Abstract Introduction Acknowledgments Methods Results Discussion Conclusions References

RESULTS

Modern Environmental Data

Text-figure 2 shows seasonal and annual variations recorded in temperature and salinity at our monitoring sites in Florida Bay. These measurements show that bay-wide salinity is less variable in the winter (dry) season than during the summer (wet) season, but the opposite is true for water temperature. Recorded winter water temperatures have varied more than 10^oC from one year to the next, ranging from a low of 16.5^oC (site 5, 2/96) to a high of 28.5^oC (site 16, 2/97). Summer water temperatures have ranged from a low of 28.3^oC (site 9, 7/99) to a high of 34.6^oC (site 12, 7/98). Seasonal standard deviations for the period of record are relatively low for water temperatures at each site, and water temperatures do not display a strong geographic gradient (Table 2).

Text-figure 2. - Plot of temperature (oC) versus salinity (ppt) recorded at sites from February 1996 through July 1999. Histogram on x axis shows frequency of observations in 5 ppt increments. Histogram on y axis shows frequency of observations in 5oC increments. Bimodal nature of temperature readings reflects seasonality. [larger image]

In contrast to temperature, salinity does display a strong geographic gradient. Text-figure 3 illustrates seasonal changes in salinity and Table 3 lists standard deviations calculated for each site. In general, standard deviations for salinity decrease as you move to the south and west in the bay, indicating more stable salinity on a seasonal and annual basis. Summer salinity standard deviation values are significantly higher at most sites than the winter values. Salinity at three sites (1, 2, and 8) in the northern transition zone has varied over 20 ppt during the July sampling period from 1996-1999, compared to winter variations at the same sites of 5-12 ppt. Bay-wide salinity has ranged from a low of 1.4 ppt (site 2, 7/97) to a high of 41.3 ppt (site 22, 7/99).

[click on maps above to view larger images]
Text-figure 3. - Maps showing 5 ppt salinity gradients and the prominent cluster designations from the presence-absence Q-mode analysis for the sampling periods from February 1996 through July 1999. Contours are primarily based on our own salinity measurements, however, in areas of sparse sampling we utilized data from Halley (http://sofia.usgs.gov/projects/circulation) to fill in the gaps. [click on maps above to view larger images]

Modern Molluscan Presence-Absence Data

An unweighted pair-group method (UPGM) dual-cluster analysis, using Sorenson's coefficient on the modern presence-absence faunal dataset (rares deleted), produced the clusters seen in Text-figures 4 and 5. The dual clustering method (Kovach, 1989, 1995) allows direct comparison between the separate Q-mode and R-mode analyses. Q-mode analysis produced seven primary clusters, grouping the modern sample sites based on species found alive at those sites. An examination of the clusters and the associated data matrix, sorted in dendrogram order, indicated which fauna are responsible for the clustering of sample sites seen. Text-figure 3 shows the distribution of the clusters by sampling period.

Text-figure 4. - Q-mode cluster of presence-absence data from modern monitoring sites. Cluster was generated through MVSP statistical package (Kovach Computing Services, MVSP Plus, version 3.1) using unweighted paired group method, average-linkage (UPGMA), with Sorenson's Coefficient distance measurement, dual clustering procedure, and random input order. Full data matrix sorted in dendogram order and labeled clusters can be viewed at http://flaecohist.er.usgs.gov/database/ Reference/synthesis. Table 4 indicates which species are responsible for the clusters seen. [larger image]

Cluster A groups samples from sites 6, 7, 11, 12 and 14 in eastern and central Florida Bay. The occurrence of Batillaria minima and/or Melongena corona defines this cluster; constancy (C) and fidelity (F) for Batillaria minima equal 83.3 and 80.0 and for Melongena corona 50.0 and 70.3 (Table 4). Fasciolaria spp. and Truncatella spp. also have high fidelity values for this cluster (Table 4). Sites in cluster A can be characterized as having a subenvironment with shallow (< 20 cm) water depth in relatively close proximity to mangrove islands, sparse vegetation (< 20% Halodule, scattered Thalassia may be present), and soft, "gelatinous" calcareous mud is typical (799-6 is an exception). Average salinity for the cluster is 28.4 ppt, and the range is 18.7-41.0 ppt. Average observed faunal richness² at sample sites in cluster A is three, with a maximum of six faunal groups seen alive during one observation period.

