Enhancing honey production with clover: Innovative methods to use white and alsike clover in Vermont hay fields

Final Report for ONE13-175

Project Type: Partnership
Funds awarded in 2013: $14,848.00
Projected End Date: 12/31/2015
Grant Recipient: University of Vermont
Region: Northeast
State: Vermont
Project Leader:
Dr. Sid Bosworth
University of Vermont
Expand All

Project Information


A decline in honeybee populations in Vermont over the past few decades has been attributed to many factors including mites, disease and a loss of nectar and pollen resources.  An important stress on honeybees in Vermont is the lack of food available to them throughout the summer.  

The goal of this project is to increase the acreage of flowering clovers that provide sustained nectar flows during this critical summer period.  To test this idea, an on-farm, replicated strip trial was planted in late May 2013 in Bridport, VT comparing pure alfalfa to alfalfa mixed with one of two different white clover varieties or with alsike clover at two seeding rates.  A similar trial was planted in small plots at the University of Vermont in August 2013.  

Data was collected at both sites in 2014 but only at the UVM site in 2015 (the Bridport site was terminated due to severe winter injury).  At a 2 to 3 lb. clover seeding rate when mixed with alfalfa, we found no differences in yield and quality compared to pure alfalfa.  ‘Pinnacle’ white clover far exceeded ‘Crusade’ white clover and alsike clover in flower abundance and seasonal distribution reaching flower head populations up to 280,000 per acre when mixed with alfalfa.     

In 2015, clover flower head populations were only about a third of the amount compared to the previous year due to fewer clover plants as a result of winter injury.   This project demonstrated that by mixing properly selected cultivars of white clover with alfalfa, one could achieve a flower head population that could provide nectar and pollen resources for pollinating bees that could be sustained over the summer period.   However, the lack of sustained high populations into the second year study demonstrates the year-to-year variation that can occur due to extreme weather conditions. 

More research is needed to test identify other white clover cultivars that can sustain flower abundance throughout the season and test the resilience of more diverse mixtures of forage legume species and cultivars.


Honeybee on white clover flower A decline in honeybee populations in Vermont over the past few decades has been attributed to many factors including Varroa mites, disease, pesticide exposure, and a loss of nectar and pollen resources.  Many beekeepers in Vermont believe that the lack of food during mid-to-late summer is an important stressor on honeybees in Vermont.   Historically, an important food resource for honeybees in Vermont has been from forage legumes such as alfalfa, clovers and birdsfoot trefoil.  This resource has declined over the past two to three decades due in part to
a decline in acreage of legume hay crops, particularly alfalfa, and an increase in intensity of harvest by dairy farmers in order to optimize feed quality resulting in very little bloom periods. 

The goal of this project was to increase the acreage of flowering clovers that provide sustained nectar flows during this critical summer period.   In Vermont, the best clovers for honeybees are white clover and Alsike clover since their florets are small and accessible.  Although not a true clover, Birdsfoot trefoil is also highly regarded by bee keepers for its nectar and pollen.  Red clover has larger florets more suited for larger pollinators such as bumblebees.  

With this project, UVM Extension in collaboration with The Vermont Beekeepers Association has set an initiative to promote the use of more pollinator plants that would enhance food resources for honeybees and other wild pollinators.   As part of this initiative, the VBA would like to promote hay and pasture crops that are more ‘bee friendly’ without sacrificing forage quality that dairy and other livestock farmers are dependent upon.  However, there is a need to conduct field research to determine the feasibility of various mixtures and management practices that would enhance forage legume bloom in mid-to-late summer while dairy livestock farmers still meet their feed goals of high quality forage.

Project Objectives:

The overall goal was to test the feasibility of using clovers in hay systems to enhance nectar production during the mid to late summer periods. 

