Evaluating Native Perennial Flower Strips for Enhancing Native Bees and Pollination Services on Farmlands

Evaluating Native Perennial Flower Strips for Enhancing Native Bees and Pollination Services on Farmlands

Summary

During the first year of this project (spring 2013), we established wildflower strips comprised of nine native perennial species on four participating farms, expecting them to fully bloom in 2014. Throughout that summer, we documented bee communities (using three methods) and flower densities so that we would have information on the abundance and diversity of bees and flower resources at each farm prior to introduction of the additional wildflowers. In 2014, we continued sampling of bees and background floral resources and began assessing bee communities and floral abundance in the wildflower strips. We repeated the sampling protocol during the summer of 2015, the final field season of the study. To complete the study, we will compare the abundance and diversity of bees collected in 2013, prior to establishment and blooming of the flower strips, to the abundance and diversity of native bees collected in 2014 and 2015, after flower strip establishment. We will also examine bee floral preferences and identify native plant species that are important food resources on farmlands. We have processed all the bee samples collected with nets in 2015 from the flower strips and surrounding floral resources and will complete identifications shortly. We are currently processing 2015 bee samples from bowl-traps. The 2013-2014 samples contained 31 bee genera and 115 taxa. The samples include at least one, and possibly two, new bee species (identified by experts at the USDA Bee Biology and Systematics Laboratory).

To determine the effects of flower strips on crop pollination and whether crops are pollen-limited on farms, we 1) observed bees visiting squash and sunflower plants at different distances from the wildflower strips, 2) performed hand pollinations, and 3) measured crop yields in in each year of the study. Preliminary analyses of 2013 squash data suggest squash flowers were pollen limited; the 2014 and 2015 squash data are ready to be analyzed, as are the 2013 and 2014 sunflower data. The 2015 sunflower seeds are currently being processed. We have identified all of the bees visiting squash and sunflower plants in 2013 and 2014. The 2015 bees have been processed and will be identified shortly.

As in 2014, flower strips produced enough blooms in 2015 that we were able to collect wildflower seeds to determine whether flower strips could serve an additional function to farmers as sources for wildflower seed that could be sold through local outlets. In order to sell seed in the state of Montana, germination and viability testing is mandatory. In 2015, we conducted germination and viability testing on the wildflower seeds collected in 2014. Preliminary analyses indicate differences among plant species and farms in total seed viability, which could affect the amount of marketable seed, depending on the farm. These data provide important information for our economic analyses, which will be performed shortly, regarding the costs and benefits associated with establishing flower strips on farms for seed production.

To communicate our projects’ progress and findings in 2015, we 1) participated in two field days, 2) presented four talks to audiences, including the general public, producers, land managers, and agricultural and horticultural professionals, 3) presented a guest lecture on native bees for students in the “Insects and Society” course at Montana State University (MSU), 4) set up an education booth at the Gallatin Valley Farmer’s Market where we gave away native plant seeds harvested from our wildflower strips, 5) hosted an MSU First-Year Research Experience student (FYRE), and 6) hosted an undergraduate McNair Scholar who is in the process of completing an independent research project. We were also interviewed for a story by the MSU Magazine, Mountains & Minds, as well as University Communications-MSU News, and ABCFoxMT. We are working on creating a website, as well as pictures and videos of research and outreach activities, to share further details of our project.

Objectives/Performance Targets

The four main objectives of the project are to:

1) determine the effects of native perennial flower strips on the abundance, diversity, and foraging behavior of native bees in agricultural crop fields,
2) determine the value of flower strips in improving crop pollination and yields through increases in the abundance, diversity, or behavior of native bees,
3) evaluate the potential of flower strips for native seed production and sales, and
4) execute a research-based outreach program to communicate our findings to producers, land managers, agricultural professionals (e.g., NRCS personnel and Extension agents), scientists, and the general public.

Accomplishments/Milestones

Objective 1. Determine the effects of native perennial flower strips on the abundance, diversity, and foraging behavior of native bees in agricultural crop fields.

