Effect of food shortage on the physiology and competitive abilities of sand martin (Riparia riparia) nestlings

Variation in fluctuation of food supplies experienced during ontogeny has important consequences for life history (Gebhardt-Henrich and Richner, 1998). The response of a growing organism to food shortage is largely determined by the plasticity of its developmental program (Smith-Gill, 1983; Schlichting and Pigliucci, 1998). In principle, such plasticity may take one of two forms. It may encompass a passive response to the stressful situation, which reflects invariability of the genetic programs controlling the rate of development (phenotypic modulation, sensuSmith-Gill, 1983). Some organisms, however, are capable of activating alternative developmental programs, which allows them to respond actively to adverse environmental conditions (developmental conversion, sensuSmith-Gill, 1983; for a review, see Schlichting and Pigliucci, 1998). Both forms of plasticity are well exemplified by the different reactions of altricial nestling birds to food shortage. Young starlings Sturnus vulgaris (Schew and Ricklefs, 1998), house sparrows Passer domesticus (Lepczyk et al., 1998) or song thrushes Turdus philomelos (Konarzewski et al., 1996; Konarzewski and Starck, 2000) attempt to continue their normal pace of growth and postembryonic development within the limits imposed by undernourishment. In contrast, the nestlings of aerial insectivorous birds (such as Meropidae, Apodidae and Hirundinidae) are capable of slowing the rate of development during periods of inclement weather and resuming the normal rate when the feeding conditions improve (Lack and Lack, 1951; Bryant, 1975; O’Connor, 1978; Emlen et al., 1991).

The fate of an altricial nestling is determined, however, not only by variability of the environmental conditions outside the nest but also, critically, by its ability to compete for limited nourishment with its nestmates (Mock and Parker, 1997). It is therefore likely that developmental plasticity increases the competitive abilities of the nestlings. The chances of successful fledgling can be further increased if developmental plasticity enhances behavioural adaptations that help to monopolise food supplies (Bengtsson and Ryden, 1981; McRae et al., 1993; Malacarne et al., 1994; Kacelnik et al., 1995; Koelliker et al., 1998). To test this, we studied developmental and behavioural plasticity in sand martin Riparia riparia nestlings, hand-reared under laboratory-controlled conditions. Sand martins are cavity-nesters, suffering little from predatory pressure (Martin and Li, 1992; Martin, 1995). Their growth and developmental rates are therefore primarily shaped by two other factors: environmental variability outside the nest and sibling competition. We focused on the association between behaviour, energy expenditure and digestive capacity of the gut, which should be particularly sensitive to feeding conditions mediated by within-brood competition.

Sand martin nestlings are unusually mobile for an altricial species. At the age of 5–6 days they are capable of actively leaving the nest cup and crawling up to 30cm within the nest tunnel towards its entrance (M. Konarzewski, unpublished data). We therefore hypothesised that the young can use this unique locomotory ability to intercept incoming parents and monopolise feeding bouts. We expected that such behaviour should be triggered by an increased number of competing, hungry nestmates (Godfray, 1995). More importantly, we also hypothesised that physiological responses to food shortages should be amplified by sibling rivalry.

When faced with food shortage, hirundinid nestlings are capable of reducing their metabolic rate and eventually entering torpor (Prinzinger and Siedle, 1988). However, such an option can presumably be chosen only at the cost of some impairment of locomotor abilities. We therefore hypothesised that the nestling’s optimal response would depend on the choices of other nestmates. A hungry individual could save energy by reducing activity levels, but then risks the loss of competitive abilities in confrontation with its nestmates that have chosen to stay active. Finally, we also expected that a particular decision would be reflected in the modulation of the activity of the digestive system. According to the ‘digestive adaptation paradigm’ (Karasov, 1990; Karasov, 1996; Starck, 1999), a hungry nestling can either increase the gut’s digestive capacity, compensating for food shortage by its more efficient processing or, alternatively, it can downregulate the gut’s activity to save energy.

