A brief guide to model selection,multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion

Akaike’s information criterion (AIC) is increasingly being used in analyses in the field of ecology. This measure allows one to compare and rank multiple competing models and to estimate which of them best approximates the “true” process underlying the biological phenomenon under study. Behavioural ecologists have been slow to adopt this statistical tool, perhaps because of unfounded fears regarding the complexity of the technique. Here, we provide, using recent examples from the behavioural ecology literature, a simple introductory guide to AIC: what it is, how and when to apply it and what it achieves. We discuss multimodel inference using AIC—a procedure which should be used where no one model is strongly supported. Finally, we highlight a few of the pitfalls and problems that can be encountered by novice practitioners.     

Social personality traits in chimpanzees: temporal stability and structure of behaviourally assessed personality traits in three captive populations

Animals of many species show consistency in behaviour across time and contexts that differs from other individuals' behaviour in the same population. Such ‘personality’ affects fitness and has therefore become an increasingly relevant research topic in biology. However, consistent variation in social behaviour is understudied. In socially living species, behaviour occurs in a social environment and social interactions have a significant influence on individual fitness. This study addressed personality in social behaviour of 75 captive chimpanzees in three zoos by coding observed behaviour. Fifteen behavioural variables were significantly repeatable (range 0.21–0.93) in at least two of the three zoos. The behaviours showed considerable long-term stability across 3 years, which did not differ from the short-term repeatability. The repeatable behaviours were then analysed with factor analyses. They formed five independent factors, three of which consisted of social traits and were labelled ‘sociability’, ‘positive affect’ and ‘equitability’. The two non-social behaviour factors were labelled ‘anxiety’ and ‘activity’. The factor scores were analysed for sex and population differences. Males had higher factor scores in all traits except ‘sociability’. The factor scores differed also between the zoos, implying considerable external effects in trait expression. The results show that chimpanzees show personality in a broad range of social and non-social behaviours. The study highlights the importance of assessing personality in the social behaviour, especially in cohesive social species, as only then can we understand the consequences of personality in socially living species.     

Effects of male sexual harassment on female time budgets,feeding behavior,and metabolic rates in a tropical livebearing fish (<Emphasis Type="Italic">Poecilia mexicana</Emphasis>)

Sexual harassment by males has the potential to affect almost any aspect of female behavior and life history. Using Atlantic mollies (Poecilia mexicana) as a model—a species in which males do not court but almost constantly try to forcefully mate with females—we asked whether and how male harassment influences (a) females’ time budgets and (b) feeding rates (e.g., through frequent flight from male approaches), and (c) whether metabolic rates are increased as a response to stress. Field observations in a natural P. mexicana population revealed that males (average feeding rate 15%) spent far less time feeding than females (60%), and clearly traded off frequent pursuit of females with foraging. Most importantly, females’ feeding times were dramatically reduced when being pursued by a harassing male. Also in standardized lab experiments, females spent significantly less time feeding when accompanied by a male as compared to being in the presence of another female. This effect was also observed when partner fish (male or female) were presented only visually, but could not interact physically with the focal female. It seems, therefore, that females increase vigilance when a harassing male is around, which keeps them from feeding even before males actually approach them. Based on the latter result, we asked whether a stress-induced increase in metabolic rates would be discernible. We measured oxygen consumption and gill ventilation frequencies (opercular rates) of females in different social contexts (alone, with another female, or a male). The predicted, strong body mass dependency of both physiological parameters was uncovered, but no evidence for an effect of social context was detected. We argue that male harassment represents such a constant (but non-lethal) stressor for poeciliid females that their metabolic stress responses have adapted to this through habituation.     

Egg composition in relation to social environment and maternal physiological condition in the collared flycatcher