Text-figure 5. - R-mode cluster of presence-absence data from modern monitoring sites. Cluster analysis and full data matrix as in Text-figure 4. [larger image]

Cluster B is defined by the presence of Ostrea equestris (C=76.9, F=70.5) and/or Arcopsis adamsi (C=46.2, F=87.6), with Brachidontes exustus present in nine of the thirteen samples in this cluster (Table 4). The sample sites (1, 2, 3, 8, 9, and 24) are geographically widespread, encompassing the northern transitional zone, eastern, and western portions of Florida Bay. The common characteristic of these sites is the presence of wood (mangrove roots, driftwood, or pilings) or exposed limestone bedrock that provides an attachment-point for the cemented Ostrea equestris, or bysally attached Arcopsis adamsi. Salinity values for the cluster range from 2.7-35.1 ppt, and the average is 18.6 ppt. A maximum of seven faunal groups were found alive at sites in cluster B, and the average observed faunal richness is three.

Cluster C comprises the largest group of samples (70) from sites 1-17, and 20-22 in the northern transitional, eastern and central portions of Florida Bay. The nearly ubiquitous occurrence of Brachidontes exustus (66 of our 70 samples, C=94.3, F=35.5) distinguishes this cluster (Table 4). Within the larger cluster C are six smaller groupings, characterized by species that co-occur with Brachidontes exustus. Average salinity for cluster C is 28.1 ppt, and the range is 10.9-41.3 ppt. The environments represented by the 20 sites in this cluster are varied, including mud banks, mangrove islands, open basins, and channels, but all have at least some subaquatic vegetation present. Sites within cluster C attain the highest values for observed faunal richness; the average is five and the maximum value is 20.

Cluster D includes samples from eastern, central, western, and Atlantic transition zone sites (3, 5-7, 10, 11, 13, 15-18, 20, 21, and 26). The occurrence of Pteria longisquamosa [= Pinctada radiata Turney and Perkins] (C=85.2, F=45.7) distinguishes cluster D; and the occurrence of Modulus modulus (C=51.9, F=71.8) is significant (Table 4). The environments represented by these samples can generally be characterized by the presence of healthy Thalassia beds (exceptions, 6, 10, and 26). Average salinity for the samples is 29.9 ppt and the range is 19.8-39.7 ppt. Observed faunal richness at sites in cluster D averages four, and the maximum value is eleven.

A small group of five samples from the northern transitional zone, and eastern and central Florida Bay (sites 2, 3, 9, and 21) constitute Cluster E. The presence of Laevicardium mortoni (C=100, F=85.5) distinguishes this cluster (Table 4). All of the sites have a subenvironment with exposed substrate and sparse vegetation. The average salinity for the cluster is 24.0 ppt and the range is 12.7-34.7 ppt. Observed faunal richness ranges from 2-6, and the average is 4.

Cluster F includes two distinct sub-clusters, one distinguished by the co-occurrence of Turbo castanea (C=50, F=50.5), Tegula fasciata (C=38.9, F=66) and Columbella spp. (C=55.6, F=77.7), the other by Columbella spp., Cerithium spp. (C=50, F=64.4), Chione cancellata (C=27.8, F=52.2) and Carditamera floridana (C=33.3, F=66.9)(Table 4). The cluster includes sites from the central, western, Gulf transition, and Atlantic transition zones of Florida Bay (13, 17, 18, 21, 23, 24, and 26). The sites can be characterized as having lush subaquatic vegetation and bare sediment areas available. Average salinity for cluster F is 34.6 ppt, and the range is 29.0-39.3 ppt. Average observed faunal richness is four and the maximum value is eight for the sites in cluster F.

