HYM. RES. J. Vol. 19(1), 2010, pp. 144-158 Colony Social Organisation of Halictus confusus in Southern Ontario, with Comments on Sociality in the Subgenus H. (Seladonia) M. H. Richards, J. L. Vickruck, and S. M. Rehan Department of Biological Sciences, Brock University, St. Catharines, Ontario — Abstract. Halictus (Seladonia) confusus Smith is one ofthe mostcommonbees in NorthAmerica. Classifiedaseusocial,itscolonysocialorganizationisknownonlyfromqualitativedescriptionsofa population in Indiana. We studied the phenology and social behaviour of this bee in the Niagara RegionofsouthernOntario,usingnestexcavations,dissectionsandn\easurementsofadultfemales, and pan trap samples offoragingbees to elucidate key elements ofcolony social organisation. The colony cycle in Niagara is typical of temperate-zone halictines, with overwintered foundresses producing a firstbrood ofworker-sized females and a few males, followedbyproductionofBrood 2, consisting of gynes and more males. Many Brood 1 females become reproductive: about one- quarter of Brood 1 females dissected exhibited levels of ovarian development rivalling queens. In contrast, only about one-quarter of Brood 1 females become classically altruistic, sterile workers. High rates of worker reproductivity may result from early queen mortality and supersedure or from the inability of viable queens to control worker behaviour - the average queen-worker size difference was only 5.6%, and queens were not always larger thanthe workers in their ownnests. Comparisons with the Indiana population suggest a geographic component to variation in colony social organisation. Comparisons with other members of the subgenus for which detailed information is available, suggest that in Seladonia, as in other eusocial halictines, queen control of worker behaviour depends on the ability of queens to dominate small numbers of small-bodied workers. — Key words. Halictidae, eusociality, pan traps, sweatbee In halictine bees, evolutionary transi- understood, this in turn may help to tions from solitary to eusocial behaviour illuminate patterns observed at higher involve two components, a demographic taxonomic levels, such as differences change from univoltine to multivoltine among subgenera or genera. For instance, colony phenology, and a behavioural socially polymorphic sw^eat bees such as change from maternal care by a lone Lasioglossum calceatum and Halictus rubi- foundress, to associationsbetweenmothers cundus, exhibit solitary, univoltine colony and daughters that raise brood coopera- cycles in regions w\i\\ short breeding tively (Schw^arz et al. 2007). Likev^ise, seasons, and eusocial, bivoltine (actually, evolutionary transitions from eusociality double-brooded) colony cycles in regions to solitary behaviour, involve the reverse w^ith long breeding seasons (Sakagami and changes in demography and behaviour. Munakata 1972; Eickw^ortetal. 1996). There Therefore, to understand evolutionary are also obHgately eusocial species, such as transitions between solitary and social H. ligatus and L. malachurum that exhibit behaviour, it may be particularly fruitful considerable demographic variation, with toexamine species thatexhibitintraspecific colonies growing to larger sizes in areas variability in eitherorboth ofthese traits. If with longer breeding seasons (Michener the adaptive significance of intraspecific and Bennett 1977; Knerer 1992). These demographic and social variability can be intraspecific patterns suggest that one Volume 19, Number 1, 2010 145 cause of the phylogenetic lability of social of Brood 1 females become reproductives, behaviour observed in several halictine rather than sterile workers, suggesting that genera, might be geographic or temporal the population contains a mix of solitary variability in the harshness of local envi- and social strategies, as well as univoltine ronmental conditions. Indeed, this predic- and bivoltine phenologies. We also com- tion is borne out by recent evidence that pare H. confusus to other well studied halictine sociality may have first evolved members of the subgenus Seladonia, in during a period of global climate warming order to assess the level of social variation (Brady et al. 2006), when it would have in the subgenus as a whole. been possible for univoltine halictine METHODS lineages to adopt bivoltine or multivoltine — nesting phenologies. Study sites. All study sites were on or One of the most common eusocial within walking distance of the Brock halictines in North America is Halictus University Campus in St. Catharines, On- (Seladonia) confusus Smith, but detailed tario (W 79 14' 57" N 43 07'11"). We information on its nesting and social excavated nests from a small nesting biology are distinctly lacking. Dolphin aggregation on the