SEARCH

SEARCH BY CITATION

Keywords:

  • digestive constraint;
  • food intake rate;
  • molluscivory;
  • optimal foraging;
  • prey selection

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

1. Studies of diet choice usually assume maximization of energy intake. The well-known ‘contingency model’ (CM) additionally assumes that foraging animals only spend time searching or handling prey. Despite considerable empirical support, there are many foraging contexts in which the CM fails, but such cases were considered exceptions rather than the rule.

2. For animals constrained by the rate at which food is digested, CM does not necessarily lead to maximal energy intake rates because the time for digestion is not part of the selection criteria. In the main model developed to explain diet choice under a digestive constraint, the ‘digestive rate model’ (DRM), time lost to digestive breaks is minimized so that energy intake over total time (searching, handling, digestive breaks) is maximized.

3. It is increasingly acknowledged that most animals may face digestive constraints as prey capture rates vary over time and as it would be a waste to carry around heavy digestive machinery that can rapidly process food under all circumstances: this is only needed in times of high demand, provided that enough food can be found.

4. In molluscivore shorebirds ingesting hard-shelled prey such as red knots (Calidris canutus), the predictions of DRM were held up so far, whereas those of CM were rejected. However, most tests were carried out under controlled experimental conditions. Red knots overwinter in coastal areas over much of Western Europe and we capitalized on this variation by comparing, during a single winter, observed diet composition with predictions of DRM, CM and a null model assuming no prey selection (‘no-selection model’, NSM).

5. The observed diets were best predicted by DRM followed by CM. NSM poorly predicted observed diet choice. Under the present conditions, diet choice based on DRM would on average have yielded an energy intake rate twice as large as one based on CM. By adjusting the size of their gizzard (held constant in the present simulations), red knots could have lifted their energy intake rate further. We suggest that application of the DRM can help many diet studies forward, especially those previously seen as exceptions to the classical CM-based rule.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

It is usually assumed that natural selection ensures that free-living animals adopt strategies of food selection that minimize time and energy spent for a given quantity of food ingested (Stephens & Krebs 1986). The original, and still most frequently used, optimal prey-selection model is the so-called ‘contingency model’ (CM) (Charnov 1976; Stephens & Krebs 1986). CM assumes that prey items are selected according to the energy intake per unit handling time (an expression called profitability), i.e. encountered prey items are only included in the diet when their profitability exceeds the forager’s long-term energy intake rate. In their review, Sih & Christensen (2001) concluded that most empirical tests of this simple optimal diet choice model have achieved successful quantitative and qualitative fits. However, they show that particularly in herbivores, successful prey-selection models include additional criteria that account for nutritional and/or digestive constraints.

An animal that is able to find and capture prey items faster than it can process them is said to be ‘digestively bottlenecked’. Jeschke, Kopp & Tollrian (2002) propose that most foragers face a digestive constraint, whereas Verlinden & Wiley (1989), Hirakawa (1997) and Whelan & Brown (2005) present diet choice models that include both the handling and the digestion of prey. In these ‘digestive rate models’ (DRM), maximum intake rate is limited by rates of digestion and prey items are selected on the basis of digestive quality (i.e. energy content per unit indigestible bulk mass) rather than profitability. In this way, an animal can spend more time searching for items of higher digestive quality instead of losing time to digestive pauses. Nevertheless, the question of whether to use CM or DRM, even in digestively constrained animals, may depend on the time horizon over which energy intake is maximized. When digestively constrained foragers aim to maximize their long-term intake rate, they should ‘obey the rules’ of DRM, but the same animals should follow the CM in case they aim to maximize their intake rate during active foraging only (i.e. searching and handling), the so-called short-term or instantaneous intake rate (Farnsworth & Illius 1998; Bergman et al. 2001; Fortin, Fryxell & Pilote 2002). So-called ‘time-minimizers’ aim to maximize short-term intake rate, as they only require a certain amount of (daily) energy and aim to minimize the daily time devoted to collect (i.e. search and handle) this food (Schoener 1971).

A common molluscivore shorebird wintering in Western Europe (Van de Kam et al. 2004), the red knot Calidris canutus islandica, ingests its hard-shelled prey whole and is therefore easily digestively constrained (Van Gils et al. 2003a). Red knots have relatively large muscular gizzards (up to 10% of fresh body mass) and a similarly large intestine that is able to withstand the stress of rapidly passing shell fragments (Piersma, Koolhaas & Dekinga 1993b; Piersma, Gudmundsson & Lilliendahl 1999; Battley & Piersma 2004). It has been experimentally established that the maximum long-term food intake rate in red knots is determined by the rate at which shell mass can be processed (Van Gils et al. 2003a). Despite their trophic specialization (Piersma et al. 1998), red knots feed on a variety of molluscs and crustaceans that differ in profitability as well as digestive quality (Van Gils et al. 2005c). This variation makes it possible to distinguish between the predictions of CM and DRM (Verlinden & Wiley 1989; Van Gils et al. 2005b,c), and is the basis to explore whether overwintering red knots behave as long-term rate-maximizers or as time-minimizers.