Cluster G comprises five samples from the western-most sites (23 and 24) in the Gulf transition zone. Pinnidae (C=60, F=80.2), and Pleuroploca gigantea (C=40, F=100) define this cluster (Table 4). Average salinity is 34.1 ppt, and the range is 31.9 to 36.8 ppt. These sites are predominantly calcareous sand bars, with lush beds of mixed subaquatic vegetation present, including Thalassia, Halodule, Syringodium, and many species of calcareous algae, and macro-benthic red, green and brown algae. The sites at cluster G have the lowest observed faunal richness, with an average of two and a maximum of five faunal groups found alive at the sites.

R-mode analysis produced two primary divisions of the faunal categories, based on the samples in which they are found (Text-fig. 5). In general, fauna in cluster I tolerate a wide range of conditions and can be found in the northern transitional, eastern, central and western portions of the bay. Fauna in cluster I were found alive in salinities ranging from 2.7-41.3 ppt. The remainder of the faunal groups fall into cluster II, with the exception of Muricidae sp., a rare species that is an outlier on the dendrogram. Cluster II groups fauna that are typically found in the western portions of the Bay, or at sites with near normal marine salinities. The average salinity recorded at sites where fauna from cluster II are found is 33.3 ppt and the range is 17.6-40.0 ppt. The exception in cluster II is Melampus coffeus. The presence of Melampus is indicative only of the presence of a supratidal environment and the presence of mangroves.

Modern Vegetation Sample Data

Sixteen vegetation samples and one surficial mud sample were analyzed from sites 8, 12, 13, and 20, collected in February and July 1998.¹ Brachidontes exustus and Bittiolum varium account for 95% (10,981 individuals) of the 35 molluscan faunal groups found alive in the samples from all four sites. These two species were most frequently found in polytypic sub-aquatic vegetation samples, including Thalassia and macro-benthic algae. Crepidula spp., Cerithium muscarum, Pteria longisquamosa and Prunum spp. combined constitute only 2.25% (260 individuals) of the living fauna collected.

Modern Push Core Data

Twenty-eight push core samples were analyzed from sites 8, 12, 13, and 20, collected in February and July beginning in 1995.¹ The push core samples represent death assemblages at each site, and thus the push core data provide a link between modern living assemblage data and down-core assemblages from the core data set. Two cluster analyses were done (Text-fig. 6); one included all specimens, regardless of preservation; the other was limited to only pristine and broken shells. The purpose was to contrast the death assemblage at each site with the living or recently expired fauna. The two clusters produced similar results. All samples from site 8 formed a distinct cluster in both analyses. In general, samples tended to cluster with other samples from the same site in both analyses.

Text-figure 6. - Q-mode cluster of push-core data from modern monitoring sites. Top cluster illustrates the results when only the pristine and broken (retaining original material and more than 50% of shell) material are included. Bottom cluster includes all identifiable molluscan components. Clusters were generated using MVSP statistical package (Kovach Computing Services, MVSP Plus, version 3.1) on log-ratio transformed and centered percent abundance data, using unweighted paired group method, average-linkage (UPGMA), with cosine theta distance measurement, dual clustering procedure, and random input order. [larger image]

Brachidontes exustus and Bittiolum varium are among the four most abundant species in the push cores at all four sites (8, 12, 13, and 20), based on the cumulative data. When the individual push core data are examined for each site, Bittiolum varium is among the four most abundant species and Brachidontes exustus is among the six most abundant species. Transennella sp. is among the four most abundant species at sites 12, 13, and 20, and Crepidula spp. at sites 12 and 13. The discrete clustering of site 8 can be explained by the prominence of Hydrobiidae, Truncatella bilabiata, and Anomalocardia auberiana.