The USDA Natural Resources Conservation Service (NRCS) offers cost-share programs, such as the Environmental Quality Incentives Program (EQIP), to encourage producers to provide habitat that enhances pollinator communities on agricultural lands using native plants. However, the effects of these plantings on native bees and pollination in agroecosystems have not been rigorously assessed in most cases. Using a variety of resources as a guide, including the USDA-NRCS publication “Montana Native Plants for Pollinator-Friendly Plantings” and results from research being conducted at Montana State University on wildflower drought tolerance and pollinator attraction, we selected a set of nine native perennial wildflowers to incorporate into flower strips being planted at four participating farms in the Gallatin Valley (Table 1). We chose plants with different bloom times, colors, and floral morphologies that would span the entire growing season (see 2015 report for bloom times). In 2013, we focused on establishing flower strips and keeping them watered and weeded throughout the summer (see 2015 report).

In the spring of 2014 and 2015, we recorded winter mortality/survival of individual plants. In 2014 we replanted anything that did not survive using additional plants that we grew in a greenhouse. In general, plants had low winter mortality in both years. In 2014, compared to 2013, we saw a marked increase in vegetative growth and flowering of all nine species at all four farms. In 2015, plants appeared similar in size and abundance of blooms compared to 2014. The majority of plants seemed to have reached their maximum size in 2014. We will confirm this observation when 2015 bloom data has been entered.

To determine whether the addition of flower strips leads to changes in the community composition, relative species abundance, or plant visitation rates of native bees, we sampled the bee community in 2013, 2014, and 2015. In all three years, we also measured floral density of naturally-occurring plant species found at each farm at two different times during the summer to better understand background floral densities before and after flower strip establishment. From the onset of flowering in 2014 and 2015, we conducted weekly surveys of floral abundance of the nine species in the flower strips to assess the amount of supplemental food resources available to foraging bees. We sampled bees weekly from May-September in 2013, 2014, and 2015 at all farms using three methods, including yellow bowl-traps, insect nets, and trap-nests.

Bees in yellow bowl-traps. In 2015, as in the previous two years, we used yellow bowl-traps filled with water and a small amount of soap to passively sample the bee community at each farm. We placed bowl-traps along four linear transects (i.e. along the flower strip and at three distances (20, 60, and 180 m) from the flower strip). All of the 2013 and 2014 samples have been processed and identified at least to genus, and to subgenus or species when adequate keys are available (see Bee species identifications). We use the term “taxa” to refer to these different levels of identification collectively. We are in the process of pinning and labeling all of the 2015 bee samples, at which point they will be identified.

Among the four farms and over two years, we have collected 4,305 bees using bowl-traps (1,906 in 2013 and 2,399 in 2014). Preliminary analyses of 2013 data indicate differences in total bee abundance among farms (Figure 1), sampling weeks, distances (position relative to flower strips), and distances by farm. With respect to distances, at three of the fours farms in 2013, total bee abundances were highest at the sampling transects furthest from the flower strips (at 180 m), although flower strips were not blooming much during this first year of the study. The lowest total bee abundances were found at the 20 m transects while bowls alongside the flower strips and at the 60 m transects were similar in bee abundance.

Preliminary analyses of 2014 data indicate no differences in total bee abundance among farms (Figure 1) and differences only among sampling weeks and distances. As in 2013, we found the highest bee abundance at the 180 m transects on all farms. We found the lowest bee abundance in bowls alongside the flower strips, which were established and blooming extensively in 2014. These preliminary findings may seem counter to expected, but there are several factors that could explain our results. For example, it is possible that bowl-traps are less attractive when near flowers. In the absence of flowers, farther from the flower strips, bowls may act as magnets that are highly attractive to bees seeking floral resources. We need to re-examine both the 2013 and 2014 data with total surrounding flower abundance added as a covariate even though the surrounding natural vegetation did not appear to change over the three years. The flower strip floral abundance did, however, increase across sites from 2013 to 2014. This increase in floral abundance could have contributed to the loss of a site effect in total bee abundance from 2013 to 2014 if it affected the attraction of bees to our farms. It is also possible that differences in bee abundance over the two sampling years are simply due to year-to-year variation, which is common in natural systems. Once our 2015 bees are identified in the coming months, we will have two years (2014 and 2015) of bowl-trap data post-flower strip establishment for analyses and comparison to our bowl-trap data (2013) pre-flower strip establishment to further our understanding of these patterns.  

