ABSTRACT

Total epiphyte and epiphyte chlorophyll loads, seagrass short shoot size and the elemental composition of seagrass leaf tissue are compared with water column nutrient concentrations. The spatial patterns in these parameters are described across a nutrient availability gradient in Florida Bay. Multivariate and univariate statistical analyses were performed on these parameters to test for correlations with epiphyte levels. Total epiphyte loads were not correlated with any measure of nutrient availability, but 18% of the variation in epiphyte chlorophyll loads could be explained by phosphorus availability. The measurement of total epiphyte loads and observations of epiphytic species composition along a transect adjacent to a point source of nutrients revealed that the effect of nutrient enrichment on epiphyte levels is pronounced but very localized. The elemental composition of seagrass leaf tissue may be a better indicator of nutrient availability than either water column nutrient concentrations or epiphyte levels.

INTRODUCTION

Seagrass meadows rank among the most productive ecosystems in nature, rivaling those of tropical rain forests and tidal marshes (Odum 1957, McRoy & McMillan 1977, Zieman & Wetzel 1980). Seagrass epiphytes are an important component of these highly productive ecosystems, often contributing greater than one-third to the total above ground biomass (Penhale 1977, Heijs 1984, Tomasko & Lapointe 1991) and as much as 30% to combined seagrass/epiphyte productivity (Penhale 1977, Morgan & Kitting 1984, Heijs 1984, 1985, 1986). Though most seagrass biomass must enter the detrital food web before it is available to higher trophic levels (Odum & de la Cruz 1967, Fenchel 1970; Fry & Parker 1979), epiphyte production can be a direct source of carbon available to higher trophic levels through grazing (Fry & Parker 1979, Kitting et al. 1984). Epiphytes may also provide a more assimilable detrital carbon source than their seagrass hosts due to the large amount of refractory structural tissue and phenolic compounds in seagrasses (Harrison 1989).

In addition to being an important component of the seagrass community, epiphytes can also be detrimental to seagrasses. Seagrasses form extensive beds in many coastal embayments worldwide, and seagrass-dominated ecosystems have fared poorly in anthropogenically-influenced coastal regions because of the sensitivity of seagrasses to changes in water quality (e.g. Orth & Moore 1983, Giesen et al. 1990, Larkum & West 1990). The general pattern of changes in benthic vegetation accompanying eutrophication is from rooted macrophytes at low nutrient availability to domination by epiphytic and water column algae at high nutrient availability (see Duarte 1995 for review); losses of submerged aquatic vegetation have been attributed to overgrowth by epiphytes (Phillips et al 1978, Kemp et al. 1983, Cambridge et al. 1986). Epiphyte loads reduce the productivity of macrophytes by shading and reducing nutrient availability (Bulthuis & Woelkerling 1983, Twilley et al. 1985, Tomasko & Lapointe 1991, Neundorfer & Kemp 1993, Neckles et al. 1993, 1994). Epiphyte abundance has been experimentally demonstrated to be a function of nutrient availability (Tomasko & Lapointe 1991, Neckles et al. 1993, 1994; but see Bulthuis et al. 1992). Seagrass epiphyte density may increase along nutrient availability gradients. Increased epiphyte levels have been observed adjacent to point sources of nutrient addition (Cambridge & McComb, 1984; Silberstein et al., 1986). Borum (1985) found that epiphyte biomass increased 50^-100 fold along a nutrient gradient in Roskilde Fjord, Denmark.

Seagrass epiphytes are sessile plants and animals that grow attached to their seagrass host (Harlin, 1980, Frankovich & Zieman, 1994). Both plant and animal epiphytes are significant producers of calcium carbonate and may be the primary source of carbonate mud in shallow tropical and subtropical embayments (Land 1970, Patriquin 1972, Nelsen & Ginsburg 1986, Bosence 1989, Frankovich & Zieman 1994). Epiphyte carbonate often accounts for as much as 70-80% of epiphyte standing stock (Heijs 1984, Frankovich & Zieman 1994). This large amount of non-autotrophic mass in the epiphyte community may complicate the relationship between nutrient availability and epiphyte loads of seagrasses. Because autotrophs are the primary component of a community that respond to nutrient availability, some measure of autotrophic epiphyte biomass (e.g. epiphyte chlorophyll) should be more sensitive to nutrient availability than total (plant + animal) epiphyte load.

