Animals commonly steal food from other species, termed interspecific kleptoparasitism, but why animals engage in kleptoparasitism compared with alternate foraging tactics, and under what circumstances they do so, is not fully understood.
Determining what specific benefits animals gain from kleptoparasitism could provide valuable insight into its evolution. Here, we investigate the benefits of kleptoparasitism for a population of individually recognizable and free-living fork-tailed drongos (Dicrurus adsimilis) in the southern Kalahari Desert.
Drongos engaged in two foraging behaviours: self-foraging for small insects or following other species which they kleptoparasitized for larger terrestrial prey that they could not capture themselves. Kleptoparasitism consequently enabled drongos to exploit a new foraging niche.
Kleptoparasitism benefitted drongos most in the morning and on colder days because at these times pay-offs from kleptoparasitism remained stable, while those from self-foraging declined. However, drongos engaged in kleptoparasitism less than expected given the overall high (but more variable) pay-offs from this behaviour, suggesting that kleptoparasitism is a risky foraging tactic and may incur additional foraging costs compared with self-foraging.
This is the first study to comprehensively investigate the benefits of facultatively engaging in kleptoparasitism, demonstrating that animals may switch to kleptoparasitism to exploit a new foraging niche when pay-offs exceed those from alternate foraging behaviours.
For over 30 years, animals have been known to engage in alternative foraging tactics when these increase foraging success (Pyke, Pulliam & Charnov 1977; Stephens, Brown & Ydenberg 2007). However, the adaptive benefits animals gain from stealing food from other species rather than obtaining it for themselves, termed interspecific kleptoparasitism (Brockmann & Barnard 1979), are not fully understood. Interspecific kleptoparasitism is widespread in the animal kingdom, occurring in birds, fish, mammals, reptiles, arthropods, molluscs and several other taxa (reviewed by Iyengar 2008). Many species only employ kleptoparasitism opportunistically, and this behaviour provides a small contribution to their diet (Brockmann & Barnard 1979; Iyengar 2008). But for others, kleptoparasitism is a more commonly used foraging tactic; for example, spotted hyaenas (Crocuta crocuta) obtain 20% of the carcasses they feed on by kleptoparasitism (Honer et al. 2002), and the spider Curimagua bayano appears to be an obligate kleptoparasite, gaining all its food from this behaviour (Vollrath 1978). For animals to invest a significant proportion of their time and energy in searching for kleptoparasitic opportunities, they must gain consistent benefits from kleptoparasitism, and pay-offs must be greater than those available from alternate foraging tactics (Kushlan 1978; Brockmann & Barnard 1979; Broom & Ruxton 1998; Iyengar 2008). However, to date, there has been little investigation of the pay-offs kleptoparasitism provides compared with alternate foraging tactics.
Kleptoparasitism is predicted to be of principal benefit when the pay-offs from other foraging behaviours decline (Brockmann & Barnard 1979; Broom & Ruxton 1998; Stillman et al. 2002; Yates & Broom 2007), such as when typical food sources are scarce (Clair et al. 2001). Kleptoparasitism could be particularly advantageous under these conditions when it enables species to gain access to a new foraging niche otherwise inaccessible to them (Brockmann & Barnard 1979; Morissette & Himmelman 2000; Morand-Ferron, Sol & Lefebvre 2007). Research investigating the pay-offs gained from kleptoparasitism has mainly focussed on seabird breeding colonies, where birds returning to their young with fish are frequently kleptoparasitized by other species or conspecifics (Furness 1987; Iyengar 2008). However, the majority of these studies only consider pay-offs gained from kleptoparasitism in a specific ecological context, do not compare pay-offs with those available from other foraging tactics and do not investigate whether kleptoparasitism enables access to a new foraging niche (Osorno, Torres & Garcia 1992; Vickery & Brooke 1994; Ratcliffe et al. 1997). Research has shown that roseate terns (Sterna dougallii) gain greater pay-offs from kleptoparasitism than from finding food themselves (self-foraging) when at their breeding colony (Shealer & Spendelow 2002). In addition, kleptoparasitic roseate terns (Shealer et al. 2005) and common terns (Sterna hirundo) (Garcia, Becker & Favero 2011) have greater reproductive success than their non-kleptoparasitic conspecifics. But, the broader relevance of kleptoparasitism for these species is unclear as only 10 roseate terns (3–5% of the population) and 10 common terns (1–2% of the population) were observed to kleptoparasitize food in these studies (from Shealer et al. 2005 and Garcia, Becker & Favero 2011, respectively). To date, the only studies to have compared kleptoparasitism with alternative foraging tactics in a frequently kleptoparasitic species were undertaken on the sedentary marine snail Trichotropis cancellata, which steals food from the mouths of polychaete worms (Pernet & Kohn 1998; Iyengar 2002). This research illustrated the benefits gained by snails associating with worms but did not consider whether snails facultatively switch foraging tactics depending upon the pay-offs. Furthermore, the adaptive benefits of kleptoparasitism could be markedly different for highly mobile animals stealing food from multiple host species, where pay-offs from foraging tactics are likely to vary.
