Archeomalacology in Texas: Western Sites

Ken Brown

In the June, 2018 issue of this newsletter, I reviewed some of the reasoning and methods that underpin paleoenvironmental research involving snails, and I pointed out that the southwestern quadrant of Texas (basically, west of longitude 100º W and south of latitude 32ºN) is almost a blank slate for archeomalacologists. We know that the better-watered eastern half of the state often produces abundant and informative snail faunas from sites in alluvial deposits, but what about sites in dryland alluvial settings? Dryland floodplains often lack fringing gallery forests and are essentially open habitats with high insolation, constant exposure to wind, and a conspicuous lack of leaf litter from deciduous tree canopy. Trees, if present at all, are likely to be represented only by scrub brush. Woody species in desert settings tend to be small-leaved, because high insolation makes photosynthesis very efficient, so leaf litter is often scarce or absent. Ground cover may be sparse, unless there is good grass cover in wet years, or rock rubble in sloping areas. Conditions like these are challenging for snails, which are highly moisture-sensitive.

One of the earliest studies in the state was done by Cheatum (1966) as part of the Amistad paleoecological project at Eagle Cave, Bonfire Shelter, Devil’s Mouth, and Devil’s Rockshelter, but the  number of samples, volumetric size of samples, and number of specimens are not disclosed and  intrasite provenience is reported only by stratum. In the entire vast area of the Big Bend (more than 12,300 square miles, not counting parts of Culberson, Hudspeth, and Reeves counties), no formal archeomalacological studies have been done until recently, as far as I know.

The Genevieve Lykes Duncan site (41 BS 2615, Early Holocene to present, Brewster County)

Figure 1. Looking north at dawn, up Green Valley on the O2 Ranch. The Genevieve Lykes Duncan site sits at the foot of the ridge system (andesite and basalt) in the distance.

This site lies 54 km south of Alpine, toward the western side of the sprawling O2 Ranch, a 272,000-acre ranch that actually sprawls over into Presidio County (Fig. 1). This is the eastern part of the Basin and Range province, and the site sits at the north end of Green Valley, on an interfluve between Terlingua Creek and Davenport Draw. Both streams have contributed sediment in the past, and the deepest component consists of stratified Late Paleoindian (Early Holocene) occupations: sparse artifact scatters associated with rock-lined hearths or small earth ovens.  An earlier occupation at about 10,400-11,080 cal BP lies about 2.3 to 2.5 m below the ground surface, with two rock-lined features, and five smaller rock-lined features dating about 8630-9535 cal BP are somewhat later. There are also two undated (but presumably Late Paleoindian) features that lack charcoal. As of 2016, at least 11 features had been found exposed in the walls of the recent arroyo that revealed the site, over a distance of about 24 meters (since then, three more have been found). There are also a few Middle and Late Holocene features exposed in more distant parts of the arroyo (Cloud et al. 2016:Fig. 6).

An extensive series of radiocarbon assays (chiefly on saltbush, mesquite, and creosote charcoal) ranging from 7934±25 to 9545±25 RCYBP places these Late Paleoindian occupations in the Early Holocene; Cloud et al. 2016:Table 2). These early occupations are buried by Allostratigraphic Unit 1, a thick deposit of clay loam that began accumulating sometime before 10,730 RCYBP and continued until the Middle Holocene. The Paleoindian deposits lie in a dark paleosol bracketed by two sheet gravel deposits (the lower one probably laid down by Terlingua Creek and the upper one by Davenport Draw; Cloud et al 2016:27). Allostratigraphic Units 2 (Middle to Late Holocene) and 3 (Late Holocene) overlie this unit. There is a series of global post-Younger Dryas cold events that punctuate the Early Holocene, but for the most part they do not seem to correspond to the Paleoindian occupations here, which follow the Preboreal Oscillation and precede the well-known 8200 cal cold event.

