Geographic background

Understanding past camelid herding practices requires characterizing the environmental conditions to which they are adapted. The Andean topography presents a distinct vertical succession of ecozones, with each elevation possessing unique characteristics and economic potentials (Fig 1). As highlighted by Sandweiss and Richardson III [23], these zones range from the arid Pacific coastline to the cold high-elevation sierras, with changes between zones characterized by decreasing daily temperatures and increasing annual precipitation as the elevation rises.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Ecological zones of Peru, ranging from the coastal lowlands (Chala) to the high-altitude mountains (Janca) [10].

The locations of key archaeological sites used in this study are mapped in a north-to-south progression, highlighting their specific altitudinal contexts [24].

https://doi.org/10.1371/journal.pone.0306205.g001

The Chala region on the western Pacific Coast, reaching up to 500 meters, is characterized by arid deserts interspersed with narrow, fertile valleys [25]. These valleys offer some grazing potential for camelids though the predominant arid conditions limit grazing sources, making it suboptimal for herding. Spanning elevations from 500 to 3,500 masl is the "Maize Cultivation Corridor" within the Andes, characterized by its more consistent rainfall and moderate climatic conditions conducive to intensive maize cultivation [1, 26, 27]. At the lower spectrum of this corridor, between 450 and 2,300 masl, is the Yunga ecoregion. This altitudinal zone is particularly suitable for the cultivation of maize (Zea mays), coca (Erythroxylum coca), and pink peppercorn (Schinus molle). While the Yunga does not serve as primary habitat for camelids due to the low density of grazing habitat as compared to the puna grasslands at higher elevations, its prolific agricultural yield offers potential fodder resources such as stored maize or maize stalks in post-harvest fields [2, 28]. The Quechua ecoregion, situated between 2,300 and 3,500 masl, offers a more temperate climate favorable for the growth of domestic staples, such as maize and potatoes. Historically, the Quechua zone has been a nexus of both agricultural and pastoral practices, supporting a significant camelid population that occasionally benefited from the region’s agricultural byproducts [29].

Higher in altitude is the Suni region, spanning 3,500 to 4,100 meters, which boasts rocky landscapes against the backdrop of a cold climate [30]. The harsher environmental conditions make large-scale agriculture challenging, favoring cold-resistant crops such as tubers. While camelids are able to navigate these rugged landscapes, the sparse vegetation demands more strategic herding practices. The Puna zone, situated above 4,100 meters, features expansive high-altitude grasslands and vast plateaus. This region is characterized by its cold temperatures and often harsh conditions, unsuitable for the majority of crops found in lower zones. However, the vast grasslands offer abundant natural fodder for camelids, making it their primary grazing habitat [2, 29, 31]. Beyond 4,800 meters lies the Janca region, the pinnacle of the Andes characterized by craggy mountain peaks and enduring cold and is unsuitable for human and camelid occupation.

With the rise of Andean pastoralism, camelids were herded across these varied ecozones, each offering a unique pasture and fodder profile [2, 13, 32]. Past work suggests that camelids consumed different resources depending upon where they were herded, including in some cases agricultural byproducts, with maize standing out as a prime example [9, 21, 22]. These varied feeding ecologies, reflective of differing herding practices, are captured in the isotopic signatures of camelids, enabling isotopes to provide insights into when and where camelid herding focused on wild vegetation versus cultivated fodder.

Setting and archaeological context

Camelids, specifically llamas and alpacas, were the principal livestock of Prehispanic Andean society [4, 33]. Evidence from archaeological sites suggests that by around 4,500 BCE, early Andean herders had established corrals and seasonal rotational grazing patterns in the puna grasslands [4, 34]. As these practices matured, camelids were selectively bred not just for their meat, but also for their wool, which became a vital textile resource [18, 35]. By 3,000 BCE, the influence of, and practices associated with camelid herding had spread northward into the central Andes. This northward movement, enabled by the adaptable nature of camelids and their utility as beasts of burden, led to the development of expansive trade networks [1, 13].

By 2,000 BCE, camelids emerged as pivotal entities within Andean societies, not only serving economic and practical roles but also attaining symbolic and religious significance [36]. Several studies have provided localized descriptions of herding in the archaeological record [9, 10, 14–22]suggesting that altitude-defined ecological factors significantly influenced camelid management approaches [19, 31].

Isotopic analyses of both archaeological and modern camelids from high-altitude settings have been relatively scarce. From the limited studies available, it is generally proposed that above 3500 meters above sea level, specialized pastoralism involved the full-time management of camelid herds with minimal crop production[13, 37]. These highland herders are thought to have frequently traded with lower elevation communities for necessary crops [1, 11, 38]. Agropastoralism, which combines herding with the cultivation of frost-resistant tubers and grains, is proposed to have been a strategy to mitigate the challenges of the highland environment, characterized by limited arable land and frequent frost events.

