SANTA ROSA - SWIFT CREEK IN NORTHWEST FLORIDA
Judith A. Bense
Paper Presented at the 49th Annual Meeting of the
Southeastern Archaeological Conference
Little Rock, Arkansas
This paper is an overview of the Santa Rosa-Swift Creek Middle Woodland cultural period in northwest Florida. We are currently conducting research on this period in the Pensacola area, and I will try to set the stage for the next paper which will report on our excavations at a Santa Rosa-Swift Creek site. I will describe the environment of northwest Florida and review the archaeological information about the this culture including site patterns, site types, site distribution, classes, burial mounds, material culture, and subsistence.
ENVIRONMENT OF NORTHWEST FLORIDA
Northwest Florida is in the western panhandle of the state and our study area is approximately 150 miles long located between Perdido Bay on the west and St. Andrew Bay on the east and it extends into the interior 30 and 75 miles. The interior is characterized by a series of sandy, well-dissected marine terraces and southward flowing streams organized into four drainage systems which empty into the Gulf. The Pensacola watershed is the largest in the area as it is fed by three major rivers. Choctawhatchee watershed is next in size with only one major river and the St. Andrew Bay drainage system is the smallest with only a series of third and fourth order streams draining the karst-sand hill region just to the north. This karst region is confined to a relatively small area where limestone containing the Florida aquifer is at the surface and sinkhole lakes dot the landscape along with overflow streams and wetlands.
On the coast, barrier islands and spits protect much of the shoreline and separate the bay systems from the Gulf. Only the Pensacola Bay system had a permanent opening to the Gulf. The other three bay systems alternated between being closed and open to the Gulf, and, consequently, had frequent fluctuations in salinity, relative water level, and marine life. The eastern coastline is not protected by any barriers, and tall dunes up to 75 feet high line the coast with a series of freshwater lakes behind them. These lakes are isolated, rich wetlands within the desert-like environment of dunes and scrub.
The bays have large expanses of sea grass beds and marshes in which large populations of fish and shellfish thrive. These protective environments are estuaries for a continual series of migrating fish that come in from the Gulf to spawn. In addition, there are large permanent populations of fish such as mullet and catfish. The specific species composing the marine communities varies within the bay systems primarily with salinity and depth.
The forests in this area are dominated by Southern Pine which is a fire climax community. However, in areas protected from fire, the vegetation reaches its successional climax of an oak-hickory-magnolia forest. Along the bay and sound coasts, hammocks of these mature forests occur frequently on slightly higher and well drained areas which are protected from fire by adjacent wetlands, and they have been the primary locations selected for settlement by human populations for the last 3,000 years.
SITE DISTRIBUTION ANALYSIS
Based on a review of the literature, 99 Santa Rosa-Swift Creek sites have been recorded in the northwest Florida study area. Like elsewhere, northwest Florida has not been systematically surveyed and there are gaps in survey coverage and intensity.
The first concern in any site distribution analysis is the credibility of cultural affiliation. In this area, we benefit from the fact that the regional chronology was developed by Willey in 1949 and it has stood the test of time. His Smithsonian publication, which defined the periods and diagnostic ceramics types, was widely distributed and often reprinted. Consequently, the diagnostic ceramic types for Santa Rosa-Swift Creek, shown in the slide, have been well-known by regional archaeologists for over 40 years. In addition, Willey (1949) and Sears (1962) integrated the extensive work of C.B. Moore in northwest Florida at the turn of the 20th century into the cultural chronology. Also, many of these 99 Santa Rosa-Swift Creek sites have been revisited during the last 10 years in CRM projects. In short, these factors plus my own partial cross-checking indicate that the affiliation of these 99 sites to the Santa Rosa-Swift Creek period is reliable.
Several patterns or trends in the location of Santa Rosa-Swift Creek sites have been identified. The first pattern is the concentration of sites on the costal strip. 87% of the sites are on or near the coast and only 12 small sites are in the interior. This concentration of sites on the coastal strip has been known for decades, however, there was a suspicion that this was because most archaeological surveys had been done on the coast. However, in the last 10 years there have been several surveys in the interior including the 500,000-acre Eglin AFB (Thomas and Cambell 1992) north of Choctawhatchee Bay and two Escambia River drainage surveys (Bense 1983; Jenkins and Mann 1985). Surprisingly, these interior surveys have found very few Early or Middle Woodland sites. These results support the theory the Indian population during these periods was concentrated on the coastal strip, and the interior was essentially vacant and used only for special-purpose, short-term activities.
