Variation of parental and chick diet in opportunistic insectivorous European Bee-eaters

1.   Introduction

Dietary opportunism is a common strategy adopted by many bird species (e.g., Post and Greenlaw, 2006; Sherry et al., 2016; Buelow et al., 2018). Foragers are considered opportunists when the composition of diet reflects the local availability of their prey. Within the same species, an opportunistic diet can vary in space and time, being influenced by e.g., prey phenology (Naef-Daenzer et al., 2000; Post and Greenlaw, 2006), weather conditions and geographic locations (Duijns et al., 2013) and habitat type (Tsachalidis and Goutner, 2002). Habitat is a very strong predictor for opportunistic insectivores’ diet, as prey abundance is often associated with vegetation diversity and land-use characteristics (e.g., Di Maggio et al., 2018). Additionally, diet selectivity by avoidance of certain prey species may also influence diet composition (Law et al., 2017). Individuals often prefer the most profitable prey, and during the breeding season in particular, birds may select prey to fulfil the high energetic requirements associated with reproduction (Naef-Daenzer et al., 2000). For example, males often provide larger items to females than what they consume before and during egg-laying (e.g., Avery et al., 1988). Chick growth is similarly energetically demanding, and the provision of larger prey to chicks than what adult breeders consume seems to occur often (e.g., Kaspari and Joern, 1993; Radford, 2008). For example, Great Tits (Parus major) provide chicks with larger caterpillars as soon as they become available, independently of caterpillar abundance (Naef-Daenzer et al., 2000); and Pied Flycatchers (Ficedula hypoleuca) feed larger prey over time to chicks during growth (Wiebe and Slagsvold, 2014).

The European Bee-eater (Merops apiaster, hereafter bee-eater) is an insectivorous land-bird preying mainly on medium to large-sized flying insects, with a marked preference for Hymenoptera. Nevertheless, its diet composition is quite broad (Bastian and Bastian, 2023) like in many avian insectivores (e.g., Post and Greenlaw, 2006; Bellavance et al., 2018). Bee-eater's preference for hymenopterans is clear, as this order seldom comprises less than 50% of its diet (e.g., Kristin, 1994; Kossenko and Fry, 1998; Aissaoui-Marniche et al., 2007; Farinós-Celdrán et al., 2016). Even when the availability of hymenopterans is low, bee-eaters retain this order as the main food source (Inglisa et al., 1993). Odonata and Coleoptera are commonly the second most important general prey groups in bee-eaters’ diet with frequencies varying between 7–17% and 11–58%, respectively (Costa, 1991; Inglisa et al., 1993; Kossenko and Fry, 1998; Fuisz et al., 2013; Arbeiter et al., 2014; Farinós-Celdrán et al., 2016; Lourenço, 2018), whereas Lepidoptera being recorded as occasionally important in several cases (10–12%; Krüger, 2018; Lourenço, 2018). Other insect orders, e.g., Hemiptera, Diptera and Orthoptera, typically represent less than 10% of the diet.

Several studies have reported seasonal changes in diet of breeding bee-eaters (Inglisa et al., 1993; Kristin, 1994; Arbeiter et al., 2014; Krüger, 2018). For example, bee-eaters can take advantage of the peak abundance of a particular prey (e.g., Dung Beetles, Krüger, 2018) or shift diet composition, with adults increasingly consuming bees and taking fewer dragonflies during the chick rearing period (Arbeiter et al., 2014). Other studies have reported differences in the type of prey consumed by adults and chicks (Kristin, 1994; Krüger, 2018), and there is evidence that bee-eater chicks take larger prey than adults (Massa and Rizzo, 2002; Arbeiter et al., 2014). But so far, there is no information about differences in prey selectivity between what adult bee-eaters consume and what is provided to chicks. Potential differences between adult and chick diet in terms of item size and type of prey may be related to nutritional requirements of chicks, as the nutritional quality can play an important role in prey selection of insectivores (Razeng and Watson, 2015). In altricial birds, parental choice of food provided to chicks can strongly influence chick survival and/or the condition at fledging (e.g., Wright et al., 2009). Interestingly, Krebs and Avery (1984) observed that bee-eaters do not feed exclusively on the most profitable prey and showed experimentally that chicks grow more efficiently on a mixed diet of bees and dragonflies than on a strict diet of either alone. Moreover, at early developmental stages, chicks may only be able to digest small and soft food items (Moser, 1986; Orłowski et al., 2015). The total energy demand increases when chicks grow, so it seems more efficient to meet those higher demands with larger prey items. Although several species have been reported to increase the size of prey provided to chicks across the development period (e.g., Wiebe and Slagsvold, 2014; Orłowski et al., 2015), so far, only Krüger (2018) reported differences in prey mass throughout chick rearing period of bee-eaters.