Initial archaeological research on camelid herding was primarily focused on the highland pastoral systems, specifically the mid-altitude Quechua and Yunga highland zones, ranging from 500 to 3,500 masl. In these regions agropastoralism necessitates a trade-off between agricultural activities and pastoral responsibilities. The challenges in these areas include preventing overgrazing, necessitating frequent herd movements, and managing time allocation trade-offs between agriculture and herding [2, 32]. During times of agricultural scarcity, the reliance on crops to sustain herds may have compounded economic difficulties, potentially making maize-dependent herds a less viable long-term strategy [4, 11, 39].Localized isotopic studies conducted in Peru’s maize belt from the Early Intermediate Period to the Late Horizon corroborate this agropastoral model, showing that camelids primarily grazed on C3 plants, aligning with their minimal reliance on cultivated crops for fodder [14, 20–22].

In more recent years, the majority of research on camelid herding has been concentrated in coastal regions. Along the coast, from 0 to 500 masl, where there is a scarcity of forage, subsistence herding is thought to have played a more restricted role [10, 17]. Isotopic analyses show that camelids grazing on coastal or low-altitude wild vegetation have higher carbon isotopic values, likely due to a predominance of C4 plants in their diets and significantly higher nitrogen isotopic values resulting from the arid conditions of these areas [40]. The limited vegetation along the coast restricts camelid herding, leading to a reliance, primarily or exclusively, on agricultural products or byproducts for fodder. The exact scope and stability of maize as a primary component in camelid diets across ancient Andean societies remain subject to ongoing investigation and debate [9, 19, 34, 41, 42]. Contrary to prevalent belief, diets heavily reliant on maize may not have been economically viable within subsistence agropastoral economies, particularly in the maize-belt. Heavy dependence on maize makes these systems vulnerable to crop failures, which could lead to simultaneous declines in herd viability in these unpredictable environments if maize were principal fodder for camelids [2, 11, 39]. Additionally, while there is substantial evidence of camelids transitioning from basic dietary staples to playing multifaceted roles within the political economies of coastal areas, the full nature and extent of this transformation are still being extensively studied [16, 19, 41]. The importance of camelids within the Andean economic, social, and cultural structures is clear, but a detailed understanding of how their management evolved across different historical periods and ecological regions—especially contrasting the highland agropastoral systems with the coastal political economies—remains underdeveloped.

Questions and predictions

Based on foundational isotopic studies of camelid herding strategies, we present two primary, and one secondary research questions regarding camelid isotopic compositions to analyze variations in herding practices across different ecological zones and specific historical periods.

Question 1: Is there a detectable dietary shift in camelids from 900 BCE to 1470 CE that correlates with the intensification of maize agriculture, and do these shifts manifest consistently between coastal and highland regions across different cultural periods?

Prediction 1: We anticipate that, correspondent with the growing importance of maize from the Early Horizon (900–200 BCE) through the LIP (1000–1470 CE), there will be a progressive shift in herding strategies towards the increased use of maize as fodder. This change is expected to result in progressively enriched δ¹³C values in camelid bone collagen and fur fibers across both coastal and highland ecozones.

Question 2: Does the rise of maize cultivation from the Early Horizon (900–200 BCE) through the LIP (1000–1470 CE) promote the practice of camelid foddering on enriched maize and other agricultural products? Prediction: We expect that increasing reliance on maize agriculture should incentivize increased use of dung and guano fertilizers, which would enrich the nitrogen content of maize and the nitrogen values of camelids foddered on enriched maize and other agricultural products.

Question 3 Do stable isotope analyses of camelid bone and fiber provide corroborative evidence of camelids fulfilling distinct roles in coastal versus non-coastal economies of ancient Andean societies?

Prediction: We hypothesize that the potentially varied role of camelids significantly changes the isotopic signatures between coastal and non-coastal economies in ancient Andean societies. On the coasts, where camelids played a relatively more symbolic and ceremonial role, we predict less variation in their dietary isotope signatures due to controlled feeding practices aligned with their status and use in rituals and state functions. We also expect coastal camelids to exhibit a more enriched maize isotopic signal as these camelids may have been foddered on domestic crops rather than wild C3 grasses. Conversely, in non-coastal economies, greater dietary variability is anticipated, as are relatively depleted carbon isotopic values indicative of consumption of wild C3 grasses, reflecting utilization of camelids as a subsistence economic diversification strategy for offsetting potential crop failure.