To investigate the pay-offs gained from kleptoparasitism compared with alternate foraging tactics, it is important to determine whether these behaviours are mutually exclusive (Broom & Ruxton 1998; Giraldeau & Beauchamp 1999). If foraging tactics can be undertaken simultaneously, then their benefits will not trade-off and constraints on the evolution of kleptoparasitism will be relaxed. Furthermore, while kleptoparasitism may provide large benefits compared with other foraging tactics, pay-offs may also be highly variable (Koops & Giraldeau 1996). Risk-sensitive foraging theory predicts that animals will often adopt behaviours with lower mean pay-offs, but also lower variance, that are more likely to meet their specific energetic needs (e.g. for survival over winter) (McNamara & Houston 1992). Consequently, a high-pay-off, high-risk kleptoparasitic foraging tactic may be optimal only when pay-offs from alternative foraging tactics decline.
The drongos (Dicruridae) have previously been overlooked as one of the most extensively kleptoparasitic bird families, despite the fact that at least eight of the twenty species employ kleptoparasitism (del Hoyo et al. 2009). Recent research has begun to illustrate the importance of kleptoparasitism for drongo species (Satischandra et al. 2007). In particular, studies on the fork-tailed drongo (Dicrurus adsimilis) have revealed that they frequently engage in kleptoparasitism (Herremans & Herremans-Tonnoeyr 1997; Ridley & Raihani 2007) and employ complex kleptoparasitic strategies (Ridley & Child 2009; Flower 2011; Flower & Gribble 2012). Fork-tailed drongos principally forage alone, hawking small flying insects or gleaning invertebrate and vertebrate prey from the ground and trees (Hockey, Dean & Ryan 2005). Drongos also approach and follow heterospecific foraging groups, catching flushed prey disturbed by these species or kleptoparasitizing host individuals directly for larger invertebrate and vertebrate prey (Herremans & Herremans-Tonnoeyr 1997; Ridley & Raihani 2007; Ridley & Child 2009). Species are kleptoparasitized either by physical attack (45% of attempts; 40% success) or by the use of deceptive false alarm calls (55% of attempts; 35% success) in response to which target individuals commonly flee to cover, enabling the drongo to collect abandoned prey (Ridley & Raihani 2007; Flower 2011; Flower & Gribble 2012). Following failed false alarm calls, drongos may additionally attack (34% of failed false alarm attempts) and evidence suggests that the false alarm call and subsequent attack might act synergistically to increase success in such kleptoparasitism attempts (63% success, Flower & Gribble 2012). The false alarm strategy employed by drongos is perhaps best considered a form of ‘stealth kleptoparasitism’, whereby the kleptoparasite surprises a host and steals a food item before the host can react (Giraldeau & Caraco 2000; Ridley & Child 2009). As drongos principally kleptoparasitize terrestrially foraging species and have been reported to do so most frequently in winter (Herremans & Herremans-Tonnoeyr 1997), kleptoparasitism has been suggested to provide drongos with access to a novel foraging niche that could be of particular benefit when pay-offs from self-foraging decline during cold winter conditions (Ridley, Child & Bell 2007).
To determine what benefits drongos gain from kleptoparasitism, this study therefore investigates (i) whether kleptoparasitism exploits a novel foraging niche and the extent to which it is mutually exclusive of other foraging tactics, (ii) whether the time drongos spend following other species and the amount of food obtained by kleptoparasitism is affected by environmental conditions and (iii) what pay-offs drongos gain from following other species and engaging in kleptoparasitism, compared with alternate foraging tactics.
