Subsistence strategy mediates ecological drivers of human violence.

Inter-personal violence (whether intra- or inter-group) is a pervasive yet highly variable human behavior. Evolutionary anthropologists suggest that the abundance and distribution of resources play an important role in influencing differences in rates of violence, with implications for how resource conditions structure adaptive payoffs. Here, we assess whether differences in large-scale ecological conditions explain variability in levels of inter-personal human violence. Model results reveal a significant relationship between resource conditions and violence that is mediated by subsistence economy. Specifically, we find that interpersonal violence is highest: (1) among foragers and mixed forager/farmers (horticulturalists) in productive, homogeneous environments, and (2) among agriculturalists in unproductive, heterogeneous environments. We argue that the trend reversal between foragers and agriculturalists represents differing competitive pathways to enhanced reproductive success. These alternative pathways may be driven by features of subsistence (i.e., surplus, storage, mobility, privatization), in which foragers use violence to directly acquire fitness-linked social payoffs (i.e., status, mating opportunities, alliances), and agriculturalists use violence to acquire material resources that can be transformed into social payoffs. We suggest that as societies transition from immediate return economies (e.g., foragers) to delayed return economies (e.g., agriculturalists) material resources become an increasingly important adaptive payoff for inter-personal, especially inter-group, violence.

Introduction

Inter-personal violence, whether intra- or inter-group, is a persistent attribute of human societies, though the degree of violence varies [1]. Explanations for coalitional violence include a wide variety of cultural [2-5] and evolutionary [6-17] hypotheses. These long-debated explanations center on the ultimate causes of collective violence while evolutionary approaches focus on the adaptive payoffs for individual participation. Recent research also explores the co-evolution of inter-group violence with group cooperation, altruism, and the emergence of complex social systems [16, 18, 19]. Those that propose inter-group violence to be a rare or maladaptive behavior [20-22] have been effectively countered in many cases [23-26], although important debate persists [7].

Here we examine how ecological and economic factors influence the adaptive payoffs for inter-personal violence and in doing so explain cross-cultural variation. We focus on the proposed adaptive payoffs for participation in inter-group violence–although these payoffs should impact intra-group violence as well–which are most often posited to be (a) social rewards such as status, mating opportunities, and alliance formation [8, 11, 13, 27], or (b) the procurement and defense of scarce resources [6, 28–30]. Our central question is whether macroscale variation in the abundance and distribution of local resources has a structuring effect on the adaptive payoffs for violence, and whether subsistence economy mediates this relationship. It is critical then that we introduce how social and material payoffs relate to violent interactions.

Many scholars hypothesize that inter-group violence serves as a resource procurement strategy and, thus, propose that the benefits of violence are high when individuals compete for high-ranking resources that are scarce and distributed in patches that can be effectively privatized [6, 28–30]. In the most general sense, inter-group violence is proposed to be a subsistence activity to procure or defend valued resources. Typically, these strategies target group-level rewards, such as territory, that elicit collective action problems [11, 31, 32]. As such, participation in resource violence must be accompanied by social rewards, such as status [11], or a disproportionate share of the loot that can then be mobilized to enhance fitness. Scarce, patchily distributed resources, thus, provide the initial incentive for violence, with social rewards or the accumulation of personal wealth providing the impetus for individual participation.

Other researchers propose that participation in inter-group violence is best explained as a strategy to directly gain, co-opt, or defend status, alliances, or mating opportunities (i.e., social rewards), with resource procurement and defense playing an irrelevant or epiphenomenal role [11, 13, 27, 33–35]. This may arise when socioeconomic conditions (e.g., mobility, resource surplus, privatization) preclude the use of violence to amass and own resources. In this sense, participation in violence is motivated by direct social rewards and should be more frequent when the abundance of local resources allows for energetic surpluses to be allocated into violent competition [30, 32, 36–38]. In this case, rich environments more frequently “finance” costly inter-group violence.

Relating these alternative explanations to resource conditions (Fig 1), we expect that (1) if inter-group violence is incentivized by the procurement of material resources and attendant fitness payoffs, then rates of violence should be highest in marginal, heterogeneous environments with scarce, clustered resources that can be monopolized [6, 28–30]. Alternatively, (2) if inter-group violence is primarily a strategy to directly acquire social rewards, with resource acquisition being an epiphenomenal motivation [11, 33, 34], then rates of violence should be highest in environments where abundant and predictable resources provide energy surpluses that can be diverted into violent inter-group (and likely intra-group) competition [30, 32, 36–38]. A third, somewhat less obvious possibility is that subsistence economy plays a key mediating role, whereby the transition to delayed-return economies (e.g., agriculturalists) allows for the accumulation of wealth, generating greater incentives for resource violence.

Fig 1

Conceptual model.

Expectations for the two primary explanations. Punnett squares display the expected payoffs for violence as a function of environmental productivity (resource abundance) and heterogeneity (resource dispersion).

We evaluate the importance of resource conditions relative to levels of violence using a global archaeological dataset of delayed-return agricultural societies and a global ethnographic dataset of immediate-return forager and mixed-economy horticultural societies (Fig 2). We combine archaeological and ethnographic datasets because (a) archaeological violence datasets are extremely limited for foragers and horticulturalists, while ethnographic research lacks data on violence among complex, agricultural societies–particularly non-industrial and non-western populations prior to the 20th century, and (b) it is vital that we assess whether socioeconomic features mediate the ecological drivers of violence, as discussed above. We combine these datasets to test how levels of violence are affected by the spatial distribution of net primary productivity (NPP), both its mean (resource abundance) and standard deviation (resource dispersion). The novel dataset presented here expands on previous evaluations of coupled archaeological and ethnographic data [e.g., 18] and allows us to quantitatively explore the question of whether levels of violence differ by ecological conditions and modes of subsistence.

Fig 2

Map of observations.

The global distribution of societies with violence data used in this study; color and shape coded by mode of subsistence.

Materials and methods

Following Bowles [18], we developed a database of proportional violence (hereafter referred to as “PV”) using archaeological and ethnographic data and categorized all sampled groups into three modes of subsistence (MoS): foraging, horticulture, and farming (see, S1 File). We define MoS simply as benchmarks along the continuum from immediate-return to delayed-return economies along with the related features of subsistence and social organization such as mobility, surplus, storage, and privatization [39, 40]. Pastoralists are not included in this study due to the lack of sufficient observations to run a separate empirical model. In all, the database contains 53 populations from seven World regions (Fig 2 and S1 and S2 Files).

While archaeological and ethnographic measures of violence are not identical, they have been used to good effect in several prominent studies [18, 41]. The dataset developed here builds off prior studies by providing additional archaeological and ethnographic samples and improves precision by excluding archaeological groups with small sample sizes.

