Abstract

In cooperatively breeding species, helpers and parents commonly face two decisions when they find a food item: first, whether to feed the item to a young group member or to eat it themselves; and second, which offspring to feed. Little is known about the factors that influence these decisions in cooperative mammals, though optimal foraging theory provides a basis for a range of predictions. In this article we describe pup feeding behavior by helpers and parents in a cooperative mongoose, the meerkat (Suricata suricatta). When meerkat pups begin accompanying the group, they beg food from older group members, who dig up dispersed prey items. As predicted, the probability of a prey item being fed to a pup shows a positive relationship with prey size and a negative relationship with pup distance. Meerkats apparently follow a “feed the nearest pup rule” and are more likely to feed the nearest pup if it is hungry. Hungrier pups beg more and follow older group members more closely. Across all age categories, females feed pups more frequently than males, both in terms of the relative frequency of feeds, and the proportion of prey biomass found by each individual that is fed to pups. Females also feed female pups significantly more than male pups, while males feed pups of both sexes equally. These sex biases in feeding contributions may result from female group members benefiting more than males from higher pup survival, and in particular higher female pup survival, because females are the philopatric sex.

In cooperatively breeding mammals and birds, individuals exhibit parentlike behavior toward offspring that are not genetically their own. This helping may include constructing nests or burrows, feeding, incubating, grooming, and guarding (see Solomon and French, 1997; Stacey and Koenig, 1990). Perhaps the most widespread helping behavior is providing food to young. Through increasing the provisioning rate of young, helpers may increase the number and improve the condition of offspring raised by the group, and this is likely to explain both why there is a positive correlation between helper number and recruitment in many cooperative species (e.g., Brown et al., 1982; Emlen and Wrege, 1989; Moehlman, 1979) and why cooperative mammals often have larger litters than noncooperative relatives (Moehlman and Hoffer, 1997).

Although offspring typically benefit from helping behavior, it is often not clear whether helpers derive direct or indirect benefits from their actions (see Emlen, 1991). Answers to two questions may help us distinguish between these possibilities. First, which helpers provide the most help? Second, do particular helpers favor particular offspring and, if so, which? These questions are most relevant when individuals experience different costs and benefits to helping, perhaps due to variation in their relatedness to the offspring, their capacity to help, or the likelihood that the offspring will provide some future benefit. Almost no field studies of cooperative mammals have quantified the contributions made by different helpers (but see Owens and Owens, 1984), and even studies of cooperative birds have typically measured only relative contributions to helping, running the risk that these may not accurately reflect different levels of effort if foraging abilities differ between helpers. Similarly, little is known about the factors influencing food allocation between offspring in cooperative mammals.

Here we describe the provisioning behavior of helpers and parents in a cooperative mammal, the meerkat (Suricata suricatta). Meerkats live in groups of 2-30 individuals in the drier regions of southern Africa (Skinner and Smithers, 1990). Multiple adult males and females are normally present in each group, with one member of each sex socially dominant, and these dominant individuals are the parents of more than 75% of pups born in the group (Clutton-Brock et al., 2000; Griffin, 1999). Meerkat groups typically contain males and females of varying ages who were born in the group and are related to the dominant female, and one or more adult immigrant males (who are normally brothers). The dominant male in each group is normally an immigrant, and incest has never been recorded (Griffin, 1999). Meerkat pups are born in a burrow and are initially entirely dependent on their lactating mother for nutrition (although occasionally other females allolactate; Clutton-Brock et al., 1999b). When the pups begin moving with the group, they are initially incapable of finding prey themselves and instead noisily beg food, which consists mainly of burrowing invertebrates, from older group members (Doolan and Macdonald, 1996). In this study, we considered the factors that affect pup feeding rates and compared the contributions of different sex and age categories of helper. Because we were able to record every food item captured by foraging meerkats, we were able to make comparisons based on both the proportion of captured prey biomass that individuals fed to pups and the relative frequency of pup feeding events. We also investigated how pup hunger levels and spacing behavior influence feeding rate and considered what factors determine which pups are fed by older group members.

METHODS

Study area

Data were collected at two study sites in the South African Kalahari, one in the Kalahari Gemsbok National Park (25°17′ S, 20°32′ E; the “Park”) and one on ranchland (26°59′ S, 21°50′ E) close to Van Zyl's Rus about 120 km to the southeast (the “ Ranch”). Both areas comprise a mixture of dry river-beds, river terraces, and sparsely vegetated sand dunes (Leistner and Werger, 1973; Rooyen et al., 1991). Annual rainfall at both sites averages around 250 mm (Clutton-Brock et al., 1999a), most of which falls in the hot, wet season between October and April. Temperatures can exceed 40°C during the hottest months, while minimum temperatures may drop below freezing in the cold, dry season (May—September).

Data collection

We collected data during more than 7000 h of observation of seven Park and nine Ranch meerkat groups between 1995 and 1999. All meerkats in the study groups could be recognized individually and were habituated to observation within a few meters. We sexed individuals soon after they emerged from their natal burrow and classified them as pups from 0 to 3 months, juveniles from 3 to 6 months, subadults from 6 to 12 months, and adults at 12 months and older.

During group foraging sessions, we attempted to record all pup feeds, locating pups by the begging calls, which they emit almost continuously. The proportion of feeding events that we detected is likely to have been affected by various factors including the number of observers, size of the group, number of pups, group dispersion, and habitat type (e.g., height of grass). We therefore used these ad lib data to make comparisons within breeding events to determine relative feeding contributions by different group members and to establish which pups were fed most and what types of prey items they were fed. We were careful to ensure that all individuals were observed equally to avoid any systematic bias in our data. For each pup feeding event, we noted the identities of the donor and recipient, and the type (e.g., scorpion) and size of prey item.

We divided prey items into three size categories: small (items fitting entirely within the mouth of an adult), medium (items too large to fit inside the mouth, but where less than half the item was protruding), and large (items which mainly hung outside the mouth). Analysis of dropped or intercepted prey items revealed that small items had an average (wet) mass of 0.172 g, medium items 0.721 g, and large items 3.990 g. We used these average weights to estimate the biomass of prey items captured and fed to pups.

In addition to the ad lib records of pup feeds, we conducted continuous focal watches of foraging individuals during which we recorded the size and identity of every prey item captured and whether or not the prey was fed to a pup. If the prey was fed, we recorded the distance and identity of the recipient and whether or not it was the closest pup to the foraging individual. If the foraging individual consumed the prey instead of feeding it to a pup, we recorded its distance to the nearest pup. Each focal lasted 20 min, and we usually conducted at least 10 focals on each helper and parent during the first 6 weeks of pup feeding. When focal data are used in analyses, they are referred to as “proportion of items captured,” distinguishing them from ad lib data, which are termed “relative contributions.”