Materials and methods
Data were collected between March and August 2008 on a wild population of fork-tailed drongos located in an area of xeric savanna in the Kalahari Desert (26°58′S, 21°49′E). Details of the habitat and climate have been published elsewhere (Clutton-Brock et al. 1999). The study population consisted of 25 drongos, habituated to observation at <5 m. Each individual was captured using walk-in traps baited with mealworms and given unique colour rings to enable the detailed observation of behaviour. During capture, 50 μL blood samples were collected through brachial venipuncture; blood samples were transferred to 700 μL of lysis buffer (Longmire's solution), DNA was extracted and PCR techniques were used to establish the sex of each individual (Griffiths et al. 1998). Drongos were classified by age as either adults or fledglings according to their plumage coloration (Hockey, Dean & Ryan 2005). Individual drongos were located for data collection by searching their typical range (1 × 1 km c.). Drongo ranges overlapped with the ranges of 14 groups of meerkats (Suricata suricatta) and 10 groups of pied babblers (Turdoides bicolor) whose members were habituated and individually recognizable (Clutton-Brock et al. 1999; Ridley & Raihani 2007). Interactions between drongos and these species could therefore be observed at <5 m. Drongos were also observed interacting with 23 unhabituated bird species (see Table 1 for species list); to avoid disturbing these interactions, observations were made at a distance of 20–30 m with binoculars. The species that drongos targeted foraged in the open, and it was therefore possible to see the behaviour of both the focal drongo and target species at all times.
Table 1. Species kleptoparasitized by drongos
Body mass relative to drongo
Number of drongos that followed species (total = 25)
Proportion of follow time with species (% per drongo)
To investigate drongo foraging behaviour, 292 focal observations (average focal duration 54 ± 1 min; mean ± 1 SE) were undertaken on 25 drongo individuals with a minimum of six focals collected per individual (mean total observation time per drongo: 10·47 h, range 4·02–20·10). During focal observations, a Palm T/X® (Palm Inc, Sunnyvale, CA, USA) hand-held computer was used to record: (i) the time (seconds) drongos were engaged in different foraging behaviours, (ii) all foraging attempts to catch a prey item, (iii) whether these were successful and (iv) the size and type of prey item drongos foraged for. When it was not possible to observe the behaviour of a focal drongo, focal observations were paused and the observer moved to a location at which focal observations could be restarted. Drongos had diverse foraging behaviour and engaged in two foraging modes: foraging alone where the drongo scanned the environment widely for food items, moving its head rapidly and not watching other species in its vicinity, and following other species where the drongo approached another target species or mixed species group to <20 m, watched them and followed them as they foraged. Following other species was always an option available, as species drongos kleptoparasitized were always observed in the drongo's territories. Within the foraging modes, drongos caught food using a number of foraging tactics. Whether foraging alone or following a species, drongos could self-forage by hawking flying prey or gleaning prey from substrates. When drongos were following other species they could additionally obtain food by catching flushed prey disturbed by the species followed or by kleptoparasitism directly from individuals of the species followed for prey they held. All attacks, false alarms and attacks followed by false alarms were considered to be kleptoparasitism attempts because they were all initiated when a target host individual was in physical possession of a food item (Brockmann & Barnard 1979; Iyengar 2008).
Prey items drongos caught were categorized by size relative to drongo bill length, and data were available for the mean wet mass in grams (g) of fifty prey items representative of each size category (Raihani & Ridley 2007). Size categories were as follows: tiny = <¼ bill length (0·02 g), small = ¼ – 1 bill length (0·11 g), medium =1 – 2 bill lengths (0·45 g) and large =2 – 3 bill lengths (0·84 g); items larger than this were scored as multiples of ‘large’. Prey items were additionally classified into groups that could be accurately distinguished in the field and were broadly representative of different aerial and terrestrial prey types as follows: (i) termites and ants (Isoptera and Hymenoptera (ants only)), (ii) flies (Diptera), (iii) bees, wasps, butterflies and moths (Hymenoptera (excluding ants) and Lepidoptera), (iv) crickets (Orthoptera), (v) mantids, stick insects, spiders, solifuges and scorpions (Mantodea, Phasmatodea and Arachnida), (vi) larvae, (vii) lizards and geckos or (viii) unidentified insect.
Drongos were typically inactive in the middle of the day, the time period when temperatures peak in the Kalahari (Smit & McKechnie 2010), and focal observations were therefore classified as occurring in either the morning (<13:00) when temperatures were typically lower or in the evening (≥13:00) when temperatures were typically high (Smit & McKechnie 2010). Minimum daily temperature (°C) was recorded to determine its influence on drongo foraging behaviour. Insects, and particularly flying insects, are less active in the morning when temperatures are low (Mellanby 1939; Taylor 1963), and their abundance declines during the cold winter months of the Kalahari between May and September (Doolan & Macdonald 1996). Time of day and minimum daily temperature are therefore an approximate indication of insect activity and abundance over the course of the day and throughout the year.
