The last few decades have seen major changes in the archaeological evidence for maize (Zea mays ssp. mays) and squash (Cucurbita spp.) histories in interior portions of temperate northeastern North America. Regional archaeologists in the past assumed that the two crop species were adopted by hunter-gatherer populations (e.g., Ritchie Reference Ritchie1969) or introduced through the migration of agriculturists from elsewhere (e.g., Snow Reference Snow1995) by around AD 1000. This began to change in the 1990s with reports of directly dated macrobotanical remains of gourds dating to ∼3000 BC in Maine and Pennsylvania and edible squash to ∼450 BC in Pennsylvania (Hart and Asch Sidell Reference Hart and Sidell1997; Petersen and Asch Sidell Reference Petersen and Sidell1996). These coincided with the report of directly dated maize macrobotanical remains as early as ∼AD 500 in southern Ontario (Crawford et al. Reference Crawford, Smith and Bowyer1997). Although there have been recent direct dates on maize macrobotanical remains as early as ∼AD 600 and ∼AD 700 in the riverine interior (Comstock and Cook Reference Comstock and Cook2024; Simon Reference Simon, Raviele and Lovis2014), the Maine, Ontario, and Pennsylvania dates have remained the earliest directly dated maize and squash macrobotanical remains in the interior Northeast.
In southern New England comprising Connecticut, Rhode Island, and islands in the Block Island Sound, AMS-dated maize macrobotanical remains mostly postdate the thirteenth century AD. Thirty direct dates on maize are reported from Connecticut in the Canadian Archaeological Radiocarbon Database (CARD; Martindale et al. Reference Martindale, Morlan, Betts, Blake, Gajewski, Chaput, Mason and Vermeersch2016), the earliest of which is 690 ± 60 BP (95.4% cal AD 1228–1399) from the Housatonic River site in western Connecticut. Most dates on maize in Connecticut are from two sites in the Connecticut River valley: Burnham-Shepard (n = 15) and Morgan (n =11). Bayesian modeling of these and other dates on annual macrobotanical remains indicates the sites date to the sixteenth and to the fourteenth and fifteenth centuries AD, respectively (Hart et al. Reference Hart, Dotzel and Winchell-Sweeney2026). Only nine direct dates on maize are reported from Rhode Island in CARD. The earliest is 965 ± 22 BP (cal 2σ AD 1025–1158) at RI-110 in southern Rhode Island (Waller et al. Reference Waller, Leveillee and Forrest2019). Other dates on maize from this and two other sites are three to nine centuries later. Although, there are ethnohistoric reports that squash and gourd were cultivated in southern New England by Indigenous farmers at contact, no macrobotanical remains have been reported from precontact archaeological sites (Lavin Reference Lavin2013).
In contrast with the northeastern macrobotanical record, beginning in the early 2000s, maize phytoliths were recovered from charred cooking residues adhering to interior pottery sherd surfaces (hereafter, residues) directly dated to as early as ∼300 BC in New York and the St. Lawrence River valley in Québec and squash phytoliths as early as ∼1000 BC in New York (Hart et al. Reference Hart, Brumbach and Lusteck2007; St-Pierre and Thompson Reference St-Pierre and Thompson2015). Kováčik and coworkers (Reference Kováčik, Cummings and Milligan2016) recovered Zea-type phytoliths from three sherds in thirteenth-century AD contexts at RI-110, Salt Pond in southern Rhode Island. Other analyses of phytoliths in eastern North America have been limited to Michigan (Albert et al. Reference Albert, Kooiman, Clark and Lovis2018) and the northern (Boyd and Surette Reference Boyd and Surette2010) and central Plains (Adair et al. Reference Adair, Duncan, Young, Bozarth and Lusteck2022). Each of these studies was based on phytoliths extracted from residues, expanding regional understandings of maize and squash histories. Here we report on the results of the first analysis of phytoliths extracted from directly dated residues from southern New England.
