Ent. Tidskr. 134 (2013) Critical resources of a mining bee Critical floral resource levels and nesting biology of the mining bee Andrena marginata (Hymenoptera: Andrenidae) Magnus sTEnMark stenmark, M.: Critical floral resource levels and nesting biology of the mining bee An- drena marginata (Hymenoptera: andrenidae). [Hur många blommor behövs för att föda en population av guldsandbi Andrena marginata (Hymenoptera: Andrenidae), och hur ser deras bon ut?] – Entomologsk Tidskrift 134(3): 135-148. uppsala, sweden 2013. Issn 0013-886x. as a result of changes in farming practices, the mining bee Andrena marginata (andreni- dae) is declining over most of its distribution in Europe, including sweden. The species is specialized on the pollen of Dipsacaceae. The present study explored the role of pollen availability in relation to the nesting biology of this bee at a site in sweden. The food-plant at the site was Succisa pratensis. an average nest (11.2 cells) was provisioned in ca. 20 days and contained about 1,300,000 pollen grains. Male cells were produced late in the season toward the lower parts of the nests that were progressive (i.e. constructed top-downwards). an average cell contained approximately 116,000 pollen grains. The bee population of A. marginata utilized 44.0% of the pollen production of the food-plant. Each nest required the whole-season production of ca. 6.9 pollen-producing plants (53.8 inflorescences). a pollen budget was modelled to predict changes in the bee population size during a hypothetical decrease or increase of the food-plant population. accordingly, a very small bee popula- tion (≤20 female individuals) was found to require up to 184 food-plant individuals. This pollen budget has potential to predict carrying capacity of bee populations in other habitats. Magnus Stenmark, Faunistica, Kungsbäcksvägen 56, 802 67 Gävle. E-mail: [email protected] The mining bee Andrena marginata is a rare cluding butterflies (Dempster et al. 1995, gutiér- and charismatic species. The bee requires open rez & Thomas 2000) and solitary bees (Minckley sandy areas for nesting and the aggregations can et al. 1994). a correlation between resource avail- reach several hundred nests. The species is oli- ability and reproductive success has been shown golectic on Dipsacaceae, and most known swed- for several solitary bee species (rosenheim et al. ish populations are utilizing Succisa pratensis 1996). Moreover, this insect group exhibits ad- for pollen. In at least two areas in sweden the aptations to food limitation such as a facultative bee population tend to favour Knautia arvensis bias towards the offspring sex that requires fewer for pollen-collecting and on these sites the popu- resources (Torchio & Tepedino 1980, Frohlich & lation has adapted the flight season to the earlier Tepedino 1986, kim 1999, Paini & Bailey 2002). phenology of Knautia arvensis. In one site near The pollen removal in plants with generalized Lund in the province of skåne the bee Andrena pollination systems can be traced to a number of marginata population has adapted to collect pol- abiotic and biotic factors. abiotic factors can be len from Scabiosa canescens. pollen lost through wind and rain and biotic fac- The amount of nectar or pollen available limits tors are e.g. contamination from flower-visitors, the population of many insects (Price 1984), in- pollen-feeding and pollen-collecting by flower- 135 Magnus Stenmark Ent. Tidskr. 134 (2013) visitors such as female solitary bees that collect pollen for provisioning brood cells. a few studies have explored specialized bees in terms of the degree of pollen utilization, de- fined as the proportion of the total pollen pro- duction that is packed in brood cells to feed off- spring. Müller et al. (2006) studied a number of specialized solitary bee species and suggested a general degree of pollen utilization of ca. 40%. Other solitary bee species have been reported to utilize about 68% of the pollen produced by He- lianthus annuus (Minckley et al. 1994), 96% of the pollen in Campanula rapunculus (schlind- wein et al. 2005), and 33% of the pollen produc- tion in Knautia arvensis (Larsson 2005). These studies have based their estimates on brood cell pollen requirements compared to the total pollen produced by the local plant population. so far, Figure 1. A female of Andrena marginata with pollen load of no studies have estimated the degree of utiliza- Succisa pratensis. tion and incorporated it into calculations of bee En hona av guldsandbi med fyllda skopor med pollen av population size. ängsvädd. Figure 2. A typical nest aggregation of Andrena marginata, below on the right side a nest entrance equipped with a tumulus. En typisk aggregation av bon som hör till guldsandbi. Det infällda fotot visar en boingång med den typiska tumulus. 