In this paper, we compare epiphyte loads, seagrass short shoot size, nutrient availability to seagrasses, and water column nutrient concentrations across a large tropical embayment. Large gradients in benthic and water column nutrient availability have been documented for this ecosystem; nutrient availability ranged from limiting to seagrass and phytoplankton growth to adequate for dense seagrass growth. We document the spatial patterns in epiphyte abundance across the ecosystem, and test the hypothesis that epiphyte biomass would be correlated with the gradient in nutrient availability.

METHODS

Study Site. This work was done in a shallow subtropical embayment, Florida Bay, Florida, U.S.A (Figure 1). Seagrasses, predominantly Thalassia testudinum, carpet ca. 2000 km² of Florida Bay. There is a strong gradient in seagrass density across the bay, with seagrass density in the western bay an order of magnitude higher than the eastern parts of the bay (Zieman et al. 1989). This gradient is the result of a strong gradient in benthic phosphorus availability; seagrass biomass in the eastern parts of Florida Bay is P-limited (Fourqurean et al. 1992a, 1992b). In addition to the gradient in benthic nutrient availability, there is a corresponding gradient in water column nutrient availability, which leads to P-limitation of phytoplankton biomass (Fourqurean et al. 1993). We sampled seagrasses, epiphytes, and water column nutrients from 24 stations distributed across the Florida Bay seagrass and nutrient availability gradient. A bird colony island in eastern Florida Bay, Porjoe Key (Figure 1), was also sampled for seagrasses and epiphytes; this bird colony is a local point source of nutrients in an otherwise oligotrophic portion of Florida Bay (Powell et al 1991, Fourqurean et al 1992a).

Seagrass and Epiphyte Sampling. On March 26-27, 1994, six arbitrary short shoots of Thalassia testudinum were collected from 24 long-term water quality monitoring sites within Florida Bay (Figure 1). During August of 1991 and 1992, 10 short shoots of T. testudinum were collected at each of three equally-spaced points located along a 45-meter transect perpendicular to the shoreline of the bird colony island (Figure 1). Macroscopic epiphytic organisms occurring on the leaves of the sampled short shoots were recorded. Organisms were identified to the lowest taxon practical, and a qualitative estimate was made of the relative abundance of each organism. Epiphytes were separated from the seagrass leaves by gentle scraping of lyophilized seagrass leaves (Frankovich & Zieman 1995). The dry weight of the leaves of each short shoot (ss) and the dry weight of the epiphytes removed from each short shoot were recorded. Total epiphyte load (mg(dry wt of epiphytes) g(dry wt of seagrass leaf)^-1), was determined from the dry weight of the separated epiphyte material and the dry weight of scraped seagrass leaves.

For each of the 24 sites sampled in March 1994, autotrophic epiphytes were quantified by extracting chlorophyll-a from the lyophilized epiphytes. The separated epiphyte material from each shoot was steeped in either 5.0 or 20.0 ml (depending on total epiphyte weight) of 90% spectrophotometric grade acetone, and held in the dark below 0øC for a minimum of 36 h. Chlorophyll-a (Chl-a) content of the acetone extracts was determined fluorometrically (Strickland & Parsons 1972). Epiphyte chlorophyll load (ug Chl-a g(seagrass leaves)^-1) was determined from the Chl-a content of the acetone extracts and the dry weight of scraped Thalassia testudinum leaves. In order to detect differences in the amount of algal epiphytism relative to faunal epiphytism, an autotrophic index (AI = ug Chl-a mg(epiphyte)^-1) was determined.

The availability of nutrients to the seagrasses was assessed by analyzing the carbon, nitrogen and phosphorus content of the seagrass leaves at each site (Fourqurean et al. 1992a, b). C and N content was determined by analyzing triplicate samples of the scraped seagrass leaves using a Carlo Erba CN analyzer. Phosphorus content of the same material was determined following a dry oxidation/acid hydrolysis procedure (Fourqurean et al. 1992a). Ratios of C, N and P were calculated on a molar basis.