Offspring survival can be influenced by resources allocated to eggs, which in turn may be affected by the environmental factors the mother experiences during egg formation. In this study, we investigated whether experimentally elevated social interactions and number of neighbouring pairs influence yolk composition of collared flycatchers (Ficedula albicollis). Social challenge was simulated by presentation of a conspecific female. Experimental females spent more time near the cage and produced eggs with higher androgen concentration, but local breeding density did not affect yolk androgen level. Moreover, we found that females exposed to more intra-specific interactions and those that bred at higher density produced eggs with smaller yolk. These females may be more constrained in foraging time due to more frequent social encounters, and there might be increased competition for food at areas of higher density. In contrast, the present study did not reveal any evidence for the effect of social environment on yolk antioxidant and immunoglobulin levels. However, we found that yolk lutein and immunoglobulin concentrations were related to the female’s H/L ratio. Also, yolk lutein and α-tocopherol levels showed a seasonal increase and were positively related to the female’s plasma carotenoid level. Mothers may incur significant costs by transferring these compounds into the eggs, thus only females in good physiological condition and those that lay eggs later, when food is probably more abundant, could allocate higher amounts to the eggs without compromising their defence mechanisms. Our results suggest that environmental circumstances during egg formation can influence conditions for embryonic development.     

Prior contest information: mechanisms underlying winner and loser effects

Animals’ contest performance is influenced by their recent contest experiences. This influence could either be exerted by individuals re-estimating their own fighting ability (self-assessment) or by their opponents responding to status-related cues (social-cue mechanism) or both. Individuals of Kryptolebias marmoratus, a hermaphroditic killifish, were given different contest experiences to examine how two opponents’ prior experiences combined to determine their contest interaction and to test both of these mechanisms as potential causes of the observed experience effect. Our data showed that losers’ decisions to retreat at different stages of a contest were influenced by their own but not by the winners’ contest experience—a result consistent with self-assessment but not with the social-cue mechanism. An association between the fish initiating and winning contests thus probably arose because both were correlated with an individual’s assessment of its fighting ability, but not because initiating contests made opponents more inclined to retreat.     

Social context-dependent immune gene expression in bumblebees (<Emphasis Type="Italic">Bombus terrestris</Emphasis>)

Social insects are prone to attack by parasites as they provide numerous resources of food and brood, homeostatic nest conditions and a high density of individuals, enhancing the transmission of parasites. The defence of social insects might occur at different levels, the individual and the group. Individual defence occurs in part via the innate immune system resulting in the expression of antimicrobial substances. Group level defences, summarised as ‘social immunity’, represent a suite of behavioural and organisational features. Here, all effects contributing to social immunity except for the social context were removed from bumblebee (Bombus terrestris) workers, kept either in groups or solitarily. The gene expression of six effector molecules of the immune system was monitored in both groups and in controls from the same source colonies. The social treatment has a highly significant effect on immune gene expression, with groups exhibiting higher levels of two antimicrobial peptides (AMPs) and two lysozymes. Phenoloxidase is affected at the regulatory level, with a strong upregulation of its suppressor Spn27A in groups suggesting a trade-off with antimicrobial activity. AMPs are strongly upregulated in groups, whereas lysozymes are strongly downregulated in solitary treatments suggesting another trade-off. Clearly, social immunity impacts elements of individual immunity.     

Model averaging,missing data and multiple imputation: a case study for behavioural ecology

Model averaging, specifically information theoretic approaches based on Akaike’s information criterion (IT-AIC approaches), has had a major influence on statistical practices in the field of ecology and evolution. However, a neglected issue is that in common with most other model fitting approaches, IT-AIC methods are sensitive to the presence of missing observations. The commonest way of handling missing data is the complete-case analysis (the complete deletion from the dataset of cases containing any missing values). It is well-known that this results in reduced estimation precision (or reduced statistical power), biased parameter estimates; however, the implications for model selection have not been explored. Here we employ an example from behavioural ecology to illustrate how missing data can affect the conclusions drawn from model selection or based on hypothesis testing. We show how missing observations can be recovered to give accurate estimates for IT-related indices (e.g. AIC and Akaike weight) as well as parameters (and their standard errors) by utilizing ‘multiple imputation’. We use this paper to illustrate key concepts from missing data theory and as a basis for discussing available methods for handling missing data. The example is intended to serve as a practically oriented case study for behavioural ecologists deciding on how to handle missing data in their own datasets and also as a first attempt to consider the problems of conducting model selection and averaging in the presence of missing observations.     

