Patterns of diversity and density in the Ichthyofauna of the Anclote Anchorage, Florida

Patterns of diversity and density in the Ichthyofauna of the Anclote Anchorage, Florida

Material Information

Patterns of diversity and density in the Ichthyofauna of the Anclote Anchorage, Florida
Rolfes, J. Kenneth
Place of Publication:
Tampa, Florida
University of South Florida
Publication Date:
Physical Description:
viii, 70 leaves : ill. ; 29 cm.


Subjects / Keywords:
Fishes -- Florida -- Anclote Anchorage ( lcsh )
Dissertation, Academic -- Marine science -- Masters -- USF ( FTS )


General Note:
Thesis (M.S.)--University of South Florida, 1974

Record Information

Source Institution:
University of South Florida
Holding Location:
Universtity of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
029035747 ( ALEPH )
01675765 ( OCLC )
F51-00011 ( USFLDC DOI )
f51.11 ( USFLDC Handle )

Postcard Information



This item is only available as the following downloads:

Full Text


.Graduate Counci 1 University of South Florida Tampa, Florida CERTIFICATE OF APPROVAL MASTER1S THESIS This is to certify that the Masters Thesis of J. Kenneth Rolfes with a major in Marine Science has been approved by the Examining Committee as satisfactory for the thesis requirement for the Master of Arts degree at the convocation of December, 1974 Thesis committee: Thesis supervisor: Ronald C. Baird, Associate Professor Member: Norman J. Blake Assistant Professor Member: Thomas L. Hopkins, Associate Professor Member: Thomas E. Pyle, Associate Professor


ACKNOWLEDGMENTS The author would like to express his appreciation to his advisor Dr. R. C. Baird not only for his guidance and assistance, but most of all for the freedom of inquiry and latitude he allowed in this research. The thesis committee members Dr. N. J. Blake, Dr. T. L. Hopkins and Dr. T. E. Pyle also deserve thanks for their efforts, suggestions, patience and editing. The graduate students and staff of the Anclote Project deserve special commendation, especially my co-workers in the 11Fish Group.11 Without their efforts this study could not have been accomplished. My wife Laine offered invaluable aid in data reduction and the preparation of the manuscript. Financial support for this project was provided through a grant from the Florida Power Corporation. ii


TABLE OF CONTENTS LIST OF TABLES . LIST OF FIGURES. INTRODUCTION. . . AREA DESCRIPTION. MATERIALS AND METHODS Preface. . . Sampling Stations. Sampling Gear. . . Sampling Methods Scope of Data. . Data Reduction and Management. Data Presentation. SAMPLING BIAS AND LIMITATIONS Sampling Program Data . . Analysis ... RESULTS . . Pre face. . . . Surface Water Temperatures Patterns of Fish Density .. Patterns of Fish Diversity . . Overall Considerations of Pattern .. DIS CUSS I ON. . . . . Comparison with Previous Work. Speculations . Apologia ...... iii . . v vi 1 3 6 6 7 10 1 1 12 1 3 1 8 21 21 22 22 23 23 23 25 29 37 42 42 49 52


TABLE OF CONTENTS (contd) Page SUMMARY . . . . 54 CONCLUSION. 56 REFERENCES CITED. . . 57 APPENDIX A Special Chi-square Technique. 62 APPENDIX B Sample Calculations . . . . 65 i v


Table 1 2 3 LIST OF TABLES Description of Stations Summary of Collections, Dates and Procedures . . . Summary of Observed Patterns. v 9 1 4 38


LIST OF FIGURES Figure 1 Geographical location and major features of the Anclote Anchorage . . . . 4 2 Location of fish sampling stations in the Anclote Anchorage (4-station procedures) 2 3 Maximum and minimum surface water temperatures recorded during pre-dredging 4-station sampling procedures, presented a) in chronological order and b) as a compound year. . . . . . 24 4 Seasonal density patterns--Trammel net a) overall pattern, b) diurnal components, c) relative dispersion in overall pattern and chi-square comparisons of d) adjacent months, e) and f) alternate months . . . . . 26 5 Seasonal density patterns--Otter trawl a) overall pattern, b) diurnal components, c) relative dispersion in overall pattern, and chi-square comparisons of d) adjacent months, e) and f) alternate months . . . . . 27 6 Habitat (station) density patterns-Trammel net a) day night and overall patterns, b) relative dispersion in overall patterns, c, d) chi-square comparisons between habitats, e) chi-square comparisons diurnally within habitats. 30 7 Habitat (station) density patternsOtter trawl a) day, night and overall patterns, b) relative dispersion in overall patterns, c, d) chi-square comparisons between habitats, e) chisquare comparisons diurnally within habitats . . . . . . . 31 vi


LIST OF FIGURES (contd) Figure 8 Seasonal diversity patterns--Trammel net a) overall pattern, b) diurnal components, c) relative dispersion in overall pattern and chi-square comparisons of d) adjacent months, e) and f) alternate months .................. 33 9 Seasonal diversity patterns--Otter trawl a) overall pattern, b) diurnal components, c) relative dispersion in overall pattern and chi-square comparisons of d) adjacent months, e) and f) alternate months . . . ........... 34 10 Habitat (station) diversity patterns-Trammel net a) day, night and overall patterns, b) relative dispersion in over-all patterns, c, d) chi-square compari-sons between habitats, e) chi-square comparisons diurnally within habitats ... 35 11 Habitat (station) diversity patternsOtter trawl a) day, night and overall patterns, b) relative dispersion in overall patterns, c, d) chi-square comparisons between habitats, e) chisquare comparisons diurnally within habitats ................. 36 12 Comparison of seasonal density patterns for a) the present studys overall trawl data (UECR-#), b) Gunter (1938) catch per trawl for 1932, c) Gunter (1938) catch per trawl for 1933, d) Reid (1954) trawl catch per month reduced to catch per trawl, e) Springer and McErlean (1962) seine catch per month, and f) Massman (1962) trawl catch per month reduced to catch per trawl ................. 43 vii


Figure 1 3 LIST OF FIGURES (cont'd) Comparison of seasonal diversity patterns for a) the present study's overall trawl data (UECR-SP) b) Gunter (1938) species trawled per month for 1932, c) Gunter (1938) species trawled per month for 1933 d) Reid (1954) species trawled per month, e) Springer and McErlean (1962) species seined per month, f) Massman (1962) species trawled per month .... vii i 47


1 INTRODUCTION The Marine Science Department of the University of South Florida, with the support of the Florida Power Corporation, has been examining the Anclote River estuary on the Gulf Coast of Florida since August 1970. The Anclote Environmental Project is a multi-discipline effort to characterize and document the environmental conditions before power plant construction, to monitor changes associated with construction and to assess the ecological impact of the plant operation after completion. Patterns of fish distribution are an expression of, and result from, organisms interactions with their environments (Hutchinson, 1953) and are a fundamental aspect of an areas ecology. This report presents a part of the ichthyofaunal section of the Anclote Project and is a first order analysis of the patterns of distribution of numbers of fish and of diversity as numbers of species with respect to seasonal, habitat, and diurnal factors. It defines these patterns in the 11before11 condition of Anclote for comparison and evaluation with data after plant start-up, is part of the first study of the fish at Anclote, and is the only study on the Gulf .Coast of Florida to examine these patterns diurnally with more than one type of gear.


2 While this study deals with fundamental ecological parameters, the results cannot be far removed from the methods used to obtain them. That different gears result in different catches both in terms of kinds and number of fish is well known, and at least one author (Margetts, 1969) suggests that fish behavior be investigated on the basis of these differences. The trammel net and otter trawl used for this study are very different types of gear, each subject to its own particular sampling bias. Patterns shown by data from both types of gear would be more likely to reflect the actual patterns of distribution than a pattern deduced from data obtained by one type of gear. Conversely, differences between the patterns displayed by the data from each gear, considered in terms of the biases of the gear, would be expected to elucidate specific components of these patterns. This study, therefore, treats the data obtained from each gear as separate but parallel lines of evidence.


3 AREA DESCRIPTION The study area, on the Florida Gulf Coast at the mouth of the Anclote River, is a region some 8 km. (5 n.m.) square centered upon latitude 28 11' Nand longitude 82 49' W. This area includes the Pinellas-Pasco county line about 30 miles north of the entrance to Tampa Bay. It is generally shallow with depths seldom exceeding 3 meters (10 feet, M.L.W.) and is bounded offshore by Anclote Key. The Anclote River is a shallow meandering stream, draining an area of about 29,000 hectares ( 113 sq. miles, Mohler, 1962). Stream flow is low and varies seasonally in response to rainfall. In the estuarine portion of the lower river, tidal influence is predominant and little vertical stratification is present (Baird et 1972}. A fossil fuel power plant, located on the north shore of the river, will draw its cooling water from the river and discharge the heated effluent in the anchorage to the northwest (see Figure 1). Dredging for cooling water canals began on June 12, 1973. This marked the end of the "undisturbed condition," and provided a tangible boundary for baseline data. Detailed descriptions of the area are available in annual project reports (Humm 1971; 1972; 1973;


N f.Gl_ Land L1l Figure 1. Geographical location and major features of the Anclote Anchorage. 4


1974). Briefly, it may be described as a barrier island estuary. Sea grasses dominate the shallow margins to a depth of 1.5 M (5 feet, M.L.W.) while the central deeper areas are basically sand bottoms. 5


MATERIALS AND METHODS Preface 6 Fish sampling began in August 1970, and through stages of reconnaisance and exploratory fishing evolved to a more formal program. The primary factors considered in establishing the formal sampling program were: 1. The effect of habitat upon the distribution of fish both with respect to depth and substrate. 2. The seasonal variation in distribution of fish. 3. The diurnal variation in distribution of fish. 4. The requirement for baseline to assess future impact. 5. The effect of gear bias. 6. The desirability of replicate samples. 7. The limitations of project resources. 8. The operational characteristics of the available fishing gear. 9. The need for quantification of the data. Resolution of these requirements, lead to the establishment of the monthly, two-gear, four-station program with replicated day and night samples described below. Additional seine and fyke net programs assess near shore and river habitats but are not reported here.


