Nest sharing and provisioning activity of females of the digger wasp, Cerceris rubida (Hymenoptera, Crabronidae)

Abstract

Using individually marked females and nests, we investigated aspects of the nesting biology of Cerceris rubida, the only European species of its genus suspected to cooperatively share nests. Up to 11 females were marked from a single nest, but only 2–6 shared a single nest at a given moment, thus probably not all the wasps remained in their natal nests after emergence. Only 1–3 of the females sharing a nest collected brood cell provisions at a given moment, sometimes joined by other females that took few additional prey to the nest. The wasp has two generations a year, and the first generation females were larger than those of the second generation. A guard wasp was constantly present at the nest entrance and rejected non‐nestmate females and parasites. Females typically made 1–18 provisioning trips in quick succession before stopping their hunting activity. They then remained in the nest for 1–148 min before re‐starting provisioning. Provisioning females generally foraged simultaneously. All provisioning females showed well‐developed ovaries with small to large oocytes ready for oviposition. Wasp size was not correlated with the number of provisioning flights it performed. Mandible wear and wing wear seemed to be independent of provisioning activity, suggesting that females are involved in digging and provisioning across their life span. We discuss these findings in relation to the known biology of other social Cerceris and other social apoid Hymenoptera.

Keywords:

• Cerceris
• sociality
• provisioning
• size

Introduction

In most species of digger wasps of the family Crabronidae, females are solitary, digging and provisioning a nest without the help of conspecifics (Evans & West‐Eberhard 1970; Bohart & Menke 1976). Yet other observations have revealed a greater variety of social interactions in a number of crabronid species; this could be of interest for studies of the evolution of sociality in aculeate Hymenoptera (Matthews 1968; Evans 1973; Alcock 1974; Ross & Matthews 1989). Although eusociality has been observed in only a few crabronid species confined to a small group of genera (Microstigmus Ducke, Arpactophilus Smith) (Matthews 1968; Matthews & Naumann 1989), an increasing number of species have been found to exhibit communal or subsocial behaviours (Evans 1973; Elliot et al. 1986; Evans & Hook 1986; McCorquodale & Naumann 1998).

In particular, the genus Cerceris Latreille, the richest in species within the family, shows an interesting variety of social behaviours, including solitary and social species. For this reason it could parallel the sweat bees (Halictidae) in the study of social evolution (Evans & Hook 1982; Field & Foster 1995). In fact, both groups include fossorial, mass‐provisioning species that often nest in dense aggregations, and they could have been shaped by similar selective pressures toward sociality (e.g. parasite pressure, limited key resource availability) (Michener 1974; Evans & Hook 1982; Wcislo 1984; Rosenheim 1990). In solitary Cerceris species, interactions between females are widely diffused, but they are mostly competitive (Elliot & Elliot 1987; Field & Foster 1995; Alexander & Asís 1997). When nest sharing does occur, various kinds of female associations have been observed: newly emerged wasp–older wasp (Elliot et al. 1986), mother–daughter and sister–sister (McCorquodale 1988, 1989).

In the present study we give data on nest sharing and cooperative interactions in C. rubida Jurine, the only species of the genus in Europe in which more than one female shares a burrow. The few studies to date on this species (Grandi 1944, 1961; Jacobi 2003) consist of short field observations carried out without using any technique to recognize females individually (i.e. individual marks). The only information available concerns the simple presence of a number of females entering and exiting from a single nest opening.

In particular, the following questions were addressed.

1. How many females share a single nest in a given period of the nesting season?

2. Is there any difference in the activities performed by the different females of a nest?

3. Are there, among females living in the same nest, any differences in size or other morphological traits? Could these differences be related to wasp activity?

Materials and methods

Field observations were carried out from 19 July to 20 August 2004 and from 18 to 30 June 2005 at the study area, located near Alberese (province of Grosseto, Tuscany, Italy) in the Parco Regionale della Maremma. We first identified, during preliminary observations in June 2004, a small nest aggregation consisting of 50–70 nests of C. rubida. This area is situated on a trail bounded by cultivated fields (maize, Zea mais, and wheat, Triticum sp.). The climate of the site is typical Mediterranean, characterized by rain mostly in October–November and highest temperatures in July–August.