Age- and state-dependent reproductive effort in male mountain goats, Oreamnos americanus

In mammalian polygynous mating systems, male reproductive effort consists mainly of male–male competition and courting of females, which entail substantial somatic costs. Males are thus expected to adjust their reproductive effort according to their age and condition. In this study, we examined how activity budgets of male mountain goats (Oreamnos americanus), a polygynous ungulate, varied with age in a marked population over two periods: (1) summers 1995–2006 and (2) ruts 2004–2006. We then assessed if the proportions of time spent in male–male competition and courtship behaviors were influenced by age-specific body mass and social rank during the rut. Males spent most of their time foraging and resting during summer, and rested more and foraged less with increasing age. During the rut, pronounced shifts in activity budgets occurred as juveniles (1–2 years) increased time spent foraging, whereas adults (≥3 years) increased standing and time spent in social interactions at the expense of foraging. At old age, reproductive effort either stabilized or decreased slightly, providing weak support for the ‘mating strategy–effort’ hypothesis, predicting that courtship behaviors should peak in prime-aged males. Age-specific body mass did not affect time spent in male–male competition, but was positively related with time spent in courtship behaviors, providing support for the ‘individual quality’ hypothesis, predicting that males with more resources at the start of the rut should spend more time in mating-related activities. Age-specific social rank did not affect reproductive effort. Surviving to prime age while increasing mass each year should thus allow male ungulates to gain greater ability to court estrus females.     

Extra-pair paternity in alpine marmots, <Emphasis Type="Italic">Marmota marmota</Emphasis>: genetic quality and genetic diversity effects

Assuming that a male’s genetic characteristics affect those of his offspring, extra-pair copulation has been hypothesized to increase heterozygosity of the progeny—the “genetic compatibility” hypothesis—and the genetic diversity within litters—the “genetic diversity” hypothesis. We tested these two hypotheses in the alpine marmot (Marmota marmota), a socially monogamous mammal showing a high rate of extra-pair paternity (EPP). In a first step, we tested the assumption that a male’s genetic characteristics (heterozygosity and genetic similarity to the female) affect those of his offspring. Genetic similarity between parents influenced offspring heterozygosity, offspring genetic similarity to their mother, and litter genetic diversity. The father’s heterozygosity also influenced litter genetic diversity but did not affect offspring heterozygosity. Hence, heterozygosity seems not to be heritable in the alpine marmot. In a second step, we compared genetic characteristics of extra-pair young (EPY) and within-pair young (WPY). EPY were less genetically similar to their mother but not more heterozygous than WPY. EPY siblings were also less genetically similar than their WPY half siblings. Finally, the presence of EPY promoted genetic diversity within the litter. Thus, our data support both the “genetic compatibility” and the “genetic diversity” hypotheses. We discuss further investigations needed to determine the primary causes of EPP in this species.     

An oligarchy of nest-site scouts triggers a honeybee swarm’s departure from the hive

Animals that travel in groups must synchronize the timing of their departures to assure cohesion of the group. While most activities in large colonies of social insects have decentralized control, certain activities (e.g., colony migration) can have centralized control, with only a special subset of well-informed individuals making a decision that affects the entire colony. We recently discovered that a small minority of individuals in a honeybee colony—an oligarchy—decides when to trigger the departure of a swarm from its hive. The departure process begins with some bees producing the worker-piping signal (the primer for departure) and is followed by these bees producing the buzz-run signal (the releaser for departure). In this study, we determined the identity of these signalers. We found that a swarm’s nest-site scouts search for potential nest cavities prior to the departure of the swarm from its hive. Furthermore, we found that the predeparture nest-site scouts are the sole producers of the worker-piping signal and that they are the first producers of the buzz-run signal. The control of the departure of a honeybee swarm from its hive shows how a small minority of well-informed individuals in a large social insect colony can make important decisions about when a colony should take action.     

Determining association networks in social animals: choosing spatial–temporal criteria and sampling rates

Social Network Analysis has become an important methodological tool for advancing our understanding of human and animal group behaviour. However, researchers tend to rely on arbitrary distance and time measures when defining ‘contacts’ or ‘associations’ between individuals based on preliminary observation. Otherwise, criteria are chosen on the basis of the communication range of sensor devices (e.g. bluetooth communication ranges) or the sampling frequencies of collection devices (e.g. Global Positioning System devices). Thus, researchers lack an established protocol for determining both relevant association distances and minimum sampling rates required to accurately represent the network structure under investigation. In this paper, we demonstrate how researchers can use experimental and statistical methods to establish spatial and temporal association patterns and thus correctly characterise social networks in both time and space. To do this, we first perform a mixing experiment with Merino sheep (Ovis aries) and use a community detection algorithm that allows us to identify the spatial and temporal distance at which we can best identify clusters of previously familiar sheep. This turns out to be within 2–3 m of each other for at least 3 min. We then calculate the network graph entropy rate—a measure of ease of spreading of information (e.g. a disease) in a network—to determine the minimum sampling rate required to capture the variability observed in our sheep networks during distinct activity phases. Our results indicate the need for sampling intervals of less than a minute apart. The tools that we employ are versatile and could be applied to a wide range of species and social network datasets, thus allowing an increase in both the accuracy and efficiency of data collection when exploring spatial association patterns in gregarious species.     

