have no consistent relationship to fish distribution. In Roosevelt, virtually all species were found above the thermocline, when it was present. By contrast, in Apache a majority of fish species were found below the thermocline. It is important to keep in mind that the thermocline in Roosevelt tended to form at a somewhat greater depth than 10 m, while in Apache it formed at less than 10 m, and that most fishes, in both lakes, were in this upper 10-meter portion of the water 59 column. It becomes evident from this comparison between reservoirs, involving the same species of fishes, that temperature was not limiting. It also is evident, however, that during periods of holomixis all species tended to be dispersed to greater depths. This was undoubtedly due in part to increase in dissolved oxygen at depths which were oxygen deficient during summer stratification. Oxygen alone is not sufficient, however, to explain the depth distribution of all the fishes, since the entire zone of high (adequate) oxygen concentration was not utilized in summer, nor was the entire water column used during circulation periods by the majority of fishes. It is probably best to say that dissolved oxygen concentration imposed outer limits on fish distribution. Other factors were then limiting within the zone of sufficient oxygen concentration for fish life.
The single factor which appeared to be most important in affecting the depth at which fishes are found, depth of euphotic zone, could not have acted directly since most fish were caught at night.
It must have acted indireotly, in a more subtle manner. The depth to which light penetrates is viewed and discussed in terms of the euphotic zone or the "usable" portion of the water column. "Usable" refers here to the ability of phytoplankton to maintain productivity in excess over respiration. First, the reason for suspecting the euphotic zone as being of primary importance is that approximately
60% of all fishes in Roosevelt, and 68% in Apache were captured within 6o this zone. Aside from dissolved oxygen, this was the most consistent positive relationship of fish dispersion patterns with any measured environmental factor in the two reservoirs.
The mechanism by which the depth of the euphotic zone deter- mined fish depth distribution is postulated as follows. Both chi a and net plankton concentrations were highest in the upper part of the water column (Rinne,1973; Portz,1973). This was due directly to the depth of the euphotic zone, particularly in the cases of chi a.
Zooplankton graze on phytoplankton, and thus their distribution, at least partially, was dependent on the depth distribution of phyto- plankton. Haskell (1959) demonstrated that threadfin shad in Arizona are pelagic plankton feeders. Baker and Schmitz (1971) also reported threadfin shad as limnetic feeders, utilizing diatoms most frequently on an annual basis. These data suggest that vertical light penetration may be an indirect limiting factor for the species.
Furthermore, shad were the major forage fish in Salt River reservoirs, and light penetration may have a significant role in determining distributions of predator species, such as bass, crappie, and catfish.
Since these species comprised the major part of the predatory fish fauna, depth distribution of fishes in these lakes seemed to be largely due to the indirect effect of depth of euphotic zone through its effect on shad dispersion. The deeper distribution of carp, particularly in Apache, most likely reflected its bottom feeding habits. Similarly, buffalofish have the second deepest mean depth 61. of capture in both reservoirs, reflecting their on-or-near-bottom feeding habits (Minckley, et al.1 1970). Sunfish of the genus
Lepomis which are largely insectivorous, are found around the lake perimeters, and therefore in shallower waters.
Computer analyses, although indispensible, presented problems in interpretation, and more importantly, in applicability of results of this study. This is primarily because, in most cases, the degree of variability explained in most comparisons was very low, indicating the highly variable and unique nature of data acquired from these desert impoundments. Or, if this factor was acceptable, it tended to be in cases where "n" was low. It is important to note, however, that in nine of 21 cases among the biological MLRP runs where a correlation was established at level 1 or 2, a food-chain relationship was indicated. LITERATURE CITED
Bailey, R. M., J. E. Fitch, E. S. Heard, E. A. Lachner, C. C. Linsley, C. R. Robins, and W. B. Scott. 1970. A list of common and scientific names of fishes from the United States and Canada. 3rd ed. Amer. Fish. Soc. Spec. Publ. 6: 1-150.
Baker, C. D., and E. H. Schmitz. 1971. Food habits of adult gizzard and threadfin shad in two Ozark reservoirs, p. 3-12. In Hall, G. E. (ed.), Reservoir Fisheries and Limnology. Amer. Fish. Soc. Spec. Publ. 8, Washington, D. C.
Berst, A. H. 1961. Selectivity and efficiency of gill nets in South Bay and Georgian Bay of Lake Huron. Trans. Amer. Fish. Soc. 90: 413-418.
Borges, H. M. 1950. Fish distribution studies, Niangua Arm of the Lake of the Ozarks, Missouri. J. Wild. Manage. 14: 16-33.
Bryan, P., and H. H. Howell. 1946. Depth distribution of fish in Lower Wheeler Reservoir, Alabama. J. Tenn. Acad. Sci. 21: 4-9.
Brylinsky, M., and K. H. Mann. 1973. An analysis of factors govern- ing productivity in lakes and reservoirs. Limnol. Oceanogr. 18: 1-14.
Creitz, G. I., and F. A. Richards. 1955. The estimation and charac- terization of plankton populations by pigment analyses. III. A note on the use of "millipore" membrane filters in the estimation of plankton pigments. J. Mar. Res. 14: 211-216.
Dendy, J. S. 1945. Fish distribution, Norris Reservoir, Tennessee 1943. Depth distribution of fish in relation to environmental factors, Norris Reservoir. J. Tenn. Acad. Sci. 20: 114-135.
