This study of Basque herding is based in the Soule Valley (Xiberoa, in Souletin Basque hereafter SB) on the north-facing western Pyrenees Mountains (Figure 1). It is the smallest of the three Basque Provinces in France with a territory of 785 km² centered on the axis of the Saison River where we have conducted place-based investigations for several years on the onset and progression of pastoralism above 800 masl. The western Pyrenees Mountains form the hinge between the Ebro River Valley to the south and the Aquitanian Plain to the north. They are central not only to the agropastoral domestication of European landscapes but the rise and fall of renowned lords and kings in their battles for control over people and territory. Our investigations in the Soule Valley include stratigraphically sampling slope wash deposits (i.e., colluvium from zero-order watersheds) back to the Late Pleistocene ca. 20,000 BCE to recover multiproxy evidence for landscape fire (Leigh et al., 2015, 2016). We have also carried out ethnographic and historic research on the use of pastoral fire to manage common grazing lands as well as the processes of land use change and household abandonment (Coughlan, 2013; Coughlan and Gragson, 2016; Coughlan et al., 2022).

The use of high-mountain pastures in the Soule Valley can be summarized as a summer event recurring from time-out-of-memory that involves hundreds of Basque herders from dozens of villages across the valley converging with thousands of sheep on to the commons in the parish-commune of Larrau. The sheep consume grass which they transform into milk that the herders transform into cheese. Much has been written about herding and pastoralism in the western Pyrenees beginning with the well-known accounts by Lefebvre (1928, 1933) and Cavaillès (1931a,b). Ott (1993) conducted research on the parish-commune of Saint Engrâce bordering the parish-commune of Larrau, both located within the Soule Valley. None of these works substantively address the process of herders forming alliances to the potential benefit of the 30,000 inhabitants in the Soule Valley living in 68 towns in the early 19th century. Törbel, by comparison, is a single closed-corporate community with a maximum population of about 700 in 1950 CE.

The Coutume de Soule first published in 1521 CE (Grosclaude, 1993) and older documents divide the Soule Valley into community lands and common lands. Community lands are enclosed within the defined boundaries of a parish-commune centered on a church. Common lands generally lie above 800 masl and consist of open rangeland and forest areas only available to residents of parish-communes within the Soule Valley (Noussy Saint-Saëns, 1955).

The parish-commune of Larrau contains 62% (9,257 ha) of the high common rangelands in Soule, and elevations that range from 300 to 2,000 masl. Flatter lands in the lower elevations (300–800 masl) contain the hamlet of Larrau centered on the parish church with houses, businesses and administrative buildings clustered around it. Privately held agricultural parcels are concentrated in the flatter areas peripheral to the hamlet. Steeper lands between 800 and 1,300 masl contain communal woodlands and heathlands as well as fragmented private inholdings (borda, SB). Named households (etxe, SB) owning the borda use them to harvest hay for feeding and bracken for bedding livestock in the winter (Palu, 1992; Gragson et al., 2020). Lands above 1,300 masl are the alpine and subalpine common grasslands and heaths used by herders from throughout the Soule Valley. Only residents of Soule have access to the communal summer pastures and only those who are members of a Cayolar have a right to graze sheep in the grazing districts.

Before turning to details of herding on the commons in Soule, we provide succinct definitions of relevant terms based on key collective action sources (Olson, 1965; North, 1990; Ostrom, 1990; van Zomeren et al., 2008; Van Zomeren and Iyer, 2009; Cronk and Leech, 2013; Jagers et al., 2020; Thomas et al., 2020). Collective action means action taken by two or more individuals to improve the status, power, or influence of a group, while a collective action dilemma (or collective action problem or social dilemma) refers to factors limiting collective action. Limitations to collective action are most commonly the result of individuals free-riding on the efforts of others or a failure to coordinate the action of individuals. For the former limitation, individuals cannot be excluded from the benefits that others provide so they are unmotivated to contribute to the effort themselves. In contexts involving common-pool resources such as open grasslands used for grazing, resource units removed from the common-pool reduce the quantity available for others yet the characteristics of the resource pool (e.g., size, etc.) limit the possibility of excluding potential beneficiaries. The failure to coordinate results from the uneven distribution of information and knowledge across individuals resulting in conflicts of interest between them as they attempt to find the solution to a social dilemma.

