9
Soils and Agricultural Carrying Capacity

A History of Soil and Related Study

The modern environment of Chunchucmil challenges farmers today and probably challenged ancient farmers because of its aridity and skeletal soils (see chapter 6 for a discussion of potential differences between modern and ancient sea level and rainfall). For decades, researchers (Beach 1998a; Dahlin et al. 2005; Dunning and Beach 2010; Farrell et al. 1996; Hixson 2011; Hutson et al. 2007; Vlcek et al. 1978; Magnoni 2008; Ardren 2003; Sweetwood et al. 2009) have investigated the economy and soil resources of the Chunchucmil region as part of the Pakbeh Regional Economy Program (PREP) in order to understand the reasons Chunchucmil prospered from the Early Classic to the Middle Classic. All of these investigations conclude that subsistence was difficult here because of the large population, both at Chunchucmil (chapter 6) and its hinterland (chapter 8), living in this environmentally challenged region. The present chapter endeavors to solve this riddle of how the ancient Maya of Chunchucmil fed themselves through the lenses of soils, food, water, and sustainability practices. Whereas chapter 6 provided a broad portrait of the different vegetative and geomorphic zones within the Chunchucmil Economic Region (CER), this chapter looks very closely at soils and carrying capacity.

The paucity of soil resources, the low rainfall, the enormity of the site, and the substantial settlement in Chunchucmil’s hinterland has led us to reevaluate the assumption that Maya households were agriculturally self-sufficient (Drennan 1984a, 1984b; Sanders and Webster 1988). Chunchucmil project participants sought multiple lines of evidence to examine and reexamine the subsistence economy (Dahlin et al. 2005). For example, Beach (1998a) and Dahlin et al. (2005) suggested that the lack of food-producing structures, insufficient sustaining area, and poor soil resources may indicate that agriculture was based on atypical crops or methods, or that food trade would have been required to supplement the food resources that could not have been produced at Chunchucmil under traditional agricultural methods.

To answer the subsistence conundrum of Chunchucmil, we set the following main objectives:

Agriculture Today

As discussed in chapter 5, our evidence suggests that few people have lived near Chunchucmil since the Terminal Classic. Some historic human disturbances included henequen plantations and, more recently, a few minor agricultural programs supported by the Mexican government: papaya (Carica papaya), aloe (Aloe vera), citrus (Citrus spp.), and chili (Capsicum spp.) (Dahlin2003). The contemporary village of Chunchucmil has about 1,000 people (Luzzadder-Beach2000), of which a small percentage currently practice milpa agriculture (shifting cultivation, slash-and-burn agriculture) (Beach1998a), while the rest supplement their income with ranching or jobs in nearby cities. These few milperos (maize farmers) have the opportunity to cultivate the more fertile soils of the region, and yet compared to the average world maize yields of 5.55 MT (metric tons)/ha (USDA 2015), their yields are still extremely low and erratic: as high as 0.25 to 1 MT/ ha in a good year (Beach 1998a) and as low as 0.1 MT/ha (Shuman1974). Even during the henequen era, the plantation at Chunchucmil produced half as much as those plantations farther east (Vlcek et al.1978). We are unsure if ancient agriculture was more productive in the ancient Maya lowlands (Hester1953), though evidence is growing for many new crops from ancient Maya studies (Dunning et al. 2015; Beach, uzzadder-Beach, Cook, et al. 2015.

As discussed in chapter 6, the northwest Yucatán receives little rain. Annual rainfall for the region has been reported to vary between 700 and 1000 mm, yet the average rainfall recorded at a weather station in the village of Chunchucmil over the last 17 years is 640 mm (see also chapter 6). In contrast, southern Belize’s annual rainfall is between 3,000 and 4,000 mm. The Chunchucmil region has enough water to grow maize, but it lies close to the maize minimum of 500 mm per year (Wellhausen Fuentes, and Hernánde 1957) and a small reduction in rainfall could greatly diminish maize yields. In addition, 80–90 percent of that rainfall occurs from June to October, and is highly variable and localized. All point to marginal circumstances for traditional maize cultivation today. We should note that our knowledge of water availability at Chunchucmil is limited by a shortage of data regarding contributions from the nearby water table (perhaps about 1 m deeper in antiquity). Furthermore, as we discuss below, soils might have been deeper in the past and therefore with greater water-holding capacity.

Early studies of the modern Yucatec Maya showed that maize made up about 85 percent of their diet (Emerson1953), which equates to a little more than 0.2 MT of maize per year per capita (Shuman1974; Steggerda1941). This equates with a land requirement of 0.2–2 ha/capita in an area with crop yields of 0.1–1 MT/ha. Dahlin et al. (2005) conservatively estimated the sustaining area of Chunchucmil to be 1,600 km² using a Thiessen polygon, but he stated that what was actually available to them must be significantly less because of major overlapping with Siho’s and Oxkintok’s sustaining areas to the south and east, respectively. The area drops to roughly 800–1,200 km² when we exclude structures and areas with no soil cover (25–50%). The seasonally inundated savanna is considered unfit for cultivation (Garza Tarazona de González and Kurjack1981) and reduces the sustaining area to Beach’s (1998a) proposed estimate of 600 km².

Our teams observed several families about 5 km east of Chunchucmil growing dry-season maize in May 2006. Farmers carried groundwater in buckets to every stalk nearly every day. This is labor intensive but similar to what we would expect in ancient Maya dry-season farming. We hypothesized that if ancient farmers incorporated dry-season maize into Chunchucmil’s subsistence systems, we should find a plethora of wells or sascaberas dug to the water table. As discussed in chapter 2, we found 18 unquestionably ancient wells. There may have been many more that are now obscured by soil or collapsed architecture. Of the 270 sascaberas mapped, we do not know how many of them reach the water table. To complicate the situation further, intensive irrigation on milpas or home gardens could potentially reduce crop yields and damage susceptible crops (especially seedlings) due to salt accumulation (Luzzadder-Beach 2000). The groundwater at Chunchucmil has an average electrical conductivity (EC) of 1.2 dS/m and ranges of 0.3 to 3.1 dS/m. This means Chunchucmil groundwater is hard and has a high dissolved ion load that can reduce maize yields up to 35 percent if exclusively applied (Luzzadder-Beach, 2000). In areas at the higher EC range, water quality would have limited maize agriculture near Chunchucmil, but we did observe farmers irrigating healthy crops with this water. Moreover, what little we can say from the few bone data indicates Chunchucmil relied less on maize than Yaxuná and sites in the Petén and Belize (Mansell et al. 2006).

The typical model we use for swidden agriculture indicates maize cultivation can produce one crop per year for about two to three years. Several studies have attributed the short cultivation period to the diminution of crop yield from decreased organic matter, soil moisture, and nutrient availability, and to increased weed competition (Emerson 1953; Cowgill 1960; Reina 1967; Reina and Hill 1980; Beach 1998a; Weisbach et al. 2002; Dahlin et al. 2005; Dalle and de Blois 2006).

