Editor's evaluation: Domestication and lowland adaptation of coastal preceramic maize from Paredones, Peru
Year: 2022
Type: peer-review
Abstract: Full text Figures and data Side by side Abstract Editor's evaluation eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Archaeological cobs from Paredones and Huaca Prieta (Peru) represent some of the oldest maize known to date, yet they present relevant phenotypic traits corresponding to domesticated maize. This contrasts with the earliest Mexican macro-specimens from Guila Naquitz and San Marcos, which are phenotypically intermediate for these traits, even though they date more recently in time. To gain insights into the origins of ancient Peruvian maize, we sequenced DNA from three Paredones specimens dating ~6700–5000 calibrated years before present (BP), conducting comparative analyses with two teosinte subspecies (Zea mays ssp. mexicana and parviglumis) and extant maize, that include highland and lowland landraces from Mesoamerica and South America. We show that Paredones maize originated from the same domestication event as Mexican maize and was domesticated by ~6700 BP, implying rapid dispersal followed by improvement. Paredones maize shows no relevant gene flow from mexicana, smaller than that observed in teosinte parviglumis. Thus, Paredones samples represent the only maize without confounding mexicana variation found to date. It also harbors significantly fewer alleles previously found to be adaptive to highlands, but not of alleles adaptive to lowlands, supporting a lowland migration route. Our overall results imply that Paredones maize originated in Mesoamerica, arrived in Peru without mexicana introgression through a rapid lowland migration route, and underwent improvements in both Mesoamerica and South America. Editor's evaluation In this important article, the authors characterize ancient DNA from maize unearthed in archaeological contexts from Paredones and Huaca Prieta in the Chicama river valley of Peru, recovered by painstakingly controlled excavation. The genetic evidence, while from a small number of samples, is compelling, although the dating evidence has to rely on archaeological context, which fortunately is excellent. The difficulties of direct radiocarbon dating of the samples in this case are appropriately discussed by the authors. https://doi.org/10.7554/eLife.83149.sa0 Decision letter Reviews on Sciety eLife's review process eLife digest The plant we know today as maize or corn began its story 9,000 years ago in modern-day Mexico, when farmers of the Balsas River basin started to carefully breed its ancestor, the wild grass teosinte parviglumis. Recent discoveries suggest the crop may have started to travel to South America before its domestication was fully complete, leading to a complex history of semi-tamed lineages evolving in parallel in different regions. For example, 5,300-year-old corn specimens found in Tehuacán, in central Mexico, still genetically and morphologically resemble teosinte. Meanwhile, cobs harvested about 6,700 to 5,000 years ago on the northern coast of Peru – 3800km away from where maize was first domesticated – look like the ones we know today. Vallebueno-Estrada et al. aimed to explore the evolutionary history of this Peruvian maize, which was discovered at the archaeological coastal site of Paredones. To do so, they extracted and sequenced its genetic information, and compared these sequences with those from modern varieties of lowland and highland maize, as well as from teosinte parviglumis and teosinte mexicana. The analyses showed that the ancestor of the Paredones maize emerged from teosinte parviglumis like any other lineage, but that it was already domesticated when it started to spread South; by the time it was present in Peru 6,700 years ago, it was genetically closer to modern-day crops. This early departure is consistent with the fact that the Paredones specimens lacked teosinte mexicana genetic variants; this highland relative of lowland parviglumis is believed to have interbred with maize lineages from Central America more recently, when these were brought to higher altitudes. The presence of genetic marks tailored to low-elevation regions suggested that the Paredones maize lineage migrated through a coastal corridor connecting Central and South America, arriving in northern Peru about 2,500 years after first arising from teosinte parviglumis in Central America around 9,000 years ago. Under the care of rapidly developing Central Andean societies, the crop then evolved to adapt to its local conditions. Maize today has spread to all continents besides Antarctica; we produce more of it than wheat, rice or any other grain. How our modern varieties will adapt to the environmental constraints brought by climate change remains unclear. By peering into the history of maize, Vallebueno-Estrada et al. hope to find genetic variations which could inform new breeding strategies that improve the future of this crop. Introduction Maize constituted 12% of global crop production in 2019, second only to sugar cane (Food and Agriculture Organization of the United Nations, 2021). Like many crop plants, global maize production is threatened by climate change, especially in the middle to low latitudes (Li et al., 2022) where maize is dominant. Maize has the allelic diversity to adapt, but much of this variation is partitioned differentially across populations (Hufford et al., 2012b; Romay et al., 2013; Zila et al., 2013). Understanding the development of population dynamics in maize not only allows better understanding of the evolutionary processes that produced a globally important crop but will highlight populations that can be used in breeding to adapt to changing climates. Although the origins of maize (Zea mays ssp. mays L.) based on archeological data puzzled the scientific community for several decades (Beadle, 1939; Mangelsdorf, 1974; Merrill et al., 1940), the integration of genomic, archeological, and botanical evidence has identified the Balsas basin in central Mexico as the only center of origin for maize (Iltis, 1983; Matsuoka et al., 2002; Ranere et al., 2009), and that the divergence from its wild ancestor, Zea mays ssp. parviglumis (hereafter parviglumis), occurred about 9000 years ago (Matsuoka et al., 2002). Domestication occurred in a single event creating a monophyletic clade that includes all domesticated maize landraces and diverges from both parviglumis and Zea mays ssp. mexicana (hereafter mexicana) populations (Matsuoka et al., 2002). Genomic investigations of archeological samples from the Tehuacan highland site suggested that the dispersal of maize to the highlands of México was complex, as early-arriving maize populations retained higher levels of genomic diversity than expected for domesticated plants (Ramos-Madrigal et al., 2016; Vallebueno-Estrada et al., 2016). The constant gene flow between domesticated maize with already divergent populations of parviglumis and mexicana has contributed to the adaptation of maize to new environments and remains embedded in the genetic structure of its populations (Swarts et al., 2017; van Heerwaarden et al., 2011). Geographic areas of contact have been stable over time, as these teosinte populations have maintained a discrete distribution in central Mexico since the last glacial maximum (Hufford et al., 2012a). Gene flow from a sympatric mexicana population to domesticated maize populations has been associated with an altitudinal cline in the highlands of Mexico and Guatemala (Hufford et al., 2013; van Heerwaarden et al., 2011; Wang et al., 2017), and the genetic introgression from mexicana in the form of distinct chromosomal inversions has been associated with adaptation of maize to central Mexican highlands (Hufford et al., 2013; Wang et al., 2017). The genetic contribution of teosinte mexicana to Mexican highland landraces is about 20% (Hufford et al., 2013; van Heerwaarden et al., 2011), with an estimated time for this introgression of around 1000 generations (Calfee et al., 2021). Archaeological evidence supports the dispersal of maize populations to South America to be associated with a Pacific lowland coastal corridor (Bonavia and Grobman, 2017; Dillehay et al., 2008; Randolph, 1959). Population substructure and differentiation patterns suggested independent adaptations to highland environments in Mesoamerica and South America; meanwhile, minimal population sub-structuring was detected between the lowlands of Mesoamerica and South America (Swarts et al., 2017; Takuno et al., 2015) indicating continuous gene flow over time. While the introgression from mexicana of chromosomal inversions on chromosomes 3, 4, and 6 has been shown to contribute to adaptation to Mesoamerican highlands (Romero Navarro et al., 2017), those regions were not detected in highland maize populations of South America (Wang et al., 2017) or North America (Swarts et al., 2017) which were also isolated from direct gene flow with parviglumis (Kistler et al., 2018). Although these inversions are specific to Mexican landraces, many maize populations across the Americas including South America show genome-wide admixture with mexicana (Swarts et al., 2017). Highland South American landraces also show phenotypic diversity relative to lowlands (Bonavia, 2013), as well as specific cytogenetic characteristics such as the absence of supernumerary highly heterochromatic B chromosomes in Peruvian landraces, resulting in the so-called Andean Complex (McClintock et al., 1981). These unique characteristics have puzzled the scientific community regarding the origin and adaptation of the Andean Complexes (Bonavia, 2013; Goodman and Bird, 1977; Grobman, 1961; McClintock et al., 1981; Randolph, 1959; Wilkes, 1979). The archeological expeditions at the sites of Paredones and Huaca Prieta in the coastal desert of north Peru yielded a robust collection of ancient maize remains that provide a unique opportunity to investigate the chronology, landrace evolution, and cultural context associated with early maize dispersal in South America (Bonavia and Grobman, 2017; Dillehay et al., 2012; Grobman et al., 