So far, tests of diet choice criteria were based on a rather limited range of conditions, a range that certainly did not do justice to the variable foods encountered by red knots in Europe (Bocher et al. 2007). Here, we used data on food availability and diet choice collected during a single winter at eight feeding sites covering much of the European wintering range. We compared diet choice with predictions based on CM, DRM and a ‘no-selection-model’ (NSM; Fig. 1). Predicted intake rate realized over the total foraging time (searching, handling and digestive pauses) by the three models are also compared. We will discuss which costs and benefits and constraints induce overwintering red knots to adopt either a ‘time-minimization’ or ‘energy-maximization’ strategy, with special consideration of the temporal scale over which foraging decisions are made.

image

Figure 1.  Recapitulated scheme of hypothesis testing by confronting the outputs of the three diet-selection models with the observed diets. In the ‘digestive rate model’ prey is mainly selected on the basis of digestive quality (and only a little bit of the basis of profitability), while in the ‘contingency model’ prey is selected on the basis of profitability only, while the ‘no selection model’ predicts all available prey to be accepted. Characterizing prey type i by metabolizable energy content ei, ballast mass ki and handling time hi, then profitability equals ei/hi and digestive quality equals ei/ki.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Study sites

Study sites were chosen to cover as much as practically possible of the winter range of red knots along the European Atlantic coast. The British estuaries, the Dutch Wadden Sea and France harbour respectively 64%, 20% and 6% of the wintering population (Stroud et al. 2004). In the Dutch Wadden Sea, eastern England and the French coast north and south of Brittany, eight major feeding sites were selected (Fig. 2), the details of which are assembled in Table 1.

image

Figure 2.  Study sites along the European Atlantic coast, which have been selected such that they cover the large geographical scale of the red knots’ wintering distribution.

Table 1.   Study sites, their locations, periods of sampling, numbers of benthos and dropping samples, textural classification of sediments (Flemming 2000) and number of overwintering red knots between 2001/2002 and 2005/2006 (NL: SOVON unpublished data; UK: Musgrove et al. 2007; F: Mahéo 2001–2006)
AreaCoordinatesSampling periodsn samplesSedimentsn Red knots
SitesLatitudeLongitudeBenthosDroppingTextural classesMean ± SD
Dutch Wadden Sea
 Engelsmanplaat53°26′ N06°02′ E25 November to 27 November 20037920Sand45 413 ± 15 362
 Griend53°16′ N05°18′ E30 November to 4 December 200327721Sand
The Wash
 Stubborn Sand53°53′ N00°27′ E6 January to 12 January 200411722Slightly muddy sand83 098 ± 34 587
 Breast Sand53°50′ N00°17′ E8 January to 14 January 200426472Slightly muddy sand & Muddy sand
Mont-Saint-Michel Bay
 Cherrueix48°37′ N01°41′ W19 January to 27 January 200444027Slightly muddy sand5656 ± 2656
Pertuis Charentais
 Aiguillon Bay46°17′ N01°10′ W8 February to 8 March 200445921Slightly sandy mud16 315 ± 2050
 Moëze45°54′ N01°05′ W18 February to 22 March 200419612Slightly sandy mud
 Oléron45°55′ N01°12′ W21 February to 24 March 20042099Muddy sand

The Dutch Wadden Sea is a shallow coastal sea bordered by the Friesian barrier islands. Two major feeding grounds on sandflats were selected: (i) Griend, a core area in the western Wadden Sea (Piersma et al. 1993a); and (ii) Engelsmanplaat, 60 km eastward. The Wash is a large intertidal bay in eastern England, where we focused on two major feeding grounds: (iii) Breast Sand, a slightly muddy to muddy sandflat in the south; and (iv) Stubborn Sand, a slightly muddy sandflat in the eastern part of the bay (Goss-Custard, Jones & Newbery 1977). Mont-Saint-Michel Bay is the third main wintering ground for red knots in France (Deceuninck & Mahéo 2000). It is located along the North French Channel coast at the boundary between Brittany and Normandy. Feeding of red knots is concentrated in the south-western part of the bay (Le Dréan-Quénec’hdu 1999), where we established (v) Cherrueix, a slightly muddy sandflat study site. Pertuis Charentais et Breton, the southernmost wintering area of islandica-knots in Europe, is located between Loire and Gironde estuaries. In the north, (vi) Aiguillon Bay is the second main wintering area for knots in France (Deceuninck & Mahéo 2000). Moëze-Oléron Bay is situated 40 km south of Aiguillon Bay and is the most important French wintering area. This bay was subdivided into two feeding sites: (vii) Moëze, a bay with sandy mudflats on mainland coast; and (viii) Oléron, seagrass-covered flats bordering the east coast of Oléron island.