Focal observations revealed that drongos followed a large number of different host species but often spent only a short period of time with each. Consequently, the time drongos spent following different species during a single focal observation and the corresponding biomass intake were summed to provide a single measure of each (see Table 1 for species details). The total time drongos engaged in different foraging behaviours (min) within a focal observation, and the biomass intake (g) obtained from these behaviours was also calculated. Drongos commonly failed to capture any prey when engaged in a foraging behaviour during a focal observation. Consequently, for all analyses of rates of biomass intake (g min−1) from behaviours, it was necessary to undertake two separate analyses considering (i) the likelihood a food item was caught and (ii) the rate of biomass intake when at least one prey item was caught.
Statistical analyses were conducted in Genstat 10.0 (Lawes Agricultural Trust, Rothamsted, Harpenden, UK). For analyses where a single categorical predictor was tested on a response term with a normal distribution, repeated measures anovas were conducted with drongo identity as a blocking factor. For multivariate analyses involving repeated sampling of individuals, General Linear Mixed Models (LMMs) with identity link functions or Generalized Linear Mixed Models (GLMMs) with logit link functions and binomial distributions were used. Variance components were estimated using the restricted maximum likelihood method for LMMs and the method of Schall (1991) for GLMMs. In each model, all potential explanatory terms were entered together and then dropped sequentially, checking for interactions, until only those terms that explained significant variation remained (the minimal model). Each dropped term was then put back into the minimal model to obtain the level of nonsignificance and to check that significant terms had not been wrongly excluded.
In all models, explanatory terms included drongo age (juvenile or adult), drongo sex (male or female), time of day (morning or evening) and minimum daily temperature; drongo identity and date were included as random terms. Foraging mode (follow or alone) was included as an explanatory factor for analyses of (i) the overall rate of biomass intake when following other species or foraging alone and (ii) the rate of biomass intake from self-foraging when following other species or foraging alone. In all analyses of rates of biomass intake, the sum of time (minutes) engaged in a foraging behaviour during a focal observation was included to control for any effect it had. Levene's tests were used to determine the variance in pay-offs from different foraging modes, (conducted in Minitab 15; Minitab Ltd, Coventry, UK). Data were log10-transformed where they differed significantly from normality (Shapiro–Wilk tests), and all statistical tests were two-tailed unless otherwise stated. Where multiple comparisons were undertaken, Dunn-Sidak procedures were used to determine pairwise differences. Means ± 1 SE are quoted except where otherwise stated.
Prey intake from kleptoparasitism
Drongos spent on average 71 ± 3% (range 43–94%) of their foraging time foraging alone; when foraging alone, they were only able to self-forage, and this provided 61 ± 5% (range 17–97%) of their total biomass intake (Fig. 1a,b). However, drongos also spent 29 ± 3% (range 6–57%) of foraging time following other species, providing 39 ± 5% (range 3–83%) of their total biomass intake: 23 ± 4% from kleptoparasitism (with all individuals observed attempting to obtain food by kleptoparasitism), 10 ± 2% from catching flushed prey and 6 ± 1% from self-foraging (Fig. 1a,b). Although drongos were able to catch prey by self-foraging when following other species, this formed a small proportion of biomass intake and both the likelihood that a drongo captured prey and the rate of biomass intake from self-foraging were lower when following other species than when foraging alone (see later). Consequently, following other species appears to be incompatible with continuing to self-forage effectively.
When self-foraging, 81 ± 2% of the food drongos captured consisted of small flies, insects, termites and ants. Conversely, while following host species, drongos obtained larger terrestrial prey types including lizards, larvae, mantids, stick insects, spiders, solifuges, scorpions and crickets by kleptoparasitism or capture of flushed prey. These large prey types collectively accounted for a significantly greater proportion of biomass intake from kleptoparasitism (58 ± 6%) and flushing (47 ± 6%) per drongo per focal observation than from self-foraging (19 ± 2%), (repeated measures anova: F2, 42 = 14·12, P ≤ 0·001, N =22 drongos). Furthermore, the mean mass of prey items caught by kleptoparasitism (0·45 ± 0·05 g) was greater than those caught by flushing (0·14 ± 0·01 g) or self-foraging (0·05 ± 0·004 g), (repeated measures anova: F2, 42 = 123·72, P = <0·001, N =22 drongos).
Variation in kleptoparasitism with changes in environmental conditions
Drongos spent more time following host species in the morning than in the evening (P ≤ 0·001; Fig. 2a, Table 2), and males spent more time following than females (P =0·036; Fig. 2b, Table 2). Male drongos also gained more of their biomass intake from kleptoparasitism in the morning than evening, while females did not (P =0·019; Fig. 2c, Table 2), and results indicate that males gained more biomass intake from kleptoparasitism overall than did females (P =0·002; Table 2). Overall, drongos gained more of their biomass intake from kleptoparasitism when daily temperatures were low (P =0·008; Fig. 2d, Table 2).
