Obsidian, a black volcanic glass, is a ubiquitous find in Mesoamerican archaeology. Prized for its properties—sharpness, color, and shine—obsidian was a popular material among many of the precolonial communities of Mesoamerica: it was used to manufacture flakes, blades, and bifaces and was involved in an array of subsistence and ritual activities. The exploitation of obsidian is well attested by its frequency in archaeological sites from Mexico to western Honduras. However, that frequency progressively dwindles toward the east, and obsidian is rarely found in most of southern Central America. This uneven access to the resource is related in part to the absence of a quality obsidian source in the territories between current-day Nicaragua and Panama (Braswell Reference Braswell and Smith2003:141; Sheets et al. Reference Sheets, Hirth, Lange, Stross, Asaro and Michel1990).
Located on the edge of what is frequently referred to as southeastern Mesoamerica, Honduras presents uneven patterns of obsidian use and access. The three main sources known in Honduras—La Esperanza, San Luis, and Güinope (Aoyama et al. Reference Aoyama1999; Joyce et al. Reference Joyce, Shackley, McCandless, Sheptak, Fajardo and Avalos2004; Sheets et al. Reference Sheets, Hirth, Lange, Stross, Asaro and Michel1990; Sorensen and Hirth Reference Sorensen and Hirth1984)—supplied the bulk of obsidian for precolonial sites beyond the wider Copan area. Research suggests that the exploitation of these sources was neither centralized nor regulated by political actors or polities, thereby affording the chiefdoms of Honduras relative independence from the networks operated by their state-level neighbors to the west (Braswell Reference Braswell and Smith2003:155; Golitko and Feinman Reference Golitko and Feinman2015). However, little is known about the role of obsidian exploitation in the internal socioeconomic networks among small-scale societies in the region, due in part to the paucity of geochemical obsidian sourcing studies in eastern Honduras.
The geochemical properties of obsidian make it particularly suitable for the study of past socioeconomic systems. Five decades of provenance studies in Central America have provided reliable data about how to geochemically distinguish sources, which have generated valuable insights into local exploitation and interregional exchange (e.g., Aoyama Reference Aoyama2017; Braswell Reference Braswell and Smith2003; Golitko and Feinman Reference Golitko and Feinman2015; Hendon Reference Hendon2004; Johnson and Canuto Reference Johnson, Canuto, Robinson and Davies2023; Joyce et al. Reference Joyce, Shackley, McCandless, Sheptak, Fajardo and Avalos2004; McFarlane and Schortman Reference McFarlane and Schortman2017; Sheets et al. Reference Sheets, Hirth, Lange, Stross, Asaro and Michel1990; Sorensen and Hirth Reference Sorensen and Hirth1984; Stroth et al. Reference Stroth, Otto, Daniels and Braswell2019). Instrumental geochemical characterization through neutron activation analysis (NAA), inductively coupled plasma mass spectrometry (ICP-MS), and portable X-ray fluorescence spectroscopy (pXRF) proved particularly successful in correct source attribution through elemental analysis (Braswell and Glascock Reference Braswell and Glascock1998; Lynch et al. Reference Lynch, Locock, Duke and Weber2016; Stroth et al. Reference Stroth, Otto, Daniels and Braswell2019). However, ICP-MS and NAA are both costly and destructive analyses, and their application can be limited by local heritage protection laws. By contrast, pXRF’s accuracy has improved significantly in recent decades, and it offers a reliable low-cost alternative for the sourcing of obsidian, thereby facilitating the study of larger datasets (e.g., Aoyama Reference Aoyama2017; Feinman and Riebe Reference Feinman and Riebe2022; Glascock and Ferguson Reference Glascock and Ferguson2012; Moholy-Nagy et al. Reference Moholy-Nagy, Meierhoff, Golitko and Kestle2013; Stroth et al. Reference Stroth, Otto, Daniels and Braswell2019).
In this article, we discuss the results of the diachronic study of the entire collection (N = 713) of obsidian recovered at two sites on El Tigre Island in the Gulf of Fonseca. This analysis sheds light on the integration of the small-scale societies of the southern edge of Mesoamerica in localized socioeconomic networks and long-distance transference systems between the Terminal Classic and the Late Postclassic (AD 800–1521). To contextualize our results, we additionally analyzed 54 artifacts from two coastal Honduran sites bordering the Gulf of Fonseca. All 767 artifacts were geochemically sourced using pXRF. We provide new data for the Terminal Classic (AD 800–1000) and Early Postclassic (AD 1000–1200), critical periods in regional trajectories (e.g., Geurds Reference Geurds2018:8–10; Reference Geurds, Stanton, Taube, Coltman and Camacho2023:647), and present new insights into the elusive Late Postclassic (AD 1350–1521), a period little represented in the archaeological record of the northern Isthmo-Colombian area (Geurds Reference Geurds, Stanton, Taube, Coltman and Camacho2023:647).