Water Quality Sampling. Mean values of water quality data were determined for the two-month period prior to sampling (the seagrass/epiphyte turnover period for Florida Bay, Zieman et al. 1989). A longer-term record (March 1991-March 1994) of surface salinities at site was used to describe the salinity variability at each site. Water quality parameters collected for comparison to epiphyte loads were: concentrations of total organic nitrogen (TON), ammonium (NH4+), nitrate (NO3-), nitrite (NO2-), total phosphorus (TP), soluble reactive phosphorus (SRP), total organic carbon (TOC), water column chlorophyll-a (Chl-a), salinity, and turbidity. Methods for these water quality measurements are described elsewhere (Fourqurean et al. 1993).

Statistical Methods. Before statistical analyses, all data were checked for normality, and were log-transformed where appropriate. Statistical treatment of the total epiphyte loads measured at the three sites located along the nutrient point source transect was performed utilizing a two-way ANOVA to detect significant differences in mean total epiphyte loads between sites and times. In the absence of a statistically significant time effect, a one-way ANOVA with Scheffe's multiple comparison procedure (0.05 level of significance) was performed to detect significant differences between the time-averaged transect point means. In order to define the independent underlying patterns in the nutrient availability data, a Principle Components Analysis, using the correlation matrix, was done on the nutrient availability data. The final solution was rotated to allow for interpretation of the principal components using a VARIMAX rotation. Stepwise multiple regression was used to determine which of the nutrient availability principal components explained significant portions of the variability in the epiphyte loads.

RESULTS

Seagrass distribution. There was a strong gradient in the amount of green leaf material per short shoot of Thalassia testudinum across Florida Bay (Figure 2). Short shoot size ranged from a low of 34 mg(green leaf) ss^-1 at station 19 to a high of 525 mg(green leaf) ss^-1 at station 13. There were relatively few stations with large short shoots; the median for the whole bay was 80 mg(green leaf) ss^-1. In general, short shoot size was smallest in eastern and central Florida Bay, and largest along the western boundary of the bay (Figure 2). This gradient in short shoot size was mirrored by a gradient in P, but not N, content of the leaf biomass. C:P of T. testudinum leaves generally increased from a minimum of 486 in the northwest part of Florida Bay to a maximum of 1807 at station 8 in eastern Florida Bay (Figure 3a). The pattern in C:N ofT. testudinum leaves was quite different, with a minimum of 11 at station 11 in the center of the bay. C:N increased in all directions from the center of the bay, the maximum C:N was 21 at station 7 (Figure 3b). C:P was strongly correlated with short shoot size (Figure 4), but C:N was not (r² = 0.00, P = 0.807). The one point that deviated strongly from the relationship between C:P and shoot size was station 11, where the shoot size was much smaller than predicted by the P content of the leaves (Figure 4).

Epiphyte abundance. We identified seven macroscopic algal and eleven faunal taxa that were epiphytic on the leaves of the sampled short shoots during the March 1994 sampling period (Table 1). All identified macroscopic algae were red algae (Rhodophyta). The epiphytic faunal composition was more diverse than the algal composition. Five phyla (Annelida, Bryozoa, Chordata, Hydrozoa, and Mollusca) were represented. Epiphyte occurrence and abundance was dominated by the coralline red algae (Melobesia membranacea and Fosliella farinosa ) and the polychaete Spirorbis sp., which is easily identified by a characteristic planospiral carbonate tube. Spirorbis was the most ubiquitous epiphyte taxon, occurring at 75% of the sites, and was the most abundant epiphyte at 58% of the sites. The coralline red algae were the most abundant and ubiquitous of the algal species, occurring at 50% of the sites and most abundant at 42% of the sites. Heavy carbonate encrustatations were formed by these species at those sites in western Florida Bay, where water is exchanged more freely with the Gulf of Mexico and the Atlantic Ocean.