Being a widespread species in the Western Palearctic, bee-eaters breed in colonies surrounded by a variety of habitats, but forage in open landscapes like grasslands, cultivated land or oak woods (Fry, 1984). In the Mediterranean Basin, they catch prey at a mean distance of 850 m from the nest, foraging preferentially in oak forests and riparian areas (Universidad de Extremadura, 2006). Although diet composition of bee-eaters is well studied in several parts of its range (e.g., Inglisa et al., 1993; Kossenko and Fry, 1998; Fuisz et al., 2013; Arbeiter et al., 2014; Bastian and Bastian, 2023), diet studies are still scarce in the Iberian Peninsula (but see Herrera and Ramirez, 1974; Costa, 1991; Farinós-Celdrán et al., 2016; Lourenço, 2018) and none has yet assessed diet variation between habitats and age classes (i.e., adults and chicks) in this region. In this study we (1) first assess the variation in the diet of adult bee-eaters between colonies located in distinct habitats and (2) explore differences between diet composition of chicks and adults. In addition, we (3) explore the variation of prey size provided to chicks throughout development; and investigate (4) how diet selection of adults varies between two periods of the breeding season: before incubation and during chick rearing.

2.   Materials and methods

2.1   Study area and habitat characterization

The study was carried out between April and July of 2016 in Portugal. We selected five bee-eater colonies surrounded by different habitats (Table 1). In each colony, we mapped the predominant habitat (habitat that occupied the larger area) and one or two secondary habitats (for more details see below and Table 1). Because bee-eaters catch prey in the vicinity of the colonies during breeding season (Fry, 1984), we considered the area within an 850-m radius from the colony as being representative of the average foraging range (Universidad de Extremadura, 2006). We used Google Earth satellite images to create a digital habitat map of each study area. All habitats within an 850-m radius from each colony were mapped using software QGIS version 2.18.24, checked and confirmed by ground observation. We calculated the proportional area occupied by each habitat and classified each colony according to its predominant habitat (see Table 1) as Oak–shrubland—oak woodland with understory and including a Rockrose (Cistus ladanifer) shrubland area, Oak Forest—cork and holm oak forest with no understory, Pasture—irrigated pasture, Mixed forest—managed mixed forest surrounded by an urban area, Meadow—meadow patches within a rural area.

Table  1.  Description of the bee-eater colonies sampled, indicating its location and the type and proportion of the main and secondary habitats (when available), as well as a description of the land-use characteristics surrounding each colony.

Colony		Oak−shrubland	Oak forest	Pasture	Mixed forest	Meadow
Location		38.1° N, 7.0°W	38.7° N, 8.8° W	39.8° N, 7.1° W	38.6° N, 9.1° W	38.6° N, 8.9° W
Main habitat	Type	Cork and holm oak woodland	Cork and holm oak forest	Meadow	Mixed forest – pine, eucalyptus and oaks	Meadow
	Area (%)	88.98%	100%	94.24%	100%	94.24%
Secondary habitat 1	Type	Rockrose Shrubland (Cistus ladanifer)		mixed forest – pine and eucalyptus		mixed forest – pine and eucalyptus
	Area (%)	8.87%		5.76%		5.76%
Secondary habitat 2	Type	Streams
	Area (%)	2.15%
Land-use		Extensive rotational grazing by cattle. Understory was always present in several patches during the study period.	Extensive grazing by cattle. No understory during the study period.	Extensive rotational grazing by sheep. Pasture was irrigated during spring. Understory was always present in several patches during the study period.	Managed forest surrounded by a large urban area and used as leisure space. The area covered by buildings was not considered as potential foraging habitat.	Meadow patches surrounded by a rural area. Hay harvested between May and June. The area covered by buildings was not considered as potential foraging habitat.

 | Show Table

DownLoad: CSV

2.2   Diet sampling and reconstruction

Diet remains contained in adult bee-eaters’ pellets were sampled at each colony during three periods: pre-incubation, incubation and chick rearing period. Pre-incubation period was defined as the time between the arrival of the first birds at the colony and the date of the first egg-laying. Incubation period was defined as the time from the first egg-laying until the first hatching, and chick rearing period as the time from the first hatching until the first chick fledges. Arrival of the first birds was determined using a sensor-activated trail camera (model NUM’AXES 1027) placed at each colony monitoring the main perches well before birds arrive (10–20 days). Egg-laying dates were back-calculated from hatching dates (assuming 20 days of incubation), which was determined by nest inspection (Costa et al., 2020a).