Obsidian Samples
Located at the intersection of hypothesized trade routes along the Pacific coast of Central America (see Callaghan et al. Reference Callaghan, Fitzpatrick, Montenegro, Beekman and McEwan2022:Figure 2.5) and of riverine networks flowing through inland El Salvador, Honduras, and Nicaragua, the Gulf of Fonseca offers a unique setting to examine the embeddedness of coastal and island communities in long-distance trade systems (Figure 1). Holding a geographically central position in this gulf is the island of El Tigre. The first archaeological investigation of this island was conducted in 2022 under the auspices of the Proyecto Arqueológico Manglares e Islas del Golfo de Fonseca (PAMIGOLF). Eight sites were identified from surface features in the survey. Based on feature conservation, two mounds were selected for stratified test pits at the site of La Tigüilotada and one at the site of Gualorita. Obsidian artifacts were recovered from surface collection in six sites and from excavation in all three test pits, totaling 747 samples.
Figure 1. Location of the Gulf of Fonseca on the Pacific Coast of Central America, highlighting the sites mentioned in text. (Color online)
In this study, we focus on the 713 samples recovered in excavation and surface collection context from La Tigüilotada and Gualorita for which we have reliable indication of the chronology through AMS radiocarbon dating of six contexts. The 47 surface features documented at La Tigüilotada make up a complex site plan, with elongated earthen mounds in plaza-like organization in the southeastern part of the site and a semicircular organization of shell mounds in the northwestern part (Figure 2). At this site, a sample of 450 obsidian fragments associated with surface features were recovered in surface collection. Mound 13, an oval shell mound of 18 × 35 m, was selected for excavation of a 1 × 1 m test pit in which 55 obsidian artifacts were recovered. The dates obtained from this context indicate the use of Mound 13 likely started at the beginning of the fourteenth century AD and lasted into the sixteenth century AD.Footnote 1 Mound 3, an elongated earthen mound of 11.3 × 45.7 m, was selected for the excavation of a 2.5 × 10.0 m trench to investigate the plaza-like section of the site: 134 obsidian artifacts were recovered there. The material and charcoal samples recovered from Mound 3 suggest an occupation starting at the latest in the ninth century AD and likely lasting until the end of the twelfth century AD.Footnote 2
Figure 2. Map of La Tigüilotada, El Tigre Island, Honduras.
The site of Gualorita is located in a contemporary village in a zone of higher construction density, which has negatively affected the preservation of surface features. Because of poor site conservation, only 25 structures were documented, from which no specific site organization could be extrapolated. However, surface scatter and excavation in one of the main intact structures seem to indicate that the site was principally made up of shell mounds. During the survey, 64 obsidian artifacts were collected from the surface in association with visible features. A single 1 × 1 m test pit was excavated in the best-preserved mound, from which only 10 obsidian fragments were recovered. A single AMS radiocarbon date from the middle layer of this mound indicated 417 ± 26 BP,Footnote 3 suggesting that Gualorita was likely inhabited contemporaneously to Mound 13 in La Tigüilotada and may still have been occupied in the early colonial period.
To situate these results in a contemporary regional context, we gained access to samples from two mainland coastal sites from Honduras: La Pegajosa and La Danta (see Figure 1). Both sites were documented in 1965 by a French expedition led by archaeologist Claude Baudez (Reference Baudez1965, Reference Baudez1966, Reference Baudez and Sauter1973, Reference Baudez1976), and their obsidian artifacts are currently held at the Musée du Quai Branly. Located on a 100 m long sandbank between the mangrove forests and the coast on the mangrove island of Güeguensi, La Pegajosa is a large salt-making site comprising salt trenches and shell middens. Despite an assemblage dominated by nondiagnostic briquetage, this site is hypothesized to have dated to the Late Classic. Fourteen obsidian artifacts obtained from surface collection from the best-preserved portion of this site were analyzed for this study.
The site of La Danta is located in the southern part of the plain of Choluteca and comprises 21 documented low mounds. The 40 obsidian artifacts from the Quai Branly collection we integrated into our analysis came from the excavation of two parallel, elongated earthen mounds. Recalibration of the only AMS radiocarbon-dated charcoal sample from La Danta (reported in Long and Mielke Reference Long and Mielke1967:377), which was collected from the lowest levels of the structure, produced large age ranges spanning from the Early Classic to Middle Postclassic.Footnote 4 However, the presence of Vallejo and Papagayo polychromes, as well as Tohil, at this site suggests an Early to Middle Postclassic occupation. It is therefore unclear whether the occupation of La Danta would have been contemporary to the sites documented on El Tigre.
Obsidian Use on El Tigre
Compositional Analysis
To assess the variability and provenance of the obsidian artifacts, we chose to take a nondestructive compositional approach using a portable X-ray fluorescence unit (Bruker Tracer 5g with a resolution of less than 140eV at 250000 cps at Mn Ka; see Supplementary Material 1 for the full protocol; and Supplementary Material 2 for accuracy and precision of measurements).Footnote 5 To evaluate chemical data, we compared bivariate diagrams of elements deemed essential for this region. The obtained geochemical data were then contrasted with published XRF and NAA reference data (Supplementary Material 3). Averages of the references were used as a generalized approach to assess the consistency and attributions of the data.
