Seasonality and alternative floral resources affect reproductive success of the alfalfa leafcutting bee, Megachile rotundata

Introduction

The alfalfa leafcutting bee (ALCB), Megachile rotundata (F.) (Hymenoptera: Megachilidae), is one of the most economically important managed bees in North America, particularly for its use as the major pollinator of seed alfalfa (Medicago sativa L.) (Pitts-Singer & Cane, 2011). ALCB is a solitary bee whose females construct linear series of brood cells within tunnels, including artificial versions provided for management. Each cell is lined by the female with cut leaf pieces and provisioned with pollen and nectar for a single larva. Alfalfa seed farmers rear adult bees for release each season from brood cells produced the previous year. Bee population losses during growing seasons and overwintering result from multiple causes often related to management practices, including diseases like chalkbrood, parasitoids and predators, dispersal, and thermal stresses (James & Pitts-Singer, 2013; Donahoo et al., 2021). In the U.S., fewer than 50% of bees released for pollination are replaced during reproduction, requiring growers to import costly bees from Canada in some years (Pitts-Singer, 2008; Pitts-Singer & Cane, 2011). Thus, improved management strategies are needed that promote the long-term health and sustainability of this pollination system.

Most research aimed at increasing the sustainability of ALCB populations has focused on the management of natural enemies (e.g., Whitfield & Richards, 1985), mitigation of disease (e.g., James, 2005), and optimizing rearing strategies (e.g., Richards, Whitfield & Schaalje, 1987; Pitts-Singer & James, 2009; Yocum et al., 2010; O’Neill et al., 2011). Another important consideration for sustaining bee populations is that bee reproductive success also depends on the quality and quantity of available floral resources (Roulston & Cane, 2000). Typical ALCB management in the U.S. involves releasing high densities of bees to increase alfalfa seed set, however, this practice quickly depletes floral resources (Strickler & Freitas, 1999; Pitts-Singer, 2008, 2013), limiting late-season brood production. Thus, management decisions can result in tradeoffs between seed yield and bee reproduction (Strickler, 1996).

One potential strategy to support pollinators in agroecosystems is to plant alternative, non-crop flowering plants to enhance and/or extend floral availability beyond the short-lived bloom of monoculture crops (e.g., Blaauw & Isaacs, 2014; Williams et al., 2015; Burkle, Delphia & O’Neill, 2020; Klatt, Nilsson & Smith, 2020; Graham et al., 2020). In alfalfa, as floral resources decline in late summer, female bees may be forced to fly farther afield. This decreases their foraging efficiency and increases their own energy needs, resulting in production of a higher ratio of sons to the daughters (Peterson & Roitberg, 2006a) that are the next year’s primary pollinators. Additionally, when faced with lower alfalfa floral resource levels, they produce fewer and smaller female offspring (Peterson & Roitberg, 2006b), reducing not only their own reproductive success but potentially that of the following generation because smaller females produce smaller eggs (O’Neill, Delphia & O’Neill, 2014). Adding wildflower strips that provide an alternative source of pollen and nectar might help alleviate this problem, provided those species (1) flower after peak alfalfa bloom so as not to impact alfalfa pollination via competition for pollinators (Lander et al., 2011; Nicholson et al., 2019) and (2) do not decrease alfalfa seed purity levels should plants later spread into alfalfa fields.

Wildflower plantings have the potential to provide food resources to ALCBs because females are known to forage on a wide variety of plant families (Small et al., 1997). A study in Montana found that non-alfalfa pollen comprised over 40% of the pollen grains carried by females from mid-August to early September (O’Neill & O’Neill, 2011), during a time when fewer alfalfa flowers are present. These alternative pollens came from diverse weed species from five plant families (Asteraceae, Brassicaceae, Chenopodaceae, Fabaceae, and Scrophulariaceae) that grew within or adjacent to alfalfa fields; yellow sweetclover (Melilotus officinalis (L.) Lam.) and a mustard species (Brassicaceae) were the two most abundant pollens collected. However, the potential value of these species as food resources for bees in alfalfa seed fields conflicts with weed management practices. Therefore, the introduction of late-blooming, non-weed flowering plants seems promising.