During fallow, secondary growth of shrubs, weeds, vines, grasses, and young trees quickly overtakes the milpa and allows soil organic matter and nutrient renewal (Emerson 1953; Dalle and de Blois 2006). Although in a wetter part of the Maya world, research on Lakandon Maya swidden suggests a highly complicated system with multiple uses and management of the fallow and longer planting periods (Nigh and Diemont 2013). But the time required for fallowing depends on the type of soil and the length of time of cultivation (Weisbach et al. 2002). Historically, the system requires seven to 15 years of fallow for every two to three years of cultivation in the Maya lowlands, a need of two to seven times more land in fallow than in cultivation (Reina, 1967; Reina and Hill, 1980). Weisbach et al. (2002) concluded that for northwest Yucatán a 12-year fallow restored most of the nutrient status, but recommended a 25-year fallow to have the most significant improvement. Recent demand for increased crop production has dictated that the recommended fallow time be cut in half or more, requiring about a 10-year fallow (Dahlin et al.2005). Nigh and Diemont (2013) suggest careful management could increase the cropping to eight-year fallow cycles in an area with greater soil resources.

1990s Soil Transects

For more than a decade from 1994, we sampled soils across this broad landscape from the current coastline at Celestún eastward to the Sierrita de Ticul, along the edge of the Puuc hills, trying to capture the diversity of soilscapes in the region’s six vegetation zones (figure 9.1): beach ridge, swamp-estuary, peten, tzekel, savanna, and karst plain (Farrell et al. 1996; Beach 1998a; Dahlin et al. 1998). Vlcek (1978) characterized much of the region as having little or no soil. In 1994, Bruce Dahlin led a field trip across soil landscapes from the coast to the Sierrita de Ticul. Tim Beach, Sheryl Luzzadder-Beach, Pat Farrell, and Bruce Dahlin followed this up with soil pits in a typology of soil landscapes (Farrell et al.1996). The goal of this work was to find the regional resource base and soil diversity. Our work started on the Celestún peninsula because Dahlin had connected Chunchucmil with this broader hinterland and the sea, including the port site of Canbalam. We interpreted that the Celestún Peninsula’s berms and swales had accreted sometime after sea level rise stabilized about 5,000 years ago and may have produced a soil chronosequence of older soils farthest inland to younger ones progressively closer to the Gulf of Mexico. We also reasoned that these soils held some evidence for past usage and found some elements of anthrosols in one of the soils (Dahlin et al. 1998). Indeed, there was the possibility of agricultural soils in the central zone of the peninsula, less influenced by salt water and more by fresh groundwater.

Figure 9.1. (a) Landscape elevation model from the Gulf Coast to the Sierrita de Ticul and (b) vegetation zones traversed by that transect.

Along the coast are beach ridges along with swamp and estuary lands with petenes or freshwater springs (figure 9.1). The beach ridges and swales have high concentrations of surface salts, as much as 14 dS/m in one swale soil, because of a high water table, making swales unfit for cultivation, though the central part of the peninsula may have had lower salinity in antiquity with the lower sea levels. The soils of the swamp/estuary zone vary between histosols, inceptisols, and entisols, and the swale zones are often covered with periphyton (algal detritus and minerals that accumulate in shallow water) (Beach 1998a; Dahlin et al.2005; Sedov et al.2007). Some have argued ancient Maya farmers used periphyton for fertilizers, which is plausible, given its high nutrient levels and organic matter (Sedov et al.2007). We noted in our work that indeed Maya farmers today borrow organic matter from wetlands (Beach 1998a), which could well have meant collection of periphyton. This hypothesis still needs testing at Chunchucmil, potentially through biomarkers, stable isotopes, and fossil shells (Dahlin et al. 2005). Morrison and Cozatl-Manzano (2003) did the latter at Yalahau and found some evidence for allochthonous shells, perhaps indicating human transplantation.

The beach-ridge soils formed along a soil toposequence across each ridge and chronosequence from younger to older soils across the more than 120 beach ridges and swales that span the nearly 3-km-wide Celestún Peninsula (Beach 1998a; Dahlin et al. 1998). Because these beach ridges and swale complexes prograded since the middle Holocene to form the recurved peninsular spit, we assume the soils get younger westward from the Ria Celestún. A survey of 11 berms from the coast to the middle of the peninsula found increased melanization across this 1,047-m-long transect (figure 9.1). These soils are shelly, calcareous sand to fine sandy loam, and classify as aquents, psamments, and aquepts (Dahlin et al. 1998; Beach 1998a).

We found very little evidence for modern agriculture in this zone, and indeed local informants say most people mainly used the area as a source of organic matter to borrow for their gardens (Beach 1998a). We found one significantly more mature soil in the landscape near Hacienda Real de Salinas (see figure 1.2). A 300-m-long road-cut exposure near here had a well-developed paleosol buried by 89 cm of eolian deposition, which in turn had evidence of two episodes of topsoil formation at 47 cm and at the surface. The surface A horizon (from 0 to 20 cm) was faintly melanized with pale brown (10YR 6/3) silt loam on a C horizon (to 47 cm) of lighter (10YR 7/2) silt loam, which in turn lies above a granular Ab horizon (48–55 cm) that is more melanized, brown (10YR 5/3) and built upon a 2C horizon (to 89 cm) that is a very pale brown (10YR 7/3) fine sandy loam. The buried soil below this had a thicker, more granular, darker (10YR 3/2) A horizon (89–100 cm) built on columnar, very pale brown (10YR 7/3) Bw horizon. These buried soils represent the oldest soils we studied on the peninsula because of their well-developed A and Bw horizons, perhaps columnar because of high quantities of sodium.

Soil Transects in the 2000s

We base most of the findings and discussion below on research conducted in the 2000s. The bulk of these findings came from three approximately 2-km transects—northeast, east, and northwest—starting within the core area of Chunchucmil and collecting soil profiles every 100 m. We also sampled rural sites such as Bon and Nah Caña, two tertiary sites without site cores or temples, and at Pocholchen and Ikmil, medium secondary sites that possessed both site cores and pyramid structures (Hixson 2011). These sites are about 6 to 12 km west of ancient Chunchucmil through several different ecosystems and soil types.

Soil Classification

The Yucatec Maya developed a soil classification based on what they could see and feel such as color, texture, and stoniness (Dunning and Beach, 2004; Bautista et al., 2005). Sahkab lu’um or saklu’um (sak, “white”; lu’um, “earth or soil”), boxlu’um (box, “light black”), and kancab lu’um (kan, “yellowish”; cab, “reddish syrup”) are the three dominant soil types in ancient Chunchucmil’s sustaining area (Weisbach et al., 2002). Much like modern classification systems, there are subgroups, but given that we have classifications distant in time and space, we use these broad groups to describe the general sustaining area. In reality, the Chunchucmil region is a complex mosaic of shallow soils, buried soils, structures, and rock outcrops (figures 9.1a and b).