2012). These findings include one charred cob fragment dated 6775–6504 calibrated BP (at 95.4% probability), and other burned cobs stratigraphically dated to similar and later ages (Bonavia and Grobman, 2017; Dillehay et al., 2012; Grobman et al., 2012), representing the most ancient maize macro-specimens found to date. Strikingly, and in contrast to Mexican cob fragments from Guilá Naquitz dating approximately the same time or slightly younger (6235 calibrated BP) (Benz, 2001), the Paredones and Huaca Prieta specimens are robust, slender and cylindrical 2.4–3.1 cm long cobs, with eight rows of kernels consistent with the hypothetical Proto-Confite Morocho landrace (Benz, 2001; Dillehay et al., 2012; Grobman et al., 2012). Recently, it has been proposed that the first maize lineages arriving in South America were partial domesticates, locally evolving the full set of domestication traits due to reduced gene flow from wild relatives that enhanced anthropogenic pressures (Kistler et al., 2020; Kistler et al., 2018). However, it is not clear how fast this process could have been and if the earliest archeological samples found in South America were partially or fully domesticated. In addition, the expectation on the phenotype of those hypothetical samples is not clear. Here we present the genomic analysis of three ancient specimens belonging to the earliest cultural phase of Paredones and radiocarbon dating 6775–6504, ~5800 to 5400 (dated by direct association with wood charcoal), and 5583–5324 2σ calibrated years BP at 95.4% probability. To reveal the population context of their origin and domestication, we conducted comparisons with parviglumis, mexicana and extant maize landraces. Also, to explore if these ancient maize samples exhibit some evidence of mexicana gene flow, we performed D-statistics under several experimental designs, comparing them to extant maize and parviglumis populations. Finally, we did a comparison with previously published data from extant highland and lowland Mesoamerican and South American landraces, to identify signatures of specific adaptation that could bring insights into the specific improvements that this maize went through in both Mesoamerica and South America. Our results provide evidence that ancient Peruvian maize originated in Mesoamerica as all landraces found to date, followed by a rapid dispersal into the lowlands of South America, and subsequently was subjected to local adaptation processes. Results Paredones ancient maize sampling The maize macroremains were collected as part of published excavations at the Paredones and Huaca Prieta sites (Grobman et al., 2012). Macroremains from both sites were excavated in deeply stratified and undisturbed cultural floors. Stratigraphic Units 20 and 22 at Paredones are the archeological component with the largest and most diversified amount of maize remains, with the oldest 14C dated cobs. The oldest cobs derive from near the base of this unit, in a single, discrete and intact floor of ~2 cm in thickness and at 5.5 m in depth from the present-day surface (Bonavia and Grobman, 2017; Grobman et al., 2012). The dated remains at both sites are chrono-stratigraphically bracketed by and in agreement with more than 165 dates from mound and off-mound contexts that were obtained by Accelerator Mass Spectrometry (AMS) and Optically Stimulated Luminescence (OSL) (Bonavia and Grobman, 2017; Dillehay, 2017; Dillehay et al., 2022). No taphonomic or other disturbing cultural or geological features were observed in any excavation units that would have altered the integrity and intactness of strata containing the maize remains (Material and Methods). All radiocarbon-dated remains were assayed by the SHCal04 Southern Hemisphere Calibration 0–11.0 calibrated kyr BP curve (McCormac et al., 2004). In 2020, Dillehay and geologists Steven Goodbred and Elizabeth Chamberlain carried out excavations in a Preceramic domestic site (S-18) located ~3.2 km north of Huaca Prieta. Preceramic corn remains were encountered consistently in hearths and in the upper to lower intact cultural layers of the site. As with parts of the Paredones and Huaca Prieta sites, the lower hearths and strata contained uncharred cobs 2.6–3.1 cm long, slender and cylindrical with eight rows of kernels, of the smaller and earliest type of identified corn species Proto Confite Morocho (Grobman et al., 2012). The middle to upper strata yielded the known later and slightly larger Preceramic varieties of Confite Chavinese and Proto Alazan. An OSL date from a discrete and intact lower layer containing a hearth with two unburned cob fragments of the Proto Confite Morocho variety assayed ~7000 +/- 630 years ago or 7560–6300 BP (Chamberlain, 2019). Wood charcoal from the hearth was processed at 7162–6914 calibrated BP (at 95.4% probability, AA75398), suggesting that the associated cob fragments date ~6800 years ago. In South America, maize micro remains (e.g. starch grains, pollen, phytoliths) have been estimated to date 7200–7000 calibrated BP (Piperno, 2006; Piperno and Pearsall, 1998; Zarrillo et