Diet composition of adults was determined through the analysis of regurgitated pellets collected below perches, which had been intentionally placed in the colonies. Because birds from different nests frequently used the same perch, it was not possible to assign the collected pellets to any specific nest. At each colony, we collected several pellets every week, from arrival of the first birds to the end of the breeding season. We collected only intact pellets while ensuring these were freshly regurgitated. After each sampling session, we removed the remaining pellets and prey remains below the perches to avoid any mixing of pellets from previous sampling events. Prior to analysis (see below), we randomly selected ten pellets from each colony and period (totalling 30 pellets per colony).

To assess diet composition of chicks, we collected the prey remains (i.e., regurgitated pellets) from nests chambers after the chicks had fledged, using a portable vacuum cleaner connected to a 2-m hose. Prey remains from each nest were thus vacuumed into individual bags. Although some prey remains from the nest may belong to adults that regurgitated inside the nests during incubation, given the accumulation of a large amount of fresh remains and the trampling of the earliest remains by the chicks, we considered that such potential contamination was negligible. Additionally, prey remains from previous years were never found in a reused nest as adults clean the nest chamber before egg laying. From colonies in Oak−shrubland and Pasture, we sampled the content of ten nests. From colonies in Meadow, Mixed forest, and Oak forest habitats, we sampled seven, four and three nests, respectively. To standardise the volume of prey remains between nests, we sub-sampled 20 mL of prey remains from each nest sample.

Pellets and nest contents were stored on a dry place in individual plastic bags (i.e., one pellet or the content of one nest per bag) for a period of six months after sampling. As only “hard” chitinous parts were used for insect identification, a potential deterioration of the material during the time period between sampling and analysis can, therefore, be excluded. Prey remains from pellets and nest contents were sorted and identified with a magnifier to the level of order and, when possible, to family, genus or species. To identify prey items, we used mainly head capsules (Hymenoptera, Hemiptera), wings or elytra (Coleoptera), mandibles (Orthoptera) and cerci (Dermaptera). Prey fragments were subsequently matched to determine the minimum number of individuals per prey category. We considered each head and a pair of wings, elytra, cerci or mandibles as one individual. Prey items were identified using Chinery (2007) and own reference collection. In total, 15 prey categories were identified as belonging to 5 insect orders: Hymenoptera, Coleoptera, Hemiptera, Dermaptera and Orthoptera. Among the Hymenoptera, we assigned insects to four families (Apidae, Vespidae, Formicidae and Scoliidae) and one species (Apis mellifera). For Coleoptera, we assigned insects to five families (Carabidae, Scarabaeidae, Curculionidae, Staphylinidae and Silphidae) and within Hemiptera, we identified insects belonging to the family Cicadidae. The prey items from Hymenoptera, Coleoptera and Hemiptera that could not be identified to a lower taxon than order were grouped as “others” within their categories (Appendix Table S1–S3).

2.3   Prey size during chick rearing

To explore variation in prey size provided to chicks, we observed adult bee-eaters carrying food to their chicks in three colonies: Oak-shrubland, Pasture and Meadow. The age of chicks was determined by using a photographic guide for ageing of chicks (Costa et al., 2020a) and nests (which contained broods) were subsequently categorized into four classes of development according to chick age (weeks 1–4). Nests containing eggs were classified as week zero (i.e., being incubated, no chicks). Prey observations were made each 3rd to 4th day in each colony for 15 min (totalling 240 min). During each observation, a telescope (magnification 30×) was aimed at one perch, where adults stop with prey in their bill before entering the nests and delivering it to chicks. Bee-eaters are usually single-prey loaders carrying only one prey item during each feeding event (Fry, 1984), which allows to observe the item before the bird carries it into the nest. Each prey item was classified according to its relative length compared to the bee-eaters’ bill. We defined size “S” as a small item (essentially a visible “dot”) in the tip of the bill; Size “M” a prey item larger than “S” and up to half the length of the bill; and size “L” a prey item of more than half the length of the bill (average bill length: 31.3 mm; Costa et al., 2020b).