Benthos sampling and treatment

Study sites were sampled between 25 November 2003 and 24 March 2004 (Table 1). Macrofauna was sampled systematically at stations arranged in a grid of 250-m interval (Bocher et al. 2007; Kraan et al. 2007, 2009). Stations were localized using a handheld gps (Garmin 45 and 12; Garmin International, Lenexa, Kansas, USA) with WGS84 as the geographic coordinate system. Across all eight study sites, a total of 2041 stations were visited. Most sampling stations were reached by foot during low tide (70%), except for stations in Engelsmanplaat, Aiguillon Bay and Moëze, which were visited by boat during high tide. At each station reached by foot, we took one sediment core of 15 cm diameter down to a depth of 20–25 cm. The upper 4 cm was separated from the rest of the core to distinguish prey that were accessible from prey that were not accessible to red knots that can probe up to 4 cm with their 3·5-cm long bill (Zwarts & Blomert 1992). Subsequently, these two parts were sieved over a 1-mm mesh. Hydrobia ulvae, a small gastropod which is potentially present in very high densities, was sampled by taking a core of 7 cm diameter to a depth of 4 cm and subsequently sieved over a 0·5-mm mesh. At stations reached by boat, we took two cores of 10 cm diameter down to a depth of also 25–40 cm. One score was subsequently sieved over a 1-mm mesh and the other over a 0·5-mm mesh (the latter was used to also sample H. ulvae). As for the samples collected on foot, living molluscs were collected and stored in a freezer (−20 °C).

In the laboratory, molluscs were identified to species level, counted and size (length) was measured with a precision of 0·1 mm. After separating the flesh from the shelly part, each individual’s flesh ash-free dry mass (AFDMflesh) and shell dry mass (DMshell) was determined with a precision of 0·1 mg (Piersma et al. 1993a; Kraan et al. 2007).

Reconstruction of diet from dropping samples

Near the stations at which we sampled the macrofauna, we also collected droppings of red knots. Each dropping sample consisted of 25 droppings and was geo-referenced using a gps. In total, 204 dropping samples were collected across all study sites (Table 1). In the laboratory, diet was reconstructed from these samples following the protocol of Dekinga & Piersma (1993). Briefly, shell fragments of bivalves and gastropods were retrieved from the dried droppings and the shell lengths of prey ingested could be reconstructed using predictive allometric equations between shell length and hinge height (or width of first turn in the case of gastropods). The relative DMshell-contribution of each prey species in the diet followed from the weighted shell fragments sorted by species. Subsequently, the size- and species-specific AFDMflesh-to-DMshell ratios, determined per study site, were used to calculate the relative AFDMflesh-contribution of each prey species.

Modelling diet selection

We tested the CM using our field data on diet composition by considering each mm-length class of each species as a prey type i, each with a unique value for energetic contents ei (equivalent to AFDMflesh), handling time hi and encounter rate λi (being the product of type-specific searching efficiency and density). Searching efficiencies and handling times were similar to the ones used in Van Gils et al. (2005c) (for parameter details see Tables S1–S4). Prey selection for the CM is modelled by the standard algorithm (Charnov 1976) in which prey types are ranked in decreasing order of their profitability (ei/hi). In this order, prey types are then added to the diet until:

  • image( (eqn\, 1))

The first j types are included in the diet and all other types are excluded (the so-called ‘zero-one rule’; Stephens & Krebs 1986).

In addition to the parameters included in the CM, the DRM takes account of each prey type’s ballast mass ki (equivalent to DMshell). In this model, long-term energy intake rate is constrained by the long-term ballast intake rate, the latter cannot exceed a specific limit. Under this constraint, DRM predicts a selection based on the ranking of prey type in which digestive quality outweighs the importance of profitability, where digestive quality is expressed as the energy content per unit indigestible bulk mass (ei/ki). It is beyond the scope of this study to explain the mathematics underlying the optimal diet selection; we refer to Hirakawa (1995) or Van Gils et al. (2005c). However, Hirakawa (1995) developed a nice graphical procedure to explain the functioning of the DRM, which we have illustrated in Fig. 3.

image

Figure 3.  The optimal diet under a digestive constraint can be found by plotting for each prey type (given by a dot) the energy intake while handling this type (profitability ei/hi in Watts, scaled on vertical axis) against the ballast intake while handling this type (ki/hi in g s−1, scaled on horizontal axis). Prey types that lie above the optimal diet line (ODL) are selected (black-filled dots), while those under the ODL are ignored (white-filled dots; the one prey type in (a) on the ODL is partially selected; grey-filled dot). Given the characteristics of all prey types together (λi, hi, ei, ki), the balloon shape defines the boundaries within which a forager’s feasible long-term energy and ballast intake rate fall, depending on diet composition and in case there would be no digestive constraint. However, in this case there is a digestive constraint (upon the long-term ballast intake rate), and this is indicated by the vertical line, defined as C. Thus, the maximum long-term energy intake rate under this constraint is found where the vertical line intersects the upper boundary of the balloon. (a) In the DRM, the ODL is tangent to this intersection, and this yields the optimal diet that maximizes the long-term energy intake rate (asterisk). (b) In the CM, the ODL is tangent to the highest point in the balloon. Under a digestive constraint, this yields a suboptimal diet, leading to a suboptimal long-term energy intake rate (asterisk). (c) In the NSM, all prey types are accepted, and henceforth the ODL overlaps with the horizontal axis. This diet composition yields an even lower long-term energy intake rate (asterisk). In both (b) and (c), the triangle gives the maximum intake rate under the considered diet selection policy if there would be no digestive constraint. By taking digestive pauses to such extent that the long-term ballast intake rate no longer exceeds the digestive processing constraint (vertical line), the forager achieves its maximally feasible intake rate under a digestive constraint (asterisk).

