The prehistory of Puerto Rico consists, most broadly, of two successive chrono-stratigraphic periods, termed ArchaicFootnote 1 and Ceramic. A traditional cultural history (e.g., Rouse Reference Rouse1992) cites fundamental differences between these periods in material culture (aceramic/ceramic), subsistence economy (food foraging/food producing), and social structure (egalitarian/ranked), among other features. Although much about the earliest inhabitants of the island—point(s) of origin, timing of arrival, and so on—remains unclear (Napolitano et al. Reference Napolitano, DiNapoli, Stone, Levin, Jew, Lane, O'Connor and Fitzpatrick2019), recent analyses have assigned greater complexity to many aspects of their lifeways than previously assumed (Pagán Jiménez et al. Reference Pagán Jiménez, López, Baik and Narganes Storde2005; Rodríguez Ramos Reference Rodríguez Ramos2007; Ulloa Hung Reference Ulloa Hung, Antonio Curet, Dawdy and Corzo2005). However, many gaps still exist in our knowledge of these early peoples; for instance, the dynamics and decisions governing their foraging behaviors are poorly understood. Here, we analyze the relationship of one aspect of these foraging behaviors—choice(s) made about which marine habitats to exploit for the purposes of shellfish collection—in light of variations in chronology, site type, and paleoenvironment. In doing so, we acknowledge that even shellfish collection practices, let alone the whole suite of foraging patterns, are multifactorial and complex phenomena.
For reasons that are not well understood, southwestern Puerto Rico encompasses an abundance of archaeological sites of the island's earliest inhabitants. A review of master site files found that more than 60% of the sites in the region are from this early period, a notable departure from broader insular patterns. Two such sites in the southwestern municipality of Cabo Rojo are considered here (Figure 1). CRNWR_P13 (Pestle et al. Reference Pestle, Laguer-Díaz and Rámos2018, Staudigel et al. Reference Staudigel, Peter Swart, Laguer-Díaz and Pestle2019) is located 2 m above mean sea level (m amsl), on the shore of the Laguna Fraternidad, a tidal salt flat, and 0.5 km from the sea. Excavations there revealed a 2,500 m2 shell midden consisting of abundant marine shells and few other artifact types (see Supplemental Table 1). Radiocarbon assays indicate an overall use life stretching from 810 cal BC to cal AD 280 (Supplemental Table 2). On the basis of site location, the abundance of marine bivalves, and the paucity of other artifacts, we hypothesize that CRNWR_P13 was a recurrently used marine resource exploitation site. The Ortiz site (ICP-CAT-CR-88-03-02), 7 km north of CRNWR_P13, includes a greater than 500 m2 midden consisting of marine invertebrate remains, lithic artifacts representing all stages of tool manufacture, and five burials. Seven radiocarbon dates indicate long-term or recurrent use or habitation from 2330 cal BC to cal AD 310 (Supplemental Table 2). Based on location (on a 25 m amsl prominence 650 m from a sheltered bay), the variety of artifacts recovered, and the presence of burials, we hypothesize that Ortiz was a habitation site.
Maps showing location of Puerto Rico (top), with study area noted; CRNWR_P13 and Ortiz sites and local marine habitats (bottom). (Color online)
Methods and Materials
Marine shells, collected in situ and using 6.4 mm screens, were sorted by excavation context, identified taxonomically, and weighed. We calculated edible meat weight by taxon using the following allometric formulas, where w = edible weight and s = shell weight: for bivalves (w = 100.02 × (s 0.68)); for thick-shelled gastropods, such as Aliger and Strombus (w = 10−0.68 × (s 0.88)), and for thin-shelled gastropods (w = 10−0.16 × (s 0.92); Reitz et al. Reference Reitz, Quitmyer, Stephen Hale, Scudder and Wing1987). Given pronounced disparities in the shell to meat ratio among taxa, an allometric approach provides a superior means for quantifying dietary contribution. Each taxon was assigned to one of three habitats: hard substrate, soft substrate, and mangrove, following Rosenberg and colleagues (Reference Rosenberg, Moretzsohn, García, Felder and Camp2009) and Turgeon and coworkers (Reference Turgeon, Lyons, Mikkelsen, Rosenberg, Moretzsohn, Felder and Camp2009). The combination of habitat determinations and edible weights enabled us to determine the proportion of meat harvested by habitat type for each excavation context.