Stable isotope analysis

To elucidate the historical dietary patterns and herding practices of camelids in Peru, we conducted analyses focusing on δ¹³C and δ¹⁵N values from collagen extracted from bone and keratin from fiber [45]. The carbon (δ¹³C) and nitrogen (δ¹⁵N) isotope compositions in consumer tissues mirror the isotope profiles of their diet during tissue formation [44, 46–49]. The type of tissue influences stable isotope values. Keratin in fiber, such as hair or wool, forms during the maturation of each strand and does not undergo subsequent remodeling [18, 43, 50]. Given the rapid growth rate of these fibers, their isotopic signatures capture dietary and environmental conditions over shorter periods, perhaps spanning weeks to months. This contrasts with bone collagen, which offers a longer-term perspective. In bones, collagen undergoes remodeling in response to factors like tissue type, growth, and external stimuli [10, 19]. Prior studies indicate that although there might be minor variations in the diets of individual animals these fluctuations are unlikely to markedly alter long-term indicators, like the isotopic composition of bone collagen [48]. Therefore, seasonal changes in camelid diet may only be captured by assessing fiber isotope values [45] while bone provides the adult average diet.

Plant δ¹³C values are primarily determined by their photosynthetic pathway, resulting in different carbon isotopic compositions [44, 51]. The C3 pathway leads to discrimination against 13C, yielding lower δ¹³C values. In the Andes, the significant elevation gradient spans from tropical climates at lower elevations to cold alpine conditions at higher altitudes, leading to a broad range of δ¹³C values in plant isotopes due to varying environmental factors [52]. Temperature, humidity, atmospheric pressure, and differences in precipitation and soil types all influence isotopic fractionation; for instance, higher elevations generally have cooler, wetter conditions resulting in less fractionation and lower δ¹³C values, whereas lower, warmer, and drier areas see increased water stress, contributing to higher δ¹³C values [53]. These variations are further complicated by plant type differences; grasses and shrubs are typically more enriched in δ¹³C, while subshrubs present lower values, influenced by root distribution and leaf traits such as cuticles and stomatal density [54]. Lichens specifically can exhibit higher δ¹³C values as they lack stomata, reducing stomatal-related fractionation [55]. Additional local effects such as salinity near coastal areas can increase δ¹³C values, deviating from broader regional patterns. This combination of factors leads to C3 plants presenting with a 13C range of −35‰ to −20‰ in the region [55].

In contrast, the C4 pathway incorporates 13C more efficiently during photosynthesis, resulting in higher δ¹³C values in plant tissues. In the Andean region, C3 plants are prevalent; however, the landscape also hosts C4 species [56]. According to Szpak et al. [56], δ¹³C values show distinct variations: wild C3 plants average –26‰ within a range of −35‰ to −20‰, C4 plants average –12‰ ranging from −15 to −7‰. In the Andes, C3 plants, constituting the common highland grasses used as fodder, contrast with C4 plants, which, including maize alongside certain Andean amaranths (Amaranthus sp.), are more suited to warmer, arid conditions [57]. The carbon isotope composition of animal proteins, such as collagen found in bone and dentin, and keratin in hair and nails, largely mirrors the protein portion of their diet [45]. This is because essential amino acids are directly transferred from the consumed food to the body’s tissues. However, this relationship is not directly proportional. Typically, the δ¹³C values in the collagen of large herbivores are approximately 5‰ higher than the average values found in their diet [48, 58].

CAM plants, including cacti, succulents, and epiphytes, utilize Crassulacean Acid Metabolism, a photosynthetic pathway that results in intermediate δ13C values ranging between −22 and −10‰ [40]. As browsers, camelids may consume a mix of C3, CAM, and C4 plants, averaged δ¹³C values camelids consuming CAM or mixed C3 and C4 diets could potentially look similar isotopically [59]. However, most CAM plants in the Andes are succulents and cacti and were not a significant part of highland camelid diets [15, 60–63]. On the coast, local lomas vegetation can include a few CAM plants like Tillandsia, but the dominance of C3 plants in this system suggests limited consumption of wild CAM or C4 plants by coastal camelids. As such, δ¹³C values from camelid bone collagen primarily differentiate between C3 and C4 plant consumption. The prevalence of C3 and C4 plants in the Andes varies with altitude and rainfall; at higher altitudes (above 3500 meters above sea level), where rainfall can reach up to 400 mm per year, the vegetation is predominantly composed of C3 plants [40, 47, 64]. This is particularly evident in the puna ecosystem, characterized by bofedales—wetlands that provide both water and vegetation, creating ideal grazing areas for camelids [65]. C3 plants are also dominant in various ecological zones across northern Peru and in the central and south-central Andes. At lower altitudes, there is a shift in plant composition, with C4 species favoring areas of lower elevation and presumably warmer, drier climates [40]. In Peru’s lowlands, particularly the hyperarid coastal regions, suitable areas for camelid pasturing are scarce, mainly found near beaches, along river courses, and in lomas, or ’fog oases,’ located in the Andean foothills [15]. These lomas, flourishing seasonally due to ocean fog humidity, comprise various species of grass, shrubs, and trees. A wild C4 species, Distichlis spicata (Grama salada), is known to grow in coastal areas, forming extensive patches in certain locales. Valleys, fed by mountain water, provide natural grazing areas, including both coastal and highland (sierra) regions, though these same spaces tend to be the best locations for farming. Agricultural practices in these valleys involve managing irrigated fields for diverse crops like maize (C4), beans (C3), peanuts (C3), and avocados (C3). Camelids may have access to these crops, leading to the practice of foddering, especially with maize. This practice was previously suggested for the southern Peruvian highlands by Finucane et al. [9] and for the northern coast of Peru by Dufour et al. [10, 16].