The nests of C. rubida were intermixed with several other nests (>300) of the eusocial bee Lasioglossum malachurum Kirby (Halictidae) and a few nests (<30) of another social sweat bee, Halictus scabiosae Rossi. In 2005, we also discovered in the nesting area very few nests (<10) of other three Cerceris species: C. sabulosa (Panzer), C. bupresticida Dufour and C. circularis (Fabr.).

Nests observations were carried out from 9.00 to 17.00 (solar hours). Six nests (four of them considered for subsequent analyses) in 2004 and seven nests in 2005 were marked with a drawing‐pin located close to the nest entrance. The respective owners of all these nests were marked with acrylic paints on the thorax. The combination of colours was unique for each wasp, as commonly used in this kind of behavioural study (see, for example, Field et al. 2000; Casiraghi et al. 2001). Data on activity collected in 2004 were used for provisioning rhythm analysis, while both years' observations provided data on the number of nest‐sharing females and their tasks.

In 2004, 50 C. rubida females were captured. Of these, 37 from the six nests were individually marked; in 2005, the number of marked females was 36 from seven nests. We measured the head width (all wasps) and the forewing length (only in 2004) using callipers to the nearest 0.01 mm.

In 2004, 14 wasps were collected in individual tubes, their ovaries examined and then preserved in ethanol (70%) for morphological observations with a dissection microscope. We defined as kind “A” mandibles with no wear (with the two teeth on the internal side clearly visible and the distal part clearly pointed) and kind “B” those with more worn mandibles (with teeth not visible and the distal part rounded). We defined forewings as kind “A” (those lacking laceration on the external margins), or kind “B” (with visible laceration).

At the end of August and in mid‐September 2004 we excavated three of the six marked nests to obtain data on their underground structure and to collect all the individuals occupying them.

Statistical analysis was performed using both parametric (ANOVA, Student's t‐test) and non‐parametric (Mann–Whitney test, Kolgomorov–Smirnov test, Spearman correlation test) approaches, depending on the characteristics of the data pools. Average numbers are given±SD.

Results

In the six nests of 2004, we were able to mark from 4 to 10 different females per nest during the observation period (6.16±2.22 on average, n = 37). For provisioning rhythms analysis, we considered four of these nests (figure 1). All wasps from these four nests (n = 26) were seen to perform provisioning flights on at least one of the days of observation. In 2005, we marked 1–7 females per nest (5.14±2.26 on average, n = 36). No statistical differences resulted between the number of wasps marked per nest in the two years (Mann–Whitney test: U = 35.5, P = 0.41, NS). A sample of prey, collected from the wasps returning to their nests or from brood cells, comprised small beetles of four families (Chrysomelidae, Hydrophilidae, Nitidulidae, Curculionidae). No more than six females in 2004 and four in 2005 were found living in the same nest simultaneously. In 2004, nests 1, 3 and 4 were observed for 4–7 days in succession and were associated always with the same females; however, nest 2, observed for three successive days in middle July and for three successive days in mid‐August, had five females in the first period and three (all different but one) in the second. C. rubida has two generations a year at the study locality, the first one starting around the middle of June (first females were observed on flight on 10 June in 2005).

Figure 1 Number of provisioning wasps from four nests(n = 26) and their average (percentage) provisioning activity (prey/day), in 2004 (second‐generation activity period). The dashed line represents the average of all wasp provisioning activities (percentages).