Thirty-six, 4-day-old sand martin (Riparia riparia L., 1758) nestlings were collected in June 1999 from a breeding colony near Bial′ystok, northeast Poland. To avoid the possible effects of relatedness and hatching order, we did not use last-hatched young, collected no more than two nestlings from each nest, and always assigned them to different experimental groups. The nestlings were collected with permission from the nature conservancy authorities (permit no. 4201/296/99).

Nestlings were taken to the laboratory and placed in artificial nest cavities made of transparent plastic. The cavity consisted of a cylindrical nest chamber (12cm diameter, height 15cm) with the bottom modelled to mimic the nest cup, and the attached nest tunnel (a 30cm long tube, diameter 5cm). The cavities were surrounded with cardboard, such that the tunnel entrance was the only place through which light could enter. The experiment was carried out in a climatic chamber at 33°C, 90% relative air humidity and 16h:8h L:D photoperiod. These conditions were found in a pilot experiment to ensure normal growth and development of nestlings.

In the colony studied, typical broods consist of 4–5 asynchronously hatched nestlings. However, we created artificial broods of six nestlings, which allowed us to accommodate the broods within the limited space of our climatic chamber and reduce the workload of hand-feeding. We created six such broods and randomly assigned them to two, 3-day-long feeding regimens. Twelve nestlings (i.e. two whole broods, hereafter referred to as FR-nestlings) were fed every 45min between 08:00h and 20:00h with a limited amount of food. This resulted in growth retardation, of the magnitude frequently observed in the wild for undernourished nestlings (P. Brzęk, personal observations). In the remaining four broods, birds were assigned into two subgroups. In each of these broods, three young were fed until they stopped begging every 45min between 06:00h and 22:00h (hereafter referred to as fed ad libitum, AL nestlings), while their three nest-mates (hereafter referred to FR/AL nestlings) were fed according to the same feeding protocol as FR nestlings. Here, we attempted to simulate differences in food supply occurring in natural, asynchronously hatching broods facing food shortage. Under natural conditions, 1–2 nestlings hatch 1–2 days later than their nestmates and these individuals typically experience undernourishment when food is scarce (P. Brzęk, personal observations).

Each day, we weighed the nestlings, to the nearest 0.1g, to estimate daily body mass increments. Upon completion of the third day of experiments the metabolic rates of the young were measured. 18 (five FR, seven FR/AL and six AL nestlings) were subsequently killed; the remaining birds were returned to the breeding colony. During the course of the experiment, one of the FR nestlings was excluded from the trial because of abnormally low body temperature; this individual was returned to the natal nest.

Nestlings were hand-fed alternately with equal amounts of crickets and a special formula designed for the nestlings of insectivorous birds: fresh, soft cheese, glucose, rice flour and maize flour, mixed in the mass proportions 300:30:15:12 (J. Desselberger, personal communication). This mixture was subsequently added to hard-boiled hens’ eggs in the proportion 3:1 and enriched with the vitamin mixture Vitaral (Polfa, Poland). Fresh portions of food were prepared twice a day.

Daily food intake was measured individually for each nestling. Subsamples of food were dried at 60°C to determine water content, and energy content was determined in a Berthelot-type calorimeter. Regrettably, because of the high locomotor activity of the nestlings, we were unable to measure their individual daily faecal output.

Oxygen consumption was measured with a positive-pressure open-circuit respirometry system fitted with two-mass flow controllers (Sable Systems TR-1 oxygen analyser, Henderson, UT, USA). Depending on the mass of the nestlings, the air flow rate was set to 100 or 200mlmin⁻¹. The air temperature was maintained using a water bath at 33°C. The air stream vented from the metabolic chamber (100ml volume) was dried and scrubbed of CO₂ before measuring O₂ levels. For each individual nestling the trial lasted 1h, and measurements were collected during the second 30min of this period. We estimated the resting metabolic rate (RMR) by taking the lowest 4min value recorded. Metabolic data were analysed using Sable Systems DATACAN V software (Sable Systems, 1991). We calculated oxygen consumption rates using equation 4 of Hill (Hill, 1972). For estimation of RMR, we took the lowest 4min value that did not change by >0.01% in O₂ concentration.