The total abundance of bees collected in bowls in 2013 and 2014 comprise five families: Andrenidae, Apidae, Colletidae, Halictidae, and Megachilidae. In both years, bees in the family Halictidae (sweat bees) were the most abundant on all farms (Figure 2). Preliminary analyses of the total abundance of bees in the family Halictidae for 2013 and 2014 revealed the same differences in treatment variables as those seen for total bee abundance in each year. Therefore, halicitids are likely the main driver of the differences we observed for total bee abundance. Analyses of the remaining families revealed differences among farms in 2013 and 2014, but no distance effect for the families Andrenidae or Apidae. Perhaps andrenids and apids, which tend to be larger bodied, are able to fly further and cover greater distances, and therefore are more evenly distributed among bowl-traps at different distances on farms. These patterns might also reflect differential attraction of different families and taxa to bowl-traps or real variation in relative abundance.

We have collected a total of 29 genera (excluding Apis) in bowl-traps (27 in 2013 and 24 in 2014; Figure 3). Only five genera were not collected in both years, including three genera in the family Megachilidae (Anthidium, Dianthidium, and Coelioxys), one in the family Colletidae (Colletes), and one in the family Apidae (Epeolus). The number of genera varied by farm over two years from 22 to 28 (Gallatin Grown = 20, Gallatin Valley Botanical = 22, Rocky Creek = 28, and Towne’s Harvest Garden = 25). The most abundant genera (>100 individuals) across all farms were the same, though their relative abundances varied between years. Sweat bees of the genus Lasioglossum spp. were the most abundant bees in 2013 (848) and 2014 (1,144; Figure 3). In 2013, the three next most abundant genera include bees of the genera Halictus (358), Hylaeus (149), and Panurginus (146). In 2014, the three next most abundant genera were Panurginus (481), Halictus (328), and Hylaeus (116). Among these 29 genera we have identified 87 taxa with many more yet to be identified (see Bee species identifications).

Bees collected in flower strips using insect nets. We captured bees visiting flowers on the surrounding vegetation (2013, 2014, and 2015) and the flower strips (2014 and 2015) during timed observations. All of the 2013 and 2014 samples have been processed and identified (see Bee species identifications). The 2015 samples have also been processed and we will begin identifications shortly.

Because seven of the nine species in the flower strips started blooming in late summer in 2013, we began collecting bees visiting the strips at that time. We collected 223 non-Apis bees visiting flower strips on farms on six sampling dates from late-July through mid-September. Preliminary analyses indicate no differences in total bee abundance among farms, but there were differences between sampling weeks and plant species. Across all farms, the highest proportion of bees was collected visiting H. maximiliani (37%), while the lowest proportion was collected visiting G. viscossisimum (7%; Figure 4). (Note-E. speciosus (1%) was blooming only at one farm, whereas the other six species did so at all farms). The total bees visiting the flower strips in 2013 comprised the same five families collected in bowl-traps. Bees in the family Apidae were the most abundant across all farms (63% of all bees collected) followed by Halictidae (21%; Figure 5). We collected a total of 18 bee genera visiting flower strips in the first year. Netting bees added one genus not collected in two years of bowl-trapping, Xeromelecta (Apidae), a brood parasite that does not collect pollen to provision her offspring. Among farms, the most abundant bee genera captured on flower strips included bumble bees, Bombus (100) and Lasioglossum (23). Among these 18 genera collected we have identified 26 taxa. Netting added four species of bees not captured in bowls. This demonstrates the importance of using multiple collecting methods for assessing the bee community. Our 2013 flower strip collections provide an idea of what bees were present on farms, but they are not necessarily indicative of the types of bee-flower associations that would occur on plants that are established.