We estimated time budgets of pups and older group members by conducting instantaneous scans (Altmann, 1978) at 10-min intervals during foraging periods. When a begging pup was recorded within 2 m of a foraging adult, we recorded the identity of the adult (hereafter this behavior is termed “social foraging”). We calculated time budgets by taking the proportion of scans an individual spent in each activity each day and then taking the average across days.

Because our analysis of focal data revealed that most pup feeds occurred during the first 40 days after the pups began accompanying the group, our other analyses were restricted to this period unless otherwise stated. The end of this 40-day period is, on average, approximately 70 days after the pups were born, which is equal to the minimum interbirth interval for females. Restricting the analyses to this period therefore also removed any confounding effects of another litter being born into the group.

Pup feeding experiment

To investigate the effect of pup hunger levels on helper feeding and pup behavior, we provisioned pups in 12 litters. We randomly assigned half the pups in each litter to an experimental group and fed each of these pups with about 12 g of hard-boiled egg at the start of morning and evening activity periods. During each day of the experiment, we recorded the number of feeds given to each pup and compared the average number of prey items given to experimental and control pups. We also used scan data to calculate the proportion of time that each pup spent social foraging, and, in nine of the experiments, the proportion of time pups spent begging. Each experiment lasted between 2 and 5 days; this variation was due to pups in some litters disappearing through natural events.

Statistical analysis

Our analyses involve data collected from between 1 and 8 breeding events in each of 16 study groups. This represents a total of 97 breeding events (each of which was assigned a unique litter code), though some analyses used fewer litters (specified in the text) because all types of data were not available for every litter. We present analyses that are based on both within-litter and between-litter comparisons. The former typically involve the use of proportional data which were normalized using an arcsine-square root transformation. Where this transformation failed to normalize the data, we either fitted the data to a binomial distribution using generalized models or used nonparametric tests. The comparisons across breeding events usually involved data that were unbalanced and not orthogonal and so we used residual maximum likelihood (REML) models or, if the data could not be normalized, generalized linear mixed models (GLMM; Schall, 1991). REML and GLMM models are equivalent to general linear regression (with normal errors) and generalized linear models (with non-normal errors), respectively, except they allow both random and fixed effects to be defined. In our analyses it was necessary to define litter code as a random term because of repeated sampling (multiple helpers and multiple pups) within breeding events. In all analyses, we found no significant effect of group identity, and hence breeding events, rather than groups, were used as our independent points. We included all likely explanatory variables in the maximal regression models and dropped terms sequentially until the models only included terms whose elimination would significantly decrease the explanatory power of the model. The significance of terms was tested using the Wald statistic, which is distributed as chi-square. Statistical analyses were performed using Genstat 5.4 and Minitab 12. Multiple comparisons tests follow Zar (1999).

Pup age analysis

To investigate the effect of pup age on the probability of an item being fed, we pooled all focal data collected on each day from each litter. These data were analyzed using a GLMM fitting the total number of prey items fed on each day to a binomial distribution with the number of items found as the denominator.

Analysis of biomass of captured prey fed to pups

For each individual, we pooled all data from focal watches conducted during the first 40 days of pup feeding and calculated the proportion of the total biomass of captured prey that was fed to pups. These data were then arcsine-square root transformed and analyzed using a REML model.

Analysis of the effect of pup sex

To investigate whether individuals preferentially fed pups of one sex, we excluded single-sex litters and used the ad lib data to calculate, for each group member, the relative proportion of prey items fed to each pup each day and then took the average across days. Because these relative proportions are affected by the number of pups present, we obtained a standardized measure for each feeder by taking the average contribution to pups of each sex and then substracting the value for males from the value for females and dividing the result by the mean of these two values. This gives a standardized index of the degree of any sex bias, with negative values indicating that males are fed more, and positive values indicating that females are fed more. These data were then compared across litters using a REML model. We used a similar method of analysis to investigate whether pups preferentially associated with helpers of a particular sex.

RESULTS

The onset and duration of pup feeding behavior

Pups began accompanying the group an average (± SE) of 29.3 ± 0.43 days after they were born (range 17-42 days, n = 97 litters). For these 97 litters, there was no difference in the age at which pups began moving with the group at the Ranch or Park (Ranch: 29.0 ± 0.53 days, Park: 29.7 ± 0.72 days, t test: t = 0.57, n = 70, 27, p =.57). Pups in larger litters began accompanying the group later: for 76 breeding events in which only one female gave birth, there was a positive correlation between the size of the litter and the age at which the pups first began accompanying the group (r =.384, n = 76, p <.001). Pup feeding behavior effectively began on the first day that the pups moved with the group. Lactating mothers were not fed by other group members, and little food was taken to pups down their natal burrow: fewer than 30 out of more than 32,000 observed feeding events were to preemergent pups.

Most pup feeds occurred within the first 40 days after the pups began accompanying the group. A GLMM of focal data showed that feeders gave pups approximately 10% of all medium and 25% of all large prey items captured on the first day that the pups accompanied the group. This increased to approximately 20% and 45%, respectively, by the 20th day and decreased thereafter, until pup feeding had effectively ceased by the 60th day. On average, less than 5% of small items were fed, even at the peak feeding age (Figure 1, Tables 1 and 2).

Figure 1

Percentage of captured prey items fed to pups of different ages. The graph shows the predicted values from a generalized linear mixed model of the effect of pup age (days since the pups began accompanying the group) and prey size (large, medium, and small prey items; see also Table 1).

Figure 1

Percentage of captured prey items fed to pups of different ages. The graph shows the predicted values from a generalized linear mixed model of the effect of pup age (days since the pups began accompanying the group) and prey size (large, medium, and small prey items; see also Table 1).