Table 2. Time spent following other species and biomass intake from kleptoparasitism
Effect ± SE
Generalized Linear Mixed Models of the factors affecting: (i) the proportion of foraging time spent following other species (292 focal observations on 25 drongos) and (ii) the proportion of biomass intake gained from kleptoparasitism (282 focal observations on 25 drongos).
Significant terms are highlighted in bold. In all analyses, degrees of freedom equal 1 for each fitted term.
(i) Proportion of foraging time spent following other species
Time of day (even./morn.)
−1·78 ± 0·20
0·98 ± 0·28
0·63 ± 0·28
−0·035 ± 0·019
−1·33 ± 0·24
(ii) Proportion of biomass intake gained from kleptoparasitism
Pay-offs from following other species versus foraging alone
The likelihood that drongos captured prey was greater when following other species than when foraging alone (P =0·033; Table 3). More specific consideration of the rate of biomass intake showed that drongos gained a greater rate of biomass intake when following other species than when foraging alone in the morning, but these behaviours did not differ in the evening (P =0·008; Fig. 3a, Table 3). Both when foraging alone and following other species, the rate of biomass intake decreased when daily temperatures were low (P =0·012; Fig. 3b, Table 3). An investigation of the variation in pay-offs from following other species and foraging alone revealed that the mean rate of biomass intake gained when following other species had a significantly greater variance than the rate from foraging alone, both in the morning and in the evening (Levene's test: = 2·92, P ≤ 0·001, N =25 drongos; Fig. 3c). However, the standard deviation of the mean rate of biomass intake from following species in the morning may be expected to be proportionately greater than that from foraging alone (owing to the larger mean value), which could account for the greater variance in rate of mass intake from following other species at this time. Consequently, the coefficient of variation for the samples (which accounts for differences in sample mean values) was compared, further indicating that the variance in the rate of mass intake from following species was greater than from foraging alone (follow species: morning = 0·80, evening = 1·15; forage alone: morning = 0·67, evening = 0·46).
Table 3. Pay-offs from foraging alone versus following other species
Effect ± SE
Effect ± SE
Likelihood prey item caught
LOG Rate of biomass intake
Factors affecting the likelihood that a prey item was caught (Generalized Linear Mixed Models), and the rate of biomass intake (Linear Mixed Models) when at least one prey item was caught, while foraging alone or following other species (likelihood = 466 observations on 25 drongos; rate of biomass = 384 observations on 25 drongos).
Significant terms are highlighted in bold. In all analyses, degrees of freedom equal 1 for each fitted term.
To determine why pay-offs from following other species remained more stable in the morning, we compared the specific pay-offs drongos gained from kleptoparasitism when following other species, from capturing flushed prey when following other species, and from self-foraging both when following other species and foraging alone. When self-foraging, the likelihood that a prey item was captured and the rate of biomass intake were lower in the morning than in the evening (likelihood food item caught: P ≤ 0·001; rate of biomass intake: P ≤ 0·001; Fig. 4a, Table 4). The likelihood that a prey item was captured by self-foraging also decreased when daily temperatures were low (P =0·012; Table 4), and at low temperatures, the rate of biomass intake from self-foraging when following other species was less than when foraging alone (P =0·035; Fig. 4b, Table 4). Females were also more likely to obtain prey by self-foraging than males (P =0·003; Fig. 4c, Table 4). When considering flushed prey, analyses found that the rate of biomass intake from this foraging tactic was lower in the morning than in the evening (P =0·039; Fig. 4d, Table 4). The likelihood that flushed prey were captured and the rate of biomass intake from capturing flushed prey were unaffected by daily temperature (Table 4). When considering kleptoparasitism, analyses found that the likelihood drongos obtained prey by kleptoparasitism and the rate of biomass intake from this foraging tactic did not differ in the morning compared with the evening and did not decline as daily temperatures decreased (Likelihood: P =0·109; Rate: P =0·913; Table 4). However, males were more likely to obtain prey by kleptoparasitism than females (P =0·042; Fig. 4e, Table 4).