Consistent personality differences in house-hunting behavior but not decision speed in swarms of honey bees (<Emphasis Type="BoldItalic">Apis mellifera</Emphasis>)

Speed-accuracy tradeoffs are a common feature of decision-making processes, both in individual animals and in groups of animals working together to reach a single collective decision. Individual organisms display consistent differences in their “impulsivity,” and vary in their tendency to make rapid, impulsive choices as opposed to slower, more accurate decisions. However, we do not yet know whether groups of animals consistently differ in their tendency to prioritize decision speed over accuracy. We challenged 17 swarms of honey bees (Apis mellifera) to simultaneously choose a new nest site in each of three locations, and measured their decision speeds in each trial. We found that swarms displayed consistent personality differences in the number of waggle dances and shaking signals they performed and in how actively they scouted for new nest sites. However, swarms did not consistently differ in how long they took to choose a nest site. We suggest that house-hunting A. mellifera swarms may place an especially high emphasis on decision accuracy when choosing a nest site, and that chance events—such as the time when each swarm discovers a sufficiently high-quality nest site—may consequently play a greater role in determining a swarm’s decision speed than intrinsic characteristics such as a swarm’s “impulsivity.”     

Differential ingestion of zooplankton by four species of bivalves (Mollusca) in the Mali Ston Bay,Croatia

This study provides information about differences in composition of ingested zooplankton amongst bivalve species coexisting in the same area in a period from May 2009 to December 2010. The study was conducted at the Mali Ston Bay (42°51′ N, 17°40′ E)—the most important bivalve aquaculture area in the eastern Adriatic Sea. Stomach content analysis was performed on cultured species—Ostrea edulis and Mytilus galloprovincialis, and commercially important bivalve species from their natural environment—Modiolus barbatus and Arca noae. Results confirmed carnivory in bivalves, both from natural and cultured populations, but cultured species had higher numbers of zooplankters than those living on the seabed. The most abundant taxa were bivalve larvae, followed by tintinnids, copepods, unidentified eggs and gastropod larvae. Recorded numbers of bivalve larvae in M. galloprovincialis stomach were the highest so far reported and show that mussels impact the availability of natural spat.     

Effect of the presence of the shore crab, <Emphasis Type="Italic">Carcinus maenas</Emphasis>, on burrowing behaviour and clearance rate of the common cockle, <Emphasis Type="Italic">Cerastoderma edule</Emphasis>

Bivalves demonstrate various morphological and behavioural adaptations to reduce the risk of being attacked by predators. This paper examines how the presence of the crab Carcinus maenas (L.), a natural predator of the cockle Cerastoderma edule (L.), affects its burrowing depth and clearance or feeding rate. Cockles were placed in experimental tanks and treated with three levels of predatory disturbance: (1) unfed crab loose inside the tank, (2) unfed crab inside a cage suspended in the water column and (3) no crab present. Cockles’ burrowing depth was measured in two sediment types: mud and sand. Cockles burrowed more deeply in treatments with no crabs. Burrowing depth in sand was significantly greater than in mud. Two factors may contribute to the reduction in burial depth of C. edule in the presence of C. maenas: the change in the vertical orientation of the cockle and the ‘cough response’. No significant difference was found in the cockles’ clearance rate among the different levels of predator threat.     