Individuals resolve collective action problems by either (a) supplying a new set of institutions, i.e., constraints that shape social interaction often expressed in the form of rules, (b) making credible commitments to those they cooperate with and/or (c) mutual monitoring to ensure everyone “follows the rules.” These solutions require shared, common knowledge that translate to individual collaborators knowing the rules, knowing their collaborators know the rules and knowing their collaborators know they know the rules. Such rules are classified as operational, collective, and constitutional:

A Cayolar is a syndicate of herders associated with a named inholding that serves as a gateway to the common lands surrounding it [Cayolar is a Gascogne word used in French equivalent to Olha in Souletin Basque and rendered as Kaiolar in Standard Basque (Ott, 1993; Barandiaran and Manterola, 2000)]. The inholding itself minimally contains a stone-walled and wood-roofed shelter (either a hut or a barn) used by the herders, a corral to hold the sheep and is generally near water (Figure 2). The most frequently referenced description of a Cayolar is from the 1521 Coutume de Soule (Grosclaude, 1993) with additional details derived from other sources (Lefebvre, 1933; Noussy Saint-Saëns, 1955; Ott, 1993; Etchegoyhen, 2012). Cayolar members participate in the syndicate by contributing a share (txotx, SB) equal to a finite, partible number between 45 and 60 head of milch ewes. The txotx is in fact the “purchase price” of a proportional title to the inholding. It is thus a credible commitment to the other members of the organization that the herder will follow the norms of the syndicate. The fundamental norm being that the herder assures the other members that they will contribute a flock of sheep equal in size to their txotx.

To achieve success in operating a Cayolar herders must: (a) supply a set of rules they agree to follow, (b) make credible commitments to each other, and (c) mutually monitor members' conformance to the rules (Ostrom, 1990). The appropriators must trust each other to succeed. As all are inhabitants of the Soule Valley, their shared history and expected future together substantially reduces each individual's future discount rate to collaborate in the present. Cayolar organizational performance is assured by the articulation between individual herders and the population of the Soule Valley through a set of nested spatial-institutional levels.

We conducted an archaeological pedestrian survey of ~4,710 hectares of pastures above 800 m asl during which we located 42 sites and associated surface features used in seasonal herding activities excluding Cayolar huts still actively used in herding (Champagne et al., 2014). Surface features included small mounds called tertre, livestock enclosures or corrals and stone foundations of shepherd huts (Figure 3). Tertre are ~2 m in diameter at the base and 1 m high; corrals have variable dimensions and consist of bermed earth, bedrock, or stacked stone structures with an opening at one end. Shepherd hut foundations are morphologically consistent with extant though not modern shepherd huts and are roughly rectangular alignments of rocks measuring ~6 (range: 4.5–8 m) by 4 m (range: 3–5.6 m).

We used a 10 cm bucket auger to sample the soil profile at 10 cm intervals within and adjacent to huts, corrals and tertres at each site and screened each sample at 0.45 and 0.2 mm to recover cultural material and macro-charcoal for dating. Forty-eight radiocarbon samples were recovered from seven, widely distributed sites and submitted for dating to the Center for Applied Isotope Studies (UGA). Multiple 14C samples were obtained from each site but no sample provided more than one date. Samples were processed following a standard protocol with 14C/13C ratios measured with a 0.5 MeV accelerator mass spectrometer and 13C/12C ratios measured separately on a stable isotope ratio mass spectrometer. Table 1 lists the details on the archaeologically recovered samples.

We discovered in examining cadastral documents in the archives of the Department of Pyrénées-Atlantiques (Pau, France) a supplement to the 1830 Napoleonic Cadaster titled “co-owners of indivisible otherwise ‘collective' property.” The supplement is not dated, but cross-referencing the named Cayolar members with other records indicate the supplement dates between 1830 and 1832 CE. Soule faced a series of legal and fiscal challenges between the fall of the Napoleonic Empire in 1814 CE and the creation of the Commision Syndicale du Pays de Soule in 1838 CE and it seems the supplement was compiled to support claims by or against the parish-commune of Larrau.