Historical Piston Core Data

Details of the individual core analyses have been previously published (Brewster-Wingard et al., 1997, 1998a, 1998b; Brewster-Wingard and Ishman 1999; Ishman et al., 1996; Wingard et al., 1995).³ Text-figures 7-10 illustrate the down-core distribution of key species identified in the modern presence-absence data analysis and in previous core analyses. Shannon's diversity index, faunal evenness, the number of faunal groups, and the number of specimens present are also shown on the down-core plots (Text-figs. 7-10). The age model for cores 6A and 19B is based on sedimentation rates calculated from ²¹⁰Pb analysis. The rate for core 6A is 0.75 cm/yr ± 0.08 and for core 19B is 1.27 cm/yr ± 0.08 (Holmes, et al., this volume).

Text-figure 7. - Down-core changes in core T24 from the mouth of Taylor Creek in Little Madeira Bay. Columns illustrate changes in Q-mode clusters (see Text-fig. 11), significant molluscan fauna (shown as percent abundance), Shannon's diversity index, evenness, number of faunal groups, and number of individual specimens. RXIc group refers to the R-mode cluster shown in Text-figure 12; this group is composed primarily of species that prefer salinities less than 18 ppt. Depth is shown in cm. [larger image]

Text-figure 8. - Down-core changes in core PK37 from south of Pass Key in eastern Florida Bay. Columns illustrate changes in Q-mode clusters (see Text-fig. 11), significant molluscan fauna (shown as percent abundance), Shannon's diversity index, evenness, number of faunal groups, and number of individual specimens. RXIc group refers to the R-mode cluster shown in Text-figure 12; this group is composed primarily of species that prefer salinities less than 18 ppt. Depth is shown in cm. [larger image]

Text-figure 9. - Down-core changes in core RB19B from Russell Bank in eastern Florida Bay. Columns illustrate changes in Q-mode clusters (see Text-fig. 11), significant molluscan fauna (shown as percent abundance), Shannon's diversity index, evenness, number of faunal groups, and number of individual specimens. Rainfall data is compiled from NOAA NCDC website http://www.ncdc.noaa.gov/. RXIc group refers to the R-mode cluster shown in Text-figure 12; this group is composed primarily of species that prefer salinities less than 18 ppt. Depth is shown in cm. Age is based on 210Pb analysis (Holmes et al., this volume). [larger image]

Text-figure 10. - Down-core changes in core BA6A from Bob Allen mudbank in central Florida Bay. Columns illustrate changes in Q-mode clusters (see Text-fig.11), significant molluscan fauna (shown as percent abundance), Shannon’s diversity index, evenness, number of faunal groups, and number of individual specimens. Rainfall data were compiled from NOAA NCDC website http://www.ncdc.noaa.gov/. Depth is shown in cm. Age is based on 210Pb analysis (Holmes, et al., this volume). [larger image]

Compiled Data Set

Percent abundance data were compiled from the modern vegetation samples and push cores from sites 8, 12, 13, and 20, and from the corresponding piston cores. The compiled database was analyzed using the unweighted pair-group method (UPGM) dual-cluster procedure, with cosine theta distance measures on a log-transformed data matrix (Text-figs. 11 and 12). Q-mode analysis produced three main clusters, grouping the individual samples based on the fauna they contain. Faunal groups responsible for the Q-mode clusters were determined by an examination of the associated data matrix, sorted in dendrogram order.⁴