From an analytical perspective, Sr/Zr and Rb/Zr plots seem to carry the most potential for provenance approaches (Glascock Reference Glascock2020:39). Most samples show relative constricted signatures that are consistent with the published ratios of the obsidian sources of Güinope, La Esperanza, and Ixtepeque (Figure 3). No other signatures are present in significant amounts, which shows a clear selection of and access to these sources over time. Only five artifacts have a divergent analytical result that cannot be readily attributed to one of these three sources; four of these outliers fall just beyond the currently documented source range of La Esperanza, and one may fall within the margins of El Chayal data. Visually these samples do show a slight variability from the general assemblage and remain unattributed.
Figure 3. Bivariate diagram of Sr/Zr (ppm) versus Rb/Zr (ppm) showing source attributions for all measurements of the complete obsidian assemblage toward Ixtepeque, La Esperanza, El Chayal, San Martin Jilotepeque, and Güinope reference data.
Sites
El Tigre: La Tigüilotada. The 639 obsidian artifacts from La Tigüilotada, from both excavation and survey contexts, cluster around three sources: 212 (33.2%) fragments are sourced to La Esperanza, 149 (23.3%) to Güinope, and 274 (42.9%) to Ixtepeque. Based on mass, La Esperanza represents 30.4% of the assemblage Güinope, 28.4%, and Ixtepeque, 41.2%. Four samples remain unassigned (Table 1; Figure 4a). In contrast, assemblages from the two excavation contexts at La Tigüilotada exhibit pronounced differences in provenance and technology. Whereas in Mound 3 La Esperanza makes up only 4.5% of the sample, Güinope makes up 48.8% and Ixtepeque another 56.7% of the assemblage (based on mass, the percentages are 7.2%, 45.8%, and 47%, respectively). By contrast, in Mound 13, La Esperanza makes up 81.8% of the assemblages, whereas the percentages for Güinope plummet to 14.5% and for Ixtepeque to 3.6% (based on mass, the percentages are 77%, 21.5%, and 1.5%, respectively). The assemblage of Mound 3 reflects a focus on flake-core technology across sources.
Figure 4. Bivariate diagram of Sr/Zr (ppm) versus Rb/Zr (ppm) of measurements for obsidian artefacts organized per site: (a) La Tigüilotada, (b) Gualorita, (c) La Pegajosa, and (d) La Danta against source reference data.
Table 1. Source Attribution of Technotypes per Recovery Context on El Tigre.
Waste and byproducts of manufacture, such as platform installation removals or platform rejuvenation flakes, indicate local production associated with the import of small cortical nodules from Güinope and large nodule fragments from Ixtepeque. The assemblage of Mound 13 reflects a technology targeting the production of prismatic blades and, to a lesser extent, of bifaces. Points fashioned from retouched blade blanks, a regionwide feature in Middle to Late Postclassic Honduras, are also present. The assemblage of technotypes sourced to La Esperanza from surface collection in La Tigüilotada suggests the obsidian might have been traded as blade cores or as nodules.
El Tigre: Gualorita. The 74 obsidian objects recovered from Gualorita cluster around the same sources of La Esperanza, Güinope, and Ixtepeque, with one flake fragment remaining unassigned (Table 1; Figure 4b). The frequency of all three sources corresponds to Mound 13 of La Tigüilotada, with obsidian from La Esperanza dominating the assemblage at 89.2%. The technotypes present at this site, both on the surface and from excavation, also correspond to the use of La Esperanza obsidian documented for La Tigüilotada, with a predilection for the manufacture of blades and arrow points from blade blanks. The presence of both Güinope and Ixtepeque obsidian is insignificant at this site.
La Pegajosa. Of the 14 obsidian artifacts that make up this assemblage, 12 could be sourced to La Esperanza and one each to Güinope and to Ixtepeque (Figure 4c). Nine of these artifacts are blades, two are flakes, two are bifacial points, and one is a unilaterally retouched core-face rejuvenation flake. The one sample that was sourced to Ixtepeque is a large, retouched flake, and the one sourced to Güinope is a blade.
La Danta. Of the 40 obsidian artifacts from La Danta that we were able to analyze, 36 can be sourced to Güinope and four to La Esperanza (Figure 4d). Of the technotypes documented in the assemblage, 36 are associated with the manufacturing process, with the assemblage dominated by cores and their byproducts: platform installation removals, core-face rejuvenation flakes and blades, and distal maintenance fragments. The frequency of (neo)cortex indicates that small nodules served as support for the production of most pieces. With the exceptions of two flake cores, most core fragments seem to be associated with blade technology. Among the additional four pieces, three are bifacial projectile points, and one is a flake with bifacial retouch. The Güinope obsidian seems to be geared toward local blade production fashioned from cortical nodules. By contrast, the association of La Esperanza obsidian with two of the four bifacial pieces in this assemblage could represent a preference for La Esperanza obsidian for their manufacture.
Diachronic Perspectives on Obsidian Use on El Tigre
Our analysis of these assemblages shows clear diachronic differences between an early and a late occupation of El Tigre. The occupation phases are associated with different networks of obsidian procurement and seem to privilege the manufacture of different types of end products.
