Ecosystem Services in Hedgerow Restorations: Pollination Function and Nesting Habitat

Final Report for GW12-024

Project Type: Graduate Student
Funds awarded in 2012: $17,882.00
Projected End Date: 12/31/2014
Grant Recipient: University of California, Berkeley
Region: Western
State: California
Graduate Student:
Principal Investigator:
Dr. Claire Kremen
University of California, Berkeley
Expand All

Project Information


We evaluated the ability of hedgerow restorations to augment hybrid sunflower pollination by the native bee community. We also investigated whether rates of nesting were increased in fields adjacent to hedgerows, as well as whether the presence of hedgerows enhanced the diversity and abundance of the native bee community at different distances into fields.


Pollination is a critical component of the crop production cycle, directly contributing to reproductive success for pollinator-dependent crops. Sufficient pollination enhances seed quantity and quality, which directly correlates to profits. Bees are the most important crop pollinators, providing an estimated $200 billion in pollination services worldwide (1). Drastic declines in honey bee (Apis mellifera) populations due to Colony Collapse disorder and other factors have demonstrated the risk of relying on a single pollinator. Many native bee species are capable of providing pollination services and could act as insurance against fluctuations in honey bee supply (2). However, native bees have also declined in some regions of North America due to agricultural intensification. As a result, there are increasing calls for on-farm management actions that can enhance agroecological resilience by restoring habitats that support pollination services (3). Hedgerows, field edge plantings of native shrubs and forbs, are a common re-diversification technique; however, the efficacy of hedgerows or the scale at which they affect crop yields remains unknown.

Native bees depend on both floral and nesting resources. While floral abundance has been extensively studied, nesting resources are understudied despite their role regulating bee communities. Native bees are predominantly solitary ground-nesters, requiring direct access to open soil. Disturbances, such as irrigation and tillage, in intensively farmed areas can inhibit nest availability. In addition, native bees are central-place foragers, typically foraging within a 300-1000 m radius from their nest site, depending on body size (4). Thus, the proximity of nesting habitat to crops could affect the distribution and evenness of pollination services within agricultural fields. Availability of nesting resources is essential for the success of on-farm habitat enhancement projects targeting native bee conservation. Hedgerow restoration appear to provide more nesting resources than unmanaged field margins (5), however whether this translates into increased rates of nesting bees has not yet been evaluated.

We studied the provision of pollination services associated with hedgerow restoration. Specifically we examined whether native bees pollinating crops also utilized hedgerow resources. We then determined whether abundance and diversity of native bees affected calculating rates of seed set in hybrid sunflower (Helianthus annuus), a mass-flowering pollinator-dependent crop. We then determined where bees nested within the agricultural landscape, and whether hedgerow presence increased nesting rates.

1. Gallai, N et al 2009. Ecol. Econ. 68:810-821 2. Winfree, R et al 2007. Ecol. Let. 10: 1105-1113 3. Garibaldi, LA et al. 2014. Front. in Ecol. and the Env.12(8): 439-447 4. Greenleaf, S 2007. Oecologia 153: 589-596 5. Morandin, LA and C Kremen 2013. Ecol. App.23(4): 829-839.

Project Objectives:

Pollination Services

  1. Assess the spatial distribution of pollinators in agricultural fields
  2. Assess pollination function in fields with and without hedgerows

Economic Evaluation

  1. Generate an economic model that evaluates the marginal benefits of hedgerow restoration along field edges


  1. Assess the nesting bee community in hedgerows and fallow field edges
  2. Correlate habitat features in hedgerows and fallow field edges to nest occurrence


Click linked name(s) to expand/collapse or show everyone's info
  • Dr. Claire Kremen


Materials and methods:

Study System

Fieldwork was carried out in 18 hybrid sunflower fields in California’s Central Valley from June through August 2012- 13. Half of the fields were adjacent to bare or weedy edges (hereafter called controls) and half were adjacent to hedgerows. Each hedgerow was 250-300 m long and 3-6 m wide (Photo 1). Control margins contained only non-native plant species or were maintained as bare, weed-free areas. Hedgerow and control sites were paired based on the timing of the sunflower bloom, the sunflower variety (specific to company), and the landscape context, and sampled during a similar time period. Pairs were a minimum of 900 m apart to maintain independence.

Pollinator Distribution

This work was conducted by a senior thesis student (Kathleen Tom) in 2012. She used fluorescent dyes (Photos 2-3) to mark 100 medium-sized bees along field edges in eight sites, half adjacent to hedgerows. She then recorded the amount of dye deposited over a single day up to fifty meters into the field.