For the four nests analysed in 2004, it appeared that, at a given moment, some wasps were more active in provisioning activity than others (figure 1). To discriminate which females could be considered the “primary” provisioners of a nest, we considered the average of all wasp provisioning activity (percentage of provisioning trips per day), i.e. 2.9% (dashed horizontal line in figure 1), and we defined as a “primary” provisioner a wasp that had an average provisioning activity higher than this value. All other provisioner females were termed “secondary” provisioners. We observed that two or three (in one case) females per nest correspond to this definition, with provisioning activity often much higher than the 2.9% limit. χ²tests showed that in each nest all the females with a provisioning activity >2.9% made a number of provisioning flights higher than all other females with provisioning activity <2.9% (table I). This confirmed the validity of the 2.9% limit to distinguish the two kinds of females. A primary provisioner wasp brought an average number of 7.1±3.5 prey/day (n = 9), while a secondary provisioner wasp carried 1.4±1.3 prey/day (n = 15) during the observation period. In 2005, 1–2 females matched the definition of “primary” provisioner at a given moment, but very rarely (two nests on just two days) were these females provisioning together with “secondary” provisioners on a given day.

Table I. χ² tests results to verify the validity of the 2.9% limit for definition of primary provisioners and secondary provisioners (2004 data). Tests compared the number of provisioning flights of primary provisioners to all the secondary provisioners of the same nest.

Primary provisioner	Nest	χ^2	Df	P
4	1	62.26	4	<0.01
6	1	60.3	4	<0.01
8	2	36.12	7	<0.01
11	2	39.69	7	<0.01
15	2	43.35	7	<0.01
18	3	68.76	4	<0.01
22	3	15.80	4	<0.01
24	4	56.00	2	<0.01
26	4	28.80	2	<0.01

Provisioning females were observed to perform two kinds of provisioning flights with respect to the time spent in the nest after they had entered with prey: (1) quickly coming out (SP = short presence), (2) staying inside for some minutes before leaving (LP = long presence). The time spent in the nest during a SP was often less than 10 s (range = 1–40; average = 5.96±9.93), while the time spent in the nest during a LP was always more than 1 min (range = 1–148; average = 22.1±31.2). Only three times did the period inside the nest last between 10 s and 1 min. On most occasions (46 out of 56), presence in the nest during a LP lasted from 1 to 30 min; within this range, the most frequently observed interval was 1–10 min (27 out of 46). Presence in the nest tended to be longer in the morning (in particular from 10.00 to 11.00; figure 2), but was not significantly different from presence at other hours (we considered only the period from 9.00 to 14.00 because of the lack of LP after then) (Bartlett's test for homogeneity of variance: df = 5, P = 0.002; ANOVA: df = 5, F = 0.57, P = 0.71, NS).

Figure 2 SP and LP frequencies(average per day) and LP average duration (minutes) across the day (2004 data). SP, short presence in the nest after a provisioning flight; LP, long presence in the nest after a provisioning flight.

SP frequency was higher in the middle of the activity period (from 12.00 to 13.00), decreasing sharply in the afternoon (from 15.00), while LP did not show any particular peak during the day, simply decreasing in the afternoon like SP (figure 2). The daily distributions of SP and LP differed (Kolmogorov–Smirnov test: D = 0.12, n _SP = 338, n _LP = 174, P = 0.044).

We calculated the sequence of SP and LP, i.e. how many SP were performed, on average, between two subsequent LP. This analysis was made per nest and per wasp. We observed that a nest received 0–21 prey (SP) between two LP (average = 2.08±2.81, n = 4), though most times just 0–3 SP (figure 3), and that a single wasp (A = primary provisioner, a = secondary provisioner) carried 0–18 prey (SP) into her nest between two LP (by any two wasps x and y) (all wasps, i.e. LP _x –ΣSP_a,A–LP _y : average = 0.26±0.34, n = 25; primary provisioner wasps, i.e. LP _x –ΣSP_A–LP _y : average = 0.56±0.47, n = 9), most of times 1–5 SP (figure 4). Moreover, on average, a primary provisioner wasp made 2.61±2.41 SP (n = 9) between two of her own LP (LP_A–ΣSP_,A–LP_A).

Figure 3 Average number of SP between two subsequent LP received in a nest(2004 data, n = 4). Abbreviations as in figure 2.