north shore of Lake (1966) studied the nesting biology and Moodie that contained nests of Halictus social behaviour of this bee in Indiana, confusus and H. ligatus, and hibemacula of USA, from 1963-1965. Although many H. rubicundus. The nests were on a gentle, crucial details were never published. Dol- south-facing slope. Nests were excavated phin suggested thatH. confusus was demo- using a standard technique in which baby graphically and socially polymorphic. His powder wasblown in at the nest entrances study population contained nests that to coat the sides of the burrows, which produced one, two, or three broods, com- were then carefully exposed using a prising both solitary and eusocial colonies. kitchen knife. Nests were excavated in the Eickwort et al. (1996) commented that H. morningbefore the entrances were open in confusus, presumedbyKnerer and Atwood or in the late afternoon after they were (1962) to be solitary in boreal Ontario, is closed. All adultoccupants werepreserved social in New York. These tantalizing in95% ethanol,whilebrood were placed in descriptions suggest that H. confusus may wax-lined petri dishes indented with small exhibit considerable demographic and so- chambers andbroughtbackto the lab tobe cialvariabilitywithin andbetweenpopula- raised to adulthood. When these died or tions. Understanding the ecological factors emerged as adults, they also were pre- associated with such variation is key to served in ethanol. investigating h3^otheses about the origins In addition to nest excavations, we used and extinctions of sociality in bees. pan traps to capture flying bees at six sites In this paper, we describe the colony on the Brock University campus and at the phenology and social organisation of H. contiguous Glenridge Quarry Naturaliza- confusus in southern Ontario. We studied a tionSite;pantrap siteswerewithin2 kmof mixed nesting aggregation of halictine the nesting aggregation. At each site, 30 bees, including a small number of nests of pan traps were laid out in an X or other H. confusus. We also used pan traps to space-filling pattern, alternating yellow, collectadultfemales and males throughout white, and blue pans at 10m intervals, the breeding season, in order to supple- accordingto standardprotocols (Lebuhn et ment the information from colony excava- al. 2003). Pans were set out weekly from 1 tions. We show that while H. confusus is May to 30 September 2006 at six locations. predominantly eusocial in southern On- Bees caught in pan traps were used to tario, there is evidence that large numbers determine the timing of important pheno- 146 Journalof Hymenoptera Research logical events, including nest founding, the visible oocytes, were assigned OD scores of first and second brood-provisioning or 0.1, respectively. phases, and brood emergence from the Caste assignments for females were nests. Since trapping effort was constant based primarily on seasonal activity pat- over the course of the summer, the num- terns and secondarily on body size, based bers of bees caught per week should on the assumption that in Niagara, Halictus provide a consistent estimate of bee den- confusus would exhibit the bivoltine phe- sity and flight activity. Weeks were num- nology typical of primitively eusocial ha- bered starting with 1 May 2006 as the lictines in the temperate zone (Schwarz et beginning of week 1. al. 2007). The term 'foundress' is used for — Dissections. Adult bees were measured, overwintered females that excavate bur- assessed for wear, and females were dis- rows and forage in spring. The term sected. Body size was measured in terms of 'worker' is used for Brood 1 females. After head width (HW, the distance across the workers emerge, a foundress may be widest part of the head, including the referred to as a 'queen'. A gyne is a Brood compound eyes) and length of the fore- 2 female that will overwinter and found a wing costal vein (CVL, from the stigma to nest the following spring. A 'replacement the end of the marginal cell); the head queen' is a Brood 1 worker that takes over widths of pupae were also measured. the role of queen from a dead or moribund Queen-worker size difference was calcu- foundress. lated as (queenHW-workerHW) / (queen Caste designations were assigned to HW) * 100. Mandibular wear (MW) was females caught in pan traps and nests assessed on a scale of0-5, with represent- based on the following criteria. When ing completely unworn mandibles with newly emerged from hibernation, foun- sharp teeth and 5 representing mandibles dresses are unworn, becoming progres- so worn as to be completely blunted. Wing sively more worn as they excavate nests wear (WW) was also assessed on a scale of and provision brood cells. Thus in mid- 0-5, representing wings with no damage