136 Ent. Tidskr. 134 (2013) Critical resources of a mining bee among hymenopterans, the allocation of pa- Materials and Methods rental investment to offspring is a decision by Study system the female: she either uses stored sperm to fer- Andrena marginata Fabricius (Fig. 1) is one of tilize the egg (for female offspring) or does not several solitary bee species highly specialized (for male offspring). Because offspring show a in the collection of pollen from Dipsacaceae sexual size dimorphism in most wasps and bees (Westrich 1989). The nests of A. marginata are (i.e. females are considerably larger than males), constructed as burrows and the females often the provision mass per brood cell is larger for prefer exposed sandy patches such as river em- female offspring. as predicted by Fisher (1930), bankments, disturbed patches created by cattle, the investment sex ratio, (i.e. the relative re- paths or road banks (kocourek 1966, Westrich source required for the total male vs. female off- 1989). The nests entrance is surrounded by a spring), has been shown to be 1:1 in several bee tumulus (Fig. 2), which at the study site con- species (Tepedino & Torchio 1982, Frohlich & tained sand of a slightly different colour than Tepedino 1986, Johnson 1988, Helms 1994). In the soil surface, a feature which makes nests these cases, the numerical sex ratio, i.e. the rela- easy to locate and count, even if they have been tive number of male vs. female individuals, is abandoned. The nest architecture, brood cell biased towards the smaller-investment sex (e.g. construction, larval development and hiberna- Visscher & Danforth 1993). tion have hitherto not been described. The flight Most bee taxa utilize pollen as resource for range of A. marginata is unknown, but other their progeny. specialized bees utilize pollen specialized mining bee species of about the from a few or even a single species of plant. same size tend to have a home range up to 500 m among solitary bees, each female constructs her (Eickwort & ginsberg 1980, gebhardt & röhr own nest, gathering pollen and distributing it in 1987, gathmann & Tscharntke 2002). separate brood cells for each offspring. Thus, the The main pollen source for A. marginata resources required for bee progeny may be mea- in nW and C Europe is Succisa pratensis and sured by quantifying pollen provisions in nests, Knautia arvensis, but A. marginata individuals and the available resources may be measured may also visit Scabiosa columbaria, S. ochro- by quantifying the number of pollen grains pro- leuca and S. canescens (Westrich 1989, Pekkar- duced over one bee generation within the flight inen 1997). among these potential food-plants area of the target bee population. of A. marginata, only S. pratensis was present in The aim of the present study was to develop the study area. The inflorescences of S. pratensis the concept of a “pollen budget analysis”: a are nectariferous, and attract a broad spectrum tool that potentially can be used for predicting, of nectar-feeding, pollen-feeding and pollen- maintaining and restoring viable wild-bee popu- collecting flower-visitor taxa (knuth 1898). The lations. To explore the relationship between the herb S. pratensis is gynodioecious (i.e., popula- population sizes of an obligate pollen-feeder tions are composed of females and hermaphro- and its food-plant, I chose a system involving dites). a variable (0-30%) proportion of female the plant Succisa pratensis and its specialized individuals have been reported from different bee Andrena marginata. I first present the hith- areas in Europe (Müller 1883, knuth 1898, kay erto undescribed nesting biology of the species 1982). Only hermaphroditic individuals produce A. marginata, then estimate the number of pol- pollen. len grains and food-plant individuals required to Succisa pratensis is found in calcareous open meet the needs of one brood cell, one nest and grasslands, moist meadows and pastures, and the bee population as a whole. Finally, I explore has suffered a notable decline due to changes how the pollen budget analysis can be used to in land management (Ekstam et al. 1998, Ver- predict how a bee population size is affected by geer et al. 2004). as a result, the specialist bee changes in the population of the pollen-produc- A. marginata has declined in most areas in C ing plant. and nW Europe (Westrich 1989, Falk 1991, Pe- kkarinen 1999), and is currently red-listed in a number of European countries. 