Download figure to PowerPoint

In red knots, the limit upon long-term ballast intake rate goes up quadratically with increasing gizzard mass (Van Gils et al. 2003a). Here, we applied a gizzard mass of 9 g, which more or less represents the average gizzard mass found in red knots overwintering in Western Europe (Van Gils et al. 2003a, 2005a; Battley & Piersma 2004).

As a null-model, we included the case where foragers accepted each prey they encountered. In this so-called ‘no-selection model’ (NSM), diet composition (expressed as the relative contribution to the total amount of flesh mass consumed) simply depends on the encounter rate λi and the energetic contents ei of each prey type.

Because of the spatial heterogeneity in prey distribution within each study site, there will be some heterogeneity in diet selection as well. For this reason, we chose to analyse our benthos data at a spatial scale smaller than the scale of an entire site, and selected benthos samples that were collected within 500 m from a dropping sample as input into the different models. Thus, data from a maximum of nine nearest benthos sampling stations were used. Only available (i.e. accessible and ingestible; Zwarts & Blomert 1992) prey densities were used as input. Prey too large to be ingested were at least 15 mm long for Cerastoderma and at least 18 mm long for Macoma (Piersma et al. 1993a). Density, DMshell and AFDMflesh were consequently averaged per prey type (i.e. by species-specific 1-mm length classes) around each dropping sample. Therefore, the predictions were made specific for each dropping sample.

To test the goodness of fit of predictions to observations for each observation, we calculated the Euclidian distance across the three main prey species as (equivalent to the ‘root mean-squared deviation’ method proposed by Kobayashi & Salam 2000):

  • image( (eqn 2))

where Cer is the proportion of AFDMflesh in the diet observed (obs) or predicted (pred) for Cerastoderma edule, Hyd denotes Hydrobia ulvae and Mac denotes Macoma balthica. Note that we only concentrated on these three species as they made up 98% of the average diet, except for the Oléron site, where M. balthica was replaced by Scrobicularidae (as values for searching efficiency and handling times were not available for Scrobicularidae, we assumed them to be similar to the values in the more or less similarly shaped M. balthica). Euclidian distances were log-transformed to normalize them before testing the goodness of fit between models by paired t-tests.

We compared the long-term energy intake rates predicted by the three models, i.e. a situation in which the long-term energy intake rate is constrained by the specific limit (called C; Fig. 3) upon long-term ballast intake rate set by a 9-g gizzard. Thus, in those cases where a CM-diet or an NSM-diet would lead to a digestive constraint, the actual long-term energy intake rate (asterisks in Fig. 3) equals the ‘unconstrained’ long-term energy intake rate (small triangles in Fig. 3) multiplied by the ratio between C and the ‘unconstrained’ long-term ballast intake rate (this ratio thus represents the proportion of time that CM- and NSM-foragers can devote to active foraging; the rest of their time is lost to digestive breaks).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Most of the diet was made up of three mollusc species: the bivalves Baltic tellin M. balthica (average contribution to AFDM in observed diets ± SE: 20 ± 4%) and edible cockle C. edule (23 ± 4%) and the mudsnail H. ulvae (55 ± 4%) (Fig. 4: left most histograms ‘observed diet’). Two other bivalves, Scrobicularia plana and Abra tenuis (both Scrobicularidae) were only found in droppings from Oléron. Diet composition varied not only between sites, but also within sites (Fig. 4). Hydrobia dominated the diet at Moëze and Oléron with respectively 87 ± 5% and 80 ± 9%. On Engelsmanplaat, Breast Sand and Aiguillon Bay, Hydrobia contributed c. 70% of the estimated biomass intake (respectively 72% ± 10%, 69 ± 5%, 73 ± 24%). Macoma was the dominant prey at Cherrueix with 87 ± 5% of the energy intake. At Stubborn Sand, Cerastoderma dominated the diet (88 ± 9%) and in the Wadden Sea Cerastoderma and Hydrobia were selected in similar proportions with respectively 40 ± 14% and 54 ± 14% in the average diet.

image

Figure 4.  Comparison of observed diet against predicted diet for each of the three models on each of the eight study sites. In the histograms, the percentages refer to the relative contribution of AFDMflesh consumed of the following species: Cerastoderma edule (cer); Hydrobia ulvae (hyd); Macoma balthica (mac); Scrobicularidae (scr) is represented by Scrobicularia plana and Abra tenuis for the Oléron-site. Error bars represent SE, denoting the variation between dropping samples within each site. For the three models, bars are filled when a model shows the significantly shortest Euclidean distances to the observations, bars are not filled otherwise (Table 2). In the most right column, box plots represent long-term energy intake rate (mg AFDM/s) as predicted by the three prey-selection models. Bottom boundary of the box indicates the 25th percentile, line within box marks the median, and top boundary of the box indicates the 75th percentile. Errors bars above and below indicate 90th and 10th percentiles. Horizontal lines in the top of these graphs indicate significant differences (at the 0·05 level) between the intake rates predicted by the different models.