First Phase of Occupation (AD 800–1200). Flake production represents the majority of the assemblage from the excavation of Mound 3, most of which can be traced to the Ixtepeque source. In contrast, blades make up only 16.3% of the assemblage. The presence of waste and cortical flakes suggests that the assemblage is made up of the byproducts of domestic flake-core industries, with most cortical flakes attributed to the Güinope source. Although the proportion of blade fragments is higher for the Ixtepeque source, there does not appear to be a preferential use of either source. The low frequency of blades from Ixtepeque obsidian and the absence of obvious byproducts of blade manufacture in the assemblage could indicate their trade as finished products, parallel to the trade of Ixtepeque nodules. Although some Ixtepeque flakes seem to correspond to biface thinning-flakes, the low frequency of Ixtepeque bifaces does not allow us to ascertain whether the inhabitants of Mound 3 engaged in their production.
Second Phase of Occupation (AD 1350–1520). The second phase of occupation of El Tigre is represented at the sites of La Tigüilotada and Gualorita, with both excavation assemblages being dominated by blades manufactured from La Esperanza obsidian. In both contexts, other sources of obsidian are insignificant and could have come from down-the-line circulation. Points fashioned from blade blanks are also associated with this later occupation (Figure 5). The small amount of waste pieces for this occupation seems to indicate that blade production was efficient and specialized and that Esperanza obsidian was likely imported as blade core preforms (Figure 5). The two bifacial pieces found in Mound 13, as well as the five bifacial pieces sourced to La Esperanza from surface collection, including a bifacial point broken during manufacture (Figure 5), indicate the presence of a local bifacial technology in this second phase.
Figure 5. Examples of obsidian artifacts from surface collection at La Tigüilotada. Top left to bottom right: three points attributed to Ixtepeque, an unknown source and La Esperanza; a blade core and a blade attributed to La Esperanza; and three bifaces attributed to Güinope, Güinope and La Esperanza. (Color online)
Implications for Settlement History. A clear pattern emerges from the analysis of obsidian recovered from excavated contexts at La Tigüilotada and Gualorita. The first phase of occupation seems dominated by obsidian from Ixtepeque and Güinope, whereas the second phase is represented mostly by La Esperanza obsidian. This pattern allows us to better understand the provenience results from the recovered obsidian. If our hypothesis holds true—if Ixtepeque and Güinope act as markers of the Terminal Classic / Early Postclassic, and La Esperanza functions as a marker for the Late Postclassic occupation—then this would suggest that Gualorita is a fully Late Postclassic site; in addition, the complex settlement pattern of La Tigüilotada can be separated into an early and late occupation based on surface finds associated with mapped features (Figure 6). By categorizing features chronologically based on the source attribution of surface obsidian, we can split the site into two clusters: (1) the plaza with earth and stone mounds in the southeast section of the site, which dates to the earlier occupation, and (2) the semicircular shell mound cluster in the northwestern section of the site, dating to the later occupation. The source attribution of the surface finds suggests the reuse of certain structures across both periods of occupation. This hypothesis is further supported by the ratios of technotypes recovered per source. In fact, for the survey and excavation assemblages, respectively, 80.2% and 86.6% of Güinope technotypes are associated with flake-core technology. The same trend holds true for La Esperanza, where blades make up 66.8% of the survey assemblage and 67.7% of the excavation assemblage. A stronger divergence can be observed with Ixtepeque obsidian, for which flake-core technologies make up 66.2% of the technotypes in the surface collection and only 57.5% in the excavation assemblage. The different procurement strategies and associated technologies, along with differences in building strategies and settlement planning, allow us to posit a cultural discontinuity between the first and second occupation.
Figure 6. Maps of la Tigüilotada. Top, Representation of the source attribution ratios per mound; bottom, representation of the attributions to hypothesized periods of occupation based on obsidian sourcing data.
Long-Term Dynamics of Obsidian Procurement in the Gulf of Fonseca
Comparative data from obsidian provenience analyses from the island of Conchagüita (El Salvador) and from several sites from northwestern Nicaragua allow us to situate the procurement strategies for the islands of El Tigre and Güegüensi, as well as for the southern coastal plain site of La Danta, in regional obsidian transference networks (Figure 7). This comparison enables the identification of diachronic and diatopic patterns in the obsidian procurement strategies across the Gulf of Fonseca.
Figure 7. Location of comparative sites mentioned in text (based on Braswell et al. Reference Braswell, González, Fletcher and Glascock2002; Brown et al. Reference Brown, García Vasquez and de García2014; Colón Reference Colón2018; Gomez Reference Gomez2010). (Color online)
Conchagüita Island, 10 km from El Tigre, holds important interpretive value in the comparison of obsidian provenience datasets. The obsidian analyzed from Conchagüita stems from transitional Late Postclassic and early colonial contexts, likely contemporary to the late occupations of La Tigüilotada and Gualorita. Of the 46 samples from Conchagüita analyzed by ED-XRF, 33 were sourced to La Esperanza (Gomez Reference Gomez2010:Appendix B).Footnote 6 Gomez (Reference Gomez2010:124) found that the only debitage shatter recovered was assigned to La Esperanza, which implies that the blades from Guatemalan highland sources may have been circulated as finished products and that the La Esperanza obsidian was either traded as nodules or as prepared cores and worked in situ. The dominance of La Esperanza obsidian, recovered in a shell midden context, is consistent with the dynamic of obsidian procurement and use observed in the Late Postclassic occupations on El Tigre—suggesting that in this period both islands participated in the same networks of obsidian transference. The technotypes recovered in both contexts also suggest that both communities relied on La Esperanza obsidian primarily for the local manufacture of blades and blade blanks for projectile points.
