Reconstructing 290 years of a data-poor fishery through ethnographic and archival research: The East Pacific green turtle (Chelonia mydas) in Baja California, Mexico

2.1 Study area

At a regional level, the study area comprises ~14,400 km² in the central desert, and comprises two important C. mydas feeding areas with key contributions to the 20th century fishery in the modern-day communities of Bahía de los Ángeles, Baja California (28°57′N, 113°33′W), on the Gulf of California, and Guerrero Negro, Baja California Sur (27°57′N, 114°3′W), on the shores of Laguna Ojo de Liebre (hereafter, Laguna Ojo de Liebre). These constitute the two primary study locations. Both sites are warm-temperate feeding areas where C. mydas is the predominant sea turtle species, have a shared cultural and economic history, and were important contributors to the commercial green turtle fishery during the 20th century (Early Capistrán, 2014b; Koch, 2013; Seminoff, Resendiz, & Nichols, 2002). The study area also includes the adjacent regions historically under the administration of the missions of San Borja and Santa Gertrudis in the 18th and 19th centuries. Additional fieldwork was conducted at said mission sites and the former mining communities of El Arco and Campo Alemán (Figure 1).

2.2 Ethnography

There is an important body of work on the use of fishers' knowledge (Johannes, Freeman, & Hamilton, 2000; Sáenz-Arroyo & Revollo-Fernández, 2016; Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera, 2005) and local ecological knowledge for fisheries research (Beaudreau & Levin, 2014; Huntington, 2000). We have built upon the methods developed by Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera (2005) and Sáenz-Arroyo, Roberts, Torre, Cariño-Olvera, & Enriquez-Andrade (2005) for quantifying fishers' knowledge gathered through semi-structured interviews, by incorporating an ethnographic approach—in terms of methods and epistemology (Bernard, 2011; Denzin & Lincoln, 1994; Guber, 2015)—to data collection and analysis. We used ethnography to gather detailed, long-term information which informed our parameter calculations, provided hard data for capture reconstructions, and helped provide broad narratives of environmental and social change. Furthermore, we hope to advance the integration of culture to marine historical ecology (Anderson, 2006; Bolster, 2006; Van Sittert, 2005).

Ethnography is a holistic approach to the study of a social system, which includes qualitative and quantitative methods and has distinctive epistemological characteristics (Bernard, 2011). Ethnographers study social systems, rather than isolated phenomena (Guber, 2015; Harris, 2001). This requires an open-ended approach, in which data are gathered broadly over topic areas and new questions are continually developed over the course of fieldwork (Guber, 2015). Ethnographers attempt to understand social systems from an “emic” perspective: from the ethnographic contributors' point of view, based on their explanations, categories and observations. This requires establishing rapport with communities, working with sensitivity to the social group's rules and norms, and developing an understanding of the social system on its own terms. Ethnographers also include “etic” perspectives: the researcher's accounts, categories and explanations (Harris, 2001). This requires ethnographers to collect data, comment on both facts and data collection, and carry out meta-analysis of both processes (Table 1). These analyses and meta-analyses help identify biases, both those of the ethnographic contributors and the researchers (Bernard, 2011; Guber, 2015).

Ethnographers use a varied toolkit distinguished by participant observation, in which the researchers immerse themselves in a social group as an active participant during extended periods of time (weeks, months or years; Table 1). Over the course of 106 working days and 1,696 person-hours of ethnographic fieldwork in 2012 and 2013, two of us (M.M.E.C. and G.G.M.) conducted participant observation and informal (n = 186), semi-structured (n = 33) and in-depth interviews (n = 20) in the communities of Bahía de los Ángeles and Guerrero Negro, compiled 2,003 pages of field journals, video recordings (n = 63), audio recordings (n = 59), historical photographs (n = 31), ethnographic photographs (n = 212) and collaborative maps (n = 32). All audio recordings, video recordings and photographs were gathered with contributors' informed verbal consent. We recorded field notes and journals in as much detail as possible and covered all observations, beyond the principal research topics. We systematized, coded and indexed all data captured in the field, and separated observations from analysis and commentary (Bernard, 2011; Denzin & Lincoln, 1994; Table 1; Supporting Information, Section 1, Table S1). Fieldwork was carried out in accordance with the Code of Ethics of the Latin American Society of Ethnobiology (Sociedad Latinoamericana de Etnobiología 2014).

Through a deliberate hierarchical sampling method, we worked in-depth with experts on green turtle fishing, commerce and processing (Bernard, 2011). In each community, we interviewed over 90% of living fishers who participated in the legal sea turtle fishery before 1990, using the above-mentioned methods for broad data collection and integrating recurring questions based on those of Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera (2005) and Sáenz-Arroyo, Roberts, Torre, Cariño-Olvera, & Enriquez-Andrade (2005) to obtain systematic quantitative data on sea turtle captures (Tables 2, 4 and 5; Supporting Information, Section 1, Table S2). We used verification methods such as cross-questioning, independent corroboration of data between contributors and data sources (oral, written, visual, etc.), and electronic data capture, as well as familiarity generated by extended stays in communities (Bernard, 2011; Denzin & Lincoln, 1994; LeCompte & Goetz, 1982). This multifaceted approach generated detailed and cross-referenced information that could not be obtained through closed questions or surveys alone, and helped identify biases by analysing data within the social, cultural and historical context in which they were generated (Bernard, 2011; Denzin & Lincoln, 1994; LeCompte & Goetz, 1982).

We must point out that ethnography has important limitations. Social systems are inherently complex, and the number of variables involved in their observation and analysis makes specific approaches to each study an inherent necessity in ethnographic research: different tools and theoretical approximations are required in each case (Bernard, 2011; Guber, 2015). For example, if we were to conduct this same study across the Gulf of California, with the Comcáac (Seri) nation, we would need at least a functional grasp of a new language (Cmiique Iitom) and would require far more time in the field (a year or more) to understand “emic” categories and touch upon the subtleties of deeply embedded cultural links between humans and turtles (Bernard, 2011; Nabhan, 2003). Ethnography also requires long time spans for fieldwork, data processing (transcription, indexing, categorizing and coding) and analysis (Bernard, 2011; Denzin & Lincoln, 1994). Finally, ethnographic data are primarily qualitative (Bernard, 2011). However, by integrating this approach with existing methods for quantifying fishers' knowledge (Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera, 2005; Sáenz-Arroyo, Roberts, Torre, Cariño-Olvera, & Enriquez-Andrade, 2005), we hope to expand upon the methodological frameworks available in marine historical ecology.