Archaeological measure of violence

The archaeological database derives from several dozen peer-reviewed publications that report proportions of violent skeletal trauma. Bioarchaeologists have robust methods for distinguishing intentional violent trauma from accidents and post-mortem damage [42-44]. This database operationalizes PV by defining it as the number of individual skeletons with evidence of violent trauma divided by the total skeletal sample for each archaeological observation (i.e., region or society). Published data that reports ante-mortem (healed) and peri-mortem (non-healed) skeletal trauma were included if they (1) reported proportion of violent trauma that exclude accidents and post-mortem damage, (2) contained sample sizes of approximately n = 100 individuals or greater (with the exception that small islands contain smaller samples sizes), and (3) contained site or region-specific location data. In all, 34 samples from seven World regions were included from 25 sources (S1 and S2 Files). Since most sources do not report peri-mortem trauma rates, we rely on generalized inter-personal violence. Because this type of data does not distinguish intra- vs inter-group violence, we treat the database as a long-term record of generalized inter-personal violence for a given area, which is standard practice for cross-regional bioarchaeological studies [e.g., 6, 42, 45–48]. However, as intra-group lethal and non-lethal violence tend to be low in human societies [1, 2], we expect that changes in population-level inter-personal violence will largely reflect fluctuations in inter-group violence. Furthermore, the payoffs for violence outlined in this paper (e.g., mating, status, and resource competition) should influence general inter-personal violence, regardless of whether it is intra- or inter-group. While certain global regions are not included due to lack of data with adequate sample sizes, the current samples in the archaeological database span a variety of net primary productivity (NPP) values (Fig 3), which allows for robust hypothesis testing. To provide a more representative sample of archaeological data, research will be needed to produce and disseminate high-resolution trauma datasets from a wider variety of economic and ecological contexts.

Fig 3

A density plot for the observations (points) in our dataset.

Color represents the density of observations (societies) along both environmental dimensions (mean NPP and standard deviation in NPP).

Ethnographic measure of violence

Our ethnographic database derives from peer-review publications and online scholarly sources. This database operationalizes PV by defining it as the number of observed or reported acts of lethal violence divided by the estimate of the total ethnographic population size. Proportions of lethal violence are obtained first through aggregate sources (e.g., ourworldindata.org) and then corroborated using eight original published sources (SS2 File). In all cases, proportions of violent death are obtained from the most recent publications. The ethnographic database contains 19 populations from seven World regions (Fig 2).

Data compatibility

Because the archaeological database contains generalized inter-personal trauma and the ethnographic database contains violent deaths, these two databases are not perfectly compatible. Perhaps most importantly, we cannot compare these two datasets to evaluate differences in absolute rates of violence. We must acknowledge that the ethnographic rates would be higher if they included all skeletal trauma, and the archaeological rates would be lower if they included only lethal trauma. That said, both data types produce similar types of measurements: incidents of inter-personal violence divided by population estimates. In addition, lethal and sub-lethal violence tend to co-occur, as does intra- and inter-group violence [49-52] and should offer comparable measures of relative rates of violence in each society, thus informing us about the functional relationship between that relativized rate and the environment.

What does this mean for our models? Primarily, it means that they cannot provide an unbiased estimate of the intercept, understood as the marginal or baseline rate of violence. However, they should offer unbiased estimates of the coefficients, where those capture functional responses to environmental conditions. Fortunately, for this analysis, the intercepts are largely irrelevant, for all we need to evaluate our hypotheses are the model coefficients. That is, we are more concerned with relative rather than absolute rates of violence.

Environmental productivity

As a proxy for environmental productivity, we relied on terrestrial net primary productivity (NPP), which approximates photosynthesis, measuring the amount of energy that is turned into mass and thereby approximating the amount of new growth biomass available to consumers. The rasters containing mean and SD NPP were compiled between 2000 to 2015 at a 1-km resolution using remotely sensed data from the MODIS instrumentation on NASA’s Terra satellite, processed and provided by the Numerical Terradynamics Simulation Group at the University of Montana [53, 54]. These were converted to kg/C/m2 by multiplying by the scaling factor 0.0001 [55]. We then extract NPP estimates for a 50km buffer generated around a geospatial centroid for each of the societies in our sample and map out the densities of the observations over space (Fig 3).

Modern NPP has been used to predict numerous prehistoric phenomena including population density [56-58], habitat colonization [59], resource scarcity [6] and more. While NPP provides a measure of modern productivity, the relative NPP ranking of each region should have remained consistent over time due to their broad geographic range representing general physiographic regions rather than variants within single ecosystems [60, 61]. To better illustrate this point, we take as an example two regions included in our sample: the North American Colorado Plateau (CP) and the Illinois river valley (IRV). While the absolute NPP of these two regions certainly fluctuated throughout the centuries, there was no point during which the “mountain grassland and scrublands” of the CP were more environmentally productive (higher NPP) than the “tropical and subtropical dry broadleaf forests” of the IRV [60, 61]. The relative differences in environmental productivity between these two regions are due to their differing physiographic characteristics resulting in a marginal ecology with the former and a productive ecology with the latter, which have remained comparatively consistent for millennia. The same can be said about the relative variation in NPP between the other regions compared in our analysis and the observations clustered therein (e.g., the lowland neotropics of the Amazon Basin versus the Canadian boreal forests). At most, changes in absolute NPP would only serve to exaggerate or flatten the modeled curves but would not lead to trend reversals.

Analytical methods

We first evaluate our sample for potential spatial autocorrelation in proportion of violence (sigma) using Monte Carlo simulations of a Moran’s I Test as implemented by the moran.mc function in the R spdep package [62] (SS2 File). This includes an analysis across all MoS and within each MoS. We find the former to be significant (MI = 0.37, p = 0.03) but primarily driven by autocorrelation among farming populations (MI = 0.39, p = 0.045), as there was no autocorrelation among foragers (p = 0.663) or horticulturalists (p = 0.295). To test whether this may indicate a latent spatial process an additional Moran’s I Test on the residuals of the full model (see below) revealed no significant spatial autocorrelation (MI = 0.169, p = 0.165), suggesting that model coefficients captured any latent spatial process in the response.

To test our research questions, we fit three generalized linear models (GLMs) with a binomial distribution and log link appropriate to proportional data using quasi-likelihood estimation to account for overdispersion in our response variable (PV). We generate a null model without predictors, a base model with only our environmental predictor variables (mean and SD NPP), and a complete explanatory model with environmental predictors and MoS, specified as an interaction term to test for differences in intercepts and slopes, with farming as the reference class. We then perform a likelihood ratio test to evaluate whether each increase in model complexity provides sufficient gain in explanatory power. For each model, we report PV as a function of mean and SD NPP along with coefficient estimates and standard errors shown as log of the odds ratio. All analyses are conducted in the R programming environment and language [63] (For more details about our analysis, please see SS2 File).