The first study objective was to test the feasibility of improving nectar flow for honeybees by introducing Dutch white clover into grass hay cropping systems and assess its impact on flower production and foraging honeybees.  The 'Dutch' white clover are of the shortest types and are often found in lawns and other intensely defoliated areas.  Our hypothesis is based on the observation that there can sometimes be a flush of white clover bloom in between cuttings of grass hay.

The second study objective was to test the feasibility of growing mixtures of flowering clovers with alfalfa managed for hay and assess its impact on flower production, forage yield and forage quality.  Our hypothesis was that white clover and/or alsike clover would bloom before the alfalfa was ready to cut and sustain a bloom until the stand was cut.


Click linked name(s) to expand/collapse or show everyone's info
  • Tom Duclos
  • Art Huestis
  • Charles Mraz


Materials and methods:

Objective 1 - In April 2013, a strip trial was planted into an existing grass hay field at the Duclos-Thompson Farm in Weybridge, Vermont with the assistance of Mr. Tom Duclos.  Treatments included a no-treatment control plus four different seeding scenarios using Dutch white clover: two seeding rates (2 vs. 4 lbs./acre) using two methods of seeding: (no-till planted with  a  Haybuster 107C drill or broadcast seeding with a spinner-spreader (Figures 1 and 2).  Strips were 50 feet wide and 150 feet long.  There were two replications.  Stands were monitored for clover germination  and clover head counts were made throughout the season.  On each day that clover heads were counted, ten 20” x 20” quadrat counts were made per treatment strip and used to estimate clover head populations. 

Figure 1 Figure 2







Figure 3


Objective 2 - Two trials were implemented in 2013 (Table 1).  The first one was a replicated strip trial planted at the Huestis farm in Bridport, VT on a Vergennes moderately well-drained clay.  Mr. Art Huestis had managed this site for over 40 years successfully growing alfalfa for his dairy herd.  The sites was tested for soil nutrients and received recommended rates of phosphorus and potassium.

The study included seven treatments (Table 1) replicated four times, including two seeding rates of three clovers.  Each plot was 30 by 180 feet.  Mr. Huestis planted the whole area on May 3, 2013 with a single seeding rate of alfalfa at 10 lbs. per acre and one bushel of oats.  Clover treatments were broadcast planted the same day within each respective plot using a hand spinner spreader.  The clovers included ‘Pinnacle’ and ‘Crusade” white clover and ‘common’ alsike clover seeded at two seeding rates, 2 and 4 lbs. per acre, respectively.  There was also a non-clover control consisting of alfalfa only. 

After planting the site, there was a month of record rainfall which kept the site saturated with water.  Due to the inclement weather, there was only one cutting made in early August.  An assessment of the stand early September showed that the percentage of alfalfa in the stands ranged from 47 to 55 percent in the clover mixture plots and 71% in the alfalfa-only control plots.  The clovers ranged from 21% to 41% of their respective stands depending on clover variety and seeding rate.  The alfalfa-only control had about 10% clover reflecting what was in the soil seedbank prior to planting.

A second small plot, replicated study was seeded August 22, 2013 at the UVM Horticultural Research Farm on a well-drained, Deerfield fine sandy loan.  The clover treatments were the same as those at the Huestis farm, but the seeding rates were 3 lbs. and 5 lbs. per acre, respectively, and the alfalfa seeding rate varied (see Table 1).  In addition, this study also included a mixture of alfalfa with all three clovers combined seeded at two different seeding rates.   The site was slow to establish due to a dry August and September.  However, the stand was in adequate shape by spring of 2014 although it had to be hand weeded.

Clover head populations were collected from each plot every one to two weeks starting in June after the first harvest was made and ending in early September.  For each strip plot at the Huestis farm, counts were made from ten 20” x 20” quadrats run diagonally across the strip.   At the Hort farm, three quadrat counts were made per plot.  White clover and alsike clover heads were recorded separately per plot.

