ABSTRACT. A long-term (13 year) study was carried out to analyze the response of estuarine fishes. (numbers, biomass and trophic organization) to seasonal ...
BULLETIN OF MARINE SCIENCE. 60(3): 984-1004. 1997
TROPHIC RESPONSE OF ESTUARINE FISHES TO LONG- TERM CHANGES OF RIVER RUNOFF
Robert J. Livingston ABSTRACT A long-term (13 year) study was carried out to analyze the response of estuarine fishes (numbers, biomass and trophic organization) to seasonal and interannual variations of Apalachicola River flow and associated habitat changes of the receiving estuary (East Bay, Apalachicola Bay system, Florida). Periodic peak floods and prolonged droughts were important events that led to altered patterns of individual fish distribution in terms of numerical abundance and biomass. There was considerable interannual variation in the temporal distribution of the dominant fish species over the study period. Individual estuarine fish species use the estuary as a nursery ground with species-specific ontogenetic feeding patterns that are defined by the complex productivity patterns of the system. In East Bay, there was a dichotomous response of the estuarine trophic organization with herbivores and omnivores (dominated by infaunal and epifaunal macroinvertebrates) directly responsive to river-associated physicochemical factors whereas the carnivores (dominated by the fishes) responded to biological factors such as predation and competition. Estuarine fish organization was indirectly responsive to changes of river flow through prey responses to state habitat and productivity variables associated with river flows. This suggests that the fish associations were strongly dependent on interannual patterns of Apalachicola River flow but that such relationships were primarily caused by biological interactions as defined by specific predator/prey relationships. A prolonged drought led to reduced fish species richness and trophic diversity; such habitat stress was related to enhanced instability of the biological components of the estuary as a function of changes in nutrient cycling. The food web was simplified while overall fish biomass and individual species populations were numerically reduced. The trophic response times of fish assemblages were measured in years from the point of the initiation of the drought. Changes in flow rates that exceeded specific natural levels of variance could be followed by identification of the subtle yet important changes in estuarine productivity and relatcd changes of fish representation within the food web. The use of individual fish species as indicators of such responses is not consistent with the processes that define the long-term behavior of estuarine populations.
Considerable effort has been expended to evaluate the relationship of freshwater input to estuarine processes (Snedaker et aI., 1977; Cross and Williams, 1981) and the effects of salinity and other physico-chemical variables on estuarine populations (Sheridan, 1978, 1979; Sheridan and Livingston, 1979; Laughlin, 1979; Laughlin and Livingston, 1982; Cloern et aI., 1983; Wilber, 1992). Human activities have had various effects on estuaries around the world; such impacts include reduced or altered freshwater flow, changes of temperature and dissolved oxygen, nutrient enrichment and hypereutrophication, fishing pressure, and many forms of pollution (Livingston, 1990). Upland and offshore nutrient loading and rapid recycling within the estuary contribute to the generally high primary production of river-dominated estuaries (Nixon, 1981). Complex processes of nutrient dynamics comprise the source of the naturally high productivity of estuarine systems (Livingston, 1984a; Peterson and Howarth, 1987; Howarth, 1988; Baird and Ulanowicz, 1989; Livingston, 1991), and, together with wind and tidal subsidies and associated phytoplankton productivity, are considered to be responsible for the valuable nursery functions of estuaries for a variety of fishes. However, despite a long history of intensive research concerning various aspects of estuarine pop984
LIVINGSTON: LONG-TERM TRENDS OF ESTUARINE FISHES
985
ulation dynamics, relatively little is known concerning the long-term relationships of freshwater input and changes in fish populations and assemblages. There is considerable information concerning the fishes of shallow estuaries of the Gulf of Mexico. The spatial/temporal connections between primary production and the secondary food webs have been inferred but rarely delineated. The Apalachicola estuary has been extensively studied (Sheridan, 1978; Laughlin, 1979; Sheridan, 1979; Sheridan and Livingston, 1979, 1983; Livingston, 1981a, 1984a, 1984b, 1991, 1997). River-dominated estuaries such as Apalachicola Bay (Fig. 1) are characterized by complex interactions of freshwater input, nutrient loading and processing, and primary production (Livingston, 1997; Livingston et aI., 1997). Interactions of such processes are mediated by short- and long-term cycles of various factors that include water temperature, salinity, and river-associated nutrient loading (Livingston, 1976, 1979; Meeter et aI., 1979; Mahoney and Livingston, 1982). Livingston