Across all study sites, DRM-based predictions of diet showed significantly shorter Euclidean distances to the observed diets than CM-based diets (average ± SE Euclidean distances: 0·23 ± 0·02 vs. 0·48 ± 0·04, t-test, < 0·0001; Table 2), while CM-based diets were closer to the observed diets than random (NSM) diets (0·48 ± 0·04 vs. 0·54 ± 0·03; t-test: = 0·009; Table 2). At six of eight sites, DRM-based predictions were significantly nearer to observed diets than CM-based predictions (t-test: Engelsmanplaat, = 0·027; Breast Sand, < 0 0·0001; Cherrueix, = 0·027; Aiguillon Bay, = 0·024; Moëze, = 0·035; Oléron, = 0·012). At Griend and Stubborn Sand, Euclidean distances to the true diets did not differ significantly between DRM and CM (t-tests, respectively = 0·371 and = 0·222). On Stubborn Sand, distances did not also differ significantly from those of the NSM prediction (t-tests: vs. DRM: = 0·178; vs. CM: = 0·378).

Table 2.   Sample sizes, means and standard errors of Euclidian distances between predictions of each of the three prey-selection models and diet observations on the eight study sites
Area sitesSample sizeDigestive rate modelContingency modelNo selection model
n Dropping locationsn Benthos stationsMeanSEMeanSEMeanSE
  1. Mean is printed bold when a model yielded the significantly lowest distance among the models (paired t-test).

All Sites733400·230·020·480·040·540·02
Dutch Wadden Sea
 Engelsmanplaat7330·300·070·680·100·610·07
 Griend9550·260·070·200·060·450·07
The Wash
 Stubborn sand7330·110·040·130·070·130·07
 Breast sand291070·270·050·580·040·610·03
Mont-Saint-Michel Bay
 Cherrueix9410·250·060·550·150·720·13
Pertuis Charentais
 Aiguillon Bay4240·160·060·560·160·680·12
 Moëze3130·120·040·660·120·450·14
 Oléron5340·100·030·360·070·320·06

On average, a diet based on DRM would permit an energy intake rate over total feeding time that is twice that obtained for a CM-instructed diet (t-test: < 0·0001; right most column of graphs in Fig. 4). A diet based on DRM would yield long-term intake rates as much as 2·5 times as those based on NSM (t-test: P < 0·0001). Also a CM-diet yields a 1·2 times higher long-term intake rate than an NSM-diet (t-test: P < 0·0001). Within each study site, except for Stubborn Sand, DRM-based diets yielded a significantly higher intake rate over total time than a CM-based and an NSM-based diet. At four sites (Griend, Breast Sand, Cherrueix and Oléron), a CM-based diet yielded a significantly higher long-term intake rate than an NSM-based diet.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

From the comparison between the observed diets and the predicted diets, we can conclude that the DRM best predicted diet composition, followed by the CM, which in turn was followed by the NSM. Given that red knots are digestively constrained most of the time (for a knot with a gizzard of 9 g, we predict a digestive bottleneck at 98% of the sampling stations; Fig. 5a), adherence to the rules of the DRM maximized their long-term intake rate.

image

Figure 5.  (a) Histogram giving the percentage of cases where knots would face a digestive constraint if they would follow the long-term rate-maximizing prey-selection policy and how this varies as a function of gizzard size (i.e. the percentage of sampling stations where obeying the CM would not maximize the long-term energy intake rate). With an increase in gizzard size, the digestive constraint alleviates. (b) Mean Euclidean distances Eobs–pred (±SE) between the observed diet and the predicted DRM-diet (dots), the CM-diet (triangles) and the NSM-diet (crosses). Error bars represent SE, denoting the variation between dropping samples. The DRM performs closest to the actual data at a modelled gizzard mass of 9–10 g, which is the size mostly observed in overwintering islandica-knots.

Although the average knot in winter carries a gizzard of c. 9 g, there is much inter-individual variation, certainly at the vast spatial scale of W-Europe (Piersma et al. 1993b; Van Gils et al. 2003a; Dietz & Piersma 2007). Therefore, we performed a sensitivity analysis to explore the effect of gizzard size on the goodness of fit of the DRM model. We did so by varying gizzard size between 1 and 20 g (which covers the natural range in gizzard masses; M.W. Dietz, A. Dekinga & T. Piersma, unpublished data) in steps of 1 g (Fig. 5). This shows that, not surprisingly, the percentage of sampling stations at which knots would face a digestive constraint fell from 100% (1 g gizzard) to 51% (20 g gizzard; Fig. 5a). More importantly and encouragingly, based on the analysis of the Euclidian distances, the DRM performs best at a modelled gizzard mass of 9–10 g (Fig. 5b), the most frequently observed values in overwintering islandica-knots. This is consistent with our conclusion that a DRM-based diet best reflects reality. For gizzard sizes larger than 9–10 g, a DRM diet starts to mimic a CM diet (relating this to the conceptual plot in Fig. 3: the vertical line C indicating the digestive constraint would shift to the right and thus the optimal diet line ODL would become more horizontal). DRM foragers with a gizzard size smaller than 9–10 g would only select the highest quality prey. At very tiny gizzards (2 g), a DRM-based diet would be so narrow and specialized that the DRM would be outcompeted in performance by the CM, and even by the NSM (with respect to Fig. 3: the digestive constraint would shift to the left and thus the optimal diet line ODL would have a very steep slope). We also analysed the sensitivity of the model output to variation in the other parameters (handling time, energy content, shell mass and encounter rate). As shown in Fig. S1, the DRM performed best across the entire range of parameter values studied (followed, in almost all cases, by the CM).

