Soils, Climate, and Agricultural Engineering

Buried A horizon soils occur frequently in Oaxaca, especially in the Mixteca Alta. We use dates from these soils and nearby annual precipitation proxies to show how the fluvial stratigraphy was linked to global climate shifts; how climatic conditions influenced cycles of stream aggradation, downcutting, and relative stability at both local and regional scales; and how human activities also played an important role in shaping the environment.

Here, Leigh and Price describe a meta-analysis of 84 radiocarbon-dated buried A horizons from terminal Pleistocene and Holocene alluvium using the dataset compiled from published field projects (Supplementary Table 1). Almost all samples (81 of 84) are from the Río Culebra and Río Verde watersheds in the Mixteca Alta; three are from the Central Valleys (Figure 1). The alluvial stratigraphic record has seen repeated cut-and-fill cycles, and the latest phase of arroyo incision (ca. 0.5–0.35 ka) commonly exposes 10–15 m tall cutbanks that for decades have been the object of archaeologic, stratigraphic, soil, and radiocarbon sampling.

Small- to medium-sized tributaries (<40 km of total networked stream length) contained 66 of the dated samples, 15 came from mainstem river valleys (60–610 km of total networked stream length), and three were indeterminate. There is no correlation between drainage network length and sample age (R = 0.004).

Thirty-two of the 84 samples are from charcoal, and the remaining 52 are from bulk soil samples. Acknowledging the potential for samples to have been made “older” by incorporation of detrital charcoal and soil, dates from stratigraphic columns with age inversions and samples that disagreed with nearby cultural chronologies were eliminated. One sample, UGAMS-8648 at 23,529 ± 64 ¹⁴C yr BP, was manually excluded as an outlier.

A summed probability distribution (SPD) was generated by aggregating the calibrated probability distributions from all 84 samples into annual bins using the R package “rcarbon” (Crema and Bevan Reference Crema and Bevan2023), resulting in the total sum value of 84 for all 16,000 years (Figure 2). This allows visualization of relative probabilities of encountering dated A horizons.

Figure 2. Summed probability values of encountering a buried A horizon. Each of the 84 samples has a probability of 1.0, so the sum of all of the plotted vertical bars equal 84. The LOESS smoothing line has a data window of 25% (4,000 years) calculated with a first order polynomial. The data peaks (above gray shading) are values greater than the 50th percentile of positive residuals above the LOESS-smoothed line. (Color online)

Paleoclimate context was provided by the ¹⁸O isotopic record of precipitation from two stalagmites in Guerrero, JX-6 from Juxtlahuaca Cave (Lachiniet et al. Reference Lachniet, Pablo Bernal, Asmerom, Polyak and Piperno2012, Reference Lachniet, Asmerom, Bernal, Polyak and Vaquez-Selem2013) and CBD-2 from Diablo Cave (Bernal et al. Reference Bernal, Lachniet, McCulloch, Mortimer, Morales and Cienfuegos2011), about 265 km and 300 km southwest of the Mixteca Alta sites, respectively (NOAA-NCEI 2024). A LOESS model (NIST/SEMATECH 2024) with a data window of 25% (4,000 years) was used to smooth the probability values with a first order polynomial to visualize the steadily increasing values through time and to define data peaks as 50th percentile or higher positive residuals above the LOESS-smoothed line.

The probability values of buried A horizons are not randomly distributed, as there are distinct peaks and valleys (Figure 2). Average probability values tend to increase and become more peaked and broad-based, particularly after 7 ka (in this case we use ka customary in studies of this long timespan). There are very prominent peaks at 13.5–12.6 ka, 3.9–3.5 ka, 1.3 ka, and 0.9–0.5 ka.

The prominent peak at 13.5–12.6 ka represents some bias resulting from 10 of 17 samples from one site, Yuzanú 50 (Lohse et al. Reference Lohse, Borejsza, Joyce, Lohse, Borejsza and Arthur2021). After 12.6 ka, there is a distinct minimum, including complete absence of dated A horizons over many centuries, until a small peak in probabilities emerges at 10.2 ka. These terminal Pleistocene peaks indicate that A horizons were commonly forming during the Bølling-Allerød (14.7–12.9 ka) warming and then disappeared with the onset of the Younger Dryas (12.9–11.7 ka) cooling and during the earliest millennia of the Holocene.

Scattered throughout the Holocene are 13 small peaks of short duration (<100 years), beginning with one at 10.2 ka. Three prominent peaks occur at 3.9–3.5 ka, 1.3 ka, and 0.9–0.5 ka. Dry climate intervals overlap with 10 of the 13 small peaks and all three of the prominent late Holocene peaks, indicating that formation and preservation of A horizons is favored by relatively dry climate. Almost all the A horizon peaks within dry intervals occur >100 years after the apparent onset of relatively dry climate, except for a very short-lived peak at 2.35 ka at the very beginning of a dry interval.