Using supplemental floral resources to extend the seasonal availability of pollen and nectar in alfalfa seed fields could lead to greater numbers of offspring, and/or larger and healthier offspring (Torchio, 1985; Minckley et al., 1994; Strickler & Freitas, 1999), provided that the alternative pollens are of sufficient nutritional value (Vaudo, Dyer & Leonard, 2024). In solitary bees, body size is influenced by the amount of pollen and nectar they receive as larvae from their mothers (Klostermeyer, Mech & Rasmussen, 1973; Roulston & Cane, 2000). However, adult female condition has been shown to decrease later in the summer (O’Neill, Delphia & Pitts-Singer, 2015) and can also influence offspring body size and sex ratios due to reduced provisioning efficiency (Sugiura & Maeta, 1989; Seidelmann, 2006). Previous studies indicate that larger ALCB females start the nesting season with proportionally greater lipid stores, a likely indicator of offspring condition (O’Neill, Delphia & O’Neill, 2014). These lipid stores decline by about a third within 1 week after adult females begin nesting activity in fields, likely because they are used to produce eggs, which also decline in size as the nesting season progresses (O’Neill, Delphia & Pitts-Singer, 2015). These and other age-related declines in the condition of adult bees (e.g., O’Neill, Delphia & Pitts-Singer, 2015) could negate the potential benefits of additional food resources if females are unable to capitalize on their availability. Alternatively, providing food resources late in the summer could support newly emerging 2^nd generation bees, which are common in seed production fields in many locations in the western U.S. (Johansen & Eves, 1973; Pitts-Singer & Cane, 2011). Whether late-season supplemental floral resources (and the quality of those resources) can enhance bee reproduction or offspring condition (e.g., body size or lipid stores) was the hypothesis that we aimed to test.

To explore the potential benefits of using wildflower strips for increasing ALCB reproductive success, we established alfalfa field plots with and without adjacent flower strips on a research farm in southcentral Montana, USA. Our objectives were to evaluate, during two field seasons, the effects of adding floral resources that bloom after peak alfalfa bloom on (1) ALCB foraging behavior, (2) ALCB reproductive success (including offspring production, survival, and condition), and (3) alfalfa seed yields. We hypothesized that addition of wildflower strips that flower after peak alfalfa bloom would lead to increases in ALCB offspring production without negatively impacting alfalfa seed yields. Our experimental design also allowed us to examine seasonal changes in offspring condition over the course of the nesting season. Due to unavoidable issues with flowering phenology and drought that limited our ability to fully address our initial objectives, we report results from one summer of the field study, along with a field-cage study that aimed to overcome some of the limitations of the open-field design. In the field-cage study, nesting bees were provided either alfalfa, wildflowers, or a combination of the two. We modified and extended our objectives to evaluate the effects of alternative floral resources on (1) ALCB reproductive success and (2) ALCB pollen provisioning, as well as (3) the effect of provision quality on offspring condition.

Materials and Methods

Study site. This research was conducted in southwest Montana at Lutz Farm (45.8132°N, 111.0550°W), an Agricultural Experiment Station operated by Montana State University (MSU) north of Bozeman, MT. The farm is 650 acres (ca. 0.80 × 2.83 km) and broken up into smaller fields of varying sizes planted primarily with wheat, barley, and other small grains. Because the surrounding crops were primarily wind-pollinated, it limited available floral resources for bees to those we provided in our experimental design.

Field study: Do late-season supplemental wildflower resources enhance ALCB reproduction and offspring condition?