Sampling Pollinator Community on Sunflower and in Edges

We established two 200 m transects within each field, perpendicular to the field edge or hedgerow. We netted and observed pollinators at four distances along these transects (10, 50, 100, and 200 m from the edge). We netted bees visiting sunflower blooms for eight minutes at each distance along each transect, and for sixteen minutes along hedgerow or control edges. We then visually observed sunflower visitation for two minutes each in two male-fertile and two male-sterile 2 x 1 m plots at each distance and in eight plots containing floral blooms in hedgerows or edges.

Seed Set

To determine ambient pollination rates, we collected three sunflower heads at each distance/transect combination when they were ready to be harvested. Heads were dried, measured, and all mature seeds were removed, weighed, and counted.


In the 18 sunflower sites, we set ten .6 m2 emergence traps (Bug Dorm; Photo 4) spaced 20 m apart along a single transect in field edges with and without hedgerows and five emergence traps at 0, 10, 50, 100, and 200 m along each of the two 200 m parallel transects extending into each field. Each e-trap was equipped with a kill jar at its apex filled with soapy water. The edges of the e-traps were secured with soil to prevent any bees from entering or exiting. We placed traps at dusk, after bees had retired to their nests, and emptied them approximately 20-22 hours later. Only female bees are considered in analyses, as male bees may have been resting in vegetation and are not indicative of nesting rates. 

Soil and Nesting Characteristics

Soil characteristics may influence nesting incidence and potentially provide a proxy for nesting habitat suitability in pollination models. We measured mean soil particle size and local soil heterogeneity (soil classes within in field and edge). We collected four soil samples at 10 cm depth at each site, two along a transect in the edges at 40 and 60 m and two in each field at 10 m on T1 and 100 m on T2. We calculated average particle size with a laser diffraction particle size analyzer (Sequoia LISST Portable XLR).

Key habitat features that influence nesting have also been examined as proxies for nesting. We therefore measured percent bare ground, percent vegetative cover, percent leaf litter, presence of cracks and cavities, dead wood, rocks, slope of the ground, and soil surface compaction within each e-trap.

We sampled 16 edge sites; eight with hedgerows. Each site was sampled three times; during spring, summer, and late summer to capture seasonal variability. We set thirty 1 m2 emergence traps along three parallel 350 m transects in field edges with and without hedgerows. Traps were spaced 30 m apart and transects were on the edges of the hedgerow and in line with the hedgerow plants.


Pollinator Distribution: To determine the rate of decay of powder marks we used a non-linear least square regression. To account for the differences in sampling effort, we weighted the number of observations at each site by the number of bees marked with luminous powder. We then assessed the effects of the ratio of female to male bees marked, distance from the marked row, and treatment on the number of weighted powder observations using a generalized linear mixed model with a Poisson distribution with site nested within pair as a random factor. To visualize the distribution of bees in a single field, we simulated nest distribution (see Nesting subsection below) using a log Gaussian Cox process. We then calculated the foraging range from these nests using an exponential decay rate of one, approximately what we found in our powder decay study, which is also the rate of decay utilized in the InVest pollination services model (Lonsdorf et al. 2009). The resulting incidence of bee nests within a field then depicts the expected pollination coverage from medium-sized bees predicted by our data. 

Pollinator Community in Sunflower and Edges: To determine the impact of hedgerow presence and field location (field or edge) on wild bee species richness and abundance (from aerial net data) we used general linear models with poisson and negative binomial distributions respectively. Both models included an interaction between hedgerow presence and field location and a random effect of site nested within pair. We then evaluated the differences between the community of bees in control edges, hedgerows, and crop fields using a perMANOVA on their Chao1 dissimilarities. 

Seed Set: To determine which factors impacted sunflower seed set, we used negative binomial generalized linear models that accounted for overdispersion in the seed data. We examined the effect of wild bee abundance and richness on seed set from net and visitation data separately. Sunflower seed set was the dependent variable and explanatory variables were hedgerow presence, wild bee abundance, wild bee species richness, sunflower company, distance into the field from the edge, and an interaction between netted wild bee abundance and honey bee visitation. Site nested within pair was included as a random effect. 

Nesting: We checked nesting resource variables for collinearity. Because of strong negative correlation with percent bare ground, we removed percent leaf litter but retained all other nesting variables. We visualized the relationship between the standardized nesting variables using a PCA ordination in the R package vegan (Fig. 1). PCA1 explained 22.9 of the variation, and PCA2 explained an additional 17.2. A low score indicated high bare ground and steeper slopes where as a high score indicated more vegetative cover and flatter ground. Along the axis, a positive score indicated more cracks and cavities in the soil, while a negative score indicated an absence of cracks and cavities.