The supplement contains comprehensive information on 68 named Cayolar operating on the commons in the parish-commune of Larrau and 231 Cayolar members from 35 parish-communes within the Soule Valley. Nine Cayolar are identified as the property of a parish-commune leased or rented to an ad hoc group of herders from one parish-commune. Twenty-three Cayolar are owned by a single individual who is sometimes indicated as renting out the property and in others is recorded as a nobleman or a public official who most likely also rented out the property. While a Cayolar refers to an organized group of herders, in this article we use the term owner/herder since 32 entries in the supplement indicate an owner for the inholding but not the individuals using the inholding.

We obtained coordinates for waterways, parish-commune boundaries and village church centroids from the BD Carto (IGN) thematic map series then verified and modified the information to match the information from the 1830 Napoleonic Cadaster. We obtained parish-commune name variants and membership in districts and provinces from Orpustan (2010) Nouvelle Toponomie Basque and the Basque Onomastics Database (Euskaltzaindia, 2022). We georeferenced the locations of Cayolar inholdings and the footprint of shepherd huts either during the pedestrian survey or from the digitized maps from the 1830 Napoleonic Cadaster. We combined all placement information into an ArcGIS analytical geodatabase.

We investigated two different chronological models using OxCal version 4.4.4 and the IntCal20 calibration curve (Bronk Ramsey, 2009; Reimer et al., 2020). We conditioned the models to constrain date ranges (Hamilton and Krus, 2018) using our knowledge of the landscape and stratigraphic context of samples. We set the site abandonment date (i.e., terminus ante quem, TPQ) either to 1829 CE for sites that had been abandoned at an unknown date prior to the 1830 CE cadaster otherwise we used the site's known, post-1829 CE abandonment date. We consider a model to be significant when the A_model and A_overall values >60, and the convergence value (C) >95% (Bronk Ramsey, 1995; Manning and Birch, 2022), and report dates in calendar years CE rounded to the nearest 5 years.

In Figure 4 we provide a summary of the dates using kernel-density estimation (KDE) with default parameters (Bronk Ramsey, 2017) that shows they span 705 and 915 years (p = 95.4%) with a distinctly bimodal distribution with peaks at c. 1300 CE and c. 1700 CE (A_model = 90.9, A_overall = 92.2, and C > 97.2). We excluded one sample (65633) that dates between 405 and 540 CE (p = 95%, C = 97.5) as it derives from the deepest sampled stratigraphic position at the site and may represent landscape burning rather than site use. In Figure 5 we present the results of our simple bounded-phase chronological model (Bronk Ramsey, 2009), which again excludes sample 65633 and sets the end-of-phase to 1958 CE when the last site in this set was abandoned. Dates in this model (A_model = 88.6, A_overall = 88.3, and C > 99.2) are assumed to derive from a continuous activity phase that began between 1035 and 1150 CE (p = 95%, C > 99.2). The use-phase lasted between 805 and 920 years with a median duration of 850 years. The distribution is again bimodal with a break between the peaks at c. 1450 CE. The KDE and simple bounded-phase model results are very similar giving us confidence in the overall duration and the general shape of the distribution.

Figure 6 contains the results of our bounded overlapping phase model that assumes the occupational phases of individual sites are independent and possibly overlapping. This allows us to estimate the start date and end date of use for each site. We constrained site use in this model to begin between 840 and 850 CE (contains the 842 CE founding date of Leyre Monastery discussed below) and the last possible date each site was abandoned. This robust model (A_model = 89.2 and A_overall = 90.5, and C > 95.9) provides greater chronological control on each site yet still aligns with the results of the KDE and simple bounded-phase model.

We previously used UCINET (Borgatti et al., 2002) to analyze the three interdependent sub-networks of the Soule herding network (Gragson et al., 2021) and characterize its structural features: (a) any herder selected at random can reach any other herder in the network irrespective of village of residence or Cayolar membership: (b) any herder selected at random is connected to more than half the total number of herders participating in the herding network; and, (c) there is no “central authority” determining where individual herders reside, who they form alliances with, or the sheep share they contribute. In summary, each sub-network as well as the overall network form a densely knit, tightly bound valley-wide structure, yet the sub-networks and the overall network are socially emergent rather than autocratically directed.

