Abstract Introduction Acknowledgments Methods Results Discussion Conclusions References

METHODS OF INVESTIGATION

Field Methods

Twenty-six sites throughout Florida Bay (Text-fig. 1, Table 1), visited biannually between February 1995 and July 1999, constitute the source of modern observations. Biannual observation of sites always occurred between February 12-23 and July 6-13 for seasonal consistency. Sites 1-14, under regular observation since February 1995, have been visited consistently twice a year. Sites 15-26, added to the observations after February 1995, were typically less accessible in the 'wet' season (May-October) and as a result underwent more regular observation during 'dry' seasons (November-April) and less consistent observation during 'wet' seasons. Observations were obtained by wading or snorkeling at each site and recording the substrate, fauna and flora present (living and dead), and microhabitats. Observation procedures did not include digging within the sediment for infaunal organisms or a systematic search of the site, but represent a careful examination of the area. One hundred and seventy-five detailed site observations from February 1996 to July 1999 are analyzed statistically in this paper.

Text-figure 1. - Map of Florida Bay showing location of 26 monitoring sites and four cores. Boundary of Everglades National Park is shown by bold dashed line. Table 1 gives latitude and longitude. [larger image]

Each observation was coupled with a characterization of the water at the site and a sampling of the sediment. Water was characterized by salinity, temperature, depth, and clarity. Salinity values were calculated using a Hydrolab Surveyor 4, YSI Model 30/10 FT, or a refractometer. Temperature values were obtained using a Hydrolab Surveyor 4 or YSI Model 30/10 FT.

In February and July 1998, seventeen sub-aquatic vegetation samples (macro-benthic algae and seagrass) were collected at sites 8, 12, 13, and 20. Sub-aquatic vegetation was sampled randomly, to characterize the types of flora present, and the associated epifaunal organisms. Sub-aquatic vegetation was collected without the roots or holdfasts (where present) to reduce the chance of contamination by infaunal organisms. Vegetation samples were collected with a small amount of concomitant water in an attempt to delay the effects of biodegradation before laboratory examination of the samples.

Sediment samples were collected using a 1.5 inch diameter PVC tube (push core). Push cores were pressed into the substrate, and a vacuum was maintained to extract the sediment. Push core tubes were cut at the water-substrate interface and sealed to prevent disruption of sediment during transport to the laboratory. The twenty-eight push cores described in this study were taken from four sites in northern and eastern Florida Bay (Text-fig. 1) biannually between February 1995 and July 1998 (collection at site 20 began February 1997), and represent a subset of the total number of push cores obtained. Push core samples were chosen for this study based upon proximity of modern sampling sites to the location of the four piston cores.

The four short piston cores (<2m) examined were taken from Florida Bay localities (Text-fig. 1) between February 1994 and March 1996 by researchers from the U. S. Geological Survey (St. Petersburg, Florida), in cooperation with South Florida Water Management District, Everglades National Park, and National Oceanic and Atmospheric Administration (NOAA). Modern sites 12, 13, and 20 occur at the same location as piston cores RB19B, BA6A, and PK37, respectively. Site 8, located at the mouth of Little Madeira Bay, is the closest modern observation site to the site of piston core T-24, located at the mouth of Taylor Creek in Little Madeira Bay.

Laboratory Methods

Vegetation samples were characterized by type of sub-aquatic vegetation present and condition of the vegetation. All vegetation samples were then carefully examined for extant epifaunal organisms before processing. Samples were vigorously scrubbed to remove epifaunal organisms, washed through 850µm and 63µm sieves, and dried at 50ºC. Dry vegetation material from each sample was weighed, re-hydrated, and washed through 850µm and 63µm sieves a second time to help remove any remaining epifaunal organisms.