The abundance of epiphytes, when measured on a per short shoot basis, was greatest in western Florida Bay and decreased eastward (Figure 5a). There was a broad range (1.7 - 824 mg ss^-1) in epiphyte abundance across the bay, with a median abundance of 79.7 mg ss^-1. Given the spatial gradient in short shoot size, short-shoot based measures of epiphyte abundance do not accurately reflect the effect of epiphytes on seagrasses, however. Normalizing this measure by the size of the short shoots gives a better estimate of the epiphyte load on the seagrasses. Total epiphyte load ranged from 23 - 1569 mg(epiphyte) g(seagrass)^-1, with a median load of 376 mg g^-1. The spatial pattern in total epiphyte load was not as clear as the pattern in epiphyte abundance (Figure 5b). In general, epiphyte load was highest in western Florida Bay, but there were also peaks in epiphyte load at stations 18 and 20 in central Florida Bay.

Similar to total epiphyte abundance, autotrophic epiphyte abundance also was highest in northwestern Florida Bay (Figure 6a). The median epiphyte Chl-a abundance was 2.7 ug ss^-1, with a range of 0.1 - 106 ug ss^-1. Epiphyte chlorophyll loads ranged from 1.5 ug g^-1 at station 1 in the east to 298.7 ug g^-1 at station 14 in northwestern Florida Bay (Figure 6b). The median epiphyte chlorophyll load was 34.8 ug g^-1. Almost two-thirds (62.5%) of the sites had mean epiphyte chlorophyll loads less than 50 ug g^-1. Aside from the high epiphyte chlorophyll loads measured at stations 18 and 20, mean epiphyte chlorophyll loads generally decreased from higher values in northwest Florida Bay to lower values in extreme northeastern Florida Bay (Figure 6b).

There was not a constant assemblage of autotrophs and heterotrophs making up the total epiphyte load. AI ranged from 0.02 ug mg^-1 at station 8 in northeastern Florida Bay to 0.68 ug mg^-1 at station 17 in southwestern Florida Bay. More than half of the sites (54.2%) had autotrophic indices less than 0.1 ug mg^-1. Autotrophic indices tended to be lower in northeastern Florida Bay and Barnes Sound with higher values towards the southwestern and northwestern areas of Florida Bay (Figure 7).

At the 24 water quality monitoring sites, all measures of epiphyte abundance and load were low in eastern Florida Bay relative to western Florida Bay, owing to the general oligotrophic nature of eastern Florida Bay. This pattern does not hold near Porgy Key, the point source of nutrients in eastern Florida Bay. Mean total epiphyte loads measured along the transect near the point source ranged from 10.6 mg g^-1 to 399.0 mg g ^-1 (Figure 8). The results of the two-way ANOVA indicate significant differences only between transect points (Table 2). As a result of the lack of a significant difference between sampling times, Scheffe's multiple comparison procedure was performed on the time-averaged transect point means. Time-averaged mean epiphyte loads were significantly higher (F = 13.2, P < 0.001) at the transect point located in closest proximity to the bird island. Mean epiphyte loads were 3 to 36 times higher at the transect point located in closest proximity to the island than at the transect points located further from the nutrient source. Significant differences in the time- averaged mean epiphyte loads were only evident between the closest point to the nutrient source and the two more distally located points (Scheffe's, P < 0.05).

Water quality sampling. In February and March 1994, Florida Bay was a polyhaline estuary, with salinities from 18.3 psu in the northeast to 35.4 psu in the center of the bay (Table 3, Figure 9a). Fully 2/3 of the bay was characterized by near marine salinity water ( ca. 34 psu). This is in contrast to the historic salinity characteristics of the bay. Salinity can vary widely in this system, from freshwater in the northeast reaches of the bay to over 60 psu over much of the central and eastern portions of the bay. The southwest margin has the lowest variability in salinity, with a range of 10 psu over the period 1990-1994, while salinity in waters adjacent to the mainland have varied over 48 psu (Figure 9b).