In July, 2013, I visited the site for several days with the auspices of the Center for Big Bend Studies, and with the assistance of Sam Cason (who did most of the work), collected a continuous column of matrix samples 60 cm wide and 2.74 m high from the southeast wall of the arroyo, immediately adjacent to the main excavation block (Fig. 2). The original plan was to collect samples in 5 cm increments, but the clay and silt-rich alluvium was so thoroughly indurated that this proved impossible, so the sampling interval was increased to 10 cm. Altogether 30 samples were collected (there are two extra samples because intervals were subdivided where stratigraphic breaks were crossed). The lowest sample cut only 10 cm into the lower gravel deposit (Fig. 2). This level is unassayed, but Andy Cloud (personal communication) estimates it dates to about 10,200 RCYBP, based on assays from the adjacent excavation block. Thus, the sample column documents the entire span of the Holocene (including the Late Paleoindian component), but presumably includes no Younger Dryas sediments.

Figure 2. Looking southeast at the completed snail sample column. The top portion looks dark simply because it is recessed farther into the arroyo wall. The tag to the right of the “paleosol” label marks an artifact. The sample column penetrated the lower gravel only partway. The main excavation block is out of view to the right.

In all, 465.6 kg (about half a ton) of sediment was collected. For the entire column, the mean sample size processed was 14.3 liters (range, 13-15 liters) and the mean analyzed weight was 15.05 kg; the total weight of sediment used in the analysis was 451 kg (a few samples had excess amounts). Samples were soaked overnight in tapwater, then passed through nested 18-inch geologic sieves with mesh sizes of 2 mm, 1 mm, and 0.5 mm. Residue from each grade was dried and bagged for picking. All snails, snail shell fragments, charcoal, seeds, animal bones, and possible microdebitage were saved and counted or weighed. After picking, the remaining residue was turned over to Brittney Gregory for heavy mineral separation studies at LSU.

Despite the fact that nearly half a ton of sediment was processed, only 843 specimens (many of them damaged or fragmentary) were recovered. This amounts to a density of only about two specimens per liter of sediment, compared to about 10-130 per liter from other sites in the eastern half of Texas where the same methods have been used. There are only four kinds of terrestrial snails: Succineidae, Gastrocopta pellucida, Helicodiscus singleyanus (= Lucilla singleyana), and Pupoides albilabris. The Succineidae are unidentified. Although succineids are usually considered to be wetland fringe inhabitants, there are perhaps a couple of borderland species (Succinea luteola and S. solastra) that are found in arid upland areas. Thus, it is not clear whether the succineids indicate wet or dry conditions (but perhaps the latter). There are only two kinds of aquatic snails: a planorbid, believed to be juvenile Planorbella trivolvis, and Physa sp., probably Physa acuta. A few examples of a third aquatic species, Gyraulus parvus, was not found in the sample column but was recovered in wet-screening by the field crew.

All of these snails are resilient habitat generalists, and all appear to be very arid-tolerant species (although the identity of the succineids is uncertain). Almost all the Succineidae, the Helicodiscus, and the aquatic snails are juveniles, hinting at high juvenile mortality. Generally speaking, the aquatic taxa are tolerant of sluggish, poorly oxygenated, warm water with high solute levels. All of the aquatic snails are confined to the bottom half of the column, disappearing in the upper gravel, just before the midpoint of the Holocene; otherwise, the same kinds of snails are found throughout the column, although fluctuating in abundance.

The Genevieve Lykes Duncan snail assemblage is characterized by low specimen density, low taxonomic diversity, apparently high juvenile mortality for many taxa, and a preponderance of arid-tolerant, resilient habitat generalists. When I began sampling this site, I expected to find lingering evidence of greater moisture conditions from the Younger Dryas at the base of the column. My expectation was that despite mounting aridification in the Early Holocene, one or two samples at the base of the column would yield greater numbers of snails, more diversity, and perhaps some taxa that are more moisture-dependant. This proved not to be the case. Instead, the snail assemblage suggests local conditions were already arid by the start of the Holocene. This is consistent with the saltbush, mesquite, and cholla identified in charcoal, and the Cheno-Ams, Artemisia, thistle, Asteraceae, and Ephedra identified in a pollen sample overlying Feature 1 (Cloud et al. 2016:55-56). My sampling stopped in the top of the lower gravel (a depositional event that might signal significant climatic reorganization), and if there is an assemblage adapted to wetter conditions, it must lie in the unsampled sediments under that gravel bed.