It is possible to hypothesize that the island of Güegüensi took part in this network: of the 14 analyzed artifacts, 12 were sourced to La Esperanza. The predominance of blades in this assemblage along with the presence of bifacial points seems to correspond to both Late Postclassic contexts on Conchagüita and El Tigre. The overlaps between the assemblages of La Pegajosa, Conchagüita, and El Tigre’s late occupations allow us to posit that Güegüensi is, in fact, a Late Postclassic site, and not a Late Classic site as previously thought. This implies that the communities that established this salt production site had access to obsidian from the same networks of exchange as the island communities of Conchagüita and El Tigre. This chronological assignment is further supported by the presence of pecked and ground platforms, which seem to be another marker for Middle and Late Postclassic occupations in our assemblages.
The largest dataset of sourced obsidian in the region is found in assemblages from 12 sites in the Nicaraguan province of Chinandega. These 12 sites represent occupations spanning from the Late Preclassic until the Early Postclassic. Among the 2,871 obsidian artifacts analyzed from the 12 sites, 2,820 were attributed to the Güinope source, 25 to the La Esperanza source, and 24 to the Ixtepeque source (Colón Reference Colón2018:72). At most sites, the presence of La Esperanza and Ixtepeque obsidian is insignificant and is generally associated with prismatic blades, which are interpreted as products acquired in their final form (Colón Reference Colón2018:76). The only site presenting some shatter from the Ixtepeque source, which would indicate that a core might have been (re)worked locally, is the coastal plains site of Cosmapa Oriental. At this site, La Esperanza is present at a similar frequency as Ixtepeque (10%), and a core fragment suggests that part of the reduction process occurred as well (Colón Reference Colón2018:98). This indicates that the communities inhabiting Cosmapa Oriental had access to either core preforms or nodules. Although this site has a Late Preclassic component, the Ixtepeque and La Esperanza materials are thought to be chronologically associated with the Early Postclassic occupation (Brown et al. Reference Brown, García Vasquez and de García2014:89; Colón Reference Colón2018:Tables 11 and 12), which would have been at the same time as the first occupation of La Tigüilotada. Most other occupations with Terminal Classic and Early Postclassic components exhibit 0.7%–4.0% of blades from Ixtepeque and La Esperanza obsidian.
Across all periods of occupations, it appears that coastal plains sites and sites from western Chinandega had the most access to Ixtepeque and Esperanza obsidian (Colón Reference Colón2018:130). The steady presence of cortical flakes and production byproducts across contexts in Chinandega suggests the acquisition of Güinope obsidian in nodule form. The dynamics observed among the Chinandega sites seem to be reflected among two Terminal Classic sites in the Madriz department (Nicaragua)—Güiliguisca and Cacaulí I—where Güinope obsidian represents 84.8% and 90.9% of the assemblage, respectively (Braswell et al. Reference Braswell, González, Fletcher and Glascock2002:Table 1), and most fragments correspond to flake-core technology (Braswell et al. Reference Braswell, González, Fletcher and Glascock2002:28). The assemblage of the Early/Middle Postclassic site of La Danta, dominated by Güinope cores that exhibit cortex, fits within the pattern observed in the sites on the coastal plains in Nicaragua.
Although the insignificant presence of Esperanza obsidian among Terminal Classic and Early Postclassic coastal plains sites is consistent with the quantities recovered at La Tigüilotada for the same phase, the pattern for Ixtepeque obsidian is vastly different. In fact, Ixtepeque obsidian dominates the first phase in the excavation context, with a percent ratio of 56.7% against 38.8% of Güinope. Although cortical flakes of Ixtepeque are not present in our excavation assemblage, the frequency of flakes and waste suggests that decortified nodules or core preforms were worked in situ. This supports the idea that the communities of El Tigre had preferential access to transference systems circulating Ixtepeque obsidian, whereas coastal plains sites seem to have operated on the periphery of this network of exchange. It is possible to postulate that, in this period, islands in the Gulf of Fonseca acted as nodes of redistribution of Ixtepeque obsidian for coastal plains communities. The lack of comparative data for coastal plains sites in the Late Postclassic at this time does not allow us to determine whether this peripheral status of coastal plains communities in obsidian transference systems was maintained in later periods with La Esperanza obsidian.