summer, we can use wear scores to to the margin and 5 representing wings distinguish worn, late-foraging foun- with the margin completely obliterated by dresses from unworn, early workers. In nicksandtears. Atotalwear(TW)scorewas late summer, we used wear scores to obtained by summing mandibular and distinguish worn workers from unworn wingwearscoresforeachfemale. Aswings gynes. Ovarian development was not used canbe nicked duringhandling andbecause to assign caste designations, thus avoiding unworn mandibles sometimes appear teleological complications in comparisons somewhat blunt, bees were categorized as of the reproductive status of queens and worn if TW>2. workers. All adult females caught in nest Females were dissected to determine excavations, as well as the majority ofpan- mating status (whether the spermatheca trapped females (all foundresses, all gynes, was opaque, indicating that it was filled and 100 workers) were measured and with sperm, or transparent, indicating that dissected. — it was empty) and ovarian development. Interspecific comparisons. To examine in- For the latter, all developing oocytes were terspecific variation in the subgenus Sela- assigned fractional scores of %, Vi, Va, or 1, donia, the best approach would be to map indicating their size relative to a fully these data onto a phylogeny and then developed oocyte. These scores were then investigate evolutionary correlations summed to make a total ovarian develop- among the various traits (Felsenstein ment (OD) score. Females with undeve- 1988). However, in the absence of a loped or only thickened ovaries but no phylogeny, several authors have used VOLL-ME 19, Nl-mber I, 2010 147 principal components analysis (PCA) to wintered foundress. Since only two foun- quantify social variation among halictine dresses were captured in the first 3 weeks, bees and to construct hypotheses about they likely emerged from overwintering how social traits co-evolve (Michener 1974; diapause in late April and early May, but Breed 1976; Packer and Knerer 1985). mostly did not venture out of their nests Hypotheses constructed without a phylo- until mid-May when the weather became genetic framework, can then be tested more suitable. Foundresses continued tobe when an appropriate phylogeny becomes caught in pan traps for about eight weeks, available. For comparisons among Selado- with the last foraging foundress caught on nia populations, we used five variables 28 June (week 9). Most foundress foraging commonly assessed in studies of halictine and provisioning of Brood 1 probably sociality: the proportion of males in Brood occurred from weeks 4-8. 1, the number of workers per nest (or the There was a sharp increase in the number of females produced in Brood 1), number of females caught beginning in the proportion of workers with developing week 8, many of them small and unworn. ovaries, the proportion of mated workers, Large numbers of brood continued to be and the queen-worker size difference caught until week 11 after which pan trap based on head width. Values for each catches declined. Weeks 8-11 thus repre- variable were either taken directly from sented thepeakemergenceperiod ofBrood the literature, recalculated from figures in 1 and the peak worker foraging period. In the literature, or recalculated as midpoints the population as a whole, there was no ofranges. The initial PCA was based on all quiescent period between the foundress five variables, retaining factors with eigen- and worker foraging periods, as the first values > 1.0. However, since Kaiser's Brood 1 females (which were small and Measure of Sampling Adequacy (MSA) unworn) were caught on 21 June (week 8) with aU five variables had a value of only when clearly identifiable foundresses 0.55, the variable with the lowest commim- (large, worn females) were still flying. ality measure (proportion of workers ma- The first Brood 1 males were caught in ted) was dropped from the PCA. With the week 9, so emergence of Brood 1 was remaining four variables, MSA=0.77, slightly protog}^nous. Based on pan trap which exceeds the 0.6 criterion. We present samples from weeks 8-11, the proportion both factor loading scores (the degree to of males in Brood 1 was about 1.9%. which each variable influences the inferred The emergence of Brood 2 was marked factors) and communality estimates (a by a smallincrease intrap numbers ofboth reliability score which estimates the pro- males and femalesbeginning around week portion of variance in each variable that 15 (7-13 August), with the majority of is jointly explained by aU three factors). Brood 2 emerging between weeks 18-20 Note that the interspecific comparisons (Fig. 1). Week 15 was marked not only by based on the PCA are presented in the last the appearance of large, unworn females section oftheDiscussion, rather thaninthe from Brood 2, but also by the last capture Results. of smaU, unworn females deemed to be from Brood 1, suggesting that the last of RESULTS Brood 1 had emerged as adultsbyweek 15. — Colony cycle. In southern Ontario, H. The worker foraging period was mostly confusus exhibits a foraging and nesting finished by week 17, although one small, cycle t3rpical of temperate zone, eusocial worn forager was captured in week 19. halictines (Fig. 1). The beginning of the Based on pan trap samples from weeks 12- foundress foraging period was marked by 20, the proportion of males produced in the capture on 1 May 2006 of an over- Brood 2 was about 22%. 