The Sayles Adobe site (41 VV 2239, Late Holocene, Val Verde County)

Figure 3. Looking down Eagle nest Canyon toward the Rio Grande. Steve Davis, Erwin Roemer, and Steve Black stand in Horse Trail Shelter contemplating my toolbox. Sayles Adobe lies hidden in green brush on the opposite side of the canyon, indicated by the arrow. This view shows the confined nature of the canyon.

In May, 2016, I visited the Sayles Adobe site in Eagle Nest Canyon (Fig. 3) near Langtry. Eagle Nest Canyon is a narrow, deep, rock-floored canyon that drains southward into the Rio Grande, and two famous sites, Eagle Cave and Bonfire Shelter, lie farther upstream in the canyon. Sayles Adobe is a sandy terrace site that sits perched high (about 11 meters) above the canyon floor not far from the confluence with the Rio Grande. It sits at the foot of the talus slope descending from Skiles Shelter (41 VV 165) and may have served as an adjunct area during the occupation of that site. Tori Pagano, a Texas State University graduate student, has been excavating Sayles Adobe for her MA thesis in geoarcheology, and my report on the snails is to appear as an appendix in her thesis. Tori collected a discontinuous series of eight pilot samples for me from two separate excavation blocks, representing about a 2500-year span of the Late Holocene. Here, instead of a continuous column, the goal was simply to find out if microsnails were present in these sediments, and if so, what kind and in what quantity.

The thick package of alluvium at this site consists of sandy sediments derived from limestone terrain in the canyon watershed, along with some fine-grained sediments (including mud drapes) contributed by backflooding from the nearby Rio Grande. Two of the samples were box-shaped removals 20 cm thick (“borrow pit” block), and six were 22-42 cm in thickness (“sandbox” block; Fig. 4). The samples come from two separate excavation blocks and represent a maximum elevation difference of about 2.9 m. These samples cover about a 2500-year span of the Late Holocene, from about 3167 cal BP at the lowest sample to a point somewhat later than 675 cal BP for the uppermost sample (Tori Pagano, personal communication 2018). This uppermost sample, at about 1275 AD, occurs well into the Medieval Climatic Anomaly. Sedimentation rates were high at this site. An Early Holocene radiocarbon assay of 8236±34 RCYBP was obtained on guajillo charcoal from an auger test at a depth of 5.7 m, so there are sediments equivalent in age to the Lykes Duncan site, but no excavation or snail sampling has been done at this depth. Late Holocene sedimentation rates were evidently high at this site, so the unusually thick samples obtained (up to 42 cm) probably do not imply much time-averaging.

Figure 4. Looking south at profile wall of the “sandbox” excavation area at Sayles Adobe. Three of the eight snail samples are shown. Rapid Late Holocene deposition produced these sediments.

For the 8 pilot samples, average volume was 12.7 liters (range, 11.3-15.0). The processing methods used were the same as those at the Lykes Duncan site. From the 101.35 liters of sediment processed, only 559 specimens were recovered; almost 81% of these came from the 0.5 mm mesh sieve. Specimen density is fairly low, about 5.51 specimens per liter. This is much lower than the 15.7 specimens per liter at Bonfire Shelter, doubtless reflecting the much higher depositional rates at Sayles Adobe. Low specimen density could be a result of poor habitat quality, the diluting effect of rapid sediment deposition, or both.

The samples produced one adult Rabdotus (species indeterminate), three juveniles, and some spires of indeterminate age. Other than this, the snail fauna is very similar to that from Genevieve Lykes Duncan. The assemblage includes Gastrocopta pellucida adults and juveniles, Succineidae adults and (mostly) juveniles, Helicodiscus singleyanus (Lucilla singleyana) of indeterminate age, a few Pupoides albilabris adults, and perhaps two aquatic species: small, mostly juvenile snails tentatively identified as Gyraulus parvus (but with a somewhat anomalous shape), and two small planorbids that look different from Gyraulus (an embryo and a columellar fragment) and suggest the possibility of one or more different species.