Results

The base empirical model that includes all forager, farmer, and horticulturalist populations results in a significant model improvement from the null model (X = 0.47, p = 0.0458, Table 1) and shows that violence positively covaries with mean NPP (β = 0.513, p = 0.0475), and has a limited response to NPP SD (β = -1.355, p = 0.1771), suggesting that the overall proportion of violence is slightly higher in productive environments (r2 = 0.08), as predicted by the social rewards hypothesis (Table 1).

Table 1

Results of binomial GLMs evaluating proportional violence (PV) as a function of mean and SD NPP.

Coefficient estimates and standard errors are shown as log of the odds ratio. Coefficients in the full model are relative to the farming reference class.

Covariate	Coefficient	Std. Error	P-Value
Null Model	-1.5041	0.1202	<0.0001
Base Model
Intercept	-1.5696	0.2239	<0.0001
NPP	0.5134	0.2526	0.0475
NPP SD	-1.3545	0.9895	0.1771
Full Model
Intercept
Farming	-1.7953	0.2448	<0.0001
Foraging	-0.0017	0.7239	0.9981
Horticulture	0.1434	0.5506	0.7957
NPP
Farming	-2.782	0.737	0.0005
Foraging	5.251	1.285	0.0002
Horticulture	3.546	0.826	0.0010
NPP SD
Farming	6.956	2.114	0.0020
Foraging	-11.266	3.030	0.0006
Horticulture	-8.785	2.402	0.0007

The more complex empirical model accounting for MoS (forager, horticulture, farmer) results in a significant model improvement from the base model (X = 2.13, p < 0.0001, r2 = 0.34, Fig 4). The proportion of violence as a function of environmental productivity significantly differs between foragers and farmers (p = 0.0002), between farmers and horticulturalists (p < 0.0001), and between foragers and horticulturalists (p = 0.0001) (Table 1 and Fig 4). The proportion of violence as a function of environmental heterogeneity also differs significantly between these MoS. For foragers and horticulturalists, violence positively co-varies with environmental productivity and homogeneity, while for farmers violence negatively co-vary with productivity and homogeneity.

Fig 4

Model results showing the predicted response of proportional violence (PV) for each subsistence mode to every combination of mean environmental productivity (NPP) and the standard deviation in environmental productivity (NPP SD).

Predicted values are constrained to their observed range in the data. Marginal response plots for each MoS are available in S1 File.

To summarize, including MoS as an interaction term significantly improves model fit, showing that proportions of generalized violence among foragers and farmers in particular exhibit opposite covariance with environmental productivity and heterogeneity.

Discussion

The empirical model shows that farmers have higher proportions of violence in low productivity, high heterogeneity environments and the lowest proportions of violence in high productivity, homogeneous environments. The opposite trend is present among foragers (and to a lesser extent, horticulturalists), with the highest proportions of violence occurring in high productivity, homogeneous environments and the lowest proportions of violence occurring in low productivity, heterogeneous environments. These results show that in order to explain ecologically driven variation in human violence it is necessary to account for different subsistence economies. The opposite covariance along subsistence lines warrants explanation in relation to the adaptive motivations for violence.

These results support the expectation that the adaptive payoffs for inter-group (and inter-personal) violence are increasingly mediated by resource availability as societies develop delayed-return economies reliant on resource surpluses, storage, and privatization, all of which enable the accumulation and distribution of wealth that can be mobilized to enhance or maintain fitness rewards such as status, marriage opportunities, and alliances [64-66]. For immediate-return economies, like mobile foraging, the benefits of violence are high when resources are abundant, homogeneous, and predictable, as their economic and social organization renders resource procurement an epiphenomenal payoff relative to direct social rewards. The result of this transition is a fundamental change in the socioecological conditions that promote collective violence. While our explanation is tentative and will require further testing, it provides a single evolutionary rationalization for these seemingly divergent findings. We further unpack this explanation below.

Violence over social and material rewards

Violence used to gain, co-opt, or defend direct mating and marriage opportunities has long been a topic of controversy and empirical debate [e.g., 67]. Ethnographers report evidence showing adult males frequently fight over mating opportunities and real or perceived infidelities [13, 33, 67]. Others find that violent conflict can result in the disproportionate accumulation of wealth that can be used to gain direct mating opportunities via social systems such as bride price [12]. In either case, those who hypothesize direct mating opportunities as a prime motivation for violence emphasize its potential to directly elevate inclusive fitness [8, 12, 13, 33]. However, explanations that promote mating opportunities without reference to status competition are problematic, as reputation is tied up in all competitive activities, including mating contests.

Afterall, competition takes numerous forms, including behavioral strategies aimed at signaling fitness quality to potential mates, allies, and competitors [68-71]. Many scholars provide evidence that participation in coalitional violence acts as a form of costly signaling intended to deter competitors, encourage alliance formation, and attract mates [8, 11, 28, 35]. Observers of these costly displays benefit from honest signals that are useful for conflict avoidance, mate choice, and competitor evaluation. As the costs of participating in coalitional violence tend to be high [6, 72], dishonest signals should be rare and easily detectable. These payoffs have clear fitness consequences, as status consistently links to reproductive success [73]. Violence as a costly display also includes fights over direct mating opportunities, which signal one’s willingness to defend or co-opt access to reproductive partners. To summarize, we argue that the available evidence points to status competition as a key adaptive payoff for human violence, with conflict over mating opportunities being a component of larger reputational contests. Environmental conditions then must be linked to the variable payoffs for violent status contests, and how they are mediated by subsistence economy.

For foragers and horticulturalists, organization or participation in collective violence can produce or signal embodied and relational wealth [74] that may translate into reproductive success [73]. For agriculturalists, organization or participation in collective violence can produce material wealth that can enhance fitness [65, 73] as well as be passed down inter-generational channels [65, 74]. The key distinction is that foragers receive high payoffs for direct behavioral competition while farmers (and other delayed-return societies) benefit from resource-based competition. Seeming exceptions prove the rule: in prehistoric California, delayed-return hunter-gatherers who rely on the storage of privatized resources follow the violence pattern observed here among agriculturalists [6].

The positive covariance between violence and environmental productivity among foragers and horticulturalists speaks to the capacity of rich environments to finance frequent costly displays and render other forms of signaling less effective. For example, when high-ranking prey items are abundant and predictable, they may require less knowledge and skill to capture relative to environments where similar game is scarce and difficult to acquire [75]. In this case, the same strategy (e.g., big game hunting) will produce larger signaling payoffs in the poor environment relative to the rich one. Such ecological constraints on the number of honest signals that can be conveyed may result in a greater emphasis on status-driven violence, where abundant resources permit greater energy being allocated into violent competition. More frequent displays in productive environments may intensify competition, with positive feedback resulting in higher rates of violence. The negative covariance between violence and standard deviation in environmental productivity suggests that homogeneous environments promote violence either because abundant resources are spread evenly across the landscape, permitting a greater number of individuals to pay the steep costs of participation, or because other venues for status competition are rendered less effective when resources are evenly distributed.