Soil development began in the Late Miocene to Pliocene (Pope, Ocampo, et al. 1996). With an average annual soil temperature of 29.4°C, the soil temperate regime is isomegathermic and the soil moisture regime is ustic (Van Wambeke 1987; Eswaran et al. 1997). Ustic moisture regimes are moisture limited except during a certain period, here the June to January wet season (Soil Survey Staff 2003).

Saklu’um consists of extremely shallow (3–17 cm), moderately well-drained calcareous soils over caliche or petrocalcic pavements. Saklu’um is grayish brown (2.5Y 5/2), sandy clay loam (endoaquent and petraquept) with effervescence, and slightly alkaline pH. These soils formed in sandy and loamy marine sediments from the Quaternary and Pliocene (Dahlin et al. 2005). Saklu’um is found in the swamp and in areas of the tzekel that are seasonally inundated. Clay content is on average 29 percent.

Kancab consists of shallow (8–50 cm), moderately well-drained soils over a caliche or petrocalcic horizon (Km) underlain by less dense, frail carbonate rock (sascab) (Beach 1998a). Kancab is reddish brown (5YR 4/4), clay loam (paleustalfs, paleustolls, and haplustalfs) that is noneffervescent with neutral pH. Clay content is 32–34 percent in the younger and shallower savanna kancab and increases to 34–36 percent in the karst plain.

Boxlu’um consists of extremely shallow (3-52 cm), well-drained soils over fractured limestone. Boxlu’um is black (10YR 2/1), skeletal, very gravelly clay loam (calciustolls, paleustolls, and haplustolls) with slightly alkaline pH. Boxlu’um is most commonly found on raised areas, as in the tzekel (tzekel lu’um [tzekel, “flat stone”]) or in areas of ancient limestone structures. Clay content is on average 28 percent; however, clay content in many of the boxlu’um soils of the tzekel is difficult to determine because of hydrous oxides and high organic matter. The very fine granular aggregates were hydrophobic and would not wet and disperse.

Field reconnaissance of densely occupied areas of ancient Chunchucmil showed that boxlu’um was present on the house mounds and platforms despite the presiding vegetative zone and surrounding kancab soils. Even in areas of boxlu’um of the tzekel, the boxlu’um within settlement structures differed in soil structure, color, and chemical properties from the boxlu’um outside settlement. The boxlu’um on the house mounds and platforms, hereafter called boxlu’um-o (o, occupied), developed after abandonment in lime plaster. Our surveys found no areas of intentional black-earth formation and we think the boxlu’um soils simply developed in the abandoned structures; hence ancient farmers would not have had boxlu’um-o to exploit in ancient times. Boxlu’um-u (u, unoccupied) developed before, during, and after occupation.

Soil Depth

The most distinctive features of the soils of Chunchucmil are their absence and thinness (Beach 1998a; Dahlin et al. 2005; Weisbach et al. 2002). Approximately 55–80 percent of the area has thin to no soil, and between 25 and 50 percent lacks any soil at all (Dahlin et al. 2005). With slopes of less than 1 percent, this area has the thinnest soils and is the most planar area of all of the Maya lowlands (Beach 1998a; Dahlin 2003). Often in the Maya lowlands shallow soils result from erosion (Beach, uzzadder-Beach, Cook et al. 2015, but the lack of slope here suggests slow soil erosion rates. Rather, Kellman and Tackaberry (1997) ascribe thin soils to the porous nature of the karst topography and slow soil development.

Curtis et al. (1996) and Beach (1998a) hypothesized that the present fertility and depth of the soils would not have been much different from the Maya occupation period because no substantial soil erosion previously existed, due to low gradient slopes and clayey soils. Yet, research in 2006 in modern quarries near Chunchucmil and conversations with Dr. Sergey Sedov (UNAM) about his work in northeastern Yucatán spurred us to reinterpret this hypothesis (Dunning and Beach 2010; Cabadas Báez et al. 2010). Although San José Chulchaca, the nearest coring site to Chunchucmil, shows no increase in sedimentation during the Maya Classic and slow sedimentation over all (Leyden et al. 1996), this site is 25 km north of Chunchucmil and erosion would be highly local in this flat terrain. Although the regional surface gradient is low, the gradient from the surfaces into sinkholes is locally steep over short distances. Over the years of the Chunchucmil projects, we measured deep sediments in many small sinks, and during the last years of the project we began to observe deep soil deposits in modern quarries near Chunchucmil. Soils in these cavities could have derived from slow erosion after limestone bedrock arose from the ocean or from periods of accelerated erosion either during devegetation from climate change or land-use change. We also observed that areas of so-called limestone pavements (similar to karren or lapis), which are all bare limestone karst features, cover some 25–50 percent of the landscape but that much of the stone surface of this zone also has deep-red staining indicative of oxidized iron in tropically weathered soil profiles. Hence, we hypothesize that the soil cover extended over these areas sometime in the past, and the most likely period of soil erosion was during the period of Maya land-use changes. Indeed, Cabadas Báez et al. (2010) have argued this from their findings from a series of quarries in northeastern Yucatán, based on several lines of evidence, including two AMS dates from soils buried in cavities from the Maya Classic. This is not a new idea, because two ublications (Conservation Foundation 1954; Robles Ramos 1950) suggested this much arlier(Beach et al. 2006). In sum, soils around Chunchucmil may have been deeper and thus potentially more fertile and better at storing moisture in antiquity (Dunning and Beach 2010). In this scenario ancient farming would have shifted sediment from the ground surface to cavities in the bedrock (compare with Fedick 2014).

The rate of soil formation for boxlu’um-o in and around Chunchucmil can be estimated since the site’s decline (i.e., post–ad 700; Dahlin, 2003). Average soil profile depth at Pochol Ch’en, located in the savanna (see chapter 8), for boxlu’um-u is 6 cm and for boxlu’um-o, 12 cm. Soil accumulation above a large platform floor at a profile in Nah Caña, also located in the savanna (figure 7.1), and above a patio floor in Chunchucmil at Profile NT12, was 6 and 11 cm deep, respectively (Sweetwood et al. 2009). The soils formed in Nah Caña and Chunchucmil during 1,100 years of abandonment at a rate of 0.05 mm/yr and 0.10 mm/yr. Other areas in the Maya lowlands, such as the Petexbatún (Beach 1998b) and Piedras Negras (Fernández et al. 2005), exhibit comparable rates: 0 to 0.11 mm/yr and 0 to 0.096 mm/yr, respectively. Johnson et al. (2007) calculated soil formation rates of 0.12–0.15 mm/year above a patio floor at Aguateca.