Reproductive competition and fecal testosterone in wild male giant pandas (<Emphasis Type="Italic">Ailuropoda melanoleuca</Emphasis>)

The giant panda (Ailuropoda melanoleuca) is notoriously difficult to study in the wild, but its interesting reproductive ecology makes the effort worthwhile. Perhaps more than most species, the panda is energy-limited, which alters the cost/benefit analysis of its reproductive ecology. Using global positioning system/very high frequency radiocollars to locate mating aggregations, we used behavioral observations and fecal testosterone assays to gain insight into male panda reproductive effort and strategies, and test theories relating to reproductive competition. Male pandas initially competed fiercely for access to females that were about to be fertile, but once male competitive status was determined, aggression rates declined. Contact aggression was only observed during the first 2 days of mating aggregations; thereafter, it was replaced with noncontact aggression and avoidance. Agonistic interactions were highly asymmetrical, with contest losers (subordinates) showing less aggression and more avoidance than contest winners (dominants), both before and after contest outcome was established. The competitively superior male displayed mate-guarding tactics and secured all observed copulations. Contrary to theoretical predictions, testosterone levels did not predict aggression levels or contest winners and also were not affected by winning or losing a contest. Body size appeared to be the primary determinant of contest outcome. We discuss our findings in light of theoretical predictions, such as those arising from the “challenge hypothesis,” in the context of the giant panda’s foraging and nutritional ecology.     

Coastal retreat and/or advance adjacent to defences in England and Wales

Retreat and advance of shoreline position occurs naturally, and also as a result of defences which are constructed to prevent erosion and flooding. Retreat more commonly manifests itself down-drift of defences due to a sediment deficit causing the coast to become ‘set-back’. Advance normally develops due to sediment accumulation up-drift of a barrier inhibiting longshore drift, resulting in the coast becoming ‘set-forward’. Many examples of set-backs and set-forwards are recorded, but their location, number and cause is not known on a national scale. Using the Futurecoast aerial photographs, approximately 200 localities were identified as set-back or set-forward in England and Wales, with half situated in the Eastern and South East regions of England. Half of the total set-backs or set-forwards were on cliffed coasts, and half on low-lying coasts. Without local knowledge it is difficult to distinguish between set-backs and set-forwards. Set-backs often indicate higher retreat rates, thus threatening cliff-top infrastructure which requires defence upgrade and extensions, as well as raising maintenance costs. Monitoring set-backs is important for shoreline management, because as retreat continues, set-backs evolve and artificial headlands form and grow. This is reinforced by the shift from hard defence policies towards softer engineering approaches, managed realignment and limited intervention.     

Mate choice games, context-dependent good genes, and genetic cycles in the side-blotched lizard, Uta stansburiana

According to mate choice models, a female should prefer males with traits that are reliable indicators of genetic quality which the sire can pass on to their progeny. However, good genes may depend on the social environment, and female choice for good genes should be context dependent. The side-blotched lizard, Uta stansburiana, exhibits genetically based throat colors (orange, blue, or yellow) that could be used as a sexually selected signal since they reliably predict the genetic quality of mates. The frequencies of male and female morphs cycle between years, and both male and female morphs have an advantage when rare; thus genetic quality will depend on morph frequency. A female should choose a sire that maximizes the reproductive success of both male and female progeny. We examine a game theoretical model that predicts female mate choice as a function of morph frequency and population density. The model predicts the following flexible mate choice rule: both female morphs should prefer rare males in ’boom years’ of the female cycle (e.g., ’rarest-of-N rule’), but prefer orange males in ’crash years’ of the female cycle (’orange-male rule’). Cues from the current social environment should be used by females to choose a mate that maximizes the future reproductive success of progeny, given the social environment of the next generation. We predict that the cue is the density of aggressive orange females. In the side-blotched lizard, cycling mate choice games and context-dependent mate choice are predicted to maintain genetic variation in the presence of choice for good genes. Received: 8 March 2000 / Revised: 26 August 2000 / Accepted: 4 September 2000     