Water column nutrient concentrations varied across Florida Bay for the two months preceding the March 1994 seagrass and epiphyte sampling (Table 3). Total organic carbon (TOC) ranged from 219 uM to 1410 uM with the higher values from sites in central Florida Bay. Organic forms of nitrogen and phosphorus dominated the nutrient pool with both consisting of approximately 90% of their respective total concentrations. Concentrations of both DIP (range 0.01 - 0.33 uM, median = 0.03 uM) and TP (range 0.2 - 1.7 uM, median = 0.4 uM) were quite low baywide (Figure 10). Maximum DIP was found in the center of Florida Bay, while maximum TP was found in northwest Florida Bay. In contrast to P, concentrations of TN (range 19.6 - 142.1 uM) and DIN (1.2 - 22.0 uM) were relatively high (Figure 10). Peak DIN (22.0 uM) was found in eastern Florida Bay; TN was maximum in the north-central part of the bay. Water column Chl-a concentrations were low across the bay (range 0.6 - 6.9 ug L^-1, median =1.6 ug L^-1).

Many of the indicators of nutrient availability were correlated (for example, compare the distribution of C:P of Thalassia testudinum leaves, Figure 3a, with the distribution of water column TP, Figure 10a). In order to define the independent underlying patterns in the nutrient availability data, a Principle Components Analysis was done on the log- transformed nutrient availability data. Three Principal Components (PC1 through PC3) were extracted from the data that described 83% of the variation in the original data set (Table 4). PC1 was highly correlated with measures of inorganic nitrogen concentration in the water column, and accounted for 29.9% of the variance in the original data set. PC2 was positively correlated with TN, TON, TOC, and DIP concentrations in the water column and negatively correlated with C:N of Thalassia testudinum leaf tissue. We interpreted this factor, which explained 28.4% of the variation, to be an indicator of N availability in both the water column and the benthos. A further 24.9% of the variation in the nutrient availability data set was explained by PC3, which was positively correlated with TP, turbidity, and Chl-a concentrations in the water column and negatively correlated with C:P of T. testudinum leaves. This factor represented P availability in both the water column and the seagrass bed.

Stepwise multiple linear regression indicated that the total epiphyte load on Thalassia testudinum from Florida Bay was not significantly related to any of the three principal components of nutrient availability (partial F for all factors <0.9, P for all > 0.10). In contrast, epiphyte chlorophyll load was significantly related to P availability. An example of the raw data indicates the pattern of the relationship between P availability and epiphyte chlorophyll load (Figure 11a); the generalized relationship between epiphyte chlorophyll load and P availability, as represented by PC3, is significant (Figure 11b), although P availability explains only 18.5% of the variation. None of the principal components was significantly correlated with the AI of the epiphyte community, however.

DISCUSSION

We measured gradients in water column nutrients, benthic nutrient availability, short shoot size of the seagrass Thalassia testudinum, and the abundance of epiphytes per short shoot in Florida Bay. Total epiphyte load, measured as the mass of epiphytes per mass of seagrass leaves, was not correlated with measures of nutrient availability, but the load of autotrophic epiphytes was correlated with nutrient availability. We also measured the strong influence of a point source of nutrients on total epiphyte loads in the oligotrophic eastern part of Florida Bay. Our results demonstrate that both water column and benthic nutrient availability may play a role in structuring the epiphytic community in this subtropical seagrass bed, but that the connection between nutrient availability and epiphyte load is not always as strong as indicated in the literature.

There are many factors affecting the distribution of epiphyte loads across both seasonal and regional scales. These factors can be categorized into structural, physico- chemical, grazing, and taxonomical factors. Structural factors affect the space and time for epiphyte colonization and growth and include the variables: seagrass host species, seagrass leaf area and seagrass leaf turnover rate. Physico-chemical factors include nutrient availability, temperature, salinity, depth, current, and light availability and quality. Grazing factors include grazer abundance and grazer species composition. Taxonomical factors include the epiphytic species composition and species dominance. Decreased seagrass leaf turnover rates and elevated nutrient availability have been implicated as the main causal factors for increased epiphyte loads in seagrass beds (Borum 1985; Twilley et al. 1985; Tomasko & Lapointe 1991). Increased epiphyte loading has contributed to declines in seagrass populations (Phillips et al. 1978; Kemp et al. 1983; Cambridge et al. 1986). Due to relatively small regional variation in leaf productivity rates, seagrass leaf turnover rate will effect epiphyte dynamics more in the temporal domain than the spatial domain (Humm 1964, Borum 1985, Zieman et al. 1989, Tomasko & Lapointe 1991, Frankovich 1996). Spatial differences in nutrient availability must therefore be the most important factor controlling the regional variation in epiphyte loads.