Where soil profiles were present, profile depth generally increased from 6 cm in an area 5 km west of Chunchucmil to approximately 29 cm deep in an area 3 km east of the site. The site center of Chunchucmil was dominated by thin boxlu’um-o with thicker soils at foot slopes and especially near areas of plaster mass wasting. Since much of the site center matches the depth of soil that would have formed at the calculated soil formation rate after abandonment, much of Chunchucmil was likely denuded of soil cover during the period of occupation. It is likely that patios and high-traffic areas were swept clean of soil but thin soils could have been purposely placed and maintained in garden areas (Beach 1998a). The denseness of structures and lack of soil likely limited central Chunchucmil to small home gardens, except for certain fruit trees that survive adequately in sparse soils (Hutson et al. 2007).

Soil Moisture

The farmers of Chunchucmil today prefer boxlu’um for agriculture (Dahlin et al. 2005), but other farmers in northwest Yucatán have stated that kancab is more productive (Weisbach et al.2002). These seemingly contradictory statements find validation in the strong link between soil moisture and nutrient availability. Low soil moisture reduces the mobility of nutrients and decreases plant uptake. Since boxlu’um tends to have low soil moisture and hydrophobic properties in the upper horizons, the available nutrients do not reflect the fertility of this soil class. Both boxlu’um and kancab hold moisture in their subhorizons longer through most dry seasons. This plays out in contemporary milpa management by modern Maya farmers because milperos tend to have multiple milpas in diverse locations to ensure crop success and decrease the probability of a disaster from variable amounts of rain (Nigh and Diemont 2013).

We can test water repellency (hydrophobicity) by using the water droplet penetration test (WDPT), which tests the amount of time water takes to penetrate the soil (King 981). For water repellency there are the following seven classes: Class 0, wettable, non–water repellent (infiltration within 5 s); Class 1, slightly water repellent (5–60 s); Class 2, strongly water repellent (60–600 s); Class 3, severely water repellent (600–3,600 s); and extremely water repellent (> 1 h), which is further subdivided into Class 4 (1–3 h), Class 5 (3–6 h), and Class 6 (> 6 h) (Dekker et al. 2001). Boxlu’um-o (1 s), kancab (0 s), and saklu’um (1 s) are wettable and non–water repellent. Boxlu’um-u is severely water repellent at 39 min or more for droplet penetration (Sweetwood et al.2009, 1,215–1,216).

Nearly half of boxlu’um-u soils, mainly developed in the tzekel, are hydrophobic. Hydrophobic or water-repellent soils have negligible water-holding capacity and are generally infertile. Water-repellent soils are seasonal. During the rainy season the hydrophobicity eventually can disappear, but if the soil is given time to dry out, the hydrophobicity can return (Quyum 2000). This is problematic for northwest Yucatán, since rain is variable, with a dry period in the middle of the rainy season, although fire management can reduce repellency and the land is so flat that water can eventually run off.

Quyum (2000) thought that water repellency occurred because hydrophobic soil organic matter (SOM) covers soil particles. Other factors associated with hydrophobicity are fungal growth, soil microorganisms, and plant type (Quyum 2000), but we have not found studies of the causes of hydrophobicity in northwest Yucatán.

Soil Fertility

Neutral to alkaline soils similar to those of Chunchucmil often exhibit deficiencies of potassium (P), iron ( Fe), manganese (Mn), boron (B), copper (Cu), and zinc (Zn). Nitrogen (N) may also be deficient, especially where there is insufficient fallow from the region’s legume-rich natural vegetation. About 30–50 percent of the trees are legumes (Rico-Gray et al.1988), which maintain healthy (low) C:N (carbon:nitrogen) ratios in undisturbed soil (Beach 1998a). After only one year of cultivation, however, total N is reduced by approximately 20 percent in soils of the Yucatán (Weisbach et al.2002). Soil N quickly becomes a limiting nutrient in the already nutrient-limited soil.

From 89 A-horizon samples, Sweetwood et al. (2009) compared the concentrations of extractable macronutrients (P, K) and micronutrients (Cu, Zn, Mn, Fe) to general fertility recommendations (Havlin et al., 2005). Boxlu’um surface samples (n = 36) had average P, K, Zn, Mn, and Fe concentrations of 13.9, 143.8, 1.3, 16.6, and 18.6 mg/kg, respectively, which are sufficient for plant growth. Average Cu concentrations of 0.6 mg/kg were marginal. Nine surface horizons of saklu’um had average chelate extractable micronutrient concentrations of 0.7, 6.8, and 33.6 mg/kg for Cu, Mn, and Fe, respectively, and these nutrients are also sufficient. Average P, K, and Zn concentrations of 11.1, 117.7, and 0.8 mg/kg, respectively, were marginal. Kancab samples (n = 44) had average concentrations of 0.7, 24.2, and 11.4 mg/kg for Cu, Mn, and Fe, respectively, and these nutrients are sufficient. Average concentrations of 0.8 and 84.5 mg/kg for Zn and K, respectively, were marginal and average concentrations of 6.4 mg/kg for P were deficient.

Although several macro- and micronutrients were greater in boxlu’um and saklu’um than kancab, concentration doesn’t account for quantity. Kancab of the area were on average 50 percent deeper than the other two soil types and therefore could potentially provide more plant nutrients. The effective root zone is critical for soil fertility. Under typical circumstances a maize root system will grow laterally 1 m in all directions and will penetrate the soil to depths of 2 m (Feldman 1994).

We also evaluated each Maya soil class with the land-capability classification system developed by the USDA (Klingebiel and Montgomery 1961). Kancab was in class III with severe limitations that reduce the choice of crops or require special cultural practices. The limitations included shallow depths to bedrock and low fertility. Farmers would need to amend these class III soils with soil organic matter (SOM) and they should not be worked when wet. Boxlu’um was in class IV with very severe limitations that restrict the choice of crops and require very careful management. There limitations were again shallowness and low moisture-holding capacity, and, in some places, salinity. Class IV soils in subhumid and semiarid areas may produce adequate yields during years of above-average rainfall, low yields during average rainfall, and failures during years of below-average rainfall. Fruit and ornamental trees and shrubs may be suitable for some class IV soils. Saklu’um soils were in class V with little to no erosion hazard but their use is limited to rangeland, woodland, wildlife, and watershed. Some limitations included ponded areas and nearly level stony soils.

Many of the physical and chemical properties of both boxlu’um-o and boxlu’um-u were significantly different (P < 0.05). Boxlu’um-o had greater values than boxlu’um-u for calcium carbonate (CaCO₃) equivalent, black carbon (BC), Cu, Mn, and Zn, and was strongly effervescent, whereas boxlu’um-u had greater levels of total N, total soil organic carbon (SOC, usually 58 percent of organic matter), P, electrical conductivty (EC), and sodium (Na), and was very slightly effervescent.