7 Sampling Stations Inasmuch as grass and sand formed the two basic substrates in the Anclote Anchorage, and that each existed over a depth range, four stations was the minimum number required to obtain habitat coverage (i.e., deep and shallow grass, deep and shallow sand). The requirements for baseline, and assessment of future impact suggested that control stations were necessary. Strict controls would require duplication of each of these stations. Unfortunately four was the maximum number possible within the project resources. Accordingly, four stations were established in the autumn of 1971, two of these in areas of high probability of future power plant influence, and two in areas not likely to be directly influenced. At the time these stations were established, difficulty was experienced in locating a 11shallow sand11 habitat, and the only area suitable was marked and sampling proceeded. However, the following spring and summer revealed this station to be seasonally covered by grass. All stations are located by cross-bearings on fixed objects, and are marked by buoys or stakes for ease of location while sampling. Figure 2 shows the location of the four stations and Table 1 gives station descriptions developed from field notes, published 1973), and unpublished data (Rogers, 1974). Stations II and III, deep sand and deep grass respectively, are in an area adjacent to the outfall canal and thus have a high probability of future


/ I --' .. 0 1 2km. I I I I 0 5 1 nm. 24 23 22 21 20 19 18 17 16 15 14 13 12 I I 10 09 08 07 06 05 04 LEGEND SHALLOW GRASS STATION : DEEP SAND STATION DEEP GRASS STATION SEASONAL GRASS STATION Figure 2. Location of fish sampling stations in the Anclote Anchorage (4-station procedures) 8 25 1 8 17 16 15 28-11' N 14 13 1 2 05


TABLE 1 9 Description of Stations Station Depth m.* Habitat Descriptions** I I I III IV 0.7 1.8 1.5 0.6 Shallow Grass Flora: Dense mixed seagrasses, Syringodium filiforme dominant, by wt.; 1220 emergent stems/ m ) Diplanthera wrightii, and Thalassia testudinum also present. Sediment: Muddy sand. Remarks: This station is south of the river mouth, and is unlikely to receive thermal effluent. Deep Sand Flora: Beyond outer margin of seagrasses, a few very sparse patches of Diplanthera may be present. Sediment: Hard sand. Remarks: This station is in open water approximately 1 km. from the outfall canal and may receive thermal effluent. Deep Grass Flora: Moderately dense mixed seagrasses. dominant by wt.; 48 emergent stems/M ) with Diplanthera a strong (32.5% wt.; 485 emergent stems/m ). Sediment: Muddy sand. Remarks: This station is in open water .approximately .75 km. from the outfall canal and may receive thermal effluent. Seasonal Grass Flora: Seasonally variable with moderately dense Thalassia in warmer months, bare sand in colder months. Sediment: Sand and shell. Remarks: This station is adjacent to Anclote Key and is unlikely to receive thermal effluent. *mean 1 ow water. **developed from field notes, publishe d (Zimmerman, et al., 1973) and unpubl i sh e d data (Rog ers, 1974). Seagrass CfenSTty and bi amass \'lere taken at or near yearly maximum (24 June-9 Sept., 1972).


power plant influence. Stations I and IV, shallow grass and seasonal grass respectively, are rather remote from the outfall area and thus are not likely to be directly affected by the power plant. Sampling Gear 1 0 Considerations of sampling bias suggested that more than one type of gear be used for sampling each station, and that the gears used should differ as much as possible in their biases. In addition, the gears chosen should be suitable for operation at the selected stattons and be readily available. Two types, an otter trawl and trammel net, were available. Exploratory fishing indicated that these were suitable for operation at our stations, and that they differed markedly in what they caught. The trammel net is designed to remain immobile and to sample the water column, whereas the otter trawl is drawn across the bottom and captures those nearbottom fish which are unable to escape the net. Thus their characteristics are very different, and it is to be expected that their sampling biases would correspondingly differ. Both can be operated in the same habitat to provide two measures of the fauna present in that habitat. Each of these two measures are subject to an individual sampling bias, but inasmuch as these biases are different, patterns confirmed by data from both would be likely to reflect actual patterns of distribution in the sampled areas.


The otter trawl used was a commercial nylon ''Try ... net11 3.05m. (10 feet) between wings and of 3.5 em (1 .... 3/8 inch) stretch mesh. Two trammel nets were used; both were 91.4 meters (300 feet) in length with #17 nylon inner layers of 5.1 em (2 inch) stretch and #16 nylon 30.5 em (12 inch) stretch mesh outer walls. Both nets were dyed green 1 1 to reduce visibility and were weighted to fish on the bottom. One net was of 2.4 meter (8 feet) depth for use at the deep stations (stattons II and III), the other of 1.2 meter (4 feet) depth for use at the shallow stations (stations I and IV) Sampling Methods The need for quantified baseline data suitable for comparison with future data dictated that the methodology be as well defined and as closely controlled as possible within the constraints of field operation. The trammel nets were set in a straight line down-wind for a period of 45 minutes between end of set and beginning of haul. However, the time used for calculation of fishing effort was measured from the mid-point of the set to the midpoint of the haul, in effect the time an average piece of net was in the water. This ttme varied wtth the amount of catch; the more fish caught the more time is spent hauling the net. The pertod ts therefore nominal. Trawls are made parallel to the trammel net at a dtstance 20 to 30 m. from the trammel net. Trawling speed was approximately 1.25 m/sec


12 (2.5 kts.) and the distance towed was the length of the trammel net 91.4 meters (300 feet). The trawl speed was measured through the water and the times recorded to the nearest minute at beginning and end of each tow. The actual over-the-bottom speed varies as a result of water currents; however, all trawls were close to two minutes in duration and this time interval was used for calculation of fishing effort. The requirement for replicates indicated that at least two discrete samples be taken at each occupation of a station. Two trammel net sets and four trawls were taken at each such occupation. Each trammel was associated with two trawls, these being made in opposite directions to reduce (average) the effects of current upon the over-the-bottom velocity. All stations were occupied twice (once during the day and again at night) during a sampling period, in order to obtain measures of diurnal variation. Further details of gear and methods are available in the Anclote Annual Report for 1 9 71 ( B a i rd 1 9 7 2) Scope of Data Of the 838 Anclote fish collections completed before dredging began, 511 were part of the four station study described here. The four stations were originally intended to be visited on a monthly cycle, but, due to manpower availability, equipment breakdown, and weather, some 20 months (October 1971 to May 1973) were required to accomplish


sampling for eleven complete data sets. Details of the sampling dates, numbers of trawl and trammel samples and references to Anclote fish project collection numbers are given in Table 2. Data Reduction and Management Data for the monthly four-station procedures were reduced to unit-effort catch rates by dividing the total catch (in terms of numbers of fish or number of species) for each sample by the effort expended for that sample. The units of effort chosen were 11minute of trawling" (2 minutes per sample}, and 11ten minutes of trammel set11 as measured from mid-time of set to mid-time of haul (generally 50 to 90 minutes per sample}. The reduced data for each collection were averaged for each occupation of a station (day or night) and these values in turn were averaged as a data group for the appropriate analysis. Calculations were made for day, night and total catches to illustrate diurnal, as well as seasonal, variability. The numbers of fish and species in a habitat (by gear type) were similarly calculated on a day, night and overall basis. When a population is known to be heterogeneous, and when that heterogeneity has a bearing on the characteristics being studied, the population may be divided into strata and random samples of units drawn from each stratum 1967, p. 26). Since our data were known (from field observations and preliminary data analysis} to be heterogeneous with respect to habitat, seasonal, and diurnal factors, the strata were 1 3