Table 1

Generalized linear mixed model relating pup age to the proportion of prey items fed to pups: significance level of terms in the model

Model term   Wald statistic (χ2)   df  p 
Age   120.38   1   <.001  
Age2  335.70   1   <.001  
Prey size   3937.51   2   <.001  
Age × prey size   6.79   2   .03  
Model term   Wald statistic (χ2)   df  p 
Age   120.38   1   <.001  
Age2  335.70   1   <.001  
Prey size   3937.51   2   <.001  
Age × prey size   6.79   2   .03  

Table 2

Generalized linear mixed model relating pup age to the proportion of prey items fed to pups: direction of effects for significant terms

Minimal model   Average effect   SE  t  df  p 
For prey size, the average effect term represents the difference of the effect of each prey size from that of large items (whose level is set to zero).  
aFor factors with multiple levels the standard errors of the differences are shown, and the significance of the difference is calculated using a t test.  
Constant   -0.520   0.12     <.01  
Age   0.075   0.0068     <.001  
Age2  -0.0018   0.0001     <.001  
Prey size       
Large   0.00   L - M = 0.065a  19.5   46   <.001  
Medium   -1.27   L - S = 0.066a  53.8   46   <.001  
Small   -3.55   S - M = 0.046a  49.6   46   <.001  
Age × prey size       
Large   0.00   L - M = 0.0047a  2.34   46   <.05  
Medium   -0.011   L - S = 0.0048a  2.50   46   <.05  
Small   -0.012   S - M = 0.0037a  0.27   46   > .50  
Minimal model   Average effect   SE  t  df  p 
For prey size, the average effect term represents the difference of the effect of each prey size from that of large items (whose level is set to zero).  
aFor factors with multiple levels the standard errors of the differences are shown, and the significance of the difference is calculated using a t test.  
Constant   -0.520   0.12     <.01  
Age   0.075   0.0068     <.001  
Age2  -0.0018   0.0001     <.001  
Prey size       
Large   0.00   L - M = 0.065a  19.5   46   <.001  
Medium   -1.27   L - S = 0.066a  53.8   46   <.001  
Small   -3.55   S - M = 0.046a  49.6   46   <.001  
Age × prey size       
Large   0.00   L - M = 0.0047a  2.34   46   <.05  
Medium   -0.011   L - S = 0.0048a  2.50   46   <.05  
Small   -0.012   S - M = 0.0037a  0.27   46   > .50  

Factors affecting pup feeding decisions

Prey type and size

Of 21,578 identified prey items fed to pups, the majority were reptiles (mostly geckos and lizards), beetles (larvae, pupae, and adults) and scorpions (Table 3). Focal data showed that the proportion of captured prey items fed to pups increased with prey size (percentage of captured large items fed = 41.1 ± 3.3%, medium = 20.6 ± 1.7%, small = 4.2 ± 0.6%; two-way ANOVA [blocking within litters]: F2,96 = 147.0, p <.001; multiple comparisons tests revealed all sizes differ from each other). Consequently, although most prey items captured by foraging individuals were small (an average of 75.0 ± 1.5% of captured items were small, 20.1 ± 1.1% were medium, and 4.9 ± 0.6% were large), ad lib data revealed that only 31.5 ± 2.3% of prey items actually fed to pups were small, while 46.4 ± 1.6% were medium, and 22.1 ± 2.2% were large (n = 44 litters). Although pups' capacity to handle larger prey items presumably improves with age, there was no greater tendency to feed larger prey items to older pups (Figure 2, two-way ANOVA: F4,116 = 1.56, p =.19). This analysis used all ad lib records of pup feeding events to each litter, pooled within five age categories of pup (Figure 2).

Table 3

Prey items fed to meerkat pups within size categories

Prey type   % Pup feeds  
Large prey items (n = 5279)   
Reptiles (mainly geckos and lizards, some snakes)   52.24  
Beetle larvae   19.34  
Scorpions   13.81  
Solifugids   6.23  
Beetle adults   2.29  
Millipedes   2.10  
Others (including mice, crickets, locusts, caterpillars, each <2%)   3.98  
Medium prey items (n = 11166)   
Beetle larvae   49.77  
Scorpions   11.87  
Reptiles (geckos and lizards)   10.93  
Beetle adults   10.21  
Crickets and locusts   5.99  
Solifugids   4.99  
Others (including spiders, millipedes, caterpillars, tubers, each <2%)   6.25  
Small prey items (n = 5133)   
Beetle larvae   54.12  
Crickets and locusts   13.29  
Beetle adults   10.48  
Scorpions   6.29  
Beetle pupae   4.91  
Reptiles (geckos and lizards)   3.35  
Others (including spiders, solifugids, tubers, termites, ant larvae, each <2%)   7.56  
Prey type   % Pup feeds  
Large prey items (n = 5279)   
Reptiles (mainly geckos and lizards, some snakes)   52.24  
Beetle larvae   19.34  
Scorpions   13.81  
Solifugids   6.23  
Beetle adults   2.29  
Millipedes   2.10  
Others (including mice, crickets, locusts, caterpillars, each <2%)   3.98  
Medium prey items (n = 11166)   
Beetle larvae   49.77  
Scorpions   11.87  
Reptiles (geckos and lizards)   10.93  
Beetle adults   10.21  
Crickets and locusts   5.99  
Solifugids   4.99  
Others (including spiders, millipedes, caterpillars, tubers, each <2%)   6.25  
Small prey items (n = 5133)   
Beetle larvae   54.12  
Crickets and locusts   13.29  
Beetle adults   10.48  
Scorpions   6.29  
Beetle pupae   4.91  
Reptiles (geckos and lizards)   3.35  
Others (including spiders, solifugids, tubers, termites, ant larvae, each <2%)   7.56  

Figure 2

Percentage of large items in the prey fed to pups of different ages (pup age is defined as the number of days since the pup began accompanying the group). Means and SEs are shown.

Figure 2

Percentage of large items in the prey fed to pups of different ages (pup age is defined as the number of days since the pup began accompanying the group). Means and SEs are shown.

Pup distance

Captured prey items of all sizes were more likely to be fed if there was a pup nearby: the average distance that prey was carried to a pup was significantly shorter than the distance to the nearest pup when items were not fed (Figure 3; paired t tests: large items, t = 3.48, n = 24 litters, p <.001; medium items, t = 4.80, n = 26, p <.001; small items, t = 8.44, n = 25, p <.001). Medium and large prey items fed to pups were carried significantly farther than small items (Figure 3; two-way ANOVA: F2,46 = 4.64, p =.015; multiple comparison tests reveal no difference between large and medium items, but both were carried farther than small items).

Figure 3

Effect of pup distance and prey size on pup feeding decisions by feeders. For items fed to pups, we recorded the distance of the recipient from the feeder; for items not fed, we recorded the distance to the nearest pup. Means and SEs are shown.

Figure 3

Effect of pup distance and prey size on pup feeding decisions by feeders. For items fed to pups, we recorded the distance of the recipient from the feeder; for items not fed, we recorded the distance to the nearest pup. Means and SEs are shown.