2.4   Prey availability

In order to assess prey availability, we performed one session of transect counts per colony and period (i.e., pre-incubation, incubation and chick rearing), in which visible insects (i.e., flying in the air, on top of vegetation or the ground) were counted (see Finch, 2016). In Mixed forest and Oak forest, prey availability was examined only during chick rearing period. We focused on flying insects of the orders Coleoptera (beetles), Hymenoptera (bees), Orthoptera (grasshoppers), Hemiptera (true bugs and cicadas), which make up most of the bee-eater's diet previously recorded for Iberia (Herrera and Ramirez, 1974; Costa, 1991; Farinós-Celdrán et al., 2016). We did not consider Dermapterans because they were difficult to detect by this method. Each transect session consisted of eight transect counts in the predominant habitat and four transects in the secondary habitats of each colony. The start and direction of transects were randomly chosen within each habitat. Each transect consisted of 20 steps in a straight line and walking at a constant speed while counting all visible insects within 3 m to each observer's side. This allowed to attain relative abundances of prey availability systematically across colonies and periods. Transects were made only during late morning (10:00–12:00) on days when the minimum air temperature was above 25 ℃ and without moderate to strong winds (Grüebler et al., 2008; Arbeiter et al., 2016), thus ensuring that the highest levels of bee-eater feeding activity were included (these start at c.a. 10:00 with individual level variation; Inglisa and Galleotti, 1993, JSC pers. obs.), but that sampling remained below high temperatures that could negatively impact the activity level of insects (e.g., Chown and Nicolson, 2004; Blazyte-Cereskiene et al., 2010; during most of the breeding season temperatures in the colonies reach more than 30 ℃ soon after noon, often reaching 35 ℃ or higher during the afternoon).

2.5   Statistical analysis

The diet composition was assessed as the total number, mean, standard deviation and relative frequency (number of items of a prey category divided by the total number of items) of each prey category. For statistical analysis, we only considered prey categories of orders comprising more than 4% of bee-eaters’ diet (Hymenoptera and Coleoptera), thus retaining >95% of the diet at order level. Among those, we included in the analysis only the families/species present in more than 4% of general diet of both adults and chicks. Taxa below 4% were pooled in the category “other” of the respective order. Therefore, we grouped Formicidae with other Hymenoptera. Similarly, the families Curculionidae, Silphidae and Staphylinidae were pooled together with other Coleoptera.

Given the high number of zeros for prey categories (despite pooled at such a high level), GLMs did not converge even when using zero inflation adjustments. Therefore, to test differences in diet composition between colonies, we performed a Kruskal-Wallis for each prey category, having number of insects as dependent variable and colony as independent variable. When statistically significant differences were detected, a post-hoc Dunn's Test was performed.

Given that the 20 mL sub-sample of each nest content corresponded to a larger amount of prey remains compared to the volume of each adult pellet, to explore differences between diet of chicks and adults, we calculated the relative frequency of each prey category recorded in each nest content and pellet. Then, we performed a beta general linear model suitable for proportional data (bounded between 0 and 1) using R-package “betareg” (Grun et al., 2012). We defined prey category as a response variable and age class as independent variable.

To investigate if the size of prey provided to chicks changed throughout the rearing period, we performed a chi-square test on the proportion of different prey sizes across the sampling days.

To assess diet preferences, we used Jacob's selectivity index (Jacobs, 1974) comparing the proportion of prey consumed with the proportion of the prey available: I = (r – p)/(r + p – 2rp), where r is the relative abundance in the diet and p is the relative abundance in the environment. Jacob's Index varies from −1 (negative selection: prey consumed in lower proportion than what is available) to +1 (positive selection: prey consumed in higher proportion than what is available). Values of zero indicate that prey is selected in similar proportion to what is available in the environment. We calculated Jacob's selectivity index for adults and chicks considering the orders sampled in the diet (consumed) and in the transects (available): Hymenoptera, Coleoptera, Hemiptera and Orthoptera. As prey availability of these four orders decreased across the season (see below, Fig. 1), we calculated Jacob's index for adults during the pre-incubation period, when prey availability was highest, and for adults and chicks during rearing period, when the prey availability was lowest. This allowed investigating if selectivity of adult bee-eaters changed between those periods. We considered only Oak-shrubland, Meadow and Pasture for this calculation as we did not sample prey availability in Oak forest and Mixed forest during pre-incubation period. The relative abundance in diet was calculated for each order as the mean of the relative frequencies of that order in pellets (adults) or nest content (chicks). The relative abundance in the environment was calculated for each order as the weighted mean of the relative frequencies recorded across the considered habitats in each colony.

Figure  1.  Period specific prey availability recorded by visual transect counts of insects in the vicinity of bee-eater breeding colonies in Portugal. Periods encompassed pre-incubation (n=44), incubation (n=44) and chick rearing stages (n=60). Mean number (+/− SD) of insects per count had been categorized in four classes (i.e., Coleoptera, Hemiptera, Hymenoptera and Orthoptera).