In 2013, we collected 161 non-Apis bees visiting the surrounding floral vegetation across all farms on three sampling dates (late July-late August). Preliminary analyses indicate differences among farms in total bee abundance. Across all farms, the highest proportion of bees was collected visiting Tanacetum vulgare (39%). This value was likely affected by site differences because certain weeds were more prevalent at some farms compared to others and not all weeds were present at all farms. The surrounding vegetation consisted of >20 agricultural weeds (e.g., Lotus corniculatus, Melilotus officinalis, Trifolium repens, Cirsium arvense, Tanacetum vulgare, etc.), and although we have not yet analyzed the data for surrounding flower abundances, there were clear visual differences in the amounts of surrounding vegetation at each farm.

The total bees visiting the surrounding vegetation in 2013 comprised the same five families collected in bowl-traps and on flower strips (Figure 6). Similarly to flower strips, bees in the family Apidae were the most abundant across farms (44% of all bees collected) followed by Halictidae (42%), though this percentage is double that on flower strips (Figure 6). These differences are likely due to differences in both the number of plant species from which we collected bees (21 versus 7) and traits associated with the individual plant species themselves. We collected 13 genera of bees visiting surrounding flowers and 24 taxa, including four species not captured in bowls or on the flower strips. Among farms, the three most abundant genera captured on surrounding flowers included Bombus (67), Halictus (34), and Lasioglossum (31). This is similar to what we found with flower strips. Our collections on the surrounding vegetation along with bowl-traps in 2013 give us a measure of the bee community composition before flower strip establishment. We next need to include measures of the surrounding floral vegetation (i.e., total floral abundance, plant species diversity) in our analyses of the effects of the flowers strips to better understand differences in bee communities across farms.

In 2014, we collected 1,763 non-Apis bees visiting flower strips across four farms on 13 sampling dates from late May-early September. Preliminary analyses indicate differences in total bee abundance among farms, sampling weeks, plant species, and plant species by farm (Figures 7). Across all farms, the highest proportion of bees was collected visiting H. maximiliani (17%), while the lowest proportion was collected visiting C. rotundifolia (5%; Figure 8). There were differences among farms regarding the plant species on which we collected the highest proportion of bees. For example, we collected the highest proportion of bees visiting M. fistulosa at Gallatin Valley Botanical, P. hastata at Towne’s Harvest, and H. maximiliani at Gallatin Grown and Rocky Creek (Figure 8). These differences in bee visitation likely reflect differences in 1) bee species assemblages at each site, 2) floral abundances of species within wildflower strips and surrounding vegetation, and 3) timing of flowering with timing of bee species activity. These are factors that we need to examine more closely in future analyses. If different plant species attract different bee species, then some plants may be better than others for different conservation situations.

The total bees visiting the flower strips in 2014 comprised the same five families previously collected and the number of bees within each family varied by farm (Figure 9). The composition of each family visiting flower strips among farms also varied from the composition of families collected in bowl-traps during the same year and over the same sampling period (approximately June-September; Figure 2). (The 2013 flower strip data did not include bees prior to late-July, since the plants were not yet blooming). For example, Halictidae comprised 66% of all bees in bowls in 2014, but only 38% of bees collected while visiting flower strips. In contrast, Megachilidae comprised only 3% of bees collected in bowls, but 22% of bees collected visiting flower strips. Similarly, Apidae comprised 3% of bees collected in bowls and 26% of bees visiting flower strips. Furthermore, if we compare the composition of families across all farms and at each farm in bowl-traps set out along the flower strip transect only (Figure 10) to bees netted at flower strips (Figure 9), we see similar differences in the composition of families between bowls and netting. These bowls were within two feet of the flower strips and yet we still see differences in the composition of bees. Halictidae comprised 30-41% of bees collected while netting among farms (Figure 9); whereas bowls placed right next to flower strips comprise 45-75% halictids (Figure 10). In the most extreme example, no megachilids were collected in bowls set out along the flower strip at Gallatin Grown, yet 26% of netted bees visiting flower strips were megachilids. However, we still need to correct these data for collecting effort as measured by time. Bowls were set out for approximately six hours each sampling week and netting at flower strips occurred for 30 m per plant species in bloom each sampling week, though including sampling effort in our analyses should only increase the differences we see between collecting methods.