Notable milestones of terminal Pleistocene and Holocene global climate change and geochronology appear in these data. The Bølling-Allerød warming (14.7–12.9 ka) is represented by a pronounced peak, whereas the onset and duration of the Younger Dryas cooling (12.9–11.7 ka) is represented by absence of peaks. Although the 11.7 ka onset of the Holocene and the 8.2 ka event (early/middle Holocene boundary) are not represented here, the 10.2 ka event appears as the onset of small peaks. The 4.2 ka event, the boundary between the middle and late Holocene, is arguably marked by the onset and persistence of prominent broad-based peaks and several high-magnitude narrow (short-lived) peaks after 3.9 ka.

Periods of A horizon peaks are interpreted as times lacking rapid sedimentation and facilitating persistence of a stable soil surface that accumulated organic matter, a pedogenic process known as melanization. In contrast, periods that lack peaks probably had relatively rapid alluvial sedimentation. It is no surprise that peak probability values generally grow with proximity to modern time because the inherent nature of fluvial erosion and deposition favors preservation of progressively younger deposits.

The prominent peak interval at 13.5–12.6 ka is anomalous with respect to the general increase in peaks with decreasing age, which perhaps is explained by the bias imposed by Yuzanú 50. However, even if Yuzanú 50 was excluded there would be a prominent peak represented by a diverse assemblage of sites. These data suggest melanization outpaced alluvial sedimentation during the Bølling-Allerød warming, producing cumulic soils, with the opposite occurring during the Younger Dryas cooling and earliest Holocene when vertical accretion rates were relatively fast (Holdridge Reference Holdridge2016:135), or climatic conditions were so arid that no melanization occurred. Indeed, Mueller et alia (Reference Mueller, Joyce and Borejsza2012) refer to the Allerød as a wet period, and indicate the Younger Dryas was “very dry.” They found that these terminal Pleistocene cumulic A horizons are distinctive stratigraphic markers commonly at the very base of thick cutbank exposures. The probability data support their observation. The basal stratigraphic position probably helped preserve these soils from fluvial erosion. Lohse et alia (Reference Lohse, Borejsza, Joyce, Lohse, Borejsza and Arthur2021) described this soil-forming landscape as seasonally wet meadows, similar to ciénagas in the southwestern United States, but with a distinct dry season, which is a general consensus from previous studies in Oaxaca (Holdridge Reference Holdridge2016; Holdridge and Leigh Reference Holdridge and Leigh2018; Mueller et al. Reference Mueller, Joyce and Borejsza2012). Alluvial floodplains apparently had stream base levels graded down to levels similar to modern streams, the difference being that terminal Pleistocene streams probably occupied wide floodplains, in contrast with modern streams’ deep entrenchment and narrow valleys.

This competition between melanization and sedimentation likely applies to all A horizon probability peaks. Conditions favoring melanization include (1) precipitation and runoff that generates little flooding, sedimentation, or soil erosion, but provides sufficient moisture for plant growth and accumulation of organic matter; (2) stream incision and terracing (arroyo cutting and floodplain abandonment), preventing vertical accretion of flood sediments onto the soil surface until the next phase of alluvial aggradation (with lapses lasting hundreds to thousands of years); and (3) artificial check dams (lama-bordos) that trap sediment immediately above the dams and cause localized rapid sedimentation (while also potentially starving soil surfaces immediately downstream). Suitable climate conditions for melanization must have commonly occurred, but the incision and aggradation dynamics of arroyos (including intrinsic mechanisms unrelated to climate) may preclude derivation of paleoclimate trends directly from the local alluvium.