DownLoad: Full-Size Img PowerPoint

3.   Results

3.1   Overall diet composition of adults and chicks

Bee-eaters generally fed mostly on Hymenoptera (adults: 58.8%, chicks: 64.1%) and Coleoptera (adults: 37.6%, chicks: 28.6%). Dermaptera, Hemiptera and Orthoptera were consumed in much lower proportion (range 0.4–3.7%; Fig. 2, Appendix Table S1 and S3). Within the order Hymenoptera, adult bee-eaters mainly preyed on Honeybee Apis mellifera (37.3%) and other Apidae species (12.5%). While among the order Coleoptera, families Scarabaeidae (21.0%) and Carabidae (8.0%) were the most abundant taxa consumed by adults (Fig. 2A, Appendix Table S1).

Figure  2.  Habitat specific diet composition in (A) adult bee-eaters (from pellets) and (B) chicks (from nest content) from five breeding colonies in distinct habitats. Composition of main diet includes several insect taxa (orders Hymenoptera in blue and Coleoptera in green, some specific families and the honeybee Apis mellifera), taxa representing <5% are pooled to category ‘Others’. For habitat categories see material and methods. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

DownLoad: Full-Size Img PowerPoint

Considering Hymenoptera, chicks were mainly fed with honeybees (27.8%) and ‘other’ Hymenoptera (22.7%). Among the order Coleoptera, Scarabaeidae (14.3%) and ‘other’ Coleoptera (12.8%) were the main prey categories (Fig. 2B, Appendix Table S3).

3.2   Diet composition in different habitats

In general, Hymenoptera (42.3–55.7%) and Coleoptera (43.3–53.5%) were similarly consumed by adults at Oak forest, Pasture and Oak-shrubland colonies, while at Meadow and Mixed forest colonies, Hymenoptera dominated (83.8% and 95.7%; Fig. 2A, Appendix Table S1).

We found significant differences in the number of items per pellet of hymenopterans between the colonies (Apis mellifera: Kruskal-Wallis χ² = 11.1, df = 4, p = 0.02; Apidae: Kruskal-Wallis χ² = 22.6, df = 4, p <0.001; Vespidae: Kruskal-Wallis χ² = 11.0, df = 4, p = 0.02) and coleopterans (Carabidae: Kruskal-Wallis χ² = 17.3, df = 4, p = 0.001; Scarabaeidae: Kruskal-Wallis χ² = 59.8, df = 4, p <0.001; ‘other’ Coleoptera: Kruskal-Wallis χ² = 22.2, df = 4, p <0.001) consumed by adults. The number of ‘other’ Hymenoptera in pellets of adults was small and did not differ between colonies (Kruskal-Wallis χ² = 3.54, df = 4, p = 0.47; Fig. 3A).

Figure  3.  Variation in the diet composition of adult (A) and chick (B) bee-eaters across five colonies in distinct habitats. Different letters indicate statistically significant differences obtained from Dunn comparisons. Prey taxa from order Hymenoptera are shown in blue and Coleoptera in green. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

DownLoad: Full-Size Img PowerPoint

The highest proportions of Hymenoptera provided to chicks were recorded at Meadow (83.2%) and Mixed forest colonies (86.6%, Appendix Table S3, Fig. 2B) while Coleoptera was recorded in higher percentage at Oak forest (40.6%) and Pasture (38.8%).

We found significant differences in the chick diet composition between colonies for hymenopterans (other Apidae: Kruskal-Wallis χ² = 18.74, df = 4, p < 0.001; Vespidae: Kruskal-Wallis χ² = 16.61, df = 4, p = 0.002; ‘other’ Hymenoptera: Kruskal-Wallis χ² = 14.08, df = 4, p = 0.007) and Scarabaeidae (Kruskal-Wallis χ² = 17.43, df = 4, p = 0.001). Carabidae was consumed only in Pasture and at very small proportions (Kruskal-Wallis χ² = 15.46, df = 4, p = 0.003). Apis mellifera was consistently consumed in high numbers in every colony (Kruskal-Wallis χ² = 4.16, df = 4, p = 0.38), and ‘other’ Coleoptera was also an important part of chicks' diet (Kruskal-Wallis χ² = 1.75, df = 4, p = 0.78; Fig. 3B).

3.3   Dietary differences between adults and chicks

There was a clear difference in the proportions of prey taxa consumed by adults and chicks, with the latter being fed with a higher proportion of Vespidae (p = 0.01), ‘other’ Hymenoptera (p < 0.001), Scarabaeidae (p = 0.01) and ‘other’ Coleoptera (p = 0.03; Table 2). We did not find significant differences in the proportions of Apis mellifera, Apidae and Carabidae between adult and chick diet (Table 2, Appendix Table S2 and S3).