There are large gradients in nutrient availability across Florida Bay, and biomass of phytoplankton and the dominant seagrass Thalassia testudinum has been shown to be phosphorus limited (Fourqurean et al. 1992a, 1993). We measured a host of factors related to nutrient availability; Principal Components Analysis indicated that the various proxies for nutrient availability could be reduced to three factors: inorganic nutrient concentration, total nitrogen availability, and total phosphorus availability (Table 4). The water column is the ultimate source of nutrients for the benthos of Florida Bay over geologic time because Florida Bay is a depositional environment; this results in the strong correlation between water column measures of N and P concentration with the N and P content of the seagrass leaves (Table 4). Because other primary producers are P limited in Florida Bay, it is reasonable to expect that epiphytic algae within this system may also be P limited. Epiphyte chlorophyll load was significantly correlated with P availability in Florida Bay (Figure 11), suggesting that the autotrophic component of the epiphyte community may also be P limited. Unlike epiphyte chlorophyll load, total epiphyte load and autotrophic index were not correlated with either N or P availability.

P availability explained only 18% of the variation in epiphyte chlorophyll load (Figure 11), but the majority of the variability in water column chlorophyll-a concentrations can be described by water quality nutrient variables, particularly those of phosphorus availability. A stepwise multiple linear regression revealed that 83% of the variation in water column chlorophyll-a can be explained by total phosphorus concentrations (r = 0.91, P < 0.001). This disparity suggests that the factors controlling epiphyte accumulation are much more complex than those controlling phytoplankton levels.

Relative P availability is a primary factor determining the biomass of benthic plants in Florida Bay (Fourqurean et al. 1992a), but it is not the only factor. Since 1987, Florida Bay has experienced the rapid decline in seagrass biomass in areas that formerly supported dense Thalassia testudinum communities; over 40,000 ha have been effected by this die-off (Robblee et al. 1991, Durako 1994). While the causes of this phenomenon are not well understood, the changes in the ecosystem caused by the death of the dominant primary producer have been followed closely. Epiphytism did not play a role in the initial loss of seagrasses (Robblee et al. 1991 and Personal Observations), but nutrient releases subsequent to the death of seagrasses led to increased phytoplankton concentrations in areas that had experienced seagrass mortality (Phlips & Badylak 1996). We also noted heavy accumulations of benthic diatoms on the leaves of surviving seagrasses in the months following the die-off; we believe that the increases microalgae in the water column and adhering to seagrasses was fueled by nutrients released by decaying seagrasses. Data on the N and P content of seagrass leaves from both before and after the die-off show that these nutrient releases did not appreciably change the regional pattern of relative N and P availability in Florida Bay. The spatial distributions of T. testudinum leaf C:P and C:N ratios measured in this study (Figure 3) are identical to those observed prior to seagrass die off (Fourqurean et al. 1992a). This indicates that the massive seagrass die-off that drastically decreased T. testudinum biomass and shoot density did not affect the wholesale distribution of nutrients within the ecosystem.

The amount of seagrass leaf biomass determines the amount of space available for colonization and growth of epiphytic biomass. The trend of increasing Thalassia testudinum short shoot size from eastern towards western Florida Bay (Figure 2) is similar to the gradient in seagrass standing crop in the ecosystem (Zieman et al. 1989). The regional distribution of T. testudinum leaf biomass per short shoot (Figure 2) and epiphyte abundance per short shoot (Figure 5) show a similar pattern. In Florida Bay, approximately 35% of the variation in epiphyte material on T. testudinum short shoots can be accounted for by leaf biomass alone (Frankovich 1996). The increase in epiphyte abundance (on a per short shoot basis) along this gradient is likely due to the increase in seagrass leaf substratum. In general, total epiphyte loads were greatest in western Florida Bay and were lowest within the tidally-restricted areas within central and northeastern Florida Bay.