137 Magnus Stenmark Ent. Tidskr. 134 (2013) Figure 3. The site Stridsvagnshöjden in the county of Halland hosts one of the largest nest aggregations in Sweden of Andrena marginata. The site consists of large open sandy areas surrounded by wetlands that deliver numerous flowers of Succisa pratensis – the host plant. Lokalen Stridvagnshöjden vid Tönnersjömålet utanför Mästocka hyser en av Sveriges största boaggregationer av guld- sandbi. Här finns stora öppna sandytor som omges av våtmarker som producerar mycket ängsvädd. Study site a hill, covering ca. 27 ha. The plant population The study was performed in august and sep- was isolated by wetlands from other flowering tember 2003–2004 in an open heathland land- individuals of S. pratensis by >550 m. The nests scape in sW sweden (56º40’29”n, 13º15’28”E, of A. marginata were scattered and in small ag- Veinge-Tjärby parish). The heath is named gregations in the well-drained sandy soil on the Tönnersjömålet and the specific site is usually top and on the slopes of the hill; thus the nests referred to as stridsvagnshöjden (Fig. 3). This were located in a central area in relation to the heath has historically been used by farmers as a food-plant population. The greatest distance be- pasture, but since 1940 the area has been man- tween an A. marginata nest and the most periph- aged by controlled burning annually and used as eral food-plant was ca. 290 m. a military training area. This activity regularly creates patches of exposed sand. The vegeta- Nest excavation tion is dominated by Calluna vulgaris and with To quantify pollen provisions and to document presence of several herbs favoured by the man- the nest architecture, nests of Andrena margina- agement regime, e.g. Arnica montana, Succisa ta were excavated. For excavation (Fig. 4), nests pratensis and the shrub Genista pilosa. in complete or almost complete structure were The study population of the food-plant S. pra- chosen on the basis of the behaviour and shape tensis was distributed in patches on and around of the female. Prior to excavation, melted tin was 138 Ent. Tidskr. 134 (2013) Critical resources of a mining bee Figure 4. The procedure of excavating solitary bee nest requires patience and accuracy. A single nest usually demands 1-3 hours. It is important to carefully choose site and spot for excavation in order to minimize the negative effect on bee the population. Utgrävning av solitärbibon är en aktivitet som kräver både tålamod och noggrannhet. Utgrävning av ett bibo kräver vanligen 1-3 timmar. Det är viktigt att välja lokal och angreppsätt med omsorg för att minimera den negativa påverkan som en utgräv- ning har för en population. poured in each nest. Each nest was carefully ex- cavated during 1-3 h. The tin structure served as a central reference point for those branches that were already filled up with soil and therefore not reached by the melted tin. The brood cells were classified based on the contained offspring as: large containing a postdefecating larva where feces were present in the cell and the larva had a distinctly wrinkled segmentation and there were no (remaining) pollen provision; medium containing a predefecating larva where no feces were present and the larva had no wrinkled seg- mentation and there were no (remaining) pol- len provision; small still containing some pollen Figure 5. A small larva, just a few days old, of Andrena provision with a feeding larva that was longer marginata. The larva remains on the top of the pollen load than the diameter of the pollen provision; ini- where it was deposited as an egg by the female bee. tial containing a larva feeding, or close to the En larv av guldsandbi som alldeles nyligt kläckts från sitt commencement of feeding, on the pollen ball, ägg. Larven sitter på toppen av pollenbollen på samma and with the length of the larva being smaller plats där dess ägg deponerades av honan. 139 Magnus Stenmark Ent. Tidskr. 