2.3 Historiography

Building upon the methods of marine historical ecology (McClenachan et al., 2015; Sáenz-Arroyo et al., 2006; Thurstan et al., 2015), we carried out archival research in 24 libraries and online archives, analysing texts in Spanish, English, French, Latin and Classical Greek (Supporting Information, Section 2). The last two languages were included because (i) many early descriptions of the American continent were written in Latin, as it was a common language of Early Modern scholarship (Gordin, 2015); and (ii) to better understand the context in which these documents were produced, we consulted works by Classical, Mediaeval and Renaissance naturalists that informed the taxonomic categories and epistemological frameworks used by the Jesuit missionaries in their descriptions of the study area (see Supporting Information for a full list of historical and archaeological sources).

We consulted 263 historical documents. Using a strict selection process described in the following paragraphs, we compiled 31% of these documents for in-depth analysis (n = 83) and used 11% as quantitative data sources (n = 29). We compiled primary sources (n = 57), as well as secondary sources and historical publications (n = 26) covering dates from 1539 to 1976. We read documents critically, analysing their internal and external validity based on hermeneutic and semiotic analysis (Denzin & Lincoln, 1994), with sensitivity to the social, political and historical context in which they were generated and considering the impact of cultural contact, conquest and colonialism as historical processes that can bias texts (Brettell, 1998). We identified sources of bias (observer bias, informer bias and authorial ethnocentrism) by systematically analysing who collected the data; how, why, under what conditions the information was produced or collected; and towards whom the texts were directed (Bernard, 2011; Brettell, 1998; McClenachan et al., 2015).

We restricted quantitative data sources to first-hand accounts based on systematic observation pertaining to the study region or to warm-temperate C. mydas feeding areas in Baja California, using either primary sources or published compilations or scholarly works which met these criteria. These include birth and death records, historical census data, ships' logs, historical scientific literature, commercial records, customs records and mining reports (Tables 3, 4 and 5; Tables S4 and S6). This strict selection process limited quantitative data to a small number of robust sources. Primary sources not based on systematic observation (such as traveller's logs or letters) or related to a broader geographical scope (North-Eastern Pacific or Gulf of California) were read critically and used as qualitative references, along with secondary sources, historical and historiographical publications (See Supporting Information for a full list of historical and archaeological sources). We used qualitative sources to establish a long-term narrative and a theoretical framework for environmental change; to select and define analytical categories and parameters used for harvest reconstruction; to inform parameter calculations; and to corroborate information through comparative analysis.

2.4 Consumption reconstruction

We compiled quantitative and qualitative data on sea turtle captures, consumption, processing and trade—as well as human population demographics—from ethnographic, historical and archaeological sources (Tables 2, 3, 4 and 5) for four broad time periods: Pre-Hispanic (1700–1750), Mission (1750–1850), Secular (1850–1945) and Modern Fisheries (1945–1990). We used these data sources to calculate per capita and aggregate regional sea turtle consumption over time.

Using paleonutritional data, modern nutritional data and ethnographic data, we estimated per capita sea turtle consumption for the first three periods using the following equation adapted from Early Capistrán (2014b):

(1)

where c_t is the approximate annual per capita consumption of C. mydas (turtles person⁻¹ year⁻¹) in year t, Q is approximate annual per capita meat consumption for a human population (kg person⁻¹ year⁻¹), γ is the percentage of annual meat consumption from sea turtles, λ is the percentage of sea turtle tissue consumed, p is the mean weight of a green sea turtle in the region (kg per turtle), and δ is the percentage of change in weight due to processing (Table 4).

We used the mean nutritional value of muscle and adipose tissue of C. mydas (864.3 kcal/kg) to calculate the contribution of sea turtles to local diets (González Olmedo et al., 2004). We calculated the percentage of sea turtle tissue consumed (λ) using percentage values grouped by category (fillet meat, offal, fats, etc.) from a commercial report of C. mydas processing (Márquez, Elizalde, & Nodarse, 1991), and summed the categories used. Values for p were based on scientific monitoring data and corroborated with ethnographic data. Additionally, we calculated δ based on ethnographic data and food processing research (ONU-FAO 1990). We calculated values for γ and λ for different time periods, adjusting for varying dietary patterns among inland and coastal subpopulations. For Pre-Hispanic and Mission Periods, we obtained parameter values from published archaeological research and historical sources; for the Secular and Modern Fisheries Periods, we used published nutritional research (Garry, Passmore, Warnock, & Durnin, 1952; ONU-FAO 2003), historical documents and ethnographic data (Tables 4 and 5; detailed descriptions of parameter calculations are available in Supporting Information, Section 3.1, Tables S3–S5).

For the Pre-Hispanic Period, we used paleonutritional data based on stable isotope analysis for two Cochimí populations (Bahía de los Ángeles and Sierra de San Francisco;King, 1997), in conjunction with ethnohistoric data on Pre-Hispanic diet in the central desert of Baja California (Aschmann, 1959) to calculate dietary composition. These sources register proportional consumption, by weight and caloric density, of different food groups and edible taxa (marine vertebrates, marine invertebrates, terrestrial fauna, legumes, etc.), including sea turtles. We correlated these data with dietary data compiled from hunter-gatherers worldwide (Cordain et al., 2000) to obtain approximate calculations of dietary composition, in terms of kg person⁻¹ year⁻¹ (Supporting Information, Section 3.1.2, Equation S1). In this desert context, many staple plant foods were seasonal (cactus fruits from Lemairocereus thurberi and Machareocereus gummosus) or required extensive processing (such as the hearts of agaves, Agave spp., which are rendered edible only after roasting in pits for a minimum of a day; Aschmann 1959, King, 1997). In contrast, marine resources were productive and reliable, and made up a significant proportion of the diet (King, 1997).

We calculated Q by adding approximate annual consumption values for main sources of animal protein (marine vertebrates, marine invertebrates and terrestrial animals) to obtain approximate annual meat consumption for coastal (500 kg person⁻¹ year⁻¹) and inland populations (192 kg person⁻¹ year⁻¹), and used interpolated weight and nutritional density values reported by King (1997) and Aschmann (1959) to calculate the percentage of annual meat consumption from sea turtles (γ; Table 4). The very high values of animal protein consumption are consistent with a non-agricultural economy, based heavily on the use of marine resources. We corroborated both Q and γ values with 19th century ethnographic reports (McGee, 1898) of the diet of the Comcáac (Seri), an indigenous nation of the Gulf of California, which, like the Cochimí, had a hunting, gathering and fishing economy in a desert landscape (Aschmann, 1959). Given the difficulty of quantifying dietary patterns in through the archaeological record, in particular among hunter-gatherer groups whose diet varied widely in relation to resource availability, this should be considered a broad estimate.