Text-Figure 11. - (above left) Q-mode cluster of compiled data (cores, push-cores, and vegetation samples). Clusters were generated using MVSP statistical package (Kovach Computing Services, MVSP Plus, version 3.1) on log-ratio transformed and centered percent abundance data, using unweighted paired group method, average-linkage (UPGMA), with cosine theta distance measurement, dual clustering procedure, and random input order. Full data matrix sorted in dendrogram order and labeled clusters can be viewed at http://flaecohist.er.usgs.gov/database/Reference/synthesis. Relationship of clusters to samples can be seen in Text-figure 14 and Table 5 indicates which species are responsible for the clusters seen. [larger image] Text-Figure 12.- (above right) R-mode cluster of compiled data (cores, push-cores, and vegetation samples). Clusters were generated using MVSP statistical package (Kovach Computing Services, MVSP Plus, version 3.1) on log-ratio transformed and centered percent abundance data, using unweighted paired group method, average-linkage (UPGMA), with cosine theta distance measurement, dual clustering procedure, and random input order. Full data matrix sorted in dendrogram order can be viewed at http://flaecohist.er.usgs.gov/database/Reference/synthesis. [larger image]

Cluster J is divided into two very distinct components (Text-fig. 11). Cluster J1 is primarily composed of modern push core samples, the majority of these from site 8. Cluster J2 is exclusively composed of samples from core T24. A diverse faunal assemblage, blending components of R-mode clusters XIa and XIc (Text-fig. 12) defines cluster J. Anomalocardia auberiana (C=93.4, F=65.4), Acteocina canaliculata (C=93.4, F=69), and Hydrobiidae (C=85.2, F=92.9) are the dominant fauna in cluster J, and Polymesoda maritima, and Cerithidea spp. are characteristic (Table 5). Cluster J1 is further distinguished by a group of species from R-mode cluster XII (Odostomia spp., Prunum sp., Schwartziella catesbyana, Bittiolum varium juv., and Truncatella spp.). A number of faunal groups occur in cluster J that are absent in other clusters (F=100, Table 5), indicating cluster J represents a relatively unique and diverse assemblage. These results are consistent with the analysis of the push cores (Text-fig. 6); site 8 push cores formed a distinct cluster, separate from the other sites. The average diversity measures for cluster J are relatively high: Shannon's diversity index is 2.03; evenness is 0.74; and faunal richness is 16.8.

Cluster K represents a mixture of samples from cores 6A, 19B, and 37, with modern push core samples from sites 12, 13, and 20, and modern vegetation samples from sites 8, 12,13, and 20. The common unifying element for cluster K is the presence and relatively high abundance of Brachidontes exustus in all but a few samples (average abundance 32.13%, C=98.1). A number of species are present in a high percentage of samples (Table 5), but at relatively low percent abundances, forming a distinct and diverse assemblage defined by R-mode clusters XIa and XIb. Cluster K can be subdivided into four components (Text-fig. 11). Cluster K1 is predominantly samples from cores 19B and 37. In addition to Brachidontes exustus, cluster K1 is defined by the presence of Bittiolum varium, Cerithium muscarum juv., and Rissoidae. The core samples in K1 can be further divided based on the abundance of Pteria longisquamosa, Modulus modulus and marginellids in K1a and Transennella sp., Chione cancellata, Olivella pusilla, and Vermicularia sp. in K1b. Cluster K2 is faunally diverse and contains a mixture of vegetation, push core and core samples primarily defined by R-mode cluster XIa species, including Laevicardium mortoni and Tellina spp, which distinguishes this cluster from K1. The occurrence of Cerithidea spp. with abundant Brachidontes exustus isolates the four samples that constitute cluster K3. Cluster K4 is primarily vegetation samples and is lower in diversity that the rest of cluster K. Bittiolum varium, B. varium juv. and Brachidontes exustus dominate cluster K4. Average Shannon's diversity index for cluster K is 1.79; evenness is 0.71; and faunal richness is 14.3

Cluster L is predominantly samples from the lower portion of core 6A. The prevalence of Transennella sp. (C=100, F=39.7) defines this cluster (Table 5). Average abundance of Transennella sp. in samples from cluster L is 64.21%. Measures of diversity for cluster L are significantly lower than for clusters J and K. Average values for cluster L are 0.60 for Shannon's diversity index, 0.45 for evenness, and 3.7 for faunal richness.

³ All open-file reports and data tables for each core are available on line at http://sofia.usgs.gov/flaecohist.

⁴ See footnote 1.