Contrasting an immediate-return economy such as mobile foraging to a delayed-return economy like farming, the latter provides greater opportunities for resources to be stored, owned, and accumulated as wealth, which can be used to facilitate marriages or enhance status via conspicuous displays or generosity [12]. Following marginal utility theory, resource-related violence should be especially intense when resources are scarce, and stored in large, dense, monopolizable packages [6, 28–30]. This is because the utility of a resource diminishes with the amount of that resource an individual possesses [76] and risk preferences vary as a function of relative wealth [77, 78]. As a result, farmers (and likely pastoralists) with abundant resources should be more tolerant of theft and more risk-averse (in this case, violence-avoiding), whereas those with scarce resources should be less tolerant of theft and more risk-prone (violence-prone) [28]. To summarize, among delayed-return economies low productivity, high variance environments should promote violence from those seeking resources that can be translated to fitness payoffs and those incentivized to defend them [6, 12, 29].

Violence and starvation

One might argue that the link between resource scarce environments and violence may indicate starvation-induced behavior. Afterall, if cooperative solutions to acute starvation are unavailable, individuals will use zero-sum strategies (including violence) to obtain necessary subsistence resources regardless of MoS or other intervening variables such as age or sex. Nonetheless, we expect starvation-induced violence to be rare for three reasons.

First, if survivorship is frequently in jeopardy, one should see far broader demographic participation in resource conflict. Afterall, if starvation negatively impacts everyone, resource-based violence should involve a wide demographic subset of the population and far greater sex-based parity. Instead, intergroup violence most often involves the organization and participation of young adult males [28, 79, 80]. This is not to say that exceptions do not exist, but rather to highlight the general pattern.

Second, many of those who attempt to empirically tie violence to acute starvation or chronic nutrient deprivation have failed to do so effectively [e.g., 81], with violence more consistently resulting in increased food shortages [11, 34].

Third, as stated above, the ‘resource procurement’ hypothesis implies the capture or defense of group-level rewards, such as clustered resources or territory, that elicit collective action problems [25, 32]. Recent studies have suggested that collective action problems can be overcome when participation in inter-group violence serves to accumulate resource wealth that can be translated into fitness rewards, such as marriage opportunities [12] or status [11], suggesting resource procurement may be a proximate means of gaining social rewards rather than alleviating resource shortages.

This is not to say that human populations never undergo starvation, or that violence and starvation are totally unrelated. Rather, these factors indicate that starvation induced violence cannot explain the full, or even normal, range of variability in violence, and is thus not the primary target of selection. Instead, we argue that resource-based violence is primarily a strategy for obtaining material resources that can increase fitness by enhancing status and facilitating marriages and alliances.

Conclusion

We find that violence-promoting environmental conditions are opposite for foragers and farmers, with violence among foragers peaking in rich, homogeneous environments, and violence among agriculturalists peaking in marginal, heterogeneous environments. We argue that the opposite covariance reflects the increasing role of material resources in motivating violent competition over fitness-related payoffs such as status, marriage opportunities, and alliances. These results yield insight into the behavioral ecology of violent behavior by assessing how variability in ecological conditions and subsistence economies explain cross-cultural differences in rates of violence.

The results of this paper should, of course, be treated as presenting tentative and testable hypotheses for future investigation. More robust tests of the expectations laid out in this paper will require a larger pool of archaeological trauma datasets that differentiate between ante-mortem and peri-mortem trauma, especially among mobile hunter-gatherers, horticulturalists, and pastoralists. Ethnographic datasets must be broadened to evaluate rates of violence among complex agricultural societies (particularly non-western populations).

The present study represents a preliminary effort to address the behavioral ecology of violence across the ecological and socioeconomic spectrum. Future theoretically-grounded research will benefit from a focus on explaining environmentally mediated variability in violent behavior, with an emphasis on individual motivations as a function of differential payoffs. The evolutionary perspective presented here can help to address questions regarding the frequency and intensity of human conflict, thus providing a foundation for developing science-based tools that may aid in conflict mitigation both in the present and the future.

The complete dataset used for our analysis.

(CSV)

Click here for additional data file.

R markdown of our complete analysis.

(PDF)

Click here for additional data file.

22 Nov 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Raven Garvey

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf  and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

3. We note that figure 1 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

1. You may seek permission from the original copyright holder of Figure(s) 1 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

2. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

4. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This paper is based on a profound misunderstanding of violence and warfare. This is manifest in the opening abstract which ends with the statement: “We suggest that as societies transition from immediate return economies (e.g., foragers) to delayed return economies (e.g., farmers) material resources become an increasingly important motivation for interpersonal violence and warfare.” Interpersonal violence and warfare are two very different cultural phenomenon and cannot be conflated. Wife-beating is interpersonal violence and it can be followed and explained in the archaeological and ethnographic records; however, wife-beating and organized, lethal warfare between armed groups have completely different causalities and trajectories. To lump together all skeletal trauma and labeling it “violence” fails to distinguish accidental trauma (breaking one’s leg from a fall) from the kind of systematic trauma arising out of warfare (forearm fractures, fontal bone fractures).

The authors also fail to take into consideration chronological variability in the occurrence of violence/warfare. They conflate centuries or even millennia into single events, without realizing that violence varies markedly over time in any given area. This fact, by itself, argues against their primary hypothesis that there is a strong causal correlation between violence/warfare and subsistence economy. People in one area, with a particular subsistence economy can experience a century of warfare, and many centuries without violence. There is no sense of chronological control in this study and no effort to look at when warfare occurs and when it fades away. To simply impute warfare to skeletal trauma for a particular world area represents a basic misunderstanding of the data.

In an attempt to push an incredibly complex and diverse global data base into their hypothetical models, they have papered over so much variability as to make their conclusions irrelevant. I cannot finish this review without commenting that is singularly the most jargon-laden, almost incomprehensible paper on this topic that I have encountered. Had the authors given more consideration into making their arguments coherent rather than technical, they might have identified some of their basic flaws.

Reviewer #2: Dear authors

this is an interesting manuscript that I enjoyed reading. I add several comments below, which I hope you will find useful. In my view, the two critical issues of your work is to add a stronger discussion over the potential bias introduced by combining together archaeological and ethnographic data, and and on the use of NPP to measure ecological effects on violence.

Kind regards

Abstract

Line 47: I don't think 'career' is a good word here; you could use 'behaviour' or 'social behaviour'

Introduction

This section is a bit too concise, which negatively affects clarity, and it leaves several questions unanswered. A lot of work done on violence in ancestral human societies has used data on different geographic areas and cultures, including work you cite (e.g. Bowles 2009), so it is not clear why and how your cross-cultural approach is novel. Also, theories on violence from behavioural ecology make predictions that can be tested and potentially proved to be correct, irrespective of the cultural milieu in which violence is observed. You also don't cite some seminal work on the topic, including on the role of culture, for example Zefferman & Mathew 2005.