14 TABLE 2 Summary_ of Collections, Dates and Procedures a) Chronological Order Number of Data Samples Anclote Fish Set Month Sampling Period Dates Trawl Trammel Collection# 1 1 011 26 October 6 November 1971 31 16 1 7 9 227 2 2 2-15 February 1972 32 16 229 276 3 6 13-27 June 1972 32 16 328 375 4 7-8 24 July-2 August 1972 30 15 376 420 5 8 24-29 August 1972 30 15 421 465 6 10 9-18 October 1972 32 16 482 505, 507 530 7 11-1 2 27 November6 December 1972 32 16 566 613 8 1 3-10 January 1973 32 16 625 672 9 3 5-15 March 1973 32 16 690 737 10 4-5 23 April-4 May 1973 30 15 738 782 1 1 5 20-22 May 1973 27 14 783 814, 816 824 Total 340 1 71 b) Compound year (monthly or seasonal order) Month Year Data-Set 1 1973 8 2 1972 2 3 1973 9 4-5 1973 1 0 5 1973 1 1 6 1972 3 7-8 1972 4 8 1972 5 10 1972 6 1 011 1971 1 1 1 -1 2 1972 7


chosen as those samples from a set of procedures within a month, station, and diurnal period (i.e., either day or night). The standard error of the mean for our data was arrived at by use of the formula for stratified samples given below (Arkin and Colton, 1970, p. 147). Where N=total number samples all strata N =number of samples in si stratum i 15 S =standard error of the mean for stratum i 1 S =standard error of the mean for stratified samples This standard error of the mean was then used for calculating the measure of variability presented below The relative dispersion as defined below (Spiegel, 1961, p. 73) is a method by which the dispersion of dissimilar data may be compared. Values were calculated for overall seasonal and overall habitat data, using the standard error of the mean as the measure of absolute dispersion, and the arithmetic mean as the average. Relative Dispersion = Absolute dispersion = s average __x_ X Each of the eleven data sets was (as indicated above) planned to consist of 16 trammel and l2 trawl samples, for a total of 176 trammel and 352 trawl samples. However, occasional samples were missed due to temporary interruptions in the


16 collecting program resulting from extreme weather conditions. These missed samples accounted for about three percent of the planned samples (i.e., 5 of 176 trammel samples, and 12 of 352 trawl samples). Calculation of the stratified standard error of the mean (above) requires a minimum of two entries per strata. Each trawl strata consisted of (nominally) four samples, and since no more than two trawl samples were missed in any one strata, estimated values were not required to obtain this minimum number of entries (above). However, the trammel strata consisted of (nominally) two samples and consequently a missed sample required estimated data. The following technique was used to obtain these estimates. In each case one of the pair of values for the stratum was missing. The existing value was compared with values of similar magnitude for complete strata. The strata chosen for comparison were those within the same season (or data set), within the same diel period (day or night) and within the same station at a different season. The standard deviations for the strata with values of comparable magnitude were noted and the arithmetic mean of these used to obtain the estimated datum by back calculation. In practice, this was done by serial approximations of the missing and calculation of the standard deviation (Wang 700 program 1971A/ST5 statistical package) for each approximation. Inasmuch as the given standard deviation could be obtained by choice of an estimated value either above or below the existing value, the following guideline


1 7 was followed. If the missing datum was a night value the choice was made below the existing value, conversely for day values. This was conservative in that it tended to reduce the expected day-night differences 1974). Estimated values were not used for the chi-square analysis described below. The distribution of catch rates within the groupings analyzed varied markedly. Visual inspection of the data showed that most distributions were decidedly although a few appeared near normal in shape. Chi-square was therefore chosen as the test statistic. This method offers the advantage of comparing distributions without assumption as regards their shapes, thus allowing a consistent method of comparison for all the data groups. However, the choice of chi-square as a test for differences between data groupings involves some loss in precision or more correctly in our ability to detect fine differences. The chi-square procedure requires that the data to be formed into frequency of occurrence classes. The chi-square statistic itself, and the methods of calculation, g i ven a set of frequency-classed data, are well discussed by most statistical texts; however, little direction was available as to the methods of constructing the class boundaries. Therefore a formalized data-defined system for establishing floating class boundaries was developed (see A p pend i x A for details, assumptions, and derivation of system). Briefly an algorithm was developed to determine the


18 minimum number of entries (from both members of the comparison) required for each rate class. The catch rate data were plotted on frequency diagrams and the data points counted off from the higher to the lower catch rates, boundaries being established whenever the total number of entries equalled or exceeded the minimum. The frequencies were then entered in a 2xR table and the chi-square value calculated by use of the formula given by (1967, p. 597-8). The probability that the calculated chi-square value was not due to random factors was obtained by use of the Wang 700 computer program 1991A/ST2 {part of statistical package 1971A/ST5) and was converted to a percentage. Comparisons were made between adjacent data sets within the compound year, between alternate months (i.e., in steps of 2), between stations, and between day and night within stations. For samples of calculations and procedures see Appendix B. Data Presentation Each data set was numbered sequentially {1-11) in the chronological order in which the data sets were taken. Because stations were not sampled each month during any one calendar year, data from all months are combined to form a compound year (see Table 2). Presentations of the yearly temperature pattern and seasonal patterns of fish abundance and diversity are based on the compound year. Unless


otherwise specified, 11year11 will refer to the compound year. Typically, each sampling procedure required two 24-hour periods and was generally completed within two weeks. For each of these two 24-hour periods the high and low surface water temperatures recorded, have been abstracted to illustrate the temperature ranges. 19 The densities of fish are expressed as unit effort catch rates by numbers of fish (UECR-#), and diversity as unit effort catch rates by numbers of species (UECR-SP). Density, diversity and relative dispersion are plotted as annual curves for seasonal analysis and as vertical bars for habitat analysis. The seasonal data points are plotted at the midpoint of each sampling period, and the data points are coded as to calendar year with the 1971 data indicated by a square, 1972 data by a circle and 1973 data by a triangle. The results of our chi-square analysis are given as a set of graphs each associated with its appropriate comparisons. Each chi-square probability is indicated as a horizontal line connecting the two groupings being compared, the vertical location of the line indicating the chi-square probability obtained for that comparison. The presentations of seasonal data (Figures 4, 5, 8 and 9) each contain three graphs of chi-square comparisons (11d,11 11e,11 and 11f.11) Graph 11d11 presents the results of chi-square comparisons between adjacent months (i.e., January-February, February-March, etc.). Graphs 11e11 and 11f11 present the results of comparisons between alternate months. These are


20 separated into two graphs to avoid crowding. Graph 11e11 gives the comparisons January-March, March-May, etc., while 11f11 gives the comparisons February-April, April-June, etc. Presentations of habitat data (Figures 6, 7, 10 and 11) also contain three graphs of chi-square comparisons (11C,11 11d,11 and 11e11). Graphs 11C11 and 11d11 present the results of comparisons between stations, and are again separated to avoid crowding. Graph 11C11 gives the station comparisons I-II, II-III, III-IV, while graph 11d11 gives the comparisons I-III, I-IV, II-IV. Graph 11e11 presents the results of day-night comparisons within station.


21 SAMPLING BIAS AND LIMITATIONS Sampling Program The samples are limited to eleven data sets spread over almost two years (20 months), the spacing of the data sets is unequal and no sampling took place in September. The large time gaps in the data were reduced by the compound year approach but, this necessarily involves the loss of indication of between-year variability except as or when it appears as an aberrant point on the yearly curve. The use of two gear types reduces the problem of gear bias but does not remove it. There are a number of problems present in the two gear approach. Most notable perhaps is the under-sampling of certain schooling species, mullet in particular. The behavioral characteristics of these fish are such that they are not taken by our gear. Other species may be over sampled, notably the sharks which may be attracted to the trammel net by fish already caught in the net. Despite these problems the approach is a major improvement over most single gear studies. The use of four stations results in a very coarse view of habitat patterns of distribution and much fine detail is, no doubt, missed. However, the choices of 11Characteristic habitats11 provides some basis for generalization. A similar


limitation exists with relation to diurnal patterns. The samples could be characterized as being 11mid-day11 and 11mid night,11 and thus do not assay patterns related to the twilight periods. Data The treatment of the data as catch rates may limit the details of analysis to some degree. The present data processing capability limits us and future analysis is planned. However, this study represents the necessary first order look at the overall patterns. Finally, it should be noted that our trammel data have some internal damping. Increased catches result in increased effort, due to the greater time required to haul the net. Therefore, catch rates tend to increase more slowly than do the number of fish caught per set. Analysis Because of the large number of species (125 identified thus far) in our collections, the subsequent analysis and discussion will treat overall pattern, and not details of individual species distributions. Distribution, life histories and feeding ecology of the more abundant species are presently being investigated and are subjects of a number of masters theses. 22


RESULTS Preface Much of the present data has been presented in part elsewhere: reports by (1971; 1972; 1974) and Rolfes (1972; 1974) all deal with some part of the following material. However, all of these were either preliminary in nature or in some way sub-sets of the present report. The present study examines the data in terms of patterns in the density and diversity of the ichthyofauna 23 with respect to season, habitat and diurnal factors, treating each sampling method separately but in a parallel fashion. Surface Water Temperatures The highest and lowest temperatures noted, at any one of the four stations, for each sampling day during the 20 calendar-month interval are given in chronological order in Figure 3(a) and as the compound year in Figure 3(b). Highest day and night temperatures were recorded during the summer months of July and August; the lowest temperatures in February. Both the greatest and the least variation within a sampling day were observed in the winter months with a sampling day in early February (data set 2) showing essentially no difference, and a sampling day in March (data set 9) showing a diurnal range of 4.5QC. Note that the temperature