Table 4. Pay-offs from self-foraging, capturing flushed prey and kleptoparasitism
Effect ± SE
Effect ± SE
Likelihood prey item caught
LOG rate of biomass intake
Factors affecting the likelihood that a prey item was caught (Generalized Linear Mixed Models), and the rate of biomass intake from foraging tactics when at least one prey item was caught (Linear Mixed Models): (i) from self-foraging when foraging alone or following other species (likelihood = 466 observations on 25 drongos; rate of biomass = 349 observations on 25 drongos), (ii) from capturing flushed prey when following other species (likelihood = 183 observations on 25 drongos; rate of biomass = 98 observations on 25 drongos), and (iii) from kleptoparasitism when following other species (likelihood = 182 observations on 25 drongos; rate of biomass = 87 observations on 22 drongos).
Significant terms are highlighted in bold. In all analyses, degrees of freedom equal 1 for each fitted term.
(i) Self-foraging when foraging alone or following other species
(ii) Capture of flushed prey when following other species
Time in behaviour (mins)
0·073 ± 0·013
−0·010 ± 0·0036
Time of day (even./morn.)
0·23 ± 0·41
0·29 ± 0·014
−0·045 ± 0·031
−0·0019 ± 0·010
−0·29 ± 0·43
−0·0063 ± 0·14
0·20 ± 0·45
0·027 ± 0·014
0·0033 ± 0·20
−2·10 ± 0·071
(iii) Kleptoparasitism of prey when following other species
Time in behaviour (mins)
0·10 ± 0·017
−0·0046 ± 0·0025
1·06 ± 0·52
0·083 ± 0·093
Time of day (even./morn.)
−0·78 ± 0·48
−0·01 ± 0·12
−0·015 ± 0·034
0·0023 ± 0·0068
0·34 ± 0·58
0·017 ± 0·080
−0·90 ± 0·41
−1·57 ± 0·037
Benefits of kleptoparasitism
Fork-tailed drongos gained significant foraging benefits from kleptoparasitism, acquiring nearly a quarter of their food from this foraging tactic. By following other species and engaging in kleptoparasitism, drongos were able to exploit a novel foraging niche in which they caught larger terrestrial invertebrate and vertebrate prey items compared with other foraging tactics. In particular, subterranean prey items such as scorpions and solifuges are inaccessible to self-foraging drongos because they cannot dig into the substrate, and we never observed drongos catching these prey independently of their host species. Kleptoparasitism was particularly beneficial in the morning and when daily temperature was low. This appears to be because the pay-offs from self-foraging decline both in the cold mornings and on cold days, while pay-offs from kleptoparasitism remain stable at these times. Results from this study therefore support predictions from theoretical models that animals will engage in kleptoparasitism when food availability is low (Broom & Ruxton 1998; Yates & Broom 2007; Broom et al. 2010). Furthermore, they illustrate how kleptoparasitism provides drongos with significant adaptive benefits because it enables the exploitation of a new foraging niche when pay-offs from other foraging tactics decline.
Drongos were likely to gain lower pay-offs from self-foraging under cold conditions because the activity and abundance of small insects and flies declines at such times (Mellanby 1939; Taylor 1963). Conversely, the pay-offs from kleptoparasitism were likely to remain stable because the foraging tactics used by the species drongos followed (such as subterranean excavation by hosts including sociable weavers (Phietairius socius), pied babblers and meerkats: Doolan & Macdonald 1996; Hockey, Dean & Ryan 2005; Ridley & Raihani 2007) do not rely on prey activity and are less likely to be affected by temperature. Kleptoparasitism is therefore particularly advantageous because it enables drongos to exploit a foraging niche where prey availability is less affected by changes in environmental conditions compared with other foraging tactics. A similar advantage was shown for the woodpecker finch (Camarhynchus pallidus), which uses tools to get prey from holes when food available from other foraging behaviours decrease during dry conditions and in food-scarce environments (Tebbich et al. 2002).