We collected a total of 27 genera visiting flower strips in the first year after flower strip establishment (Figure 11). As in 2013, netting bees in 2014 added one genus not collected in two years of bowl-trapping, Ashmeadiella (Megachilidae). Conversely, there were three genera collected in bowls not collected using nets, Diadasia, Epeolus, and Perdita. The most abundant genera (>100 bees) visiting flower strips across all farms were Lasioglossum (373), Bombus (307), Halictus (245), Heriades (182), and Megachile (108; Figure 11). These results are different from those genera most abundant in bowls traps (Figure 3). Using both collecting methods we now have 31 genera represented among farms. Among the 27 genera collected from flower strips we have identified 73 taxa to date. In addition to adding another genus, netting bees added 23 species not collected with bowl-traps, bringing the total number of taxa up to 115 among farms and both collecting methods. The many differences we have observed in the composition of bees between collecting methods highlights the importance of using multiple methods for surveying a bee community.

Trap-nests. In 2013-2015, we prepared and placed two sets of trap-nests at each farm in late May to provide nest sites for cavity-nesting bees. We monitored trap-nests weekly (i.e., remove completed nests and replaced with empty nest tubes) from May-September. We retrieved trap-nests from each farm in October, removed and labeled all filled nesting tubes, and then placed them in cold storage for overwintering of bee offspring. In April of 2014 and 2015, we removed the nesting tubes from cold storage to initiate rearing of adult bees. We pinned all the bees that emerged from trap-nests collected in 2013 and 2014 and will identify them in spring 2016. The trap-nests collected in 2015 are currently in cold storage for overwintering until spring 2016.    

Bee species identifications. To facilitate bee identification, we traveled to Logan, Utah in March of 2015 to get assistance from bee specialists at the USDA Bee Biology and Systematics Laboratory. We spent a week working with a team of experts learning how to identify species within several genera as well as having as many species identified as possible. We are working to get all of our bees identified to species, which will contribute to our long-term goal of developing a comprehensive bee species list and reference collection for Montana. We also sent specimens within the genus Melissodes to a graduate student at the University of New Mexico, who identified 10 species among our specimens.

Objective 2. Determine the value of flower strips in improving crop pollination and yields through increases in the abundance, diversity, or behavior of native bees.

To measure pollination services and determine whether crops are pollen-limited on farmlands due to a lack of pollinators (either diversity or abundance), we established three experimental crops strips comprised of squash and sunflower plants at each farm at varying distances from the flower strips (20, 60, and 180 m). In the greenhouse in the spring of 2013, 2014, and 2015 we propagated 100 acorn squash and 100 confection-type sunflower plants and, in mid-June, planted eight squash and eight sunflower plants at each of the three distances at each farm (i.e. 24 plants of each species at each farm). We chose these crop plants because they have different bloom times and flower morphologies, and so likely attract different suites of bee pollinators. As the plants began to bloom (late July) we hand-pollinated half of the squash and sunflower plants weekly in each crop strip, while leaving the other half to be open-pollinated in order to determine whether plants are pollen-limited. (Hand pollination provides supplemental pollen to flowers so that one can determine maximum yield possible). We also conducted bi-weekly timed observations in 2013 and weekly observation in 2014 and 2015 at each of the crop strips during flowering (July-September) and captured bees visiting flowers. All of the 2013 and 2014 samples for squash and sunflower have been processed and identified. All of the 2015 samples have also been processed and we will begin identifications shortly.

In 2013, we collected 1,012 honey bees and 104 other bees during squash observations. Preliminary analyses indicate no differences among farms in total abundance of honey bees or native bees visiting squash; however, we need to reexamine these data with total squash flower abundance as a covariate. Bees visiting squash comprised only two families (Apidae and Halictidae). Other than honey bees, the most common bees were those in the genera Lasioglossum (41) and Bombus (29). In 2014, we collected 61 honey bees and 8 other bees during squash observations. They represented the same two families (Apidae and Halictidae) and the most abundant genus was Lasioglossum (6).

In 2013, we collected 8 honey bees and 50 other bees during sunflower observations. They represented three families (Apidae, Halictidae, and Megachilidae). The most abundant genera were Melissodes (26) and Bombus (13). In 2014, we collected 23 honey bees and 61 other bees during sunflower observations, representing four families (Andrenidae, Apidae, Halictidae, and Megachilidae). As in 2013, the most abundant genera visiting sunflowers in 2014 were Bombus (24) and Melissodes (21).