In the Andean region, the altitude-related increase in temperature and precipitation notably affects the nitrogen isotopic composition in soils and plant life. Hot, arid environments, predominantly found at lower altitudes, are characterized by higher δ¹⁵N values [17, 43]. This is due to biogeochemical processes that cause the loss of the lighter nitrogen isotope, δ¹⁵N. These findings are supported by studies that connect these high δ¹⁵N values in plants, and subsequently in animal tissues, to the arid conditions and related nitrogen metabolic processes [66–68].

In contrast, at higher altitudes, plant tissues show comparatively lower carbon and nitrogen isotopic compositions. This variance is further complicated in agricultural plants, where fertilization plays a significant role in isotopic composition. Use of organic fertilizers, such as camelid dung, results in a moderate increase in δ¹⁵N values (+2‰ to +4‰), whereas seabird guano application can cause a substantial rise, often exceeding +20‰ [56, 69, 70]. As a result, animals grazing in coastal areas tend to exhibit higher nitrogen isotope values than those in higher altitudes, underscoring the complex relationship between fertilization methods, plant isotopic composition, and animal diets across different ecological zones [17]. In coastal Peru, for example, the irrigation of agricultural plants, such as maize, reduces the impact of water scarcity on δ¹³C and δ¹⁵N values [70]. Wild plants in these coastal areas have higher δ¹⁵N values, in contrast to irrigated agricultural plants, which generally exhibit lower δ¹⁵N values, even without fertilization [56]. Among terrestrial plants, only those that fix atmospheric nitrogen, such as legumes, exhibit distinctly unique δ15N values [40].

Due to prior suggestions of a latitudinal aridity effect on nitrogen levels, our research utilized generalized linear models (GLMs) to examine the relationship between Peru’s latitude (a proxy for increasing aridity along the north-south gradient) and camelid nitrogen isotope levels (δ¹⁵N). The initial Generalized Linear Model (GLM) analysis with latitude as a predictor showed a negative relationship between latitude and δ¹⁵N values, although the effect was not statistically significant (p = 0.109). Comparing this model to a null model using the Akaike Information Criterion (AIC) suggested only a marginal improvement. In a more comprehensive GLM incorporating additional environmental factors like elevation and ecozone, latitude still did not show a significant effect on δ¹⁵N values (p = 0.106), and the final model suggested no significant residual relationship between latitude and corrected δ15N values after accounting for these environmental influences (S2 File).

Data preparation

Before implementing statistical analyses, we add several categorical variables to our data. In this study, we adopted Lanning’s [71, 72] standardized periodization of Andean prehistory to classify isotope values, assigning them to one of four cultural periods: Early Horizon (900–200 BCE), Early Intermediate Period (200 BCE-600 CE), Middle Horizon (600–1000 CE), and Late Intermediate Period (1000–1470 CE). This approach was utilized despite the existence of more specific cultural chronologies for certain regions, such as the Moche period in northern Peru, to maintain consistency with a widely recognized framework that facilitates broader comparisons across the Andes.

Secondly, we have divided camelid samples into two groups: coastal and highland. This categorization reflects a distinction in camelid herding based on ecoregion, as previously identified by researchers [14, 16, 19, 73]. Our study evaluates distinctions in camelid management by analyzing isotopic data across extensive regions and diverse cultural groups, enabling a detailed examination of these practices. This method allows us to explore the potential variations in herding strategies between and within coastal and highland societies, thereby addressing key questions about the influence of ecological context on pastoral practices.