Graphical spatial models: a new view on interpreting spatial pattern

Graphical models provide an important tool for facilitating communication between scientists, decision-makers, and statisticians—many complicated ecological processes can be described in terms of “box-and-arrow” conceptual diagrams (e.g., Shipley in Cause and correlation in biology: a user’s guide to path analysis, structural equations and causal inferences, Cambridge Universtiy Press, Cambridge, 2000; Clark and Gelfand TRENDS in Ecology and Evolution 21:375–380, 2006). In particular, problems in landscape ecology often involve modeling relationships among multiple physical and/or biological variables that may operate on differing spatial scales (e.g., Rossi et al. in Ecol Monographs 62:277–314, 1992; Legendre et al. in Ecography 25:601–615, 2002; Overmars et al. in Ecol Model 164:257–270, 2003; Brown and Spector in J Appl Ecol 45:1639–1648, 2008; Koniak and Noy-Meir in Ecol Model 220:1148–1158, 2008). These problems are inherently multivariate, though researchers commonly rely on univariate methods, such as spatial regression models, to address them. In this paper, we introduce a multivariate method—graphical spatial models—that extends path analysis to incorporate spatial autocorrelation in one or more variables in a directed graph. We show how both exogenous and endogenous ecological processes as defined by Legendre et al. (Ecography 25:601–615, 2002) and Lichstein et al. (Ecol Monographs 72:445–463, 2002) can be represented in a graph. Most importantly, we show how to translate graphs representing these ecological processes into statistically estimable models. We motivate our theoretical results using an example of stream health data from the Willamette Valley, Oregon. For these data we are interested in the spatial pattern within both riparian land use and an index of stream health, and whether there is an association between land use and stream health, after accounting for these spatial patterns. We use a graphical spatial model to address these ecological questions simultaneously. We find that the health of a stream decreases as the percent of developed land within a 120-m riparian buffer increases; interestingly, there is only evidence of spatial pattern within land use.     

Leaf-cutter ants with worn mandibles cut half as fast,spend twice the energy,and tend to carry instead of cut

The importance of mechanical wear in the behavioral ecology and energetics of small organisms is an open question. We investigated wear in leaf-cutter ants, Atta cephalotes, because their cutting technique can be imitated and the leaves are the main energy source for the colony. We found that a razor-sharp (50-nm radius) “V-blade” that cuts leaves between the first and second mandibular teeth was dulled (∼10-μm radius) and often nearly worn away on foragers. We found that the force required to cut standard leaves, using mandibles removed from foragers cutting in the wild, varied by a factor of 2.5 with tooth wear, defined as the difference between pupal and actual tooth length. We also found that wear significantly reduced the cutting rate. From the distribution of wear among the cutting foragers, we estimated that the wild colony would have spent 44% less of both energy and time making the observed cuts if the cutters’ mandibles had all been pristine. Finally, wear correlated with behavioral differences—foragers with the most worn 10% of mandibles almost exclusively carried rather than cut. This previously unreported form of task partitioning suggests that eusociality may extend useful lifespans by making it possible to switch tasks as skills decline. We developed a model, assuming that ants do work at a constant rate proportional to their mass, to predict the cutting rate from head width, tooth wear, and force to cut leaves with a scalpel (R = 0.62), and we used this estimate to argue that the partitioning of cutting and carrying was sub-optimal but better than random. Wear’s strong effect on performance may promote wear-avoiding behavior and wear-resistant mandible composition; it may affect leaf selection and worker lifespan and it raises the possibility that wear is a similarly important constraint for many other small organisms.     

Sex differences in the movement patterns of free-ranging chimpanzees (<Emphasis Type="Italic">Pan troglodytes schweinfurthii</Emphasis>): foraging and border checking

Most social primates live in cohesive groups, so travel paths inevitably reflect compromise: decision processes of individuals are obscured. The fission–fusion social organisation of the chimpanzee, however, allows an individual's movements to be investigated independently. We followed 15 chimpanzees (eight male and seven female) through the relatively flat forest of Budongo, Uganda, plotting the path of each individual over periods of 1–3 days. Chimpanzee movement was parsed into phases ending with halts of more than 20 min, during which individuals fed, rested or engaged in social activities. Males, lactating or pregnant females and sexually receptive females all travelled similar average distances between halts, at similar speeds and along similarly direct beeline paths. Compared to lactating or pregnant females, males did travel for a significantly longer time each day and halted more often, but the most striking sex differences appeared in the organisation of movement phases into a day's path. After a halt, males tended to continue in the same direction as before. Lactating or pregnant females showed no such strategy and often retraced the preceding phase, returning to previously visited food patches. We suggest that female chimpanzee movements approximate an optimal solution to feeding requirements, whereas the paths of males allow integration of foraging with territorial defence. The ‘continually moving forwards’ strategy of males enables them to monitor their territory boundaries—border checking—whilst foraging, generally avoiding the explicit boundary patrols observed at other chimpanzee study sites.