Line 67: It would be useful to have a definition of violence at the start of the introduction. Do you use the term for all sorts of violent-like behaviours, including bullying and passive aggression, or it's only for potentially lethal violence?

Lines 80-82: I think you should mention here that some studies have not found a positive relationship between participation to war and adaptive benefits (e.g. Ferguson 1989; Beckerman et al 2009).

Lines 83-88: I don't fully get your point here. You say you want to focus on cross-cultural variation and not on the payoffs of violence, but then your aim is to analyse the ecological and economic conditions triggering violence. Ecological and economic conditions are clearly linked to payoffs, so I don't understand the distinction you are trying to make. Moreover, ecological and cultural factors are linked together; for example, a group living in a highly seasonal environment may be forced to fight with other groups during periods of food scarcity and also develop group norms that reward warriors, both factors affecting payoff. I think this section should be clarified to make your point stronger.

Line 107: it would be useful to explain in details how your dataset represent an expansion of previous ones. Is this in relation to societies included, precision of the data, range of variables considered or else?

Methods

Lines 132-134: you should discuss here whether/to what extent you can reliably say that these signs of trauma are due to pre-/peri-mortem violence or to post-mortem rituals. These has been a lot of debate whether traumas on bones are really reliable indicators of violence; e.g. see Fry's book (2013).

Line 136: you should give the definition of NPP here, the first time you introduce this acronym

Lines 136-138: if this is a recommendation for future studies, it would be better placed in the conclusions

Line 163: how far back in time are the archaeological data from? These has been a lot of environmental changes between prehistoric times and now, and if we also include global warming I don't think NPP data collected between 2000 and 2015 is a reliable measure of habitat productivity to be linked to violence occurring thousands (or even hundreds) of years before. You claim that "While NPP provides a measure of modern productivity, the relative NPP ranking of each society should have remained consistent over time due to their broad geographic range representing general physiographic regions rather than variants within single ecological or climatological regimes", but this sounds a weak claim to me. Do you have any evidence to support this claim? is there any reference to previous work that has tested this assumption? There have been both micro-and macro-climatic changes at the end of the Holocene, that have been suggested to lead to changes in social organisation and the emergence of agriculture (e.g. Shennan 2018), so this point is really critical for the reliability of your findings.

Line 181: it would be useful to add a binary control variable, at least in the preliminary analyses, to see if you find any difference between archaeological and ethnographic data, otherwise you cannot really disentangle whether difference between type of societies are real or due to the fact that different data are available for different societies

Line 191: I believe you should compare a null model (containing no predictor variables) with your two base and full models. If there is no significant improvement from the null model, you should make no claims that NPP affects violence.

Results

I don't understand from this section or from the methods, whether you have data on MoS for the archaeological and ethnographic data or just for the latter. Why not analysing the two datasets separately, since you say above that there are differences in how they can estimate violence?

Discussion

Your results are in line with what one would predict based on previous work in behavioural ecology. However, I cannot see any discussion about the potential bias in the findings, introduced by combining together archaeological and ethnographic data, and on how reliable NPP is to test your predictions. These are key aspects that would need to be addressed or at least highlighted to the reader to help a more critical evaluation of your findings.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

6 Jan 2022

January 3, 2021

PONE-D-21-31553

Subsistence strategy mediates ecological drivers of human violence

PLOS ONE

We would very much like to thank the reviewers for their incisive and valuable comments.

Our revised manuscript and the responses presented here focus on addressing the three main concerns expressed by the academic editor. Below, we provide a point-by-point response to reviewer comments. Where major in-text changes were made we highlight revised text with red font.

Academic editor’s main points:

(1) Reviewer #1’s requests for clarification

(2) the same reviewer’s concern that environmental change complicates the assumption that modern estimates of NPP accurately reflect prehistoric ones.

(3) providing additional support for the assumption that skeletal trauma is a reliable measure of violence per se, a concern shared by both reviewers.

(4) The map figure copyright concern: the map used in our manuscript is an original creation by the authors and thus does not have any copyright violation.

Reviewer #1

Comment: Interpersonal violence and warfare are two very different cultural phenomenon and cannot be conflated. Wife-beating is interpersonal violence and it can be followed and explained in the archaeological and ethnographic records; however, wife-beating and organized, lethal warfare between armed groups have completely different causalities and trajectories.

Response: Great point regarding definitions. We clarify throughout the revised manuscript to focus on interpersonal (primarily inter-group) violence. However, we should note that intra- and inter-group violence tend to co-occur (Lutz 2007; McCool et al. 2020; Nordstrom, 1998; Tung, 2014), and that changes in intra-group violence should be affected by the process outlined in our theoretical setup. For example, resource scarcity may promote intra-group competition that increases interpersonal violence (Daly, 2017), and status and mating competition may intensify within group animosities that result in interpersonal violence (Chagnon, 2013).

Comment: To lump together all skeletal trauma and labeling it “violence” fails to distinguish accidental trauma (breaking one’s leg from a fall) from the kind of systematic trauma arising out of warfare (forearm fractures, fontal bone fractures).

Response: We apologize for the confusion. Our archaeological dataset relies entirely on publications that explicitly provide trauma data that is attributable to interpersonal violence and excludes accidental trauma and post-mortem damage. We have revised our methods section to clarify.

Comment: The authors also fail to take into consideration chronological variability in the occurrence of violence/warfare. They conflate centuries or even millennia into single events, without realizing that violence varies markedly over time in any given area. This fact, by itself, argues against their primary hypothesis that there is a strong causal correlation between violence/warfare and subsistence economy. People in one area, with a particular subsistence economy can experience a century of warfare, and many centuries without violence. There is no sense of chronological control in this study and no effort to look at when warfare occurs and when it fades away.

Response: We feel it is important to note that our primary result is not that there is “a strong causal correlation between violence/warfare and subsistence economy” (nor is it our “primary hypothesis”). It is the case with all multiregional archaeological analyses that chronological gaps will exist. Afterall, there is not a single region in which complete temporal resolution exists for data on violent trauma. Given this, we flatten temporal depth because we are interested in whether large physiographic regions differ in average rates of violence in response to variation in macroscale ecological conditions. Far from unusual, averaging diachronic variability is standard practice for multiregional studies in bioarchaeology (e.g., Allen et al. 2016; Arkush and Tung 2013 (flatten intraperiod variation); Beier et al. 2020; Delgado-Darias 2017; Fibiger et al. 2013; Kohler et al. 2014; Milella et al. 2020; Mummert et al. 2011; Pomeroy and Stock 2012; Scott and Buckley 2010:512; Standen et al. 2020; Steadman 2008:59; Vercellotti et al. 2014; Zhang et al. 2020), particularly in areas with limited radiocarbon datasets.