Using ArcGIS 10.3 (ESRI 2014) and following Kendall (Reference Kendall2002), marine patches within 5 km radii of both sites (Figure 1) were classified as hard substrate, soft substrate, or mangrove. Area and distance-to-site for all patches near each site were combined to generate a single representation of habitat distribution (Table 1).
Habitat Areas, Patch Distance-to-Site, and Combined Habitat Distribution for CRNWR_P13 and Ortiz.
Results
Shell and edible weight by site and taxon are provided in Supplemental Table 3. Given the similar (Fisher's exact test, p = 0.24) distribution of habitats near each site (Figure 2a), we predicted that, were shellfish collection practices consistent and the exploitation of resources from each habitat type proportional to their local availability, the shellfish assemblages at the two sites also should resemble one another. Surprisingly, the observed proportions of marine mollusk edible meat (Figure 2b) differ significantly between the two sites (p < 0.01); taxa from soft-bottomed habitats make up more of the assemblage at CRNWR_P13 (75%) than at the Ortiz site (55%), whereas mangrove and hard-bottomed taxa are more abundant at Ortiz (making up 19% and 26% of edible meat, respectively, vs. 3% and 22% at CRNWR_P13).
(a) Expected habitat availability and (b) observed proportions of edible meat at CRNWR_P13 and Ortiz. (Color online)
At Ortiz (Figure 3a), there is a statistically significant difference (p < 0.01) between habitat distribution and the origin of mollusk edible meat. Soft-bottomed taxa are underrepresented in the Ortiz assemblage (55% observed vs. 88% expected), and hard-bottomed and mangrove taxa are overrepresented (26% and 19% observed, respectively, vs. 3% and 9%). CRNWR_P13 also differs significantly (p < 0.01) from baseline predictions (Figure 3b); taxa from hard-bottomed habitats occur more frequently than expected (22% observed vs. 4% expected), whereas those from soft-bottomed habitats appear less often (75% observed vs. 93% expected).
Comparison of expected habitat availability and origin of edible meat at (a) Ortiz and (b) CRNWR_P13. (Color online)
This similarity diminishes, however, when the sites’ assemblages are disaggregated by time (with comparisons made using X 2 tests). At the Ortiz site (Figure 4), when comparisons are made level by level, there is no significant temporal trend (R = 0.52, p = 0.37) in the X 2 values. Conversely, at CRNWR_P13 (Figure 5), when one considers the malacological remains from Unit 1 (which accounted for more than 52% of the site's shells and for which we possess dates from the top and bottom of the midden), a distinct chronological pattern emerges. Indeed, the degree of difference between observed and expected habitat use (measured by X 2 value) decreases significantly (R = 0.91, p = 0.03) between Level 4 (730–430 cal BC) and Level 1 (40 cal BC–cal AD 180). Thus, over 400 to 900 years, the collection practices evidenced at CRNWR_P13, in terms of habitat choice, changed, ultimately coming to resemble the present-day habitat distribution.
(a) Level-by-level comparison of expected habitat availability and origin of edible meat and (b) resulting chi-squared trend for Ortiz. (Color online)
(a) Level-by-level comparison of expected habitat availability and origin of edible meat and (b) resulting chi-squared trend for CRNWR_P13. (Color online)
Discussion
Despite having similar proximate habitat distributions, the shell assemblages recovered from these sites differed significantly. Although this variation could result from synchronic differences in shellfish collection practices, given that the earliest dates for the Ortiz site predate those from CRNWR_P13, it is also possible that there were diachronic changes in habitat composition, foraging practices, or both. While both sites are “early,” this finding shows the folly of assuming uniformity in cultural practices within a given archaeological “age.”
Assuming the Ortiz site was habitational, we expected that foods gathered from a wide geographic range would have been brought back to the site; thus, the assemblage might not conform strictly to local habitat availability. This contention found support in the aggregate and on a level-by-level basis: there were significant differences between habitat exploitation choices and nearby habitat distribution. The shell remains at Ortiz would thus appear to represent a broader array of habitats than those found closest to the site.