Analytical methods

To address our research questions, we adopted a mixed methods approach that integrates Kruskal-Wallis tests with subsequent pairwise tests, General Linear Models (GLM), and Linear Mixed Effects Models (LMER) [74]. We utilized the Kruskal-Wallis test to identify significant differences in δ¹³C and δ¹⁵N values across various periods and ecological zones, which is appropriate for our skewed data [74]. Detailed comparisons for specific periods were conducted using the Wilcoxon Rank Sum test. GLMs were used to analyze the impact of factors such as altitude and time on δ¹³C and δ¹⁵N isotopic compositions, while LMERs with random effects to control for latitude addressed spatial dependencies in data showing non-independent residuals [74, 75]. All statistical analyses were conducted in the R programming environment [74] (S2 File). These statistical tools allow us to delineate dietary transitions with precision, drawing on data from bone collagen and fiber to authenticate patterns and gauge the influences of geographic location and historical periods. Additional analyses include Kruskal-Wallis tests assessing δ¹³C and δ¹⁵N differences in the Bone and Fiber datasets and across cultural groups, a GLM examining the relationship between nitrogen isotopes and latitude, and visualizations that highlight isotopic variations driven by geographical and cultural changes over time (S2 File).

Question 1: Carbon isotopes: Camelid diet and maize agriculture

The statistical analysis of our integrated dataset, using Kruskal-Wallis rank sum tests, highlights significant differences in δ¹³C corrected values across various ecozones and cultural time periods (Table 1). The analysis revealed pronounced disparities in δ¹³C values by ecozone, with a Kruskal-Wallis chi-squared of 152.25 (p < 0.0001). Focusing on regional comparisons, significant variances were noted across cultural periods within the coastal region. For camelids located along the coast, the Kruskal-Wallis test showed a chi-squared value of 101.85 (p < 0.0001), whereas for camelids from the highland region, the carbon isotope signature showed a chi-squared value of 27.316 (p < 0.0001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of Kruskal-Wallis tests and pairwise comparisons for δ¹³C and δ¹⁵N values.

https://doi.org/10.1371/journal.pone.0306205.t001

Further distinctions were identified through Pairwise Wilcoxon rank sum tests with continuity correction, which provided detailed contrasts between specific periods. For camelids from coastal sites, significant differences were observed between all four cultural time periods (Table 1; Fig 2). In the highland region, comparisons of δ¹³C values showed significant differences between the Early Horizon and the three later time periods, but no significant variation was found between the EIP, Middle Horizon, and LIP (Table 1; Fig 2). Additionally, a substantial isotopic difference was observed between the highland and coast regions for δ¹³C values, underscoring the ecological influence on isotopic signatures (p < 0.0001). These results were confirmed to be robust as the Kruskal-Wallis and Wilcoxon tests held up under 1000 bootstrap iterations, indicating the results are not due to sampling bias. Detailed analyses and bootstrapping procedures are provided in the Supplemental Material (S2 File).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Carbon isotopes.

Panel A shows boxplots of δ¹³C values in camelids from coastal and highland regions, organized by cultural periods: Early Horizon, Early Intermediate Period, Middle Horizon, and Late Intermediate Period. Panel B illustrates regression plots of δ¹³C values against elevation (MASL) for each cultural period, detailing the changes in isotopic composition with increasing altitude. Each regression is color-coded by period, with the shaded areas representing 95% confidence intervals, indicating the robustness of the linear model fit across the range of elevations.

https://doi.org/10.1371/journal.pone.0306205.g002

A Generalized Linear Model (GLM) was utilized to analyze the variation in δ¹³C values the integrated bone and fiber isotope dataset (Table 2). Initially, a null model was specified, demonstrating a significant mean level of δ¹³C values (Intercept = -15.2582, p < 0.0001). This indicates that the average δ¹³C value across the dataset is significantly different from zero. Subsequently, an additive model incorporating elevation (MASL) as a predictor revealed a significant negative relationship between elevation and δ¹³C values (Coefficient = -0.001157, p < 0.0001). This suggests that as elevation increases, δ¹³C values decrease, highlighting the influence of altitude on carbon isotope composition. The analysis was extended with a more complex interaction model that included the effects of elevation and specific time periods (early horizon, early intermediate period, middle horizon, late intermediate period). This model provided the best fit (AIC = 2794.4, BIC = 2833.85). The interaction model revealed significant interactions between elevation and time periods, particularly noting pronounced changes in δ¹³C values with elevation during the Late Intermediate Period (Coefficient for MASL: Late Intermediate Period = -0.003267, p < 0.0001). This indicates that the effect of elevation on δ¹³C values was more pronounced during this period. The GLM analysis reveals that increasing elevation is associated with decreasing δ¹³C values, and that this relationship is influenced by the time period. Specifically, during the Late Intermediate Period, the impact of elevation on δ¹³C values was more substantial compared to other periods.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Comparative analysis of δ¹³C and δ¹⁵N isotopes in camelid bone and fiber using Generalized Linear and Linear Mixed Effects Models.