Comment: To simply impute warfare to skeletal trauma for a particular world area represents a basic misunderstanding of the data.

Response: As stated in the methods section, the archaeological dataset is a measure of generalized interpersonal violence, not warfare. Our revised methods section also points out that our dataset does not include all skeletal trauma, only violent trauma. Because rates of intra-group violence tend to be low in small-scale societies (Bohem 1999; Wrangham et al. 2006) and co-occur with inter-group violence (Nordstrom 1998; McCool et al. 2020; Lutz 2007; Tung 2014) – see revised intro and methods sections – changes in rates of population-level generalized violence will largely be attributable to fluctuations in rates of inter-group violence, which should co-occur with intra-group conflict. However, as stated above, changes in intra-group violence should be affected by the process outlined in our theoretical setup. Our main goal with this paper is to assess how macroscale ecological conditions impact generalized rates of interpersonal violence, and whether subsistence economy has a mediating effect.

Comment: In an attempt to push an incredibly complex and diverse global data base into their hypothetical models, they have papered over so much variability as to make their conclusions irrelevant. I cannot finish this review without commenting that is singularly the most jargon-laden, almost incomprehensible paper on this topic that I have encountered. Had the authors given more consideration into making their arguments coherent rather than technical, they might have identified some of their basic flaws.

Response: We contend that a key strength of our dataset is that it is “complex and diverse”. It is not possible to address the reviewer's concern that “they have papered over so much variability as to make their conclusions irrelevant” without additional details regarding how the complexity and diversity of our dataset renders our model results (from which our conclusions follow) “irrelevant.” As for the comment regarding the language and clarity of our paper, it is fully in-line with the broader literature on the evolutionary and ecological drivers of human violence (e.g., Allen et al. 2016; Bowles 2009; Glowacki et al. 2017; Zefferman and Mathew 2015; etc.). We suggest that the “incomprehensibility” may stem more from divergent research traditions between the authors and reviewer, rather than incoherent writing.

Reviewer #2

Comment: Line 47: I don't think 'career' is a good word here; you could use 'behaviour' or 'social behaviour'

Response: Changed to ‘behavior.’

Comment: This section is a bit too concise, which negatively affects clarity, and it leaves several questions unanswered. A lot of work done on violence in ancestral human societies has used data on different geographic areas and cultures, including work you cite (e.g. Bowles 2009), so it is not clear why and how your cross-cultural approach is novel.

Response: We have expanded the introduction, included new citations, and worked to clarify our setup with more explicit theoretical expectations. While our primary focus is on the adaptive payoffs of interpersonal and inter-group violence, our revised introduction includes content recognizing some of the broader research efforts, including: “Recent research also explores the coevolution of inter-group violence with intra-group cooperation, altruism, and the emergence of complex social systems (Bowles 2009; Rusch 2014; Zefferman and Mathew 2015).”

We agree that a lot of excellent work has been done on cross-cultural approaches to human violence. However, the majority of the formal cross-cultural research focuses on (1) the co-evolution of coalitional violence with cooperation, parochial altruism, and cultural group selection (e.g., Bowles 2009), (2) the categories of available rewards for participation (e.g., Glowacki and Wrangham 2013), and (3) punishing free-riders (e.g., Mathew and Boyd 2011). While these studies are vitally important, our study provides a novel perspective by seeking to explain how variation in large-scale ecological and economic conditions structure the social and material payoffs for inter-group (and interpersonal) violence. To our knowledge, there is no published research evaluating how macroscale resource conditions and modes of subsistence affect the adaptive payoffs for violence, especially on the scale presented in our research. Our findings: that violence among foragers and horticulturalists peak in productive, homogeneous environments, while violence among farmers peaks in marginal, heterogeneous environments, are novel and have implications for the evolution of interpersonal violence and warfare.

Comment: You also don't cite some seminal work on the topic, including on the role of culture, for example Zefferman & Mathew 2005.

Response: Thank you for clueing us into this literature. This reference and several like it have been included in our revised introduction.

Comment: Line 67: It would be useful to have a definition of violence at the start of the introduction. Do you use the term for all sorts of violent-like behaviours, including bullying and passive aggression, or it's only for potentially lethal violence?

Response: We clarify throughout to focus on inter-group violence (warfare, raids, feuds) and to a lesser extent general interpersonal violence, although much of the processes outlined in the paper should impact both intra- and inter-group violence (e.g., status and mating competition).

Comment: Lines 80-82: I think you should mention here that some studies have not found a positive relationship between participation to war and adaptive benefits (e.g. Ferguson 1989; Beckerman et al 2009).

Response: We include Beckerman et al. 2009 in the revised introduction to show that debate persists. We do not include the Ferguson 1989 paper as Chagnon 1989 provides a robust refutation of Ferguson’s claims.

Comment: Lines 83-88: I don't fully get your point here. You say you want to focus on cross-cultural variation and not on the payoffs of violence, but then your aim is to analyse the ecological and economic conditions triggering violence. Ecological and economic conditions are clearly linked to payoffs, so I don't understand the distinction you are trying to make. Moreover, ecological and cultural factors are linked together; for example, a group living in a highly seasonal environment may be forced to fight with other groups during periods of food scarcity and also develop group norms that reward warriors, both factors affecting payoff. I think this section should be clarified to make your point stronger.

Response: We have expanded our theoretical setup to provide greater clarity and set out more explicit expectations. As suggested by the reviewer, resource violence and social rewards may very well be linked, and we address this possibility in the revised introduction and discussion sections. However, many evolutionary anthropologists studying violence find resource procurement to be an irrelevant or epiphenomenal motivation (e.g., Chagnon 2013; Glowacki and Wrangham 2013; Kelly 2005; Macfarlan et al. 2018), instead positing that direct access to social rewards (i.e., status, mating, alliances) provide the key indirect fitness payoffs. Alternatively, many scholars studying inter-group violence propose that the primary motivation for violence is the procurement or defense of scarce, monopolizable resources (e.g., Allen et al. 2016; Ember and Ember, 1992; Field, 2008; LeBlanc 1999; Vayda, 1976), with social rewards taking an epiphenomenal position.

We propose to test these hypotheses by noting that each hypothesis implies a different functional relationship between violent behavior and socioecological conditions. If the resource procurement hypothesis is correct (if the payoffs to violence are primarily material resources), then violence should positively co-vary with a scarce and variable resource distribution. If the social rewards hypothesis is correct (if the payoffs to violence are social rewards), then violence should positively covary with an abundant and predictable resource distribution. We recognize in the revised introduction that, as the reviewer states, fighting over resources can be linked to social rewards, but contend that in this case resource violence would only be initiated with resource scarcity, and would diminish as resources become increasingly abundant and predictable. In addition to this basic test, our analysis offers a novel avenue for the assessment of resource violence, in particular whether it yields increasingly high payoffs with the emergence of sedentism, storage, privatization, and wealth. We summarize these divergent expectations with a new figure (Fig 1).