Figure 4 Average number of SP between to subsequent LP performed by wasps(2004 data, n = 25). Abbreviations as in figure 2.

The “primary” provisioner females typically provisioned simultaneously, since there were no periods in the day when one wasp's activity was significantly higher than that of the others. In fact, a Spearman correlation of the number of sequential entries by a wasp (after another wasp's entry) versus the frequency of the following entry by the same wasp was significant (r = −0.86, n = 15, P = 0.0012).

As a measure of the size of the wasps, we used head width, which was correlated with forewing length (Spearman correlation test: r = 0.68, n = 50, P<0.001).

The size of a female was not correlated with its provisioning activity (both as the average number and the highest number of provisioning flights) (Spearman correlation test: average flights: r = −0.19, n = 29, P = 0.31; highest number of flights: r = −0.26, n = 29, P = 0.16). Females marked in 2005 (belonging to the first generation: average head width = 2.28±0.12 mm, ranging from 2.00 to 2.58 mm) were larger than those marked in 2004 (belonging to the second generation: average head width = 2.00±0.18 mm, ranging from 1.66 to 2.40 mm) (Student's t‐test: t = 8.36, df = 80, P<0.001).

A guard wasp was always present at the nest entrance, with its head visible. Unfortunately, we were not able to identify this wasp (the coloured marks were not visible from this position). We assume that guarding was performed by the same female for at least some hours across a day, because the guard did not change after the entry of another wasp.

Wasps sharing the same nest were generally accepted by the guard when returning to their own nest. However, in some cases (n = 6 in 2004) a wasp was rejected by the guard of her nest when returning back to it. This temporary (a few hours') rejection was observed immediately after the marking operation. We should not exclude an effect of the marking paint (but for many social insects this technique is successfully used); however, the rejected wasp was always accepted later by the guard.

Guard wasps almost always rejected a foreign C. rubida female. On just two occasions in 2004 was a foreign female, observed to provision a different nest for several days, accepted by the guard: in one case becoming a primary provisioner of the new nest and in the other staying in the new nest without provisioning (we confirmed her presence some days after, during the excavation of the nest). The guard also actively attacked, protruding its head from the entrance and snapping its mandibles at cuckoo wasps (Hedychrum gerstaeckeri Chevrier (Chrysididae)) and the scuttle flies (Megaselia leucozona Schmitz (Phoridae)), two parasites present at the aggregation. A third parasite, an undetermined miltogrammine satellite fly (Sarcophagidae), was often seen to follow the wasps carrying a prey, but it did not enter in contact with the wasp guard, because it probably larviposited landing rapidly on the nest entrance (Polidori et al., unpublished data). Moreover, the guard often attacked workers of Lasioglossum malachurum, which were frequently observed trying to enter the wasp nests.

From excavations of the nests we obtained data on their general structure. The nests consisted of a main, oblique burrow ending in 2–8 brood cells. We were able to recognize the contents of nine cells: three were empty, five included pupae and prey remains and one only prey remains. Cocoons were typically made of silk covered by the elytera of eaten prey.

Dissection of the abdomen of 14 females (2 primary provisioners, 2 secondary provisioners, and 10 females randomly collected in the field on the last days of the flight period–mid‐September) revealed that almost all of them had developed ovaries with small to large oocytes ready for oviposition. Using the ovary classification of Evans & Hook (1982) for C. australis De Saussure, we recognized among the females of C. rubida two out of the three kinds of ovaries: young adult ovaries with small oocytes (2 cases); adults with well‐developed ovaries with true oocytes fully developed (12 cases) (table II). The two females with ovaries of the first kind were collected on the 19 August from two observed nests but they were not marked, presumably because they had recently emerged.

Table II. Comparison of the morphology of primary and secondary provisioners (2004 data).