For the Secular Period and Modern Fisheries Period, we based values and parameter consumption on ethnographic and historical data, and adjusted for varying dietary patterns at inland and coastal sites. Dietary patterns had shifted drastically by this period due to the introduction of extensive cattle ranching, small-scale horticulture and non-perishable plant-based food items such as rice, beans, and wheat flour which became staple foods (Crosby, 2010). We calculated annual per capita meat consumption by adjusting mean values for the Baja California peninsula reported by ONU-FAO (2003) for the reported caloric intake of miners, who made up most of the regional population (Garry et al., 1952), such that Q = 97 kg person⁻¹ year⁻¹. We calculated the percentage of consumed meat obtained from sea turtles (γ) based on mean values of frequency of sea turtle consumption obtained through ethnographic research. In coastal communities, sea turtles were a staple protein source consumed up to three times per week (γ = 43%), and an important source of dried meat in inland communities (γ = 7%; Table 4); other sources of protein included beef, fish, marine invertebrates and wild game.

We estimated total annual consumption by multiplying per capita consumption by human population size using the following equation adapted from Early Capistrán (2014b):

(2)

where C_t is the aggregate sea turtle consumption by a human population during year t (turtles per year) and n_t is human population size during year t (humans). For the Pre-Hispanic and Mission Periods, we used demographic data from published archaeological research and historical sources (Supporting Information, Section 3.2.1, Equation S2 ,Tables S3, S4 and S6). We calculated population change outside mission settlements by interpolating late Pre-Hispanic population density data with mission records (Supporting Information, Section 3.2.1), For the Secular and Modern Fisheries Periods, we obtained demographic data from historical documents and ethnographic sources (Supporting Information, Section 3.2.2; Tables S5 and S6). We reconstructed consumption until the approximate peak years of the commercial fishery (1965 in Bahía de los Ángeles and 1975 in Laguna Ojo de Liebre; all demographic calculations and population data are available in Supporting Information, Section 3.2, Tables S7–S9).

We assumed that all captures correspond to C. mydas given the region's importance as a feeding area; regional and global market preference for the species; and the species' condition as the target of the 20th century commercial fishery in the study area, as confirmed by fishers and merchants (Early Capistrán, 2014b; Márquez, 1996; Seminoff, 2010). While hawksbill turtles (Eretmochelys imbricata) were fished commercially in the Gulf of California for their shells, their taste was considered inferior to green turtles, and they were not targeted for human consumption (Márquez, 1996; Sáenz-Arroyo et al., 2006) nor captured systematically in the study area (Early Capistrán, 2014b). We assumed that mean sea turtle weight was constant across time periods. We based our values on scientific monitoring data, corroborated with the mode weight reported by fishers as far back as 1940. We make this assumption despite the possibility that size frequency declined with fishing effort because we do not have sufficient data to adjust for this pattern. However, we consider it to be an appropriate assumption given the limitations of the data.

We assumed that dietary patterns remained stable within each historical period, and that inland and coastal subpopulations had distinct, but stable, dietary patterns. We assumed that sea turtle consumption patterns remained stable from the Pre-Hispanic to the Mission Period for two reasons: (i) the adverse conditions for agriculture resulted in famines rather than broad-scale dietary shifts (Aschmann, 1959; Rodríguez Tomp, 2002); and (ii) because of massive demographic loss during this period, the effect of contingent dietary changes would not have been significant for calculations. For the Secular Period, we assumed that dietary patterns obtained from ethnographic data could be extrapolated as far back as the 1850s, given the region's extreme geographic isolation and confirmation from ethnographic contributors that technological conditions and means of communication had changed little between the 1950s and the previous two generations.

2.5 Commercial reconstruction

We used official landing records when available. However, official data exists only for a series of 20 years (1962–1982) at Bahía de los Ángeles and 20 years (1887 and 1917–1935) at Laguna Ojo de Liebre respectively. These 40 years of official fisheries data represent 7% of the cumulative chronology (290 years at each location, for a total of 580 years). For this reason, we relied primarily on historical and ethnographic data to reconstruct commercial captures, and our methods allowed us to develop a reconstruction where no other data were available. As different sources reported landings in different units (pounds, kilograms and tonnes), all commercial captures were standardized to turtles per year by converting the annual catch volume to kg per year and dividing by p.

For the Secular Period, we reconstructed commercial captures from Laguna Ojo de Liebre using multiple historical data sources. Sea turtles were captured opportunistically by whalers for food and commerce (Drew et al., 2016; Henderson, 1972). From 1858 to 1873, we used published whaling logbooks (Scammon,1859/1970), shipping reports (Daily Alta California 1860, 1871) and published research on whaling in Baja California (Henderson, 1972; Vernon, 2009) to compile data on whaling activity and estimate sea turtle captures by American and Russian whalers in Laguna Ojo de Liebre. We assumed that reported catches were representative of the fleet, and that all catches corresponded to C. mydas based on taste preferences (Henderson, 1972; See Supporting Information, Section 4, Equation S3 for a detailed description of data standardization). To calculate the approximate annual harvest by the whaling fleet in a given year, we developed the equation:

(3)

where R_t is the mean approximate annual harvest by the whaling fleet (turtles per year) in year t, μ_w is the mean approximate annual harvest per ship (turtles ship⁻¹ year⁻¹), and s_t is the number of ships in the lagoon in year t (ships; Table 6). We obtained vessel counts (s_t) from records compiled by Henderson (1972), and used published logs and shipping reports to estimate catch (μ_w) (Daily Alta California 1860, 1871; Scammon,1859/1970).