Here we use Exponential Random Graph Modeling (ERGM) to analyze the joint probability of edges connecting an owner/herder to a named Cayolar. The data are organized as a bipartite, 2-mode network. Unlike two, one-mode projections of the same data, the bipartite network accounts for ties forming between the individuals in the network who come from multiple parish-communes and are in partnership in more than one Cayolar. We analyze the network using ERGM 4.5.0 (Krivitsky et al., 2003–2022, 2023; Goodreau, 2007) dependent on R (>4.1) (R Core Team, 2013) and running in RStudio (RStudio Team, 2015). ERGM is a bottom-up, actor-based analysis framework sensitive to the non-independence of two nodes linked by an edge (Kenny et al., 2006) which treats the social network as a dependent variable of one or more independent conditions. The estimated coefficients are log-odds ratios of establishing a network tie conditional on the rest of the network (Goodreau, 2007). ERGM offers a stronger basis for interpreting how social networks form than the descriptive measures available in UCINET (Wasserman and Pattison, 1996; Robins et al., 2001; Goodreau et al., 2009; Snijders et al., 2010; Wang et al., 2013).

The bipartite network consists of 231 owner/herder nodes and 68 Cayolar nodes joined by 270 edges (Figure 7) that we condition by the (a) parish-commune (n = 34), (b) degairie (n = 7), and (c) messagerie (n = 3) an owner/herder is associated with. Fifty-five percent of all parish-communes, and 100% of all degairie and messagerie in Soule are represented in the network. We tested the fundamental question that needs to be asked about any association of two or more individuals: do the groups observed differ from the groups that might result by chance alone. We performed the test by treating the observed network as a single observation from the distribution of all possible networks with the same number of nodes (Robins et al., 2007), then fit the dataset to the Erdös-Rényi random graph model in which each edge appears independently and with equal probability. Figure 8 is a representative graph from the model run that shows numerous unaffiliated owner/herders even though in the actual network every individual is affiliated with at least one Cayolar. The edges coefficient (−4.03) indicates the Erdös-Rényi model is a poor fit to the empirical network, which is further demonstrated by the inverse logit of θ with p = 0.02 of a tie forming between any two individuals through their common membership in a Cayolar.

The simple reason for the poor fit is that ties in the empirical network beyond that of two individuals through a single Cayolar (i.e., two-star) are dyad-dependent (Figure 9) whereas ties in the Erdös-Rényi model are equiprobable. Once two individuals are allied in the empirical network, the probability of a third joining them (i.e., three-star) is a dependent probability of the first two. The widely reported number of herders needed to organize a Cayolar is seven (i.e., seven-star) yet the empirical network ranges from one-star (single individual) through 13-star (13 members) with a strong clustering of observed cases between four and eight-star and a simulated median value c. four. The parsimonious model we identified for the observed distribution of joint probability ties is that more individuals from the same parish-commune organize into a Cayolar than expected by chance alone. This result rests on the statistically significant and positive b1nodematch estimate of 0.607, equivalent to p = 0.65 of a tie forming between any two individuals from the same parish-commune. This result amounts to a test of the social principle of homophily (i.e., sociality, ethnicity) in which individuals form alliances with other individuals with similar attributes.

The 1830 CE cadaster lists 122 inholdings within the commons associated with 68 named Cayolar. The shape and size of inholdings ranges from 0.1 ha to 167 ha with a median of 8.4 ha. As a rule, inholdings held by one individual or organized group of individuals are on the small end of the size distribution whereas inholdings held by a parish-commune are larger and include both the typical infrastructure of a Cayolar inholding as well as rangeland used for sheep grazing. Some named Cayolar furthermore comprise two or three separate inholdings at different elevations that are distinguished by suffixes to the name of the Cayolar: Pekoa-Olha (lower), Arteko-Olha (middle), and Gaineko-Olha (upper) (Lefebvre, 1928). Twelve named Cayolar organizations in the Soule Valley operated from three inholdings at different elevations, 22 operated from two inholdings at different elevations, and 32 operated from one inholding.

Inholdings are further distinguished by whether they contained structures taxed as shepherd huts (cabane, FR n = 95) or structures taxed as barns (grange, FR n = 17). These are mutually exclusive types of inholdings and there are no instances of a Cayolar inholding having both a shepherd hut and a barn. We then evaluated landscape position of inholdings overall and relative to the number of inholdings per Cayolar using a 5 m² resolution digital elevation model (DEM) by taking the elevation of the shepherd hut or the barn on the inholding (Table 2). Without distinguishing the number of inholdings or type of structure, Cayolar are located at a mean elevation of 1,192 masl (range: 1,497–689 m). This is well above the mean elevation of etxe or barns that are part of fragmented etxe inholdings within communal woodlands and heathlands (i.e., Borda). There are no significant differences between the absolute range in elevation values of shepherd huts or barns for Cayolar operating from one, two or three inholdings other than the tendency for a narrowing of the elevation range for each elevation class as the number of inholdings per Cayolar increases.