https://doi.org/10.1371/journal.pone.0306205.t002

Question 2: Nitrogen isotopes—Camelid diet and maize agriculture

The investigation of δ¹⁵N values, employing Kruskal-Wallis rank sum tests, revealed significant differences across both coastal and highland regions, as well as across various cultural periods (Table 1; Fig 3). For camelids from the coastal region, δ¹⁵N levels varied significantly over time, evidenced by a Kruskal-Wallis chi-squared value of 48.489 and a significant p-value (p < 0.0001). In the highland region, δ¹⁵N levels also showed significant variation by period (chi-squared = 11.964, p = 0.0075). Subsequent analysis using Pairwise Wilcoxon rank sum tests indicated pronounced differences in δ¹⁵N values between the coastal camelids from each of the four periods. Specifically, significant differences were found between the Early Horizon and the Early Intermediate Period (p = 0.0028), the Late Intermediate Period (p < 0.0001), and the Middle Horizon (p = 0.0028). Additionally, significant differences were observed between the Early Intermediate Period and the Late Intermediate Period (p < 0.0001), and between the Middle Horizon and the Late Intermediate Period (p < 0.0001). In the highland region, pairwise comparisons revealed no significant differences between the Early Horizon and the Early Intermediate Period (p = 0.135), the Late Intermediate Period (p = 1.000), or the Middle Horizon (p = 1.000). A significant difference was found between the Early Intermediate Period and the Late Intermediate Period (p = 0.022), and a nearly significant difference between the Early Intermediate Period and the Middle Horizon (p = 0.057). No significant variation was found between the Late Intermediate Period and the Middle Horizon (p = 1.000). Comparisons between ecozones revealed a substantial isotopic difference between the highland and coastal regions for δ¹⁵N values (p < 0.0001), underscoring the ecological influence on isotopic signatures. These results were confirmed to be robust as the Kruskal-Wallis and Wilcoxon tests held up under 1000 bootstrap iterations, demonstrating 99.9% success for the coast and 99.1% for the highland, indicating that the results are not due to sampling bias. Detailed analyses and bootstrapping procedures are provided in the Supplemental Material (S2 File).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Nitrogen isotopes.

Panel A shows boxplots of δ¹⁵N values in camelids from coastal and highland regions, categorized by cultural periods: Early Horizon, Early Intermediate Period, Middle Horizon, and Late Intermediate Period. Pannel B illustrates regression analyses of δ¹⁵N values against elevation (MASL) for each cultural period. The color-coded regressions, accompanied by 95% confidence intervals shown as shaded areas, underscore the relationship’s variability and the model’s predictive confidence at different elevations.

https://doi.org/10.1371/journal.pone.0306205.g003

Using Linear Mixed Effects Models (LMER), we examined δ¹⁵N isotope variations across bone and fiber samples from various latitudes (Table 2; Fig 3). The null model showed significant average δ¹⁵N values (Intercept = 7.5845, p < 0.0001), indicating that the mean δ¹⁵N value across the dataset is significantly different from zero. An additive model including elevation (MASL) suggested that elevation alone does not significantly affect δ¹⁵N levels (Coefficient = -0.0002689, p = 0.328). This implies that changes in elevation do not have a straightforward impact on nitrogen isotope values. Expanding to an interaction model revealed how elevation interacts with periods to influence δ¹⁵N values. The interaction model showed that the effect of elevation on δ¹⁵N values varies across different time periods. Specifically, the interaction term for elevation and the early intermediate period (MASL: EIP) was significant (Coefficient = -0.0017763, p = 0.012), indicating a more pronounced effect of elevation on δ¹⁵N values during this period. The model also adjusted the baseline δ¹⁵N values for different periods, with the intercept set at 7.006 (p < 0.0001). However, the interactions for the middle horizon (Coefficient = 0.7772, p = 0.328) and late intermediate period (Coefficient = 1.3778, p = 0.202) were not statistically significant. This analysis reveals that while elevation alone does not significantly influence δ¹⁵N values, its impact varies across different historical periods. During the early intermediate period, higher elevations are associated with a notable decrease in δ¹⁵N values. The interaction model’s complexity provided a better fit (AIC = 2383.021, BIC = 2426.822), illustrating the detailed interplay between environmental and temporal factors in shaping isotopic profiles.