Age and sex of helper

Across all age categories, females contributed a significantly greater proportion of the total number of prey items fed to pups than did males (ad lib data, Figure 4; Wilcoxon tests: juveniles z = 2.67, n = 15, p <.01; subadults z = 2.54, n = 24, p <.01, adults z = 3.26, n = 44, p <.001). Within each sex there was a significant effect of age category (Figure 4; Freidman tests: females, n = 11 litters, χ2 = 14.36, df = 2, p <.001; males, n = 11 litters, χ2 = 13.64, df = 2, p =.001). Multiple comparison tests revealed that in both sexes this result was due to juveniles feeding significantly less than other age categories, but there was no difference in the feeding rate of adults and subadults. Across the feeding period as a whole, there was no difference between the relative contributions of helpers and parents of the same sex (ad lib data, Wilcoxon tests: females z = 0.823, n = 31, p =.41; males z = 1.581, n = 23, p =.11). However, the contribution of mothers, relative to that of other group members, was significantly less to young pups than to older ones (Figure 5, Freidman test, n = 27 litters, χ2 = 13.64, df = 2, p <.001; multiple comparisons tests revealed that the contribution to pups at 1-10 days was significantly less than at all other ages).

Figure 4

Relative feeding contributions by different age/sex categories of feeder. Medians and interquartile ranges are shown.

Figure 4

Relative feeding contributions by different age/sex categories of feeder. Medians and interquartile ranges are shown.

Figure 5

Feeding contributions of mothers to different ages of pups, expressed as a percentage of the contribution of the whole group. Medians and interquartile ranges are shown.

Figure 5

Feeding contributions of mothers to different ages of pups, expressed as a percentage of the contribution of the whole group. Medians and interquartile ranges are shown.

The higher overall feeding rate by females may in part be attributed to differences in time budgets because adult females spent a slightly higher proportion of time foraging during the pup feeding period. This was particularly true among the parents, among which males spent substantially less time foraging than females (Tables 4 and 5). Other terms significant in this REML analysis were group size (individuals spent more time foraging in larger groups) and study site (individuals spent more time foraging on the Ranch than in the Park). Both of these results are likely to be due to a trade-off between foraging and vigilance because individuals are more vigilant both in smaller groups and in the Park, where the risk of predation is higher than on the Ranch (Clutton-Brock et al., 1999a).

Table 4

Residual maximum likelihood model of the percentage of time spent foraging by different age and sex categories: predicted values for individuals of each category and the significance of differences between the sexes

  Back-transformed means (%)   Transformed means      
Age category   Female   Male   Female   Male   SE  t  df  p 
Adult individuals are classed as one of two “age” categories: parents and nonparents. The analysis required the data to be normalized using the square-root arcsine transformation, and significance levels are based on transformed data. Both back-transformed and transformed mean estimates are shown; SE denotes the standard errors of the differences.  
Juveniles   85.12   85.34   67.31   67.49   1.43   0.12   39   > .5  
Subadults   78.23   77.50   62.19   61.69   1.23   0.41   39   > .5  
Nonparents   74.09   70.62   59.40   57.18   0.82   2.71   39   <.01  
Parents   78.75   65.00   62.55   53.73   1.49   5.91   39   <.001  
  Back-transformed means (%)   Transformed means      
Age category   Female   Male   Female   Male   SE  t  df  p 
Adult individuals are classed as one of two “age” categories: parents and nonparents. The analysis required the data to be normalized using the square-root arcsine transformation, and significance levels are based on transformed data. Both back-transformed and transformed mean estimates are shown; SE denotes the standard errors of the differences.  
Juveniles   85.12   85.34   67.31   67.49   1.43   0.12   39   > .5  
Subadults   78.23   77.50   62.19   61.69   1.23   0.41   39   > .5  
Nonparents   74.09   70.62   59.40   57.18   0.82   2.71   39   <.01  
Parents   78.75   65.00   62.55   53.73   1.49   5.91   39   <.001  

Table 5

Residual maximum likelihood model of the percentage of time spent foraging by different age and sex categories: significant terms in the final model

Model term   Wald statistic (χ2)   df  p 
For direction of significant effects, see text.  
Group size   43.82   1   <.001  
Study site   35.66   1   <.001  
Age category   63.55   3   <.001  
Sex   7.36   1   <.01  
Sex × age category   39.26   3   <.001  
Model term   Wald statistic (χ2)   df  p 
For direction of significant effects, see text.  
Group size   43.82   1   <.001  
Study site   35.66   1   <.001  
Age category   63.55   3   <.001  
Sex   7.36   1   <.01  
Sex × age category   39.26   3   <.001  

Overall, foraging individuals gave pups an average of 18.4 ± 1.4% of the total biomass they caught during the first 40 days of pup foraging. Females of all age categories fed pups a significantly higher proportion of the biomass of prey items that they captured than males (Tables 6 and 7). Parents fed pups a similar proportion of prey items as other adults of the same sex (when parentage is included as a term in the REML, it is insignificant: χ2 = 0.0, df = 1, p =.99). The proportion of prey biomass fed to pups was higher as group size decreased and as litter size increased (Tables 6 and 7). Adults and subadults also fed pups a higher proportion of prey biomass than juveniles of the same sex (Tables 6 and 7).

Table 6

Residual maximum likelihood model of the proportion of captured prey biomass fed to pups: significance level of terms in the model

Model term   Wald statistic (χ2)   df  p 
Sex   6.77   1   <.01  
Group size   35.76   1   <.001  
Litter size   7.16   1   <.01  
Age category   24.31   2   <.001  
Model term   Wald statistic (χ2)   df  p 
Sex   6.77   1   <.01  
Group size   35.76   1   <.001  
Litter size   7.16   1   <.01  
Age category   24.31   2   <.001  

Table 7

Residual maximum likelihood model of the proportion of captured prey biomass fed to pups: direction of the effects for significant terms