However, there is significant unexplained elevational overlap in Borda vs. Cayolar inholdings that is both subtle and largely invisible archaeologically. Indeed, both Borda and Cayolar occur as inholdings within the heathland commons and while Borda are always associated with barns, they sometimes have huts; while most Cayolar are associated with huts, some instead have barns. The main differences between the two inholdings are difficult to parse archaeologically either by ownership (Borda are held by individual etxe and Cayolar are held by a collective) or use (Borda barns are used to shelter animals and to store hay from their associated meadows while Cayolar barns are used for shelter only). To evaluate the differences between these two types of inholdings, we used ArcGIS Pro 3.2 and the 5 m² DEM to calculate slope and construct a topographic position index (i.e., deviation from mean elevation) both resampled to a 250 m window (De Reu et al., 2011). We then calculated distance to nearest stream and the amount of pasture vs. other land use types recorded in the 1830 cadaster (Coughlan, 2014) within 250 m of private Borda barns and Cayolar huts and barns. These were then used as independent variables in a simple binomial Generalized Linear Model analysis in SPSS v.29 of differences between the two farming infrastructures (Cayolar = 1, Borda = 0). Significant variables in our model include elevation, slope (250 m), and access to pasture within 250 m (Table 3). Cayolar as a class are on higher, steeper terrain with greater access to pasture than Borda even when limiting the analysis to the elevational zone in which Cayolar and Borda co-occur (c. 640–1,050 m asl).

The Cayolar organization is the day-to-day operational unit of pastoral activity on the high mountain commons in the Soule Valley. The nested assembly levels of free houses represented by the maîtres de maisons, however, hold the authorities of supply, commitment, and monitoring of collective action on the commons (Ostrom, 1990). While the regime operates within the framework outlined in the Coutume de Soule published in 1521 CE during the reign of Francis I (1515-1547 CE), this is not a “pure” Basque institution of governance. The western Pyrenees were incorporated into the administrative structure of the Kingdom of France in 1461 CE during the reign of Louis XI (1461–1483 CE). The institutional details in the Coutume were assembled in response to a 1483 CE royal order that charged royal representatives to record oral and customary practices (not impose normative French law) from all local communities of interest across the realm.

The organizational principals contained in the Coutume, if not always the details, are common to Aragonese, Basque, Béarnaise and Gascogne valley communities either side of the western Pyrenees Mountains. Like Soule, these valley communities are recognized as customary countries—pays coutumier in France and comunidad de villa y tierra or comunidad de aldeas in Spain. The first written accounts of oral tradition and customary practice in this region (called coutume in French and fuero in Spain) date to the 12th century (Noussy Saint-Saëns, 1955; Lefebvre, 1963; Cursent, 1998; Vivier, 1998; Fernández Mier and Quirós Castillo, 2015). A unique period document (Urrutibéhéty, 1983) from a court case on August 1, 1455 CE, references a public ordinance from August 15, 1395 CE that stipulates the inhabitants of Larrau were exempt from tribute and had free use of forests and pasturelands since the establishment of Larrau in 1174 CE.

There is also proxy material evidence of the Soule Valley institutional framework in the form of legacy structures. For example, the National Convention of 31 October 1793 CE suppressed parish-communes across France and created the communes still used today. When geometricians collected the field information to develop the national cadaster, in nearly every case in southern France and with known certainty in Soule, they simply recorded the geographic properties of the existing parish-communes, agricultural parcels, commons, etc. (Clergeot, 2007; Motte and Vouloir, 2017). At least five parish-communes in the Soule Valley have been in existence since the 11th century CE, and all parish-communes and many etxe with their associated stem families have been stable since at least 1377 CE (Noussy Saint-Saëns, 1955; Urrutibéhéty, 1983; Cierbide, 1994). The earliest record of a Cayolar operating on the commons within the territory of the parish-community of Larrau dates to 1024 CE and it has operated continuously up through the present on the same inholding and under the same name (Coughlan et al., 2023). The first mention of the Silviet dates to the 13th century (Noussy Saint-Saëns, 1955, p. 339; Lafourcade, 2010b).