Alfalfa plots. In early May 2016 we established two 0.05-ha (23 m × 23 m) alfalfa plots (variety ‘Cooper’, Seed Source, Inc., Toston, MT) at each of six sites (distributed in a 2 × 3 grid) at Lutz farm. Sites were at least 400 m apart to reduce the likelihood that bees would move between them (Tepedino, 1983; Gathmann & Tscharntke, 2002). To hasten blooming for sampling in year 1 (2016), we transplanted alfalfa plugs grown in the greenhouse in late winter into one randomly selected plot at each site. We hand-seeded the second plot for sampling in year 2 (2017). All alfalfa (plugs and seeds) were planted on 0.91 m centers to ensure plants received adequate ground moisture as there was no irrigation (Mueller, 2008).

In year 1, alfalfa bloom was delayed by 3 weeks, likely due to transplanting, and overlapped entirely with wildflower strip bloom, which was not our intended experimental design; data from year 1 are not reported in this study. In year 2, alfalfa bloomed earlier than in year 1, but then declined quickly when the site experienced the second driest August on record. That year the wildflower strips bloomed after peak alfalfa bloom as we planned, but most bee activity had ceased once bloom in wildflower strips began to peak, affecting what relevant data we could collect. Despite these difficulties, we obtained useful information on how offspring fitness was correlated with dates of brood cell provisioning during year 2.

Wildflower strips. We selected nine wildflower species for the flower strips: common marigold (Calendula officinalis L.), deerhorn clarkia (Clarkia pulchella Pursh), Rocky Mountain beeplant (Cleome serrulata Pursh), plains coreopsis (Coreopsis tinctoria Nutt.), garden cosmos (Cosmos bipinnatus Cav.), common sunflower (Helianthus annuus L.), showy goldeneye (Heliomeris multiflora Nutt.), desertbells (Phacelia campanularia A. Gray), and lacy phacelia (Phacelia tanacetifolia Benth.). The species list includes primarily native annuals, one native perennial that blooms from seed in its first year, and two non-native annuals. We chose these plant species because: (1) annuals establish quickly and bloom in the same year they are seeded, allowing more control over the timing of flowering by altering seeding dates; (2) native plants are a good ecological fit with the natural landscape and could also provide added value by supporting native bee populations; (3) a diversity of flowering plants with different flower colors and morphologies provides varied pollen and nectar sources, creating a mixed diet for bees; and (4) these species are reasonably easy to grow, have commercially available and inexpensive seed, and can grow in full sun with reasonable drought tolerance.

To help guide our plant selection, we also took into consideration (1) published records of M. rotundata visitation to plants (Stubbs, Drummond & Osgood, 1994; Jensen, O’Neill & Lavin, 2003; O’Neill et al., 2004; Cane, 2008; Teper, 2008; O’Neill & O’Neill, 2011); (2) species used in restoration seed mixes for rehabilitation of western rangelands (Herron et al., 2013); (3) species being used in insectary plant mixtures (Grasswitz, 2013); (4) species being tested for creating “bee pastures” for increasing bee populations (Cane, 2010); and (5) species on which bees will readily forage but that do not produce seeds that are similar in size and shape to alfalfa seeds which would not be suitable for growing near alfalfa fields because escaped plants could become weedy in alfalfa seed fields and/or decrease alfalfa seed yield purity, like yellow sweet clover, M. officinalis.

Half of the six sites received the flower strip treatment, alternating based on site location so that adjacent sites received different treatments. In mid-June of 2017, we seeded wildflower strips (4.5 × 25 m) alongside the alfalfa plots. Each wildflower strip contained three, 0.61 × 4.5 m-replicates of each of the nine species for a total of 27 subplots per strip with a 0.30-m walking path between subplots.

Bee nest shelters. We constructed six wood nesting shelters (designed after O’Neill, 2004; Fig. S1) and placed one shelter alongside the edge of each alfalfa plot. Wood-laminate nesting blocks were placed in each shelter for bees to nest (for more details see Fig. S1). Shelters were oriented so that (1) the opening of the shelter faced the alfalfa plot and (2) the flower strips were perpendicular and equidistant to the shelters (12 m from one end of the flower strip) in all plots. This meant that the cardinal direction each shelter opening faced varied based on the layout of the field at each site such that shelters faced either north, northeast, northwest, or east (field layout was determined by the available space at the research farm to accommodate our plots). Shelter orientation was examined as a potential covariate for analyses.