Minimal model   Average effect   SE  t  df  p 
For age category, the average effect term represents the difference of the effect of each age from that of Adults (whose level is set to zero).  
aFor age categories the standard errors of the differences are shown, and the significance of the difference is calculated using a t test.  
Constant   21.17   1.22     <.001  
Sex (male)   -2.58   0.99     <.01  
Group size   -1.14   0.19     <.001  
Litter size   1.71   0.64     <.01  
Age category       
Adult   0.00   A - J = 1.50a  4.71   40   <.001  
Juvenile   -7.06   A - S = 1.19a  0.11   40   > .05  
Subadult   0.13   J - S = 1.64a  4.38   40   <.001  
Minimal model   Average effect   SE  t  df  p 
For age category, the average effect term represents the difference of the effect of each age from that of Adults (whose level is set to zero).  
aFor age categories the standard errors of the differences are shown, and the significance of the difference is calculated using a t test.  
Constant   21.17   1.22     <.001  
Sex (male)   -2.58   0.99     <.01  
Group size   -1.14   0.19     <.001  
Litter size   1.71   0.64     <.01  
Age category       
Adult   0.00   A - J = 1.50a  4.71   40   <.001  
Juvenile   -7.06   A - S = 1.19a  0.11   40   > .05  
Subadult   0.13   J - S = 1.64a  4.38   40   <.001  

Factors affecting which pup is fed

Proximity to pup

Helpers almost invariably fed the closest begging pup: of 1072 feeding events by 112 individuals from 9 different groups, 1033 (96.4%) were to the nearest pup. When the pups began accompanying the groups during foraging periods, they initially spent approximately 40% of their time begging within 2 m of a foraging group member (“social foraging”). This increased to about 55% 17 days after they began accompanying the group and declined thereafter (Figure 6).

Figure 6

Percentage of time pups spend social foraging at different ages. The graph shows predicted values from a REML analysis in which pup age and age2 were fitted to the percentage of time pups spent socially foraging (data pooled within each litter for each day). Litter code was fitted as a random term.

Figure 6

Percentage of time pups spend social foraging at different ages. The graph shows predicted values from a REML analysis in which pup age and age2 were fitted to the percentage of time pups spent socially foraging (data pooled within each litter for each day). Litter code was fitted as a random term.

Social foraging data for each pup were pooled for the first 40 days after the pups began following the group and fitted to a REML model. This analysis showed that pups in larger groups spent a higher proportion of time socially foraging than pups in small groups. In addition, they socially foraged more if the sex ratio of helpers was female biased (Tables 8 and 9). Female pups spent significantly more time socially foraging than male pups, although the difference was small (< 2% of their time budgets; Tables 8 and 9). Pups socially foraged more with older individuals than with younger individuals, and more with adult females than with adult males (Figure 7; Friedman test, n = 8 litters, χ2 = 14.5, df = 5, p =.01; multiple comparisons tests revealed a significant difference between adult females and all other categories apart from sub-adult males, and a significant difference between adult males and subadults of both sexes and juveniles of either sex). There was no significant difference in the proportion of time that pups spent social foraging with parents and other adults of the same sex (Wilcoxon signed-ranks tests: females, T = 147, n = 25, p =.69; males, T = 110, n = 20 p =.87).

Table 8

Residual maximum likelihood model of the proportion of time spent social foraging by pups: significance level of terms in the model

Model term   Wald statistic (χ2)   df  p 
Pup sex   5.38   1   .02  
Group size   7.54   1   <.01  
Helper sex ratio   6.99   1   <.01  
Litter size   0.66   1   .42  
Study site   1.17   1   .28  
Model term   Wald statistic (χ2)   df  p 
Pup sex   5.38   1   .02  
Group size   7.54   1   <.01  
Helper sex ratio   6.99   1   <.01  
Litter size   0.66   1   .42  
Study site   1.17   1   .28  

Table 9

Residual maximum likelihood model of the proportion of time spent social foraging by pups: direction of effects for significant terms

Minimal model   Average effect   SE  p 
A positive effect of helper sex ratio indicates that the pups spend more time social foraging when the helper ratio is female biased.  
Constant   28.21   1.20   <.001  
Pup sex (male)   -1.50   0.65   <.05  
Group size   0.64   0.23   <.01  
Helper sex ratio   21.69   8.21   <.01  
Minimal model   Average effect   SE  p 
A positive effect of helper sex ratio indicates that the pups spend more time social foraging when the helper ratio is female biased.  
Constant   28.21   1.20   <.001  
Pup sex (male)   -1.50   0.65   <.05  
Group size   0.64   0.23   <.01  
Helper sex ratio   21.69   8.21   <.01  

Figure 7

Percentage of time pups socially forage with individuals of different age/sex categories. Medians and interquartile ranges are shown.

Figure 7

Percentage of time pups socially forage with individuals of different age/sex categories. Medians and interquartile ranges are shown.

Pup feeding experiment

Pups that were provided with egg at the start of foraging sessions spent less time begging and less time socially foraging than control pups (median percentage of time spent begging by fed pups = 51.8%, interquartile range [IQR] = 47.4-59.0%; control pups = 64.3%, IQR = 51.0-69.1%; Wilcoxon signed-ranks test: T = 0, n = 9, p <.01; median percentage of time spent social foraging by fed pups = 45.8%, IQR = 36.3-56.6%; control pups = 55.4%, IQR = 38.9-66.8%; Wilcoxon signed-ranks test: T = 10, n = 12, p <.05). Probably as a result of this, they were fed fewer prey items by older group members than were control pups (Figure 8; Wilcoxon signed-ranks test: T = 1, n = 12, p <.005).

Figure 8

Percentage of prey items given to experimentally fed and control pups. Medians and interquartile ranges are shown.

Figure 8

Percentage of prey items given to experimentally fed and control pups. Medians and interquartile ranges are shown.

Pup sex

Female group members fed female pups significantly more prey items than they fed to male pups, whereas male group members fed pups of both sexes equally (Tables 10 and 11). REML analyses showed that this was not due to female pups associating more strongly with females, as male and female pups spent the same proportion of their social foraging time with female group members. This was the case when considering all group members, and when the analysis was restricted to adults (REML analyses with litter code fitted as random term: all group members, Wald statistic for pup sex, χ2 = 0.11, df = 1, p =.74; adult group members, Wald statistic for pup sex, χ2 = 0.06, df = 1, p =.81).