The Cayolar organization in Soule lies within a class of pastoral institutions common to the western Pyrenees and Cantabrian mountains, and comparable institutions identified among mountain herders across Europe (Poumarede, 1984; Couturier, 2000; Agnoletti, 2007; Bordessoule, 2007). These pastoral institutions take the form of an annually iterated n-person public goods assurance game in which the contribution of each member is essential to the production of resource units. The most obvious resource units produced by members of a Cayolar organization are sheep wellbeing and cheese, but a variety of secondary benefits also derive from this alliance of herders (e.g., friendship, assistance-in-time-of-need, etc.). A Cayolar is an enduring place-based organization as demonstrated by our chronology from multiple sites across the commons indicating a use-span of 850 years for the infrastructure on Cayolar inholdings. Rarely is documentary evidence for the use of a Cayolar inholding available much before the 16th century with the occasional exception as the just noted Cayolar that has been operating for 1,000 years (Coughlan et al., 2023).

Cayolar organizations endure because the members have a regulatory interest in enforcing the contribution of others in the collective herding efforts and little incentive to defect since they must trust each other to cooperate (Heckathorn, 1996; Kitts, 2006; Takács et al., 2008). Hardin (1968) parable speaks to the conflicts of interest and challenges of coordination that undermine cooperation among herders leading to a future tragedy for all as each individual overexploits the grazing commons in the present. Unlike Hardin's herders, herders in a Cayolar share a past and expect to share a future as members of the Soule Valley community of interest. More immediately, each herder represents an etxe that holds proportional title to the Cayolar and contributes the sheep share the herder is cooperatively managing with the other members of the organization. Herders in both instances attribute less value to future benefits than they do to present benefits. However, Hardin's herders exist in the moment and discount the future so severely that they choose to capture all their benefits in the present. The Cayolar herders, on the other hand, do not discount the future as severely and are willing to forgo some present benefits because they or the members of their etxe can be expected to benefit in the future. The Cayolar herders are also likely to have other opportunities to benefit by virtue of the alliances they have with the other members of the Cayolar.

Our homophily test parallels the logic of how discount rates and the tradeoffs between present vs. future benefits influence a herder's decision to collaborate with others. Homophily, sociality and ethnicity are traditionally viewed as the global property of a population and thus unsuitable for explaining the agency of individuals or the alliances they form with others. ERGM reveals how an aggregate phenomenon such as homophily emerges from the tendency of variably conditioned individuals to form links with similarly conditioned neighbors (Granovetter, 1978; Goodreau et al., 2009). Netting (1981) offers a reduced set explanation for affiliation in collective herding in the Swiss Alps that is equally applicable to Cayolar herding in Soule: herders depend on each other to avoid labor shortages that would arise if households were forced to tend their own herds alone.

As for the placement of Cayolar inholdings, the results do not confirm but they do suggest the Cayolar organization emerged as a way to govern grazing lands unsuitable for the cultivation of crops or storable forage. Theory suggests that elevational constraints guide the settlement ecology of household vs. collectively owned farming infrastructure (Rhoades and Thompson, 1975; Netting, 1981). We previously analyzed parcel land use change and etxe abandonment between 1830 and 1958 CE (Coughlan and Gragson, 2016). The results reveal the complex processes by which etxe farmhouses within the Borda zone were functionally repurposed as Borda in response to changing socioeconomic conditions. There is no hard rule governing what constitutes high pasture or which terrain falls under the purview of the Cayolar institution. Temporal variability in elevational constrains on farming are climate-driven and subject to change, thus, the degree of elevational co-occurrence of the two institutions suggests flexibility over time in whether a place functioned as a Cayolar or a Borda. Figure 10 summarizes elements of this spatial flexibility in the Cayolar system.