Costs and constraints on kleptoparasitism
The pay-offs available from a kleptoparasitic foraging tactic appear to trade-off against those from self-foraging because, when following other species, drongos did not self-forage as effectively. This trade-off most likely occurs because drongos perch lower to the ground when following other species and move less frequently to remain with their host (Ridley, Child & Bell 2007). Following other species may therefore reduce the likelihood drongos encounter or see prey they can hawk or glean, as the location of the host may not be optimal for drongo self-foraging techniques. These results confirm the assumption of kleptoparasitism models that engaging in kleptoparasitism will compromise the pay-offs animals gain from other foraging behaviours (Barnard & Sibly 1981; Broom & Ruxton 1998; Giraldeau & Beauchamp 1999; Ruxton & Broom 1999; Yates & Broom 2007). However, the trade-off between self-foraging and kleptoparasitism did not result in drongos abandoning kleptoparasitism altogether as has been predicted by some models (Ruxton & Broom 1999). The incompatibility between model predictions and our findings may occur because current models (see Broom et al. 2010 for a summary of these models) are more representative of intraspecific kleptoparasitism, where animals steal food from conspecifics, than interspecific kleptoparasitism. For example, drongos kleptoparasitize other species for food items that they are unable to obtain by self-foraging while current models assume that the same resources are available from both self-foraging and kleptoparasitism (but see Bautista, Alonso & Alonso 1998). In addition, the number of self-foraging individuals in intraspecific groups (available hosts) declines as a greater proportion switch to a kleptoparasitic strategy. In contrast, drongo pairs hold stable and exclusive territories resulting in a relatively inflexible number of kleptoparasites while the number of interspecific host individuals remains stable irrespective of how many drongos adopt kleptoparasitism. We suggest that future models of interspecific kleptoparasitism account for the fact that resources obtained from kleptoparasitism may differ in value from those obtained by self-foraging and that the number of individuals engaged in kleptoparasitism does not necessarily trade-off against those searching for or handling food.
Considering the higher biomass intake gained from following other species and engaging in kleptoparasitism, one might predict that drongos should invest more time in this behaviour than foraging alone. However, there are likely to be additional costs and constraints associated with following species and engaging in kleptoparasitism (Stillman et al. 2002) that reduce pay-offs from this behaviour compared with alternate foraging tactics. Firstly, following species may be a comparatively risky behaviour because the variance in foraging pay-offs when following other species is high. Drongos may therefore optimize their behaviour by facultatively switching between foraging behaviours depending upon both the mean pay-offs available and the variance of these pay-offs. Results therefore provide support for the predictions of risk-sensitive foraging theory (McNamara & Houston 1992) and affirm that kleptoparasitism is likely to provide unpredictable pay-offs (Koops & Giraldeau 1996).
Secondly, kleptoparasitism is likely to incur higher physical costs and risks of injury than self-foraging which, although difficult to quantify, will reduce the pay-offs from this tactic (Huntingford & Turner 1987; Flower & Gribble 2012). Drongos commonly hovered above individuals with food or chased them for up to one minute (T. Flower pers. obs.) and would grapple with individuals in physical contests, including larger species which repeatedly pecked them, such as glossy starlings (Lamprotornis nitens) and pied babblers (T. Flower pers. obs.; Ridley & Raihani 2007). Such costly kleptoparasitic attacks are also frequently unsuccessful and less successful than self-foraging attempts (Ridley, Child & Bell 2007; Flower & Gribble 2012), further increasing costs compared with alternate foraging tactics. Host species may additionally employ counter strategies that affect the pay-offs available from kleptoparasitism (Stienen & Brenninkmeijer 1999; Ridley & Raihani 2007). Pied babblers are known to be directly aggressive towards drongos that follow them, which may make kleptoparasitism no longer profitable in some cases (Ridley, Child & Bell 2007). However, the physical costs of kleptoparasitism may be mitigated for drongos because they can use false alarm calls to scare other species and steal their food without the need for a physical contest (Flower & Gribble 2012). Drongos preferentially make false alarm calls in kleptoparasitism attempts on larger species which more frequently defend their food, indicating that they do employ false alarm calls when the costs of kleptoparasitism are high (Flower & Gribble 2012). Nevertheless, species may stop responding to drongo alarm calls when they are made too frequently (Ridley & Raihani 2007; Flower 2011). We would predict an equilibrium point to be reached where drongos cease following a species when pay-offs are equal to or less than those available from switching to a different species or foraging alone.
More generally, the benefits available from kleptoparasitism are likely to vary between host species and may depend upon the frequency and value of food items the host finds, the likelihood that kleptoparasitism attempts will be successful and the physical costs incurred from kleptoparasitism (Furness 1978; Barnard, Thompson & Stephens 1982; Ratcliffe et al. 1997; Clair et al. 2001; Garcia, Favero & Vassallo 2010; Flower & Gribble 2012). Future investigation should consider the factors affecting kleptoparasitic pay-offs from different host species and how species adjust their kleptoparasitic behaviour accordingly.