To monitor the health of the nestlings, we measured cloacal temperature of each individual at the artificial nest three times a day to the nearest 0.1°C using a 0.5cm long thermocouple probe attached to a BAT-12 (Physitemp Instruments, USA) electronic thermometer. However, for the sake of brevity below, we report only the body temperatures for each nestling measured at the nest, after 2.5 days of experiments and at the end of the RMR trial.

Immediately after measurement of RMR, the nestlings were killed by overdose with ether, and their intestines and pectoral muscles were dissected out, cleared of adherent fat and weighed to the nearest 0.001g. Leg and pectoral muscles were also dissected and frozen for later analyses. The remaining carcasses were dried at 70°C to a constant mass, and then homogenised with an electric mill. Fat was extracted from homogenates with petroleum ether in a Soxhlet extractor. The residues were then redried, and the fat content was calculated as the mass lost during extraction (Sawicka-Kapusta, 1975).

We quantified the intestinal digestive activity as brush-border uptake of l-proline using the everted sleeve technique (Karasov and Diamond, 1983). We were not able to evert very tiny and fragile sleeves from the distal intestinal region. We therefore used sleeves from the proximal and middle regions only. However, some were torn during eversion. For this reason, the final sample size for the proximal region of the intestines of AL nestlings and the middle region of FR/AL young was five.

Briefly, small intestines were washed out with cold avian Ringer’s solution and cut into sections 1cm long. The sleeves were everted, mounted on metal rods and pre-incubated for 5min in oxygenated avian Ringer’s solution at 37°C. They were subsequently incubated for 2min at 37°C in Ringer’s solution containing 50mmoll⁻¹l-proline plus the same ³H-labelled nutrient as the radioactive tracer, and [¹⁴C]polyethyleneglycol as an adherent fluid marker. Sleeves were then removed from the rods, put into vials, weighed, solubilised with Soluene-350 (Packard Bioscience Company) mixed with scintillation cocktail (Hionic-Fluor, Packard Bioscience Company), and counted for both radiolabels. Uptake rate was expressed as nmoll-prolinemin⁻¹mg⁻¹ intestinal tissue.

The locomotor activity of the nestlings was monitored during the second and third day of the experiment, and quantified for each brood and feeding regime as the ratio of the number of feeding bouts when the young were found in the nest tunnel to the total number of bouts. Young found in the nest tunnel were always returned to the artificial nest cup upon completion of the feeding bout.

Within-individual values of all measured traits were highly autocorrelated between subsequent days of life (e.g. in the case of locomotor activity r=0.77, P<0.001) and, therefore, they were analysed for the whole 3-day period (except for behavioural observations, which were analysed for 2-day periods).

Many physiological and morphological variables (e.g. metabolic rate, size of internal organs) need to be corrected for body mass prior to comparisons. This is usually performed by means of analysis of covariance (ANCOVA), with body mass as a covariate. However, the almost twofold differences between body mass (BM) of nestlings fed ad libitum and food-restricted might potentially bias statistical comparisons, because the calculation of BM-adjusted means in ANCOVA would require extrapolation of within-treatment relationships between BM and the variable of interest outside their range of BM variation (e.g. Fig.1). This could confound the analysis (Cochran, 1957), since the grand mean of BM lies beyond the range of its within-treatment variation.

Unfortunately, to our knowledge there is no commonly accepted remedy to this problem. Thus, we restricted the use of ANCOVA to statistical comparisons between the food-restricted groups, which did not differ with respect to BM (Fig.1). To analyse the differences between nestlings food-restricted and fed ad libitum, we computed mass-specific values of observations for the traits potentially affected by BM and analysed them using a Mann–Whitney U-test. Although the use of ratios is rightly criticised (e.g. Packard and Boardman, 1987; Jasienski and Bazzaz, 1999), we think that in our case it allowed us to account satisfactorily for the effect of BM (Tracy and Sugar, 1989). In all statistical analyses, we subtracted the mass of the organ under consideration from BM, to avoid the possibility of autocorrelation.