Nesting data contained a high number of zeros and was over-dispersed; therefore, we analyzed nesting abundance using a zero-inflated negative binomial model in the R package glmmadmb. Fixed effects were distance into field, hedgerow presence (hedgerow or control edge), soil particle size, landscape heterogeneity, and year. Because sites were paired in our experimental design, we nested site within pair as a random effect in the model. We analyzed raw species richness using a generalized linear model. We then visually compared rarefied richness by site in fields and edges with and without hedgerows using 100 permutations of the random species accumulation method in vegan. To estimate total species richness across all sites we used a jacknife from the vegan package.

Soil and Nesting Characteristics: We calculated pairwise gower dissimilarities between species composition at each site as well as nesting resources. We then used a mantel test to look for correlation between these pairwise matrices and visualized the results using cluster analysis.

Research results and discussion:

Objective 1

We dyed a total of 743 medium-sized bees with luminous powder, with a median of 101 per site. 72.4% of all bees dyed were in the genus Melissodes, with 428 females and 110 males. 97.2% were sunflower specialists. We observed 464 traces of powder on sunflower heads, with 80.7% concentrated in the first row. Powder marks decayed at a rate of 0.9964 (t = 2.80, P = 0.009) from the transect in which bees were marked (Fig. 1). Distance into the field was the only factor that had a strong effect on the dye marks observed (t = -3.74, P 0.001).

Using the estimated nesting rates (see below) and foraging distances we found, we predict a spatially heterogeneous pattern of ecosystem service delivery within a single crop field (Fig. 2). Instead of allowing a bee with a foraging range as large as a the industrial monoculture fields in our system to forage uniformly across that field, the rapid decline in dye marks we observed indicates a truncated foraging range centralized around nest location. Thus, the distribution and density of nests within a given field would likely influence foraging extent. For example, spatial clustering of nests, which we did not test due to the low numbers of nests encountered, could lead to areas of higher or lower pollination.

Objective 2

We collected 670 wild bees with aerial netting representing 30 species. On average, hedgerow edges supported higher species richness (5.11 ± 0.89, mean ± standard error; Fig. 3a) than control edges (2.11 ± 0.48), hedgerow fields (1.41 ± 0.20), or control fields (2.06 ± 0.20). Similarly, we collected more bees in hedgerow edges (Fig 3b). We detected a significant interaction between hedgerow presence and location within fields (edge or field) for abundance (t = -4.36, P < 0.001) and species richness (t = -4.18, P < 0.001). 

We recorded 2,745 visits to sunflower from wild (339 visits) and honey bees (2,406 visits). We did not find a main or interactive effect of hedgerow presence or distance from the edge on visitation rates, although we detected seven times more honey bees visits than wild bee visits (t = -14.51, P < 0.001).

The communities of bees we found in fields and edges with aerial netting were strongly differentiated (F = 4.11, P = 0.001), but the communities found at hedgerow or control edges were not differentiated (Fig. 4). Bee communities in edges were dominated by generalists (e.g., Halictus tripartitus), whereas bee communities in fields contained higher numbers of sunflower specialists (e.g., Melissodes agilis). 

Within edges, hedgerows supported a more diverse community of bees; however, within fields the effect of hedgerows disappeared. The bee species visiting sunflower seem to be attracted to the crop fields and not the hedgerow plantings, thus assessing how sunflower in the landscape impacts the sunflower-pollinating community could provide more explanation for the trends we observed. 

Objective 3

Seed set was affected by netted wild bee species richness (t = 2.05, P = 0.039) but not abundance (t = -1.27, P = 0.20). We did not detect an interaction effect between netted wild bee abundance and honey bee visitation rates. In the visitation model, the interaction between wild bee and honey bee visitation influenced seed set (t = 2.04, P = 0.041). Neither hedgerow presence nor distance from the field edge impacted sunflower seed set in either the net or visitation models, whereas company strongly affected seed set (Fig. 5). This indicates that while bee richness and interactions between wild and managed bees can positively influence seed set, hedgerows did not affect bee richness within fields.

Objective 4

We collected 95 female ground-nesting bees from e-traps representing 10 species. Our total jacknifed species richness across all sites and years was 15 3. Ground-nesting bees nested in equal numbers in fields and edges (Z = -1.17, P = 0.243), although nesting rates slightly declined with distance into field (Fig. 6; t = -1.90, P = 0.057). We also detected fewer species within field centers than along field edges and in field margins (t = -2.23, P = 0.025). Hedgerow presence did not influence the abundance (Z = -0.30, P = 0.761), incidence (t = -0.51, P = 0.621), or richness (t = -0.88, P = 0.391) of ground nesters. Nesting was influenced by steeper slopes and higher amounts of bare soil ( abundance: Z = 1.68, P < 0.001; incidence: t = -3.11, P = 0.002; richness: t = -3.20, P = 0.001).