Alfalfa leafcutting bee management. In early February, we purchased alfalfa leafcutting bee cells (i.e., brood cells) from a local alfalfa seed grower (Seed Source, Inc., Toston, MT, USA) containing overwintering ALCB prepupae and placed them into cold storage (6 °C) until needed. Three weeks before alfalfa was expected to be 25–50% full bloom, we removed the bee cells from cold storage and divided them by weight into six trays. We kept them at room temperature for 24 h to acclimate before placing them in a growth chamber in early June at 28 °C to initiate their development into adults for field release. On 3 July 2017, when adult females began to emerge, we moved them to the field shelters in the alfalfa plots. We followed standard recommendations for the quantities of bees released for alfalfa seed pollination in the U.S. (i.e., 40,000–60,000 bees/acre; Pitts-Singer & Bosch, 2010); we set out ca. 6,000 bee cells per plot. Prior to releasing bees, we reared a subset of ca. 550 bee cells which resulted in adult emergence success of 53%, with 43% females for an estimated 1,367 adult females released per plot.

Alfalfa and wildflower floral density. We measured alfalfa floral density for each plot each week after bee release at 0730–1100 h. We divided each plot into thirds, and along one linear transect within each third (each chosen using a random number generator), we estimated floral density within a circular subplot (491 cm²) at three points chosen using a random number generator for both the distance into the row and the position within each row (left or right side of row). Following methods similar to Pitts-Singer (2013), we counted (1) the total number of racemes within the subplot and (2) open flowers classified as tripped (pollinated) and untripped (still available as a food resources) on three racemes that were closest to three evenly spaced markings on the hoop. The “tripped” category included flowers that were “newly tripped” (tripped that day) and “old tripped” (likely tripped the previous day or 2 days). We calculated the average number of open flowers per subplot as: number of open flowers counted per raceme × number of racemes counted in the subplot. We then used this information to estimate floral density in the entire plot. In general, alfalfa flowers were evenly distributed throughout the plots due to the way in which we spaced the plants at the start of the study. We used these same calculations for tripped flowers. For our purposes we added the open and tripped (newly tripped and old tripped) flowers together for floral density estimates because this represented the food resources available to bees within that week. Alfalfa floral density was used to understand the timing of bloom relative to wildflower strips and examined as a potential covariate for analyses.

We also measured floral density of each species within each wildflower strip weekly from the onset of bloom to understand the density and timing of wildflower bloom relative to alfalfa bloom in each plot. Each of the 27 wildflower strip subplots was divided in thirds or in half, depending on the floral density, and all open flowers were counted then multiplied by 2 or 3, depending on how the subplot was divided, to estimate total flowers for each subplot. One species, C. serrulata, did not germinate, and another, C. pulchella, produced very few blooms.

Flower visitation and pollen load composition. To examine flower visitation, we conducted weekly, timed observations of ALCB foraging in wildflower strips and surrounding weedy vegetation. All observations were conducted on calm, sunny days during 1130–1600 h. For wildflower strips, we conducted 3-min timed observations at each subplot (9 min total per plant species per plot) and recorded the abundance and sex of ALCBs visiting flowers. For surrounding vegetation, we conducted 3-min timed observations on each plant species within a 50-m radius of the alfalfa plot and recorded the abundance and sex of ALCBs visiting the flowers. The surrounding vegetation included 10 species total from the genera Liatris, Tanacetum, Solidago, Helianthus, Berteroa, Lupinus, Clematis, Achillea, Heterotheca, and Chamaenerion. All species were distributed in small patches that ranged from an individual plant to a handful of plants in areas no larger than ca. 2 m². For all observations, ALCBs were visually noted and recorded. Because bees were not individually marked, we could not avoid recording the same individual, so it is possible that the same individual could have been recorded more than once. But the duration of the observations was short, and we never saw more than three bees per observation period.