148 Journalof Hymexoptera Research DFemales Males liiLiiii JL.Lb£i 10 11 12 13 14 15 16 17 18 19 20 Week Foundresses DWorkers DGynes 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Week Fig. 1. Right phenolog)' of H. confiisus based on 2006 pan trap samples. Top: Phenolog}' of all adult bees collected in pan traps. The sharp rise in numbers of females caught in mid-summer (beginning with week 8) corresponds with the first appearance of males in week 9, suggesting that this mid-summer peak marks the emergenceofBrood 1. Bottom:Timingoffemalecasteemergencebasedonsizeandwearpatterns. Foundresses emerge in early May and continue to forage until mid-summer, slightly overlapping with females of Brood 1 (workers). Gynes firstbegin to appear in week 15. Sample size differencesbetween top andbottomgraphs are because only 100 ofthe workers caughtin pan traps was dissected. — Nest contents. Fourteen nests were ex- pupa (damaged during excavation). In cavated in total, four prior to worker week 8, three nests were excavated. The foraging and ten later in the summer. A first nest contained a queen and 3 worker single nest excavated during week 5 con- pupae; the second nest contained a queen, tained a foundress and three brood cells, one worker with worn mandibles, three comprising one provision mass with an worker pupae, and an unfinished provi- egg, one medium larva, and one earlystage sion mass; and the third nest contained a Volume 19, Number 1, 2010 149 4.8 4.6 D 4.4 J A A O D n n D o A Q A O A D D Q Q ADO D A E 4.2 O A A £ A Q O Foundresses A O A A A ABrood1females ^ 4 O A AA AA A G A OGynes A A 3.8 A A A 3.6 3.4 1.70 1.80 1.90 2.00 2.10 2.20 2.30 Head width (mm) Fig. 2. Bodysizedistributions ofpantrappedH. confusus foundresses. Brood 1 females,and gynesbased on head width and winglength (measured ascostal veinlength, CVL). queen, two workers with worn mandibles, were poorly preserved and could not be five female pupae, four male pupae, one scored). The youngestbrood were pigmen- larva that had completely consumed its ted pupae, so it is likely that older adult provisions, one provision mass with an brood had already dispersed from their egg, and one unfinished provision mass. natal nests. The average number of brood The averagebrood size of these three nests per nest had fallen to 3.4, and only 3% of was 6.3, and 22% ofthe sexablebroodwere sexable brood were males (as compared males. The latter figure is considerably with 22% in pan trap samples). Of the 23 higherthanthe estimate of2% malesbased gynes in these nests, 21 had mated and 20 on pan traps and implies that males are had noticeable fat deposits in their abdo- under-represented in pan trap samples. mens. There was no evidence that gynes The presence of workers in these nests, hadbegun digginghibemaculabelowtheir together with evident age gaps between natal nests. — younger (eggs and larvae) and olderbrood Female body size. Foundresses and (pupae), indicates that the younger brood gynes were very similar in size, and both were from Brood 2 and that within were larger than Brood 1 females; there individual nests there is a hiatus or was no indication of a body shape differ- quiescent period between Broods 1 and 2. ence between the gyne and worker castes Ten nests were excavated during weeks (ANOVA, F=8.56, df=2,147, p=0.0003; 17 and 18. None contained a live foundress Fig. 2 and Table 1). Queen and worker size (queen). Four nests contained a total of six measurementswere available fortwo nests worn adult workers. Dissections showed (bothexcavated inweek8). InNest166, the that three of these had undeveloped queen was larger than all four of her ovaries but were mated (the other three workers (adults and pupae), and the 150 Journalof Hymenoptera Research sucTcaebslser1e.fleCctastteechcnhiacraalctdeirfifsitciuclstieosfwfietmhaldeisssceacutgiohnts.inFpemaanletrsapwse.rSemaclolnesridsearmepdleassiwzoersnfoirfoMvWar>ia2noarndWmWati^n2,g and ready to lay if they contained at least one, y4-developed oocyte. Statistical comparisons of foundresses versus Brood 1 females were based