Question 1: Carbon isotopes: Camelid diet and maize agriculture

The analysis of δ¹³C values across various ecozones and cultural periods using Kruskal-Wallis rank sum tests and GLM provided substantial evidence for significant dietary shifts in camelid populations by region. These findings partially support Prediction 1, which posited that changes in isotopic composition would occur concurrently with the adoption of maize agriculture as a major economic activity and should vary by ecoregion. Regarding ecoregion, the δ¹³C values on the coast were significantly higher compared to those from the highland region, supporting the hypothesis of trans-regional differences in the adoption of maize-based foddering practices for camelids based on being coastal or non-coastal. Coastal camelids appear to have been foddered with maize to a significantly greater extent than non-coastal camelids overall.

Providing support for Prediction 1, significant temporal shifts in carbon isotopic values were observed in camelids from coastal sites, with substantial increases in δ¹³C values from the Early Horizon through to the Late Intermediate Period (LIP). These findings strongly suggest that coastal regions experienced a pronounced intensification of maize use as fodder over time, supporting the hypothesized link between the adoption of maize agriculture and the increase in carbon isotopic signatures in camelid diets [12, 19]. Coastal herders likely allowed the animals to graze on maize stalks in the fields, forage in cultivated plots, or brought maize stalks to the herds in corrals and pens [73, 77]. While the coastal increase in observed δ¹³C values appears to correspond with maize intensification temporally, there are plausible, if less likely, alternatives. The possible consumption of local loma vegetation, which was largely comprised of C4 and CAM plants could influence these isotopic profiles [40]. However, C3 plants dominate Peru’s flora, lomas vegetation is typically a seasonal and not year round phenomena, and it would likely be challenging for coastal herders to find sufficient wild C4 pastures to account for the enriched carbon isotope values observed in camelid diets.

There is also variation in coastal camelid diet, particularly in the Early Intermediate Period, and a decline in carbon values during the Late Intermediate Period. Although more individuals had diets with a high proportion of C4 plants, this does not fully capture the complexity and variability in the data. This variability and decline seem to reflect a long-term, mixed C3-C4 diet signal, suggesting a diachronic switch in diet, perhaps associated with changes in geographic location. Especially in the Late Intermediate Period, camelids may have been part of caravans moving to and from the coast, consuming a more mixed C3/C4 diet than those living exclusively on high-altitude pastures, with maize fodder provided at various points along the way [16, 73].

In the highland regions, detectable differences in isotopic values were only observed between the Early Horizon and subsequent periods, suggesting maize foddering in the Early Horizon but far less in later periods. Indeed, the pattern observed from the Early Intermediate Period through to the Late Intermediate Period shows that the camelid diet was predominantly C3-based, with the possibility of some C4 plant consumption. It is important to note that the small, localized sample size from the Early Horizon limits our comprehensive understanding of herding practices during this period. Therefore, it seems more parsimonious to suggest that highland camelids broadly maintained a consistent diet of primarily C3 plants over time, unaffected by the intensification of maize agriculture. While camelid groups from coastal sites generally exhibit greater isotopic variability compared to those from the highlands, there are some outlier camelids from each of the three later time periods. These outliers indicate a rare consumption of a C4 plant, such as maize, possibly reflective of movement between the coast and the highlands.

These results collectively demonstrate that regional and temporal variations in maize agriculture significantly influenced camelid diets, with ecological variation resulting from elevation differences playing a crucial role. The progressive enrichment of δ¹³C values in coastal regions suggests a stronger integration of maize into diets over time, contrasting with the highlands where camelids predominantly consumed non-maize plants [12, 19, 73, 77]. These adaptations to maize agriculture, although linked, occurred in non-uniform ways across different regions. Along the coast, where limited forage was available and camelids may not have been primarily used for subsistence, maize agriculture likely led to increased maize foddering [73, 77]. Conversely, despite the ecological suitability of C4 plants in the maize belt, camelids there exhibited lower δ¹³C values, indicative of a diet largely composed of C3 plants, which are abundant across all ecological zones. In highland areas, the increased adoption of maize seems to have led to a reduction in maize foddering, potentially to preserve fields for crop cultivation or as a risk mitigation strategy against crop failure [11, 39]. These findings underscore the responsiveness of past herding practices to the changing agricultural landscape, revealing that herders actively adapted their livestock’s diets in accordance with regional economic and agricultural conditions.