To examine bee pollen load composition, at each plot each week, we captured ca. 10 female bees returning to their nests with full pollen loads, placed them in vials, chilled them on ice, and then removed the pollen loads using dental toothpicks (The Doctor’s BrushPicks®). Following the methods of O’Neill et al. (2004), pollen from bees was stained using Safranin O solution (Home Science Tools, Billings, MT, USA) and fixed on microscope slides for examination using light microscopy. We identified the pollen collected from bees by comparing them to our pollen reference collection comprised of samples of pollen collected from plant species that bloomed that season; all pollen identifications were conducted by the same individual using the reference collection we created consisting of 24 co-flowering species from five plant families (Asteraceae, Boraginaceae, Fabaceae, Onagraceae, and Ranunculaceae). Pollen grains from taxa not represented by our reference collection were classified as morphospecies and reference photos were taken as they were encountered to serve as a digital reference collection. We calculated the percentage of pollen grains of each plant species in each pollen load based on at least 100 grains per slide; percentages were not corrected for size of the pollen grains. Because of concerns with the identification of Phacelia pollen from the flower strips (see Results), we provide descriptive data on percentages of alfalfa vs. non-alfalfa pollens in pollen loads only; no statistical analyses were performed.

Offspring production and survival. To determine approximate timing of when nests were occupied and cells were provisioned with pollen, we conducted weekly observations at all plots and marked the end cap of newly completed nests in each shelter with a different color of paint pen for each week.

In late September, we removed all nesting blocks from shelters. For each plot, we quantified the total numbers of completed nests per plot in each week based on their color paint markings. We then carefully separated individual bee cells from all completed nests and (1) weighed, as a group for each week, all the cells for each plot and (2) determined the average weight of 100 cells for each plot each week. We used these measures to estimate the total number of bee cells produced for each plot each week. Bee cells were then placed into cold storage (6 °C) for overwintering until they were removed the following spring to initiate their development into adults for offspring survival and fitness measures.

In spring 2018, we removed from cold storage a subset of bee cells produced during the previous field season and reared offspring to adults in a growth chamber at 28 °C. We reared bees in individual gelatin capsules (#000) with a hole punched in each end for air exchange; for each plot we chose 80 to 250 cells, depending on availability, provisioned during each of 8 weeks. Each day we checked for bee emergences and immediately freeze-killed emerged adults for examining offspring condition. After bee emergences ceased, we counted the total number of emerged adults and calculated emergence success of offspring (measured as the proportion of adults emerging from all cells containing offspring). We dissected all cells from which no adults emerged to determine whether the cells contained offspring and, if so, the stage reached by the offspring prior to death (e.g., egg, larva, prepupae, pupa, or unemerged adult). For each plot in each week, we calculated the proportion of offspring that died at each stage, hereafter offspring mortality.

Offspring condition. To evaluate offspring quality, we measured body lipid content and body size of adult female offspring that emerged following methods previously used for this species (O’Neill et al., 2011; O’Neill, Delphia & O’Neill, 2014; O’Neill, Delphia & Pitts-Singer, 2015). We randomly chose females for offspring measures from the peak female emergence date for each subset of bee cells reared (i.e., 8 weeks × six plots = 48 subsets) to standardize individuals among plots and weeks.