on ANOVA (F statistics), Kruskal-Wallis tests (H statistics), and chi square tests. Foundresses Brood1 females Gynes Statisticalcomparison Trait (n=22) (n=100) (n=28) (foundressesvs.Brood1 females) HW (mm ± 1 sd) 2.09 ± 0.07 2.03 ± 0.08 2.09 ± 0.09 F=11.22, df=l, p=0.0011 CVL (mm ± 1 sd) 4.33 ± 0.16 4.15 ± 0.16 4.25 ± 0.17 F=21.91, df=l, p=0.0001 Proportionwith worn 13/22 (60%) 32/100 (32%) 0/28 (0%) X'=5.68, df=l, p<0.0171 mandibles Proportion with worn wings 1/22 (4%) 8/100 (8%) 0/28 (0%) X'=0.32, df=l, n.s. OD score (mean and range) 1.82 (0.5-2.75) 0.58 (0-2.25) 0.03 (0-0.1) F=21.64, df=l, p=0.0001 Proportion ready to lay 16/22 (73%) 23/100 (23%) 0/28 (0%) X'=20.50, df=l, p<0.0001 Proportion mated 18/18 (100%) 39/80 (49%) 17/28 (61%) X'=15.86,df=l,p<0.0001 queen-worker size difference was 7.2% had sperm in their spermathecae. Foun- based on head width and 9.9% based on dresses also had significantly higher OD wing length. In Nest 168 the situation was scoresthanBrood 1 females,andweremore very different. The small, worn queen was likely to have at least one oocyte ready or the same size as one worker, but smaller almost ready to lay. than four others (two worker pupae were Four types of 'workers' could be distin- notmeasured), resultingina queen-worker guished based on wear and ovarian devel- size difference of negative 4.0% based on opment, each category comprising about head width and negative 1.2% based on 25% of the total among Brood 1 females wing length. Since the above calculations caught in pan traps (Table 2). The first were based on females from only two group comprised unworn (TW < 1) nests, we also calculated the average size females with undeveloped ovaries (OD < differences for pan trapped foundresses 0.1); these were evidently newly eclosed and workers: these were 2.9% based on workers. The second group were worn head width and 4.2% based on wing (TW>2) but exhibited no ovarian develop- length. ment, suggesting that they were engaged — Wear and reproductive status. Based on in nest maintenance or foraging activities, females caught in pan traps, foundresses but were not laying eggs; these bees were sustained higher levels ofmandibularwear categorized as sterile altruists. The third than Brood 1 females (Table 1). Few fe- group were queen-like, at least in terms of males had worn wings, but one notable their readiness to lay eggs: most of these exception was the queen of Nest 168 (18.3% of all Brood 1 females) contained at (excavated in week 8), with a total wear least one fully developed oocyte ready to score of 10; this female was so much more lay, while the remainder contained at least worn than other bees examined that she one y4-developed oocyte. The remaining might have been nesting for the second group of Brood 1 females can be categor- time, having overwintered twice. ized as potentially reproductive workers, Potential forreproductionbyfoundresses exhibiting a distinct degree of wear and andBrood 1 femalesiscomparedinTable 1. some ovarian development, but not suffi- All foundresses dissected (18 from pan cienttobe ready tolayeggs. Theseworkers traps and 4 from nest excavations) had likely provision both queen-laid and some- sperm in their spermathecae, whereas only times their own eggs, and could also be abouthalfofthe Brood 1 females examined referred to as 'partial altruists'. Volume 19, Number 1, 2010 151 Table2. Comparisonofovariandevelopmentand wearinBrood 1 females collectedinpan traps. Unworn femaleshad totalwear (ITW = MW+WW) scores of or 1, whereas wornfemaleshad TW > 2. Percentages represent proportions of the total (n=93). Four categories of workers can be distinguished: 'newly eclosed' females thathavenotyetaccumulated eitherwearorovariandevelopment; 'altruists',worn, workingfemales with no ovarian development, 'queen-like' females with very high rates of ovarian development, and the remainder, with intermediate levels of wear and ovarian development, that can be referred to simply as 'workers'. Sizeoflargestoocyte Unworn Worn Total None 24 (25.8%) 21 (22.6%) 45 (48.4%) Newlyeclosed Altruists % 6 (6.5%) 10 (10.8%) 16 (17.2%) Potentiallyreproductive Potentiallyreproductive workers workers •/2 2 (2.2%) 7 (7.5%) 9 (9.7%) Potentiallyreproductive Potentiallyreproductive workers workers 'A 2 (2.2%) 4 (4.3%) 6 (6.5%) Queen-like Queen-like FuU 6 (6.5%) 11 (11.8%) 17 (18.3%) Queen-like Queen-like Total 40 (43.0%) 53 (57.0%) 93 (100%) Roughly half of Brood 1 females caught p=0.02, n=95, n.s.), even when females in pan traps were mated (this value under- assumed to be newly eclosed were ex- estimates the rate of worker mating as it cluded. includes newly eclosed individuals that Comparisons ofqueenstotheworkers in might not yet have met males). Those with their own nests suggest that