There are no indigenous Basque accounts before the 9th century CE and the few period documents that do refer to the Basque are often written by individuals in open conflict with them (Larrea, 1998, p. 114). While there is physical, documentary and proxy evidence for Cayolar-based herding at high elevation in Soule beginning c. 1000 CE, neither we nor others have yet to identify structural remains conclusively associated with pastoralism above 800 masl prior to this date. We previously established that charcoal production associated with the establishment and maintenance of pastures above 800 masl in Soule increases dramatically after 1000 CE following a production hiatus that began c. 1000 BCE (Coughlan et al., 2023). The increase in charcoal production occurs in tandem with the establishment of farming households below 800 masl organized into neighborhood communities (Coughlan and Gragson, 2016; Gragson et al., 2021).

The well-known accounts of pastoralism by Lefebvre (1928, 1933), Cavaillès (1931a,b), Ott (1993) and others largely limit their analysis of the Cayolar organization to the operational level (Ostrom, 1990). When pastoralism is further colored by the debate between legal experts and social historians the organization is both taken out of time as well as separated from evidence about individual agency and affiliation networks so that it is not possible to understand the endurance of this organization over time. Ostrom shattered the conviction that local collective dilemmas could only be solved by external authorities imposing private property and centralized regulation. The emergence and persistence of Basque institutions for self-regulating grazing of the commons reveal their capacity to shape and reshape the situations in which individuals make day-to-day decisions and bear the consequences of their actions. This is precisely the space we need to consider in order to investigate the process of change over time in collective action.

Documentary evidence indicates that Basque settlement was impacted over time by military activity at the onset of the Roman era in 180 BCE, the Germanic invasions beginning c. 400 CE, military action between Franks, Visigoths and Vascones along with Bagaude/brigand marauding between 500 and 700 CE, and the Moorish invasion of 711 CE (Larrea, 1998). Major Basque settlements starting in Antiquity such as Pamplona and Calahorra were on the plains adjacent to the Ebro River while extensive agro-pastoral activities were concentrated in the pre-Pyrenees valleys of Guipuzcoa and the inland territory of Álava (Larrea, 1998; Martín Duque, 2002a; Pavón Benito, 2018; Quirós Castillo, 2020). After 700 CE, Basque expanded their territory from the Ebro Basin in Iberia across the western Pyrenees to the Aquitaine Plain in France, which was then referred to in period documents as Guasconia (Fredegar, 1981). Shortly thereafter, the Ravenna Cosmographer (Anonymous, 1860) produced the first map showing Basque lands as spanning the Pyrenees up to the Adour-Ouse River corridor in France.

After the Basque defeated the Carolingian army at the Second Battle of Roncevalles in 824 CE, Íñigo Arista established the Basque kingdom of Navarra: c. 5,000 km² centered on Pamplona with close ties to the Catholic Church against their common enemy, the Moors (Martín Duque, 2002b; Pavón Benito, 2018). Íñigo Arista set the course for reopening the Soule Valley to agropastoral use by making donations that combined with donations from Wilesindo, Bishop of Pamplona, led to the establishment in 842 CE of the Benedictine monastery of Leyre (in Navarra). Few details are available, but Leyre is known to have operated a significant pastoral enterprise in the small valley of Bezula within the territorial limits of what would become the parish-commune of Larrau (Brocas and Legaz, 2005). The monastery of Leyre was later acquired by Cistercian monks who with subsequent land donations initiated large scale sheep herding in the mountain reaches of the kingdom of Navarra including Soule. The Cistercian monks from Leyre were followed into Soule by Cistercian monks from the Sauvelade Abbey in the lower reaches of the Soule Valley in France while other monastic orders established themselves in the western Pyrenees in areas surrounding the Soule Valley (Cursent, 1994; Larrea, 1998; Unzu Urmeneta, 2021).

Following a series of land donations with grazing rights for large and small livestock, Larrau was established in 1174 CE as a priory of the Cistercian Abbey of la Sauvelade; the church of Sanctus Johannes de Larraun was built in 1193 CE as a chapel for the ospitau de Larraun that served as the nucleus of what would become the parish-commune of Larrau (Urrutibéhéty, 1983). Our chronology and documentary evidence for Cayolar-based activities on the commons within the territorial limits of the parish-commune of Larrau pre-date by over 150 years the construction of the church and establishment of the population center of Larrau. Understanding the process of settling down in the western Pyrenees and Europe have been intellectually dominated by the argument that agropastoral spaces are the inevitable consequence of first establishing a hamlet, village or town. The sequence of events in Larrau and comparable settings in northern Iberia such as Tobillas (Quirós Castillo et al., 2023) point to just the opposite. It is land use and management activities that lead eventually to the materialization of alliance in the form of a settlement and its social institutions.