Table 10

Residual maximum likelihood model of the relative proportion of prey items fed to female pups: significance level of terms in the model

Model term   Wald statistic (χ2)   df  p 
Helper age was defined as one of three categories (see text) and parentage was fitted to the model to test whether parents showed a different bias in food allocation to nonparents.  
Helper sex   4.40   1   <.05  
Helper age   2.63   2   .27  
Parentage   1.38   1   .24  
Group size   0.08   1   .77  
Model term   Wald statistic (χ2)   df  p 
Helper age was defined as one of three categories (see text) and parentage was fitted to the model to test whether parents showed a different bias in food allocation to nonparents.  
Helper sex   4.40   1   <.05  
Helper age   2.63   2   .27  
Parentage   1.38   1   .24  
Group size   0.08   1   .77  

Table 11

Residual maximum likelihood model of the relative proportion of prey items fed to female pups: direction of effects for significant terms

Minimal model   Average effect   SE  p 
The constant does not differ significantly from zero, indicating that males do not feed either sex of pup more (since the only other factor in the minimal model was helper sex). No interaction terms were significant.  
Constant   -0.034   0.056   > .5  
Helper sex (female)   0.126   0.060   > .05  
Minimal model   Average effect   SE  p 
The constant does not differ significantly from zero, indicating that males do not feed either sex of pup more (since the only other factor in the minimal model was helper sex). No interaction terms were significant.  
Constant   -0.034   0.056   > .5  
Helper sex (female)   0.126   0.060   > .05  

DISCUSSION

In meerkats, helpers begin provisioning pups when they leave their natal burrow and start begging noisily for food. Unlike helpers in some other social carnivores (e.g., black-backed jackals: Moehlman, 1979; wild dogs: Malcolm and Marten, 1982), meerkats do not feed lactating females, nor do they feed the young at the burrow as do dwarf mongooses (Rood, 1978), black-backed jackals (Moehlman, 1979), and Ethiopian wolves (Sillero-Zubiri, 1994). This is probably because meerkats feed on small, dispersed food items, and repeated trips back to the burrow would be both uneconomical and dangerous (Clutton-Brock et al., 1999a). As a result, pup feeding in meerkats is perhaps more analogous to fledgling feeding in birds than it is to offspring feeding in many other social carnivores.

Feeding rules in meerkats

When foraging individuals capture a prey item, they are faced with the choice of feeding it to a pup or eating it themselves. The probability of captured prey being fed to a pup increases as the size of the item increases, and all sizes of prey items are more likely to be fed if there is a pup nearby. In addition, large and medium items are carried farther than small items. These observations are consistent with the marginal value theorem, which predicts that it only pays individuals to work hard (in this case deciding to feed or traveling farther) if the return (in terms of biomass of food given to a pup) is high (Krebs and Kacelnik, 1991).

When individuals do feed a pup, it is almost invariably the nearest, and a similar tendency to feed the closest chick has been observed in a number of studies of nestling birds (e.g., Kacelnik et al., 1995; McRae et al., 1993). From a parent's perspective, this strategy may lead to optimal food allocation among the brood if the chicks are free to compete for position in the nest, and they compete more strongly when they are hungry and/or larger (Kilner and Johnstone, 1997; Mock and Parker, 1997). In contrast, because meerkat pups are dispersed, the nearest will often not be the hungriest, so following a strategy of always feeding the closest pup would appear at odds with status signaling explanations of resource allocation among offspring (Godfray, 1995). However, despite their overwhelming tendency to feed the nearest pup, meerkat group members are nonetheless responsive to pups' needs. Manser (1998) showed that meerkats increase their feeding rates in response to playbacks of pup begging calls and preferentially approach louder playbacks (which simulated closer pups) with food items. Thus, hungrier pups would be expected to increase their probability of being fed by begging more and by following group members more closely. This is precisely what control pups did relative to pups fed with egg in the provisioning experiment described here, and control pups were indeed fed more frequently.

Despite following a feed the nearest pup rule, helpers and parents still preferentially feed hungry pups because the decision that they make is not which pup to feed but whether to feed at all. This decision appears to be a graded response dependent on the strength or type of the begging signal from the nearest pup (it may also be influenced by the sex of the pup, as discussed below). Because meerkat groups are dispersed while foraging, this may be a more efficient strategy than attempting to compare hunger levels among pups. The extent to which this will prove a general feeding rule in other situations where offspring are dispersed remains to be seen.

What determines who helps?

The factors that determine which sex helps most in cooperative species have been the subject of considerable debate (see Cockburn, 1998). In many cooperative bird species it is primarily or exclusively only one sex that helps, typically males (e.g., long-tailed tits: Hatchwell and Russell, 1996; superb fairy wrens: Mulder et al., 1994; pied kingfishers: Reyer, 1990), although there are some species where only females help (e.g., Seychelles warblers: Komdeur, 1992). In his comprehensive review, Cockburn (1998) suggested that most cases of helping by one sex may ultimately be explained as an epiphenomenon of sex-biased dispersal to avoid inbreeding. However, some cooperative bird species (e.g., splendid fairy wrens: Russell and Rowley, 1993; Florida scrub jays: Woolfenden and Fitzpatrick, 1990) and most cooperative mammals (e.g., banded mongooses: Rood, 1975; dwarf mongooses: Rood, 1978; brown hyenas: Owens and Owens, 1984; hunting dogs: Malcolm and Marten, 1982) have helpers of both sexes. Quantifying the relative contributions by the sexes in these species may help us understand the function of helping behavior, yet few studies have done so (but see Owens and Owens, 1984).

Female meerkats fed at higher rates than males in all age categories, and this was true both in terms of the total number of items delivered and the proportion of captured prey biomass that they gave away. In doing so, females also spent a higher proportion of their time foraging, and this may have exposed them to higher predation risk (Clutton-Brock et al., 1999c). Within groups, juveniles fed less than older individuals, and this may be because giving away prey items is more costly to juveniles, who face the extra energetic demands of growth while also being less successful foragers than older individuals (Barnard JA, unpublished data). Pups also spent more time socially foraging with older individuals, especially adult females, who fed pups more frequently than adult males. However, these association patterns may result from these individuals taking more food to the pups, rather than any tendency for pups to associate with the best feeders per se.

To what extent do these results help us determine why helpers actually help (Emlen, 1991)? A number of hypotheses have been proposed to account for the evolution of helping behavior, including (1) indirect fitness benefits via kin selection derived from feeding the offspring of close relatives (Hamilton, 1964); (2) unselected behavior arising as a by-product of older individuals' innate response to begging stimuli (Jamieson, 1989); (3) a means of gaining parenting experience (Skutch, 1961); (4) “ payment of rent” in order to remain in the group (Gaston, 1978); and (5) future benefits of delayed reciprocity or group augmentation (Brown, 1987; Ligon and Ligon, 1978; Woolfenden and Fitzpatrick, 1978). In meerkats, some forms of help are performed exclusively by nonbreeders, helping is often costly, and helpers contribute at least as much as parents to feeding young and other cooperative behaviors; therefore, helping is unlikely to be an unselected byproduct of parental behavior or a means of gaining experience (Clutton-Brock et al., 1998). A recent formulation of the payment of rent hypothesis shows that it is only likely to explain the evolution of helping where there is low relatedness within groups and where the benefits of group living are low, at least for dominant individuals (Kokko H and Johnstone RA, personal communication). Neither of these situations apply in meerkats (Clutton-Brock et al., 1999b, 2000; Griffin, 1999). This leaves two remaining hypotheses that may account for helping in meerkats: kin selection and group augmentation.