Conversely, because CRNWR_P13 likely was a strategically positioned exploitation site, its assemblage should closely resemble nearby habitats. Although this was not the case in the aggregate, a level-by-level comparison found that the makeup of the assemblage better resembled modern local habitats in every successive level until, in the uppermost level, there was no significant difference between the observed habitat exploitation and the nearby habitat distribution. This finding compels examination of paleoenvironmental conditions to ascertain how ancient these “modern” conditions were and to explain the earlier levels’ deviation from the expected.
For paleoenvironmental insights, we turn to two local studies. At the Flamenco Lagoon, 35 km east of the sites presented here, Cohen and colleagues (Reference Cohen, Lara, Cuevas, Oliveras and Sternberg2016) found a succession of conditions consistent with studies of the larger Caribbean (e.g., Toscano and Macintyre Reference Toscano and Macintyre2003). Specifically, they argue that modern relative sea level (RSL) was obtained by around 3000 cal BP (approximately 1000 cal BC)—preceding the cessation of activity at the sites under study here. More locally, however, Sherman and Fernández (Reference Sherman and Fernández2014) analyzed cores collected from a hypersaline pond 4 km from CRNWR_P13 and others from the Laguna Fraternidad adjacent to the site. Based on sedimentological analysis of these cores, they found that the present-day salt flats were open to the ocean until as recently as 1,400–1,700 years ago. As such, the present-day habitats surrounding CRNWR_P13 are probably quite different from those that prevailed during the site and region's earliest occupation (at 2300 cal BC or earlier), and realization of the modern habitat configuration would thus appear to have occurred much later.
Indeed, when the earliest date for the establishment of the modern marine habitat configuration obtained by Sherman and Fernández (Reference Sherman and Fernández2014; CR16S5/OS-114198; 2030 ± 40 BP; marine shell; δ13C = 2.6‰) is compared with the date obtained from Level 1 at CRNWR_P13 (Supplemental Table 2), the two dated events’ calibrated 2-sigma ranges overlap by nearly a century (Beta-492526 cal BC 60-cal AD 280, CR16S5/OS-114198 cal AD 200–550), such that the collection practices evidenced in Level 1 and the attainment of the present habitat configuration were contemporary. Thus, the greater concordance seen in the most recent levels at CRNWR_P13 may reflect attainment of the current habitat distribution at or near the time of collection of those resources, whereas the earlier levels represent shellfish collection in a very different local environment, with a different distribution of nearby marine habitats.
Conclusion
In this work, we investigated differences in malacological assemblages at two sites associated with the earliest habitation of southwestern Puerto Rico. We found that, even at a local scale, foraging practices, evidenced here by shellfish collection from different habitats, can vary significantly. Indeed, these results show the diversity of foraging practices internal to an archaeological “age” in just one constrained local area. The observed differences between these sites may be attributable to differences in time, site function, or changes in local coastal configuration. As such, this work establishes the need for the generation of higher-resolution paleoclimate data for the region under study, as well as further study and dating of local sites and assemblages of this earliest period of human activity. Nonetheless, the present work represents a meaningful step forward in terms of characterizing the diverse foraging practices of the early inhabitants of southwestern Puerto Rico.
Acknowledgments
Portions of this work were supported by NPS-HPF funding administered by the Oficina Estatal de Conservación Historica (OECH), contract 2017-155056, under ARPA permits CRJNWR-022617, 121517, and 121918. We wish to acknowledge our partners at the OECH and the U.S. Fish and Wildlife Service, as well as the tireless efforts of the field school students who participated in survey and excavation at CRNWR_P13.
Data Availability Statement
All data for this article are presented in the text or in Supplemental Materials.
Supplemental Materials
To view supplemental material for this article, please visit https://doi.org/10.1017/laq.2021.29.
Supplemental Table 1. Artifacts/Ecofacts Recovered from Excavation of 5 m2 at CRNWR_P13 by Material Class.
Supplemental Table 2. Radiocarbon Dates from CRNWR_P13 and Ortiz.
Supplemental Table 3. Shell and Edible Meat Weight by Invertebrate Taxa at the Ortiz and CRNWR_P13 Sites.