To determine crop yields for both hand- and open-pollinated plants, we harvested all of the squash plants and sunflower seed heads at the end of the summer/early fall in 2013, 2014, and 2015. For squash, whole plants were harvested, and the vegetation was dried and weighed. The fruits were counted and weighed, and then seeds were removed, cleaned, dried, categorizing (mature vs. non-mature), counted, and weighed. We have completed processing all squash plants, fruits, and seeds, and started analyzing the 2013 and 2014 data. We just completed processing all of the 2015 squashes and entering the data and can now begin analyses. Preliminary analyses indicate squash plant biomass was highly variable among farms, distances, and distance by farm in 2013 and 2014. This was likely due to differences in locality, irrigation, and growing conditions among farms, but even within a single farm there were differences among the three locations where the crops were planted. For 2013 squash data, we also found differences for total fruit weight among farms and at different distances within farms, but there were no treatment effects (hand-pollinated versus open-pollinated). From a farmers perspective this is good news because pollination did not affect the size of the fruits (i.e., the saleable product). However, from a plant reproduction point of view, there were treatment effects for several measures of pollination success, including total number of seeds and total seed weight, which were higher for hand-pollinated plants. This indicates that squash flowers were pollen-limited on farms in 2013.

For sunflowers, we measured the height and stem width of plants before harvesting the flower heads as an estimate of biomass. Seeds were removed from flower heads, cleaned, categorizing (mature vs. non-mature), counted, and weighed. We have completed processing all of the sunflower seeds and entered the 2013 and 2014 data. We are currently processing the 2015 sunflower seeds and, once completed, will analyze the data.

Objective 3. Evaluate the potential of flower strips for native seed production and sales.

In addition to providing habitat for supporting native bees, another possible benefit of establishing flower strips are the sale of locally produced, regionally adapted, organic, wildflower seed for CSA shares, farmer’s markets, retail stores, or habitat restoration projects. To determine the feasibility of harvesting and selling wildflower seeds, we will be conducting an economic analysis. We have been keeping records (2013-2015) of costs associated with installing and maintaining flower strips (i.e., cost of plant materials, labor involved with growing plants, preparing the planting site, planting the strips, weeding, and watering), and harvesting seed (i.e., collecting, processing, and cleaning).

Flower strips were established on farms in the spring of 2013, and we saw some flowering that summer, but not in any great amounts for seed collecting. In 2014 and 2015, however, the blooms were abundant and we collected seed in both years. We hand-harvested seed weekly from all flower strips as it matured. The ease of harvesting seed was species specific. For example, some species, like P. confertus, had seed that matured essentially all at once and resulted in fewer collection weeks to gather the majority of the seed. Other species, like G. aristata, H. villosa, P. hastata, and C. rotundifolia, continued to flower as they also set mature seed, requiring many weeks of repeated collecting. In 2014, we kept the seed separated by collection date and farm for data analyses. We cleaned all of the seed by hand (quantities were too small for any type of mechanized cleaning) and time spent cleaning seeds varied by plant species and farm (see 2015 report). We also found differences among farms in the total weight of seed collected (see 2015 report), which is likely due to differences in planting (i.e. IRT plastic, bare ground, black fabric) and irrigation methods (i.e. drip tape, overhead sprinklers), as well as differences in climate/location that contributed to differences in the overall size of plants and hence the number of blooms, which are likely correlated with seed production.

Another aspect that we needed to consider is that there could also be differences among farms in the quality of the seed produced (i.e. size, total viability), which could affect salability of seeds. In late winter-early spring of 2015 we determined 1) mean weight per seed and 2) conducted both germination and viability testing on seeds of each of the nine plant species collected in 2014. We have entered seed weight data and are almost finished entering data on germination and viability. Preliminary analyses indicate differences in mean weight per seed among farms for some, but not all of the plant species. For example, H. maximiliani, P. confertus, C. rotundifolia, and G. aristata all varied in mean weight per seed among farms. Furthermore, the variation for each plant species was not consistent among farms. The farm with the highest mean weight per seed for H. maximiliani was Towne’s Harvest Garden and the lowest was Gallatin Valley Botanical, but this pattern was reversed for P. confertus. Other species, including E. speciosus, G. visocosissimum, and H. villosa, showed no site differences in mean weight per seed.