30 u 25 0 Q 20 E-t 15 10 MONTH: DATA SET: a ) Surface wate r temperature, chronological order. 30 25 u 20 15 10 MONTH: IH J lr: I r-:1 I A'"" M H I J .... \1 s ... N"" D I DATA SET:B 2 9 lOll 3 4 5 6 l 7 b ) Surface water temperature, com p ound year. =1971 DATA =1972 DATA =1973 DATA Figure 3 Maximum and m i nimum surface water temperatures recorded during pre-dredging 4 station sampling procedures, presented a) in c h ronological order and b ) as a comp ound year. 24


25 rise from winter low values to summer high is relatively gradual and displays much variability, requiring some three months of the compound year to increase from approximately 15-17C to 24C. These summer temperatures were maintained for some 5 months. The subsequent drop from summer values (25 range) was more abrupt, occurring in little more than a single month. Patterns of Fish Density Figures 4 through 7 present our density data (as numbers of fish taken per unit effort, UECR-#). Relatively consistent patterns of seasonal and diurnal variation were observed overall but differences associated with gear type are noted. Greater numbers of fish were taken during the warmer months. The summer high values for the trammel procedure (Figure 4) persisted from late May (data set 11) through late October early November (data set 1). The trawl data (Figure 5) showed the summer highs persisting from late July-early August (data set 4) through December (data set 7). The decrease in the fall for the trammel procedures occurred between November and December (data sets 1 and 7) while the major trawl decrease was between December and January (data sets 7 and 8). The summer peak therefore is sharper and narrower in the trawl data, the summer values lasting for some 4 months, as compared with 5 for the trammel data which parallels the temperature curve. In addition, the trawl peak begins 2 months later in the year. Perhaps the most striking feature is the abrupt


1 F M lA M J IJ A 1 s lo N D a) Trammel net overall seasonal pattern, UECR-#. b) Diurnal components of overall trammel 26 net seasonal patterns, UECR-#. NIGHT DAY F MIA 1M 1JIJ A s lo N D c) Relative dispersion in overall trammel data, UECR-#. 0. MONTH lv J ..,_F If M I A .,.1. MT I -:;_I J ... A J s 1.3 N D I DATA SET 8 2 9 10 11 3 4 5 6 1 7 100% 50% f.t t----t "'""'i d) Chi-square ooi .........--,.__ arisons of campadjacent [ months. :: I j I g5> e) & I _ ..J=-L_ 95% 50% f) Chi-square .: : : : : I arisons of u months. DATA SET 8 2 g lO 1\ 3 4 5 G i 7 campalternate Figure 4. Seasonal density patterns Trammel net a) overall pattern b) diurnal components, c) relative dispersion in overall pattern and chi-square comparisons of d) adjacent months, e) and f) alternate months. See text for further details.


lS 10 I p:; s ::::> 0 MONTH 0. MONTH DATA SET J 1.,f I "TM I A.;. M, l J, I J -1 A Tl s I$ + N + D I 8 2 9 10 11 3 4 s 6 1 7 a) Trawl overall seasonal pattern, UECR-#. b) Diurnal components of overall trawl seasonal UECR-#. c)' Relative dispersion in overall trawl data, UECR-#. (J) Q) .-i +J .-i .--i .-i .Q Ill .Q 0 0.. 10! : : 9S% SO ......,_. ....,__. d) Chi-square 10 0 r---1 camp adjacent Q) Ill & (J) I .-i .c u DATA so -: g I SET 2 10 11 3 4 S 9S% : I : e) Chi-square compa risons of alte r n a t e months. -"'"===< -9S% I 6 1 f) Ch i-square c omp arisons of alternate months. Figure S Seasonal density patterns -Otter trawl a ) overall pattern, b) diurnal components, c) relative d ispersion i n overall pattern, and chi-squar e comparison s of d) adjacent months, e ) and f) alternate months. See text for f urther details 27


decrease in density during the fall indicated by the trawl data. 28 The trawl data consistently display higher night than day catch rates while the trammel data, though they too reflect generally higher night catches, on occasion exhibits higher day rates. The high overall trawl values for November and December (data sets 1 and 7) are entirely functions of the night catch rates in these months. Chi-square analysis confirms the intuitive interpretation of the graphs. The trammel data (Figure 4) are relatively constant over the summer peak (except for a dip or depression noted in July-August (data set 4), and comparisons have correspondingly low chi-square probabilities. The sharper slope in the fall data has its correspondingly higher chi-square probabilities as compared with spring data. Chi square analysis of trawl data confirms the visually apparent pattern of more pronounced differences and shorter duration of peak (Figure 5). Chi-square of the trammel data considered by habitat (Figure 6) indicates a relative similarity of all the stations, the chi-square probabilities indicating no differences at the 95% level. Diurnal variation is present but not markedly so; only the day-night comparison for the shallow grass (station I) yields a probability exceeding 95%. In contrast, the trawl data (Figure 7) show more well defined (stronger) patterns. The 11deep-sand11 station (station II) differs strongly from all the others and yields


29 probabilities of 100%. The difference between the shallow grass (station I} and seasonal grass (station IV) is somewhat less marked and yields a chi-square probability of 98.8%. Inasmuch as the overall UECR1s for these stations are similar, the difference appears to be a function of the diurnal components present at each station. All stations had relatively high (over 95%} chi-square probabilities for the day-night comparisons within stattons. The diurnal pattern was strongest in the shallow grass (station I) and deep sand (station II} with chi-square probabilities of 100%, somewhat weaker in the deep grass (station III) with a probability of 99.96% and weakest in the seasonal grass (station IV) with a probability of 95.7%. Patterns of Fish Diversity Figures 8, 9, 10 and 11 present our diversity data (as numbers of species taken per unit effort, UECR-SP). The seasonal pattern of diversity generally (Figures 8, 9} parallels that found for density but is, perhaps less sharply defined. The decrease in density during the summer, noted in the trammel data is not paralleled by a decrease in diversity. Again the trammel data more closely follow the temperature curve than do the trawl data. The trawl data indicates summer high values from late July-early August (data set 4) through late November (data set 1) a period of three months, and the ensuing decrease to winter values was over a two-month period, early November (data set 1} through


30 9 --8 --7 6 ---a) Trammel net habitat patterns, day, night and overall1UECR-#. --3 -0--2 1II-II-I; 1 0 D N 0 D N 0 D N 0 D N 0 -I II III IV STATION:I II III IV h l::J 1--1 J ,......--95% H 11 DN D N DN D N STATION:I II III IV DAY NIGHT OVERALL b) Relative dispersion in overall trammel data1UECR-#. c) Chi-square comparisons between habitats, overall data. d) Chi-square comparisons between habitats, overall data. e) Chi-square comparisons diurnally within habitat. Figure 6. Habitat (station) density patterns Trammel net a) day, night and overall patterns, b) relative dispersion in overall patterns, c,d) chi-square comparisons between habitats, e) chi-square comparisons diurnally within habitats. See text for further details.


2 01 50-5--n. D N 0 STATION I I"' D N 0 D N 0 D N 0 II III IV n-1a) Trawl habitat patterns, day, night and overall, UECR-#t. DAY NIGHT OVERALL 31 : :j--J--......t!+----+1 J-+----1-1 t-1--------t--1 I b) Relative dispersion in overall trawl data1UECR-#. STATION:-I II III IV :I ] J B 1 s::r :1 : I ] : 95% 0 0 0 0 STATION:-I II III IV -95% DN D N DN D N STATION:-I II III IV c) Chi-square comparisons between habitats, overall data. d) Chi-square comparisons between habitats, overall data. e) Chi-square comparisons diurnally within habitat. Figure 7. Habitat (station) density patterns -Otter trawl a) day, night and overall patterns, b) relative dispersion in overall patterns, c,d) chi-square comparisons between habitats, e) chi-square comparisons diurnally within habitats. See text for further details.


early January (data set 8). The diurnal components of the diversity data are somewhat different from those of the density data, the night UECR-SP, being more consistently higher than the day rate. However, the difference in the day and night UECR-SP is perhaps less marked. Note that the variability in the relative dtspersion for the diversity data seems to be less than is the case for the density data. In addition, the diversity relative dispersion appears to bear an inverse relation to the overall seasonal curves for both the trammel and trawl. The pattern of diversity with respect to habitat, in 32 the trammel data, (Figure 10) appears more pronounced than, but remains similar to, the pattern of density. However, chisquare analysis indicates probabilities exceeding 95% in 4 of the 6 comparisons between stations (excepting only the comparisons of station and I-IV}. Diurnal variation within habitat is also increased as compared with the density data. The shallow grass (station I} and deep sand (station II) both have diurnal differences large enough for chi-square probabilities exceeding 95%. In contrast to the trammel results, the habitat patterns of diversity in the trawl data (Figure 11) are less pronounced than for density. This is visually apparent from the graphs and is confirmed by the chi-square analysis. The only between habitat comparisons yielding probabilities of difference exceeding 95% are those between the deep sand (station II) and all other habitats. Further, while diurnal


1.5 p.. (fJ 1.0 I p:: u l:il ::l 5 0 MONTH 1.5 p.. (fJ I 1.0 p:: u l:il ::l .5 0 MONTH .101 o:J -----------MONTH DATA SET a) Trammel net overall seasonal pattern, UECR-SP. b) Diurnal components 33 of overall trammel net seasonal patterns, UECR-SP. c) Relativ e dispe rsion in overall trammel data, UECR-SP. Chi-square comp arisons of adjacent months. e ) C hi-square com p arisons o f alt ernate m o n t h s f) Chi-square c omp arisons o f alternate m onths. Figure 8. Seasonal diversity patterns Trammel net a) o verall pattern b) diurnal components, c ) relative dispersion in overall pattern and chi-square comparisons o f d) adjacent months e ) and f) alternate months See text for further d etails.