For some variables (e.g. intestinal uptake rates of l-proline), we used either parametric analysis of variance (ANOVA) or, when a skewed distribution was detected, a non-parametric Kruskall–Wallis test (body temperature, locomotor activity). Accordingly, the significance of between-group differences was tested at P=0.05 by a least-significant difference test (LSD test) or pairwise Mann–Whitney U-test. In the case of multi-group comparisons using Mann–Whitney U-tests, the conventional P level of 0.05 was adjusted by applying a Bonferroni correction. To do this, we divided P=0.05 by the number of inter-group tests performed. All tests were carried out using the STATISTICA statistical package (StatSoft, Inc., 1997).

As expected, the different feeding regimens resulted in significant between-treatment differences in energy intake (ANOVA, F_2,32=41.8, P<0.001), which was significantly higher in AL nestlings than in the FR and FR/AL groups (LSD test, P<0.001 in both cases); the energy intake of the latter two treatment groups did not differ (LSD test, P=0.41; Table1).

Although the initial BM of the nestlings did not differ between all three groups (ANOVA, F_2,33=0.12, P=0.89), the high energy intake of AL nestlings resulted in their BM increments being larger than those observed in the food-restricted young (ANOVA, F_2,33=86.06, P<0.001; initial BM was not a significant covariate; LSD test, P<0.001 for both cases) (Table1; Fig.2). However, the comparable energy intake of FR/AL and FR nestlings was reflected by a lack of any difference in BM increments between the two groups (LSD test, P=0.52).

Feeding regime and the energy intake level of the nestmates significantly affected RMR. FR nestlings had significantly lower RMR than the FR/AL group (ANCOVA, F_1,16=5.96, P=0.027; BM was a significant covariate; Table1; Fig.3A). However, both food-restricted groups had a lower mass-specific RMR than AL young (Mann–Whitney test, for AL versus FR comparison: U=0, P<0.005; for AL versus FR/AL comparison: U=4, P<0.005; Table1; Fig.4A).

Nestlings’ body temperature measured at the artificial nest (after 2.5 days of experiment) differed significantly between treatment groups (Kruskal–Wallis test, H=18.07, P<0.001), (Table1; Fig.5). It was marginally higher in AL than in FR/AL nestlings (the difference did not reach statistical significance using a Mann–Whitney test: U=31, P=0.018). The body temperature of FR nestlings was lower than that of the other two treatment groups (Mann–Whitney test, for FR versus FR/AL nestlings: U=25, P=0.0067; for FR versus AL nestlings: U=6.5, P=0.00016).

We also detected between-group differences in body temperature measured following completion of the metabolic trials (Kruskal–Wallis test, H=13.78, P=0.001) (Table1). AL nestlings were characterised by a higher body temperature than both food-restricted groups (Mann–Whitney test, for AL versus FR group: U=0, P=0.0027; for AL versus FR/AL group: U=0, P=0.0012). There was no difference between FR and FR/AL nestlings (Mann–Whitney test, U=17, P=0.37).

Marked differences in energy intake and BM increments were also reflected in significant, relative differences in fat content normalised to lean dry body mass (LDBM). The AL group had significantly larger LDBM-specific fat stores than did both food-restricted groups (Mann–Whitney test, for AL versus FR group: U=0, P=0.006; for AL versus FR/AL group: U=0, P=0.0027; Table1; Fig.4B). Despite the similar energy intake of the FR and FR/AL nestlings, FR young had significantly smaller fat stores than the FR/AL group (ANCOVA, F_1,9=20.12, P<0.005; LDBM was a significant covariate; Table1; Fig.3B).

The food-restricted groups did not differ in pectoral muscle mass (ANCOVA, F_1,9=4.04, P=0.075; BM was a significant covariate; Table1; Fig.3C). However, AL nestlings had significantly higher BM-specific pectoral muscle mass than the other groups (Mann–Whitney test, for AL versus FR group: U=2, P=0.018; for AL versus FR/AL group: U=1, P<0.005; Fig.4C).

