Analyses use individual means. To investigate the effects of age on rates of provisioning, food-offering calls, and independent prey capture, I pooled data for each focal individual in 4 age categories: 09–20, 21–32, 33–44, and 45–56 weeks. The data regarding prey transfer have been presented in a previous publication in a slightly different manner (i.e., averaged over 3 age-blocks instead of 4: Rapaport 2006) and are reanalyzed here in order to elucidate developmental changes in provisioned versus independent prey acquisition. I calculated rates (i.e., food transfers received, food-offering vocalizations given to juveniles during interactions in which adults had prey in hand, and independent prey capture) by estimating time spent visible from the simultaneously collected focal instantaneous and continuous samples. Rates therefore represent the frequency of a given behavior exhibited by a juvenile or received by a juvenile divided by the estimated number of hours spent visible, for a given period. The rates at which food-offering calls were recorded were probably underestimated because it is unlikely that observers heard all these low-amplitude vocalizations, which were produced by adults who were often several meters away from a given focal immature at the beginning of a food-transfer interaction. Nonetheless, it is doubtful that underestimation was age biased. I used linear mixed models incorporating repeated measures to investigate the effects of age category on behavior. Additionally, group and individual within group were considered random effects in the model. To account for correlation between measurements taken on the same individual over time, a first-order autoregressive correlation structure was used. Heterogeneity of variance across ages was evaluated against equal variances across ages using Akaike Information Criteria (AIC); the heterogeneous variance structure was adopted as it yielded the smallest AIC in all analyses. Analyses were conducted in SAS. All tests are 2-tailed and significance level set at P < 0.05.

Adults not only used food-offering calls to encourage young to take food that the caller had captured. Starting at about 20 weeks of age, adults began to use the call in a new context: to alert juveniles to productive prey-foraging sites. During the study's first year, we had observed 3 instances in which an adult initiated extractive foraging, stopped without having captured prey, and then emitted a food-offering call. The juvenile who approached in response to the call began to forage immediately in the same place where the adult had just been foraging and found the prey within seconds. The adult either left as soon as the juvenile found the item or stayed but, either way, allowed the juvenile to eat the prey without further interaction (Rapaport and Ruiz-Miranda 2002). To that initial report, 12 other similar events are now added (Table 2). During these events, the targeted substrates varied but the food item was always concealed within vegetation. Adults directed juveniles to find items within the following substrates: knothole in a branch (7 instances), curled dead leaf (2), broken end of a branch (3), crevice in twisted vines (2), and thick tangle of dead vines, small leaves, and twigs (1).

In all but 2 cases, the juveniles found prey, which was most often arthropods but also included tree frogs (Table 2). In one of the exceptions, a subadult male helper of the Mais Um (MA) group gave a food-offering call after having briefly foraged in a knothole and found and eaten a small (≤1.0 cm in length) prey item. The male's 40-week-old brother approached in response to the call and foraged in the knothole but did not extract anything. Both individuals then left the area. The other exception was the only event to involve nonprey. In this instance, the breeding female of the BA group was holding and eating the relatively large fruit of a Eugenia robustovenosa tree when she emitted a food-offering call, but the group's 21-week-old juvenile did not approach to take the fruit from her hand. She then reached down and placed the fruit in a crevice of twisted vines. This was the only time that an individual did not let a fruit just drop when finished with it. The female emitted another food-offering call, at which point her juvenile daughter approached and foraged by inserting her fingers into the vine crevices around the fruit but neither the juvenile nor her mother extracted it. The remaining adult-directed foraging events were striking with regard to the speed with which juveniles extracted prey. It was far more typical for juveniles to engage in extensive manipulative foraging, in substrate after substrate, and without success. For instance, when juveniles approached and foraged on a substrate where another individual was foraging or had been foraging within the past 15 s, but in which food-offering calls were not heard, the juveniles did not often capture prey; the proportion of these juvenile-initiated social foraging bouts that resulted in successful prey capture was only 3.33% (n = 13 juveniles; range: 0.00–8.90%).

Individuals in 7 of the 8 different study groups participated in the 15 unambiguous adult-directed foraging events (Table 2). Mothers, putative fathers (i.e., non-natal adult males of a group; genetic paternity was not determined), and male helpers were seen to call their group's juveniles to vegetation substrates where prey could be found. Adult-directed foraging events were distributed among 6 different mothers (8 events), 4 putative fathers (4 events), and 2 male helpers (3 events). Ten different juveniles, 4 females and 6 males, were involved in these adult-directed events.

Most instances occurred when the juveniles were in the 2 middle age categories. Five events were observed when immatures were in age-category 2 and 8 events when they were in age-category 3. The youngest juvenile to be so directed to a substrate containing prey was 20 weeks of age and the oldest was 45 weeks of age (Table 2). Although the total number of observation days varied among juvenile age categories (an observation day being defined as that in which >2 focals were completed; repeated measures analysis of variance F = 4.324; df = 3; P = 0.012; range: 97–170 days), the only significant post hoc comparison was between the first age category and the second (9–20 vs. 21–32 weeks: t-test, t = −3.568, df = 10, P < 0.05), suggesting that the drop in number of adult-directed prey captures seen between the juvenile age-category 3 and 4 was not simply due to disparity in observation time.