Frozen female bees were placed individually into vials, dried at 55 °C until weights stabilized, then reweighed to the nearest 0.1 mg on a Sartorius TE64 balance (Sartorius, Goettingen, Germany). To extract lipids, we added 10 ml of petroleum ether to each bee for 10 days, after which the ether was decanted and bees re-dried, weighed again, and proportion body lipid and total lipid mass calculated. To assess bee body size, we measured the head width of each bee (from the outer edges of the eyes) to the nearest 0.5 mm using a microscope fitted with an ocular micrometer (O’Neill et al., 2011; O’Neill, Delphia & O’Neill, 2014; McCabe et al., 2021). We chose this metric to be consistent with our previous work on ALCB body size and because it is a more easily obtained and less ambiguous measure than intertegular distance, another common metric for assessing bee body size (Cane, 1987). We (O’Neill, Delphia & O’Neill, 2014) and others (McCabe et al., 2021) have also shown that head width and intertegular distance are highly correlated with one another in ALCBs and are therefore comparable metrics for intraspecific comparisons.

Estimating seed yields. In late fall, we randomly selected nine spatially stratified alfalfa plants in each plot, harvested all plants, and determined plant biomass and seed yields (measured as dry weight to the nearest 0.01 g).

Results

Field study: Do late-season supplemental wildflower resources enhance ALCB reproduction and offspring condition?

Alfalfa and wildflower floral density. Alfalfa peak bloom preceded wildflower bloom as intended in our experimental design (Figs. 1A and 1B). Alfalfa floral density was highest in week 2 (week of July 10) and then declined in the following weeks to ca. 15% of the highest mean floral density during weeks 5–7 (week of July 31–week of August 14), and finally to ca. 3% in week 9 (week of August 28) (F_7,35 = 49.35, P < 0.0001). The wildflowers started blooming in week 5 (week of July 31) and then increased in the following weeks (F_7,14 = 85.19, P < 0.0001).

Flower visitation and pollen load composition. Female ALCBs at our sites were not often observed in the wildflower strips. During 8.95 h of observations, we recorded 16 female bees visiting wildflower strips on flowers of P. campanularia (N = 7), P. tanacetifolia (N = 8), and C. officinalis (N = 1). During 3.5 h of observations in vegetation surrounding our plots, we recorded eight ALCBs (seven females, one male) visiting hoary alyssum (Berteroa incana L.), six on flowers and two collecting leaf pieces.

We examined the pollen loads of 367 female bees returning to shelters in 2017. Pollen identifications revealed some anomalies, mainly that we observed what appeared to be Phacelia pollen in samples 3 weeks prior to Phacelia blooming in wildflower strips and we observed this for plots from both treatments. This indicated to us that (1) there may be other wild or cultivated plant species in this same genus within the broader landscape (e.g., surrounding residential home gardens) that we were not aware of or (2) there may be another plant species within the landscape whose pollen is easily confused with Phacelia that was not in our reference collection and that we cannot account for. Therefore, we are not able to say from our pollen analyses whether bees collected Phacelia pollen from our wildflower strips. Additionally, only a single bee collected another pollen species which we identified as Calendula, possibly C. officinalis, from the wildflower strip. Consequently, we were limited to comparisons of alfalfa vs. non-alfalfa pollen.

Pollen loads from all plots and weeks combined (N = 367) consisted of a mean (± SE) of 81.1 ± 1.7% alfalfa pollen grains and 18.9 ± 1.7% non-alfalfa pollen grains. We identified non-alfalfa pollen grains in pollen loads at the onset of our bee-pollen collections (week 2), and throughout the nesting season. Two weeks after bee release pollen loads from all plots combined consisted of a mean of 29.7 ± 5.2% non-alfalfa pollen grains with a range for all plots from 19.9–48.3%. Alternative pollen resources used appeared to include those in the surrounding landscape, primarily a Berteroa type pollen, an Achillea type pollen, and three pollen morphospecies that were not in our physical reference collection.