Comment: Line 107: it would be useful to explain in details how your dataset represent an expansion of previous ones. Is this in relation to societies included, precision of the data, range of variables considered or else?

Response: We added: “The dataset developed here builds off prior studies by providing additional archaeological and ethnographic samples and improves precision by excluding archaeological groups with small sample sizes.”

Comment: Lines 132-134: you should discuss here whether/to what extent you can reliably say that these signs of trauma are due to pre-/peri-mortem violence or to post-mortem rituals. There has been a lot of debate whether traumas on bones are really reliable indicators of violence; e.g. see Fry's book (2013).

Response: Thank you for this suggestion. The revised methods section states that bioarchaeologists have robust methods for differentiating these types of trauma. For each of our archaeological samples only violent skeletal trauma is included, with post-mortem damage/rituals and accidents being excluded.

Comment: Line 136: you should give the definition of NPP here, the first time you introduce this acronym.

Response: This has been fixed.

Comment: Lines 136-138: if this is a recommendation for future studies, it would be better placed in the conclusions

Response: We include this caveat in the methods section as a preemptive response to readers who may be wondering why our samples are clustered in certain world regions. This clustering is due to the paucity of existing data rather than our own oversight. We include an additional sentence in the revised conclusion section calling for additional data in more diverse ecological settings.

Comment: Line 163: how far back in time are the archaeological data from? These has been a lot of environmental changes between prehistoric times and now, and if we also include global warming I don't think NPP data collected between 2000 and 2015 is a reliable measure of habitat productivity to be linked to violence occurring thousands (or even hundreds) of years before. You claim that "While NPP provides a measure of modern productivity, the relative NPP ranking of each society should have remained consistent over time due to their broad geographic range representing general physiographic regions rather than variants within single ecological or climatological regimes", but this sounds a weak claim to me. Do you have any evidence to support this claim? is there any reference to previous work that has tested this assumption? There have been both micro-and macro-climatic changes at the end of the Holocene, that have been suggested to lead to changes in social organisation and the emergence of agriculture (e.g. Shennan 2018), so this point is really critical for the reliability of your findings.

Response: To address this point, we added the following content to the revised methods section: “Modern NPP has been used to predict numerous prehistoric phenomena including population density (Bradshaw et al. 2019; Eriksson et al. 2012; Timmermann and Friedrich 2016), habitat colonization (Codding and Jones 2013), resource scarcity (Allen et al. 2016) and more. While NPP provides a measure of modern productivity, the relative NPP ranking of each region should have remained consistent over time due to their broad geographic range representing general physiographic regions (i.e., bioregions) rather than variants within single ecosystems (see, Bioregions 2020). To better illustrate this point, we take as an example two regions included in our sample: the North American Colorado Plateau (CP) and the Illinois river valley (IRV). While the absolute NPP of these two regions certainly fluctuated throughout the centuries, there was no point during which the “mountain grassland and scrublands” of the CP were more environmentally productive (higher NPP) than the “tropical and subtropical dry broadleaf forests” of the IRV (Bioregions 2020; Dinerstein et al. 2017). The relative differences in environmental productivity between these two regions are due to their differing physiographic characteristics resulting in very different ecologies, which have remained comparatively consistent for millennia. The same can be said about the relative variation in NPP between the other bioregions compared in our analysis and the observations clustered therein (e.g., the lowland neotropics of the Amazon Basin versus the Canadian boreal forests). At most, changes in absolute NPP would only serve to exaggerate or flatten the modeled curves, but would not lead to trend reversals.”

Comment: Line 181: it would be useful to add a binary control variable, at least in the preliminary analyses, to see if you find any difference between archaeological and ethnographic data, otherwise you cannot really disentangle whether difference between type of societies are real or due to the fact that different data are available for different societies

Response: We include a binomial model in the supplement that evaluates data type. We also include a new “Data Compatibility” section in the revised methods that works to address this concern and others mentioned below. In the revised introduction and methods section we discuss the need to used mixed archaeology/ethnography dataset due to inherent limitations in the available violence data.

Comment: Line 191: I believe you should compare a null model (containing no predictor variables) with your two base and full models. If there is no significant improvement from the null model, you should make no claims that NPP affects violence.

Response: In our revised manuscript we generate a null model and conduct a likelihood ratio test comparing the null model to the base model and the full model to the base model. The base model results in a significant improvement from the null model (X2 = 0.47, p = 0.0458), while the MoS (full) model results in a significant improvement from the base model (X2 = 2.13, p < 0.0001). More details are available in the revised manuscript and Supplement 2.

Comment: I don't understand from this section or from the methods, whether you have data on MoS for the archaeological and ethnographic data or just for the latter. Why not analysing the two datasets separately, since you say above that there are differences in how they can estimate violence?

Response: We apologize for the confusion. We categorized all samples by MoS and include an additional sentence at the beginning of the revised methods section to clarify. We do not analyze the datasets separately because, due to differential research traditions, ethnographic data is restricted to forager and horticultural groups while archaeological data almost entirely focus on agricultural groups. Thus, the ethnographic dataset would not have sufficient sample size to test the agricultural MoS while the archaeology dataset would have the same problem for foragers and horticulturalists. Thus, the combined approach taken here is necessary until ethnographic violence data is obtained for agriculturalists (especially non-western, non-state societies), and archaeological violent trauma data is generated for forager and horticultural societies.

Comment: Your results are in line with what one would predict based on previous work in behavioural ecology. However, I cannot see any discussion about the potential bias in the findings, introduced by combining together archaeological and ethnographic data, and on how reliable NPP is to test your predictions. These are key aspects that would need to be addressed or at least highlighted to the reader to help a more critical evaluation of your findings.

Response: We included additional content in the revised discussion and conclusion sections regarding the combining of archaeological and ethnographic datasets. We acknowledge that differing data types may affect our models. However, until better datasets are available (see revised methods section), the approach taken here constitutes the best practice available, and our results follow strongly from theoretical expectations. We expanded our conclusion section to include these caveats and how we propose to treat them. We also included additional content in the methods about the use of relative NPP values and physiographic characteristics.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

24 Feb 2022

7 Apr 2022

We would like to thank the editor and reviewers for their insightful commentary. Below, we provide a point-by-point response to review comments.

Reviewer comments:

Reviewer #2

Comment: Line 55, "We argue that the trend reversal between foragers and farmers": here I think farmers refer to agriculturalists, but at line 53 you use the term 'forager/farmers' for horticulturalists, so I am not sure where you draw the line between foragers and farmers in this sentence.

Response: We edited this sentence to refer specifically to agriculturalists.