al., 2008) at sites in southwest coastal Ecuador, located ~450 km north of Huaca Prieta and Paredones, and in other localities across the continent at ~6500 calibrated BP and later (Kistler et al., 2018). Excavation Units 20 and 22 from Paredones are illustrated in Figure 1A and B. Three of the recovered maize samples (Par_N1, Par_N9, and Par_N16) are well-structured maize cobs deprived of seeds and showing morphological similarities to extant landraces. Par_N1 is the most ancient specimen, a burned maize husk and shank fragment dating 5900 ± 40 14 C years BP (6770–6504 calibrated BP at 95.4% probability, OS860020), and obtained from archeological Unit 22 (Figure 1C and Figure 1—figure supplement 1). The other two samples were found in Unit 20, Par_N9, a slightly charred maize husk fragment, that radiocarbon assayed at 5582–5321 calibrated BP (at 95.4% probability, AA86932), and Par_N16, an unburned cob fragment, which stratigraphically dated by direct association with wood charcoal in a hearth at 5603–5333 calibrated BP (at 95.4% probability, AA86937; see Figure 1C and Table 1). Other charred cobs from overlying, younger strata in these units or from other units assayed between 4800–3800 calibrated BP (at 95.4% probability). These dates are stratigraphically bracketed by radiocarbon assayed intact hearths and prepared floors that are in complete chrono-stratigraphic agreement (Bonavia and Grobman, 2017; Dillehay et al., 2012; Grobman et al., 2012). Par_N1 is older than any other maize macro-specimen found to date (Torres-Rodríguez et al., 2018). Figure 1 with 1 supplement see all Download asset Open asset Archeological site and specimens of Paredones. (A) Topographic contour map of Huaca Prieta and Paredones (units U20, U22, and U27) coastal sites, showing the location excavation units. (B) The Paredones mound during archeological excavations. (C) Maize specimens Par_N1 (dating 6775–6504 calibrated years BP), Par_N9 (dating to 5800–5400 calibrated years BP), and ParN16 (dating 5583–5324 calibrated years BP); Scale bar = 1 cm. Table 1 Radiocarbon and calibrated dates of maize specimens from Paredones. Lab no.SiteAssociated Dating no.Unit / stratum14 C years BP95.4% probablityδ13C valueDated MaterialPar_N1ParedonesOS8602022/185900±406775–6504–10.3Husk/shank fragment attached to cobPar_N9ParedonesAA8693220/6b-184770±355582–5321–23.5Husk fragment attached to cob*Par_N16ParedonesAA8693720/6b-184849±315603–5333–25.8Wood charcoal in associated hearth† * Aberrant δ13C assay. Attachment to cob and molecular data confirm maize. † Maize cob directly associated with hearth. Paleogenomic characterization of ancient maize samples To determine the genomic constitution and degree of genetic variability present in the 6775–5324 calibrated years BP (at 95.4% probability) maize of Paredones, we extracted DNA and conducted whole-genome shotgun sequencing in specimens Par_N1, Par_N9, and Par_N16. Since the endogenous DNA content of all three specimens was low (0.2% for Par_N1 and Par_N9; 1.1% for Par_N16), we conducted in-depth whole-genome shotgun sequencing of high-quality libraries under Illumina platforms, generating 622 million (M) quality-filtered reads for Par_N1, 423 M for Par_N9, and 392 M for Par_N16. Due to its higher endogenous DNA content (one order of magnitude larger), we further sequenced the Par_N16 library, obtaining 459 M additional reads, to generate a total of 851 M for this sample (Table 2). Comparison with version 3 of the B73 maize reference genome resulted in 1,320,284 (Par_N1), 1,034,544 (Par_N9), and 15,023,803 (Par_N16) reads mapping to either repetitive (33.4% for Par_N1; 34% for Par_N9; 34.8% for Par_N16) or unique (66.5% for Par_N1; 66% for Par_N9; 65.2% for Par_N16) genomic regions, for a total virtual length of 52.2 Mb (Par_N1), 40.8 Mb (Par_N9), and 471.65 Mb (Par_N16) of the unique maize genome (Table 2). Average mapping quality in Phred score was 31.9 for Par_N1, 32.1 for Par_N9 and 34 for Par_N16; this is reflected in the estimated error rate of 1.19E-02 for Par_N1, 9.59E-03 for Par_N9 and 1.05E-02 for Par_N16 (Table 2). Reads contained signatures of DNA damage typical of postmortem degradation in ancient samples, including overhangs of single-stranded DNA, 13–20% cytosine deamination and fragmentation due to depurination (Dabney et al., 2013) resulting in median fragment lengths of 36 bp for all three samples. A total of 42–53% of all covered sites had signatures of molecular damage (Figure 2). This damage pattern is an indication that this is ancient endogenous DNA and does not represent DNA contamination from extant sources. After mapping reads corresponding to unique genomic regions, Par_N1, Par_N9 and Par_N16 yielded approximately 16.9 M (Par_N1), 12.1 M (Par_N9), and 334.36 M (Par_N16) unique genomic sites spread across all 10 chromosomes at an average depth of 1.2 X (Table 3 and Figure 2—figure supplements 1–4), which were used as a platform for subsequent studies. Figure with supplements see all Download asset Open asset DNA damage and fragmentation patterns of ancient maize samples. DNA