134 (2013) than the diameter of the pollen provision; egg The pollen requirements for an average nest, containing an egg; and unclosed representing a p , was calculated as: nest cell which not yet had been sealed by the female bee and still physically connected to the main p = n × p + n × p nest f f m m shaft. The brood cell angle of inclination, depth where, from soil surface and distance to main shaft of the tunnel were measured. n = the average number of female cells per nest, f To quantify the brood cell pollen require- p = the average number of pollen grains per female f ments, cells that contained pollen provisions cell, n = the average number of male cells per nest, with unhatched eggs were collected in 70% m p = the average number of pollen grains per male ethanol. In the laboratory, the provision samples m cell. were sonicated (ks101, kerry ultrasonics Ltd.) for 35 min and quantified by counting a known Pollen-collecting bouts volume proportion of the sample under a bin- To estimate the per-day foraging efficiency, the ocular microscope (modified from kearns & foraging time of a group of adult females were Inouye 1993). I used the average of 3 indepen- recorded during one full day (on 13 aug 2004, dent volume proportions from each provision between 08.10 a.m. and 18.15 p.m.). The pollen sample. amount gathered during a pollen-collecting bout To determine brood cell sex distribution with- was measured by sampling female bees at nest in nests, a number of larvae were collected for site during arrival from collecting pollen. These subsequent artificial rearing. These reared lar- bees were stored in 70% ethanol and sonicated vae differed in male vs. female sex significantly for pollen release from the bee’s scopae. The in 3 characteristics: age order in nest, distance to bees were then moved to a new jar of 70% etha- main shaft and cell angle of inclination. The age nol and sonicated again. This process was re- order was given during excavation as a within- peated 3 times and the bees were then put under nest relative measure of when the cells were a binocular microscope for visual counting of constructed. The age order is easy to see dur- any remaining pollen grains. The pollen content ing excavation as cells usually can be ordered in the jars was quantified as described for the from unclosed, egg, initial (Fig. 5), medium pollen provision samples. and large representing cells with increasing age. This order was set relative to the total number of cells. For example, the second constructed Food-plant population characteristics cell in a nest containing 20 cells was given the To estimate total pollen production, a number relative order 0.10. Distance to main shaft was of food-plant characteristics were recorded. The measured during excavation and is illustrated in population size of the food-plant was measured figure 8. The cell inclination was also measured on a single day (9 sep 2004) by counting all in- in the field and represents the cell inclination dividuals flowering or that had flowered during compared to a horizontal line. The average val- the current season. For hermaphroditic morph ues of these three characteristics were used to frequency determination, I registered the sex estimate the sex of brood cells containing pol- of 260 random individuals of S. pratensis. ad- len provisions, as most of those cells only were ditionally, I randomly sampled hermaphroditic measured and not object to rearing. a cell was individuals (females were excluded as they do estimated as male (female) if it showed >0.52 not produce pollen) in late-season, measuring (<0.38) (relative order 0-1), <2.06 (>5.23) (cm) the total number of inflorescences produced and <11.28 (>17.31) (º) for each of the 3 charac- over the season and flowers per inflorescence. teristics, respectively. Brood cells with ≥2 male anthers were collected on a single day (12 aug (or female) characteristics were considered as 2004) and placed in 70% ethanol for pollen con- male (or female), while brood cells lacking ≥2 tent quantification, as described for the pollen characteristics or with any conflicting character- provisions. The pollen production in the plant istic were not estimated to sex. population, P, was calculated as: p 140 Ent. Tidskr. 