Table  2.  Mean relative frequency and standard deviation of the proportion of each insect taxon in the diet of adult bee-eaters (n=50) and chicks (n=35), with Beta GLM results testing differences in proportions of each prey category between adults and chicks. Taxa with proportions <4% had been omitted before the analysis.

	Adults		Chicks		Model results
Prey category	Mean	SD	Mean	SD		Est.	SE	Z	p
Apis mellifera	0.34	0.30	0.27	0.11	Intercept	−0.747	0.16	−4.65	3.32e−6
					Chicks	0.09	0.24	0.39	0.69
Other Apidae	0.19	0.23	0.05	0.09	Intercept	−1.44	0.17	−8.13	4.14e−16
					Chicks	−4.43	0.23	−1.8	0.06
Vespidae	0.02	0.05	0.05	0.07	Intercept	−3.37	0.16	−20.05	<2e−16
					Chicks	0.51	0.19	2.63	0.01
Other Hymenoptera	0.09	0.15	0.24	0.11	Intercept	−2.30	0.17	−13.49	<2e−16
					Chicks	1.18	0.21	5.59	<0.001
Carabidae	0.07	0.15	0.01	0.02	Intercept	−2.64	0.17	−14.74	<2e−16
					Chicks	−0.36	0.21	−1.68	0.09
Scarabaeidae	0.11	0.15	0.13	0.11	Intercept	−2.18	0.16	−13.49	<2e−16
					Chicks	0.56	0.20	2.71	0.01
Other Coleoptera	0.09	0.17	0.13	0.08	Intercept	−3.30	0.11	−28.85	<2e−16
					Chicks	0.25	0.12	2.05	0.03

 | Show Table

DownLoad: CSV

3.4   Variation of prey size provided to chicks

In addition, prey size for chicks differed according to chick age/size, with a higher proportion of smaller insects (size S) being presented when most chicks were very young (development weeks 1 and 2). Adults provided their offspring with insects of larger size when chicks were older (development weeks 3 and 4) (size L; χ² = 162.56, p < 0.001; Fig. 4).

Figure  4.  Seasonal progress in chick development (A) and prey size (B) during the observation period from May to June (Day 1=May 1st). Chick development is given as proportion of nests containing broods with chicks in different development stages (in weeks 1–4, week 0 is the egg stage). Prey size for chick provisioning is the proportion of three prey size classes (S–small, M–medium, L–large) recorded by direct observations of chick-feeding adults.

DownLoad: Full-Size Img PowerPoint

3.5   Prey selection

The overall preferred insect taxa were Coleoptera, Hymenoptera and Hemiptera (Jacob's Indices of similarity range 0.15–0.74; Fig. 5). During pre-incubation period with high prey availability (Fig. 1), adult bee-eaters preferred Coleoptera, Hymenoptera and Hemiptera (Fig. 5A). As prey availability decreased across the season (Fig. 1), adult bee-eaters still selected Coleoptera and Hymenoptera for themselves and for chick provisioning (Fig. 5B). However, Hemiptera was only selected in higher proportions to chicks and was not recorded in the adult diet during chick rearing. Although Orthoptera were highly available across the entire breeding season (Fig. 1), they were always negatively selected (Fig. 5).

Figure  5.  Diet selectivity (Jacob index) of adults (brown) during: (A) pre-incubation, and (B) chick (green) rearing. Jacob's Index varies from −1 (negative selection/avoidance: prey consumed in lower proportion than being available) to +1 (positive selection: prey consumed in higher proportion than being available). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

DownLoad: Full-Size Img PowerPoint

4.   Discussion

Bee-eater diet during the breeding season varied across colonies in distinct habitats and between what adults consume in relation to what is provided to chicks. Overall, Hymenoptera and Coleoptera were the most important prey in the diet of both adults and chicks of colonies in Portugal. Our results show that the proportion of these prey taxa in colonies surrounded by urban or rural areas (Mixed forest and Meadow habitats) is distinct from that in colonies located in Oak forest, Oak-shrubland and Pasture, suggesting that habitat specific prey availability in the vicinity of the colony influences diet composition. Interestingly, and despite being opportunistic predators, adult bee-eaters preferred Hymenoptera and Coleoptera for themselves and to chicks, even when the availability of these insects was lower that of other prey groups (i.e., Orthoptera and Hemiptera). Furthermore, bee-eaters seem to be selective, not only regarding prey composition but also prey size, as adults provisioned their chicks with an increasing proportion of large insects through the chick growth period.