. 1946. Further studies of depth distribution of fish, Norris Reservoir, Tennessee. J. Tenn. Acad. Sci. 21: 94-104.
Doudodroff, P. 1957. Water quality requirements of fishes and effects of toxic substances, p. 403-430. In Brown, M. E. (ed.), The Physiology of Fishes, Vol. II. Academic Press, Inc., New York. 63
Eley, R. L., N. E. Carter, and T. C. Dorris. 1967. Physicochemical limnology and related fish distribution of Keystone Reservoir, p. 333-357. In Lane, C. E. (ed.), Reservoir Fishery Resources Symposium. Amer. Fish. Soc., Washington, D. C.
Ellis, M. M. 1937. Detection and measurement of stream pollution. U. S. Bur. Fish., Bull. 22: 1-437.
Emig, J. W. 1966. Largemouth bass, p. 332-353. In Calhoun, A. (ed.), Inland Fisheries Management. Calif. Fish & Game, Sacramento, Calif.
Ferguson, R. G. 1958. The preferred temperature of fish and their midsummer distribution in temperate lakes and streams. J. Fish. Res. Bd. Canada 15: 607-624.
Grinstead, B. G. 1969. The vertical distribution of the white crappie in the Buncombe Creek arm of Lake Texoma. Okla. Fish. Res. Lab., Bull. 3: 1-37.
Hartman, G. F. 1962. Use of gill-net rollers in fishery investiga- tions. Trans. Amer. Fish. Soc. 91: 224-225.
Haskell, W. L. 1959. Diet of the Mississippi threadfin shad, Dorosoma petenense atchafalayae, in Arizona. Copeia 1959: 298-302.
Heard, W. R. 1962. The use and selectivity of small-meshed gill nets at Brooks Lake, Alaska. Trans. Amer. Fish. Soc. 91: 263-268.
Horak, D. L., and H. A. Tanner. 1964. The use of vertical gill nets in studying fish depth distribution, Horsetooth Reservoir, Colorado. Trans. Amer. Fish. Soc. 93: 137-145.
Hutchinson, G. E. 1957. A Treatise on Limnology. Vol. I. Wiley and Sons, Inc., New York, N. Y. 1015 p.
Johnson, D. W., and W. L. Minckley. 1972. Variability in Arizona buffalofishes. Copeia 1972: 12-17.
Johnson, J. E., J. N. Rinne, and W. L. Minckley. 1970. Selectivity and efficiency of some commercial fishing devices in central Arizona reservoirs. J. Ariz. Acad. Sci. 6: 46-50.
Lackey, R. T. 1968. Vertical gill nets for studying depth distribu- tion of small fish. Trans. Amer. Fish. Soc. 97: 296-299.
McCombie, A. M., and F. E. J. Fry. 1960. Selectivity of gill nets for lake whitefish Coregonus clupeaformis. Trans. Amer. Fish. Soc. 89: 176-184. 614
, and A. H. Berst. 1969. Some effects of shape and structure of fish on selectivity of gill nets. J. Fish. Res. Bd. Canada 26: 2681-2689.
Mayhew, J. 1963. Thermal stratification and its effects on fish and fishing in Red Haw Lake, Iowa. Iowa State Conserv. Comm., Des Moines, Iowa. 24 p.
Miller, E. E. 1966. Channel catfish, p. 440-462. In Calhoun, A. (ed.), Inland Fisheries Management. Calif. Fish & Game, Sacramento, Calif.
Minckley, W. L., and J. E. Johnson. 1968. The common fishes of the Salt River reservoirs. Wildl. Views 15: 4-13.
, J. E. Johnson, J. N. Rinne, and S. E. Willoughby. 1970. Foods of buffalofishes, genus Ictiobus, in central Arizona reservoirs. Trans. Amer. Fish. Soc. 99: 333-342.
Portz, D. 1973. Plankton pigment heterogeneity in seven reservoirs of the lower Colorado basin. Unpubl. M. S. Thesis. Arizona State Univ. 168 p.
Reid, G. K. 1961. Ecology of Inland Waters and Estuaries. Van Nostrand Reinhold Co., New York, N. Y. 116 p.
Richards, F. A., with T. G. Thompson. 1952. The estimation and characterization of plankton populations by pigment analyses. II. A spectrophotometric method for the estimation of plankton pigments. J. Mar. Res. 11: 156-172.
Rinne, J. N. 1973. A limnological study of central Arizona reservoirs with reference to horizontal fish distribution. Unpubl. Ph.D. dissertation. Arizona State Univ. 350 p. BIOGRAPHICAL SKETCH
Peter Olof Bersell was born in Battle Creek, Michigan, on January 20, 1948. He attended Jacksonville, Illinois, public school for his elementary education. He received his secondary education at Roosevelt Junior High School in Beloit, Wisconsin, and graduated from Prescott High School, Prescott, Arizona, in 1966. He then attended the University of Arizona at Tucson for two years, before transferring to Arizona State University in Tempe, where he majored in biology. He received his Bachelor of Science degree in June 1970, and then entered the Graduate School of Zoology at Arizona State University. He held research assistantships while studying for his Master of Science degree in Zoology. He is now a student working toward the degree of Doctor of Philosophy at the Biology Department of the University of Louisville, Louisville, Kentucky, where he holds a National Science Foundation Fellowship. He is married and has one daughter.