The greater values in boxlu’um-o of CaCO₃ equivalent (24%), BC (0.9 g BC/kg soil), and SOC (1.0% BC of SOC), Cu (0.8 mg/kg), Mn (20 mg/kg), and Zn (1.7 mg/kg) were 154, 43, 225, 2,392, 146, and 588 percent greater than boxlu’um-u, respectively, and could be explained by ancient human activities. Higher CaCO₃ equivalent resulted from the broken-up and weathered building materials and stucco.

Boxlu’um-u had nearly 100 percent more exchangeable Na (17.4 mg/kg), total SOC (23.2%), total N (1.8%), and P (22.9 mg/kg), and 50 percent higher EC (1.5 dS/m) than boxlu’um-o. The greater values of total N, total SOC, and P in boxlu’um-u are explained by interactions of soil chemical and physical properties. In general, the increased concentrations of both P and N in the soils were significantly related to increased levels of SOC (P < 0.05). Retention of SOC is often attributed to clay content, base saturation, the chemistry of the soil organic matter (SOM), and microbial activity rates (Oades 1988).

Higher EC and Na in boxlu’um-u can be explained by depth to water table. Pochol Ch’en is in the tzekel zone, and in the encompassing area of boxlu’um-u the water table was visible in large fractures of the bedrock at depths of approximately 10 to 15 cm from the soil surface during the dry season. Close proximity allows wicking of groundwater and deposition of salts. The water table was not visible in the site of Pochol Ch’en. Soil profiles revealed fill for ancient patio groups, which increased depth to the water table and possibly reduced upward soil water movement.

Intensive Agriculture

To increase yields and shorten fallow time for agricultural self-sufficiency, Chunchucmil would have needed large inputs of plant-essential nutrients and SOM. Agricultural intensification like soil importation, soil amendments, and fertilizing with organic amendments over centuries would have left an imprint on the soils of Chunchucmil. Intensive agriculture could have elevated or decreased certain soil properties or chemical residues above natural background concentrations, such as ¹³C isotope, biomarkers, P, and BC. On the other hand, if the farmers at Chunchucmil maintained high-input agriculture for centuries and also had high harvests, and possibly higher harvests toward their abandonment, little or no increase of P might be evident. This would be especially difficult to identify with the overprint of subsequent land uses, thin and possibly eroded soils, and highly mixed, bioturbated soils.

Carbon Isotopes

One way to delineate probable areas of ancient agriculture is through carbon isotopic ratios (¹³C/¹²C), and several studies have used these at other sites in the Maya lowlands (Beach et al. 2011 Beach, Luzzadder-Beach, Cook, et al 2015 Beach, Luzzadder-Beach, Guderjan, and Kraus 2015 Burnett et al. 2012a, 2012b; Fernández et al. 2002; Fernández et al. 2005; Lane et al. 2008; Webb et al. 2004, 2007; Wright et al. 2009). Ancient long-term maize cultivation leaves a distinct isotopic signature in the SOM. A maize C₄ signature is formed when long-term cultivation of maize, a C₄ grass, takes place in a normally C₃ vegetative region, which leaves a δ¹³C-enriched horizon.

We ran stable carbon isotope analyses of soil profiles on our soil survey transects and on a typology of ancient land uses around Chunchucmil. For the transects, the δ¹³C values of surface A horizons varied significantly (P < 0.01) according to soil type and vegetation zone. Nearest the Gulf of Mexico is the swamp/estuary zone with highly organic soil profiles. Surface horizons of these soils had average δ¹³C values of -27.18‰, which indicates that this zone is dominated by C₃ vegetation. East of the swamp/estuary zone are the tzekel hillocks, which had average surface soil δ¹³C values of -25.44‰. This zone has mainly a high canopy with few grasses but enough C₄ vegetation to shift slightly from C₃. A small ancient rural site called Bon (see figure 7.1) with deep boxlu’um soils in the savanna had average surface horizon δ¹³C values of –23.67‰. Surface horizons of kancab in the karst plain and savanna were analyzed and had average values of -22.39‰. The decrease in discrimination of the ¹³C isotope across Chunchucmil’s landscape from west to east follows the change in vegetative zones and is an indication of increasing C₄ vegetation distribution near the site.

Next, we compared δ¹³C values of surface horizons from within structure groups of central Chunchucmil to control samples from 4 to 6 km north of Chunchucmil. Some historical depth came from buried surface-horizon samples beneath ancient structures. Surface soils from structure groups in central Chunchucmil had average δ¹³C values of -23.50‰, which was similar to that of boxlu’um and control samples with average δ¹³C values of -22.59‰. The buried A horizons under Classic structures had average δ¹³C values of -24.00‰. Statistically there were no differences between surface soils from structure groups, buried A horizons, and control samples (P = 0.90).

Of six grasses collected in the Chunchucmil region, four were C₄ and two were C₃. This mix of C₄ vegetation prevents the use of stable carbon isotopes to delineate zones of ancient maize agriculture in the savanna and karst plain of northwest Yucatán. The mixed C₃/C₄ vegetation produced humin with δ¹³C values similar to values in soil horizons of suspected ancient maize growth in a predominately C₃ vegetative region. In the shallow soils of northwest Yucatán, it would be impossible to differentiate between ancient milpas and native vegetation. Soil depth complicates the situation further because of a high rate of bioturbation and the inability to observe a change with depth. Soil samples are usually taken every 10 cm of depth for the carbon isotope analysis; however, average profile depths for boxlu’um, kancab, and saklu’um near Chunchucmil were 12, 21, and 10 cm, respectively. Even with the shallow soil of the tzekel, if long-term maize cultivation took place, then we would assume that average δ¹³C values would be similar to those of the savanna and karst plain. Instead, the δ¹³C values suggest that tzekels rarely grew maize.

We also ran stable carbon isotope analyses on and adjacent to several landscape types at Chunchucmil, including sascaberas, an ancient apsidal house base, a sacbe, a callejuela, modern solar, infields, savannas, reservoirs, the market plaza (Area D in figure 11.1), metates, and querns, which are metates with spillways (see chapter 2). Since the sample size is so small for these, we can only suggest further study. The savanna site, as in the transects above, produced the lowest, most C₃-dominated δ¹³C signature (-27.3‰), followed by the aguada sediments (-26.1‰), and the sascabera (-24.3‰). The highest δ¹³C signatures, closest to C₄ species, were a surface sample in the central market place (-20.5‰) and the three querns (-22.0‰). The market-place sample conceivably could have been influenced by ancient maize use or recent planting of tropical grasses for pasture (Beach, uzzadderBeach, Guderjan and Krause 2015. The metates were -23.1‰, which is similar to the querns but also the mean for the surface typology (-23.7‰). The one significant finding from this work was the mean of buried soils at Chunchucmil (-21.6‰ with 10 samples, versus the mean for surface samples, -23.7‰ for 23 samples). These may indicate more-intense maize cultivation in the recent past C₄ plants, or the processing and consumption of maize within the city.