4.1   Overall diet composition of adults and chicks

Bee-eaters are known to prey on a wide range of flying insects, with up to 300 insect species of 15 different orders described in their diet (Fry, 1984). The predominant prey recorded in our study was Hymenoptera (adult diet: 59%, chick diet: 64%) and the second most consumed prey was Coleoptera (adult diet: 38%, chick diet: 29%), confirming the former diet composition in bee-eaters breeding in Portugal about 30 years ago (Costa, 1991). Hymenoptera and Coleoptera also formed the most abundant prey in other breeding populations across the species range, for example, in central Italy (65–70% and 17–25%; Inglisa et al., 1993), Sicily (67–96% and 2–19%; Massa and Rizzo, 2002), Spain (45–85% and 6–30%; Herrera and Ramirez, 1974), southern France (58% and 32%; Christof, 1990) and central Asia (65% and 23%; Kossenko and Fry, 1998). However, several studies from central Europe recorded much lower proportions of Coleoptera (northern Germany: 5%; Krüger, 2018; north-eastern Germany: 6%; Arbeiter et al., 2014). This suggests that location, and thus site-specific prey availability, plays an important role in the diet composition of bee-eaters, with Coleoptera being more frequently consumed in Mediterranean regions. Dragonflies (Odonata) have an important role in the diet of bee-eaters from north-eastern Germany (Arbeiter et al., 2014) as well as from France (Krebs and Avery, 1985) and Hungary (Fuisz et al., 2013), but were completely absent in the diet at our study sites. Odonata are mostly associated with freshwater bodies (Maravalhas and Soares, 2013), and these were scarce and dry in late spring at our study sites. However, Odonata have been recently recorded in the diet of bee-eaters in Portugal (13%; Lourenço, 2018) but were likely captured in colonies near aquatic habitats.

Despite having Hymenoptera and Coleoptera as main diet items, bee-eaters also preyed on Orthoptera, Hemiptera and Dermaptera (adults: 4%; chicks 7%). These orders have been previously recorded in the diet of bee-eaters (Herrera and Ramirez, 1974; Costa, 1991; Inglisa et al., 1993; Massa and Rizzo, 2002; Farinós-Celdrán et al., 2016), but contrary to other studies, we did not record the orders Diptera and Lepidoptera (e.g., Fuisz et al., 2013; Lourenço, 2018). However, these studies used a distinct sampling method (direct observation) to identify large insects, like large butterflies, which are easily spotted in the bill of birds, whereas small insects may be overlooked (Krüger, 2018). In pellet analysis, insects with soft bodies are methodologically under-represented, like Diptera and Lepidoptera (Bastian and Bastian, 2023). In fact, previous studies relying on the analysis of pellets alone also recorded a low proportion or absence of these two insect orders (Herrera and Ramirez, 1974; Costa, 1991; Kossenko and Fry, 1998). In any case, our direct observations did not record any medium or large Lepidoptera, suggesting this prey is taken in low frequencies, if at all, in the studied colonies.

4.2   Diet composition in different habitats

Diet composition of birds is often closely related to habitats and land-use characteristics in the surrounding area (Tsachalidis and Goutner, 2002; Orłowski and Karg, 2011; Di Maggio et al., 2018). Insect species richness is known to increase with plant species richness, as a higher diversity of plants likely provides a wider spectra of resources and niches for a more diverse insect community (Siemann et al., 1998; Haddad et al., 2001). Therefore, natural habitats with high plant diversity, like grasslands, often contain a higher richness of arthropods (Söderström et al., 2001; Attwood et al., 2008; Grüebler et al., 2008). In our study, we found a clear difference in the diet composition of bee-eaters between colonies surrounded by different habitats. More specifically, Hymenoptera was the predominant prey in Meadow and Mixed forest colonies (habitats surrounded by rural or urban areas); while in the other colonies (Oak forest, Oak-shrubland and Pasture), bee-eaters consumed similar proportions of Hymenoptera and Coleoptera. This suggests that colonies in Oak forest, Oak-shrubland and Pasture habitats likely have higher plant diversity, which may translate into higher diversity and abundance of insects overall, and of coleopterans in particular, compared to the colonies surrounded by rural and urban areas. These differences may also be related to the ecology of the most consumed coleopteran families, as Scarabaeidae are linked to ruminants (Barbero et al., 1999), and Carabidae may be more abundant in open habitats like pastures but also in oak forests (da Silva et al., 2008). Furthermore, Carabidae abundance and diversity may be negatively influenced by habitat fragmentation (Niemelä, 2001). This can explain a habitat specific food availability and, thus, a higher proportion of preyed Coleoptera in Oak forest, Oak-shrubland and Pasture. Meadow and Mixed forest colonies did not have cattle or other grazing ungulates in their vicinity, and were in highly fragmented habitats surrounded by buildings (residential and industrial areas), which could have contributed to the lower availability of Coleoptera and therefore to a lower proportion of this order in bee-eaters’ diet. Faced with a lack of coleopterans in Meadow and Mixed forest colonies, bee-eaters consumed a higher proportion of hymenopterans. The consumption of honeybees is often dependent on the availability of beehives in the vicinity of colonies, and although it is a common taxon consumed both by adults and chicks (e.g., Costa, 1991), it does not always compose the main prey of bee-eaters (e.g., Fuisz et al., 2013). As reported by Galeotti and Inglisa, 2001 bee-eaters prey on bees mainly in relation to their mean availability and in our study, the honeybee, Apis mellifera, was overall the most consumed Hymenoptera, as bee-eaters had access to beehives at less than 1 km in almost all colonies. The exception was the Oak forest colony, where, in fact, the mean number of honey bees per pellet was lower in adults’ diet.