queens domi- developed ovaries (OD scores > 0.25) were nated but did not completely monopolize more likely to have mated than females oviposition. Nest 166 was excavated on 22 with no ovarian development (Likelihood June 2006 (week 8), and contained a queen, ratio chi-square, G=14.46, df=l, p<0.0001; one adultworker, three worker pupae, and Table 3), and this was significant even an unfinished provision mass. The queen when newly eclosed females are excluded (TW=6) had anOD score of2.75, including from consideration (Likelihood ratio chi- three y4-developed oocytes but no fully square, G=4.978, df=l, p=<0.0257). De- developed oocytes. The adult worker was gree of ovarian development was not slightly worn (TW=3) and had probably correlated with body size (head width: collected the pollen provisions. She was p=—0.01, n=95, n.s.; costal vein length: mated and her OD score was 0.75, com- prising a single y4-developed oocyte. Evi- Table 3. Association between mating status and dently, the queen or theworker could have ovarian development in 80 H. confusus Brood 1 had a mature oocyte to lay by the time the females collected from pan traps. The minimum OD provisionmass was completed. Innest 168, scoreforafemalewithatleastonevisiblydeveloped excavated on the same day, the queen, oocyteis 0.25. Statistical analysisis given inthe text. which was the most worn bee we found Matingstatus (TW=10) had an OD score of 1.75, com- prising one fully developed and one y4 Ovarian score Unmated Mated Total developed oocyte. Of the two adult work- OD < 0.1 29 (36%) 11 (14%) 40 (50%) ers in the nest, the one smaller than the OD > 0.25 12 (15%) 28 (35%) 40 (50%) queen was worn (TW=3), was unmated Total 41 (51%) 39 (49%) 80 (100%) and had only slightly thickened ovaries. 152 Journalof Hymenoptera Research while the one larger than the queen was a ovaries caughtin midsummer, suggestthat bit less worn (TW=2), was mated and had when foundress queens die or become an OD score of 1.0, including a y4-devel- moribund, they are replaced by one of oped oocyte. Since both workers were their Brood 1 daughters, and colonies worn, they were probably both foragers, become parasocial. but only the former would be categorized Halictus confusus nests are probably as a 'sterile altruist'. founded haplometrotically (singly), as the Dissections ofpan-trappedbees revealed few (n=4) nests that we excavated in that 4/22 foundresses, 14/100 workers, spring each contained a single foundress. and 0/28 gynes had been parasitized by Haplometrotic nest founding is more likely conopid larvae, many of them large en- when gynes overwinter away from the ough to fill their host's abdominal cavity. summer nesting sites, while pleometrotic Perhaps noteworthy is the fact that two co-founding is more likely when gynes foundresses caught in pan traps in late overwinter together near the nesting site May contained conopid parasites so large (Packer 1993; Richards and Packer 1998). as to prevent any ovarian development. Atwood (1933) and Dolphin (1966) sug- Two gynes from nests excavated in late gested that H. confusus gynes overwinter August were also parasitized by conopids. away from their natal nests, and nests that we excavated near the end of August DISCUSSION contained newly eclosed gynes but no Phenology and colony social organization in evidence that these were preparing hiber- — southern Ontario. In southern Ontario, nacula. Nevertheless, pleometrosis cannot Halictus confusus exhibits a colony cycle be ruled out entirely, as we did excavate a whichinbroad terms, is typical ofeusocial, nest in which the queen was smaller than temperate zone halictines (Schwarz et al. most of her workers. In eusocial halictines, 2007). In spring, large females excavate queens control worker body size by ma- new burrows, then provision a first brood nipulating the size of larval provision that is composed mainly of workers and a masses (Richards and Packer 1994), mak- few males. Foundresses cease provisioning ing workers that are almost always smaller shortly after the summer solstice, and then than themselves (Richards and Packer are replaced as small Brood 1 females 1996), so the finding of a very worn queen emerge from their nests, and begin to smaller than some of her own workers provision Brood 2. Most individuals of suggests that she may have been a small Brood 2 are provisionedby the end ofJuly, subordinate co-foundress that outlived a emerging as adults until mid-September. larger dominant (Packer 1986; Richards Since many queens evidently survive until and Packer 1996). mid-summer when workers emerge and In H. confusus, it appears that females begin foraging, this suggests that many produced in Brood 1 may adopt one of surviving colonies become eusocial. As in three or four different reproductive op- other halictine bees (Packer 1992; Richards tions. Some Brood 1 females become etal. 1995; Paxtonetal. 