Arroyo incision repeatedly occurred throughout the Holocene in the Mixteca Alta (Holdridge Reference Holdridge2016; Leigh et al. Reference Leigh, Kowalewski and Holdridge2013; Mueller et al. Reference Mueller, Joyce and Borejsza2012; Rincón Mautner Reference Rincón Mautner1999). It can be caused by multiple climatic, tectonic, structural, intrinsic, and anthropogenic factors (Antevs Reference Antevs1952; Bull Reference Bull1997; Miller Reference Miller2017; Waters and Haynes Reference Waters and Vance Haynes2001). Waters and Haynes noted a distinct climatic pattern wherein “flooding resulting from a period of increased precipitation following a dry period would trigger arroyo cutting” (Reference Waters and Vance Haynes2001:401) in the southwestern United States. They concluded that arroyo incision may be caused by repeated dry-wet cycles driven by climatic teleconnections of the El Niño–Southern Oscillation (ENSO) atmospheric circulation. The pattern of A horizon probabilities falling primarily within relatively dry periods and lagging behind their onset could reflect time intervals most climatically prone to arroyo incision, at least until agricultural engineering and lama-bordo construction began. Mueller et alia (Reference Mueller, Joyce and Borejsza2012) documented four arroyo incision events followed by alluvial aggradation in the headwaters of the Rio Verde watershed: (1) at or before 14 ka, (2) a time-transgressive incision moving upward through the drainage from about 6 ka until 4 ka, (3) a regionally synchronous event at about 0.95–0.8 ka, and (4) postcolonial incision ca. 0.5–0.35 ka following land abandonment. It is likely that more incision events occurred than have been securely dated. Both Bernal et alia (Reference Bernal, Lachniet, McCulloch, Mortimer, Morales and Cienfuegos2011) and Lachiniet et alia (Reference Lachniet, Pablo Bernal, Asmerom, Polyak and Piperno2012, Reference Lachniet, Asmerom, Bernal, Polyak and Vaquez-Selem2013) attribute rainfall variability during the relatively dry late Holocene to oscillations in the intensity of ENSO, which could have triggered arroyo incision leading to numerous stable terrace surfaces with A horizon development during relatively dry episodes. ENSO has existed since at least 7 ka (Moy et al. Reference Moy, Seltzer, Rodbell and Anderson2002), but it is thought to have intensified after 5 ka in southern Mexico (Bernal et al. Reference Bernal, Lachniet, McCulloch, Mortimer, Morales and Cienfuegos2011; Bianchette et al. Reference Bianchette, McCloskey and Liu2017). Three of the four prominent peaks are immediately preceded by the dated incisions noted previously, excluding the brief peak at 1.3 ka. Such incision could have stranded/stabilized terrace surfaces, favoring melanization and stratigraphic preservation. The most frequent cluster of small peaks is situated squarely within the 6–4 ka incision episode and might imply frequent arroyo incision.

Another important factor is agriculture—especially the building of lama-bordos and contour terraces. Lama-bordos are known from about 3.5 ka (Leigh et al. Reference Leigh, Kowalewski and Holdridge2013) and ultimately covered the length of almost all tributaries. They were built to block tributary flow lines, trapping water and sediment, and became strong influences on alluvial aggradation, vertical accretion, and soil cultivation (affecting both A horizon enhancement and degradation); but they also exerted huge influences when they were destroyed or breached, artificially inducing localized stream incision and downstream sedimentation. Lama-bordos undoubtedly played an important role in the formation and preservation of the five apparent A horizon peaks since 2.3 ka. The interval 3.5–2.3 ka, which lacks A horizon peaks, may relate to rapid and widespread artificially induced sedimentation caused by progressive buildup of lama-bordos. This was a time of rapid alluviation (Holdridge Reference Holdridge2016; Mueller et al. Reference Mueller, Joyce and Borejsza2012), sedentary village expansion, and especially 2.7–2.3 ka, major human population growth (Kowalewski and Recorrido Arqueológico de Coixtlahuaca Reference Kowalewski and Recorrido Arqueológico2021; Kowalewski et al. Reference Kowalewski, Balkansky, Stiver Walsh, Pluckhahn, Chamblee, Pérez Rodríguez, Heredia Espinoza and Smith2009).

The last few prominent peaks (1.3 ka, 0.9–0.5 ka) almost certainly represent soils that were highly managed and cultivated, and the latter was the time of maximum population density. The 0.9–0.5 ka peak is bracketed by the last two arroyo incision events noted by Mueller et alia (Reference Mueller, Joyce and Borejsza2012) at 0.95–0.8 ka and 0.5–0.35 ka, corresponding to the Classic/Postclassic and Postclassic/colonial transitions. These two incision events may have been stimulated by human actions/inactions that led to instability of the built-up fluvial landscape.

The availability of so many radiocarbon dates provides a perspective and scale of understanding for the linkage between climate, stream regime, and human activities that eclipses the individual tributary and site. This meta-analysis illustrates how recent studies in the Mixteca Alta and their ¹⁴C dates can now be compared in broader climatic and cultural contexts.

Three Late/Terminal Formative Centers

Our second case is a chronological juxtaposition of the building, use, and abandonment of the major civic-ceremonial buildings chosen because they had detailed published accounts of excavations and multiple radiocarbon samples taken from well-described contexts. San José Mogote was the top central place of the Etla subregion of the Valley of Oaxaca, 14 km north of Monte Albán. Tilcajete was the leading center in the southern Valley of Oaxaca, 20 km south of Monte Albán. Cerro Jazmín, 80 km northwest of Monte Albán, in the Mixteca Alta, grew through a period when most Late Formative settlements in that region were abandoned (Figure 1).

The Mixteca Alta and the Central Valleys had distinct ceramic spheres. Although there are some stylistic similarities, with assemblages found in critical excavation contexts it is often difficult to make chronological assignments finer than the ∼200–300-year ceramic phase, within one region, let alone across regions with different ceramic traditions. We are not dating the sites, all of which have much longer histories, but the construction, uses, and abandonment of particular buildings, some of the most prominent in Oaxaca at the time. Our updated calibrations and modeling confirm the interpretations of the original investigators.