Offspring production and survival. There was no effect of wildflower treatment on the total number of bee cells produced (Table 1). However, the number of bee cells completed varied by week peaking three weeks after release (Fig. 1C, Table 1). Most bee cell production (90%) occurred during the first five weeks after bee release. The remaining 10% occurred in weeks 6 (ca. 6%), 7 (ca. 2%), 8 (ca. 1%), and 9 (ca. 1%) after bee release when the wildflower strips were blooming. Bee cell production for any single plot ranged from 357–1,490 total cells provisioned during week 6, 220–536 cells during week 7, 68–148 cells during week 8, and 41–389 cells during week 9. Shelter orientation was negatively associated with bee cell production; shelters facing northwest were most associated with lower bee cell production, whereas those facing north (followed by northeast and east) were most associated with higher bee cell production.

Among just those cells in which an egg was laid, there was no effect of wildflower treatment on proportion of cells producing adult offspring after overwintering (Table 1). However, the proportion of cells producing adults varied by week of cell completion, declining in late summer (Fig. 1D, Table 1).

Among cells in which an egg was laid, but which did not produce live adults, there was no effect of wildflower treatment on offspring mortality (Table 1). Most premature deaths occurred at the larval stage with a mean (±SE) proportion of 0.34 ± 0.04 for treatments and weeks combined (N = 48). However, the week that cells were produced influenced the stage at which offspring died (Table 1). Neither shelter orientation nor alfalfa floral density was associated with offspring mortality.

Offspring condition. Wildflower treatment did not affect head width or proportion of body lipids of adult female offspring that emerged in 2018 (Table 1). However, female offspring from cells completed earlier in the nesting season had greater mean head widths and higher proportion body lipids (Fig. 2, Table 1). Neither shelter orientation nor alfalfa floral density influenced head widths. Alfalfa floral resources were negatively associated with body lipids. Shelter orientation did not influence body lipids.

Estimating seed yields. There was no main effect of treatment on alfalfa seed yields (i.e., total seed mass) (Table S6). Greater plant biomass was associated with higher total seed mass.

Cage study: Do alternative wildflower resources enhance ALCB reproduction and offspring condition?

Offspring production and survival among cages. Relative to the other treatments, females in the A+WF treatment provisioned more bee cells with an egg per cage (F_2,21 = 6.53, P = 0.0062; Fig. 3A). Of those cells containing an egg, the proportion that produced adults was 17.9–26.8% higher in A-only compared to WF-only and A+WF (F_2,21 = 9.35, P = 0.0012; Fig. 3B). More female (F_2,21 = 6.09, P = 0.0082) and male (F_2,21 = 4.45, P = 0.0245) offspring emerged per cage from A+WF compared to WF-only, and neither was different from the A-only (Fig. 3C). There was no difference among treatments in the mean male: female sex ratio per cage (F_2,21 = 1.06, P = 0.3638).

Offspring mortality was similar across treatments (F_2,21 = 0.22, P = 0.8067). Most premature deaths occurred at the larval stage (mean proportion all treatments combined = 0.33 ± 0.03, N = 24); followed by prepupal, pupal, unemerged adult, and egg stages.

Offspring condition among cages. Female offspring from A-only (N = 115) emerged with significantly greater head widths (F_2,19.27 = 3.71, P = 0.0433; Fig. 4A) and proportion body lipids (F_2,17.96 = 7.84, P = 0.0036; Fig. 4B) than offspring from WF-only (N = 67), but neither were different than offspring from A+WF (N = 145).

Pollen provisioning and provision quality among cages. Examination of pollen provisions in A-only cages confirmed that bees were foraging solely on alfalfa (mean = 98.8 ± 0.1%). In A+WF cages, provisions were comprised of alfalfa (mean = 73.9 ± 1.9%; range: 1–100 pollen grains), Phacelia spp. (24.4 ± 1.9%; range: 0–99), and H. multiflora pollen (0.5 ± 0.1%; range: 0–6). In WF-only cages, pollen provisions were comprised of Phacelia spp. (mean = 95.8 ± 0.7%; range: 70–100) and H. multiflora (2.9 ± 0.7%; range: 1–28). For all treatments, some pollen grains (≤7 in any one sample) were too distorted to determine their identity and were categorized as unidentifiable, which is why values do not sum to 100%.