The large differences in mean total epiphyte loads along the bird island transect are likely due to nutrient enrichment from bird defecation on Porgy Key, a bird colony island. Eutrophication of the seagrass community causes biomass and species composition changes close to bird colonies in Florida Bay (Powell et al. 1991). The highest mean total epiphyte loads were always located at the transect point closest to the island (distance = 15 m). Epiphyte loads decreased drastically between the closest point and the two more distally located points (Figure 8). Differences in epiphytic species composition were also observed along this transect. Various "fleshy" rhodophytes and chlorophytes (Chondria sp., Ceramium spp., Laurencia spp., and Derbesia sp.) were abundant close to the island (15m). Lesser amounts of Chondria were observed at the middle transect point (distance = 30 m), while epiphyte composition furthest from the island (45 m) was dominated by animal epiphytes(especially the epiphytic bivalves Pinctada imbricata and Brachidontes exustus ). Along a similar transect at Porgy Key, Powell et al. (1991) documented an increase in seagrass biomass and a change in seagrass species composition from Thalassia testudinum to Halodule wrightii as distance decreased from the bird island. These changes in seagrass biomass and species composition are coincident with the changes in epiphyte loads and species composition of the present study. The absence of significant differences in the total epiphyte loads at 30 and 45 m from the island and the observed change in epiphytic species composition along the transect suggest that the effect of bird colony nutrient enrichment on epiphyte levels is pronounced but very localized, not extending even 30 meters from the island. In contrast, seagrass standing crop was enhanced up to 200 meters from the bird island (Powell et al. 1991) and the nutrient content of the seagrass biomass showed elevated phosphorus as far as 90 meters from the bird island (Fourqurean et al. 1992). Although epiphyte load is often touted as an indicator of nutrient availability in seagrass ecosystems, these findings suggest that seagrass biomass, species composition, and nutrient content are more sensitive indicators of nutrient availability than epiphyte load.

The trend of decreasing epiphyte chlorophyll loads from northwest to northeast Florida Bay and Barnes Sound is coincident with a gradual epiphytic community composition change (Table 1). Qualitative observations indicate that algal epiphytes, specifically the coralline red algae Melobesia membranacea and Fosliella farinosa , were abundant in western Florida Bay. The amount of these encrusting algae decreased towards the more-restricted interior of Florida Bay. The opposite trend was noticed for the animal epiphytes (the polychaete Spirorbis sp. and the molluscs Brachidontes exustus and Pinctada imbricata ), which were more abundant in eastern Florida Bay. The relative importance of algal epiphytes to the total epiphyte community, as measured by the AI, was minimum in the northeast parts of Florida Bay, and maximum in the southwest part of the bay (Figure 7).

The observation of a shift from algal epiphytes in western Florida Bay to animal epiphytes in the more tidally-restricted central and eastern areas of the bay suggests that a relationship may exist between the level of algal epiphytism and salinity or salinity variability. Even though no relationship was observed between epiphyte load and salinity or the variability thereof (P > 0.05), the salinity and the range in salinity that an individual site experiences may be a prominent factor in the epiphytic species composition at that site. The paucity of epiphytic coralline red algae and the occurrence of the epiphytic bivalves Pinctada imbricata and Brachidontes exustus in the eastern and central areas of Florida Bay is coincident with high and very often variable salinities experienced in those areas (Figure 9, Schomer & Drew 1982; Robblee et al., 1989; Fourqurean et al. 1992). Harlin et al. (1985) attributed the decrease in the density of epiphytic coralline red algae in the interior of Shark Bay, Australia to increases in hypersalinity. In contrast, the relative abundance of the previously mentioned epiphytic molluscs in the interior of Florida Bay has been attributed to their ability to withstand "large salinity fluctuations and other effects of poor circulation" (Turney & Perkins 1972). Along a hyper-salinity gradient in Shark Bay, Western Australia, the diversity and density of epiphyte species decreased significantly as levels of hyper-salinity increased (Harlin et al. 1985; Kendrick et al. 1988). The standing crop of epiphytic diatoms on Halodule wrightii also decreased significantly along a hyper-salinity gradient in South Texas (Jewett-Smith 1991).