The understanding of settling down even when individual agency and alliance networks are acknowledged has been colored by assumptions about the length of time it must take for such a phase shift to occur. However, without refined chronologies we deny people their history and render them powerless (Whittle, 2018). Monasteries turned their attention to the Pyrenees Mountains as the Catholic Church issued its “Peace of God” policy in response to the civil unrest that had prevailed in southern France since at least 840 CE from the abuses perpetrated by the landed gentry on the peasantry. One consequence of the policy was the establishment of safe-have (sauvete, FR) churches, towns, and castles that guaranteed safety to fugitives (e.g., Sauvelade Abbey).

While safe havens were developing to the north of the Pyrenees Mountains in France, there were equally significant events unfolding to the south in Iberia. The collapse of the Califate of Cordoba in 1032 CE led to the emergence of multiple taifas (i.e., Moorish kingdoms) who paid an annual protection tax (paria, Spanish) in gold to Christian kingdoms such as Navarra. King Sancho Ramirez of Aragón and Pamplona (1042–1094 CE) and king Alfonso VI of Castilla (1040–1109 CE) used this tax to underwrite the Fuero de Jaca and the Fuero de Estella. These primordial fuero encouraged resettlement and repopulation of the Pyrenees Mountains by creating safe corridors through the mountains (Martínez González, 2012) and reducing the initial cost of developing the institutional structure for individuals to collaborate for common benefit. The result was dramatic local economic development and a significant increase in regional exchange between the Kingdoms of Navarra, Aragón and Castilla in Spain and Béarn in France.

The Fuero de Jaca (Aragón) and the “Fuero” de Estella (Navarra) were direct royal agreements with “personas libres, francas e ingenuas” (SP)—freemen able to hold property free and clear who worked the land with their own hands (Orella Unzué, 2017; Pavón Benito, 2018). The king would personally sign a fuero with representatives of a community of interest bypassing local elites, and personally promised protection for pilgrims, travelers and merchants who agreed to use the routes the king specified for crossing the Pyrenees including in the western Pyrenees the passes of Larrau, Somport and Ibañeta. Individuals who agreed to these terms were exempted from all taxes and tribute and given the authority to form councils to decide local affairs; the king in turn would sponsor construction of bridges, lodges, and hospitals using the paria taxes. The concessions and infrastructure in turn attracted representatives from diverse professions and guilds including money changers, master builders, merchants, and hoteliers accelerating the establishment of churches, monasteries, hospitals and way stations.

Two examples clarify the scale and speed of the social, economic, and political transformation in the western Pyrenees from overcoming conflicts of interest and the commitment to enforce rules supporting collective action. The Order of Cannons Regular of Prémontré formed in 1120 CE put in place at least 34 establishments in southern France including the Basque and Gascogne regions in the western Pyrenees between the 12th−13th centuries CE (Abadie, 2019). The kingdom of Béarn (France) adjacent to Soule and the kingdom of Aragón became close allies c. 1100 CE. To facilitate and protect pilgrims, traders and troops following the old Roman road from Lescar (France) to Zaragoza (Spain) crossing the Pyrenees by the Pass of Somport, the partners established the order of Sainte-Christine du Somport. By the early 13th century, the order included over 20 religious establishments linked to agropastoral domains distributed across the Ossau Valley and the eastern portion of the Soule Valley in France, and the adjoining valleys of Salazar, Roncal, Ansó and Echo in Spain (Lassègues, 2019).

There is a tendency to view monarchs through Hobbe's parable of man in a state of nature who seeks his own good yet ends up fighting with other men only avoiding ruin by the coercive, outside force of Leviathan. The primordial fueros of King Sancho Ramirez and King Alfonso VI are an example of the benign model of select benefits for the provision of public goods between advantaged and disadvantaged social groups (Van Zomeren and Iyer, 2009; Jagers et al., 2020). Opposing the categories of mobile and settled has obscured our grasp of what settling down means to individuals with agency capable of autonomous decision-making. In the same vein, we still lack an evidence-based understanding of the temporal causes of collective action involving a wise and advantaged king contracting with persecuted and disadvantaged individuals to the mutual benefit of both (Thomas et al., 2020).