For comparison with our Chunchucmil sites we evaluated samples from transects at the slope of the Sierrita de Ticul (20°22' N latitude) about 27 km southeast of Chunchucmil. These samples come from an area with few structures or artifacts. This area consisted of dry scrub forest, again mostly C₃ plant species but some grasses and cacti as well. Parts of these slopes may have been farmed and possibly stripped in antiquity, and indeed only the depression soils have deeper profiles with Bt horizons. The slopes in this region had an average δ¹³C of -24.7‰ overall (n = 35), -25.6‰ for A horizons (n = 24), and -22.8 for lower horizons (n = 11), which is nearly a 3‰ increase in these profiles. Most soils had only A and AC horizons, but three karst depression soils had deeper sequences. The 60-cm-deep soil profile that was designated the North transect (125 m) had a change in δ¹³C from -26.3‰ in its upper A horizon to -22.3‰ and -21.8 ‰ at the 25-cm and 45-cm depths, respectively. This is an increase of 4.5‰, reflecting ancient C₄ vegetation and possible maize agriculture in this deep kancab soil. The profile designated the East transect (25 m) had a similar pattern with a surface δ¹³C of -26‰ that increased to -22.7‰ at 25 cm, -22.3‰ at 45 cm, and -22.6 ‰ at 75 cm. The soil profile at 2W had -25.3‰ in the upper A and -22.3‰ at 30 cm in the B horizon. Both of these soils have δ¹³C increases of more than 3‰, which provides moderate evidence of the impacts of ancient C₄ plants. This region produced similar results to those in a similar sloping forested area in Belize, with mostly C₃ δ¹³C signatures except for two deeper-depression soils that also had significant δ¹³C enrichment in the lower soil (Beach, uzzadderBeach, Guderjan and Krause 2015 The increase in δ¹³C within the deep soils was comparable to the increases within the buried soils at Chunchucmil, which in both cases may mean more maize or C₄ species in antiquity, due to either drier climate or anthropogenic impacts.

Black Carbon

Large regions in Amazonia and some in Africa have terra preta soils, which have copious amounts of charcoal, or black carbon (BC), such that they are still highly fertile today (Woods et al. 2009; Costa et al. 2004; Schaefer et al. 2004). If the Chunchucmil Maya had used charcoal to amend soils, there would be differences in BC concentrations from unoccupied and occupied areas with higher levels of BC in the ancient fields.

BC, a product of incomplete combustion (Brodowski et al. 2005), is almost entirely made up of aromatic C (Schmidt and Noack 2000) that resists chemical and microbial decomposition and persists through geological time-scales (Taylor et al. 1998; Glaser and Amelung 2003; Glaser et al. 2001a; Dai et al. 2005). The accumulation of BC is related to climate, textural properties, concentration of SOM, and soil moisture (Glaser and Amelung 2003). BC enhances soil fertility by increasing the soil nutrient-holding capacity (Glaser and Amelung 2003; Glaser et al. 2001a), which has greatly improved crop yields of infertile Amazonian soils (Glaser et al. 2001a).

There is no doubt that the Maya produced charcoal, but little shows up. Three hypotheses might explain this absence:

We made a preliminary test of these hypotheses by mapping BC with respect to distance from settlement. If long-term soil amending occurred in milpas, we should observe elevated concentrations in unoccupied areas. Soil profiles were categorized as off-mound (no ancient structures within ~20 m), near-mound (within 20 m of ancient structures), and on-mound. We assigned a numerical value to each category and then compared their BC concentrations. Soil profiles from rural sites of Ikmil, Pocholchen, and Nah Caña were analyzed.

One transect centered over the site center of Ikmil, a large secondary site, reached to the unoccupied areas west and east of the site. A regression analysis of position versus BC concentrations shows that there is a significant correlation with proximity to ancient structures (P = 0.00, R² = 0.59). BC concentrations increased from off-mound (0.62 g BC/kg soil), to near-mound (0.78 g BC/kg soil), and then to on-mound (1.1 g BC/kg soil). BC concentrations also increased from off-mound (0.61 g BC/kg soil) to near-mound (0.90 g BC/kg soil) in Pocholchen (P = 0.01, R² = 0.45); no on-mound samples were taken.

The positive gradient of BC concentrations toward ancient structures suggests an incidental effect of ancient human activities. Cooking fires and charcoal incidental to the burning of old thatch and to stucco production may have been major sources of BC in near-mound and on-mound soils.

Sweetwood et al. (2009) used the BC digestion and analysis method of Glaser et al. (2001b) to determine that surface soils surrounding Chunchucmil contained between 0.37 and 1.37 g BC/kg soil. These Yucatán soil results were about an order of a magnitude lower in the surface horizon compared to the terra preta soils of the Brazilian Amazon region (~11 g BC/kg soil). Even the BC concentrations of the control samples surrounding the terra preta soils were approximately twice as high as BC concentrations at Chunchucmil. There was also no significant differences between the BC contents of boxlu’um-u (0.7 g BC/kg soil), kancab (0.5 g BC/kg soil), and saklu’um (0.6 g BC/kg soil) soils (P = 0.47). The even distribution of BC throughout unoccupied rural Chunchucmil suggests that the major source of natural BC has been rotating milpa and natural fires.

The source of the dark color of boxlu’um is likely related to the retention of SOM rather than to BC. Average organic C contents for boxlu’um, saklu’um, and kancab were 15.1, 8.8, and 6.4 percent, respectively. Of the soil properties analyzed, as the exchangeable multivalent cations Ca and Mg (P = 0.00, R² = 0.55) and clay content (P = 0.00, R² = 0.44) increased, SOM also increased. One mechanism of organic matter retention is cation bridging between clays and organic colloids (Oades 1988). The accumulation of SOM through introduced multivalent cations may explain the islands of dark brown soil of anciently occupied areas among the reddish brown soils of the savanna and karst plain. The dissolution of broken-up limestone from the construction of patio groups, the stucco used by the ancient Maya, and the lime used for food preparation were the major sources for elevated Ca and Mg (Fernández et al.2002). The long-term liming effects of the stucco and other construction materials has apparently enhanced the accumulation of SOM content of house-mound soils (Oades 1988; Beach 1998a).