Meerkat groups typically contain highly related individuals; the average relatedness between helpers and litters born into the group approximates that of half-sibs (Clutton-Brock et al., 2000). On average, helping behavior will therefore confer inclusive fitness benefits to the helper. However, unrelated individuals help as much as relatives, and close relatives help no more than less-related individuals, suggesting that differences in relatedness do not account for the variation in helping behavior observed within groups (Clutton-Brock et al., 2000). Nor does it explain why brothers help less than their sisters, as observed in this study.

Group augmentation, or delayed reciprocity, would predict that individuals that are most likely to associate with the pups in the future should help most. This is consistent with the observed sex differences in helping behavior because males are the dispersing sex (males disperse at 18-24 months), and females are philopatric. By helping pups to survive, female helpers therefore potentially have more to gain, both in terms of the antipredation benefits of large group size (Clutton-Brock et al., 1999a) and by increasing the number of potential future helpers should they attain a breeding position. Owens and Owens (1984) also found that female helpers fed pups more than males in a clan of brown hyenas, where, again, females are the philopatric sex. Similarly, young, female green woodhoopoes feed at a higher rate than males, and this may be because females are more likely to breed in their natal territories and subsequently benefit from delayed reciprocity if the nestlings become helpers (Ligon and Ligon, 1990a, b).

The tendency of female helpers to preferentially feed female pups, and for males to show no preference, may also be best explained in terms of future benefits. With the exception of older females who die before the male pups in the current litter disperse, females are likely to spend more of their lives associating with females from the current litter than with males. In contrast, male helpers will typically spend an equal proportion of their remaining time in the group with all pups in the current litter, and so would be expected to help both sexes equally, as observed. This study thus provides further evidence that fitness benefits accruing from group augmentation may play an important role in determining the levels of helping behavior in cooperative species.

We are grateful to the National Parks Board of the Republic of South Africa for permission to work in the Kalahari Gemsbok National Park, and to Mr. and Mrs. H. Kotze for permission to work on their land at Van Zyl's Rus. In particular, we thank E. le Riche and D. Engelbrecht, Wardens of the Park, P. Novellie and A. Hall-Martin and the National Park staff based at Nossob, including D. Ras, J Herrholdt, G. de Kock, and S. de Waal. The study would not have been possible without the support of members of the Mammal Research Institute, University of Pretoria, including J. Skinner, J. du Toit, P. Richardson, A. MacKenzie, M. Haupt, and G. van Dyk, and of A. Griffin, T. Marshall, and J. Pemberton at the Institute of Cell, Animal and Population Biology (Edinburgh). We are grateful to the following for their assistance in data collection: D. Allsop, G. Avey, Z. Balmforth, C. Britten, J. and P. Chadwick, S. Clarke, A. Crichton, S. Davies, P. Dixon, P. Elsmere, J. Garner, N. Green, S. Hodge, J. Kewido, J. Kinns, C. MacCallum, A. MacColl, C. Macleod, T. Maddox, M. Manser, A. and G. Marais, K. McKay, S. Mercenaro, H. Nicholls, I. Olyn, M. Peterson, L. Postgate, M. Shaw, S. Slater-Jones, R. Smith, R. Tait, G. Telford, B. Themen, A. Toole, A. Turner, S. White, and R. Yarnell. For advice and assistance, we thank S. Balshine-Earn, J. Barnard, T. Coulson, R. Johnstone, R. Kilner, L. Kruuk, D. Macdonald, J. Nel, P. Roth, A. Russell, I. Stevenson, and R. Woodroffe. We are also grateful to J.D. Ligon and an anonymous referee for their comments on the manuscript. Our research is funded by grants from Natural Environment Research Council (NERC) and Biotechnology and Biological Sciences Research Council (BBSRC).

References

Altmann J,
1978
. Observational study of behaviour: sampling methods.
Behaviour
 
49
:
227
-267.
Brown JL,
1987
.
Helping and communal breeding in birds
 . Princeton, New Jersey: Princeton University Press.
Brown JL, Brown ER, Brown SD, Dow DD,
1982
. Helpers: effects of experimental removal on reproductive success.
Science
 
215
:
421
-422.
Clutton-Brock TH, Brotherton PNM, O'Riain MJ, Griffin AS, Gaynor D, Sharpe L, Kansky R, Manser M, McIlrath GM,
2000
. Individual contributions to babysitting in a cooperative mongoose, Suricata suricatta.
Proc R Soc Lond B
 
267
:
301
-305.
Clutton-Brock TH, Gaynor D, Kansky R, MacColl ADC, McIlrath G, Chadwick P, Brotherton PNM, O'Riain MJ, Manser M, Skinner JD,
1998
. Costs of helping behaviour in suricates (Suricata suricatta).
Proc R Soc Lond B
 
265
:
185
-190.
Clutton-Brock TH, Gaynor DG, MacIrath GM, MacColl ADC, Kansky R, Chadwick P, Manser M, Skinner JD, Brotherton PNM,
1999
. Predation, group size and mortality in a cooperative mongoose, Suricata suricatta.
J Anim Ecol
 
68
:
672
-683.
Clutton-Brock TH, MacColl A, Chadwick P, Gaynor D, Kansky, Skinner JD,
1999
. Reproduction and survival of suricates (Suricata suricatta) in the southern Kalahari.
Afr J Ecol
 
37
:
69
-80.
Clutton-Brock TH, O'Riain MJ, Brotherton PNM, Gaynor D, Kansky R, Griffin AS, Manser M,
1999
. Selfish sentinels in cooperative mammals.
Science
 
284
:
1640
-1644.
Cockburn A,
1998
. Evolution of helping behavior in cooperatively breeding birds.
Annu Rev Ecol Syst
 
29
:
141
-177.
Doolan SP, Macdonald DW,
1996
. Diet and foraging behaviour of group-living meerkats, Suricata suricatta, in the southern Kalahari.
J Zool
 