4.3   Dietary differences between adults and chicks and variation of prey size provided to chicks

Dietary differences between adults and chicks are well documented for many insectivore bird species (Kaspari and Joern, 1993; Jiguet, 2002; Radford, 2008), including the bee-eater (Kristin, 1994; Massa and Rizzo, 2002; Arbeiter et al., 2014; Krüger, 2018). The main differences in diet composition between adults and chicks recorded in previous studies are, firstly, the higher proportion of Odonata (Massa and Rizzo, 2002; Arbeiter et al., 2014) or Lepidoptera (Krüger, 2018) and a lower proportion of Coleoptera (Kristin, 1994) by chicks. In our study, Vespidae, unidentified (‘other’) Hymenoptera, Scarabaeidae and ‘other’ Coleoptera were more frequent in the diet of chicks than in adults, suggesting that breeders are selectively provisioning their offspring with these taxa. Coleoptera have indeed high nutritional content (Razeng and Watson, 2015), and it is likely that bee-eater chicks grow more efficiently (i.e., faster) when fed with mixed diet (Krebs and Avery, 1984). Additionally, selectivity was also found in terms of prey size. Adult bee-eaters seem to feed their offspring with larger items than what they consume (Fry, 1984; Kristin, 1994; Massa and Rizzo, 2002; Arbeiter et al., 2014). This is not unusual in birds feeding on flying insects: Tree Swallows (Mccarty and Winkler, 1999) and Barn Swallows (Orłowski and Karg, 2011) had a preference for larger prey when feeding young, even if smaller insects were more abundant. With direct observations of prey provided to chicks, we recorded an increase in prey size during the rearing period from young to old chicks, suggesting a gradual selection of larger items as chicks grow. This finding contrasts with the situation recorded in a German colony, where the mean prey mass for chicks decreased during the chick rearing period (Krüger, 2018). The author argues that the local breeders follow an opportunistic strategy with increase in feeding rates, independently of size of insects, in order to satisfy the larger energy demands of older chicks.

4.4   Prey selection

Bee-eaters are known to be opportunistic but also to actively select hymenopterans, which typically dominate diets at many sites across the range (e.g., Kristin, 1994; Kossenko and Fry, 1998; Aissaoui-Marniche et al., 2007; Farinós-Celdrán et al., 2016).

To assess prey selection, it is necessary to consider diet composition and food availability. Inglisa et al. (1993) compared the diet of bee-eaters with prey availability, and found a permanent selection of Hymenoptera during the breeding season, but selection of Coleoptera and Hemiptera only during the end of the season. Our results show a clear preference for Hymenoptera, Coleoptera and, to a lesser extent, for Hemiptera, disregarding Orthoptera independently from availability. Coleopterans may be hard to digest due to its high proportion of chitin but at the same time they are slow-flying insects and possibly easier to catch than fast-flying ones. Although Orthopterans were highly available during the entire breeding period, they were not preyed by bee-eaters which is an aerial hunter. Hemiptera was positively selected during pre-incubation and during chick rearing but consumed proportionally to its availability by adults during this latter period. Our study shows that bee-eaters equally selected Hymenoptera and Coleoptera for themselves and chicks throughout the breeding season, highlighting the importance of these two insect orders for the species diet in Iberia.

CRediT authorship contribution statement

Joana S. Costa: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Steffen Hahn: Writing – review & editing, Writing – original draft, Validation, Supervision. José A. Alves: Writing – review & editing, Writing – original draft, Validation, Supervision, Methodology, Formal analysis, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We are grateful to the landowners, Jorge Turibio and Jerónimo Trigueiros. We thank Luzia Costa for her valuable assistance during fieldwork, Ricardo Correia for his support with the statistical analysis, and Prof. Artur Serrano, Mário Boieiro and Sandra Antunes for the training in insect identification.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.avrs.2024.100211.