MONTH j A S I 0 N D a) Trawl overall seasonal pattern, UECR-SP. 3 DAY ----NIGHT ,. ....... I \ -f)) 2 I hl ::J l 0 j MONTH .101 th>< 0 05 0.0 MONTH DATA SET Ul Q) .-I .j.J .-I .-I .-I ..Q 0 100 1-1 50% 0. Q) I J (J T 8 i 1o8I I 50% .-I ..c: u 0 F IF "T" 2 f DATA SET 8 2 M IM T 9 lA j IJ I M J A s lo N D b) Diurnal components of overall trawl seasonal patterns, UECR-SP. c) Relative dispersion in overall trawl data, UECR-SP. IA.,LM IJ IJ-tA ,s T T t 1$ +N +D lOll 3 4 5 6 l 7 1--t 1---t d) Chi-square camparisons of adjacent months. :::::: -1 e) Chi-square comp arisons of alternate months. 1-:-.:-1 : : f) 6 1 7 Chi-square comp arisons of alternate months. 9 10 ll 3 4 5 Figure 9. Seasonal diversity patterns Otter trawl a) overall pattern, b) diurnal components, c) relative dispersion in overall pattern and chi-square comparisons of d) adjacent months, e) and f) alternate months. See text for further details. 34


-; 1.5 -1-1.0 ---.5 0 D t 0 STATION I STATION:I ,.. D N 0 D 0 II III II III I Jl 0 IV IV a) Trammel net habitat patterns, day, night and overall,UECR-SP. DDAY aOVERALL b ) Relative dispersion in overall trammel data,UECR-SP. c) Chi-square comparisons between habitats, overall data. d) Chi-square comparisons between habitats, overall data. Bh t--4 J H -95% 0% I I I I I I I D N D N D N D N e) Chi-square comparisons diurnally within habitat. STATION:I II III I V Figure 10. Habitat (station) diversity patterns Trammel net a) day, night and overall patterns, b) relative dispersion in overall patterns, c,d) chi-square comparisons between habitats, e) chi-square comparisons diurnally within habitats. See text for further details. 35


3.0 Jil< U) 2. 2. J: 1 u 1. -5-05-o--5-fl STATIO N:STATION:D N 0 I I n D l'i 0 'r N 0 II III II III N 0 I v IV a) oDAY Trawl habitat patterns, day, night and overall, UECR-SP. 1NIGHT a-OVERALL b) Relative dispersion in overall trawl data1UECR-SP. :--+------rl :-+-------r-1 95% -: 1 :J t--95% 6 t[ I 0 0 0 0 c) d) Chi-square comparisons between habitats, overall data. Chi-square comparisons between habitats, overall data. STATION:I II III IV -95% DN bJ D 'N STATION:I II III IV e) Chi-square comparisons diurnally within habitat. Figure 11. Habitat (station) diversity patterns Otter trawl a) day, night and overall patterns, b) relative dispersion in overall patterns, c,d) chi-square comparisons between habitats, e) chi-square diurnally within habitats. See text for further details. 36


differences within habitat for the shallow deep sand, and deep grass (stations l, II, III) remain strong (98-100%}, the difference for the seasonal grass (station IV} is less marked, and yields a relatively low chi-square probability (89%). Overall Considerations of Pattern A number of features are present in our data which, although not formally confirmed by or dealt with by chi-square analysis, deserve consideration. 1} The relative homogeneity of t h e trammel as compared with the trawl data (described above), and the differences tn the patterns revealed by the two methods should be noted. 2) Table 3a summarizes some of the major features of our seasonal patterns. The shorter and more delayed summer peak for the trawl data is clear. The parallel between the trammel and temperature data is clear as is also the abrupt decrease of temperature, density and diversity in the fall. 3) The overall relative ranking o f the habitats (stattons} in terms of density and diversity by the different gears is given in Table 3b. Except for the shallow and deep grass hab itats (stati ons I and III), which are quite similar in catch rates in any case (see Figures 7 and ll), 37


38 TABLE 3 Summary of Observed Patterns a) Seasonal patterns, location and duration of features SUMMER HIGH VALUES Begin End Month Month No. of Pattern (data set) (data set) Months Water Temp. May (11) Oct-Nov(l) 5 Dens Trammel May(ll) Oct-Nov(l) 5 Trawl July-Aug(4) Dec (7) 4 Diversity: Trammel May(ll) Oct-Nov(l) 5 T raw 1 July-Aug(4) Oct-Nov(l) 3 FALL DECREASE IN VALUES Begin End Month th No. of Pattern (data set) (data set) Months Water Temp. Oct-Nov(l) Dec(?) 1 Oct-Nov(l) Dec (7) 1 Trammel Trawl Dec (7) Jan(8) 1 Trammel Oct-Nov( 1) Dec(?) 1 Trawl Oct-Nov(l) Jan(8) 2


39 TABLE 3 (cont'd) b) Habitat patterns, rank order of habitats (stations) by UECR OVERALL Pattern high low Trammel Density IV I I I I I I Diversity I I I IV I I I Trawl Density I I I I IV I I Diversity I I I I IV I I DAY Pattern high Tow Trammel Density I I IV III I Diversity I I I IV I I I Trawl II Density IV III I Diversity III I IV I I NIGHT Pattern high low Trammel IV I I I III Density I I I I I I IV Diversity Trawl III IV II Density I Diversity III I IV I I


the overall trawl data is parallel in density and dtversity. The overall trammel data, however show marked differences in rank order between density and diversity. These differences are such that only the shallow grass (station I) has the same rank for both density and diversity. 4) The day and night relative rankings of the habitats (stations) are also given in Table 3b. The trammel data again display marked differences in ranking, both between fish density and diversity and between day and night. The rankings for density change completely between day and night. The deeper habitats (stations II and III) have higher rank during the day than at night, while the shallower stations (stations I and IV) rank higher at night than during the day. For diversity the diurnal change in ranking is less complete with the deep grass (station III) maintaining its high rank in both day and night. Comparison of the rankings by density and diversity for day-time data shows that only seasonal grass (station IV) has the same rank, similarly the night data shows that the shallow grass (station I) has the same rank. The trawl rankings for diversity are identical for both day and night. The density rankings show two habitats with the same ranking (deep sand and deep 40


grass, stations II and III respectively). Comparisons of the density and diversity rankings for day-time show only the deep sand maintaining the same rank. A similar comparison of night data shows only the deep and shallow grass (stations III and I) change rank, and the night rankings are identical to the overall rankings. The deep sand (station II) maintains its lowest ranking in all of the comparisons of trawl data. In fact, the deep sand ranking is also generally low in the trammel data. 41


DISCUSSION Comparison with Previous Work Publications relating to the ecology of estuarine fish faunas along the eastern Gulf of Mexico are few; none of these were intended as baseline studies; none deals with the Anclote area. The most applicable of the existing studies were the surveys conducted by Reid (1954), Kilby (1955), Springer and Woodburn (1960], and Wang and Raney (1971). 42 For the most part, these are and will remain important contributions (if only because of the singularity) and provided background information for our study. These studies largely consisted of systematic accounts, and are valuable in this respect. However, they do not directly address themselves (and for the most part were not designed to address) the definition of the patterns we are concerned with. Other studies further afield geographically have been reviewed. Gunter's (1938} study in Louisiana, (1954) in the Escambta Rtver, Springer and McErlean (1962} at Matecumbe Key, (19731 in the Patuxent River, Dahlberg and Odum (1970} in the Newport Rtver, Georgia, and Tyler (19711 tn Passamaquoddy Bay, among others have addressed the problems of seasonal and habitat patterns. Among these, Gunter (1938), Reid (1954], and Springer