Mass-specific intestinal mass of the FR/AL group was significantly lower than that of AL nestlings (Mann–Whitney test, U=1, P=0.0043), whereas no difference was found between the FR and AL group (Mann–Whitney test, U=6, P=0.1; Table1; Fig.4D). However, when one outlying data point from the FR group (see Fig.4D) was omitted in the analysis, the latter difference became significant (Mann–Whitney test, U=0, P=0.01). By contrast, there was no difference in intestinal mass between FR and FR/AL nestlings (ANCOVA, F_1,9=0.52, P=0.49; BM was not a significant covariate; Table1; Fig.3D). This difference was still not significant when the outlying data point from the FR group was omitted from the computation.

Two-way ANOVA revealed that treatment group significantly affected intestinal uptake rate of proline (F_1,27=63.87, P<0.001), whereas the position along the intestinal length had no effect on uptake rates (F_1,27= 0.22, P=0.65). One-way ANOVA with treatment group as the main factor performed for the proximal intestine revealed that the proline uptake rate of the FR group was significantly higher than that of both the FR/AL and AL groups (LSD test, P<0.005 in both comparisons). The uptake rate of the FR/AL group was also significantly higher than that of AL nestlings (LSD test, P=0.008; Table1; Fig.6A). A similar analysis performed for the middle part of the intestines indicated that the uptake rates in FR nestlings were again significantly higher than in the FR/AL and AL groups (LSD test, P<0.001 in both cases), whereas the FR/AL and AL groups did not differ (LSD test, P=0.99; Table1; Fig.6B).

Locomotor activity differed significantly between treatment groups (Kruskal–Wallis test, H=20.03, P<0.001), (Table1; Fig.7). FR-nestlings were more active than both FR/AL (Mann–Whitney test, U=22, P=0.0039) and AL nestlings (Mann–Whitney test, U=0, P=0.00003). Locomotor activities of FR/AL and AL nestlings were not significantly different (Mann–Whitney test, U=38, P=0.05).

An association between begging, locomotor activity and within-brood competition for food has been found in the nestlings of many altricial species (Ryden and Bengtsson, 1980; Bengtsson and Ryden, 1981; Reed, 1981; Greig-Smith, 1985; McRae et al., 1993; Malacarne et al., 1994; Mock and Parker, 1997; Koelliker et al., 1998). However, their locomotor activity usually is limited to a few begging movements (such as gaping, stretching the neck or jostling for a particular position inside the nest cup), which are unlikely to incur high energy costs (McCarty, 1996; Bachman and Chappell, 1998). This limited activity is no match to the striking locomotor abilities of 5- to 6-day-old, still ectothermic sand martin nestlings observed in our study, which repeatedly crawled the 30cm distance within the tunnels of artificial nest burrows. Note that this distance travelled by the nestlings was limited by the length of the tunnel and, therefore, potentially did not reflect their full locomotory ability. Although we were not able to measure the energy expenditure of nestlings crawling towards the tunnel entrance, it seems safe to assume that they are substantial. Furthermore, it is clear that under natural conditions a nestling leaving the nest cup is exposed to increased heat loss, further increasing the possible costs of such behaviour.

Signalling models of parent-offspring conflict suggest that food-restricted nestlings should be more active than their satiated nestmates (Godfray, 1995). Indeed, locomotor activity of food-restricted nestlings was much higher than that of young fed ad libitum (Fig.7; Table1). This result strongly, although indirectly, suggests that the striking locomotor activity of sand martin nestlings is a means of monopolising feeding bouts by underfed young. However, by creating broods of FR nestlings and mixed broods of FR/AL and AL young, we also attempted to simulate the effect of within-brood competition resulting from different levels of hatching asynchrony. Despite similar food intakes and body mass increments, the locomotor activity of FR nestlings was significantly higher than in the FR/AL group (Fig.7; Table1). Thus, the presence of hungry, similar-sized siblings, rather than fully fed, bigger nestmates presumably created a competitive environment that stimulated the nestlings to show particularly high locomotor activity.

A possible explanation for this is that sibling competition would increase the survival prospects of the most mobile FR nestlings, competing against their nestmates of similar size. Increased locomotor activity, however, would be less beneficial for FR/AL young, since they have little chance of outcompeting larger/older siblings. FR/AL nestlings would benefit more from their reducing energy expenditure and thus maximising their survival prospects in the hope of an improvement in feeding conditions. This interpretation of our results supports Ricklefs’ suggestion that hatching asynchrony is an effective parental strategy to eliminate benefits to offspring from sibling competition (Ricklefs, 1993).