We employ Bayesian modeling, a statistical technique that permits the incorporation of prior knowledge, such as stratigraphy or known calendar dates, to produce more precise, contextually informed chronologies (Bayliss Reference Bayliss2009; Bronk Ramsey Reference Bronk Ramsey2009a). It reduces date ranges, resolves ambiguities, and improves interpretations of archaeological events. This case and those that follow employed OxCal v. 4.4 (Bronk Ramsey Reference Bronk Ramsey2009a), the IntCal20 calibration dataset (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey and Butzin2020), and parameters that downweight statistical outliers (Bronk Ramsey Reference Bronk Ramsey2009b). Architectural components were treated as Phases constrained by Boundaries (capitalized terms Phase, Boundary, Date refer to OxCal Chronological Query Language). Unless otherwise indicated, results are discussed as “interpretive” 68.3% highest posterior density (hpd) intervals rounded to the nearest five years in the text, with the unrounded dates and conservative 95.4% hpd intervals provided in Supplementary Code 1.

At Tilcajete, the El Mogote sector was the Middle Formative civic-ceremonial center; then it was largely abandoned and in the Late Formative a cluster of platforms and plazas was built at the Palenque sector. Here, Redmond and Spencer extensively excavated a sprawling palace (Redmond and Spencer Reference Redmond and Spencer2017) and a temple complex (Redmond and Spencer Reference Redmond and Spencer2013). From the pottery they knew these were Late Formative, Monte Albán Ic buildings. They extracted charcoal samples on floors, sealed under floors, in hearths, in mud mortar, and in adobes, and they made arguments for how these samples should date construction, use (e.g., last fire in a hearth), and abandonment (burned roof beams) (Supplementary Table 1). We selected seven palace and 17 temple samples with the clearest contexts for modeling.

The model results confirm Redmond and Spencer’s interpretation that the two complexes were contemporary. The palace use began between 270 and 80 BC and ended between 70 BC and AD 60. The temple complex was begun between 515 and 125 BC and ended between 30 BC and AD 55 (Figure 3). The date ranges for the end of the palace and temple use are both tight and in close agreement. Redmond and Spencer said that both complexes ended abruptly, the site was abandoned, and a new center founded on top of a mountain 1 km away.

Figure 3. Boundaries of construction phases at three Late/Terminal Formative centers. TJ PC = Tilcajete Palace Complex; TJ TC IC = Tilcajete Temple Complex Initial Construction; TJ TC MU = Tilcajete Temple Complex Main Use; CJ PH = Cerro Jazmín Patio Hunditos; CJ WM = Cerro Jazmín Western Mound; CJ WM FL = Cerro Jazmín Western Mound Foundation and Leveling; CJ WM C = Cerro Jazmín Western Mound Construction; CJ WM RE = Cerro Jazmín Western Mound Reconstruction and Expansion; SJM = San José Mogote; SJM LT = San José Mogote Lowest Temple; SJM ST = San José Mogote Second Temple; SJM UT = San Jose Mogote Uppermost Temple.

Cerro Jazmín, the mountaintop city in the Mixteca Alta, has two foci of monumental architecture, Patios Hundidos and Tres Cerritos, excavated during a project led by Pérez Rodríguez (Pérez Rodríguez and Martínez Tuñón Reference Pérez Rodríguez and Martínez Tuñón2020; Pérez Rodríguez et al. Reference Pérez Rodríguez, Tuñón, Stiver Walsh, Pérez Roldán and Torres Estévez2017). The excavators took charcoal samples from stratigraphically discrete construction events, ovens, and hearths. The stratigraphy and radiocarbon samples fall into three broad construction and use events at what is likely the most public of the Tres Cerritos buildings (the West Mound) and span four stages at Patios Hundidos.

We applied the same modeling procedures for 12 Patios Hundidos and eight West Mound samples. Both foci had episodes of construction and use, from 370 BC to AD 250 at Tres Cerritos and from 235 BC to AD 255 at Patios Hundidos (Figure 3). Within this continuity, Pérez Rodríguez and Martínez Tuñón found evidence of a shift in political institutions from more exclusionary activities limited to a small sector of the population, to more inclusive activities involving greater access and participation. In our modeling, this shift began between 85 and 15 BC. This timing is consistent with an influx and incorporation of people from Mixteca Alta settlements abandoned by the end of the Late Formative (including Monte Negro and many others).

San José Mogote was the primary center of the Valley of Oaxaca until the founding of Monte Albán at about 500 BC. After a possible hiatus lasting several centuries, San José Mogote saw a renaissance in the Terminal Formative, when it became the most important center of the Etla arm of the Valley of Oaxaca. The site center underwent a major civic-ceremonial building program that included a group of temples on top of the largest platform mound. The project led by Flannery and Marcus excavated these temples, including three similar two-room temples built on top of each other (Flannery and Marcus Reference Flannery and Marcus2015:227–292). They collected one sample from each temple—charcoal (probably pine) in incense burners on the stucco floors or on the floor where an incense burner had been used.