2003; Richards etal. classical, sterile, altruistic workers that 2005), foundress queens likely produce the provide provisions for eggs laid by the majority of Brood 2 gynes and males. queen but produce no offspring of their Dissections indicate that queens have own. Some Brood 1 females become re- higher reproductive potential on average, productive workers, a category that com- and that workers can have high rates of prises workers that collect provisions upon ovarian development even in queen-right which a queen will lay eggs, but whose nests. The relatively large numbers of developing ovaries suggest that they also unworn workers with highly developed will lay eggs given the chance. For many. Volume 19, Number 1, 2010 153 perhaps most, ofthese 'reproductive work- diapause preparatory to becoming foun- ers', egg-laying opportunities may never dresses the following spring, a phenom- present themselves, so observations that enon known as differential diapause and m many workers have ovarian development well documented Halictus rubicundus do not necessarily translate into high rates (Yanega 1988). It would be interesting to of worker oviposition in queen-right nests compare pan trap phenologies with de- (Packer 1992; Packer and Owen 1994). tailed nesting data for several species with Nevertheless, worker maternity in queen- different colony cycles, in order to assess right nests does occur even in strongly concordance in the patterns inferred using eusocial halictines (Richards et al. 2005) so the two types of information. in H. confusus, it is likely that at least some Geographic variation in colony social orga- — reproductive workers, successfully pro- nisation. Demographic differences be- duce brood, even in queen-right nests. tween Indiana (Dolphin 1966) and Ontario The workers with queen-like ovaries likely stem from differences in the timing would almost certainly be egg-layers, and of important colony events. In Indiana, most likely were replacement queens. We foundresses emerge from hibernation as found no new H. confusus nests in mid- early as March or April and complete summer after the first emergence of work- foraging by late May or early June, with ers, so it is unlikely that workers with first brood workers emerging from mid- queen-like ovaries were Brood 1 females May to early June, second brood workers that leave their natal nests to found new emerging in mid to late July, and gynes nests in summer, either solitarily or com- emerging from mid-July to early Septem- munally (Sakagami and Hayashida 1968; ber. In Ontario, foundresses emerge from Richards et al. 2003). hibernation in late April and forage until A curious feature ofthe flightphenology aboutthe third week ofJune, with workers of H. confusus in Niagara was the small emerging from about June until the end of number of females captured in late sum- July, and gynes from mid-August to mid- mer, following emergence of Brood 2, September. This suggests that Dolphin's compared to the far greater numbers population experienced a breeding season captured in midsummer following emer- aboutthreeweekslongerthanwe observed gence of Brood 1. Several explanations in Niagara in 2006. In Indiana, many present themselves. First, gynes might colonies produced two worker broods. have been imder-represented in pan traps This seems unlikely for our Ontario popu- relative to workers, due to changes in lation, as pan traps suggested that the flower and forage availability. Pan traps majority of Brood 1 workers emerged are known to capture relatively fewer between weeks 8 and 13, a six weekperiod foragers when flower availability increases that is only slightly shorter than the period (Roulston et al. 2007). Pan traps may encompassing most foundress foraging therefore be less attractive to gynes (and activity between weeks 3 and 9. However, males) because they are not active provi- the intriguing, small peak in captures of sioners, and because flower availability females around weeks 14 and 15, might may be higher after midsummer than have signaled the emergence of a second- before. It is also possible that the pattern ary worker brood. We did capture some oflower gyne thanworker densities is real. small, unworn females at this time, which If so, then one explanation would be high we categorized as gynes, but which were rates of colony failure prior to worker possibly workers. The ability to interpolate emergence (Richards and Packer 1995a). a second worker brood in areas with long Another possibility is that some Brood 1 enough breeding is well known in Lasio- females leave their natal nests to enter glossum malachurum, which produces a