For years 1887–1935, we used customs and landings data for green turtles imported to California from Mexico—almost exclusively from Baja California—to calculate approximate commercial harvests (Karmelich, 1935; Radcliffe, 1922; True 1887), which we standardized to turtles per year. For most of the 19th and early 20th century, turtle capture was opportunistic rather than the result of a dedicated fishery (Averett, 1920; Karmelich, 1935; O'Donnell, 1974), and documentation for this period was scarce. Import and export records provide centralized information, which we analysed in conjunction with landing reports and commercial publications. We used historical records to establish a narrative of changes in capture, market dynamics and spatial extent, and to estimate the proportion of landed green turtles captured at Laguna Ojo de Liebre over the time period evaluated (Table 6; detailed description of data standardization in Supporting Information, Section 5). Due to the lack of documentation of turtle catch over this time period (O'Donnell, 1974), our estimate should be considered conservative.

For the Modern Fisheries Period, we used official C. mydas landing data for the late-20th century commercial fishery at Bahía de los Ángeles (Márquez cited in Seminoff et al., 2008), dating from 1962 to 1982. Landings were reported in metric tonnes and standardized to turtles per year. Landing data were not available for this period at Laguna Ojo de Liebre. Based on ethnographic data, we assumed that shipment volumes were representative of commercial captures and that all captures corresponded to C. mydas. We calculated the number of turtles shipped annually from the community to urban centres by developing the equation:

(4)

where M_t is the number of turtles shipped annually (turtles per year), V_t is the approximate number of annual shipments (shipments per year) during year t, and K is the carrying capacity of the vehicles (turtles per shipment). K was a constant of 60 turtles and is the mode reported by sea turtle merchants and fishers. V_t was calculated from ethnographic data; we used parameters obtained from ethnographic data to adjust for seasonality in captures and changes in shipment frequency due to changes in infrastructure over time (all calculation procedures and parameter values are available in Supporting Information, Section 6, Equations S4–S6; Table S10).

3.1 Pre-hispanic Period (1700–1750)

During the Pre-Hispanic Period, nomadic hunter-gatherers from the Yuman-Cochimí language family relied heavily on marine resources as a source of protein (Aschmann, 1959; King, 1997; Laylander, 2010), and sea turtles appear as a food source in the archaeological record since the earliest phases of human occupation, at least 12,000 years ago (Des Lauriers, 2006; Ritter, 2012). Stable isotope analysis and ethnohistoric data suggest that for Pre-Hispanic populations in the central desert of Baja California, sea turtles comprised 3% of animal protein consumed in inland regions, and as much as 14% of animal protein consumed in coastal areas (Aschmann, 1959; King, 1997). Marine turtles also appear in artwork and burials, suggesting symbolic or religious importance (Ritter, 1998, 2010a, b).

Our earliest estimate of sea turtle consumption corresponds to two generations before the arrival of European missionaries (c. 1700), based on available paleonutritional and demographic data (Aschmann, 1959; King, 1997; Figure 2a,b). The lack of Pre-Hispanic demographic data from the central desert limits our ability to reconstruct sea turtle consumption before the early 18th century. However, it is likely that post-Pleistocene populations were small and widely dispersed, and within an order of magnitude of those recorded in early ethnohistoric documents (Laylander, 2010). Archaeological and ethnohistoric sources estimate a population of 4,000 people in the central desert—and around 12,000 in the entire peninsula—at the time of European contact between the late 17th and mid-18th century (Aschmann, 1959; Laylander, 2010; Rodríguez Tomp, 2002). We estimated annual consumption values of 535 and 740 turtles/year for Bahía de los Ángeles and Laguna Ojo de Liebre, respectively (Figure 2a,b; Table 4).

3.2 Mission Period (1750–1850)

Jesuit, Dominican and Franciscan missionaries—envoys of the Spanish Crown—were the first Europeans to establish permanent settlement in Baja California, nearly 200 years after the Spanish conquest of the Aztec empire in mainland Mexico (Crosby, 1994; León Portilla, 2001). The Jesuits in particular were among the intellectual elite of their time—versed in philosophy, theology and natural sciences. As such, they left detailed accounts of the social life and natural surroundings of the missions, which had pragmatic value in the logic of Spanish imperial expansion (Crosby, 1994). The mission system was based on the forced sedentarization of the native hunter-gatherers which, coupled with disease and unfavourable conditions for agriculture, led to mass mortality of the indigenous peoples (Table 5). Within two generations of the founding of the missions of Santa Gertrudis and San Borja (Figure 1), the population of the central desert was reduced by 90% (Rodríguez Tomp, 2002), and Pre-Hispanic populations levels were not re-established until the mid-20th century (Early Capistrán, 2014b). Detailed baptismal and census records from the missions of San Borja and Santa Gertrudis allowed us to estimate demographic change and sea turtle consumption. During this period, the massive loss of human life reduced sea turtle harvests to levels lower than those of the Pre-Hispanic Period, and reduced pressure on sea turtle populations for an extended period of time (Figure 2a,b; Supporting Information, Section 3.2.1, Equation S2).

While colonization and agriculture would have caused important dietary shifts marked by increased consumption of plant foods, this period was characterized by famine, and dietary patterns responded largely to the availability of food sources, sea turtles being chief among them (Baegert,1761/1982). In the context of mass human mortality, we consider that the effect of contingent dietary shifts over this period would not have been significant for calculations. However, we recognize that our estimates for this period may be high, as sea turtle consumption may have been reduced as a result of sedentarization.

Taxonomic distinctions between sea turtle species in this period were blurry. Categories used by Jesuits and Spanish naturalists overlap with the three taxa defined by mediaeval naturalists: the hawksbill (E. imbricata) and leatherback (D. coriacea) were recognized as distinct species, and all others were grouped within a single category (del Barco,1757/1988; Longinos Martínez,1787/1994; Rondeletti, 1554). However, we assumed that the bulk of sea turtle consumption consisted of green turtles, as this species is and was the most common at the study sites (Koch, 2013; López-Castro, Koch, Mariscal-Loza, & Nichols, 2010) and is considered the most desirable by modern-day local populations (Early Capistrán, 2014a, b; Mancini & Koch, 2009). Estimated annual consumption ranged from 8 to 757 turtles/year, and median harvest values for this period were 390 and 93 turtles/year in Bahía de los Ángeles and Laguna Ojo de Liebre, respectively (Figure 2a,b; Table 4).