239
:
697
-716.
Emlen SJ,
1991
. Evolution of cooperative breeding in mammals and birds. In:
Behavioural Ecology: an evolutionary approach
 , 3rd ed. (Krebs JR, Davies NB, eds). Oxford: Blackwell Scientific;
301
-337.
Emlen ST, Wrege PH,
1989
. A test of alternative hypotheses for helping behavior in white-fronted bee-eaters of Kenya.
Behav Ecol Sociobiol
 
25
:
303
-320.
Gaston AJ,
1978
. The evolution of group territorial behavior and cooperative breeding.
Am Nat
 
112
:
1091
-1110.
Godfray HCJ,
1995
. Signalling of need between parents and young: parent-offspring conflict and sibling rivalry.
Am Nat
 
146
:
1
-24.
Griffin AS,
1999
.
A genetic analysis of cooperative breeding in meerkats
  (PhD dissertation). Edinburgh: University of Edinburgh.
Hamilton WD,
1964
. The genetic evolution of social behaviour.
J Theor Biol
 
7
:
1
-52.
Hatchwell BJ, Russell AF,
1996
. Provisioning rules in cooperatively breeding long-tailed tits Aegithalos caudatus: an experimental study.
Proc R Soc Lond B
 
263
:
83
-88.
Jamieson IG,
1989
. Behavioral heterochrony and the evolution of birds' helping at the nest: an unselected consequence of communal breeding?
Am Nat
 
133
:
394
-406.
Kacelnik A, Cotton PA, Stirling L, Wright J,
1995
. Food allocation among nestlings: sibling competition and the scope of parental choice.
Proc R Soc Lond B
 
259
:
259
-263.
Kilner RM, Johnstone RA,
1997
. Begging the question: are offspring solicitation behaviours signals of need?
Trends Ecol Evol
 
12
:
11
-15.
Komdeur J,
1992
. Importance of habitat saturation and territory quality for evolution of cooperative breeding in the Seychelles warbler.
Nature
 
358
:
493
-495.
Krebs JR, Kacelnik A,
1991
. Decision-making. In
Behavioural ecology: an evolutionary approach
  , 3rd ed. (Krebs JR, Davies NB, eds). Oxford: Blackwell Scientific;
105
-136.
Leistner OA, Werger MJA,
1973
. Southern Kalahari phytosociology.
Vegetation
 
28
:
353
-399.
Ligon JD, Ligon SH,
1978
. Communal breeding in the green woodhoopoe as a case of reciprocity.
Nature
 
276
:
496
-498.
Ligon JD, Ligon SH,
1990
. Female-biased sex ratio at hatching in the green woodhoopoe.
Auk
 
107
:
765
-771.
Ligon JD, Ligon SH,
1990
. Green woodhoopoes: life history traits and sociality. In:
Cooperative breeding in birds: long-term studies of ecology and behavior
  (Stacey PB, Koenig WD, eds). Cambridge: Cambridge University Press;
33
-65.
Malcolm JR, Marten K,
1982
. Natural selection and the communal rearing of pups in African wild dogs (Lycoan pictus).
Behav Ecol Sociobiol
 
10
:
1
-13.
Manser MB,
1998
.
The evolution of auditory communication in suricates, Suricata suricatta
  (PhD dissertation). Cambridge: Cambridge University.
McRae SB, Weatherhead PJ, Montgomerie R,
1993
. American robin nestlings compete by jockeying for position.
Behav Ecol Sociobiol
 
33
:
101
-106.
Mock DW, Parker GA,
1997
.
The evolution of sibling rivalry
 . Oxford: Oxford University Press.
Moehlman PD,
1979
. Jackal helpers and pup survival.
Nature
 
277
:
382
-383.
Moehlman PD, Hofer H,
1997
. Cooperative breeding, reproductive suppression, and body mass in canids. In:
Cooperative breeding in mammals
  (Solomon NG, French JA, eds). Cambridge: Cambridge University Press;
76
-128.
Mulder RA, Dunn PO, Cockburn A, Lazenby-Cohen KA, Howell MJ,
1994
. Helpers liberate female fairy wrens from constraints on extrapair mate choice.
Proc R Soc Lond B
 
45
:
830
-833.
Owens DD, Owens MJ,
1984
. Helping behaviour in brown hyenas.
Nature
 
308
:
843
-845.
Reyer H-U,
1990
. Pied kingfishers: ecological causes and reproductive consequences of cooperative breeding. In:
Cooperative breeding in birds: long-term studies of ecology and behavior
  (Stacey PB, Koenig WD, eds). Cambridge: Cambridge University Press;
527
-557.
Rood JP,
1975
. Population dynamics and food habits of the banded mongoose.
East Afr Wildl J
 
13
:
89
-111.
Rood JP,
1978
. Dwarf mongoose helpers at the den.
Z Tierpsychol
 
48
:
277
-287.
Rooyen N van, Bredenkamp GJ, Theron GK,
1991
. Kalahari vegetation: veld condition trends and ecological status if species.
Koedoe
 
34
:
61
-72.
Russell EM, Rowley I,
1993
. Philopatry or dispersal: competition for territory vacancies in the splendid fairy wren, Malurus splendens.
Anim Behav
 
45
:
519
-539.
Schall R,
1991
. Estimation in generalized linear models with random effects.
Biometrika
 
78
:
719
-727.
Sillero-Zubiri C,
1994
.
Behavioural ecology of the Ethiopian wolf (Canis simensis)
  (PhD dissertation). Oxford: University of Oxford.
Skinner JD, Smithers RHN,
1990
.
The mammals of the Southern African region
 , 2nd ed. Pretoria: Pretoria University Press.
Skutch AF,
1961
. Helpers among birds.
Condor
 
63
:
198
-226.
Solomon NG, French JA,
1997
.
Cooperative breeding in mammals
 . Cambridge: Cambridge University Press.
Stacey PB, Koenig WD,
1990
.
Cooperative breeding in birds: long-term studies of ecology and behavior
 . Cambridge: Cambridge University Press.
Woolfenden GE, Fitzpatrick JW,
1978
. The inheritance of territory in group-breeding birds.
Bioscience
 
28
:
104
-108.
Woolfenden GE, Fitzpatrick JW,
1990
. Florida scrub jays: a synopsis after 18 years of study. In:
Cooperative breeding in birds: long-term studies of ecology and behavior
  (Stacey PB, Koenig WD, eds). Cambridge: Cambridge University Press;
239
-266.
Zar JH,
1999
.
Biostatistical analysis
 , 4th ed. Englewood Cliffs, New Jersey: Prentice-Hall.