134 (2013) Critical resources of a mining bee p were estimated by the plant morphology nest P = S × D × i × f × a × a analysis and the nest excavation data, respec- p herm fl p tively. In this study UT was not empirically spec. bee where, measured but calculated from the formula. S = plant population size (number of individuals), D = proportion of hermaphrodites in the popula- Statistical analyses herm tion, statistical analyses were performed with Minit- i = number of inflorescences per hermaphroditic in- ab 14.0 (Minitab Inc. 2000). The anderson-Dar- dividual, ling test was used to test normal distribution. For f = number of flowers per hermaphroditic inflores- comparing various characters of male vs. female cence, brood cells t-tests were used. Linear regressions a = number of anthers per flower (4 in S. pratensis), afl = number of pollen grains per anther. were used to test the effect of brood cell depth on p amount of pollen provision, and the effect of cell The available pollen resource order on cell depth, the lateral tunnel length and The amount of pollen remaining available for cell angle of inclination. LsD-test was used to flower-visitors, P, was calculated as P correct- test differences between brood cell categories in a p ed for number pollen grains that had not been their relation to cell depth from soil surface. For removed from anthers. The latter pollen amount mean values standard error (±sE) is provided. was quantified by counting pollen grains in withered anthers under a binocular microscope Results (sampled in the field on 9 sep 2004). Nest architecture a total of 32 nests were excavated. The nest-en- The bee population size trance was usually surrounded by a distinct tu- The utilization of pollen removal by A. margina- mulus (an asymmetric heap of sand). The tumu- ta may be calculated by measuring the visitation lus was left open when the female was foraging, frequency by all flower-visitors (see Larsson and was sometimes closed when the bee was 2005). By correcting the visitation frequency by inside the nest, and was generally closed during per-visit effectiveness (for each species or for unfavourable weather and during the night. The each group of species), the removal of pollen by main shaft of the tunnel averaged 24±0.81 cm in each species is obtained. The utilization of A. length (n=32 nests), and went downwards irreg- marginata is the contribution to pollen removal ularly at 90-100º from the soil surface. The di- relative to the total pollen removal by all flower- ameter of the main shaft was 5.4±0.12 mm (n=6 visitors. The bee population size, in this study nests). The cells were 10.9±0.26 mm (n=6 cells) defined as the number of reproducing female in length and 6.3±0.12 mm (n=10 cells) at their individuals, R, may be calculated as broadest part, forming a non-symmetric ellipse P × UT more flattened at the bottom than at the top. The R = a p spec. bee cell angle of inclination was 15.4±0.81º (n=158 nest cells). The average nest had 11.2±1.4 cells where, (n=32). The stages of the cells and their spatial UT = the utilization of pollen by A. marginata, distribution within the nest revealed that the nest spec. bee p = the pollen requirements for one average nest. was constructed progressively, (i.e. from the top nest In this study, R was estimated by counting on downwards, Fig. 6, 7a). The distance from the a single day (8 sep 2004) the number of nests cell to the main shaft indicated a decrease as that were active or had been active during the the nest construction advanced (Fig. 7b). The season. This count also included nests that had nest resembled an inverted straight cone, hav- been abandoned before completion; however ing its base 16.1±0.42 cm (n=32) below the soil such nests were also included in the excavation surface (Fig. 8). This type of nest has been de- and will therefore not affect the result. P and scribed as linear-branched (Malyshev 1926). a 141 Magnus Stenmark Ent. Tidskr. 134 (2013) m) d 1.0 a within nest (c bc cd within nest (cm)00..86 yrp2=<=000...104201+10.68x h b h e dept a a ve dept0.4 ativ elati0.2 el R R 0 large medium small initial egg unclosed 20 b y=6.24-2.37x m) r2=0.05 Figure 6. The variation in brood cell content (large larva, c p=0.001 medium larva, small larva, egg or unclosed cell) (N=306) aft ( 15 h shows that the nests are constructed from top downwards. n s Bars not sharing a letter are significantly different at p<0.05 mai 10 (LSD-test). Error bars: ±SE. o e t Variationen av födocellsinnehållet (stor larv, intermediär nc 5 a larv, liten larv, ägg eller icke försluten cell) (N=306) visar att st boet anläggs från toppen och nedåt. Kategorier som inte Di 0 delar en bokstav skiljer sig signifikant med p<0,05 (LSD- test). Linjerna visar standardfelet. 