3.3 Secular Period (1850–1945)

After Mexican independence, the secularization of mission lands in Baja California led to large-scale commercial concessions to private, mostly foreign, companies (León Portilla & Piñera Ramírez, 2011; Romero Gil, Heath, & Rivas Hernández, 2003; Table 5). The region was integrated into global capitalism, through an extractive economy tied to the international demand for commodities like whale oil, gold and seafood (Henderson, 1972); however, few permanent settlements were established and population levels remained low for much of this period as a result of demographic collapse during the Mission Period (Henderson, 1972; León Portilla & Piñera Ramírez, 2011). Historical records such as whaling logs, mining reports, scientific reports and census data allowed for a detailed reconstruction of sea turtle harvests during this period. While sea turtles were caught commercially during this period, average annual capture remained within an order of magnitude of the previous century with the exception of two outlying years (1919 and 1925) in which a mining population boom and a short-lived commercial enterprise caused very brief increases in capture.

On the Pacific coast, whales, guano, seals, otters and salt were exploited intermittently by American and Russian fleets (Henderson, 1972; Vernon, 2009). In 1857, whaler Charles Scammon was the first navigator to breach the grey whale (Eschrichtius robustus) breeding grounds at Laguna Ojo de Liebre (known in English as Scammon's Lagoon). From 1858 to 1873, whalers flocked to the previously untouched whaling grounds (Henderson, 1972). While sea turtles were not the main target species, they were captured opportunistically for subsistence and commerce (Drew et al., 2016), and green turtles from Laguna Ojo de Liebre and the Pacific coast of Baja California were sold at luxury restaurants in San Francisco and as far away as Chicago (Daily Alta California 1860, 1871; O'Donnell, 1974). Green turtles, in particular, were considered a delicacy in the United States and Britain, and had been exploited commercially in the Caribbean since the 1700s for sale in cities like Boston, New York, and London (Anson, 1748; Jackson et al., 2001; McClenachan et al., 2006). Due to the opportunistic nature of turtle capture, harvest was highly variable. However, given the intermittent and short-lived whaling activity—which ended in <20 years as grey whale populations collapsed—we estimate that overall catch by whalers was relatively low: the estimated median value for annual commercial harvest from Laguna Ojo de Liebre during this period was 43 turtles/year (Table 6).

The California gold rush drew attention to Baja California, and in the late 19th and early 20th century, gold, silver and copper mines were tapped by American, British and Mexican investors (Goldbaum,1918/1971; Romero Gil et al., 2003). Mining led to massive demographic shifts through a “boom and bust” economy, in which cities were established around veins and abandoned as mineral resources dwindled (Early Capistrán, 2014b; Romero Gil et al., 2003). Sea turtles were an important source of protein in mining communities, mainly in the form of salted meat and jerky. This processing method used only fillet meat, which lost up to 80% of its volume due to processing. Additionally, in contrast with fresh turtle consumption, edible organs and most fats were discarded. This processing pattern led to increased local consumption compared to previous years, which is particularly noticeable in 1925, when the mining towns of El Arco and Calmallí reached a peak population of ~1,000 residents (Figure 2a,b; Table 4).

Between World Wars I and II, sea turtles were fished commercially for export to California, USA, from the Pacific Coast of Baja California in years 1917–1923 and 1927–1932 (Averett, 1920; Nelson, 1922). As mining and railroad fortunes accumulated in California, investors tried their hand at importing East Pacific green turtles to high-end restaurants in San Francisco and San Diego. Large investments were made, including a canning facility in San Diego. This enterprise ran at full capacity from 1919 to 1921, when the schooner Catarina shipped up to 1,000 turtles a month from Laguna Ojo de Liebre during peak seasons (Averett, 1920; Karmelich, 1935; Nelson, 1922; Table 6). The magnitude of captures generated by this venture briefly raised concerns about the future viability of the fishery (Nelson, 1922; O'Donnell, 1974). However, the schooner shipments to San Diego ended in the early 1920s, presumably due to a lack of market demand in California, and as a result landings were reduced substantially (Karmelich, 1935; O'Donnell, 1974).

By the 1930s, turtle landings in California were limited to “one or two boats” that occasionally made shipments to San Diego, “but these are so spasmodic that a constant market cannot be maintained, with the result that the fishermen find it difficult to dispose of their catches whether large or small” (Karmelich, 1935). An account from 1931 describes the landing of 50 green turtles from Scammon's Lagoon at San Diego, on board the fishing boat “Vigilant.” For 20 days, the crew “strove valiantly to dispose of the fare,” but eventually 41 of the 50 turtles were shipped back to Mexico for lack of buyers, two were sold to “select dining resorts” in San Diego, and the rest were “butchered on board and retailed from the deck to Mexicans who came down for a piece of their favourite seafood” (The West Coast Fisheries 1931). The venture was described as “a failure, financially, and will not be repeated” (The West Coast Fisheries 1931). This is consistent with a reduction of sea turtle consumption in the United States towards the mid-20th century (Freedman, 2007). With the exception of the outlying year 1919, when ~2,686 turtles from Laguna Ojo de Liebre were imported to California, we estimate that annual commercial harvest in the early 20th century remained within an order of magnitude of captures in the past centuries (Table 6).

Local subsistence captures were carried out with harpoons, from wooden vessels powered by oars or paddles. Several factors limited fishing efficiency: the harpooners' ability (skill limited the number of turtles potentially caught per trip); weather, tides and lunar phases (their status limited the days when harpooning was viable: ideal conditions required calm seas and winds on a neap-tide, and moderate moonlight); propulsion (which determined trip duration and spatial extent of fishing); navigational knowledge and experience (which was based on triangulation, dead-reckoning, and celestial observations with limited instruments and required great expertise); and vessel capacity (open wooden vessels held no more than 20 turtles).

Additionally, commercial capacity was inhibited due to a limited market access because of (i) the isolation of the fishing sites (there were no urban population centres within 500 km); and (ii) lack of transportation and communications infrastructure including roads and telephones, respectively. In coastal communities, capture was limited to what could be used, and practically, none of the turtle was wasted: meat, offal, and blood were all consumed, and even the carapace could be boiled down to a gelatinous consistency and eaten, while oil was rendered for cooking and medicinal purposes. Bones were boiled in broth and then given to domestic dogs. The head and skin were the only by-products not considered fit for human ingestion and were left out for dogs and coyotes. Consumption patterns with minimal waste continued to be the norm in fishing communities throughout the 20th century. Sea turtle consumption ranged from 1 to 1682 turtles/year, with the maximum value corresponding to the year 1925. Median harvest values for local consumption were 71 turtles/year in Bahía de los Ángeles and 505 and in Laguna Ojo de Liebre (Figure 2a,b; Tables 4, 6).