The data presented do allow us to argue for separate networks of procurement for island and coastal communities of the Gulf of Fonseca in the Terminal Classic and the Late Postclassic. In fact, the coastal and piedmont communities from the eastern Gulf of Fonseca exhibited stable and resilient procurement strategies for Güinope obsidian from the Late Preclassic to the Late Postclassic, steadily participating in Braswell’s proposed “Lower Central American exchange sphere.” By contrast, the networks supplying island communities alternated between Braswell’s (Reference Braswell and Smith2003) “Southeast Maya Exchange,” “Central Honduran,” and “Lower Central American” exchange spheres. Although they favored higher-quality obsidian, these procurement strategies seem to have been comparatively more unstable and short-lived and possibly were dependent on the cultural affiliations of the islands’ inhabitants.
Obsidian Procurement Networks
Least-Cost Paths
To evaluate possible trade routes from the different sources identified to El Tigre, we performed a least-cost path (LCP) analysis. In this analysis, we selected time of travel from the obsidian sources to the archaeological site as the “cost.” To define the cost surface, we considered the distance and elevation changes of the terrain (slope map). Navigable rivers, lakes, and the Pacific Ocean were speed-enhancing factors, whereas non-navigable rivers and mangrove forests were barriers, increasing the cost. Because human movement speed over the slope decreases proportionally to the incline and decline of the terrain, researchers created various algorithms to model the movement speed over slopes. For this analysis, we used Tobler’s hiking function’s (Tobler Reference Tobler1993) adaptation for ArcGIS (Tripcevich Reference Tripcevich2009). Although some limitations to Tobler’s hiker function for creating friction surface have been reported, it remains a standard in archaeology for LCP analysis, and its effectiveness has been tested in ethnographic studies (Aldenderfer Reference Aldenderfer1998; Gowen and de Smet Reference Gowen and De Smet2020).
Ethnohistoric sources report human movement on foot on land routes and by canoe on rivers and lakes and near the coast of the Pacific Ocean in Central America (Ponce Reference Ponce1872; Roberts and Shackleton Reference Roberts and Shackleton1983; Stone Reference Stone, Eckholm and Willey1966; Wafer Reference Wafer1903 [1699]). However, archaeologists modeling LCP in Central America have almost exclusively focused on land routes, either excluding rivers from the model (e.g., Benfer Reference Benfer2023; Munoz Reference Munoz2017; Rivas Reference Rivas2014; Silva de la Mora Reference Mora and Flavio2023) or using rivers as buffers (e.g., Rosenswig and Tuñón Reference Rosenswig and Tuñón2020). Analyses conducted on mobility between Early Postclassic sites on the Balsam Coast Range, El Salvador, considered the Pacific Ocean as a possible path for modeling intersite mobility but did not report on the assigned costs (Escamilla Rodríguez Reference Rodríguez and Marlon2022).
Considering the location of El Tigre, which is surrounded by dense mangrove forests and river deltas, and the location of the obsidian sources, most of which had no direct access to navigable rivers, we constructed a multipurpose cost surface that allowed the incorporation of foot transport and canoe travel for rivers, the Gulf of Fonseca, and the Pacific Ocean (for full protocol, see Supplementary Material 4). It is important to keep in mind that the proposed LCPs have many limitations, because they are based on a simplified representation of the natural landscape and do not consider the varied social aspects of travel. Our model could not account for considerations such as deviations from the route or delays caused by sociopolitical or practical factors. The resulting model depicted in Figure 8 is to be considered a fact-driven hypothesis, rather than a representation of exact routes.
Figure 8. Calculated least-cost paths for travel from the sources of Ixtepeque, Güinope, and La Esperanza to El Tigre. (Color online)
Procurement Strategies in the Gulf of Fonseca
Güinope. The obsidian procurement strategies documented in the eastern section of the Gulf and in most of southeast Honduras seem consistent with a strong and stable reliance on Güinope obsidian starting in the Late Preclassic (e.g., Braswell et al. Reference Braswell, González, Fletcher and Glascock2002; Colón Reference Colón2018:72). Within this established transference system, the presence of Güinope obsidian on El Tigre is unsurprising. The proposed riverine LCP into the Gulf of Fonseca is a likely candidate to have supplied coastal lowland communities in the Gulf. Piedmont communities along the highlands separating Honduras and Nicaragua could have also be part of this network or may have relied on a strategy based more on the use of land routes (see Benfer Reference Benfer2023:244–245); perhaps they took the precolonial road passing through Icalupe and Macuelizo in Nicaragua (Espinoza Pérez and Rivas Reference Espinoza Pérez and Rivas2013; Incer Barquero Reference Incer Barquero1985:378). Although the use of these transference systems likely continued in Chinandega, these networks seem to have been abandoned in the resettlement of La Tigüilotada in the Late Postclassic. Unlike other contemporary sourced contexts around the Gulf, Güinope at no point dominates the assemblage in the Terminal Classic / Early Postclassic occupation.