It is common to report BC as a proportion of SOC, because it helps describe factors of BC accumulation (Dia et al. 2005). Terra preta soils of Amazonia have up to 35 percent BC as a proportion of the SOC (Glaser et al. 2001a). In contrast, boxlu’um, kancab, and saklu’um had much lower values of 0.71, 2.35, and 0.57 percent BC of SOC, respectively, but were significantly different between each soil class (P = 0.00). Kancab generally had greater soil moisture than saklu’um and boxlu’um during the dry season. Clay content was also greatest in kancab. These two factors tend to play a role in BC accumulation (Glaser and Amelung2003). These three soils also represent three very different vegetative zones, which may greatly differ in the quantity and type of plant material, and frequency of burning.

Phosphorus Concentrations and Biomarkers

Because so many people lived in relatively tight quarters at Chunchucmil and they had few fertilizer choices, it seems likely that farmers used night soils. To test this, we analyzed stanol biomarkers and P. An enrichment of one or both properties should appear in areas of ancient croplands if amended with fecal residues (Fernández et al. 2002), though we also recognize that coprostanol would likely decompose and P was in short supply in this environment. Coprostanol is formed in the intestinal tract of most mammals, and has considerable potential as an indicator of ancient manuring and night soiling (Bull et al. 1999). Hutson et al. (2007) and Sweetwood et al. (2009) examined 10 surface samples from both contemporary and ancient houselots at Chunchucmil for stanol biomarkers that could indicate ancient human waste disposal and manuring.

Extractable Soil P concentrations correlated with the change in vegetation and with densely populated regions, but there existed no anomalies of elevated P above normal background concentrations in potential outfield areas. Soil P concentrations were naturally elevated in the swamp/estuary (9.3–14 mg/kg) and tzekel (14.1–22.3 mg/kg) and then declined in the savanna (5.7–6.4 mg/kg) and karst plain (6.5–7.2 mg/kg). Of the 104 soil profiles collected, the range of soil P was 2–46 mg/kg. Soil P concentrations found in middens and suspected marketplaces in central Chunchucmil reached concentrations upwards of 250 mg/kg (Dahlin et al., 2007). There is no evidence of increased accumulation of P above background levels that would suggest the ancient Maya performed widespread night soiling.

Geostatistical Analyses

Dense settlement of ancient Chunchucmil left an imprint of both physical and chemical properties. This is most notable when observing selected soil properties mapped over part of Chunchucmil’s sustaining area (52 km²) using geospatial analysis in ArcMap^®. Along with soil P, we explored K (Olsen method), trace elements, Cu, Mn, Zn, and Fe (Diethylenetriamine-pentaaceticacid DTPA method), exchangeable ions, Ca, Mg, Na, and K, and several other physical and chemical soil properties as possible indicators of human activity in occupied areas and land usage in unoccupied areas. Two separate methods (Olsen and DTPA extractable) allowed us to compare the effectiveness of K to indicate ancient activity.

The urban outline of Chunchucmil coincided with concentration isopleths of each soil chemical property, and intensity of these concentrations generally increased toward the center of the site. For example, centered over Chunchucmil was an elevated island of P (7.3–9.2 mg/kg), K (Olsen; 116–262 mg/kg), SOC (8.4–13.0%), and Mg (26–63 mg/kg).

Exchangeable Ca was also elevated in Chunchucmil, greater than 561 mg/kg, and background concentrations decreased gradually from east to west, from 538mg/kg in the karst plain to 489 mg/kg in the swamp/estuary. Conversely, percent CaCO₃ equivalent decreased from west to east of the site, from greater than 40 percent in the swamp/estuary to 4–11 percent in the karst plain. Ikmil and Chunchucmil had slightly elevated CaCO₃ equivalent, but the contrast from background levels is not as pronounced as exchangeable Ca.

There are probably many reasons for elevated concentrations of P, K (Olsen), SOC, Mg, and Ca in central Chunchucmil. Soil P and K initially accumulated after centuries of discarded food and waste. Increased SOM, likely caused by increased polyvalent cations from broken-up limestone and stucco, stimulated the retention of additional P and K.

Exchangeable K (DTPA) gradually increased from west (13.4 mg/kg) to east (31.9 mg/kg) but did not share the same patterns as K (Olsen). Fernández et al. (2002) used exchangeable K in soils from a modern Maya house lot and discovered that exchangeable K was elevated in food-preparation areas beneath a thatched roof. From an abandoned house lot with three years of exposure to weather, exchangeable K was slowly leached and concentrations were only slightly elevated above background levels. Thus after a millennium of disuse at Chunchucmil, we cannot measure ancient human activity with exchangeable K. Ancient human activity is illegible with DTPA extractable K but K (Olsen) may be a more efficient indicator of ancient human activity within settlement for this area.

The isopleth maps of extractable Fe and Cu did not follow vegetation change as well as other soil properties because of high variability. Even with the greater variation, some patterns emerged. In general, there were elevated concentrations of both Fe and Cu in the swamp/estuary (40–52 mg/kg and 1.0–1.7 mg/kg, respectively) and mildly levatedconcentratons n central Chunchucmil (23–34 mg/kg and 0.7–1.7 mg/kg, respectively), and background concentrations were 1–23 mg/kg and 0.4–0.7 mg/kg, respectively

Soil concentrations of DTPA extractable Zn exhibited a peculiar pattern. Concentration gradients were high in the swamp/estuary zone (1.1–2.9 mg/kg) and low (0.4–0.9 mg/kg) in the tzekel, savanna, and karst plain except for two locations. Concentrations were high (1.1–2.9 mg/kg) in between Ikmil and Chunchucmil and on the northeast periphery of Chunchucmil.

Soil concentrations of Mn were relatively even throughout the mapped region, between 7 and 19 mg/kg, except northeast of Chunchucmil, where concentrations rise sharply to 31 to 44 mg/kg. Although Linderholm and Lundberg (1994) connected Mn, Zn, Fe, and Cu with ancient human activity, the anomalies here are more likely inconsistencies in parent material and/or an increased cation exchange capacity (CEC).

Maya soils often retain the footprint of society in the physical and chemical properties of each site, despite nutrient cycling or minimal post-ancient Maya cultivation. A lack of some chemical such as phosphorous in cultivable land may suggest, but does not prove, the lack of agricultural intensification.

Spatial Analysis

Land settlement patterns can illustrate preferences of agricultural resources (Fedick 1995). Based on vegetation, we could predict to which potential Maya soil class each site pertained. The majority of secondary sites, 21 of 24, were in kancab, two were in boxlu’um, and one was in saklu’um. Tertiary sites exhibited a similar pattern with 11 sites in kancab, two in boxlu’um, and 3 in saklu’um. The majority of the rural sites on the karst plain lie on kancab soils.

If we assume the major occupation for the rural population was agriculture and settlement location was in close proximity to milpas, then the ancient Maya preferred cultivating in kancab north and east of Chunchucmil. The soil east of Chunchucmil is deeper, has a slightly better capability class, and is laterally more continuous than the savanna or tzekel.