et ai. (1997) have indicated that Apalachicola River flow is a key element in the control of the biological organization of East Bay, and that such control is both direct (effects of salinity and light penetration) and indirect (related trophic relationships). Species distributions and the basic structure of the fish populations are indirectly related to the highly variable estuarine environment through alterations in the food web which are defined by interannual changes in river flow. The biological processes involved in the predator/prey interactions are thus closely but indirectly associated with the physical and chemical constraints of the estuarine system that are ultimately defined through cyclic changes in productivity and direct responses of the herbivore/omnivore populations to such changes. The Apalachicola Bay system remains in a relatively unaltered state with fresh water flow as a major controlling variable (Livingston, 1984a, 1997). The Apalachicola River is one of the last major free-flowing, unpolluted alluvial systems left in the conterminous United States. East Bay (Fig. I) is located at the head of the Apalachicola estuary. Because of its nursery status (providing habitat and food for various juvenile populations), commercial net fishing is prohibited in East Bay which, together with a relatively pollution-free habitat and unaltered river flows, has contributed
to continued
natural pattems
of fish recruitment
and
growth. Such processes remain unaltered relative to estuaries that have been strongly affected by human activities (Livingston, 198Ib). Interacting factors that control fish population distribution in the Apalachicola Bay system have been described (Livingston, 1976, 1979, 1981a, 1981b, 1984a). It is shallow (mean depth; 2 m) with mean river flow rates of approximately 690 m3/s (1958-1980; U. S. Army Corps of Engineers, Mobile, Alabama). Annual high flows average 3,000 m3/s (Leitman et aI., 1982; Leitman et aI., 1991). Together with wind, tides and local rainfall which provides runoff from Tate's Hell Swamp (Fig. 1) into the northeastern portions of East Bay, the Apalachicola River controls the salinity of this part of the estuary (Livingston, 1984a). Surface water temperatures in East Bay normally range from a low of 5-8°C from December to February to highs near 32°C in August (Livingston, 1984a). Seasonal trends and long-term (11-14 years) biotic changes in the Apalachicola Estuary are related in various ways to the river flow (Livingston, 1991, 1997). Seasonal variation is strongly related to river flow and associated habitat changes. Fish assemblages are numerically dominated by anchovies (Anchoa mitchilli), Atlantic croaker (Micropogonias undulatus), spot (Leiostomus xanthurus) and sand seatrout (Cynoscion arenarius). Anchovies are dominant during fall-winter periods whereas sciaenids such as Atlantic croaker and spot are dominant during winter-spring months. Spot and croaker are in direct competition for food, es-
986
BULLETIN OF MARINE SCIENCE, VOL. 60, NO.3, 1997
............................................
.. ::.:::::::::::::::: ..
..
.
O· .. ,
:$¢~~~::::::::::::::':: 2' . '3 .. 4'
. '5' . . . . . ..
TATE'S'HEI-l' .. ~W~~~::::: .
.
·•••IC:::I•••c:::II•• :KM::::.::::::::::::::·
::::t'::::::: .'
"
.....
.
. ::::::
.
. .
N::::
APALACHICOLA BAY
Figure
I.
Location
map of the East Bay system showing
stations
used in the long-term
studies,
pecially as young populations that are numerically high during the winter-early spring months of high river flow (Sheridan, 1979, Sheridan and Livingston, 1979). The sand seatrout is a piscivorous fish that feeds primarily on anchovies (Sheridan and Livingston, 1979), reaching peak numbers during late spring and early summer. There are indications that long-term changes in population distribution of these species could be related to competition for food (Livingston, 1991). These fishes seem to respond in a relatively stereotypic fashion to seasonal changes in the primary variables. The interannual trends relative to such variables are less certain, however. The assumptions implicit in the above observations include the relative importance of feeding patterns in the distribution of fishes in space and time. The purpose of this paper is to present a series of field analyses and the results of extensive trophic studies in an effort to define the long-term responses of fish population and community indices in the Apalachicola estuary to factors that are related to seasonal and interannual fluxes of freshwater runoff and periodic extreme events such as floods and droughts. The interactions of the fish trophic organization with specific patterns of Apalachicola River flow were examined in an attempt to define how river flow and associated alterations in the physico-
LIVINGSTON: LONG-TERM TRENDS OF ESTUARINE FISHES
987