Ixtepeque. If the procurement strategies of Ixtepeque correspond to the proposed LCP (Figure 8), whether by direct procurement or down-the-line trade, we would expect the continued presence of Ixtepeque obsidian in archaeological sites along that path until about AD 1250. Although this analysis is limited by the small amount of Early Postclassic sites with chemically sourced obsidian both in eastern El Salvador and in southern/central Honduras, we propose a fluvial route running through the northern portion of El Salvador as likeliest for the procurement of Ixtepeque obsidian at La Tigüilotada. It is of note that this LCP overlaps with one of the LCPs proposed by Johnson and Canuto (Reference Johnson, Canuto, Robinson and Davies2023:Figure 12.3) for Late Classic obsidian trade, which follows footpaths along the riverbeds. It also overlaps in part with the trajectory followed on foot by Hernando de Soto’s army in 1525 (Fowler Reference Fowler, Figueroa, Johnson and Goodwin2021:281). This hypothesized LCP aligns with some sites with Terminal Classic and Early Postclassic components with the known presence of Ixtepeque obsidian in their assemblage in El Salvador, such as Tazumal/Chalchuapa, which lies south of the LCP, and Cihuatan (Braswell Reference Braswell and Smith2003; Fowler et al. Reference Fowler, Kelley, Asaro, Michel and Stross1984). However, other monumental centers in proximity to the proposed LCP such as San Andres, El Tanque, La Laguneta, and Quelepa (Braswell et al. Reference Braswell, Wyllys, Andrews and Glascock1994; Fowler and Earnest Reference Fowler and Earnest1985; Johnson and Canuto Reference Johnson, Canuto, Robinson and Davies2023), although exhibiting use of Ixtepeque obsidian throughout the Late Classic, fall into disuse after the Terminal Classic demographic change. The network of procurement for Ixtepeque obsidian along Salvadoran rivers, predating the settlement of La Tigüilotada, would then have been impervious to these demographic changes along its path. The likeliest alternative for our proposed LCP would be a route connecting Ixtepeque and Chalchuapa to the Pacific coastal plains and extending into the Gulf of Fonseca, as has been suggested for the Middle Preclassic (Sheets Reference Sheets and Sheets1983:220). An alternative path for procurement from Ixtepeque directly through Guatemala’s south coast and Pacific routes of exchanges is considered unlikely. Although Middle postclassic sites on Guatemala’s southeastern Pacific coast do exhibit a predominance of Ixtepeque obsidian (Bove et al. Reference Bove, Genovez, Batres, Dillon and Dillon2012:253), El Chayal and Jilotepeque obsidian predominate in the Terminal Classic occupations in this region such as El Baúl (Chinchilla Mazariegos Reference Chinchilla Mazariegos2021:70). In fact, neither southeastern Guatemalan coastal sites nor El Salvador’s western Pacific coastal sites exhibit continuous prevalence of Ixtepeque throughout the Terminal Classic and Early Postclassic, partly because of the population decline and abandonment of sites at the onset of the latter period.
An alternative path that could be considered goes south and parallel to the proposed LCP. Little archaeological evidence is available to evaluate this path, which follows historical roads taken by Ponce and Ciudad Real (Incer Barquero Reference Incer Barquero1990:Figure 25) and by Spanish armies under the leadership of Pedro Alvarado (Fowler Reference Fowler, Figueroa, Johnson and Goodwin2021:281). The frequently posited control of Quirigua or Copan over the distribution of Ixtepeque obsidian in the central valleys of Honduras and into El Salvador seems unlikely to be of relevance here because of evidence for the sustained use of Ixtepeque obsidian into the Terminal Classic and Early Postclassic (Braswell Reference Braswell and Smith2003:146). Notably the continuous circulation of Ixtepeque in the Ulua Valley, where it dominates assemblages until the end of the Terminal Classic (Braswell Reference Braswell and Smith2003:Table 20.1; Hendon Reference Hendon2004:7), and in Los Naranjos, where it continuously represents one-fifth of the assemblage throughout Late Classic and Early postclassic contexts (Sion et al. Reference Sion, Andrieu, Rodas, Deras and Sion2022:216), seems to suggest that those communities were active in networks of procurement that were impervious to demographic changes in the Maya world. The presence of Ixtepeque obsidian on El Tigre during the Early Postclassic then highlights the resilience of the network and its associated fluvial LCP and may be related to Braswell’s (Reference Braswell and Smith2003:146) proposed expansion of the “Southeast Maya Sphere” for that period. Within that sphere, this would be the most eastern example of a Terminal Classic–Early Postclassic obsidian assemblage dominated by Ixtepeque obsidian.
Although the continued and stable presence of Ixtepeque obsidian from the Terminal Classic into the Early Postclassic would not seem atypical in other parts of Honduras, it does stand out in the Gulf of Fonseca. It seems to suggest a preferential access of the community of La Tigüilotada to the Ixtepeque transference system, which coastal lowland communities from Honduras and Nicaragua did not seem to actively take part in or have access to. The stable presence of Ixtepeque from the earliest levels of La Tigüilotada suggests a consistent access to those transference systems during 400 years of occupation. We can draw two implications from this observation: (1) it is likely that the communities were already active within this network before settling La Tigüilotada, and (2) this network of procurement remained stable during a period of stark demographic change both in the highland Maya world and in most parts of Honduras.