0 0.2 0.4 0.6 0.8 1.0 Relative order within nest Figure 7. The relationship between brood cell order within the nest and – a) the depth of the brood cell and – b) the distance from the brood cell to the main shaft. Kopplingen mellan födocellernas inbördes ordning – a) och djupet och – b) avståndet mellan födocellen och boets hu- vudschakt. Development were therefore pooled in further analysis. an The egg of A. marginata is white and shiny and average cell contained 116,465±4,286 pollen 2.5±0.07 mm (n=5 eggs from different nests) in grains (n=49). The amount of cell provision length. The egg is laid on the top of the pollen did not increase with increasing distance to the provision and attached to the pollen mass with main shaft of the tunnel, with depth, or order of one end. after consumption of the pollen provi- the brood cell (all p>0.05). This indicates that sion, the larva defecates and develops wrinkled the female tend to provision the same amount segmentation and turns into a prepupa. The spe- to male cells regardless when or where in the cies hibernates as prepupae. nest the cell is built. However, only estimated male cells were sampled. Probably, female cells Pollen provision are provisioned with more pollen and as female Cells were provisioned with 111,910±6,091 cells were not sampled we do not know whether pollen grains (n=23) for male estimated cells female cells are provisioned differently with in- and 120,494±6,017 pollen grains (n=26) for creasing distance to the main shaft of the tun- offspring not estimated to sex (t-test: t=-1.00, nel, with depth, or order of the brood cell. an p=0.321). no pollen provisions were estimated average pollen provision represents pollen that to belong to a female cell due to their character- was produced by ca. 2.0 hermaphroditic inflo- istics in the nest – all sampled pollen provisions rescences or 0.3 individuals of S. pratensis dur- had male cell characteristics. all cell provisions ing a complete season. 142 Ent. Tidskr. 134 (2013) Critical resources of a mining bee Foraging and pollen load a Females foraged between 08.10 a.m. and 18.15 p.m. no females were found to provision more b than one nest simultaneously. There were no in- dications (e.g. rubbed, reduced wing surface of females initiating a new nest) of females con- structing a second nest. Pollen-foraging females spent on average 74±6.7 min (n=8) outside the nest per pollen-collecting bout. Females return- ing to their nests carried 8,861±1,769 pollen grains (n=14). Consequently, an average cell required 14 pollen-foraging trips for its comple- tion. an average nest (11.2 cells) consumed 1,299,029 pollen grains, thus requiring ca. 150 collecting bouts. under favourable weather con- ditions, females were able to make 7-9 pollen- foraging trips during one day. Therefore, about 20 days with good weather conditions were needed to provision an average nest. Artificial rearing and sex ratio Out of the collected brood cells (n=111), 51 individuals (10 males and 41 females) of A. marginata were successfully reared in the labo- ratory. The sex ratio was 19.6% males. The re- Figure 8. Schematic architecture of the nest (a) and one maining cells were reared unsuccessfully as the brood cell (b) of Andrena marginata. The number, the length and the relative position of the lateral burrows rep- larvae were killed by fungal or bacterial parasit- resent an average based on 32 nest excavations. A grid ism (n=57) or starvation (n=3) and sex deter- illustrates the lateral tunnels that have been re-filled with mination could not be done. Female cells were soil by the female bee. located further away from the main shaft of the Schematisk arkitektur av boet (a) och av en födocell (b) för tunnel when compared to the male cells (t=3.77, guldsandbiet. Antalet, längden och den relativa placeringen p=0.001). The angle of inclination of the female av sidogångar representerar ett genomsnitt baserat på 32 cells was steeper when compared to male cells utgrävda bon. Mörkade sidogångar indikerar att de fyllts (t=2.37, p=0.031). Moreover, by measuring the igen av honan. Table 1. Results of comparative analysis of male, female and parasitized cells of A. marginata artificially reared in the laboratory. Standard error is indicated in parenthesis. Resultaten av en jämförande analys av hanliga, honliga och parasiterade guldsandbiceller uppfödda i lab. Inom parantes finns standardfelet. Brood cells Male Female Parasitized N 10 41 57 Order (no.) in nest1/ Ordningsnummer i boet 6.29 (1.23) 5.66 (0.65) 8.53 (0.82) Relative order within nest2/ Relativ ordning i boet 0.52 (0.04) 0.38 (0.04) 0.47 (0.04) Depth (cm) from soil surface/Djup (cm) under markytan 19.38 (0.92) 19.89 (0.46) 21.31 (0.41) Relative depth within nest/ Relativt djup under markytan 0.24 (0.09) 0.44 (0.04) 0.45 (0.04) Distance (cm) to main shaft2/ Avstånd (cm) till huvudgången 2.06 (0.59) 5.23 (0.60) 4.73 (0.45) Inclination (°) of the cell2/ Lutning (°) på celler 11.28 (2.00) 17.31 (1.57) 16.56 (2.06) 1 parasitized vs. female+male cells significantly different with P<0.05 2 male vs. female cells significantly different with P<0.05 143 Magnus Stenmark Ent. Tidskr. 