Chelonia mydas nests mainly on tropical beaches, and nesting activity in the warm-temperate study area is rare (Koch, 2013; Seminoff, 2004). As a result, eggs were not traditionally consumed, and only 9% of fishers recalled having tasted sea turtle eggs at some point. Additionally, areas surrounding key nesting beaches in the Mexican Pacific were geographically isolated and sparsely populated until the second half of the 20th century. For example, there were no permanent human settlements near the most important green turtle nesting beaches of Colola nor Maruata, in Michoacán, until the 1950s, and egg harvests at these key nesting beaches were minimal (Alvarado & Figueroa, 1992; Clifton et al., 1995; Márquez, 1996).

3.4 Modern Fisheries Period (1945–1990)

Urban growth along the Mexico-U.S. border increased demand for sea turtle products: from 1940 to 1970, the population of the state of Baja California increased by 1100%, mostly in cities along the Mexico-U.S. border such as Tijuana, Ensenada and Mexicali (Instituto Nacional de Estadística, Geografía e Informática 2015), which became the main markets for green turtle products (Figures 1 and 3). Fishing and commercial capacity grew thanks to new technologies: gillnets eliminated the need for skilled harpooners, increasing catch efficiency; fibreglass vessels boosted carrying capacity to 30 or more turtles per boat; and outboard motors greatly increased the spatial and temporal extent of fishing. The Transpeninsular Highway, inaugurated in 1974, shortened the trip to the Mexico-U.S. border from 2 weeks to 2 days, greatly increasing market access (Early Capistrán, 2014b).

Harvest peaked in the late 1960s and early 1970s, as estimated annual catches exceeded those the past 250 years by an order of magnitude (Figure 2a,b). During this period, we estimate that the median harvest value for local consumption at Bahía de los Ángeles was 282 turtles/year, compared to a median commercial harvest of 2,370 turtles/year. At Laguna Ojo de Liebre, the median harvest value for local consumption was 922 turtles/year, in contrast with a median commercial harvest of 5,220 turtles/year (Figure 2a,b; Table 4).

Unregulated harvests led to swift declines in green turtle abundance in the Eastern Pacific, reflected in nesting data and descriptions of population levels. Gravid females were captured in the fishery and, simultaneously, settlements and roads were built around key nesting beaches in Michoacán that had previously been unpopulated or harvested at subsistence levels (Clifton et al., 1995; Márquez, 1996). During the 1960s and 1970s, close to 100% of eggs were harvested until index beaches were protected in 1980 (Clifton et al., 1995; Márquez, 1996). While further information on recruitment patterns and stock composition is needed to directly evaluate the impact of egg harvest on C. mydas populations in the study area (Bjorndal & Bolten, 2008; Casale & Heppell, 2016; Koch, 2013), this process undoubtedly contributed to declines in abundance.

State intervention increased throughout the 1970s through licence restrictions, seasonal bans, nesting beach protection and re-population programs (Early Capistrán, 2010; Márquez, 1996; Seminoff et al., 2008). Unfortunately, these efforts came too late: the commercial green turtle fishery collapsed in the early 1980s (Figure 2a,b; Seminoff et al., 2008). A nominal ban on captures of C. mydas in 1983 was followed by a total ban on sea turtle captures in 1990, which remains in effect today (Márquez, 1996; Secretaría de Medio Ambiente y Recursos Naturales 2010).

4.1 Subsistence vs. market economy

Estimated annual sea turtle capture increased by an order of magnitude due to demographic and economic shifts, both at regional and international scales. Furthermore, technologies such as gillnets, outboard motors and fibreglass vessels increased fishing efficiency. Additionally, improved infrastructure increased market access. From 1940 to 1970, the population of cities along the Mexico-U.S. border grew almost exponentially (Figure 3; Instituto Nacional de Estadística, Geografía e Informática 2015). Border cities became the main markets for green turtle products (Figure 1), and sea turtle restaurants and stands—known locally as caguamerías—were regularly supplied with green turtles from the central peninsula. Caguamerías became immensely popular, to the degree that tacos and other street foods are today in Mexico. This unregulated market led to a fast decline in sea turtle populations, in contrast with the local subsistence captures which had been limited by small human populations and minimal waste and had proved sustainable over long time spans.

The pattern of marine resource depletion as a result of national and international market dynamics has been repeated worldwide since the early days of capitalism (Langton, 2003; Roman & Palumbi, 2003; Schwerdtner Máñez et al., 2014). This has also been the case with sea turtle fisheries in other locations in Mexico, as well as the Caribbean, where captures were mainly for non-local luxury markets (Costa-Neto & Márques, 2000; Early Capistrán, 2010, 2014c; Nietschmann, 1974). Cinner et al. (2016) showed that market gravity—a metric of potential interaction with urban centres or markets measured in terms of the relative size of markets and their distance from fishing communities—is the strongest predictor of reef fish biomass loss: more so than population pressure, environmental conditions or national socio-economic context. Similarly, a strong correlation has been found between the demand for megavertebrates in international luxury markets and extinction risk (McClenachan, Cooper, & Dulvy, 2016). Furthermore, McClenachan and Kittinger (2012) found that contrasting social and historical trajectories greatly affect the long-term sustainability of fisheries, and that high economic connectivity and human population density, coupled with a lack of customary management systems, caused rapid overexploitation of marine resources.

We suggest that market forces were the main driver of the green turtle fishery collapse in the temperate feeding areas of Baja California. Decline was not caused by local subsistence fishing, but by a combination of (i) unprecedented and unregulated demand from urban centres; and (ii) resulting supply in the form of increased sea turtles capture made possible by improved fishing efficiency and market access. Demand increased in response to demographic and economic growth along the Mexico-U.S. border. Simultaneously, supply increased as technologies such as gillnets and outboard motors improved fishing efficiency and improved infrastructure increased market access.

4.2 Turtles in time

In broad terms, human impacts on large marine vertebrate populations have shown a similar pattern worldwide: slow changes over millennia, rapid depletion in recent centuries, and accelerated decline in the 20th century (Jackson et al., 2001; Lotze & Worm, 2009). For example, Caribbean green sea turtle populations were decimated by large-scale commercial fisheries for export to Europe as early as the 18th century (Bjorndal & Jackson, 2003; McClenachan et al., 2006). This was due greatly to the Caribbean region's fast integration into the global economy, and relative proximity to European markets with important demands for sea turtle products (Nietschmann, 1974). According to sea turtle expert Archie Carr, “more than any other dietary factor, the green turtle supported the opening of the Caribbean” (Carr cited in Nietschmann, 1974). Remnant populations were exploited throughout the 20th century, and technologies such as nets and outboard motors permitted more efficient captures and accelerated their decline (Nietschmann, 1974).