The majority of foraging studies have obtained data consistent with long-term rate-maximization (Stephens & Krebs 1986; Sih & Christensen 2001), and our results are no exception. However, as overwintering knots only require a certain amount of energy per day to maintain energy balance, a priori we may have expected overwintering knots to be time-minimizers, in which case they should have obeyed the CM-based rules. Also with respect to the observed gizzard sizes in winter, we might expect a time-minimizing strategy: the observed 9–10 g gizzard seems to be the size required to maintain energy balance; only during late winter/spring, when fuelling before commencing their non-stop long-distance flights, do knots further enlarge their gizzard such that the rate of storing fuel is maximized (Van Gils et al. 2003a, 2005a). As digestive breaks are generally not mutually exclusive with other activities (Fortin, Boyce & Merrill 2004; Van Gils & Piersma 2004), why did knots then not follow a strategy that would minimize their time devoted to active foraging, i.e. searching and handling, such that the remaining time can be spent to other useful activities?

We propose that overwintering knots do not actually opt to minimize active foraging time (searching and handling time), but opt to minimize total foraging time instead (i.e. active foraging plus digestive breaks). This is achieved by maximizing long-term intake rate (i.e. intake over searching, handling and digestive breaks), thus by adhering to the DRM. As knots have high digestive turnover rates (median food retention time = 56 min, average transit time = 25 min; Van Gils 2004, pp. 293–294), digestive breaks cannot be postponed until roosting at high tide (such as in oystercatchers Haematopus ostralegus; Kersten & Visser 1996). Instead, knots have multiple short digestive breaks in between active foraging bouts (e.g. these multiple short breaks amounted to 80% of total foraging time in Van Gils et al. 2003b). Thus, the effect of minimizing total foraging time for a knot is that total time on the mudflat is minimized, whereby daily time on the roost is maximized. This may be a relatively safe option due to risk dilution and shared predator detection (Clark & Mangel 1986; Cresswell 1994; Fernandez-Juricic, Siller & Kacelnik 2004; Caro 2005) as flock sizes are largest on the roost (Piersma et al. 1993a; Van den Hout, Spaans & Piersma 2008), in addition to thermostatic costs being relatively low on roosts (Wiersma & Piersma 1994). We have strong indications that knots aim to maximize their roosting time: daily time off the roost decreased strongly as a function of gizzard mass, from 16 h (4 g gizzard) to 9 h (12 g; Van Gils et al. 2005b), and it has been suggested earlier that just-arrived knots grow somewhat larger gizzards to minimize the total foraging time (Van Gils et al. 2007).