Estimated pollen provision protein content varied considerably among treatments, being highest in WF-only cells (mean = 55.5 ± 0.2%, N = 62) followed by A+WF (28.8 ± 0.7%, N = 143) and A-only (19.6%, based on Stace (1996), N = 112).

Provision quality and offspring condition in A+WF cages. Closer examination of the A+WF treatment revealed that pollen protein content was not significantly related to head widths of emerging adult females (F_1,141 = 2.04, P = 0.1555).

Discussion

Research on the use of flower strips in agricultural landscapes for enhancing pollinators has focused primarily on two general questions: to what degree do flower strips (1) enhance pollination services in crops and (2) aid conservation purposes by supporting high wild bee diversity? Regarding bee conservation, there is ample evidence that wildflower plantings have positive effects on the abundance and richness of eusocial and solitary wild bees (Blaauw & Isaacs, 2014; Williams et al., 2015; Campbell et al., 2017; Rundlöf, Lundin & Bommarco, 2018; Burkle, Delphia & O’Neill, 2020; Capera-Aragones et al., 2024). However, few studies have directly assessed whether flower strips can increase bee reproduction (Klatt, Nilsson & Smith, 2020; Ganser, Albrecht & Knop, 2021) or the quality of offspring, both of which are important species-specific factors for increasing bee populations.

Our research focused on the reproductive success of a single species of solitary bee, the alfalfa leafcutting bee (ALCB) M. rotundata, that is often intensively managed for its pollination services, especially for alfalfa grown for seed production (Pitts-Singer & Cane, 2011). We combined open-field and field-cage studies to examine whether adding late-blooming floral resources in the form of wildflower strips would enhance the reproductive success of nesting females and the condition of their offspring. The results we present here add to our understanding of the relationships between floral resources and bee reproductive success, and are important for guiding other studies exploring floral resource provisioning strategies for supporting ALCBs and other solitary bees in agroecosystems.

Conclusions

Our study was motivated by the general interest in using wildflower plantings to support bees in agricultural systems and by the specific problem that some alfalfa seed growers face sustaining ALCB populations without purchasing replacements each growing season. With our experimental design in the open-field study and with our choice of wildflower species, however, the supplemental late-season floral resources did not enhance ALCB reproduction. In addition, female offspring produced later in the season emerged the following year with smaller body sizes and with lower lipid stores compared to those produced earlier in the season when alfalfa was in full bloom. Those two effects may place constraints on the reproductive success of those offspring and, perhaps, their value as pollinators. If this seasonality effect was due, in part, to maternal aging rather than just the decline in alfalfa resources, it may be that late-blooming wildflower strips will have limited value in sustaining healthy ALCB populations. Parallel results in our cage study revealed that the availability of both wildflowers and alfalfa did not result in any measurable reproductive benefits beyond alfalfa only. The availability of wildflowers in addition to alfalfa did not produce more offspring, larger offspring, or increase their lipid stores. Furthermore, when only wildflowers were available in the cage study, it resulted in smaller female offspring with lower lipid stores compared to those on alfalfa. If this effect was the result of pollen quality and not simply the amount of pollen available in WF cages, it could exacerbate the reproductive success of the already reduced condition that we observed for adult female offspring produced late in the season the previous year.

Our results highlight the importance of measuring multiple metrics of bee reproductive success, since using only one metric could lead to erroneous conclusions. For example, one might conclude that WF-only diets are not different from A-only diets if one examined only offspring numbers. But, if offspring size and body lipid content are important for future reproductive success and pollinator efficacy, the WF-only diet we chose may provide limited benefits to the long-term health and fitness of the population. To better guide future floral resource provisioning strategies, more research is needed to understand how alternative floral resources and nutrition, particularly pollen quality and quantity, and the temporal availability of those resources affects the reproductive success of ALCBs (and wild bees more generally), as well as the success of subsequent generations.

Supplemental Information