The second pattern observed in Santa Rosa-Swift Creek sites is in their configuration. There are three types of midden configurations: ring middens, linear middens, and small midden dumps. Ring middens are large (about 100m in diameter), with well-formed rings a meter or more high, and clean central plazas. Four Santa Rosa-Swift Creek ring middens have been identified in northwest Florida. Two are circular, such as the Old Homestead site on the north shore of Choctawhatchee Bay, and two are U-shaped, such as the 3rd Gulf Breeze site on Santa Rosa Sound near Pensacola. Investigations have been conducted at both of the U-shaped ring middens. Testing at the Gulf Breeze ring midden revealed that the ring was well formed, about a meter high, 95m in diameter, and it was definitely built during the Santa Rosa-Swift Creek period (Willey 1949; Houston and Stoutamire 1982; Doran and Piateck 1985). The plaza inside the arms of the horseshoe is sterile. Interestingly, the shell midden extends outside the shell ring in the flats and is 50-75cm thick with abundant cultural material. The midden eventually thins out about between 50 and 80m from the ring. The surrounding midden contained ceramics and faunal material and it was high in phosphate.
The Horseshoe Bayou site on the south shore of Choctawhatchee Bay has been extensively excavated by Thomas and Cambell (1990, 1992) and Wright, Jr. The shell midden from the Santa Rosa-Swift Creek period was confined to the ring and a few elevated spots inside the ring and the plaza is characterized as "swept clean." This shell midden is described as being composed of discrete lenses of shells with over 330 features including trash pits, shell concentrations, fire hearths, and almost 200 postmolds. It is stated that the overwhelming majority of the features were affiliated with the Santa Rosa-Swift Creek culture.
The third pattern observed in Santa Rosa-Swift Creek sites is site class. Phelps (1969) and Penton (1974) defined three site classes for Santa Rosa-Swift Creek period two decades ago: multi-mound centers, middens with mounds, and middens without mounds. There are no multi-mound centers in northwest Florida, and the nearest such center is the Mandeville site on the Chattahoochee River near Dothan, Alabama, about 250 miles to the east. However, there are four class B sites in northwest Florida, middens with associated mounds, and scores of class C sites, middens without mounds. On closer examination, though, there appears to be clusters of Class B and C site types. There are four Class B clusters (middens with a mound), identified in the slide by dashed circles, and four Class C cluster (middens without mounds), identified by dashed squares in the slide. In both types of clusters, there is usually one large midden while the others are much smaller. In two Class B village-mound clusters, Strange Bayou and Gulf Breeze, the largest sites are ring middens. Similarly, two of the Class C clusters, villages without mounds, have a ring midden as the largest site. Curiously, there are two mounds in the St. Andrew Bay system, Alligator Bayou and Anderson's Bayou, that do not have any recorded middens nearby.
In reflecting on the reality of the eight Santa Rosa-Swift Creek clusters, one naturally questions whether they are simply a result of where archaeological surveys have been conducted or whether they truly reflect the Santa Rosa-Swift Creek settlement pattern. In their survey of Eglin AFB, Thomas and Cambell specifically addressed this question (Thomas and Cambell 1992). After documenting the Basin Bayou village-mound cluster on Choctawhatchee Bay, the adjacent stream drainages both on the east and west were intensively surveyed and only one small Santa Rosa-Swift Creek site was found. Therefore, at least in one case, the clustering of middens around a mound is real and not a product of research. I suspect that with the long history of interest in coastal sites in this area, the numerous surveys, and the easily identifiable diagnostic ceramic types, that most of the clusters of Santa Rosa-Swift Creek sites are real.
There are six Santa Rosa-Swift Creek burial mounds in northwest Florida. Available information from four mounds indicates that both oval and circular shaped mounds were constructed and they were between 50 and 75 feet wide at the base, and 2.5 to 6 feet high. Most mounds appear to have been constructed in a single event. The Manly Mound in the Gulf Breeze cluster, has a ramp on the western side which leads to the summit. Two mounds had a basal layer of oyster shells, and one of these (Baker's Landing) had nine burials under the oyster shell layer. Cremations were likely present in one mound Moore excavated (Anderson's Bayou) as well as multiple skull burials. Artifacts recovered from the mounds were primarily pottery. One mound in the St. Andrew Bay system, Alligator Bayou, had 66 vessels in an east side cache.