Assuming that the samples represent close to the last use of each temple, plus the age of the pine trees used to make the charcoal, and assuming that closure of one temple was soon followed by the construction of the next, the modeled end Boundaries at the 68.3% hpd level should estimate the last use of each temple. These are AD 30–130, AD 75–185, and AD 95–260 (Figure 3). Repeating Flannery and Marcus’s proposition, “Although not conclusive, such intervals make it hard to rule out the possibility that San José Mogote commissioned a new temple . . . every 52 years” (Reference Flannery and Marcus2015:282).

This interregional comparison draws attention to concordant continuity and change, in which state-building played out with varying trajectories at different places. The earliest sector at Tilcajete was the large plaza at El Mogote. That was abandoned in favor of El Palenque, where a new palace and temple were built in an architectural form that would have a long history in Oaxaca. Activity at El Palenque ceased abruptly between 70 BC and AD 60; then, a new center was built on a nearby mountain.

San José Mogote was eclipsed in importance by the early growth of Monte Albán, but it was redeveloped as a secondary center in the Terminal Formative, and sometime between 50 BC and AD 260 it had the succession of three two-room temples.

Valley of Oaxaca centers El Palenque and San José Mogote had some volatility in their histories of new construction and abandonment. Yet the architectural styles (two-room temple, palace with sunken patio) exhibit uniformity and continuity. Cerro Jazmín, in contrast, endured and even grew during tumultuous times that saw other settlements in the Mixteca Alta abandoned. Its public architecture, however, suggests distinctiveness, innovation, and shifting institutional strategies across the entire period.

Three Late Classic Centers

The Valley of Oaxaca ceramic chronology of Monte Albán IIIB-IV/Xoo Phase, broadly AD 500–1000, is short on decorative design, rather static, and under debate (see Feinman Reference Feinman, Huntington, Arnold and Minich2018). Recent excavations in Classic period contexts, with their well-documented ¹⁴C dates, provide an empirical way to describe events more precisely.

We assess continuity and change in public architecture AD 500–1000 at three contemporaneous secondary centers in the Valley of Oaxaca: Lambityeco, El Palmillo, and Macuilxochitl. This is the time when the primary city, Monte Albán, grew to its maximum size and then began to fragment and decline in size. This transition was marked by the construction of exclusive facilities for local nobility at the expense of public spaces.

Lambityeco was an important center with 32 platform mounds more than 2 m high (Macuilxochitl had 29 at the time; Lind and Urcid Reference Lind and Urcid2010), best known for the excavations of Mounds 190 and 195 directed by John Paddock (Lind and Urcid Reference Lind and Urcid2010; Paddock et al. Reference Paddock, Mogor and Lind1968). Those studies documented a major architectural change when a series of superimposed high-status houses was suddenly topped with a new pyramid and patio arrangement. Paddock clearly understood the radiocarbon process and obtained stratigraphically consistent dates on six well-documented samples. His project’s radiocarbon treatment was state of the art in 1968. Unfortunately, the calibrated dates have such wide error bars that the sequence of architectural renovations could have taken place with equal probability any time between about AD 650 and 900.

A recent investigation at Lambityeco (Feinman and Nicholas Reference Feinman, Nicholas, Ray and Götz2019; Feinman et al. Reference Feinman, Nicholas, Lapham, de León, Ríos Allier and Morehart2016) with 70 new ¹⁴C dates (reported here) places in time another significant change in public architecture just 100 m south of Mound 195. This transition consisted of making a ballcourt smaller, blocking the public access to its north end, doing away with a processional access between a temple (Mound 170) and the ballcourt, and changing the character of associated rituals, all suggesting greater exclusivity. When did this occur?

The transition (referred to as Surface 4 [S4] to Surface 3 [S3]) appears in the excavations of Mound 165, Mound 170, and the plaza between the two mounds. We modeled 32 ¹⁴C samples from Mound 165, including charcoal, maize cob, and cacao seeds, resulting in a date range of AD 690–720 for the transition (Figure 4). For the plaza, we modeled 29 ¹⁴C samples including charcoal and carbonized seeds, to come up with an estimated AD 690–720. At Mound 170, the transition, as modeled in the same way with 12 charcoal dates, has range of AD 580–675, interpretable as some decades earlier than Mound 165 and the plaza, or roughly contemporaneous as it is within the 95.4% hpd bracket. Whether the architectural change at Mounds 190/195 happened at the same time as the major remake at Mounds 165/170 could be determined by new assays on carbon recovered by the 1960s project and curated in Oaxaca.

Figure 4. Boundaries from dated Phases at Lambityeco: Mound 165, Mound 170, and the plaza that connects them. Highlighted Boundaries represent the transition from S4 to S3.