chemical structure of the bay are associated with longterm fluctuations of estuarine fish populations. MATERIALS AND METHODS Field Collections.- The field data were taken at fixed stations in East Bay (Fig. I). Methods for the field collection of physicaUchemical data are given by Livingston (1979, 1980, 1981a, 1982, 1984a, b) and Livingston et al. (1974, 1976). Local rainfall data were provided by the Florida Department of Agricultural and Consumer Services, Division of Forestry, East Bay. River flow information was provided by the U. S. Geological Survey (Sumatra, Florida gage). Water chemistry samples (surface and bottom) were taken with I-liter Kemmerer bottles. Dissolved oxygen and water temperature were taken with Y. S. I. dissolved oxygen meters and stick thermometers. Dissolved-oxygen anomaly was calculated from the field measurements as the difference between the measured dissolved oxygen and the oxygen solubility at a standardized temperature and salinity (Weiss, I970). These measurements reflected potential biological effects due to photosynthesis and respiration. Salinity was measured using temperature-compensated refractometers and salinity probes at least once each month from March 1972 through July, 1984. Turbidity was determined with a Hach model 2 IOO-A turbidimeter. Water color was measured by an American Public Health Association platinum-cobalt standard test (colorimeter, Pt-Co Units). Biological samples (infaunal and epibenthic macroinvertebrates, fishes) were taken monthly at fixed stations in East Bay (Fig. I). These stations were based on a preliminary habitat stratification of the East Bay system and haphazard samples taken within the bounds of the predetermined areas (Livingston, 1976, 1987). Infaunal macroinvertebrates were taken using coring devices (7.6-cm diameter, IO-cm depth). The number of cores taken was determined using large initial samples (40 cores) and a species accumulation analysis based on rarefaction of cumulative biological indices (Livingston et aI., 1976). Based on this analysis, multiple (10) core samples were taken randomly at each station with the assured representation of at least 80% of the species taken in the initial sampling. All infaunal samples were preserved in 10% buffered formalin in the field, sieved through 500 11m screens and identified to species wherever possible. Epibenthic invertebrates and fishes were taken with 5-m otter trawls monthly from March 1972 through July 1984. The otter trawls (1.9-cm mesh wing and body, 0.6-cm mesh liner) were towed at speeds of about 3.5-4 km·h-I for 2 min resulting in a sampling area of about 600 m2 per tow. Repetitive samples (2 or 7) were taken at each site; sampling adequacy was determined by Livingston (1976). All organisms were preserved in 10% buffered formalin, sorted and identified to species, counted and measured (mm standard length). Total numbers of species were represented as species richness. Representative samples of fishes and invertebrates were dried and weighed and regressions were run so that data from the biological collections could be converted into dry and ash-free dry mass. The numbers of organisms (infauna, epibenthic invertebrates, fishes) were then converted to biomass·m-2• Trophic Analyses.-All biological data (as biomass·m-2mo-') were transformed from species-specific data into a new data matrix based on trophic organization as a function of ontogenetic feeding stages of the species found in East Bay over the multi-year sampling program. Infaunal macroinvertebrates were organized by feeding preference based on trophic analyses and a review of the scientific literature (Livingston, unpublished data). Ontogenetic feeding units were determined from a series of detailed stomach content analyses carried out with the fishes and epibenthic invertebrates in the region (Sheridan, 1978, 1979, Sheridan and Livingston, 1979, 1983; Livingston, 1980, 1982, I984b, unpublished data; Stoner and Livingston, 1980, 1984; Laughlin and Livingston, 1982; Stoner, 1982; Clements and Livingston, 1983, 1984). Details of this reorganization of the data base by trophic units are given by Livingston et al. (1997). Data from the various stations in East Bay were averaged (gm-2) so that station location was not a factor in the analysis. The data were summed across all taxonomic lines and translated into the various trophic levels that included herbivores (feeding on phytoplankton and benthic algae), omnivores (feeding on detritus and various combinations of plant and animal matter), primary carnivores (feeding on herbivores and detritivorous animals), secondary carnivores (feeding on primary carnivores and omnivores) and tertiary carnivores (feeding on primary and secondary carnivores and omnivores). All data are given as ash-free dry mass m-2mo-1 or as percent ash-free dry mass m-2mo-l• Statistical Analyses.-Data Transformations. Scattergrams of the long-term field data were examined and either logarithmic or square root transformations were made, where necessary, to approximate the best fit for a normalized distribution. These transformations were used in all statistical tests of significance. Analysis of Variance. Analysis of Variance (ANOVA) models were run using Systar@> and SuperAnova®. The ANOVA determinations were run using year-by-season data sets. Three-month intervals, starting with March of each year, were used to define the seasonal patterns. These seasonal
988
BULLETIN OF MARINE SCIENCE. VOL. 60. NO.3. 1997