Comment: Lines 55-60: I am ok with your argument, but it doesn't clearly follow from your finding that resource availability affects differently foragers and farmers. For example, why would farmers acquire "material resources to be transformed into social payoffs" only or more in unproductive environments?

Response: Farmers should acquire fitness-linked resource in all environments, but where resources are scarce and unevenly distributed individuals will receive higher relative payoffs for using violence to gain resources. Those with abundant resources should be more tolerant of theft and less willing (relative to those in poor environments) to use risk-prone strategies such as violence to obtain resources. As such, we don’t argue that farmers use violence to obtain resources only in poor environments, but that the payoffs for violence will be higher relative to those living in productive, homogeneous environments. We provide more detail on these relative payoffs using marginal utility theory on lines 365 – 374.

Comment: Line 73: I really don't think you can use data on chimpanzees to make a claim about all non-human animals. If you look at the literature on other species, you will find that lethal violence is probably way more common in humans in comparison to the majority of other primates and mammals.

Response: We agree and find that this sentence is distracting and have thus deleted it to save this point of discussion for later in the manuscript.

Comment: Line 276: it should read "while FOR farmers VIOLENCE negatively co-vary with productivity and homogeneity"

Response: This sentence has been fixed.

Line 321 "directly elevate inclusive fitness." I think there should be some refs at the end of this sentence

Response: We have added four references that support this statement.

Reviewer #3

I have only a few suggestions for clarification on some points, but also a recommendation for what would be a more major undertaking (dividing the foraging sample, which might require a larger sample size), and so therefore it is not required but might be considered for future.

Comment: 80/ Not sure that all would agree that those seeing inter-group conflict as rare/maladaptive have been effectively countered, though the authors do acknowledge that this debate continues.

Response: Fair point. Our objective with this statement is to acknowledge that important debate continues, but that debate is not the focus of the present research. We added additional text to emphasize that the ‘effective counters’ are for most, but not all, cases.

Comment: Odd to rely on the ethnographic record alone for foragers and horticulturalists when there is considerable archaeological evidence that might be drawn upon, though it is admittedly patchy. Some of the previous studies cited do drawn upon the archaeological record for foragers.

Response: We do include several archaeological samples for horticulturalists, but these data are indeed rare. As for foragers, most of the data, while extremely valuable, are highly localized and often suffer from exceedingly small samples sizes that are below our sample size cutoff (lines 163-164). We are in the process, over the next several years, of putting together a much larger archaeo-ethnographic database which will expand upon the present database and will hopefully provide a more comprehensive archaeological database on foragers. This does assume however that the recent increase in bioarchaeology studies on violence among small scale societies continues. For the time being however, we must rely on the limited information available and look forward to the production and release of additional datasets. We hope with the publication of this manuscript to encourage future work generating trauma datasets.

Comment: 159/ “This database operationalizes sigma by defining it as the number of individual skeletons with evidence of violent trauma divided by the total skeletal sample for each archaeological observation”; repeated line 187

�  Surely this defines prevalence rather than sigma (a measure of variability), though from ln 258 it is being used here as a measure of the ‘rate of violence’?

Response: We changed “sigma” to “PV” for “proportional violence” and changed “rate of violence” to “proportion of violence” throughout.

Comment: 317/ “Ethnographers report evidence showing adult males frequently fight over mating opportunities and real or perceived infidelities.”

�  This seems to contradict the previous statement regarding the low incidence of in-group violence conflict, assuming that opportunities for such infidelities are likely to be mainly within the group

Response: We feel this is an important point to make, as intra-group hostilities do certainly occur, even if the rates are much lower relative to inter-group rates. Our point here is that when mating rewards are at stake, motives for inter-personal violence should be consistent regardless of whether antagonists are within group or outsiders.

Comment: 345/ “Seeming exceptions prove the rule: in prehistoric California, delayed-return hunter-gatherers who rely on the storage of privatized resources follow the violence pattern observed here among agriculturalists.”

�  So this is not taken into account in the analysis, and the ‘foraging’ mode of subsistence includes all h-g? Not clear how this follows the violence pattern seen among agriculturalists as stated, as h-g practicing storage are often in relatively rich environments (such as those of CA and the NWC), whereas the discussion here focussed on the increased violence among agriculturalists being predicted by more marginal/unpredictable environments. This seems a discrepancy rather than an accord.

Response: This an excellent point about CA and NWC in relation to environmental productivity and conflict. Unfortunately, we cannot further investigate empirically how these two areas fit into our models for two reasons: First, the NWC does not contain sufficient violent trauma data to be included in our models; Second, the Central California bioarchaeological database does not contain a measure for generalized violent trauma (which is how violence is measured for all of our archaeology samples), as the database divides all individual traumas into either blunt or sharp force. As such, individuals could potentially be double counted if they exhibit both sharp and blunt force trauma. We cite the CA study because these largely delayed-return foragers follow a pattern similar to the farmers in our model. While the Central CA coast is indeed productive, we would need to model this relationship to assess where they fit in along the NPP and violence spectra, which we cannot, unfortunately, do.

�  Why not divide foragers? The division between immediate- and delayed-return h-g is well established and in many ways it is odder to lump them than to divide them.

Response: We completely agree that this analysis could be improved by converting categorial subsistence economies to a continuous scale from immediate- to delayed-return. However, we should note that the foraging societies included in our database are immediate-return, and have revised the text to include this information (line 127). For the present analysis, the pool of available data is simply not large enough to further divide these data into more fine-grained economic categories, especially in relation to model sample sizes. For now, we are specifying the best possible models with the given data, which we hope will improve in size in quality in the coming years. We hope our research here will encourage future research on prehistoric violent trauma, particularly among foragers and pastoralists.

Comment: 382/ “First, if survivorship is frequently in jeopardy one should see far greater demographic parsimony regarding participation in resource conflict”

�  Not clear that this follows, as there is little point in all and sundry engaging in violence (children, the elderly?) when they are more likely to be a hindrance than help to those better suited to violent conflict through both age and skills (e.g., hunting).

Response: Fair point. We do not mean to suggest that demographic participation should be universal if violence functioned to prevent starvation, but merely to state that participation should be much broader than the typical pattern of young adult males. Certainly, if basic survivorship were in jeopardy and violence was the only means of somatic maintenance we should expect far greater sex-based parity in participation, yet this does not commonly occur. The point we make is simply that the common pattern of young adult male organization and participation in collective violence heavily implies motives other than basic somatic maintenance in the face of starvation. We added a statement to clarify that a key point is the notable absence of sex-based parity in participation.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

26 Apr 2022

Subsistence strategy mediates ecological drivers of human violence

PONE-D-21-31553R2

Dear Dr. McCool,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing any required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Raven Garvey, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

13 May 2022

PONE-D-21-31553R2

Subsistence strategy mediates ecological drivers of human violence

Dear Dr. McCool:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr Raven Garvey

Academic Editor

PLOS ONE