..c: {/).-I ri 4-l Ill l-l =#o.j.l l-l O'> 0. o

44 and McErlean (1962) all present data which are amenable to the extraction of seasonal patterns. Massman's (1962} data for Chesapeake Bay as abstracted and presented by Tyler (1971} are also amenable to extraction for seasonal patterns. Ftgures 12 and 13 compare our trawl density (Figure 121 and diversity (Figure 13} data given earlier wtth data abstracted from Gunter (1938} for Barataria Bay, Louistana, Reid (1954} for Cedar Key, Florida, Springer and McErlean (1962} for Lower Matecumbe Key, Florida, and Massman (1962) for Chesapeake Bay. Gunter, Reid and Massman's density data include some correction for effort. Gunter's data were presented as average catch per trawl. Massman's data were presented per month for 4 trawls, and were converted (by us} to a per trawl basis for comparison. Reid presented his data as total catch, but inasmuch as he also indicated the number of trawls, his data were also reduced to average catch per trawl for comparison. Springer and McErlean's density data and all of the abstracted diversity data are uncorrected for effort. With the possible exception of those of Springer and McErlean, whose seining effort was indeterminate (they sampled until no more new species were obtained), the abstracted data points represent few replicates at any one station. Reid's 106 trawls were spread over 12 months and several stations, Gunter's data represent 1 trawl at each of 3 stations each month, and Massman's data represent 1 trawl at each of 4 stations each month. The higher density of data


45 points in the present study (representing 30-32 trawls spread amongst 4 stations) probably accounts for the smoother curve obtained. Examination of Figure 12 reveals a certain amount of non-uniformity among the results of the various studies, and some interesting para1le1s as we11. The general irregularity of the density data ts of some concern and prevents detailed analysts. It may in part be a result of inadequate sampling, both in numbers of samples and in lack of nighttime sampling. Springer and McErlean's and data all have a density peak in the approximate location of the mtnor May peak found in this study. rn each, the peak resulted from high catches of few species. The peak in the present study was primarily due to Lagodon rhomboides; in data it was again due to a single species, Micropogon undulatus; Reid's peak resulted from a combination of h rhomboides and Orthopristis chrysoptera; Springer and McErlean's from Atherinids; and Massman's from Anchoa mitchilli, Brevoortia tyrannus and Urophycis regius. All of the studies (to one degree or another} parallel the present study's finding of lower densities in the winter and early spring. Finally, Springer and McErlean's seine data seems to show a sharp decrease in catch rates between December and January similar to that found for trawling by this study. The diversity data (Figure 13} within a given study are, in general, smoother than the density data, but differ


46 widely between studies and are not directly correlated. The seasonal patterns tend to be similar to those of the present study but are relatively flat and featureless (possibly a scale effect in presentation). They otherwise present no consistent pattern for analysis, and inasmuch as they are all uncorrected for effort and have few replicates this is unsurprising. The high variability in Massman's data may reflect increased short term seasonality in a more temperate habitat. The remaining papers, although not presenting data as amenable to comparison as those abstracted, do contain information of interest. Springer and Woodburn (1960} and Kilby (1955) both had indeterminate sampling efforts and thus did not provide data which allowed abstraction for direct comparison of overall patterns. However, revtew of their species accounts indicate that seasonality and habitat are major features of the distribution of individual species. D a h 1 b e r g an d 0 d u m ( 1 9 7 0 } r e f e r en c e t h e i r '' T a b 1 e 2 11 a s indicating unit effort seasonal patterns of density and diversity. However, the table seems to have been lost or altered in the editorial process and the only data avatlaole are in their text account of the seasonal patterns of 11Sprtng (42 species}, Summer (41) and Autumn (42} two winter periods (36 and 38).11 HcErlean .!}_. (1973) present quarterly time series data for a 4-year period. The pattern of high in summer, low in winter is present, but because of the widely spaced data points, details are not visible.


MONTH:-p.. Ul I p:; u 1 [il 0 0 lll.C: 20 Q)+J o-1 s:: u 0 Q) s 0.. 10 Ill 1-i Q) '*1:0.. 0 lll.C: Q)+J o-1 s:: 20 u 0 Q) s 0.. Ill 1-i Q) 10 '*1:0.. 0 lll.C: Q) +I 40 o-1 s:: u 0 Ill 1-i 20 Q) '*1:0.. 60 lll.C: Q) +I o-1 s:: 40 u 0 Q) s 0.. 20 Ill 1-i Q) '*1:0.. 15 lll.C: Q) +I o-1 s:: 10 u 0 Q) s 0.. Ill 1-i 5 Q) '*1:0.. 0 J Ml I I Jl F A M J A sl 0 Nl ol a) Trawl, average catch rate, numbers of species (Present._ study). b) Number of species trawled per month, 1932 (Gunter 1938). c) Number of species trawled per month, 1933 (Gunter 1938). d) Number of species trawled per month (Reid 1954). e) Number of species seined per month (Springer and McErlean 1962). f) Number of species trawled per month (Massman 1962) MONTH:-I J If 111 I A IM jJ IJ lA ,s IO IN ID I Figure 13. Comparison of seasonal diversity patterns for a) the present study's overall trawl data (UECR-SP) b) Gunter (1938) species trawled per month for 1932, c) Gunter (1938) species trawled per month for 1933, d) Reid (1954) species trawled per month e) Springer and McErlean (1962) species seined per month, f) Massman (1962) species trawled per month. 47


48 The Wang and Raney (1971) paper on Charlotte Harbor presents a seasonal curve. Although the pattern is dominated by two in the other in November, the apparent winter low period December through March is somewhat similar to th&t found in the present study. However, the aut h o r s con c 1 u de that 11 a bun dance . was r e 1 ate d to temperature and saltntty ... salinity dropped with it the fish catch. Likewise .. with low winter temperatures." The drop from the June peak is indeed, associated with a drop tn salinity albeit under stable temperature conditions. Unfortunately for interpretation, the rise to the November peak coincided with dropping temperatures and relatively stable salinity and the decline from the November peak took place under essentially the same rate of temperature decrease as did the rise. There may well be a temperature-salinity relationship with catch rate but, from the data presented, it remains obscure. General patterns relating to habitat and diurnal variation were more difficult to obtain from the literature. Those papers providing species accounts give details of the individual species preferences, but provide little in the way of overall patterns. On occasion, problems of methodology make attempts at comparison questionable (at best). Springer and Woodburn (1960} for used 4 types of gear and collected at eight stations in their survey of the Tampa Bay region. However, the same gear was not used at each station and some stations were sampled only at night,


others only during the day. Thus between station (habitat) differences were also subject to between gear and diurnal differences. Since, Each.area was collected until it was subjectively cons1dered that additional collecting would not evince new information from the station. effort was variable and, as the authors acknowledge, the resulti"ng abundance figures were subjective ("at best"). The results of our study demonstrattng strong gear and diurnal components of variation would seem to indtcate that their judgement was generous. In addition approximately 20% of the spectes noted were not taken by sampling dures but were the result of sightings, newspaper records, flotsam and commercial landings. 49 Roessler (1965} investigated the variation 1n the trawl catches in Biscayne Bay during the sprtng and summer of 1963. His results, although limited to part of the yearly cycle, in part parallel those of the present study. Catches (by numbers and species) decreased with decreasing sea grass coverage of the bottom, and night catches tended to be higher than day. Speculations A number of our results may be functions of gear ectivity interaction with habitat, diel and seasonal factors. However, since gear selection is a result of both the physica l apparatus and the ecology and behavior of the fish involved, such. phenomena may profitably (Margetts, 1969) be


used to infer something about fish ecology and behavior. Certainly the low trawl catch rates in the deep sand habitat (station II) reflect the paucity of near bottom fauna over sand as compared with grass and the concomitant low catches with gear that selects for these forms. 50 In addition, the relative uniformity of catch rates with the trammel in different habitats (as compared with the trawl) argues for interpretation as a reflection of the relatively continuous uniform nature of the pelagic habitat as compared with the patchiness of the benthic habitat. Further the relative uniformity of trammel catch rates with respect to season implies a relatively greater uniformity in the seasonal availability of larger fish (which it selects for} vs. the smaller forms which the trawl fishes for. The closer parallel of the trammel seasonal pattern with the temperature pattern may reflect the greater ability of the larger forms to move with the weather. The differences between the patterns observed with the trawl and trammel may be (in part} a tion of a fish nursery effect, that is, the juveniles may not become vulnerable (i.e., may be too small) to the trawl until later in the year, and may remain in the nursery until a life history stage is complete rather than respond directly to thermal cues. The relative dtsperston for dtversity data (Figures 8 and 9} ts less variable than for denstty data (Figures 4 and 5} and seems to behave inversely wtth relatton to the overall seasonal curves of diversity for both trammel and trawl. The


51 relative smoothness may imply that the faunal patchiness with respect to species is less than that wtth respect to numbers. The inverse relation may reflect the increased probability of encountering rarer with increased catches of species. In a mathematical artifact may be present. Given a relatively constant (or some finite minimum} value for the absolute measure of (the standard error of the mean) the relative dtspersion will increase with decreased catch rates. This effect may become important at very low catch rates. The consistent diurnal ranking of the habitats displayed by the trawl dtverstty data (see Table 3b) may be interpreted as the effect of the trawl's sampling a resident fauna. The diurnal differences in absolute numbers of both species and individuals are accounted for by the fish's increased vulnerability at night. Conversely the diurnal variability in the diversity ranking of habitats in the trammel data may be interpreted as a sampling of diurnal transients by the trammel net. Differences in diurnal location of these transients would account for the differences in the rankings. That the deeper habitats tend to rank higher during the day than at night (in terms of density), and that the shallower habitats tend to rank higher at night than during the day, tends to support this inference. Again the diurnal differences in the absolute numbers of both spectes and individuals could be accounted for by differential activity and vulnerability at night.