As with seed weight, preliminary analyses of germination and viability data entered to date indicate differences among farms in the total viability of seeds depending on the species. For example, despite differences in seed weight for H. maximiliani seeds across sites, we found no differences in total viability, which ranged from 91-97% among farms. This is good news from a seed sales perspective. Of 259 H. maximiliani seeds tested, only one seed germinated before we performed a Tetrazolium Chloride (TZ) test. This is a quick germination test that uses a chemical reaction between living tissues in the seed and tetrazolium chloride, which turns the living tissue reddish in color. These seeds are therefore considered “dormant by TZ,” meaning they are viable but cannot germinate without breaking seed dormancy. Depending on the plant species, seed dormancy can be broken through a variety of methods, including scarification, stratification, and age. This is very common among wildflower species, which is one issue that makes their cultivation difficult. Looking at G. aristata, which also exhibited differences among sites in mean seed weight, we also saw significant differences among farms in total viability of seed ranging from 10-45%. Of 320 G. aristata seeds tested, 243 were unfilled (76%), which is a common occurrence with this species, 44 germinated (14%), and 31 were TZ tested and viable (10%). Lastly, there were no differences in mean seed weight among farms for H. villosa and similarly there were no differences in total viability, which ranged from 46-67%. Of 662 H. villosa seeds tested, 214 were unfilled (32%), 262 germinated (40%), and 125 were TZ tested and shown to be viable (19%).    

These preliminary findings indicate that seed weight is not a good predictor of seed viability. It also appears some farms may have better conditions for growing certain plant species and producing quality seed than others. Perhaps variation in the bee community at each farm affected pollination and subsequent seed set of different species within wildflower strips. If we initiate seed sales, it might be best to keep seeds separated by farms. It will be interesting to see if the seed we collected in 2015 exhibits similar patterns in seed weight and viability among plant species and farms. As part of an undergraduate research project, a student has just started hand-cleaning the 2015 seed and will also be responsible for conducting germination and viability testing, data entry and analyses, and writing a report on her findings due in early May.

Objective 4. Execute a research-based outreach program to communicate our findings to producers, land managers, agricultural professionals (e.g., NRCS personnel and Extension agents), scientists, and the general public.

To improve producer and public awareness of using native perennial wildflowers as a management strategy to improve bee habitat on farmlands and other types of managed lands, we have taken part in numerous outreach efforts in 2015. At all outreach events, we distributed our guide showing peak bloom times of the nine species in our flower strips in 2014 (see Table 1 2015 annual report). We participated in two field days. The first, in early June, featured conservation plant solutions at the NRCS Bridger Plant Materials Center located in Bridger, Montana (approximately 50-60 participants including scientists, NRCS personnel, Extension agents, and the general public; conducted post outreach surveys after participants visited our educational station). The second field day was in late July and sponsored by Montana State University (MSU) College of Agriculture. It was located at MSU’s Horticulture Farm and the site of one of our participating farms, Towne’s Harvest Garden (approximately 35 participants including producers, scientists, and the general public). We also set up an educational booth in late summer at the Gallatin Valley Farmer’s Market (alongside the Master Gardeners with MSU Extension) and asked visitors to our booth to fill out a survey. In exchange, they received native seed collected from our flower strips.

We presented a talk for the Montana Organic Association’s (MOA) 13^th Annual Conference titled ”Identification and conservation of native bees” which was one four concurrent talks (approximately 20 participants; conducted pre- and post- outreach surveys; gave away native seed from our flower strips). Among MOA’s goals are promoting organic agriculture for the good of the environment and the meeting was comprised of producers, processors, retailers, consumers, NRCS personnel, Extensions agents, and researchers. We also presented a talk at the Pollinator Protection Symposium sponsored by the Montana Department of Agriculture (40 participants, all from the horticultural industry, including nursery owners, employees, and landscapers). We presented a talk for MSU Extension’s Small Acres Management Course, which is geared towards helping landowners and land managers create informed management plans by understanding the natural resources on these lands (14 participants; conducted pre- and post- outreach surveys). We also presented a talk in early spring for the Whitehall Garden Club (25 participants; gardeners and general public).