	>2.9% prey/day (primary provisioners)	<2.9% prey/day (secondary provisioners)
Head width (mm)	1.92±0.14, n = 9	2.00±0.17, n = 14
Ovarian development	Well developed (n = 2)	Well developed (n = 2)
Mandibular wear	2 worn (type A) and 2 unworn (type B)	1 worn (type A) and 1 unworn (type B)
Wing wear	3 worn (type A) and 2 unworn (type B)	Worn (n = 3)

We also analysed mandible and forewing wear and were able to identify two types of mandible and forewing wear (figure 5). Both primary and secondary provisioner wasps showed both kinds of mandible and forewing wear (table II).

Figure 5 Mandibles and forewings of primary and secondary provisioning wasps: a, unworn mandible (type A) (the arrows point to the two teeth; the dashed line marks the length between the distal part of the second tooth and the end of the mandible); b, worn mandible (type B) (note the worn first tooth and the shortened distance between the distal part of the second tooth and the end the mandible); c, unworn forewing (type A); d, worn forewing (type B). Bar = 0.5 mm.

Discussion

Although the genus Cerceris, comprising 900 known species worldwide, potentially represents one of the most valuable groups for the study of the evolution of cooperation in aculeate Hymenoptera due to its interspecific social variability, little is known about the social species of the genus. Most data on Cerceris wasps concern solitary species (e.g. Linsley & McSwain 1956; Genaro & Sanchez 1993; Hook 1987), in many cases supported by good investigations of female–female competitive interactions (nest usurpation). Competitive behaviours are widespread in the genus and are probably closely related to the subsequent development of cooperation (Willmer 1985; Field & Foster 1995; Polidori et al. 2006). Moreover, in some solitary species performing nest usurpation, there were recorded cases of simultaneous, short‐term provisioning of the same nest by more than one female (joint provisioning).

This last behaviour in the past was often confused with cooperation, but it has now been demonstrated to be a product of nest searching by usurped females, which occupy nests when the owner is absent and the latter does not meet the intruder for some hours (Alcock 1975; Field & Foster 1995).

The widespread habit of nest usurpation, together with reuse of the same nest generation after generation, suggests that the main limiting factor in the nesting biology of Cerceris wasps is probably the nest, which may represent one of the selective pressures favouring sociality in this group (Melo 2000; Cahan et al. 2002; Polidori et al. 2006).

Of the 20 social Cerceris species that have been studied, detailed information is available for just 10 of them (table III); the other 10 species have been investigated briefly and show only weak evidence of nest sharing by females (Tsuneki 1965; Mueller et al. 1979; Evans & Hook 1986; Evans 2000).

Table III. Data on social traits in nest‐sharing species of Cerceris. Abbreviations: W, highest number of active wasps per nest at a given moment; Pr, differences in provisioning activity among females; G, presence of a guard wasp at the nest entrance; S, size differences between wasps with different provisioning activity; O, ovarian development differences between wasps with different provisioning activity; M/W, mandible and/or wing wear differences between wasps with different provisioning activity; R, relatedness among nest‐sharing females (M–D, mother–daughter, S–S, sister–sister).

	W	Pr	G	S	O	M/W	R	References
C. acanthophila	2	–	–	–	–	–	–	Hook 1987
C. antipodes	4	yes	yes	no	no	–	M–D, S–S	Alcock 1980; Evans & Hook 1986; McCorquodale 1988, 1989, 1990
C. australis	4	yes	yes	no	yes	yes	–	Evans & Hook 1982, 1986
C. goddardi	7	yes	yes	no	no	–	–	Evans & Hook 1986
C. minuscula	4	yes	–	–	no	–	–	Evans & Hook 1986
C. rubida	5∗	yes∗	yes	no∗	no∗	no∗	–	Grandi 1944, 1961; Jacobi 2003; ∗this study
C. unispinosa	1	yes	–	–	–	–	–	Evans & Hook 1986
C. watlingensis		yes	–	–	–	–	–	Elliot et al. 1986
C. windorum	2	yes	–	–	–	yes	–	Evans & Hook 1986
C. xanthura	3	–	–	–	–	–	–	Evans & Hook 1986