El Palmillo is a steep-sided mountain at the eastern end of the Valley of Oaxaca. It is jam-packed with 1,453 residential terraces and was continuously inhabited from Late Formative to Late Postclassic times (Feinman and Nicholas Reference Feinman and Nicholas2004). The most elaborate architecture was built at the top of the hill. This included three large residences—Platform 11, Terrace 335, and Structure 35. At some point people built a formal ballcourt in the narrow space between these residences, and to do so they had to modify the existing residences. This was the most ambitious building project in the history of the elite precinct. We think the behavioral effect was to restrict, make more private, and privilege what went on at the top of the mountain.

When was this project carried out? Structure 35’s masonry directly articulates with the ballcourt’s western mound, which was built on top of Surface 3 of Structure 35, so building the ballcourt happened subsequently, in conjunction with Surface 2 (later) of Structure 35 (Feinman and Nicholas Reference Feinman and Nicholas2011). Four samples from Structure 35, Surface 2 yielded uncalibrated ages of 1283, 1291, 1293, and 1295 years, and a modeled date range of AD 685–775 (68.3% probability level). Figure 5 shows why such a tight cluster of assays has the rather wide 90-year date range.

Figure 5. (A) Date estimate for El Palmillo Structure 35 S2. (B) Examining a single date from the tightly clustered set of dates in this Phase demonstrates that the wide probability distribution derives from a double intercept with the radiocarbon calibration curve.

Macuilxochitl-Dainzu is expansive, multifocal, and has a long history as a secondary center from the Late-Terminal Formative to the Late Postclassic. Faulseit (Reference Faulseit, Dante García Ríos, Pink, Gabriela López Rivera, Javier Rojas Ortíz, Ausel and Lapham2015) directed a study of a single terrace, S25, a prominent place only 100 m from the Late Classic civic-ceremonial-market center. The most important buildings on the terrace were a high-status residence (Area B) adjacent to and contemporary with an unroofed ceremonial platform (Area A). Faulseit took 12 radiocarbon samples from contexts marking construction, use, or last use of both structures (and six more from a third contemporary house). We modeled those from the ceremonial platform, which was started in the interval AD 630–685 (all these ranges are 68.3% hpd). The building was greatly expanded between AD 765 and 875 and fell into disuse by about AD 950 (Figure 6).

Figure 6. Dated Phases from Macuilxochitl. The model employs a terminus post quem (TPQ, earliest date) for dates found between the base of the Complex C platform and bedrock. The natural separation in the dates themselves suggested earlier and later Phases of occupation.

The dating of the continuity and change in public buildings suggests that the breakdown of the Monte Albán state leading eventually to the city-states of the sixteenth century was a geographically and temporally uneven process. Local rulers at Lambityeco and El Palmillo, distant from Monte Albán, sometime around AD 700 tried to restrict access or participation to spaces that had once been more open, which may suggest assertions of independence from the regional state, the state’s declining capacity to underwrite the legitimacy of local rulers, increasing competition among nobility, or a change in the terms of the social contract between town/district citizens and their rulers. Macuilxochitl, if what happened at Terrace S25 is indicative, does not seem to have walled off or made more exclusive its plaza and ritual activities, which instead remained rather open and visible, even expanding in scope, perhaps into the AD 900s. Early in the history of Monte Albán (Feinman Reference Feinman, Gary and Marcus1998), as well as years after that city’s decline (Feinman and Nicholas Reference Feinman and Nicholas2016; Oudijk Reference Oudijk and Jeffrey2008), settlements such as Macuilxochitl were both more visible to and had closer interconnections with Monte Albán, compared to sites farther away such as Lambityeco and El Palmillo, where local rulers took steps to assert their autonomy.

Without radiocarbon, the best we could say was that the events just described took place during Monte Albán IIIB-IV/Xoo phase, a 400–500-year time interval. With multiple radiocarbon samples and processing by current best practices, events can be specified to 50 or 100 years, and as discussed in the following text, targeted studies can often narrow that range to decadal spans.

The Longevity of Houses and Households

How long did prehispanic householders reside in the same location? To answer this, we draw on dates from four sites in the Central Valleys (two terrace sites in hilltop locations, El Palmillo, and the Mitla Fortress, and two sites located on flat alluvium, Ejutla and Lambityeco). Our sample consists of 13 excavated houses—eight at El Palmillo, three at the Mitla Fortress, and one each at Ejutla and Lambityeco (Feinman et al. Reference Feinman, Faulseit, Nicholas, Timothy and Michael2018). The sample spans from AD 400 to 1200.

We assess whether homesteads were occupied for multiple generations, even centuries, or much shorter durations (Barnard Reference Barnard2016; Prindiville and Grove Reference Prindiville, David and David1987; Varian Reference Varian and John2012). Was the longevity of homesteads (residential locations) constrained by the mud/adobe and thatch architectural materials that were used to construct houses in Oaxaca to a generation or two, or were homesteads much more durable? Was there a difference between the two hilltop terrace sites and the households situated in the floodplain? Domestic units on human-built terraces at hilltop sites involved significant labor investments and more interhousehold collaboration to construct (Carballo et al. Reference Carballo, Feinman and López Corral2022; Kowalewski et al. Reference Kowalewski Stephen, Feinman, Nicholas, Heredia, Durrenberger and Martí2006). The houses in the alluvial zone were closer to flat farmland and water sources, the two key factors for agricultural production in the Valley of Oaxaca.