How can an individual nestling evaluate the level of nourishment of its nestmates? We cannot offer an unequivocal answer to this question, but we suggest that the high locomotor activity of FR broods was stimulated by the vigorous vocalisation of each individual in those broods. Although we did not quantify the level of vocalisation, it was noticeably less intense in FR/AL nestlings, presumably because of the lack of stimulation from their AL nestmates, which did not vocalise at all. Indeed, in some bird species, begging intensity appears to be stimulated by the begging of hungrier nestmates (Smith and Montgomerie, 1991; Price and Ydenberg, 1995). Particularly relevant is the behaviour of food-restricted tree swallow Tachycineta bicolor nestlings, whose begging intensity increases with the number of hungry competitors (Leonard et al., 2000). Furthermore, begging intensity increased in food-deprived broods but did not change in broods that contained some full-fed siblings (Leonard and Horn, 1998).

If high locomotor activity serves as a means for monopolising limited food resources, one would expect that it should be associated with digestive adaptations that enable the efficient utilisation of scarce food supplies. In principle, this can be achieved either by enlargement of the surface of the intestinal epithelium or by increasing the density of epithelial transporters, or both (Karasov, 1990; Karasov, 1996). In either case, enlargement of the intestinal epithelium should lead to increased intestinal mass. This was not found in our study; food-restricted nestlings had relatively lighter intestines than young fed ad libitum. Such a reduction, however, cannot be interpreted as downregulation of intestinal function, because the mass-specific intestinal uptake of l-proline was much higher in food-restricted nestlings than in nestlings fed ad libitum. This finding corroborates the results on song thrush (Turdus philomelos) nestlings (Konarzewski and Starck, 2000), who also reported higher intestinal uptake rates of l-proline in food-restricted, rather than in overfed, young. Furthermore, although the intestinal mass of FR/AL and FR young did not differ, the latter group was characterised by significantly higher uptake rates of l-proline. Interestingly, this difference was in the same direction as the difference in locomotor activity, possibly suggesting a functional link between the intensity of sibling competition and the digestive physiology of the nestlings.

The validity of our conclusions related to intestinal function clearly relies on the integrity of the tissue samples subject to the everted sleeve technique used to estimate uptake rates of l-proline. Starck et al. (Starck et al., 2000) have recently reported intestinal tissue damage resulting from handling of the samples during the everted sleeves method. In the present study, we were not able to evaluate histologically the effect of tissue damage on intestinal uptake rates, as suggested by Starck et al. (Starck et al., 2000). However, for the following reasons, we believe that our estimates of uptake rates are reliable. First, our most critical comparisons involved two groups of food-restricted nestlings subject to identical feeding regimes. Any bias due to tissue damage would therefore equally affect both groups. Second, the intestines of sand martin nestlings are of comparable size to the intestines of the smallest song thrush (Turdus philomelos) nestlings, in which the tissue damage was shown to have only second-order effects on intestinal histology compared with feeding regime (Konarzewski and Starck, 2000). Moreover, the intestinal proline uptake rates of young sand martins were very similar to those of the song thrushes, which would be unlikely to occur if the damage was extensive (Starck et al., 2000). Although we are confident that our findings represent reliable estimates of uptake rates of l-proline within our experimental design, caution must, however, be exercised if one wishes to use absolute values for between-species comparisons.

The high locomotor activity of FR nestlings was associated with their fat reserves being lower not only than those of AL young, but also of the FR/AL group. It is therefore likely that their low fat stores were due to the energetic expenditure incurred by crawling into the artificial nest tunnel, since energy intake did not differ between the two food-restricted groups.