In table III we summarize current knowledge of the social biology of these species of Cerceris. At first sight, most characteristics seem to be common to all species. In particular, the presence of a guard at the entrance of the nest is a widespread phenomenon. Guarding the nest and attacking intra‐ and interspecific intruders is a successful technique of defence in many primitively social wasps and bees (Gobbi et al. 1991; Miyanaga et al. 1999; Wcislo & Shatz 2003) and it may represent an evolutionary benefit. Fossorial bees and wasps are often attacked by parasitoids and cleptoparasites (Michener 1974; Bohart & Menke 1976), and the presence of nest guarding could be evidence of its importance in the evolution of the nesting biology of these animals (Rosenheim 1990). Whether an individual remains as guard throughout its lifetime is not known in wasps, although honeybee workers generally guard for only a very limited part of their lifespan. Unfortunately, we were unable to determine the identity of the guard wasp, because our marks on the thorax of the guard were not visible when it was inside its nest.

Nevertheless, there is some evidence that the identity of the guard does not change frequently in C. rubida (i.e. at each new entry of another wasp to the nest). These observations do not agree with the hypothesis of Grandi (1944, 1961) (based on observations of unmarked females), i.e. that every time a C. rubida female enters the nest, a different one exits, leaving the previous wasp as the guard. This is not true, because during our observations most times the same female entered and exited very quickly from the nest (SP behaviour), so the guard had to be a wasp that, at that moment, was not involved in provisioning. Moreover, from the three excavated nests, we collected marked females and few newly emerged wasps (with poorly developed ovaries and little mandibular and wing wear), thus the guard was probably a marked wasp in a non‐provisioning period of her life (or perhaps a secondary provisioner).

Many times, often in the late afternoon, we observed some wasps patrolling the nesting site with a typical “nest searcher” behaviour, i.e. flying slowly close to the soil surface and trying to enter into various nests (Field & Foster 1995; Polidori et al. 2006). Considering that up to 10 females were marked from a single nest, we suppose that not all the females stay in their natal nest after emergence, becoming “nest searchers”. The rare cases of acceptance by a guard of a foreign female (as in other social Cerceris—Alcock 1980; Evans & Hook 1986—the C. rubida guard usually rejected foreign females) that we recorded in 2004 led us to suppose that “nest searchers” could become new “nest mates” in some moments of their life. In one of these cases, we noted that the wasp was accepted in a nest that, at that moment, was occupied by only one female, since all the other wasps of that nest had been collected for morphological analysis. Foreign females might be accepted in a nest only in the case of an insufficient number of provisioners to satisfy the necessity of brood care. However, more observations are required to accept or reject this hypothesis.

Differences in the distribution of provisioning activity among nest‐sharing females are other common traits of the social organization of social Cerceris. However, one has to note that in many works only provisioner or non‐provisioner wasps have been recorded, because during excavations of the nests wasps were found that were never observed provisioning. However, since these studies were often carried out across brief periods of time, it is not clear if non‐provisioner wasps were effectively never‐provisioning wasps or simply wasps not seen to carry prey into the nest in the (short) period of observation. For C. rubida, we cannot assess the existence of these two categories, because we always recaptured all the marked females while returning with prey, at least once, and other (unmarked) wasps were generally not collected during nest excavation. Thus, in our analysis, we have preferred to identify two groups of wasps, based on their provisioning activity, as “primary” and “secondary” provisioners. This definition is obviously restricted to a limited period of observation, since a primary provisioner female could slow down her activity at any given moment, while a secondary provisioner could become, later on, a primary provisioner. During our study, primary provisioner females maintained their provisioning activity for 3–4 days. After this time, one or both primary provisioners changed. The evidence for a complete exchange in provisioning roles among females came just from one nest (nest 2), which was observed for two 3‐day periods separated by one month.

Other interesting information comes from the analysis of provisioning activity. For example, it emerged that primary provisioners were active simultaneously, alternating their trips, thus the nest probably requires at least two females bringing prey through the day to satisfy brood care. Often primary provisioners were joined by two or three secondary provisioners during the second generation, but they commonly provisioned without the help of secondary provisioners during the first generation.