definitions were based on established seasonal temperature patterns in the study area. The basic model followed that outlined by Winer (197 I) as; Y,! = m
+
a;
+
bj
+
abij
+
eij
where Yij is the response associated with the ith year and the jth season, m represents an overall effect, a; is the effect of the ith year, bj is the effect of the jth season, abiJ is the interaction between the ith year and jth season, and eij represents the random error. Specific hypotheses were tested (e.g., years and seasons have no effect on dependent variables). A two-factor analysis was carried out with Years and Seasons as the two factors. The assumptions of the model were verified by examining residual distributions. The three monthly values within each season were used as replicates even though these factors were not true replicates. An autocorrelation function (ACF) was run to check for serial correlation among the monthly residual values (positive correlations would invalidate the ANOYA test). In most cases, no positive serial correlations of residuals for sequential months were noted for the various dependent variables. Residuals were plotted against fitted values in addition to carrying out normal probability plots. Unless otherwise stated, ANOYA results showed random distributions of the residuals in normal probability plots. The Wald-Wolfowitz run test (with cutoff = 0) was run on residuals to determine the possibility of clumping of the positive and negative residuals (Wald and Wolfowitz, 1940). A lack of significance in the Wald-Wolfowitz run tests on residuals was usually found. F-values and their associated p-values were calculated for the final determinations with pair-wise comparison of the effects/interactions via post-hoc tests for further comparisons. RESULTS
River Flow Conditions.-Trends of Apalachicola River flow are shown in Figure 2. The river data are given as percent differences from long-term (40-yr period: 1950-1990) monthly mean values. The general seasonal pattern of river flow was similar from year to year with high flows during the winter-spring months and low flows during summer-fall months. During the first year of record (1972), there was a minor drought which was broken by major increases in river flow during the winter/spring and summer of 1973. This was followed by somewhat drier conditions (1973, 1974) after which there was another period of high river flow (1975). Over the next 5 years (1976-1980), winter-spring flows tended to show comparable trends of seasonal river peaks. The period from 1976 to 1977 was characterized by relatively low winter flows compared to preceding and succeeding years. Low summer flows became increasingly prolonged from 1975 to 1978. Winter-spring peak flows occurred during the period from 1978 to 1980. From May 1980 through the end of 1981, there was a major drought with substantially lower river flows during the winter-spring of 1981. Flow rates during the 20-mo period prior to the winter of 1982 were consistently below the longterm (40-yr) monthly means and were often less than 50% of what East Bay usually receives in the way of freshwater runoff from the river. In terms of duration, the 1980-1981 drought was the fourth longest period in this century with below-average flows in consecutive months (Livingston et al., 1997). The combination of individual wet and dry seasons during this period formed the third most severe drought in terms of magnitude (mean flow) over three consecutive seasons. The following 2.5 yr were characterized by a general return to the prevailing patterns of Apalachicola River flow as noted during the period from 1978 through the winter of 1980. According to ANOVA analyses of the long-term data set (Table 1), river flow highs during winter-spring periods were significantly different (p < 0.05) from summer-fall lows. River flows during 1981 were significantly (p < 0.05) different from all other years except 1978. Peak flows in 1973 and 1975 were significantly (p < 0.05) different from flows in 1981. Seasonal and Long-term Habitat Changes in East Bay.-Monthly trends of important water quality factors over the study period are shown in Figure 2. Statistical results of the ANOVA applications to these data are given in Table 1. Based
989
LIVINGSTON: LONG-TERM TRENDS OF ESTUARINE ASHES
150 100 ~50 0 -50
roui
Droll
ht
2
~ 1.5
~
ell
~
I 0.5
0 .l!l400 'C300
~
Q200
,.
~IOO 0 100 ~
50
~
0 6
"
~3
E
0 .]
..8:
oW 30
20 10
8 81S!:l8 8 ~!:l8 8:1i!:l8 8 ~ !:l1'l8 ~!:l8 8 ~!:l8
f' f' ~ f' j;; j;; l' j;;;l;!;!;l
~ ~~
8 ~!:ll;! ~ ~!:l8 8 ~!:l8 8 ~"8
~ ~ ~ ~ ~ iO iO ~ iO ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~;:;:;:;:
81S !:l81l ~!:l8 8 iii iii&! iii &i &l &i &i ~ 1.\
DATE
Figure 2. Apalachicola River flows, presented as percentage deviations from the monthly means over the study period, compared with means (and ranges from the 10th to the 90th percentiles) of leading habitat features such as Secchi depths (meters), surface color (Pt-Co Units), surface turbidity (lTU), surface oxygen anomalies (mg·liter-I) and bottom salinity (%0).
on the seasonal data, increased river flows during winter-spring months were associated with reduced salinity (p, OJ (Q
>(
:u
=c..