The affiliation of four burial mounds with the Santa Rosa-Swift Creek culture is by artifacts excavated by Moore at the turn of the 20th century and re-examined by Sears in the 1950s. The presence of Santa Rosa and Early Swift Creek ceramic vessels (Sears 1962) is the primary basis for their cultural assignment to the Santa Rosa-Swift Creek period. The cultural affiliation of these four mounds is the strongest. One mound, the Manly Mound near Pensacola, is affiliated on the basis of a surface collection and close proximity to two single component Santa Rosa-Swift Creek sites. The sixth mound(s?), at Wynnhaven, has been destroyed for over 100 years and it is tentatively assigned to this period because is on its close proximity to a very large single component site. Four of the mounds have been destroyed, one has been partially destroyed, and one is virtually undisturbed.
Eight radiocarbon dates have been obtained from four Santa Rosa-Swift Creek sites. Four dates from two sites have been obtained in both the Choctawhatchee (Thomas and Cambell 1990, 1992) and Pensacola Bay systems (Phillips 1992) shown in this slide. These dates indicate that while this cultural period lasted about three centuries in both areas, the dates are 200 years earlier in Choctawhatchee (A.D. 150 - 450) than in Pensacola (A.D. 350 - 650). The same situation as has been documented for the preceding late Deptford culture in these bay systems. The late Deptford Okaloosa Phase in Choctawhatchee Bay has been dated at three sites between 50 B.C. and A.D.150 (Thomas and Cambell 1984), while in Pensacola the culturally similar Hawkshaw Phase has been dated at one site between 0 and A.D. 260 (Bense 1985).
From these first few dates, it appears that the Deptford and Santa Rosa-Swift Creek phases in the Pensacola Bay system were increasingly late in making and using the South Appalachian check and complicated stamping styles of ceramic decoration. This is can be explained by diffusion lag, considering these styles originated far to the east and the Pensacola area was the farthest west that these pottery styles were ever made and used on a regular basis. Also, remember that we have dated only one Deptford site in the Pensacola Bay system. The Santa Rosa ceramic series, part of the Gulf pottery tradition which diffused from the west along the coast, is part of both the Deptford and Santa Rosa-Swift Creek assemblages in northwest Florida. Santa Rosa series pottery types were present in both Pensacola and Choctawhatchee Bay by about 50 B.C. and apparently were used in Pensacola until about A.D. 650.
There are only five Santa Rosa-Swift Creek midden sites in northwest Florida for which artifact counts are available. Four of these sites are in the Pensacola Bay system and one is on Santa Rosa Sound. We have recently tested one site in Pensacola which will be reported by John Phillips in more detail in the following paper. When the data is scrutinized from the five sites for which artifact counts are published, the following material culture has been documented for Santa Rosa-Swift Creek coastal sites.
From the two sites which have produced more than 550 sherds, over 75% are plain and the Santa Rosa and Swift Creek series together make up less than 15%. The temper of most pottery is usually micaceous sand, but often the Santa Rosa series sherds are thin and very well made with a clay temper Baytown-like paste. Rims of plain or complicated stamped vessels are usually notched or crenelated in a pie-crust fashion or they are wavy or undulating. Swift Creek Series and plain containers are usually large open bowls which appear to be utilitarian cooking vessels. Containers decorated with the Santa Rosa Series or which are plain and clay-tempered are usually small bowls or beakers which do not show signs of cooking (soot). These same vessel-decoration-paste correlations were documented in the Pensacola Bay system for the preceding Late Deptford period at the Hawkshaw site which was extensively excavated (Bense 1985).
Other ceramic artifacts in Santa Rosa-Swift Creek sites in northwest Florida include solid clay female figurines and a ceramic paddle stamp with a complicated pattern on one side and a check pattern on the reverse side (Phelps 1969).
Few details are known about the rest of the material culture of Santa Rosa-Swift Creek in northwest Florida. The chipped stone point style associated with the ceramics is small with a triangular blade and a stem made by corner or side notching. Phelps refers to these points as Swift Creek points and they are similar to the Columbia type (Bullen 1974). The points are usually made out of Tallahatta Quartzite which occurs just north of the state line in south central Alabama. Other documented lithics include pieces of the local ferruginous sandstone and small split quartzite pebbles.
Bone artifacts include bipointed points, some with transverse lines, which likely were used as fishing gouges or parts of composite fishing harpoons or spears. The few other bone artifacts include drilled animal teeth and vertebrae.