First, a note on how we assess longevity. Modeling and reporting follow the same parameters outlined for the Late/Formative centers case study mentioned previously. We employ the Interval function, and present results at the 68.3% hpd level in the text discussion. This is the conservative choice in matters of duration since at the 95.4% hpd level the measure of house longevity increases substantially (Supplementary Code 1).

Based on our modeling, we found that all 13 house locations were occupied continuously for many generations, minimally more than one century and probably for multiple centuries (Table 1). For example, the excavated house from Ejutla, a valley floor location, was occupied for about 200 years; the house on Terrace 507 at El Palmillo probably lasted between 375 and 505 years; the elaborate house of Mound 165 at Lambityeco was occupied for about three centuries, perhaps longer because the remains of the last stages are poorly preserved. The many generations or several centuries duration fits Smith’s (Reference Smith and Knapp1992:12) concept of household series (see also Hirth Reference Hirth, Robert and Kenneth1993:25), a sequence of households successively occupying the same location or homestead. The duration of residences at terrace sites is particularly remarkable, especially at the Mitla Fortress but also at El Palmillo. One might think that being away from good land and water, uphill, and in closely packed proximity, might mean more ephemeral or temporary homesteads, but the opposite effect seems to be the case.

Table 1. Longevity of Houses.

For the eight domestic units excavated at El Palmillo (Feinman and Nicholas Reference Feinman and Linda2013:142), the three large elaborate houses (Terrace 335, Structure 35, and Platform 11) situated near the apex of the hill were occupied roughly 50 years longer on average than the five smaller domestic units on the slopes below. The three domestic units on the lowest set of terraces (Terraces 1162, 1163, and 1147/48) had a somewhat shorter average duration than the domestic units near the middle or at the top of the site. The start dates for the three large palatial structures (as defined by Barber and Joyce Reference Barber, Joyce, Christie and Sarro2006) at El Palmillo were somewhat later than the start dates for most of the other domestic units at the site. When Feinman and Nicholas (Reference Feinman and Linda2013:142) excavated these palaces, they found relatively early pottery in deposits at the base of the structures but no intact architectural remains. This leaves open the possibility that these residences had even longer durations, as the spaces where they were positioned were unlikely to have been open when lower tiers of this compact hilltop settlement were occupied. Given the long duration of the domestic units at the Mitla Fortress, there is little difference in duration between the three palatial residences and ten smaller residences in this sample.

In sum, we have established that these excavated homesteads typically endured for centuries. People were able to circumvent the material limitations of mud-brick architecture in part because of the foundations of cut stone that anchored adobe brick walls (e.g., Barnard Reference Barnard2016), and by sustained attention to replastering, remodeling, roofs, and rebuilding. Such minor architectural modifications were documented during the excavations of each of the residences.

Early Colonial Transformations

The fourteenth to sixteenth centuries in Oaxaca were a time of tremendous change and political turbulence as multiple empires, foreign and domestic, vied for control of territory and routes of commerce. The Spanish invasion attracts the most attention, but those of us who work in Oaxaca also talk about Mexica empire building and Mixtec and Zapotec plays for power across the highlands, Isthmus, and coast.

Much of what we know comes from written records, including codices, maps, and religious texts and documents created for the Spanish colonial authorities such as lienzos, testaments, and wills. Archaeological research is more limited and the challenge becomes connecting specific historical events described in documents with archaeological materials that are frustratingly vague.

Further complicating matters is the radiocarbon curve for these centuries. The challenge is a major plateau and reversal in the calibration curve such that any dated event occurring between AD 1480 and 1630 will typically intercept the curve three times, yielding an unhelpful age range. The radiocarbon curve thus does not narrow the time frame closer than what archaeologists already likely knew.

Nejapa and Tavela in the southern Sierra Sur of Oaxaca are situated along a trade corridor through which Aztec, Zapotec, and Spanish migrants and conquerors traveled. Recent work by Michel Oudijk (Reference Oudijk2019) and Guillermo Ramón Celis (Reference Ramón Celis2024) dates the initial migration of Zapotecs to the Isthmus to around AD 1350, with a much larger migration around AD 1450. According to Francisco de Burgoa (Reference Burgoa1934 [1674]), the Zapotecs secured their route to the Isthmus by establishing armed garrisons, including perhaps in Nejapa (King et al. Reference King, Workinger, Konwest, Elvis Badillo and Jarquín Enríquez2014). Later, in the 1490s, they successfully defended themselves against an Aztec attempt to conquer them in the Isthmus. In the early 1500s, the Zapotecs allied with the Spanish against the Aztecs to maintain control in the Isthmus (Oudijk Reference Oudijk2019). Their political sovereignty was honored until their ruler was arrested and accused of idolatry in the 1550s, at which point the Spanish seized political control.