52 In any case, these patterns all have components wh.ich remain to be analyzed. Life history strategies, ontogenetic changes of habitat and feeding ecology, as well as, growth change of gear vulnerability with change in size are all, no doubt, factors contributing to the observed patterns. Apologia Inasmuch as: 1) The concept of pattern is basic in ecology (Hutchinson, 1953) 2) It is readily apparent (Hopkins, 1974, personal communication) that seasonal, diel and habitat patterns of density and are important ecological characteristics of marine inshore fishes, 3) The effect of gear bias is well discussed tn the fisheries literature (see Ben-Tuvia and Dickson, 1969, for numerous examples). I am unable to account for the lack of consideration given one or another of these factors in most available studies of inshore or estuarine fish. As indicated above this report is part of a larger continuing study. This study has attempted through well controlled and systematic sampling, through investigation and definition of overall patterns as the requisite framework for more detailed work, and through analysis of these patterns, to deal with some of the factors usually missing in previous


53 studies. That the factors considered when establishing the sampling program were important, and that the resolution of these factors in our program was appropriate, is well by our results. Patterns of variation were demonstrated with respect to all of the factors enumerated. The extensive consideration given to methodology both as to field operations, and mathematical manipulations reflects the difficulty of both the problems encountered in collecting and in coming to grips with a large and complex mass of data. Finally, further investigations are planned or are in progress. Examinations of species composition and of resident-transient components similar to Tyler's (1971) study in Passamaquoddy Bay seems a logical next step in the overall analysis of patterns, perhaps combined with or closely followed by size class analysis. In addition other investigators are examining single species or species-complexes.


SUMMARY 1 Over a 20-month period (from October 1970 to May 1973), diurnal sampling using trammel and trawl procedures was conducted at four stations, prior to channel dredging. 2. Eleven sets of samples, totalling 340 trawls and 171 trammel sets, were compounded into one twelve-month period. 3. Catches were reduced to unit of effort catch rates by numbers of fish (UECR-#) and numbers of species (UECRS p ) 4. Highest surface water temperatures were observed in July-August (data set 4) with a maximum of 30.2C and minimum of 28.5C. Yearly lows occurred in February (data set 2) with a minimum of 14.5C and the greatest temperature fluctuation in March (data set 9) with a range of 4.5C. 5. Catch rates (both UECR-# and UECR-SP) were highest from late spring to mid-fall. 6. The period of high trawl catch rates (both UECR-# and UECR-SP) occurred later in the year and persisted for a shorter period than did the trammel catch rates which paralleled the temperature data. 7. Catch rates (both UECR-# and UECR-SP) were generally higher at night than in the day. 54


8. Catch rates (UECR-# and UECR-SP) were generally over the sand habitat (station II) whereas higher rates were noted in the more complex environs of dense, mixed seagrasses (stations I, III, and IV). 9. The trawl appeared to sample the diurnally resident fauna while the trammel net appeared to sample more of the diurnally transient fauna. 10. A period of relatively high density (UECR-#) in late May-early June (due to the abundance of one or more at most a few species) appears in trawling and seine data and may be a common occurrence in estuarine faunas. 11. The observed patterns are a reflection of the interaction of fish and their environment. The well defined character and distinctness of these patterns indicate that they are an important ecological feature of the estuarine fish fauna. 55 12. The definition of the patterns provides a basis for a more complete characterization of the baseline condition of the fish fauna of the Anclote Anchorage.


CONCLUSION This study has defined some of the overall ichthyofaunal patterns of distribution in the Anclote Anchorage. It has demonstrated that these are d ifferent for each of the sampling methods used, thereby confirming the need for more than one type of gear. 56 The patterns noted were not as numerous lines of evidence have established the existence of seasonal, diurnal and habitat factors in the distribution of many taxa. However, these patterns required definition for the Anclote area in order to provide a baseline for comparison and to provide the framework for more detailed study. The need for this more detailed analysis and definition of the patterns is the final conclusion, and while it is fully realized that similar conclusions are reached by many (if not most) investigators, it remains an important and val i d one.




REFERENCES CITED Arkin, H., and R. R. Colton. 1970. Statistical Methods, Fifth Ed., Barnes & Noble, Inc. New York. 344 pp. Bailey, R. M., H. E. and C. L. Smith. 1954. Fishes from the Escambia River, Alabama and Florida, with ecologic and taxonomic notes. Proc. Acad. Nat. Sci. Phila., 106:109-164. Baird, R. C., K. L. Carder, T. L. Hopkins, T. E. Pyle, and H. J. Humm. 58 1972. Anclote Environmental Project Report 1971. University of South Florida, Marine Science Institute, St. Petersburg, Florida, Contribution #39. 251 pp. Baird, R. C., K. L. Carder, T. L. Hopkins, T. E. Pyle and H. J. H umm. 1973. Anclote Environmental Project Report 1972. University of South Florida, Department of Marine Science, Contribution #41. 220 pp. Baird, R. C., K. L. Carder, T. L. Hopkins, T. E. Pyle, H. J. Humm, N. J. Blake and L. J. Doyle. 1974. Anclote Environmental Project Report 1973. University of South Florida, Department of Marine Science, St. Petersburg, Florida. 136 pp. Ben-Tuvia, A. and W. Dickson (editors). 1969. Proceedings of the Conference on Fish Behavior in Relation to Fishing Techniques and Tactics, Bergen, Norway, Oct. 19-27, 1967. FAO Fish. 62(2) :461. Croxton, F. E., D. J. Cowden, and S. Klein. 1967. Applied General Statistics. Third Ed., Prentice Hall, Inc., Englewood Cliffs, N.J. 754 pp. Dahlberg, M. D., and E. P. Odum. 1970. Annual cycles of species occurrence, abundance and diversity in Georgia estuarine fish populations. Amer. M idland Naturalist, 83(2):382-392.


Gunter, Gordon. 1938. Seasonal variations in abundance of certain estuarine and marine fishes in Louisiana with particular reference to life histories. Ecological 8"(3) :313-346. Humm, H. J., R. C. Baird, K. L. Carder, T. L. Hopkins, and T. E. Pyle. 1971. Anclote Environmental Project Annual Report 1970. Marine Science Institute, University of South Florida, St. Petersburg, Contribution #29. 172 pp. Hutchinson, G. E. 1953. The concept of pattern in ecology. Proceedings of the Academy of Natural Sciences, (Philadelphia). Vol. 105:1-12. Kilby, J.D. 1955. The fishes of two Gulf coastal marsh areas of Florida. Tulane Studies in Zoology, 2(8) :173-247. McErlean, A. J., S. G. oconnor, J. A. Mihursky, and C. I. Gibson. 1973. Abundance, diversity and seasonal patterns of estuarine fish populations. Estuarine and Coastal Marine Science, 1:19-36. Margetts, A. R. 1969. Comparative and Experimental fishing as methods for studying fish behavior in the natural environment in Ben-Tuvia and Dickson, 1969 131-138. Massman, W. H. 1962. Water temperatures, salinities, and fishes collected during trawl surveys of Chesapeake Bay and York and Pamuniky Rivers, 1956-1959. Va. Inst. Sci. Spec. Sci. Rep. 27. 51 pp. vide Tyler, 1971. Mohler, Frank C. 1962. Anclote River Basin pilot study. Div. of Water Resources and Conservation, State Board of Conservation, Tallahassee, Florida. February 1962. Mimeo. Reid, George K. Jr. 1954. An ecological study of the Gulf of Mex1co fishes, in the vicinity of Cedar Key, Bull. of Mar. Sci. of the Gulf and Car1bbean. 4(1):1-94. 59


Roessler, Martin. 1965. analysis of the variability of fish populatlons taken by trawl in Biscayne Bay Florida. Trans. Am. Fish. Soc., 94(4) :311-318: Rogers, Scott W. 1974. Personal communication. Rolfes, J. R. C. 1972. Rolfes, J. R. C. 1974. K., B. D. Causey, D. Milliken, W A. Fable, and B a i rd. Problems of quantitative sampling in relation to near shore and estuarine fishes. Abstract Journal of the Florida Academy of Sciences, Vol. 35, supplement to No.1, p. 32. K., A. Feinstein, R. A. Dietz, D. M. Milliken, Baird, W. A. Fable, and B. D. Causey. Ecological Base-line Study of the Fish of the Anclote Thermal Ecology, J. W. Gibbons and R. R. Sharitz (eds.), AEC Symposium Series, (Conf-730505). Spiegel, M. R. 1961. Theory and Problems of Statistics, Schaum Publishing Co., New York, 359 pp. Springer, V. G. and K. D. Hoodburn. 1960. An Ecological study of the Fishes of the Tampa Bay area. Professional Papers Series No. 1, Florida State Board of Conservation, 1-104. Springer, V. G. and A. J. McErlean. 1962. Seasonality of fishes on a south Florida shore. Bull. Mar. Sci. Gulf and Carib. 12(1):39-60. Tyler, A. V. 1971. Periodic and resident components in communities of Atlantic fishes. J. Fish. Res. Bd. Can. 28:935-946. Wang, J. C. 19 71 S., and E. C. Raney. Distribution and fluctuations in the fish fauna of the Charlotte Harbor estuary, Florida. Charlotte Harbor Estuarine Studies, Mote Marine Laboratory. 56 pp. and appendices. 60 Zimmerman, N. J. 1973. R. J., R. A. Dietz, T. E. Pyle, S. W. Rogers, B 1 a k e and H J H u mm Benthic Community--Seagrasses. in il, 1973. 115-141.