The implications of our study may be wide-ranging. Digestive constraints are faced by many more species than just red knots. It has been suggested that rates of digestion, rather than rates of encounter and handling, delimit the intake rate in many foragers during much of their live (Jeschke et al. 2002). And this makes much sense from an optimization point of view. Food encounter rates vary widely in time and space, and it would be costly and wasteful to maintain and carry large digestive organs (e.g. Piersma et al. 2003) that can process food faster than any rate at which food is encountered and ingested. Rather, maintaining a somewhat smaller digestive machinery is cheaper and outweighs the costs of facing regular digestive constraints. This study shows that the costs of facing a digestive constraint can be kept to a minimum when sticking to the rules of the digestive rate model.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The sampling of the eight study sites was only possible with the precious help of numerous people from the three involved countries. We would like to thank the staff of ONCFS for providing additional help: H. Audebert, M. Claise, T. Dodin, A. François, D. Gaillard, Y. Limouzin, P. Mallassage, J. Marquis, J. Moreau and G. Puaud. We thank D. Lamoise and M. Allain for their participation in the sorting of droppings. Logistic help, including the provision of Weld accommodation, was generously arranged or provided by P. L. Ireland, P. Atkinson, N. Alligner, N. Clark, E. Feunteun and P. Miramand. We are especially grateful to the managers of the nature reserves in Aiguillon Bay, F. Meunier of the Ligue pour la Protection des Oiseaux (LPO) and E. Joyeux of the Office National de la Chasse et de la Faune Sauvage (ONCFS), as well as the manager of Moeze-Oléron, P. Delaporte of LPO. Financial support was received from the Conseil Général de Charente-Maritime, the Programme Environnement, Vie et Société CNRS Micropolluants Marennes–Oléron, Zone Atelier du Mont Saint-Michel (PEVS CNRS), Réserve Naturelle de la Baie de l’Aiguillon (LPO) and the bilateral Van Gogh programme administered by the Netherlands Organisation for Scientific Research (NWO) and the French Ministry of Foreign Affairs. Finally, we thank the editors and two anonymous reviewers for important feedback on the study.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Battley, P.F. & Piersma, T. (2004) Adaptive interplay between feeding ecology and features of the digestive tract in birds. Physiological and Ecological Adaptations to Feeding in Vertebrates (eds J.M.Starck & T.Wang), pp. 201228. Science Publishers, Enfield, New Hampshire, USA.
  • Bergman, C.M., Fryxell, J.M., Gates, C.C. & Fortin, D. (2001) Ungulate foraging strategies: energy maximizing or time minimizing? Journal of Animal Ecology, 70, 289300.
  • Bocher, P., Piersma, T., Dekinga, A., Kraan, C., Yates, M.G., Guyot, T., Folmer, E.O. & Radenac, G. (2007) Site- and species-specific distribution patterns of molluscs at five intertidal soft-sediment areas in northwest Europe during a single winter. Marine Biology, 151, 577594.
  • Caro, T. (2005) Antipredator Defenses in Birds and Mammals. Chicago University Press, Chicago.
  • Charnov, E.L. (1976) Optimal foraging: attack strategy of a mantid. The American Naturalist, 110, 141151.
  • Clark, C.W. & Mangel, M. (1986) The evolutionary advantages of group foraging. Theoretical Population Biology, 30, 4575.
  • Cresswell, W. (1994) Flocking is an effective anti-predation strategy in redshanks, Tringa totanus. Animal Behaviour, 47, 433442.
  • Deceuninck, B. & Mahéo, R. (2000) Synthèse des dénombrements et analyse des tendances des limicoles hivernants en France 1978–1999, pp. 83. LPO – Wetlands International, Paris, France.
  • Dekinga, A. & Piersma, T. (1993) Reconstructing diet composition on the basis of feces in a mollusk-eating wader, the knot Calidris canutus. Bird Study, 40, 144156.
  • Dietz, M.W. & Piersma, T. (2007) Red knots give up flight capacity and defend food processing capacity during winter starvation. Functional Ecology, 21, 899904.
  • Farnsworth, K.D. & Illius, A.W. (1998) Optimal diet choice for large herbivores: an extended contingency model. Functional Ecology, 12, 7481.
  • Fernandez-Juricic, E., Siller, S. & Kacelnik, A. (2004) Flock density, social foraging, and scanning: an experiment with starlings. Behavioral Ecology, 15, 371379.
  • Flemming, B.W. (2000) A revised textural classification of gravel-free muddy sediments on the basis of ternary diagrams. Continental Shelf Research, 20, 11251137.
  • Fortin, D., Fryxell, J.M. & Pilote, R. (2002) The temporal scale of foraging decisions in bison. Ecology, 83, 970982.
  • Fortin, D., Boyce, M.S. & Merrill, E.H. (2004) Multi-tasking by mammalian herbivores: overlapping processes during foraging. Ecology, 85, 23122322.
  • Goss-Custard, J.D., Jones, R.E. & Newbery, P.E. (1977) The ecology of the Wash. I. Distribution and diet of wading birds (Charadrii). Journal of Applied Ecology, 14, 681700.
  • Hirakawa, H. (1995) Diet optimization with a nutrient or toxin constraint. Theoretical Population Biology, 47, 331346.
  • Hirakawa, H. (1997) Digestion-constrained optimal foraging in generalist mammalian herbivores. Oikos, 78, 3747.
  • Jeschke, J.M., Kopp, M. & Tollrian, R. (2002) Predator functional responses: discriminating between handling and digesting prey. Ecological Monographs, 72, 95112.
  • Kersten, M. & Visser, W. (1996) The rate of food processing in Oystercatchers: food intake and energy expenditure constrained by a digestive bottleneck. Functional Ecology, 10, 440448.
  • Kobayashi, K. & Salam, M.U. (2000) Comparing simulated and measured values using mean squared deviation and its components. Agronomy Journal, 92, 345352.
  • Kraan, C., Piersma, T., Dekinga, A., Koolhaas, A. & Van der Meer, J. (2007) Dredging for edible cockles (Cerastoderma edule) on intertidal flats: short-term consequences of fisher patch-choice decisions for target and non-target benthic fauna. ICES Journal of Marine Science, 64, 17351742.