At the Huestis site, yields of the 2nd and 3rd harvest were determined using a falling plate method (Rayburn, 1997). Ten falling plate heights were collected per strip one day before the second and third harvest, respectively.   The falling plate was calibrated by clipping 12 to 16 quadrats associated with a falling plate height measurement.     At the Hort Farm site, yields were determined for each cutting using a Carter small plot harvester by weighing a harvested sample cut from a 3’ by 16’ strip out of the center of each plot.  Hand sub-samples were then collected from each plot, bagged, weighed, dried at 60oC, and reweighed to determine dry matter content.

In 2014 at the Hort Farm, samples from selected treatments were collected during the 3rd harvest to determine crude protein, neutral detergent fiber, and digestibility.  Botanical separations between alfalfa, clover, and weeds were made at the same time.

Table 1


Research results and discussion:

Objective 1: Utilizing Dutch White Clover in Grass Hay Fields

Within a month after planting, there was a higher number of white clover seedlings at the 4 lb./a compared to the 2 lb./a seeding rate regardless of seeding method (data not shown); however, by the middle of the summer, the actual flower head populations were similar across treatments (Figure 5).  Only during the second growth period were the seeded treatments higher in flower counts compared to the control; however, the control plots had a large number of white clover flowers indicating that there was already a large seedbank that existed on this field.    It is likely that flower population did not differ between seeding rates because white clover flower numbers are more affected by stolon population rather than actual plant population.  In addition, it is likely that existing white clover seed already present in the soil may have germinated and contributed to the stand in 2013.


In 2014, the study was dominated by volunteer red clover.  It is likely the seedbank was quite high in red clover and conditions in early 2014 encourage germination.  Due to this domination, there was significantly less white clover across all treatments and no measurements were made that year.  In 2015, flower head data was collected after the first harvest to see if there was any residual effects of the 2013 seeding treatments.  Due to the extremely wet month of June, the first cut was not made until the first of July.  An extended dry period followed and the second harvest was not made until mid-August.  Overall, there were no significant differences amongst the treatments (Figure 5).  There was no way to distinguish between any Dutch white flower heads from any other cultivar or ecotype that may have already been in the soil seedbank at this site.  Clover populations were lower in 2015 than 2013 at this site.  This may reflect the year to year variation observed in clover abundance in pastures and hay fields due to variation and extremes in rainfall, dry periods and temperature patterns.

Figure 5

Objective 2: Enhancing Summer Nectar Production with Alfalfa-Clover Mixtures

Forage Yield and Stand Persistence – At the Heustis site in 2014, there were no significant differences in yield of either the second or the third harvest for any of the clover mixture treatments nor the pure alfalfa (data not shown).  Our observation was that the clover was most prevalent in the areas of the field with the thinnest alfalfa.  This was particularly true for the alsike clover and somewhat for the ‘Crusade’ white clover.  ‘Pinnacle’ white clover tended to be better dispersed across the strips as determined by clover flower head counts.  Unfortunately, the complete stand at this site was killed out during the winter of 2014/2015 so we only got data from the 2014 year.  It was a very frigid winter with variable snow cover in the Champlain Valley region resulting in a high amount of alfalfa winter injury or winter kill across the region.

At the UVM Hort Farm location, total seasonal yields were generally not affected by most single-clover treatments (Table 2).   Only the high seeding rate of three-way clover seeding (Treatment #9) had consistently lower yields in both 2014 and 2015.  The combination of all three clovers and the high clover seeding rate may have created too much competition for the alfalfa in the seeding year resulting in a lower density of alfalfa plants.  Observations showed that the majority of the clovers in this treatment were white clover.

Table 2

In 2014, the higher proportions of weeds in the alsike clover treatments (#6 and #7) most likely offset potentially lower alfalfa-clover yields for those treatments in 2014.  There was also a lower amount of clover in the alsike treatments compared to the other clover-alfalfa treatments.  Weed content was lower in the other treatments indicated a more dense stand of alfalfa and clover.  The white clover cultivars made a significant contribution to the mixture, especially at the 5 lb. seeding rate (#3 and #5).     