All push cores were cut to 10 cm below the water-substrate interface, and this upper 10 cm of sediment was retained for processing. Twenty-eight push core samples were analyzed for faunal content. Piston cores collected for historical reconstruction were x-rayed and visually examined for signs of sediment disruption, including bioturbation. Samples from the piston cores were submitted for ²¹⁰Pb analysis (see Robbins, et al. in press, for method; Wingard et al., 1995, and Brewster-Wingard et al., 1997 for ²¹⁰Pb results on cores discussed here). All four piston cores were sampled at 2-cm intervals. All push core and piston core samples were washed through 850µm and 63µm sieves to remove carbonate mud. Samples were dried at 50ºC.

All recognizable molluscs and molluscan fragments were picked from the 850µm-size fraction from each vegetation, push core, and piston core sample. Picked specimens were sorted, identified, and characterized by condition of the shell material present. Specimens were characterized by stage (adult or juvenile) as well as pristine (intact and retaining original shell material), whole (intact but lacking original shell material or color), broken (retaining original shell material and more than 50% of shell), worn, or fragmented (retaining less than 50% of the shell but still recognizable at the generic level).

Age models for the piston cores 6A and 19B are based on ²¹⁰Pb analysis of the <63µm-size fraction of the samples. For details of the method, see Holmes, et al. (this volume).

Analytical Methods

Univariate and bivariate statistical analyses, including minimum, maximum, mean, median, and standard deviation, were conducted on environmental data (salinity and temperature), and individual species distributions using Microsoft Excel 97. For site observations, the presence- absence data were simplified for statistical analysis by deleting samples with no molluscs present and removing faunal groups that occur less than two times throughout the dataset. Observation data were translated into a presence-absence data matrix ¹. For the piston core, push-core and vegetation data, absolute species abundances were standardized by calculating percent relative abundance of the total molluscan component of the sample. Worn and fragmented molluscan specimens were removed from the analysis of vegetation samples, and from one run of push core samples, in order to isolate extant material from older material.

All multivariate analyses were done using MVSP (Kovach Computing Services, MVSP Plus, version 3.1) following the methods described by Kovach (1989, 1995). A cluster analysis of the presence-absence data was conducted using unweighted paired group method, average-linkage (UPGMA), with Sorenson's Coefficient distance measurement, dual clustering procedure, and random input order. Cluster analysis of percent abundance data were conducted with log-ratio transformed and centered data, then clustered using UPGMA, with cosine theta distance measurement, dual clustering procedure, and random input order. Each cluster analysis was run several times using random input order to determine robustness of clusters.

Constancy and fidelity measures were calculated for each cluster in both the presence-absence data analysis and the percent abundance data analysis. Method was adapted from Hazel (1977). Constancy is a measure of how frequently a species occurs within a given cluster, expressed as a percentage. The formula is

A species with 100% constancy is present in every sample in a given cluster. Fidelity is a measure of how unique a species is to a given cluster (expressed as a percentage). The formula is

Thus, it is possible for a species to have a high constancy value, if it is present in most samples in a cluster, but a low fidelity value if it also is commonly present in other clusters. Conversely, a species can be very rare and have a low constancy for a cluster, but a high fidelity because it occurs only in that cluster. A species with 100% fidelity is unique to that cluster.

Several measures of diversity were calculated on all piston core samples: 1) total number of individual molluscan specimens; 2) faunal richness; 3) evenness; and 4) Shannon's diversity index. Faunal richness in this study is a measure of the number of faunal groups present in a given sample (in the case of the presence-absence data observed at a given site). In some cases, the fauna are grouped into genera and occasionally broader categories (eg. marginellids), so this is not "species" richness in the usual sense. Evenness and Shannon's diversity index were calculated using MVSP. Evenness is a measure of how evenly dispersed the total number of individuals are among the faunal categories; the higher the value, the more evenly dispersed the individuals are. Shannon's diversity index (Shannon and Weaver, 1949; Ludwig and Reynolds, 1988) is a measure of the degree of uncertainty that a randomly selected individual will belong to a certain species. The formula is as follows:

	s
H' = -		pi log pi
	i=1

where s is the number of species of know proportions (p₁, p₂, p₃, . . . p_s) (see Ludwig and Reynolds, 1988, p. 92, for complete explanation).