The species has at least two generations a year in this locality, covering the period from mid‐June to late September. Grandi (1961) also supposed that this species had two generations a year.

The clear distinction between SP and LP is quite puzzling. One hypothesis is that LP corresponds to the moment when a wasp completely fills its empty cell with prey items that have been temporarily stored in the nest during SP, a widespread habit in the genus (Bohart & Menke 1976). This hypothesis can be rejected, because the average number of SP between two subsequent LP (either for wasps or for nests, see Figures 3 and 4) was too low to represent the entire cell contents (50 prey items or more, according to Grandi 1961, p. 175; about 50 prey items according to Aptel 1931, quoted by Grandi 1961). However, we cannot exclude the possibility of a partial fill of the cell that may happen or immediately after having brought a prey (SP = 0 between two LP) or when one or more prey are already stocked outside the cell (SP1 between two LP).

As in the other analysed social Cerceris, all C. rubida females had developed ovaries, suggesting that effectively all females are able to reproduce and that the distribution of provisioning activity varies among females during the season, probably according to the brood care necessity of each female after oviposition. In C. australis, in contrast, provisioner females had clear ovarian reduction, suggestive of a division of labour between females, i.e. true workers (Evans & Hook 1982). Provisioning activity was not related to wasp size in C. rubida, probably because all females forage. Anyway, females of the first generation were larger than females of the second generation. We may speculate that this could happen because a lower number of females provisioned a nest (on a given day) during the first generation activity period, and this could result in less food provided to each larva, producing a smaller‐size second generation.

Mandible and wing wear analyses suggested that all the females of a nest contribute, possibly in different periods of their life, to all the operations related to parental care, i.e. nest cell construction, nest expansion, provisioning and prey hunting. However, in C. australis there is evidence for some females with very worn mandibles and others with weakly worn mandibles; it is suggested that this represents a clear division of activities, with provisioner females using mandibles less than non‐provisioner ones, that are more involved in digging (Evans & Hook 1982).

Finally, we highlight comparisons between the social traits of Cerceris species and other primitively social Apoidea.

The presence of the guard is widespread in eusocial halictid bees (Miyanaga et al. 1999; Richards 2000; Wcislo & Shatz 2003), but is not a common trait of communal species in this family (Kukuk & Sage 1994; Wcislo 1997) or in other communal bees (Michener 1974; Paxton & Tengö 1996; Giovanetti et al. 2003). Moreover, in socially advanced crabronid wasps, defence of the nest is apparently often undertaken by males or by both sexes (Gobbi et al. 1991; Matthews & Naumann 1989; Melo 2000). From this point of view, social Cerceris have unique behavioural traits, since they combine the lack of differences in reproductive capacities of females with some behavioural characteristics (nest guarding by females) typical of eusocial apoid Hymenoptera. Moreover, social Cerceris differ from communal fossorial crabronids (e.g. Spilomena subterranea McCorquodale & Naumann, Philanthus gibbosus (Fabricius)), in which both males and females share the nest (Evans 1973; McCorquodale & Naumann 1998). Another difference between communal bees and social Cerceris is relatedness between nest‐sharing females; this seems to be low in communal bee species (Kukuk & Sage 1994; Danforth et al. 1996; Paxton et al. 1996) and high in Cerceris species (although only one study was carried out on this topic, see table III). These data suggest that nest sharing has evolved several times not only in other apoid Hymenoptera but also in Crabronid wasps.

Acknowledgements

Thanks are due to Fabrizio Rigato and Carlo Pesarini (Museo di Storia Naturale di Milano) for the determination of Cerceris rubida and its prey. We thank Josè Tormos (Universidad de Salamanca) for the suggestions that improved the manuscript. We are very grateful to Robert Paxton (Queen's Univesity of Belfast) for helpful critique and English revision. We thank the Maremma Regional Park for kind support. The present work was supported by a three‐year grant FIRB (Fondo per gli Investimenti della Ricerca di Base) RBAU019H94‐001 (2001).