The materials excavated by Moore in four Santa Rosa-Swift Creek burial mounds were part of the basis for Sears' definition of the Green Point burial complex (Sears 1962). The burial mounds in northwest Florida contained only ceramic vessels including Early Swift Creek Complicated Stamped, Basin Bayou Incised, and Alligator Bayou Stamped. In fact, of all the traits noted for the Yent or Green Point burial complexes defined by Sears, only pottery has been recovered from mounds in northwest Florida. Thus far, no exotic items such as copper panpipes or copper covered ear spools, such as illustrated in Sear's definitive work in 1962, shown in this slide, have been found in northwest Florida.
From the preliminary summaries of shell midden analyses (Thomas and Cambell 1990, 1992; Phelps 1969), the bulk of the shell middens is made up of either oyster (Crassostrea virginica) or marsh clam (Rangia cuneata) which differ in their salt tolerance. In areas with low salinity, marsh clams dominate the shell middens and in areas which are more saline, oysters dominate. Other shellfish regularly exploited were the quahog (Mercenaria), lightning whelk (Busycon), coquina, scallop (Pecten), and conchs.
While shellfish remains dominate the bulk of the shell middens, it was fish that made up the bulk of the diet. At the Horseshoe Bayou ring midden on Choctawhatchee Bay, 26 species of fish were identified including blue runner, Crevalle jacks, sheephead, catfish, Atlantic croaker, flounder, red and black drum, speckled trout, and sea bass. Ten mammal species were identified which were dominated by deer, eight reptile species, seven bird species, and one crustacean species were also identified. These faunal remains indicated that this site was occupied for most, if not all, of the year. The presence of sea bass and coquina indicated that the inhabitants ventured to at least the shore of the Gulf of Mexico for resources, but most marine food was procured from the bays.
Unfortunately, no botanical studies of have been reported for any of the 99 Santa Rosa-Swift Creek sites.
SUMMARY AND CONCLUSIONS
In sum, the best-known material trait of the Santa Rosa-Swift Creek period is their pottery. The Santa Rosa and Swift Creek pottery series are diagnostic of the period and we have known that since 1949. Radiocarbon dates place these pottery series in northwest Florida between A.D. 150 and 650. While the seventh century A.D. may seen too late for either of these pottery series, as it did for Phelps (1969) who rejected a A.D. 600 date from the Gulf Breeze ring midden, it should be remembered that this was a fringe area for both the South Appalachian and Gulf Tradition pottery traditions at that time. Also, only four sites have been dated providing only eight dates for this 150 mile area. Another important recent contribution has been the documentation for the long-suspected coastal preference for the Santa Rosa-Swift Creek population. The interior apparently was vacant and used only on a limited basis.
Coastal site types are large ring middens, long linear middens, or small midden dumps. Sites types include Class B (middens with a mound) and type C (middens without a mound) which appear to be organized into clusters. Each cluster had a large site which often was a ring midden. While subsistence remains are sketchy, it appears that fish were the mainstay of their diet. These fisher-folk used nets, traps, and probably spears to procure large quantities of fish. The burial mounds from this society indicate participation in the Hopewell interaction sphere, but there is a curious absence of the more exotic items found just to the east of the Appalachicola. Ceramics were the only mound and grave goods.
Much remains to be learned in our research on the Santa Rosa-Swift Creek culture and our research has just begun.
Bense, Judith A.
1985 Hawkshaw: Prehistory and History in an Urban Florida. Reports of Investigations No. 7, Office of Cultural and Archaeological Research, University of West Florida, Pensacola.
Penton, Daniel T.
1970 Excavations in the Early Swift Creek Component of Bird Hammock (8Wa30). Unpublished M.A. Thesis, Department of Anthropology, Florida State University, Tallahassee.
1974 The early Swift Creek Phase In Northern Florida: Internal Expressions and External Connections. Paper presented at the Annual Meeting of the Society for American Archaeology. Washington.
Phelps, David S.
1969 Swift Creek and Santa Rosa in Northwest Florida. Institute of Archaeology and Anthropology, University of South Carolina Notebook 1, pp. 14-24.
Thomas, Prentice M., Jr. and L. Janice Cambell
1990a The Santa Rosa/Swift Creek Culture on the Northwest Florida Gulf Coast: The Horseshoe Bayou Phase. Paper presented at the Southeastern Archaeological Conference, Mobile.
1992 Eglin Air Force Base Historic Preservation Plan: Technical Synthesis of Cultural Resources Investigations at Eglin Santa Rosa, Okaloosa, and Walton Counties, Florida. New World Research, Inc. Report of Investigations No. 192.
Willey, Gordon R.
1949 The Archaeology of the Florida Gulf Coast. Smithsonian Press. Washington, D.C.