  • Kraan, C., Van Gils, J.A., Spaans, B., Dekinga, A., Bijleveld, A.I., Van Roomen, M., Kleefstra, R. & Piersma, T. (2009) Landscape-scale experiment demonstrates that Wadden Sea intertidal flats are used to capacity by molluscivore migrant shorebirds. Journal of Animal Ecology; in press. doi: DOI: 10.1111/j.1365-2656.2009.01564.x
  • Le Dréan-Quénec’hdu, S. (1999) Paramètres influençant la répartition des limicoles: sédiments et parasites. PhD thesis, University of Rennes I – Natural History Museum, Rennes, France.
  • Mahéo, R. (20012006) Limicoles séjournant en France (Littoral). Rapports Wetlands International/ONCFS, Paris, France.
  • Musgrove, A.J., Collier, M.P., Banks, A.N., Calbrade, N.A., Hearn, R.D. & Austin, G.E. (2007) Waterbirds in the UK 2005/06: The Wetland Bird Survey. BTO/WWT/RSPB/JNCC, Thetford.
  • Piersma, T., Hoekstra, R., Dekinga, A., Koolhaas, A., Wolf, P., Battley, P. & Wiersma, P. (1993a) Scale and intensity of intertidal habitat use by knots Calidris canutus in the Western Wadden Sea in relation to food, friends and foes. Netherlands Journal of Sea Research, 31, 331357.
  • Piersma, T., Koolhaas, A. & Dekinga, A. (1993b) Interactions between stomach structure and diet choice in shorebirds. Auk, 110, 552564.
  • Piersma, T., Van Aelst, R., Kurk, K., Berkhoudt, H. & Maas, L.R.M. (1998) A new pressure sensory mechanism for prey detection in birds: the use of principles of seabed dynamics? Proceedings of the Royal Society of London. Series B: Biological Sciences, 265, 13771383.
  • Piersma, T., Gudmundsson, G.A. & Lilliendahl, K. (1999) Rapid changes in the size of different functional organ and muscle groups during refueling in a long-distance migrating shorebird. Physiological and Biochemical Zoology, 72, 405415.
  • Piersma, T., Dekinga, A., Van Gils, J.A., Achterkamp, B. & Visser, G.H. (2003) Cost-benefit analysis of mollusc-eating in a shorebird. I. Foraging and processing costs estimated by the doubly labeled water method. Journal of Experimental Biology, 206, 33613368.
  • Schoener, T.W. (1971) Theory of feeding strategies. Annual Review of Ecology and Systematics, 2, 369404.
  • Sih, A. & Christensen, B. (2001) Optimal diet theory: when does it work, and when and why does it fail? Animal Behaviour, 61, 379390.
  • Stephens, D.W. & Krebs, J.R. (1986) Foraging Theory. Princeton University Press, Princeton.
  • Stroud, D.A., Davidson, N.C., West, R., Scott, D.A., Haanstra, L., Thorup, P., Ganter, B. & Delany, S. (2004) Status of migratory wader populations in Africa and Western Eurasia in the 1990s. International Wader Studies, 15, 1259.
  • Van de Kam, J., Ens, B.J., Piersma, T. & Zwarts, L. (2004) Shorebirds: An Illustrated Behavioural Ecology. KNNV Publishers, Utrecht.
  • Van den Hout, P.J., Spaans, B. & Piersma, T. (2008) Differential mortality of wintering shorebirds on the Banc d’Arguin, Mauritania, due to predation by large falcons. Ibis, 150, S219S230.
  • Van Gils, J.A. (2004) Foraging Decisions in Digestively Constrained Long-Distance Migrant, the red Knot (Calidris canutus). PhD thesis, University of Groningen, Groningen, The Netherlands.
  • Van Gils, J.A. & Piersma, T. (2004) Digestively constrained predators evade the cost of interference competition. Journal of Animal Ecology, 73, 386398.
  • Van Gils, J.A., Piersma, T., Dekinga, A. & Dietz, M.W. (2003a) Cost-benefit analysis of mollusc-eating in a shorebird. II. Optimizing gizzard size in the face of seasonal demands. Journal of Experimental Biology, 206, 33693380.
  • Van Gils, J.A., Schenk, I.W., Bos, O. & Piersma, T. (2003b) Incompletely informed shorebirds that face a digestive constraint maximize net energy gain when exploiting patches. American Naturalist, 161, 777793.
  • Van Gils, J.A., Battley, P.F., Piersma, T. & Drent, R. (2005a) Reinterpretation of gizzard sizes of red knots world-wide emphasises overriding importance of prey quality at migratory stopover sites. Proceedings of the Royal Society of London. Series B: Biological Sciences, 272, 26092618.
  • Van Gils, J.A., Dekinga, A., Spaans, B., Vahl, W.K. & Piersma, T. (2005b) Digestive bottleneck affects foraging decisions in red knots Calidris canutus. II. Patch choice and length of working day. Journal of Animal Ecology, 74, 120130.
  • Van Gils, J.A., De Rooij, S.R., Van Belle, J., Van der Meer, J., Dekinga, A., Piersma, T. & Drent, R. (2005c) Digestive bottleneck affects foraging decisions in red knots Calidris canutus. I. Prey choice. Journal of Animal Ecology, 74, 105119.
  • Van Gils, J.A., Dekinga, A., Van den Hout, P.J., Spaans, B. & Piersma, T. (2007) Digestive organ size and behavior of red knots (Calidris canutus) indicate the quality of their benthic food stocks. Israel Journal of Ecology & Evolution, 53, 329346.
  • Verlinden, C. & Wiley, R.H. (1989) The constraints of digestive rate – an alternative model of diet selection. Evolutionary Ecology, 3, 264273.
  • Whelan, C.J. & Brown, J.S. (2005) Optimal foraging and gut constraints: reconciling two schools of thought. Oikos, 110, 481496.
  • Wiersma, P. & Piersma, T. (1994) Effects of microhabitat, flocking, climate and migratory goal on energy-expenditure in the annual cycle of red knots. Condor, 96, 257279.
  • Zwarts, L. & Blomert, A.-M. (1992) Why knots Calidris canutus take medium-sized Macoma balthica when six prey species are available. Marine Ecology – Progress Series, 83, 113128.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Fig. S1. Parameter-sensitivity of mean Euclidean distances between the observed diet and the predicted DRM diet, the CM diet and the NSM diet for each prey species

Table S1. Estimates of handling time and how they relate to shell length

Table S2. Estimates of searching efficiency

Table S3. Densities and shell lengths of available prey within 500 m of dropping samples

Table S4. Observed relationships for flesh mass and ballast mass as a function of shell length

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JANE_1608_sm_Supmat.doc517KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.