La Esperanza. The proposed LCP for La Esperanza obsidian seems like a likely candidate for obsidian procurement (Figure 8). Although the linguistic affiliation of the precolonial inhabitants of El Tigre is beyond the scope of this article, it is likely that El Tigre was inhabited at the time of colonization by potones, a Lenca-speaking group (Gomez Reference Gomez2010:38). With La Esperanza being in traditional Lenca territory, it is possible to postulate a privileged trading relationship between highland Lenca groups and Lenca-speaking potones inhabiting the western coast and islands of the Gulf of Fonseca. The direct access from the highlands to the Gulf of Fonseca through the Goascorán River offers a likely fluvial route for the transference of obsidian goods, which overlaps with the footpaths and mule trails documented in early colonial times (Brady Reference Brady1996:Map 10; Ibarra Rojas Reference Ibarra Rojas2020:Figure 4). Unfortunately, the lack of site studies and obsidian sourcing studies for the northeastern border of El Salvador and Honduras does not yet allow us to assess whether the overlap between the proposed LCP for Ixtepeque and La Esperanza obsidian is of significance. Future studies in this area might reveal mechanisms of reappropriation of this route across periods.
Summary and Conclusions
Our results show that there were distinct systems of transference for the acquisition of obsidian on El Tigre for both the Terminal Classic / Early Postclassic and the Late Postclassic occupations. In the earlier period, the community of La Tigüilotada relied on Ixtepeque and Güinope obsidian primarily for flake-core industries, whereas in the later period, inhabitants of La Tigüilotada and Gualorita favored La Esperanza obsidian and blade production. In both periods, obsidian seems to have been readily accessible across the community, with no indications of centrally regulated distribution.
The LCP analyses indicate likely fluvial access for all three sources for the islands of the Gulf of Fonseca. The LCP for the acquisition of Ixtepeque obsidian suggests an interaction with communities of northern El Salvador in the Terminal Classic and Early Postclassic. The presence of Güinope obsidian in the first occupation is consistent with the pattern of procurement for communities in the eastern section of the Gulf of Fonseca from the Late Preclassic to the Late Postclassic (Colón Reference Colón2018): it was likely acquired through partially overlapping networks. The LCP indicates that the procurement of La Esperanza obsidian was likely tied to the navigation of the Río Goascorán and may indicate ties between the potones (Lenca) islanders reported at conquest and highland Lenca-speaking communities of the highlands of La Esperanza. Although all the proposed LCPs are still significantly limited by the lack of obsidian sourcing studies and the knowledge of contemporary settlements along potential routes in El Salvador and southern Honduras, they highlight the embeddedness of small-scale chiefdoms of the Gulf of Fonseca in large-scale exchange networks.
Finally, the results of the analysis of obsidian from La Danta and La Pegajosa, compared with previous sourcing studies around the Gulf of Fonseca, allow us to posit that island communities operated within different networks of procurement from their mainland counterparts. This hypothesis is limited by our knowledge of obsidian procurement patterns in the western part of the Gulf of Fonseca, where future obsidian sourcing studies could reveal the role that mainland communities of eastern El Salvador played in obsidian exchange from the Late Classic to the Late Postclassic. The dwindling presence of Ixtepeque and La Esperanza obsidian east of El Tigre nonetheless suggests that the island was one of the last actively embedded communities southeast of these networks. Results from the present studies then highlight the differences in participation (either due to access or preference) in obsidian exchange networks east and west of the Gulf and the key position that island communities held in brokering social and economic landscapes.
Acknowledgments
We thank the Instituto Hondureño de Antropología e Historia (IHAH) for granting permits for the field research and the export of obsidian artifacts for analysis (009-G.-2022). We would additionally like to thank Carmen Julia Fajardo and Alejandro Figueroa for providing geological reference samples, as well as the Carrera de Antropología (UNAH) for facilitating laboratory space. We are grateful to the Musée du Quai Branly for providing access to the comparative collection. We are indebted to the members and volunteers of the 2022 season of PAMIGOLF for their support in the collection and processing of lithics finds; to Leiden University students Lisa Anderson, Agnieszka Gawin, and Anouk Roggema for their help in pXRF scanning; and to Dasha Derzhavets for her help with the database. We would like to thank Katherinne Guerra Cheva for editing the Spanish abstract. Lastly, we acknowledge the communities of Amapala for their collaboration, support, and warm welcome.
Funding Statement
This work was supported by the Nederlandse Wetenschappelijke Organisatie under a PhD in the humanities grant no. PGW.20.033 and by the Fondation Martine Aublet under a doctoral fieldwork grant. The radiocarbon dates were provided by Vilnius Radiocarbon Laboratory under their PhD student academic support scheme.
Data Availability Statement
Data for this article are available on request.
Competing Interests
The authors declare none.
Supplementary Material
The supplementary material for this article can be found at https://doi.org/10.1017/laq.2024.46.
Supplementary Material 1. Protocol for the compositional analysis (text).
Supplementary Material 2. Accuracy and Precision of Measurements (table).
Supplementary Material 3. Overview of Published Geochemical Obsidian Data Used as Reference for This Study (table).
Supplementary Material 4. Protocol for the least cost analysis (text).
Open access