Sex differences in kleptoparasitism
Sex differences in kleptoparasitism may reflect variation in both the ability of males and females to capture prey by self-foraging versus kleptoparasitism and the energetic requirements of the sexes. Males were more likely to capture prey by kleptoparasitism, engaged in kleptoparasitism more frequently and obtained a larger proportion of their food intake by kleptoparasitism than females. Conversely, females were more likely to capture prey by self-foraging than males. Males may possess characteristics that improve kleptoparasitic ability; for example, male birds typically have a larger call repertoire (Searcy 1992) which could increase the efficacy of their false alarm-calling strategy (Flower 2011). Alternatively, males may engage in kleptoparasitism more frequently than females because they suffer less from gaining more variable energy returns associated with kleptoparasitism, (e.g. female egg production requires consistent energy returns) (Wright & Radford 2011) or because they have larger energy requirements than females that force them to adopt a more risky foraging tactic with higher energy returns (Barnard & Brown 1985). Selection may also have operated on drongos to partition foraging niches between the sexes (Radford & du Plessis 2003), as pairs occupy the same territory throughout the year and may therefore compete for food (Hockey, Dean & Ryan 2005). Such partitioning could be particularly important for kleptoparasitism, as an increase in the number of individuals engaged in kleptoparasitism is predicted to reduce pay-offs from this behaviour (Barnard & Sibly 1981). This may ultimately result in the sexes evolving different adaptations to different foraging niches (Radford & du Plessis 2003).
Evolution of kleptoparasitism
Fork-tailed drongos possess many of the characteristics predicted to facilitate the evolution of kleptoparasitism. They follow host species and catch food flushed by host foraging activity (Herremans & Herremans-Tonnoeyr 1997; Ridley, Child & Bell 2007), and this association may have led to opportunities to steal food (Brockmann & Barnard 1979). Drongos also possess exceptional flying ability, are highly aggressive towards other species including predators and employ diverse foraging behaviours to catch invertebrate and vertebrate prey (Hockey, Dean & Ryan 2005), all of which are common features of kleptoparasitic species (Brockmann & Barnard 1979; Morand-Ferron, Sol & Lefebvre 2007). Perhaps most importantly, drongos have large brains relative to their body size (D. Sol, unpublished data), which recent evidence suggests is one of the most prevalent features of kleptoparasitic species (Morand-Ferron, Sol & Lefebvre 2007). Brain size is predicted to be important because kleptoparasitism is likely to involve complex tactics which may have an important learnt component (Morand-Ferron, Sol & Lefebvre 2007). This could be particularly relevant for fork-tailed drongos, which use a comparatively sophisticated deceptive alarm-calling strategy during kleptoparasitism (Ridley, Child & Bell 2007; Flower 2011) and specifically target individuals from which they are more likely to gain food (Ridley & Child 2009). However, we know little about the mechanisms that underlie such behaviour. Future research on drongos should therefore investigate the development of kleptoparasitic behaviour and in particular the acquisition of deceptive alarm calling.
Kleptoparasitism appears to have become a significant foraging tactic for fork-tailed drongos because it enables them to exploit a novel foraging niche providing large benefits at times when pay-offs from other foraging tactics decline. Furthermore, the costs and constraints associated with kleptoparasitism are likely to affect the benefits of kleptoparasitism for drongos less than for many other kleptoparasitic species for several reasons. First, despite a reduction in the pay-offs, drongos could continue to self-forage when following other species. Secondly, by using false alarm calls to kleptoparasitize other species for food, drongos may avoid the physical costs of kleptoparasitism (Flower & Gribble 2012). Finally, drongos have predictable access to many of the species they follow and gain consistent pay-offs from kleptoparasitism when following these species. For the majority of kleptoparasites, which only engage in this behaviour opportunistically, the benefits are likely to be insufficient to make it a preferable foraging tactic. For other species that obtain a significant proportion of their food from kleptoparasitism, there is evidence that they, like drongos, primarily exploit other species when pay-offs from finding food themselves decline (Clair et al. 2001) or where kleptoparasitism provides generally high pay-offs (Iyengar 2002; Shealer et al. 2005). By comprehensively examining the functional benefits of kleptoparasitism for known individuals compared with alternate behaviours, this study provides a unique insight into the factors which favour the evolution of kleptoparasitism.
Access to the study site and meerkats was kindly provided by the Kalahari Research Trust, T. H. Clutton-Brock and M. Manser. The Department of Environment and Conservation for Northern Province, SA, gave permission for research (FAUNA 270/2008), and the Percy Fitzpatrick Institute, University of Cape Town, provided ethical permission (2010/V20/TF). Many thanks to C. O'Ryan at the University of Cape Town for providing laboratory access for drongo sexing. SAFRING kindly provided permits for drongo ringing (Licence: 1263 & 1546). We thank N. Davies for supervision and M. Nelson-Flower and R. Sutcliffe for their help and advice. We also thank three anonymous reviewers for their invaluable comments, which we believe greatly improved this manuscript. Work was funded by the Natural Environment Research Council and supported by Clare College, Cambridge.