Except for the presence of a few Rabdotus and the absence of Physa, this assemblage is much the same as that at Genevieve Lykes Duncan. The succineids are mostly very tiny embryos, and I suspect represent one of the two upland, arid-adapted species in the Succineidae family.


Overview: Snails in Dryland Alluvium

These two sites differ in many ways. Genevieve Lykes Duncan lies in an open, lowland basin at about 1180 meters, in sediments generated from igneous and metamorphic bedrock, and spans the entire Holocene, at the confluence of two meandering, unconstrained streams. Sayles Adobe lies in a narrow, rock-floored limestone canyon at 350 meters, partially sheltered from wind and insolation, with samples that are limited to the Late Holocene (although older but unsampled sediments are at the base), with sediments produced both by limestone bedrock terrain and by Rio Grande backflooding (although the there seems to be no particular evidence of a Rio Grande influence in the snail assemblage).

Annual precipitation is similar in the two areas, but it is distributed differently. At Langtry under the current climatic regime, it  is about 37.26 cm/year (based on 1981-200 normals) with 32% occurring in the summer. There is no permanent weather station on the O2 Ranch, but average annual precipitation was about 36 cm in 1914-1928, and 42 cm in 2015-2016. May to October are the wettest months, with August the peak, and 44% of precipitation occurring in the summer months (June to August).  Insolation rates are very high here and evapotranspiration exceeds precipitation for 11 out of every 12 months. Although the antiquity of the North American Monsoon system is not entirely certain, Genevieve Lykes Duncan participates in this system today, receiving Pacific moisture, often from convectional thunderstorms, mostly in the summer, when plant growth and snail activity are enhanced, but evaporation rates are higher. Sayles Adobe lies farther east and receives more moisture from the Gulf of Mexico, so its precipitation pattern is a hybrid of the West Texas monsoon pattern and the bimodal eastern Texas pattern of rainfall peaks in May and September. At Langtry, April-May and September-October are the rainiest months, but significant rainfall also occurs in June-August (Fig. 5).

Figure 5. Average annual precipitation (expressed as proportions of the annual total) on the O2 Ranch and at Langtry. The O2 data are limited and mostly not recent, but the influence of the North American Monsoon system can be seen here. Notice also the greater month to month variability.

Despite these differences, both sites have similar snail assemblages, characterized by low specimen density, a very limited range of resilient, arid-adapted taxa, and apparent high juvenile mortality in some of the species. Both assemblages seem to indicate  arid, drought-prone (and probably variable) conditions that only resilient habitat generalists can tolerate. These findings raise an interesting question: if the fauna consists entirely of resilient habitat generalists, can we rely on it to register small-scale fluctuations in past climates? In faunal studies, it is often the rare, infrequent species with specialized habitat preferences that carry the most paleoenvironmental information payload. If none of these are present, can we rely on fluctuating specimen counts to indicate changing environmental conditions?

We probably need to apply the FRIN principle (Further Research Is Needed).



Cheatum, E. P.

1966        Report on Mollusk Shells Recovered From Four Archeological Sites in the Amistad  Reservoir. Pages 227-243 in Dee Ann Story and Vaughn M. Bryant, Jr. (assemblers), A   Preliminary Study of the Paleoecology of the Amistad Reservoir Area. Final Report of  Research Under the Auspices of the National Science Foundation (GS-667).

Cloud, William A., Richard W. Walter, Charles D. Frederick and Robert Mallouf

2016        Late Paleoindian Occupations at the Genevieve Lykes Duncan Site, Brewster County,       Texas. The Journal of Big Bend Studies 28:1-82. Center for Big Bend Studies, Alpine.


Manuscripts Resulting from This Research


Brown, Kenneth M.

n.d.a        Analysis of a Column of Snail Samples from the Genevieve Lykes Duncan Site.   Manuscript in progress, estimated length about 80 pages, to be published by the Center    for Big Bend Studies.


n.d.b        Appendix G: Pilot Sampling of the Snail Fauna at Sayles Adobe. Appendix to appear in

MA thesis (2019) by Victoria Pagano, Texas State University, estimated length about 25  pages.

Leave a Reply

Your email address will not be published. Required fields are marked *