We did not detect a preference for nesting within field edges, instead finding that bees nest directly within planted, tilled agricultural fields. Whether their offspring survive the following year, however, is another question. Ullmann and Williams conducted tillage experiments using squash bees (Peponapis pruinosa) and found that only half the offspring emerged following a deep till event (to 16 inches beneath the soil; unpublished data). This indicated that fields might be sink habitat. We also found that bare, sloped ground increased nesting incidence. Most farms in the Central Valley are very flat, thus increasing topography and leaving bare, undisturbed patches could promote ground-nesting bees.

Additional nesting study: In 2013 we began an additional project evaluating the role of irrigation on nesting rates within fields. Collette Yee, through Berkeley’s Sponsored Undergraduate Projects Program, conducted this research as her senior thesis project. She set out 20 traps along two parallel transects into fields, from 0 – 100 m. Her study was carried out in 10 sites, half of which were furrow irrigated and half of which were drip irrigated. She did not find a difference in nesting rate or the abundance or foraging bees in sites with either irrigation management style (Fig 7).

Objective 5

We did not find a correlation between the nesting resources and which species nested in control and hedgerow sites (r = 0.088, P = 0.287). Neither species nor resources clustered based on hedgerow presence (Fig 8). This suggests that either the nesting resources we assess are not focusing on the nesting attributes most important to bee species in nest site selection, or that hedgerows do not alter the nesting resources available in agricultural landscapes above those provided in unmanaged field edges.

Participation Summary

Research Outcomes

No research outcomes

Education and Outreach

Participation Summary:

Education and outreach methods and analyses:

Extension Publications: In collaboration with UC Cooperative Extension, we published four infographics on bees nesting (see attachments). These publications will be made available on the UCCE website, the Berkeley Food Institute website, the Hedgerow Working Group site, and my research blog.

Sardiñas, H.S., R.F. Long, M. Fahey, K. Ullmann, M. Vaughn and C. Kremen. 2014. Native Bee Nest Locations in Agricultural Landscapes. University of California Cooperative Extension, College of Agriculture and Natural Resources.

Sardiñas, H.S., R.F. Long, K. Ullmann, and C. Kremen. 2014. Managing Native Bees in Agricultural Landscapes. University of California Cooperative Extension, College of Agriculture and Natural Resources.

Sardiñas, H.S., R.F. Long, M. Fahey, K. Ullmann, and C. Kremen. 2014. Selecting Hedgerow Plants for the Sacramento Valley. University of California Cooperative Extension, College of Agriculture and Natural Resources.

Workshops: We participated in two pollination workshops put on by UC Extension to over 60 growers in January and July 2014. We presented strategies to increase bee nesting at EcoFarm 2014. We are planning a nest identification workshop for summer 2015.

Conferences: We presented our work at the 2013 Entomological Society of America Conference, 2014 Pacific Branch Entomological Society of America Conference, 2014 Society for Conservation Biology, and the 2014 Ecological Society of America Meeting.

Public Outreach: We published a popular article in the Berkeley Science Review

Two years ago we created a website focused on our research and findings Originally it was updated during the field season, but we are currently updating it at least twice a month with summaries of current scientific research on pollination, other relevant new stories and pertinent information or web-based resources. It also serves as an educational tool, with information about hedgerows and native bee biology. To date, we have received 4,560 views from 32 countries. We average seven views per day.

Public Talks: In June 2014, H.S.S. gave talk at East Bay Nerd Nite on bee conservation. She attended numerous tabling and community events for UC Berkeley garden. She has even given a talk on bee diversity and their importance to the food system at an event highlighting local foods and local honey.

Lectures: H.S.S lectured in agroecology, conservation biology, restoration ecology and student led courses at UC Berkeley and UC Santa Cruz.

Data Sharing: We shared our data with two meta-analyses that are examining native bee crop pollination and the role of habitat enhancements. These studies combine data across a variety of crops and ecoregions and will allow better understanding of trends affecting native pollinators that can translate into management actions. 

Applications: Our findings will be useful to the NRCS, specifically WHIP and EQIP programs that participate in habitat enhancement projects, including hedgerows. Findings may be relevant to other mass flowering crop systems. They also shed light on nesting dynamics. Our findings will also help local non-profits that plant hedgerows, including Audubon California and the Xerces Society for Invertebrate Conservation.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.