The sparse ancient settlement in the tzekel and swamp/estuary with their shallow soils confirmed that these areas were not preferred for cultivation; rather, the ancient Maya probably used the tzekel zone for wood and other forest products and for hunting and gathering. It would have been better to use the tzekel zone for certain economic species, like agave, nopal, and fruit trees, that do not require deep soils (Hutson et al. 2007). Nearly all secondary and tertiary sites in the savanna lie on the edge of the tzekel in an ecotone between cultivable land to the east and hunting and gathering land to the west.

Rural settlement and land use is an issue of interest for many geographers (Chisholm 1979). Research from all over the world of prehistoric and historic land use has shown that agricultural activity occurs concentrated within a 1-to-2-km radius from settlement and beyond, and activities decline with distance and often terminate at around 5 km (Stone 1991).

Modern Maya milperos follow a similar trend and generally choose locations for cultivation based on location, soil type, and distance to milpa (Reina 1967). To minimize movement costs, these milperos live near their milpas and arrange them so that they spend no more than an hour on the trail traveling between each milpa (Reina 1967). With a radius of 5 km, the area of cultivable land surrounding all known ancient settlement in the savanna and karst plain at Chunchucmil would be 445 km², below the initially proposed sustaining area of 600 km² (Beach 1998a).

At optimum crop yields and shortest fallow, this area would only sustain 22,250 persons using Conklin’s (1957) equation. For the 42,400 Maya in the core area of ancient Chunchucmil, the land requirement using the highest crop yields for this region and lowest fallow cycle would have been 848 km². This estimate does not include areas with no soil cover, which would raise the estimate to over 1,000 km². The enormous land requirement for just the core area means that a milpero would have been required to walk as much as 25 km from Chunchucmil if agriculture solely took place in the savanna and karst plain. Even without suburban and rural population estimates, it is improbable that the ancient Maya traveled this great distance to cultivate. Known rural settlement only extended as much as 13 km away from Chunchucmil. But even after more than 20 years from the start of the Chunchucmil project, the great unknowns for subsistence remain possibly unnoticed evidence for agricultural intensification and of deeper soils that would have had more fertility and greater water-storage capacity.

Conclusions

The many lines of evidence used to assess the agricultural resources surrounding Chunchucmil make us question its agricultural self-sufficiency. It is reasonable to conclude that poor building materials, shallow rocky soils, low fertility, variable rains, seasonal inundation, and water-repellent soils would deter any sustained large and dense population as it does today. Historic agricultural yields using traditional methods could not have supported the ancient population during Chunchucmil’s major period of occupation. But we also recognize that past farmers could have used any of a number of intensification techniques to increase crop production. We also hypothesize that the ancient Maya soil environment might have been one more conducive to crop production, which may have been eroded some time during the Maya period and might be a part of the answer to the riddle of Chunchucmil’s subsistence and even to its decline in the Late Classic period. These are testable hypotheses, which we hope to test with further study of quarries, dating of buried soils, and estimating how long bare surfaces were exposed by cosmogenic nuclide dating using ³⁶Cl (Matsushi et al. 2010).

Of the three dominant Maya soil classes, we found that kancab was the most consistently cultivable soil in Chunchucmil’s sustaining area. Saklu’um had high salts, level stony soils, and ponding, and is unsuitable for cultivation. Boxlu’um-u had greater concentrations of nutrients for crop growth than kancab, but the often hydrophobic SOM, low soil moisture, and shallow depth negate the higher concentrations, especially when precipitation is already low or variable. Besides the fact that kancab covers a greater region, kancab must have been agriculturally important for the ancient Maya because it provided some security with higher soil moisture, greater soil depth, and improved nutrient transportation. There still exist complications with some areas with kancab, mainly ponding that can hinder crop development (Beach 1998a). Where we could excavate deeply enough, soils are deeper and bedrock had more and larger fractures east of Chunchucmil, away from where case hardening inhibits water and gas movements through soils. This may explain why most of the rural settlement is east of Chunchucmil.

Carbon isotopic signatures of ancient maize agriculture proved unsuccessful in delineating agricultural soils of the area surrounding Chunchucmil. Shallow soils and native and introduced vegetation of C₄ and C₃ plants mask the isotopic signature of maize.

The land-use capability of the karst plain with kancab was ranked as more favorable than all other main soil types. The land capability has severe limitations in the swamp/estuary and tzekel. The lack of rural settlement within these zones suggests little ancient use for cultivation.

We found no evidence of agricultural intensification of Chunchucmil soils by night soiling and soil amendments with charcoal. The stanol biomarkers likely decomposed quickly in the warm, seasonally wet environment, if they existed, and soil P concentrations in unoccupied areas did not exhibit patterns or concentrations of long-term night soiling. BC (g BC/kg soil) levels were low in comparison to the terra preta soils of Amazonia. We found a few incidentally elevated concentrations of BC on ancient structures and within settlement but not in the cultivable land surrounding each site. Thus we can infer that the Maya of Chunchucmil did not amend their soils with charcoal.

The distributions of soil physical and chemical properties should have a buildup of chemical residues or altered physical properties if intensive agricultural occurred; however, we found little evidence in the surrounding landscape. The traditional method of shifting cultivation leaves little input of any source and the distribution of soil physical and chemical properties should resemble those observed in Chunchucmil. Based on this, ancient Maya agricultural practices at Chunchucmil were likely shifting cultivation, orchards, and solares, but we also recognize that eroded soil could hold more evidence for intensification.

The ancient Maya of Chunchucmil during the Middle Classic (ad 400–700) have yet to fully reveal their secrets of how they fed themselves. We could find no evidence that the ancient habitants of Chunchucmil used anything other than traditional methods. Atypical crops were possible as an alternative for maize but they would have been subjected to the same poor soil conditions and the same natural hazards like flooding and drought that plague the northwest Yucatán. Based on these findings, it seems more likely that Chunchucmil traded perishable goods to places like the nearby (30 km) Puuc hills. After all, Chunchucmil lay between the agriculturally rich Puuc and the maritime and estuary resource-rich Canbalam (27 km west), which was a stop on one of Mesoamerica’s major maritime trade routes (Dahlin et al., 1998) for a host of marine and estuary products.

When we started the Chunchucmil project, we planned to study soil, sea levels, world systems connections, water tables, climate changes, and evidence for craft production. We and the broader scientific community made progress on these topics, but there is much left to future research. To answer the question of subsistence at Chunchucmil will require scholars to identify its coastal and interior connections better by characterizing its connections with Canbalam and the Puuc, its world systems networks with distant cities like Teotihuacan, its sea-level contexts over time, and what evidence exists for deeper past soils and their proxies for past soil uses.