Overall clover content was significantly lower in 2015 compared to 2014, most likely due to the harsh winter of 2014/15.  This was particularly the case for alsike clover and ‘Crusade’ white clover.  ‘Pinnacle’ white clover was also reduced not as much.  Alfalfa did not appear to be affected by winter injury at this location.   Other than treatment #9, the alfalfa tended to have a significant increase in content in 2015 compared to the 2014 levels in their respective treatments.  The dry weather from July through September most likely favored the alfalfa over the other species and was the most dominant species by the time the botanical separations were made on August 14.  There were generally more weeds in 2015 in the higher clover seeding rate plots due to the lower density of alfalfa caused by the clover competition in the seeding year.

Forage Quality –Only selected treatments were assessed for quality during third harvest in 2014.  The samples were hand separated, weighed and re-composited for analysis (Table 3).   Our hypothesis was that white clover would have little or no negative impact on the mixture since most of its biomass is made up of leaf material and some petiole.  This turned out to be the case.  There was no significant difference in crude protein, acid detergent fiber or neutral detergent fiber amongst the alfalfa-clover mixtures compared to the pure stand of alfalfa.   However, NDF digestibility was significantly higher in the two white clover mixtures compared to pure alfalfa or the alsike mixture, with a strong relationship between clover content in the mixture and NDF digestibility (Fig. 6).   The alsike clover made such a small contribution to the mixture and was essentially the same as the pure stand of alfalfa.

Table 3

Figure 6

Clover Flower Abundance and Distribution – White and alsike clover flower populations from the two research sites in 2014 are shown in Figures 7, 8 and 9.   White clover was much more prolific at producing flowers compared to alsike clover.  After each harvest, white clover would produce bloom within the first week to 10 days after regrowth and would continue to bloom until the next harvest.   On the other hand, alsike clover flowers took longer to develop than white clover and would not have a significant number of flower heads until at least two to three weeks into regrowth.  Most of the bloom from alsike clover was in June with less bloom in the later part of the season; however, the response of alsike clover was quite different between locations.  At the Heustis site (Figure 7), there was hardly any alsike clover heads; whereas at the UVM site (Figure 8), there were a significant amount of alsike clover flowers in 2014 in both the June and July growth period.

There were differences in white clover cultivars.  ‘Pinnacle’ white clover had the most abundant flowers and the most widely distributed throughout the season (Figure 7, 8 and 9).  Mid-summer ‘Pinnacle’ flower populations reached 250 to 300 thousand heads per acres at both locations.  ‘Crusade’ white clover, had its most flower production in June but tapered off in July and August.

Figure 7

Seeding rate did not appear to have much effect on clover flower head populations at the Huestis Farm (Figure 7), whereas,  the higher seeding rate did result in more flower heads at the Hort Farm location in 2014, particularly during the June growth period (Figure 8).    However, yields were slightly depressed at the higher seeding rates (discussed in earlier section).   Overall, a seeding rate of 2 to 3 lbs./acre seems satisfactory to for white or alsike clover when mixed with alfalfa.

The Hort Farm location also included a treatment that combined all three clovers with the alfalfa at two seeding rates (Treatments #8 and #9 in Table 1).   We found most of the clover heads to be white clover (data not shown) and the overall populations (white and alsike clover heads combined) were still less than the ‘Pinnacle’ white clover treatments at their respective seeding rates (Figure 8).   There appeared to be no advantage of having the diverse mixture treatment in this study under these conditions.