Methods and Materials
Phytoliths
Residues were sampled from sites, with collections curated at the University of Connecticut in Storrs, the Institute for American Indian Studies in Washington, Connecticut, and the Public Archaeology Laboratory in Pawtucket, Rhode Island (Figure 1). A maximum of 20 sherds with residues were selected for sampling from each site as an arbitrary means of maintaining the number of samples at a manageable quantity for analysis. The chronological and spatial distributions of sites included in the samples were determined by the available collections and the presence of residues (Table 1).
Locations of sites discussed in the article. (Color online)
Figure 1 Long description
The map displays archaeological sites in the northeastern USA, specifically around Long Island Sound and the Atlantic Ocean. Sites are marked with symbols indicating their presumed era of occupation. Green triangles represent 550 BC to AD 800, pink diamonds indicate AD 800 to 1550 and blue circles denote 500 BC to AD 1500. Key sites include Eagle Hill, Grannis Island, Butternut Knoll, Burnham-Shepard, Morgan, Kasheta, Selden Island, Hamburg Cove, Cooper, RI 110, RI 1428 and RI 1407. A detail map inset shows the broader regional location. A scale bar and compass rose are present for orientation.
Summary Data for All Analyzed Residues.
Table 1 Long description
The table summarizes analyzed residue samples by archaeological site, presumed occupation period, environmental zone, number of samples, total residue processed in milligrams, and counts of maize and squash phytoliths. Overall totals are 126 samples, 1,292.1 milligrams processed, 21 maize phytoliths, and 44 squash phytoliths, indicating squash remains are more common than maize in these residues. The largest sample counts are Burnham-Shepard with 20 samples and RI 110 with 19, while the greatest residue mass processed is Burnham-Shepard at 291.2 milligrams. Maize phytoliths are most abundant at RI 110 with 8, followed by Morgan with 4 and Kasheta with 3; several sites report none. Squash phytoliths peak at Morgan with 9 and Butternut Knoll with 8, with Selden Island close behind at 7. Coastal or estuarine sites include both zero-count locations and sites with notable squash presence, such as Selden Island, while inland or riverine sites include multiple locations with higher squash counts. One environmental zone entry for RI 1428 appears misspelled, and phytolith counts are small, so differences between sites may reflect limited recovery as well as real variation.
Phytoliths extraction was done in the University of Connecticut’s phytolith and starch laboratory following a protocol developed at the University of Minnesota Laboratory of Archaeometry (Raviele Reference Raviele2010). Before sampling, residues were treated with compressed air to remove any accumulation of dust or debris from storage. Between 0.5 and 40 mg of residue were removed from each sherd, depending on the quantity of material present (Table 1). After initial experimentation, 20 mg of residue was found to be the ideal quantity: more than that often yielded too much material to fit on a single slide. Each sample was placed in a test tube with 20 ml of Schulze solution (a 3:1 ratio of nitric acid to dissolved potassium chlorate), which was then placed in a 95°C water bath for ∼24 hours until the carbonized organic matter dissolved. Each sample was then rinsed three times with deionized water in a centrifuge. The remaining material was then dried and mounted on a slide with Canada balsam.
We examined each slide under a combination of 200× and 400× magnification using a Zeiss Axio Lab.A1 or Leica DM2700 M microscope equipped with a Digital Insight DS-Fi3 camera. Each slide was scanned for diagnostic phytoliths of economically important taxa, including maize, common bean, squash, and sedge. To establish a representative sample of all phytoliths as described in Pearsall (Reference Pearsall2015), we counted up to 200 of all other identifiable phytoliths, if present. Phytoliths were classified by shape according to established standards (Bozarth Reference Bozarth, Rapp and Mulholland1992; Madella et al. Reference Madella, Alexandre and Ball2005; Pearsall Reference Pearsall2015; Piperno Reference Piperno2006; Twiss et al. Reference Twiss, Suess and Smith1969). Diagnostic squash phytoliths—scalloped forms comprising of spheroidal or hemispheroidal forms with deep concavities (Bozarth Reference Bozarth1987)—were identified using modern comparative materials. Maize phytoliths were identified by wavy-top and ruffle-top rondels, using modern comparative materials, Bozarth’s (Reference Bozarth1998) and Pearsall and associates’ (Reference Pearsall, Chandler-Ezell and Chandler-Ezell2003) descriptions of wavy-top rondels, and Piperno’s (Reference Piperno2006) description of ruffle top rondels. The latter also occur in the fruit cases of teosintes, but because teosinte is not expected to be found in northeastern North America during the era of this study, they were used to identify maize.
Radiocarbon Dating
AMS dates were obtained on samples of cooking residue containing diagnostic phytoliths at the University of California-Irvine W. M. Keck Carbon Cycle Accelerator Mass Spectrometer (KCCAMS) Facility. Details on pretreatment, graphitization, isotope ratio mass spectrometry (IRMS), and AMS dating are available on the KCCAMS website (https://www.ess.uci.edu/w-m-keck-carbon-cycle-accelerator-mass-spectrometer-kccams-facility).
Calibrations were performed using OxCal v. 4.4.4 (Bronk Ramsey Reference Bronk Ramsey2009). All sites have evidence of marine resource exploitation including fish bone, marine bivalve shell, or both. It is possible that carbon from both terrestrial and marine resources contributed to the formation of residues. To account for this, we used estimated marine reservoirs (∆R; see Supplementary Table 1 for details) in radiocarbon date calibrations using the IntCal20 northern Hemisphere terrestrial (Reimer et al. Reference Reimer, Paula, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey and Butzin2020) and Marine20 calibration curves (Heaton et al. Reference Heaton, Köhler, Butzin, Bard, Reimer, Austin and Ramsey2020). OxCal uses a Markov Chain Monte Carlo algorithm to find the best mix of the two curves that fit the data. Because we cannot directly determine the percentages of marine carbon that may have contributed to residue formation—and knowing that terrestrial resources contributed carbon to residue formation based on the phytolith results—the calibrations were done using a range of 0%–50% marine carbon using local ∆R estimates (see Supplementary Table 2). Stipulating a uniform distribution allowed the models to calculate an unbiased estimate of the marine carbon content of the residue within the percentage range constraint (Bronk Ramsey and Lee Reference Bronk Ramsey and Lee2013:722).
Results
Twenty-one residues produced diagnostic maize, squash phytoliths, or both. We only report here the 14 directly dated residues that produced diagnostic phytoliths unless otherwise noted. Funding for dates was limited, and we chose to date the samples that were likely to be the oldest from each site. Descriptions of sites from which phytoliths are reported here are provided in Supplementary Text 1. Results of the phytolith and residue AMS dates and calibrations are presented in Table 2. Sherd provenience data and complete phytolith analysis results for the dated residues are presented in Supplementary Table 3.
Calibrations of AMS Dates on Residues Containing Diagnostic Maize and/or Squash Phytoliths.
Table 2 Long description
The table lists AMS radiocarbon results for residue samples that contain diagnostic maize and/or squash phytoliths, organized by laboratory number and site. For each sample it reports the radiocarbon age with uncertainty, a carbon isotope value when available, the phytolith type present, and calibrated calendar ranges given as two probability intervals (a narrower and a wider interval). Most calibrated ranges fall in the late first to early second millennium AD, commonly spanning roughly AD 900 to 1300, including Selden Island (maize about AD 668 to 945; squash about AD 888 to 1128) and multiple Morgan samples (several squash and maize results centered around about AD 1157 to 1406). Earlier ranges occur at Morgan for a squash sample around AD 438 to 656, and Hamburg Cove shows a broad range around AD 649 to 877. Butternut Knoll and RI 110 maize and squash samples largely cluster around about AD 995 to 1281, while one RI 110 maize sample extends later, with ranges mainly from about AD 1420 into the 1600s. Several entries lack isotope measurements because the samples were too small, and the calibrated ranges reflect modeling assumptions about possible marine carbon and omit very small probability portions.
Notes: Calibrations assume 0-50% marine carbon; distributions <1% are not listed.
a All 14C ages were corrected for δ13C fractionation on prepared graphite using the AMS spectrometer. Listed δ13C measurements were made on aliquots of the original sample CO2 with an IRMS. Samples indicating n/a were too small to provide a CO2 aliquot for IRMS δ13C measurement.
b Highest posterior density.
The earliest squash phytoliths (n = 5), AD 425–705 (90.5% highest posterior density [hpd]), were recovered from Morgan sample UCIAMS-243990. Although this sample was from a thin, undecorated body sherd that would be considered contemporaneous with later-dated occupations (Hart et al. Reference Hart, Dotzel and Winchell-Sweeney2026), even assuming a 90% marine carbon contribution, the 95% hpd is from AD 430 to 987. One residue from Hamburg Cove on the Connecticut River estuary yielded 11 squash phytoliths. Although this residue was not directly dated, a second residue, UCIAMS-244017, which did not produce diagnostic maize or squash phytoliths, produced a date that calibrates to AD 578–991 (95.4% hpd), consistent with McBride’s (Reference McBride1984) estimated age of the site; however, we cannot state with any certainty that the two sherds are contemporaneous. Squash phytoliths from four additional sites indicate the crop’s use in the region up to the contact period.
The earliest maize phytoliths were from sample UCIAMS-243973 on a dentate-stamped body sherd recovered at RI-1428 on Block Island with a 95.4% calibrated date range of AD 671–1021. As reported in Dotzel (Reference Dotzel2021), the sherd was recovered in a feature that had radiocarbon dates on wood charcoal of 2110 ± 210 BP (95.4% cal 757 BC–AD 340), 2380 ± 120 BP (95.4% cal 795–197 BC), and 2420 ± 110 BP (95.4% cal 803–210 BC; Tveskov Reference Tveskov1992). Obviously, the sherd and the charcoal on which these dates were obtained are not contemporaneous. Samples with maize phytoliths from this and other sites extend evidence for maize use from the centuries immediately before AD 1000 forward to the time when it is common in the macrobotanical record in the fourteenth to sixteenth centuries (Table 2).
Discussion and Conclusions
The evidence presented here for maize does not substantially alter, but does extend and broaden, the crop’s documented history in southern New England. Maize was present on Block Island and probably at Selden Island before AD 1000. Maize phytoliths were common in residues that postdate AD 1000.
The new evidence, does, however, substantially alter our understandings of squash history in the region. Squash phytoliths are common in the residues (Table 1), with their earliest occurrence dating to the mid- to first millennium AD. This result demonstrates a long history for this crop in southern New England despite its absence in the macrobotanical record. This study also found that maize and squash crops were common across southern New England, with evidence of maize or squash at all but one sampled site.
A limitation of this study is that, in the samples available for analysis, both the number and quantity of residues were biased toward later sites because of the limited availability of earlier residues. However, the results do indicate the potential of phytolith analysis to expand our understanding of crop histories in southern New England. Future analysis of earlier residues has the potential to lead to extended histories of maize and squash as in nearby New York and Québec.
Acknowledgments
We thank Kevin McBride for help in typological descriptions of the sherds used in the study, Thomas C. Hart for providing a second opinion on phytolith identifications, Susan Winchell-Sweeney for drafting Figure 1, and Christian Gates St-Pierre for editing the French abstract. No permits were required for this research.
Funding Statement
This work was supported by the National Science Foundation under Grant 2006327 to Dotzel.
Data Availability Statement
All data are available in the report and supplementary material.
Competing Interests
The authors declare none.
Supplementary Material
The supplementary material for this report can be found at https://doi.org/10.1017/aaq.2026.10226.
Supplementary Text 1. Site Descriptions.
Supplementary Table 1. Marine Reservoir Calculation Details.
Supplementary Table 2. OxCal Code for Calibrations.
Supplementary Table 3. Phytolith Counts for Residues Discussed in the Report.
Open access