around and DNA relative to the and the two for damage signatures to and to are shown in and while other of are shown in Table Paleogenomic characterization of three ancient maize samples from Paredones. number of number of quality of sequences mapping to of sequences mapping to repetitive of sequences mapping to the unique length length quality rate / Table 3 number of unique genomic sites covered at in ancient Paredones samples. compared to the B73 reference Par_N1, Par_N9, and Par_N16 yielded and single To any by postmortem all corresponding to a cytosine (C) to or to (A) were not for subsequent analysis damage et al., et al., 2001; Table All corresponding to or were also a previously et al., Par_N1, Par_N9, and Par_N16 yielded and with in the maize diversity most of which were only covered at 1 X due to the low amount of endogenous DNA recovered (Table this low the to a previously for Par_N1, for Par_N9, and for Par_N16), suggesting that this an of maize that can be used to determine its evolutionary To determine if the specific of C to and to A could the results in of maize than teosinte an additional was in which all were only were in Par_N16 it was the only sample with sequencing data to this This second of between this sample and the this we conducted parallel which results are shown in the corresponding Table of and recovered from ancient Paredones samples. number of number of in in between ancient maize, extant landraces, and Balsas teosinte To better the origin and domestication of South American maize, we the evolutionary between Paredones teosinte parviglumis and mexicana, and extant maize landraces. We a through patterns of population divergence to genome-wide the three ancient Paredones samples were (Figure 2—figure supplement were for DNA based on the of between of the samples and the maize that includes B73 as a reference genome and 22 maize landraces several in teosinte parviglumis two of teosinte mexicana, and a single of as the 1). In the case of Par_N16, the resulting shows all maize landraces and teosinte into two distinct all from as previously (Hufford et al., 2012b; Matsuoka et al., 2002). Par_N16 is in a clade that includes extant maize landraces, and this is for all Par_N16 is not in its clade but with – the only extant Peruvian landrace in the reference – in a with South American landraces such as and These that the ancient samples are monophyletic with modern maize, supporting a single domestication and that they are most to modern samples from the same suggesting an between them and modern South American (Figure 3 and Figure supplement 1). In the case of Par_N1 and Par_N9, and although with were and the resulting is with both samples at the same as Par_N16 (Figure supplements and As other ancient samples previously Paredones samples to have long in which can be by by time. the other the fact that 3 independent samples present the same in the that molecular which is is not their A parallel analysis that only showed the same where Par_N16 with the South American landraces the maize monophyletic clade (Figure supplement This shows that the of the Paredones ancient samples is not by the molecular damage Thus, based on genome-wide Paredones maize with extant domesticated Andean landraces, supporting a single origin for maize and that these Peruvian samples were already domesticated by ~6700 Figure 3 with supplements see all Download asset Open asset domestication of ancient Peruvian maize. between ancient Par_N16 maize and its wild and from an of genome-wide regions of the reference maize The teosinte is in the maize landrace in and the ancient maize sample from Paredones in The teosinte and landrace previously and are in the The Par_N16 was for a with the complete can be in Figure supplement of gene flow from mexicana Par_N16 was the only sample with DNA data to this All the samples showed the same was to be of ancient Paredones maize. To investigate the genetic of this maize with teosinte mexicana, we estimated D-statistics in the form mexicana, that the of due to of across different divergence an gene flow over alleles from parviglumis to and mexicana to (Figure supplement 1). We used highland as a in the form mexicana, and lowland as a in the form mexicana, The results of show that the with from the gene flow mexicana. Meanwhile, remains in with around The D-statistics distribution of Par_N16 mexicana, is similar to the distribution of and significantly different from the distribution of (Figure supplement 2). The of all 1000 is in all suggesting that are consistent across the These results with the analysis in which only were used (Figure supplement showing that the absence of gene flow between Par_N16 and mexicana is not by the molecular damage To further confirm the absence of mexicana introgression in
Source: eLife
Author: Detlef Weigel
Institution: Max Planck Institute for Biology
Language: English
Subfield: Paleontology
Field: Earth and Planetary Sciences
Domain: Physical Sciences
SDG: Zero hunger
Open Access status: gold
Estimated APC paid: $2,500
OpenAlex ID: W4366313437