The first criterion by Caro and Hauser is assessed by excluding the possibility that tamarin food-offering calls are given regardless of the presence of immatures. Previous studies have demonstrated that food-offering calls are made only in the presence of young and can be categorized unequivocally as parental and alloparental behaviors (Feistner and Price 1991; Joyce and Snowdon 2007). In my study population, almost exclusively it was immatures, not adults, who responded to the calls (93/94 calls). In the single instance in which an adult responded to a call, the food-possessor did not relinquish the prey it had captured (Rapaport and Brown 2008). Juveniles, in contrast, did not emit food-offering calls. All the main study groups contained immature group members while under observation, but ad libitum observations were conducted for brief periods on 2 additional groups that did not have immatures. One, comprised 4 adults, was observed for 8 all-day follows; the other consisted of 3 adults and was observed for 11 days (≥6 h group-contact time per day). Food-offering calls were never heard in these adult-only groups nor were any events observed that could be construed as adult-directed prey capture. What is more, adults in the main study groups varied their calling behavior in developmentally appropriate ways. Importantly, adults used the vocalization to call attention to productive prey-foraging sites only when juveniles were at the developmental stage in which they were often foraging for prey, but their prey-capture success rates were either low or increasing; that is, when the young were inexperienced. The behavior was extinguished when the young were effectively foraging for prey on their own. Thus, the first criterion of Caro and Hauser (1992) is met.

The second criterion by Caro and Hauser (1992) stipulates that tutors sustain cost or gain no immediate benefit from their actions. Food-offering vocalizations may cause a caller to incur a cost because prey that could have been eaten by the caller ends up in the hands of another individual. Helpers hypothetically may gain immediate benefit for their generosity in the form of increased tolerance from breeding group members, as has been suggested for other infant-caretaking behaviors (Clutton-Brock et al. 2000; Sánchez et al. 2002; Ginther and Snowdon 2009) although this supposition has not been tested.

Adults attract juveniles to them when they call with prey-in-hand and as a result food transfer is facilitated. Calling to offer captured prey in general, therefore, is an efficient means to supplement juvenile nutrition and must be excluded as teaching according to the second criterion of Caro and Hauser (1992). However, experimental evidence suggests that food transfer, in addition to serving as an efficient means to supplement nutrition, also may act to inform young callitrichids about diet. Young tamarins (Leontopithecus rosalia: Rapaport 1999; Saguinus oedipus: Joyce and Snowdon 2004) and common marmosets (Vitale and Queyras 1997; Voelkl et al. 2006) are more likely to receive novel foods from adults than foods with which they are familiar (but see Brown et al. 2005). Young common marmosets are responsible for any information gained because they beg more for novel foods (Brown et al. 2005; Voelkl et al. 2006). Immature tamarins, on the other hand, beg indiscriminately, whereas adults more readily relinquish acceptable food items that are novel to the immatures (Rapaport 1999; Joyce and Snowdon 2004). By varying their provisioning behavior depending on the novelty of the item, tamarin adults may increase the likelihood that naive immatures will learn to identify appropriate foods. It is not yet known whether immatures do, in fact, learn to incorporate novel foods in their diet from provisioning interactions.

Are adult-directed prey captures an efficient means by which to augment juvenile feeding and do adults therefore benefit from these interactions? During prey transfers accompanied by food-offering calls and during adult-directed prey captures, the adult typically engages in manipulative foraging, locates a prey item, gives a food-offering call, and waits for a juvenile to approach. The salient differences between the 2 types of interactions are the identity of the individual who captures the prey item and the greater time lag between initial prey detection and extraction in adult-directed prey captures. When an adult finds a prey item but does not capture it and instead calls and waits for a juvenile to search for it, the adult runs the risk that the prey will escape or that the juvenile will not find it, as may have occurred in the incidence between brothers of the MA group. If capture outcomes of immatures are less likely than those of adults, even after prey has been located for the juveniles, then immediate nutritional benefits to juveniles would be optimized if the adults were to catch the prey immediately and offer it to them. On the other hand, if juveniles do not experience higher rates of nondetection or prey escape once they have been alerted to the presence of the item, adults might obtain small energetic savings by allowing the juveniles to sustain the cost of prey extraction. Determination of the comparative rates at which common prey species escape detection and capture by wild-born golden lion tamarin adults and juveniles would help to delineate the costs or benefits to adult-directed prey captures. If juveniles were found to experience higher failure rates, then the second criterion by Caro and Hauser (1992) would be satisfied. In the long-term, though, both parents and related helpers will benefit if these behaviors function to reduce the period of provisioning and/or increase offspring survival (Thornton and Raihani 2008).