Food restriction also appeared to be associated with reduced growth of the pectoral muscles. This is in contrast to other studies demonstrating that underfed altricial nestlings preferentially invest in pectoral muscles and other body parts that enhance the chances of successful fledgling (Donazar and Ceballos, 1989; Nilsson and Svensson, 1996; Konarzewski et al., 1996). However, young sand martin are characterised by a relatively long nestling period of approximately 22 days (Turner and Bryant, 1979). This long duration may mean that a delay in the growth of the pectoral muscles during the first week of life can be compensated for later, prior to fledgling. This, therefore, allows a reduction of energy expenditures in the early stages of post-hatching growth.

The nestlings subject to food restriction responded with a reduction in body temperature and RMR. A decrease in metabolic rate in response to food shortage has also been reported for the nestlings of another hirundinid aerial insectivore, the house martin (Delichon urbica; Prinzinger and Siedle, 1988). Young house martins are also capable of entering torpor when undernourished (Prinzinger and Siedle, 1988). The sand martin nestlings used in our experiment were partly poikilothermic (Marsh, 1979), whereas the ability to enter torpor requires strict control of body temperature. Furthermore, in our experiment they were reared at an ambient temperature of 33°C, which did not allow for conspicuous reduction of body temperature and metabolic rate. However, food-deprived song thrush Turdus philomelos and starling Sturnus vulgaris nestlings, reared under similar conditions to the present study, maintained body temperature and RMR at the same level as overfed young (Schew and Ricklefs, 1998; Konarzewski and Starck, 2000). Thus, unlike song thrush and starling nestlings, young sand martins responded to food shortage by reducing their body temperature and energy expenditure. Such an important difference suggests that under natural conditions, aerial insectivores face frequent, unpredictable fluctuations in food abundance that could select for mechanisms enhancing energy savings.

The high locomotor activity of FR nestlings is especially puzzling because of their reduced RMR and body temperature (Table1; Fig.3A, Fig.5). Strikingly, both RMR and body temperature (measured at the nest) were lower than those of the FR/AL group. The higher body temperature of FR/AL nestlings at the nest can possibly be attributed to the effect of huddling with warmer nestmates fed ad libitum. This is presumably why the difference in body temperature between FR and FR/AL young disappeared when measured immediately after 1h long metabolic trials. However, RMR was measured individually for each nestling, during exposure to the same temperature, and therefore the lower value in FR nestlings cannot be attributed to their lower temperature at the nest. Thus, the low body temperatures of FR nestlings measured at the nest were likely to be due to the reduced RMR, not vice versa. However, the magnitude of the reduction of RMR and body temperature in FR/AL and FR nestlings was inversely related to the satiation level of the nestmates and therefore, the level of their locomotor activity. Thus, at least under the present experimental conditions, a lowered metabolic rate did not impair the nestlings’ locomotor activity.

Energy costs of maintenance of the gut are an important component of RMR (Kersten and Piersma, 1987; Konarzewski et al., 2000). Food-restricted song thrush nestlings, whose growth was reduced to a similar extent as in the FR young of the present study, maintained both high intestinal uptake rates of l-proline and high RMR (Konarzewski and Starck, 2000). It is therefore surprising that food-restricted sand martins were able to downregulate RMR, and upregulate energy-dependent, intestinal uptake rates of l-proline. This suggests that the energy costs of such upregulation are relatively low.

In conclusion, our experiment demonstrated that undernourished nestlings reduced energy expenditure, upregulated digestive capacity and increased locomotor activity. These behavioural and physiological responses were clearly amplified by the presence of hungry siblings. We interpret such an amplification as an attempt to increase competitive abilities. This may lead to an escalation of indirect sibling competition among undernourished nestmates of similar size. It is therefore likely that developmental and behavioural plasticity played an important role in the evolution of sibling rivalry.

Our study would not have been finished without the help of Emilia Burak, Malgorzata Lewoc, Anna Roszkowska and Urszula Szymska. Andrzej Gebczyňski and Jan Taylor helped us skilfully in laboratory trials. Mariusz Cichoň, Kimberly A. Hammond, Wiliam H. Karasov, Regina McDevitt, Thenius Piersma, Robert Ricklefs, Jan Taylor, J. Matthias Starck and an anonymous reviewer made many helpful comments on the manuscript. The study was supported by Polish KBN grant 6 PO4F 054 14 to P.B. and KBN grant 6 P204 072 to M.K.