We presented a guest lecture on native bees and native plants for a class titled ‘Insects and Society’ at MSU (19 students). We hosted two students in the lab and will be hosting another this coming semester. The first was a First Year Research Experience (FYRE) student in the Undergraduate Scholars Program at MSU who worked primarily in the lab processing insect specimens but also assisted at the BPMC field day. The goal of the FYRE program is to connect freshman with hands-on research experiences. The second student is in the McNair Scholars Program and is working on an independent research project examining variation the quality (i.e., protein content) of pollen among the nine native plants in our flower strips and how this might influence bumble bee foraging. The McNair program is targeted toward students often underrepresented in graduate schools (first-generation/low income students or minorities). A third student (a senior in Organismal Biology) just joined the lab and will be conducting an undergraduate research project examining our 2015 flower strip seed data. The student will be cleaning, weighing, and counting seed, as well as conducting seed germination and viability testing and data analyses to be completed this semester (May 2016).

We were interviewed for a story for the MSU Magazine, Mountains & Minds, as well as University Communications (MSU News). The MSU News story was picked up by our local newspaper (The Bozeman Daily Chronicle) and shortly after we were interviewed by ABCFox MT, which was broadcast on the local evening news.

• The Montana State University Magazine: Mountains & Minds

http://www.montana.edu/mountainsandminds/article.php?article=15531

• University Communications: MSU News

http://www.montana.edu/news/15615/montana-bees-get-washed-fluffed-and-brushed-for-msu-researchers

• ABCFox MT

http://www.abcfoxmontana.com/story/29466288/washing-brushing-and-fluffing-all-for-montana-bee-research

We have taken the prerequisite training for creating a University website using the new Web Content Management System and are working on adding our content. We hope to publish the website in March, as another means of sharing information about our project, its progress, and our findings, as well as pictures and videos of research and outreach activities during the summers of 2013, 2014, and 2015.

• Figure 2. Total bees by family collected in bowl-traps over two years

• Figure 1. Total bees collected in bowl-traps at farms over two years
• Figure 5. Total bees by family visiting flower strips in 2013
• Figure 7. Total bees visiting plant species in 2014
• Figure 10. Total bees collected in bowl-traps alongside flower strips in 2014
• Figure 8. Total non-Apis beees visiting flower species in 2014
• Figure 3. Bee genera collected in bowl-traps over two years
• Figure 6. Total bees by family visiting surrounding vegetation in 2013
• Figure 4. Total non-Apis bees visiting flower species in 2013
• Figure 9. Total bees by family visiting flower strips in 2014
• Table 1. Native wildflower species
• Figure 11. Bee genera collected on flower strips in 2014

Impacts and Contributions/Outcomes

The first year of field sampling for our multi-year project provided primarily baseline data and information on typical bee assemblages prior to flower strip establishment and blooming. We successfully established flower strips at all farms, and, after completing a third year of field sampling, we obtained the necessary data needed for comparisons of bee assemblages before and after the flower strip additions. Once we complete bee sample processing and identifications of the 2015 data in the coming months, we will conduct these analyses.

We do not yet know if flower strips have increased pollination of focal crops, but we do know that squash were pollen-limited in the first year of this study. We have two more years of squash data (2014 and 2015) ready to be analyzed, as well as two years (2013 and 2014) of sunflower data. We conducted germination and viability testing on the wildflower seed collected in 2014 from our flowers strips, providing more information for our economic analysis and potential seed sales.

We have shared information about our project and its objectives with growers, NRCS personnel, Extension agents, scientists, landowners, students, and the public through two field days, an educational booth, presentations, interviews, and volunteering opportunities, and we hope to conduct more outreach this summer by attending farmers markets and distributing surveys and informational handouts, which we will create in the coming months. We have also begun designing a website to share the progress and findings of our project and to serve as a location for pictures, videos, surveys, and future outreach materials.

Collaborators: