ACCELERATOR MASS SPECTROMETRY DATING OF MEADOWCROFT ROCKSHELTER MAIZE

ABSTRACT The Meadowcroft Rockshelter in southwestern Pennsylvania is best known for its pre-Clovis occupation. Potentially important for later times is the recovery of maize macrobotanical remains from higher strata dating as early as the 4th century BC based on radiometric radiocarbon (14C) dates on wood charcoal. These remains have been considered to be potentially as old as the earliest microbotanical evidence for maize in Michigan, New York and Québec recovered from directly dated charred cooking residues adhering to pottery. The results of accelerator mass spectrometry (AMS) dating 17 samples from maize specimens from all Meadowcroft strata producing maize, indicate that the specimens originated from historical use of the shelter, most likely after AD 1800. These results further emphasize the need to obtain direct dates on maize macrobotanical remains recovered from early contexts prior to the development and common use of AMS dating.


INTRODUCTION
The histories of the spread of maize (Zea mays ssp. mays) north and south of central Mexico where it evolved from an annual teosinte (Zea mays ssp. parviglumis) 9000-7000 years ago (Matsuoka et al. 2002), its adaptations to wide ranges of climatic and edaphic conditions, the timings of its adoptions by far-flung Native American societies, and the impacts of its adoption, if any, on regional subsistence-settlement systems remain important topics of research for archaeologists, geneticists, and paleoethnobotanists (e.g., Staller et al. 2006;Bonavia 2013;Grobman 2013;Blake 2015;Pearsall 2019). While major strides have been made in the past few decades in building knowledge on each of these topics through a variety of analytical methods and techniques, the crop's histories remain far from settled in many regions. One such region is temperate northeastern North America (hereafter Northeast), one of the last regions where maize was adopted, but where it became the main crop of agricultural systems after AD 1000-1300 (Hart and Lovis 2013). Resolving the timing of the crop's adoption is necessary to anchor maize's histories in this region and has been a long-standing focus of research that is yet to be resolved (e.g., Emerson et al. 2020;Dotzel 2021;Simon et al. 2021;Stewart 2021). Current microbotanical evidence from Michigan (Schultz site; Albert et al. 2018), New York (Vinette site; Hart et al. 2007a), and Québec (Place-Royale site; Gates St-Piere and Thompson 2015) (Figure 1), and potentially southern New England (Dotzel 2021), in the form of phytoliths and starch recovered from accelerator mass spectrometry (AMS)-dated cooking residues adhering to pottery sherd interior surfaces indicates use by at least cal. 300 BC. However, the macrobotanical evidence, which until recently was largely in line with this date for the greater Northeast, is in a state of flux.
Maize macrobotanical remains potentially older than those from Grand Banks have been recovered from various sites in the Northeast but have yet to be directly dated and subjected to IRMS (McConaughy 2008;Hart and Lovis 2013;Stewart 2021). Obtaining direct dates on these remains coupled with IRMS δ 13 C measurements to confirm identifications is needed to help clarify the histories of maize in the region.
Caves and rockshelters, two categories of archaeological site relatively rare in the Northeast, provide excellent conditions for preservation of charred and desiccated maize macrobotanical remains and have provided key evidence for early maize in Mexico, Mesoamerica, and the American Southwest (e.g., Piperno and Flannery 2001;Merrill et al. 2009;da Fonseca et al. 2015;Kennett et al. 2017;Swarts et al. 2017;Torres-Rodríguez 2018). Most prominent of such sites in the Northeast is the stratified Meadowcroft Rockshelter located on Cross Creek, an east-west-flowing tributary of the Ohio River, southwest of Pittsburgh, Pennsylvania near the West Virginia boarder (Figure 1; Adovasio et al. 1978;Adovasio 2010). Excavated primarily in 1973Excavated primarily in -1979 and sporadically thereafter, the site is best known for its pre-Clovis component (e.g., Haynes 2015;Carr 2018;Williams and Madson 2020). However, potentially important for maize history is a series of charred and desiccated cobs and cob fragments recovered from later strata as reported in Adovasio and Johnson (1981). These strata were defined chronologically with radiometric 14 C dates on wood charcoal, and as was common practice at the time, these were used to assign dates to the maize (Table 1). 14 C dates with median calibrated dates of 403 BC and 349 BC from the earliest stratum to yield maize (Stratum IV) are in-line with the earliest dates for maize microbotanical remains in the Northeast. These finds have been cited as potential evidence for early maize in the Northeast, but the need for direct dates to confirm their early age has been noted often (e.g., Crawford et al. 2006;McConaughy 2008;Hart and Lovis 2013); it is now accepted practice to directly AMS date crop remains because there is no necessary chronological relationship between spatially associated wood charcoal and the crop remains (Blake 2006). Here, we report direct AMS dates on a series of 17 samples of maize cobs/ cob fragments recovered from Meadowcroft. The results emphasize the need to directly AMS date macrobotanical remains of maize and other crops recovered and reported prior to the development and common use of AMS dating. Table 1 Maize cob data. Associated radiocarbon dates from Adovasio and Johnson (1981). Rows, grain thickness, cupule width, charring, and notes from Cutler and Blake (1977) as published in Adovasio and Johnson (1981 *Cutler and Blake (1977) indicated this cob fragment was carbonized. However, only two of the samples from this provenience in the collection are carbonized and this fragment is not. The cupule width matches Cutler and Blake's measurement as well as their statement on glumes. Adovasio and Johnson (1981) reported maize macrobotanical remains in the form of charred and desiccated cobs and cob fragments from the upper strata (IV-XI) of the Meadowcroft Rockshelter (Table 1). Identification of the maize was done by Cutler and Blake (1977) at the Missouri Botanical Garden as reported in an unpublished manuscript and summarized by Adovasio and Johnson (1981). The earliest of these was a small, charred cob fragment identified as probably 16-row popcorn, from Stratum IV associated with 14 C dates on wood charcoal of 2355 ± 75 BP and 2290 ± 90 BP. Stratum VI yielded 10-, 12-, and 14-row cob fragments associated with wood charcoal dates of 2155 ± 65 BP and 2075 ± 125 BP. Cob fragments representing 10-and 12-row maize were recovered from Stratum VII with associated wood charcoal dates of 1290 ± 60 BP and 925 ± 65 BP. Botanical remains from Stratum IX included 12-and 14-row maize associated with a date on wood charcoal of 685 ± 80 BP. Cobs and cob fragments from Stratum XI, from 8-, 12-, and 14-rowed maize were reported as dating later than 685 ± 80 BP and earlier than 175 ± 50 BP. In their unpublished report on the maize, Cutler and Blake (1977: 1) related that the maize from Meadowcroft was "surprisingly large and vigorous, the cobs firm and thickened." They indicated that three of the cobs/cob fragments, one from Stratum IX and two from Stratum XI, were possibly modern, post-1800, and recent, respectively (Table 1). This suggested the possibility of some mixing of earlier and later deposits within these strata. They attributed most of the remaining cob fragments in these strata to their prehistoric Midwest 12-Row maize category, with a few ascribed to the prehistoric Eastern 8-Row category.

METHODS AND MATERIALS
The Meadowcroft Rockshelter maize remains are curated at the Senator John Heinz History Center in Pittsburgh along with the rest of the site's collection. All maize remains are identified by catalog number, wrapped in aluminum foil, and stored in capped plastic vials. The specimens were placed on loan to the New York State Museum (NYSM) where photography and sampling were completed. For catalog numbers with fragments from multiple cobs, cupule width measurements were used to correlate them with Cutler and Blake's inventory (Table 1). Images and data for the sampled specimens are presented in Supplement 1.
Small samples of 18 cobs and cob fragments were taken under low magnification with a solvent-cleaned scalpel or razor blade. Samples from fragments of different cobs that had been assigned the same Meadowcroft catalog number were given sample numbers to distinguish them in Tables 1 and 2 (Reimer et al. 2020).

RESULTS
AMS 14 C ages and calibrated dates and δ 13 C values are presented in Table 2. The sample from specimen FS-1811.1 did not yield enough carbon after pretreatment for analysis, and samples FS-269.8 (2) and FS-130.10 (1) were too small to provide enough material for IRMS measurement. Given the results of the remaining samples, providing additional material of these specimens for assay was unwarranted. While there is no doubt based on Cutler and Blake's analysis and their physical appearance that the specimens are maize cobs/cob fragments (Supplement 1), the δ 13 C values ranging from -8.4 to -10.8 confirm their identifications as maize ( Table 2).
All dates are historical and remarkably consistent given that the samples were recovered from four separate strata. There is no record and no visual evidence of the maize remains being treated with consolidants or adhesives. The application of most of the commonly used consolidants and adhesives would result in older, not younger, ages than anticipated (Crann and Grant 2019). Three of the organic consolidants and glues analyzed by Crann and Grant (rabbit skin glue, technical gelatine, and wheat starch) produced modern ages. This suggests a heavy application that would be visible and prevent the smears of carbon that occurred when handling of the cob fragment for sampling. All consolidants and adhesives tested by Cran and Grant (2019:1062) have δ 13 C values more negative than archaeological maize in the Northeast ranging from -36.0 to -15.8‰ (median = -27.5). That the Meadowcroft maize δ 13 C values are well within the range for archaeological maize in the Northeast also suggests the absence of treatment with consolidants or adhesives. Furthermore, if any of the organic materials analyzed by Crann and Grant had been applied to the Meadowcroft maize specimens, they would have been removed by the base step of the standard acid-base-acid treatment applied to the samples at KCCAMS and have no effect on the analytical results.
The AMS dates, disprove an early presence for maize at the Meadowcroft Rockshelter and go beyond Cutler's and Blake's (1977) suggestion that three of the specimens date after AD 1800. The calibrated dates are multimodal; the largest probabilities fall within the nineteenth to early twentieth centuries, with probabilities generally <30% falling in the late seventeenth to early eighteenth centuries.  the multiyear excavations at Meadowcroft Rockshelter to note the presence of bioturbation or other disturbances; however, it is apparent that disturbances in the area which produced the maize remains were missed. It should also be stressed that the same strata which yielded the maize remains evidenced no disturbance elsewhere on the site.

DISCUSSION AND CONCLUSIONS
The timings of the adoption of maize are ongoing research topics throughout the Western Hemisphere. In the northeastern North America, two lines of evidence have been used to determine when maize becomes archaeologically visible: microbotanical remains recovered from directly dated food residues adhering to pottery and directly dated macrobotanical remains. Until recently these two lines of evidence were generally in agreement for the region as a whole with early directly dated microbotanical evidence in the eastern Great Lakes and St. Lawrence River Valley and early directly dated macrobotanical evidence from the riverine interior. The early evidence from the riverine interior was recently discredited, leaving the earliest directly dated macrobotanical evidence from the Great Lakes region in southern Ontario, some 800 years later than the earliest directly dated microbotanical evidence. The Meadowcroft maize had the potential to bridge that gap, but it joins a growing list of macrobotanical remains once thought to represent early use of maize in the Northeast that have been shown to date much later in time or to have been misidentified as maize (e.g., Murphy 1989: 348;Conard et al. 1984;Simon 2014Simon , 2017Simon et al. 2021).
At the time the specimens were recovered, the extent of potential biological disturbance was underestimated. Apparently, the maize specimens were transported downward from a higher level and the extent of the bioturbation was not perceived by the excavators. There were no reasons at the time of the Meadowcroft maize recovery to doubt the maize remains' stratigraphic sequence, association with 14 C dates, or Cutler and Blake's assignments of the majority of cobs to their prehistoric morphotypes. The Meadowcroft results further emphasize the need for AMS dates and δ 13 C IRMS measures on purported maize macrobotanical remains recovered from contexts in the Northeast that are potentially earlier than Grand Banks such as those listed in Table 3.
The gap between the earliest direct dates on maize micro-and macrobotanical remains in the Northeast continues. This situation is not untypical; maize microbotanical remains pre-date macrobotanical remains in several areas of the Americas where environmental conditions do not favor macrobotanical preservation (e.g., Pohl et al. 2007;Lombardo et al. 2020). Ultimately the two lines of evidence need to be reconciled as suggested by Dotzel (2021). This will require additional laboratory and actualistic experimentation to determine under what conditions and contexts maize micro- (Crowther 2012; Raviele 2011) and macrobotanical (e.g., King 1987; Dezendorf 2013; Whyte 2019) remains preserve in the archaeological record.