Recent work has indicated the important role that losses of epiphytes due to grazing pressure in regulating epiphyte density (Howard 1982, Hootsman & Vermaat 1985, Neckles et al. 1993, 1994). In some environments, grazing of epiphytes is necessary for seagrass survival, lest epiphyte fouling become so severe as to outcompete seagrasses for light (Wetzel & Neckles 1986, Borum 1987). The removal of epiphytes by grazers may obfuscate the relationship between nutrient availability and epiphyte biomass accumulation. By the definition applied to epiphytes in this study, seagrass epiphytes are sessile. As such, they may be more susceptible to grazing pressure. The activity of epifaunal grazers may be responsible for much of the variation in epiphyte chlorophyll loads which was not described by water column total phosphorus concentrations. The presence or absence of field densities of grazing epifauna was found to have stronger effects on epiphyte biomass than increases in nutrient loading (Neckles et al. 1993). Grazing pressure was also found to be more important than leaf turnover rate in controlling epiphyte loads on Zostera marina in Roskilde fjord (Borum 1987). It therefore seems likely that epifaunal grazers have great control of algal epiphyte standing stock, while water column nutrient parameters, particularly those of phosphorus availability, may control algal epiphyte productivity, but not standing stock, in Florida Bay.

Traditionally, increased epiphyte levels and elevated water column nutrient concentrations were pointed to as evidence of nutrient enrichment (Larkum 1976, Cambridge & McComb 1984, Silberstein et al. 1986, Tomasko & Lapointe 1991), which subsequently led to declines in seagrass ecosystem health (see Duarte 1995 for review). Recent studies have shown that these traditional indicators do not accurately describe nutrient availability within the ecosystem (Bulthuis et al. 1992, Neckles et al. 1993, 1994; Lin et al 1996, Tomasko et al 1996). Active uptake of increased nutrient loading keeps water column nutrient concentrations low (Lin et al. 1996) and top-down grazer control of epiphyte or macroalgal biomass may mask any obvious effect of nutrient enrichment (Neckles et al. 1993, 1994). It can be hypothesized that the nutrient concentration gradient in Florida Bay (particularly that of phosphorus availability), though pronounced, is under the threshold for epiphyte super-abundance. At the present level of epiphyte productivity, grazing pressure may exert sufficient control over epiphyte levels. In Florida Bay, only in the immediate proximity to the bird island, a point source of nutrients, was phosphorus availability high enough to cause increases in epiphyte levels. The uptake and storage of nutrients by seagrasses (Fourqurean et al. 1992a, 1992b) and their resistance to grazing pressure (Harrison 1989) makes them ideal bio-indicators of ecosystem health. This study has shown that the nutrient content of seagrass leaf tissue is a better indicator of nutrient availability than either water column nutrient concentrations or epiphyte levels. Elevated water column nutrient concentrations and increased epiphyte loads are obvious but, unfortunately, may be late indicators of nutrient enrichment, and these symptoms may only become evident when seagrass ecosystems are already in decline.

ACKNOWLEDGMENTS

The major portion of this work was supported by the South Florida Water Management District and the National Park Service, Everglades National Park under CA5280-2-9017 to R.D. Jones at FIU. Additional funding was supplied by grant CA5000-0-9010/2 to J.C. Zieman from the National Park Service Southeast Regional Office. The authors wish to thank R.D. Jones, Director, Southeast Environmental Research Program (SERP), Florida International University (FIU), and the entire SERP laboratory for supplying the water quality data. We thank S. Quackenbush of FIU for the use of a freeze-drier. Boat support was generously provided by D. Smith of the Southeast Research Center of the Everglades National Park. The authors wish to extend a special thanks to J.C. Zieman of the University of Virginia for introducing them to the ecosystems of south Florida and enabling them to work on seagrasses in Florida Bay. This manuscript is contribution number ? of the Southeast Environmental Research Program, Florida International University.