The golden lion tamarins' adult-directed prey capture behavior satisfies the third criterion of Caro and Hauser (1992); that is, the behavior encourages age-appropriate experience. Around the age of weaning, when young golden lion tamarins spend little time attempting to find prey, adults regularly provision them with captured prey, often calling to promote transfer. As the young begin to spend more time foraging for prey but their return-rates remain at a nadir, adults begin to habitually, if infrequently, direct them to substrates where they can forage with immediate reward. By offering the juveniles an opportunity to succeed, adult-directed foraging may act to condition the young to keep trying in the face of high failure rates. Between the second and third age category, juvenile foraging-success rates increase dramatically. It is during this same period that adults most often provide juveniles with opportunities to experience productive foraging. Thus, adults progressively modify their food-calling behavior in a manner appropriate to the skill development of their young.

The rise in the rate at which juveniles captured prey independently concomitant with adult-directed prey capture suggests that the experience the adults provide may accelerate the development of the young tamarins' foraging skills. If adults are providing food-related information, what is it specifically that juveniles may be learning? Juveniles may learn from these interactions what kinds of prey are appropriate to eat but this possibility was not addressed in my study. In wild meerkats, adults hasten the development of their pups' prey-capture abilities by providing them with progressively more challenging prey-handling opportunities. As a result of experience, the pups learn at a younger age to dispatch dangerous prey (Thornton and McAuliffe 2006). It is unlikely, though, that adult-directed foraging instructs juvenile tamarins in prey-handling techniques. In all the adult-directed foraging events in which prey was extracted, the prey was removed quickly and then immediately killed and eaten. In contrast to calls made during adult-directed prey capture, however, food-offering calls made in the context of food transfer may function to enhance prey-handling proficiency. Adults with prey in hand more frequently used food-offering vocalizations to encourage juveniles to approach and take the item when the prey was alive than when it was dead, especially when juveniles were reluctant to take the item (Rapaport 2006).

The adult wild golden lion tamarins could be offering information to their young about prey-foraging substrate properties. Adult cotton-top tamarins (S. oedipus) helped their offspring to solve a novel laboratory foraging task by partially solving the task in the presence of the pupil, exposing the hidden food, and allowing the youngster to remove it from the apparatus. Parents not only offered this active assistance but they showed sensitivity to individual juvenile foraging proficiency, as they refused more begs as soon as the juveniles had solved the foraging task. As in the meerkat study (Thornton and McAuliffe 2006), the naive young tamarins appeared to learn more quickly due to adult encouragement (Humle and Snowdon 2008). My study does not provide evidence that young tamarins do indeed learn as a consequence of responding to food-offering calls by adults. The use of playbacks of adult food-offering calls at foraging sites that have been experimentally supplied with hidden food could help to unequivocally determine whether naive wild golden lion tamarins learn to find prey more quickly or efficiently as a direct result of adult demonstration and whether the behavior serves to focus the young on appropriate substrates, thus satisfying the fourth criterion of Caro and Hauser (1992).

Unlike pied babblers, who frequently (about 0.6 times per hour) used calls to recruit their fledglings to profitable foraging patches (Radford and Ridley 2006), the tamarins only rarely called their young to forage for hidden food. I eliminated many questionable events from consideration because the adult's behavior before the interaction had not been observed. Even if most of those deleted interactions were, in fact, adult-directed prey captures, the behavior still would have been quite rare. Adult-directed prey captures might be predicted to occur at a higher frequency if these events are important for the development of competent prey-foraging behavior (Thornton and Raihani 2008). If, on the other hand, the function of these interactions is to encourage poor foragers, the events need not occur often but instead should be directed to particularly slow learners or to less diligent foragers. Of course, this would require adults to somehow track the proficiencies of individual pupils rather than merely to adjust their behavior with juvenile age. For the demonstrations to be effective, a juvenile would be expected to experience an increase in prey-foraging return rate or at the very least an increase in prey-foraging effort on similar substrates, following an adult-directed prey-capture event. Further observational and experimental studies, designed to assess juvenile behavior before and after adult-directed foraging, could resolve these issues.

Caro and Hauser (1992) divided teaching into 2 categories: “opportunity teaching” in which the demonstrator puts the pupil in “a situation conducive to acquiring a new skill or knowledge” and “coaching” in which the demonstrator “directly alters the behavior of a pupil by encouragement or punishment” (p. 166). Coaching requires that the demonstrator actively mold the pupil's behavior, whereas in opportunity teaching, the demonstrator simply creates an environment that facilitates learning by the pupil as a result of the pupil's own behavior. The tamarins in my study did not actively change their pupils' behavior in the same way that captive gorilla mothers can be said to coach their offspring when they physically alter their infant's movements to encourage increasingly independent locomotion (Gorilla gorilla: Whiten 1999). However, the adult tamarins' food-offering calls do encourage juveniles to approach a situation in which productive-substrate properties may be learned. Extractive foraging for prey requires a lion tamarin to choose a foraging substrate with limited information regarding the probability of success. Furthermore, once a prey source has been depleted it may not provide prey again because favored prey species typically occur as solitary items and because the substrate may have been destroyed in the course of foraging, as often occurs when tamarins forage on dead and dying vegetation. Thus, a foraging tamarin's challenge is to generalize from a previous productive food site to a new one.