The central desert of Baja California presents a different trajectory: we suggest that the turning point in human impact was much more recent, in the 1960s, when estimated annual captures exceeded those of the previous centuries—including late phases of Pre-Hispanic occupation—by an order of magnitude. We consider that colonization processes, economic cycles and geographic isolation had important roles in this unique scenario. First, as the American continent was colonized by Europeans, in broad terms, from east to west and south to north (in the case of the Northern Hemisphere), Baja California was colonized centuries later than the Caribbean islands or mainland Mexico: Jesuit missionaries did not establish permanent settlements in the peninsula until nearly 200 after the fall of the Aztec empire (Crosby, 1994; León Portilla, 2001). By the time colonial presence was first established in the peninsula, much of Latin America and the Caribbean were thoroughly integrated into a global economy; however, due to the adverse conditions in the desert peninsula and its geographic isolation from the colonial metropolis, trade to and from the peninsula in general, and the central desert in particular, was scarce during the 18th and early 19th centuries (Baegert,1761/1982; Crosby, 1994; León Portilla, 2001; Linck & Burrus, 1967).

From the 1850s until the 1950s, an extractive economy based on mining, whaling and fishing had important impacts on the region (Henderson, 1972; Piñera Ramírez, 1991). However, the lack of a constant market for sea turtle products in urban centres and the small local populations kept impacts on sea turtles mostly within an order of magnitude of past centuries. This is supported by reports of very high abundance from the late 19th century until the early 1960s (Averett, 1920; Caldwell, 1963; Townsend, 1916). It was not until the 1960s that a confluence of factors—market demand in new and accessible urban centres coupled with increased infrastructure and catch efficiency—led to swift declines.

We do not wish to imply that a “pristine” baseline exists at any point in the chronology. Long-term abundance of resource species has been affected both by human activity and long-term climate fluctuations (Lotze & Worm, 2009), to the degree that any point chosen as a baseline is, to some extent, arbitrary. Furthermore, the idea of the “New World” as a pristine wilderness before the arrival of Columbus is both scientifically unsupportable and embedded in colonial discourse (Denevan, 1992; Kay & Simmons, 2002). Beyond the vast empires of Mesoamerica and the Andes, hunter-gatherers—such as the Pre-Hispanic inhabitants of Baja California—had significant impact on coastal and marine ecosystems centuries before written records exist (Rick & Erlandson, 2009). Archaeological evidence, such as large shell middens, suggests that pre-historic human activity had significant impact on Baja California's marine ecosystems (Des Lauriers, 2011; Laylander, 2010). However, currently archaeological data are insufficient to reliably calculate human impacts on sea turtle populations in early phases of human occupation. In this context, we have chosen to extend our reconstruction as far as sources allow us to do so reliably in order to show processes of change over the longest time span possible.

4.3 Past and present

Green turtle catch rates in scientific monitoring conditions have increased since the early 2000s (Figure 4a,b; Comisión Nacional de Áreas Naturales Protegidas, unpublished data; Grupo Tortuguero de las Californias A.C., unpublished data; Koch, 2013; López-Castro et al., 2010). Populations at nesting beaches have also increased since the early 2000s, with marked increases from 2010 onward (Figure 4; Delgado-Trejo & Alvarado Díaz, 2012; Delgado-Trejo, 2016). This 25- to 30-year time frame corresponds roughly with the approximate generation length of East Pacific green turtles (Seminoff, 2004). These increases have been attributed to a combination of initiatives, including the total ban on sea turtle captures in 1990, along with nesting beach protection since 1980 (Márquez, 1996) and increased involvement of governmental, academic and non-governmental institutions in sea turtle conservation (Koch, 2013; Delgado-Trejo, 2016; Figure 5).

The pattern of collapse in the later years of the fisheries in the 1980s and the increase in the past 10 years are congruent with fishers' perception of changes in abundance (Figure 6). As part of a series of recurring questions, fishers were asked if there were “many fewer,” “somewhat fewer,” “about the same,” “more,” or “many more” green turtles present today as in the years they worked in the commercial green turtle fishery (Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera, 2005; Sáenz-Arroyo, Roberts, Torre, Cariño-Olvera, & Enriquez-Andrade, 2005; Supporting Information, Section 1.3, Table S2). We recognize the inherent limitations of these data, and present them only as an initial exploration of possible tendencies; 59% of fishers aged 40–64 (n = 10) and 36% of fishers 65–89 (n = 5) responded “much more.” This suggests a shifting baseline between younger and older fishers (Pauly, 1995; Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera, 2005). However, the data also suggest a positive overall outlook: none of the fishers considered that there were “many fewer” turtles at present, and all fishers who responded “somewhat fewer” (16%, n = 5) added that green turtles are currently abundant, but below the level of their years in the fishery.

Perceived changes in abundance among older fishers are particularly interesting, as our reconstructions suggest an inflection in long-term abundance in the 1960s. Since older fishers worked in the early years of the commercial fishery—and in some cases as subsistence fishermen in the 1940s and 1950s—they witnessed what could be considered a historical baseline abundance level for these two locations. These observations are vital for future evaluations of conservation status, and carrying out this type of research while older expert fishers are alive is of prime importance (Johannes et al., 2000; Sadovy & Cheung, 2004; Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera, 2005).

Evaluating current and present turtle population levels, conservation status, or recovery is beyond the scope of this study. However, our methods could be used to generate reliable baseline abundance data with which to compare current abundance levels. Further research, in the form of standardized Catch Per Unit Effort (CPUE) comparable to modern monitoring data, is needed to evaluate past and current local abundance in terms of biomass. Additionally, long-term analysis of changes at nesting beaches and changes in population structure are required to evaluate changes at species or regional levels (Casale & Heppell, 2016; Kittinger et al., 2013; McClenachan et al., 2006). Although we cannot evaluate the degree of recovery at present, recent increases provide a positive outlook for this green turtle population, and speak to the success of conservation efforts in feeding and nesting areas. (For interpretation of the references to colour in this figure, please refer to the web version of this article).

4.4 Implications for management

The recent green turtle population increase in Baja California echoes increasing population trends in various C. mydas stocks in the Central Pacific and West Atlantic (Balazs & Chaloupka, 2004; Broderick et al., 2006; Chaloupka & Balazs, 2007; Chaloupka et al., 2008). This shows that relatively simple, widespread conservation efforts, such as protection from human hazards—for example, unregulated fishing and egg harvests—can have a profound impact on population levels of once-depleted green turtle stocks (Chaloupka et al., 2008). Nonetheless, green turtles continue to face threats such as by-catch, poaching, habitat degradation, and climate change (Koch, Nichols, Peckham, & de la Toba, 2006; Mancini & Koch, 2009; Mancini et al., 2011; Seminoff, 2004).

Sound management decisions require solid recovery targets based on reliable information. With organisms subjected to long-term exploitation, we risk underestimating the degree of change by limiting decision-making to recent experimental data (McClenachan et al., 2012; Pauly, 1995; Sáenz-Arroyo, Roberts, Torre, & Cariño-Olvera, 2005; Sáenz-Arroyo, Roberts, Torre, Cariño-Olvera, & Enriquez-Andrade, 2005). Through our reconstruction of past harvests, we are confident that we have determined a point in time, between 1950 and 1960, that can serve as a temporal reference point before large-scale exploitation which can be used in the future to establish baseline abundance and recovery targets by building upon our methods.

4.5 Integrating local and scientific knowledge

This type of research is only possible through the construction of collaborative knowledge between scientists and local experts. A critical approach to non-traditional data sources should not be confused with invalidating the credibility of place-based empirical knowledge, which is based on experiential information accrued over generations, with its own particular epistemologies (Beaudreau & Levin, 2014; Idrobo & Berkes, 2012; Mistry & Berardi, 2016). Invalidating such knowledge without attempting to confront epistemological differences risks creating value judgements embedded in forms of colonial representation (Mistry & Berardi, 2016; Sáenz-Arroyo & Revollo-Fernández, 2016). Rather than seeing place-based empirical knowledge as subjective and arbitrary—in contrast with the perception of science as objective and rigorous—we must make a concerted effort to bridge epistemological gaps, recognizing that all forms of knowledge are value-laden and produced by socially situated actors (Mistry & Berardi, 2016). This dialogue between science and place-based empirical knowledge is of prime importance not only to understanding past ecosystem conditions, but also to facing current and future global challenges such as ecosystem degradation and climate change (Klenk & Meehan, 2015; Mistry & Berardi, 2016).

We must highlight the importance of recognizing and integrating fishers' knowledge as a way of decolonizing conservation. Implementing conservation policies and ideologies based on politically and economically dominant agendas further marginalizes the communities most affected by natural resource depletion, and can potentially cause them great harm (Adams & Mulligan, 2003; Langton, 2003; Mistry & Berardi, 2016). Instead, scientists must take a self-critical and collaborative approach which considers the way people perceive, allocate, and manage their natural resources (Costa-Neto & Márques, 2000; Johannes, 1993; Mistry & Berardi, 2016). When approaching conservation issues, scientists should first engage with the communities that interact closely with the natural environment, rely on it directly for their livelihood most, and are most affected by environmental degradation (Mistry & Berardi, 2016). This also implies respectfully acknowledging and understanding each community's distinctiveness and epistemology—as well as the rules, values, ethics and ways of knowing related to resource use—providing relevant scientific knowledge, and establishing self-determination as a key principle of engagement (Johannes, 1981, 1993; Mistry & Berardi, 2016; Weiss, Hamann, & Marsh, 2012).

4.6 Methodological and epistemological challenges

The use of non-traditional data for population ecology—such as place-based empirical knowledge and the historical record—requires a systematic approach based on tried methods from the social sciences (Baisre, 2016; Taylor, 2013). It requires engagement with communities and sources—placing fisheries and fishing societies in a historical, social, cultural and economic context—rather than approaching contributors and documents as mere sources for numerical data extraction (Anderson, 2006; Bolster, 2006; Harrison, 1997; Mistry & Berardi, 2016). In this sense, participation of trained social scientists is fundamental.

The epistemological challenges of integrating both historical and place-based empirical knowledge into population ecology deserve particular attention (Taylor, 2013). Bridging various modes of knowledge production requires an active engagement and dialogue with anthropology and the philosophy of science—as well as epistemology, phenomenology, hermeneutics and ethics (Alagona, Sandlos, & Wiersma, 2012; Cajete, 2004)—in close collaboration with social scientists and humanities scholars (Anderson, 2006; Bolster, 2006). In our case, this dialogue was facilitated by the inclusion of an anthropologist (M.M.E.C.) and a philosopher (G.G.M.) as part of an interdisciplinary research team. This type of constructive, multidisciplinary collaboration improves the reliability of results and contributes to solving broader theoretical issues.

4.7 Concluding comments

Developing robust estimates of past marine animal exploitation requires a solid interdisciplinary framework along with collaborative knowledge-building with local experts. Through the use of ethnography and historiography, we were able to develop detailed estimates of past green sea turtle capture in a key region of Northwest Mexico. We found that from 1700 to around 1960, sea turtle capture remained within an order of magnitude except for two outlying years (1919 and 1925). During the Pre-Hispanic and Mission Periods, harvest levels changed primarily in response to human demographics and local consumption patterns. During the Secular Period (1850–1945), harvest was driven by global economic trends, such as whaling, mining, and early industrial fishing, but remained relatively low. Between 1960 and 1980, the growth of cities along the Mexico-U.S. border and the growing, unregulated demand for sea turtle products—coupled with increased fishing efficiency and infrastructure—led to overexploitation and green turtle population collapse. These 20 years of market demand led to the depletion of a fishery that had been of fundamental importance for millennia. While recent monitoring data suggest a positive outlook for this green turtle population, further research is needed to evaluate past and current turtle abundance, as well as to monitor conservation status.

Through this regional study, we have developed a methodological framework that can be applied widely to reconstruct past marine animal exploitation patterns in data-poor contexts. This methodology can be used to develop time series for other heavily exploited organisms and may help reconstruct and understand long-term change where ecological or fisheries data are unavailable. By incorporating methods from social sciences to solve the epistemological difficulties entailed by this type of research, we hope to contribute to the development of reliable approximations to the study of long-term change in the oceans. This dialogue between the natural and social sciences, place-based empirical knowledge and the humanities could prove vital for understanding both past environmental conditions and addressing current and future global challenges.