APPENDIX A Special Chi-square Technique The chi-square procedure requires that the data be formed into frequency of occurrence classes. Furthermore since it is advantageous to maximize the number of these classes, a system of data-defined floating class boundaries was developed. This procedure allowed establishment of the 62 maximum number of rate classes possible, within the requirements of the chi-square procedure, in a formal rigorous manner. Campbell (1967, p. 71) and Arkin and Colton (1971, p. 140) both indicate that rate classes should be structured. such that the expected frequency of occurrence in any rate c 1 as s i s at 1 east 5 Ark i n and Co 1 ton ( 1 9 71 p 1 3 7) g i v e the theoretical frequency (expected frequency) as f = n rni c -N where f = theoretical frequency c nr = total of row entries con-taining comparison n. = total of column entries 1 containing comparison N = total number of observa-tions in table by re-arrangement of this formula (with some relabeling of terms) we arrive at an algorithm for determining the minimum number of entries for each rate class (N row)(N c olumn) N total for 2xR tables, c


63 (n co 1 } sn tot 5::; (n tot .J (n rowl > ... n col ::; n row :::: n. 1 5 n tot = n. min. and for columns with unequal entries .. [n col.(min}] 1 = min # of row entries in rate class for f (generally c n min ::; 1 0 or 111. 1 The catch rates for trawling data existed i n steps of 0. 5 catch rate units (a mathematical artifact due to use of 2-minute trawls} and were so tallied on frequency diagrams. The trammel data, however, existed as a more or less con-tinuous function which was arbitrarily iterated in steps of 0.1 catch rate untts when tallied on frequency diagrams. A value of n. min. (above) was then calculated for each com-, parison. The frequency diagrams were counted off from the higher rates toward the lower, frequency class boundaries being established whenever the total number of entries (from both members of the comparisons) equalled or exceeded the n. min. value. The frequencies for each of the classes were 1 then entered in a 2xR table and the calculated chi-square value obtained by use of the formula given by (1967, pp. 597-8). 2 X R Table Cl., : O..i n .... b, bz. . b.: n., degrees of freedom 'rl, l'h n, 2 N2 2 2 X = n nb a n. N 1 N = (R-1) (C-1), C = 2...,..R-l df.


Chi-square probabilities were obtained by entering the calculated and degrees of freedom in a Wang 700 computer program 1991A/ST2 (part of statistical package 1971A/ST5). The output of this program is the probability that the observed chi-square value is not due to random factors. The decimal probabilities were converted to percentages. 64




66 APPENDIX B Sample Calculations 1 ) Calculation of unit effort catch rates (UECR) a) Sample trawl calculation Dataset 1 Gear trawl Time Collection No. No. Time UECR UECR No. Fish (min) # SP 180 16 4 2 8 2 181 18 7 2 9 3.5 183 17 4 2 8.5 2 184 19 6 2 9.5 3 Mean (xs) = 8.75 2.6 (by Wang 700 program) Std. deviation (Ss) = .65 .75 Number. of samples (Ns) = 4 4 b) Sample trammel calculation Dataset 1 Gear trammel Time day Collection No. No. Time Effort UECR UECR No. Fish (min) Units # SP 179 98 11 77 7.7 12.73 1. 43 182 92 8 76 7.6 12. 11 1. 05 Mean (xs) = 12.42 1. 24 (by Wang 700 program) (Ss) .44 .'l.1 Std. deviation = Number of samples (Ns) = 2 2


2) Calculation of the average of means and the stratified standard error of the man (example given is for seasonal grouping). Dataset 1 Station I II III IV Mean=x= Gear Trammel 12.42 4.13 7.05 5.27 7.22 s s .438 3.93 3.95 .977 N= Data UECR-# Time Day N s. N 2 1 X S. s 1 1 2 .38 2 30.96 2 31 1 6 2 1. 91 8, 64.41 = [_(N; ) xs. i 1 67 Stratified standard error of the mean s : si . ) =too x N1. (per Arkin and Colton [1971] p. 147) 3) Calculation of Relative Dispersion (per Speigel, 1961) s 1 00 Rel. Disp. = Absolute dispersion average = x = 7 2 2 = 1 3 9 -X 4) Estimation of Missing Values Trammel net, dataset __ 5_ station __ !_ a) existing datum UECR-# =2.931 = x b) datum for complete strata, data-set 4, sta. I, day x = 3.47, S = .366 s s c) datum for complete strata, data-set 6, sta. I, day = 3. 41 s = .451 Xs s d} no similar values within data set 5 e) mean of s for data sets 4 and 6 = .4085 = s estimate s s


f) day sample, therefore choose estimated value above existing value, try 3.3 (run on Wang 700 program) x = 3.12 S = 2.61, not acceptable s g) try 3.5 (run on Wang 700 program) 68 xs = 3.22, Ss = .402 which is an acceptable approximation 5) Conversion of catch rates to frequency diagram a) The data Station III, Day-Night comparison. UECR-# trammel CATCH RATES Data Set Q2L Night 8 .77 1. 11 196 1.4 2 1. 89 3.89 1. 92 3.09 9 1.8 3.12 1.6 4.44 10 3.94 7.46 3.33 9.42 11 7. 41 5.65 6.03 7.57 3 6.56 5.00 9.85 4.68 4 8.57 9.58 6.72 9.24 5 5.08 7.08 7.05 5. 15 6 8.5 9.94 4.21 10.79 1 9.84 8. 14 4.26 7.29 7 2.75 2.04 3. 28 2.83


0 2 8 1.2 1.5 1.7 1.9 2.0 2. 1 2.8 2. 9 3. 1 3.2 3.3 3.4 3.9 4.0 4.3 4.5 4.7 5. 1 5. 2 5. 7 6. 1 6.6 6.8 7. 1 7. 3 7. 5 7.6 8.2 8.6 9.0 9. 3 9.5 9.6 9 . 9 69 b) the above data as tallied on a frequency diagram Frequency diagram (tabular version of graphic method) UECR-Class 2 UECR-ft . 8 -1.2 -1.5 -1.7 1.9 2.0 -2. 1 -2.8 -2.9 -3. 1 ... 3. 2 ... 3.3 -3.4 -3.9 4.0 -4.3 4.5 -4.7 -5. 1 -5.2 -5. 7 -6. 1 -6.6 -6.8 -7. 1 -7. 3 -7.5 -7.6 -8.2 -8.6 ,... 9.0 -9. 3 -9. 5 -9.6 -9.9 ... 10.8 # Occurrences ft Occurrences Night 1 0 l 0 0 1 0 1 1 0 2 0 1 0 0 1 1 0 0 1 0 1 0 1 1 0 l 0 0 1 1 0 2 0 0 1 0 1 1 1 0 1 0 1 1 0 1 0 1 0 1 1 0 1 1 1 0 1 0 1 2 0 0 1 0 1 0 1 0 1 2 0 0 1 Frequency Class Boundary (number entries in class) ( 1 4 entries in class 0.0 -3.3 UECR-#) Class boundary ( 1 0 entries i n class 3.3 -5.2 UECR-#) Class boundary ( 1 0 entries in class 5.2 -7.6 UECR-#) Class boundary (10 entries in class 7.6 10.8 UECR-#) 22 entries 22 entries


70 c) The minimum number of row entries is calculated by the algorithm developed in Appendix A 5 n tot. ;:: n. min n c 0 l min. 1 { n tot = 22 + 22 = 44 n col. min = 22 5 44 = 10 entries required from both columns in any 22 row d) The table in 5b) above is counted off from the bottom e) with frequency class boundaries being established whenever the total entries in that class exceeds 10. The rates within entered in a 2xR Rate Class 7.6 -l 0. 8 5. 2 -7.6 3.3 -5.2 0.0 -3.3 each class table (see Day (a ) 1 4 5 5 8 N = 22 a established Appendix A) Night 6 5 5 6 N = 22 b (Croxton, 1967, pp. 597-8) 4 [11. 17-11] = .68 above are then ( b ) N. 1 1 10 l 0 10 14 N =44 'V = DEGREES OF FREEDOM = R-1 = 4-l = 3 f) Enter Wang 700 program with X?.= 68 degrees of freedom = = 3 output p (X-.. 1r) = .1221 = 12% probability of calculated chi!square being due to non-random factors.


f{...jC/ C"D L L L.D I \ flom I ct 7 L/


Download Options

No images are available for this item.
Cite this item close


Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.


Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.


Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.


Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.