134 (2013) Table 2. Population characteristics of the food-plant S. pratensis included in the pollen budget. Populationsspecifika data för värdväxten ängsvädd som användes i pollenbudgeten. Population characteristics N Mean Min Max SE Inflorescences per individual (no.)/ Antal blomhuvden per individ 22 7.8 1 42 1.80 Flowers per inflorescence (no.)/ Antal blommor per blomhuvud 51 84 52 122 2.29 Pollen grains per anther (no.)/ Antal pollenkorn per ståndare 22 168 105 236 7.82 Pollen left in withered anthers (grains per anther)/ Pollen kvar i vissna ståndare (pollenkorn per ståndare) 100 6.0 0 28 0.74 relative order within the nest (=order/number of estimated, by counting nest entrances, to include cells in the nest) (t=2.57, p=0.025), female cells 580 pollen-collecting females. This bee popula- were found to have been constructed earlier and tion required 3,998 hermaphroditic individuals thus must be positioned closer to the soil surface (or 4,014 if adjusted for the presence of female when compared to male cells. However, the par- individuals) of S. pratensis. This empirical bee ticular cells brought for artificial rearing (Table and food-plant relationship was extrapolated 1) did not show that female cells were closer to over hypothetical sizes of the bee population. soil surface when compared to male cells. This By assuming a similar degree of utilization is thought to be an effect of the limited num- (44.0%), the respective required resource was bers of sampled cells (n=51 of non-parasitized calculated for a range of sizes of the bee popula- cells) and the fact that these sampled cells were tion (Table 3). a small bee population number- brought from late season. When comparing the ing only 20 females were found to require up successfully reared cells (male and female in one to 184 individuals of the food-plant S. pratensis group) with parasitized cells (Table 1), I found (Table 3). that the parasitized brood cells were produced later (t=-2.77, p=0.007), and were placed deeper Discussion below the soil surface (t=-2.62, p=0.010). The results of the present study provide an esti- mation of the critical pollen resources required Pollen utilization for one nest and for an entire bee population as The food-plant population characteristics (Table a whole. The model corrects for pollen removed 2) were used to calculate total pollen produc- by other flower-visitors. Thus, such pollen bud- tion. The plant population consisted of 4,014 get analysis may open a new tool-oriented ap- individuals and had a hermaphroditic morph proach in conservation, where the relationship frequency of 99.6%. The number of anthers in between specialized solitary bees and their S. pratensis flowers is fixed and includes 4 an- food-plants is expressed in the currency of easy- thers per flower (knuth 1898). The solitary bee counted plant individuals. Further variables of A. marginata ♀ was found to utilize 44.0% of all the food-plant (e.g. density) can be easily ex- pollen produced, 52.4% was removed by other plored and included in the model. flower-visitors and 3.6% of the pollen remained in anthers. When considering only pollen avail- Pollen budget able for flower-visitors, A. marginata utilized Pollen budget data (Table 3) allow the prediction 45.6%. of bee population size in response to a change in the food-plant population. For example, the pol- Bee vs. food-plant population size len budget in this study predicted that 346 indi- an average nest of A. marginata represented viduals of the food-plant can support 50 nesting the whole-season pollen production of 22.7 in- female bees. Most of the current populations of florescences or 2.9 hermaphroditic S. pratensis A. marginata in sweden are small and probably individuals. However, when all other pollen re- include less then 50 reproducing females (L. a. moval agents were included, the average nest nilsson and M. stenmark, personal observa- required 53.8 inflorescences or 6.9 hermaphro- tions). For these already small bee populations, ditic individuals. The entire bee population was a small reduction in food-plant numbers or en- 144