Question 2: Nitrogen isotopes—Camelid diet and maize agriculture

Our study’s analysis of δ¹⁵N isotope values in camelid bone and fiber aimed to determine if nitrogen values changed over time, potentially as a result of changes in manuring and fertilizing practices associated with agricultural intensification. Controlling for the latitudinal aridity effect on nitrogen, our Linear Mixed Effects Models demonstrated that both elevation and time period play crucial roles in modulating δ¹⁵N levels. Camelids herded along the coast are typically thought to exhibit high δ¹⁵N values because plant tissue δ¹⁵N values tend to be higher under conditions of limited water availability [49, 78]. In contrast, highland plants receive more precipitation, and consequently, camelids feeding in the highlands are typically expected to exhibit lower δ¹⁵N values than those at lower elevations. However, our data shows an intriguing opposite pattern: the average δ¹⁵N values for highland camelids are generally higher than those for coastal camelids over time. This is an unanticipated result. It may suggest that highland camelids may have consumed more fertilized plants, which are enriched in δ¹⁵N, while coastal camelids primarily grazed on unfertilized plants, resulting in lower δ¹⁵N values.

For camelids from coastal regions, Kruskal-Wallis rank sum tests demonstrated marked variability in δ¹⁵N levels over time, culminating in significant differences, particularly between the Early Horizon and the Middle Horizon, followed by a slight decline in nitrogen levels during the Late Intermediate Period (LIP). This overall pattern suggests a pronounced increase in nitrogen values parallels the intensification of agricultural practices and increased consumption of maize by camelids, suggesting substantial maize foddering in coastal irrigated and/or fertilized fields [56, 79]. Indeed, the temporal correlation between the high δ¹³C and δ¹⁵N values indicates a diet of maize grown in fields fertilized with guano or manure which has been previously identified to increase δ¹⁵N values [56, 79]. The high carbon values coupled with the arid environment makes camelid consumption of non-irrigated maize unlikely. However, coastal camelids might have also fed on loma vegetation, primarily composed of C4 and CAM plants, or grama salada, a salt tolerant C4 grass. Additionally, some diets with higher δ¹⁵N values may have also included a high percentages of fish remains, such as Peruvian hake (Merluccius gayi peruanus) and Peruvian anchoveta (Engraulis ringens), found in camelid coprolites [76]. Sea algae, with δ¹⁵N values below +2‰ or +3‰ (similar to terrestrial plants), cannot explain the high δ¹⁵N values in these archaeological remains [40].

Camelids from the highland region display intriguing patterns in nitrogen isotope values, reflecting changes in agricultural and herding practices over time. Controlling for latitudinal effects, the nitrogen isotope values (δ¹⁵N) for highland camelids largely fall within the expected range for the consumption of fertilized agricultural plants, indicating that either intentional fertilization techniques in agricultural practices or unintentional fertilization (repeated camelid use and defecation) in pastures may have directly influenced isotopic signatures. While the carbon isotopes remain low, the enriched δ¹⁵N values suggest that herding strategies likely involved the use of foddering C3 plants, including tubers and quinoa, which have been identified through microbotanical analyses as part of the diet of herded camelids in the highlands [22]. The increase in δ¹⁵N values is unlikely to be the result of consumption of nitrogen-fixing plants like legumes, which are rare in the highlands and have not been identified in camelid diets. During the Early Horizon and LIP, the broad variation in δ¹⁵N values indicates significant diversity in husbanding practices. This variation aligns with more intensive and varied agricultural practices, including the sporadic consumption of fertilized C3 crops such as tubers and traditional C3 grass pastures, which influenced dietary changes.

Our analysis partially supports the second hypothesis that the temporal intensification of maize agriculture and fertilization practices was a pivotal factor in pastoral evolution in Peru. While there is a temporal increase in nitrogen values for both coastal and highland camelids through time, this increase is not solely due to maize agriculture. The data suggest that the enrichment in δ¹⁵N values is not exclusively tied to increased maize cultivation but rather to a broader use of fertilization practices that enhanced nitrogen levels in various crops. For camelids residing along the coast, the rising nitrogen values reflect changing diets directly related to significant changes in agricultural techniques for maize. In contrast, camelids from the agricultural highlands maintained a more varied diet that included foddered C3 plants. Consequently, the observed isotopic shifts in camelids point to a complex interplay of evolving agricultural strategies and environmental factors influencing camelid diets over time.

Research question 3: Isotopic evaluation of varied camelid socioeconomic roles

Our study highlights an intriguing temporal divergence in camelid diets between coastal and maize-belt areas. Among C4 taxa, maize is the most significant component in the diet of the New World archeological agrarian populations [40]. It has been identified that significant δ¹³C deviation toward C4 plant range is a sign of intentional maize feeding of camelids, as a part of organized herding following the intensification of maize agriculture [9, 10, 15]. However, the gradual intensification of maize cultivation did not uniformly influence camelid diets across these diverse regions [9, 80–83]. Coastal camelids show the anticipated shift towards a heavily maize-influenced diet while conversely, their non-coastal counterparts exhibited a decrease in carbon isotope signature. Though unexpected, this pattern appears to follow the divergent socio-economic purposes camelids filled in these distinct ecological zones.