The story we get from the archaeology is far blander (King Reference King, Christine and John2020, Reference King, Hendon, Overholtzer and Rosemary2021). Here we highlight Majaltepec, a town in the mountains of Nejapa (subject to Santiago Nexapa), which according to Spanish sources was occupied by Mixe-language speakers. Based on colonial court documents, the people who lived in Majaltepec used the Spanish legal system to fight for the right to have horses, use a saddle and tack, and travel/work as merchants; others requested land for raising sheep and pigs and another group filed a complaint about being forced to provide labor to build the church in Nexapa and work in the mines. Dominican church authorities in Nexapa also complained in court documents about having to travel so far to minister to rural Indigenous communities including Majaltepec, which may have led to Majaltepec’s community members launching legal complaints against Spanish efforts to relocate their town from the mountains to the valley in the early 1600s congregación campaigns, which at some point succeeded. The last documents that mention Majaltepec as an occupied town date to the 1760s when residents were party to complaints about a corrupt colonial representative alcalde mayor (King and Konwest Reference King, Konwest and Rani2019). The town now lies abandoned and has been for many generations. When was Majaltepec abandoned? When was Majaltepec first settled?

On three 10-day field excursions from 2009 to 2013, King and colleagues mapped and excavated Majaltepec alongside members of the town of Santa Ana Tavela, the municipality on whose land Majaltepec lies. Many residents of Tavela are descendants of people who once lived in Majaltepec. The excavations documented the building that likely served as the church—the open churchyard, the entrance steps to the church, and its interior rooms, which were built with wide stone foundations and upper walls of adobe (the building is known locally as “siete cuartos”). The project documented a few other buildings nearby, including a house constructed with patterned adobe block walls and earthen floors. Eight individuals were buried beneath the floor of this house in our limited (3 m²) excavations. Interred with them were nearly 450 glass beads. No imported obsidian was found, suggesting that the networks of obsidian exchange had already been impacted by Spanish colonialism (King and Konwest Reference King, Konwest and Rani2019).

The team collected charcoal and burnt wood from several contexts: from below and above the church floor, fragments of burnt wood from the post on the stairs of the church, and charcoal found with the burials beneath the floor in the patterned adobe house. The radiocarbon dates once again initially provided us with the two-sigma calibrated date range “AD 1430 to 1630”—rather unhelpful.

The beads provide a better age estimate. The residents of Majaltepec most likely acquired beads through their interactions with Dominican authorities; gifting beads was a common church practice. The Dominican monastery in Santiago Nexapa was founded in 1553, after a priest had been there for four years (Gerhard Reference Gerhard1993). The Spanish villa of Nexapa was formally settled in 1560, but by that time Dominicans had already been active and ministering to rural communities. All but 55 of the Majaltepec beads correspond to types found at the Franciscan Mission of St. Catherine’s Island in Georgia, in the United States. This mission operated from 1605 to 1680, and church authorities received and distributed beads through similar mechanisms (Blair et al. Reference Blair, Lorann and Peter2009). We found 38 beads that correspond to Blair et alia’s Type 18, blue glass beads made with the a speo method, produced in France between AD 1560 and 1750. This date helps refine the occupation of Majaltepec (Konwest et al. Reference Konwest, King and de León2020).

We modeled the ¹⁴C samples for the construction and use of the church and the use of the house with the burials with the glass beads. Given the Dominican presence and the earliest possible appearance of the glass bead types, the model employs a terminus post quem of 1554. With that strong constraint and incorporating the contextual information, the dates for the church’s construction and use (including the sequential floor deposits), resolve rather well: there is a 68.3% probability of the church being used between 1580 and 1615 (95.4% between 1560 and 1630; Figure 7). A date on maize found with the burial and its beads falls in the interval 1595–1630 (68.3%), or more conservatively 1575–1640 (95.4%).

Figure 7. Modeled dates for the Majaltepec church. The model employs a “hard” TPQ of 1554, marking the arrival of Dominican Friars and considers architectural elements independently and in Sequence within the Phase. Maize from a burial not associated with the church itself is considered outside the Phase but in the overall Sequence.

Even with the difficult 1400–1600 plateau and reversal in the calibration curve, radiocarbon work can be quite valuable given carefully selected, well-described samples and Bayesian modeling using documented prior constraints. The Nejapa case can be compared to Yucundaa (Teposcolula Pueblo Viejo), where early colonial events (including deaths due to Old World epidemic disease) were identified and assigned to narrow time brackets with ¹⁴C dates (Tuross et al. Reference Tuross, Warinner, Robles García, Spores and García2014).