Figure 8

Figure 9

There are two possible reasons to explain the differences in flower head production amongst the three clover treatments – 1) differences in their response to day length or 2) differences in their ability to compete with the alfalfa.   The most likely reason that the white clover cultivars differed in floral abundance and seasonal distribution was due to their response to day length.  When looking at the 2014 Hort farm data during the June growth period, flower head populations were somewhat similar between the two cultivars in the study; however, as days shortened in July and August, ‘Crusade’ white clover produced far fewer flower heads than ‘Pinnacle’ white clover.  This occurred regardless of if each of the clover cultivars were grown in mix with alfalfa or when grown alone (Figure 10) indicating a day length response.  On the other hand, alsike clover had similar populations during the June and July/August growth periods, only falling off in September.    The competition from the alfalfa also played a role since all the clovers had fewer flower heads in the mixed stands compared to the pure clover stands.  However, the relative differences in clover flower head populations between the pure stands and the mixed stands for each respective clover treatment was not that different; therefore, we cannot conclude that the reduction in late season flowers of the ‘Crusade’ white clover and alsike clover treatments was due to competition. 

Figure 10

Clover populations were much lower in 2015 compared to the previous year resulting in half to a third of the flower head populations (Figure 11).  The only treatment that exceeded 200,000 flower heads per acre was the three way clover, high seeding rate (# 9) but that treatment resulted in lower yields and more weeds in the stand.  ‘Pinnacle’ white clover performed best but was still far lower in floral abundance than the previous year.

Figure 11

Floral Resources and Bee Observations

There is limited information in the literature on reported white clover flower populations and most are from seed production fields.  Weaver (1965) reported an average of 17.3 seed head per square foot in a bee behavior study in Mississippi.  That would extrapolate to 753,588 flower heads per acre.   Goodwin et al. (2011) reported white clover flower head populations between 371,033 and 727,449 per acre in two New Zealand seed fields during peak flower.  Our research showed that populations of ‘Pinnacle’ white clover flower heads were within the lower end of this range throughout the season in 2014, indicating that there was likely enough flower abundance to provide a critical level of nectar production for honeybees.  ‘Crusade’ white clover had a moderate abundance in June at the Hort farm site but failed to provide sufficient flowers at the Heustis site nor in the later summer period at either location.  Alsike clover was too low in abundance in most cases.  The combination of all three (Treatments 8 and 9 at the Hort farm site) was also adequate but most likely was made up primarily of ‘Pinnacle’ white clover. 

At the Duclos site which consisted of mixed grasses, white and red clovers and weeds, our overall populations of white clover were much lower than in the seeded alfalfa/clover fields.  This may have been due to the competition of the other species in the mixture including many weed species particularly smooth bedstraw and dandelion. 

Bee observations collected in 2014 and 2015, showed that the most predominate pollinators were honeybees (Apis spp.); however, we also found native bumblebees (Bombus spp.) and sweat bees (Augochlora pura) working the clover flowers.   The plots were too small at the Hort farm to make any conclusions about preferences for clover type and cultivar and at the Huestis farm, there were not enough flowers from the alsike and ‘Crusade’ strips to make any comparisons.  Overall, our limited bee observations indicated they had no discretion between the clover types.


Goodwin, RM, HM Cox, MA Taylor, LJ Evans and HM McBrydie.  2011.  Number of honey bee visits required to fully pollinate white clover (Trifolium repens) seed crops in Canterbury, New Zealand.  NZ J. Crop and Hort. Sci. 39: 7-19.

Weaver, Nevin. 1965.  Foraging behavior of honeybees on white clover.  Insectes Sociaux, Vol. XII, pp. 231-240.

Research conclusions:

This project demonstrated that by mixing properly selected cultivars of white clover with alfalfa, one could achieve a flower head population that could provide nectar and pollen resources for pollinating bees that could be sustained over the summer period.  We also demonstrated that this could be achieved without negatively impacting forage yield and quality.  

The reduction in clover populations in the second year of the study demonstrates the year-to-year variation that can occur with clovers due to extreme weather conditions.  Therefore, for this practice to be sustainable and reliable, it may be necessary to formulate more diverse mixtures of legumes.  However, we did not test diverse mixtures in this study and cannot make any firm recommendations at this time.

As a result of this project, we have raised awareness of the issue of bee decline in Vermont and New England.  We have partnered with a New England Pollinator Working Group (sponsored by USDA IPM) and developed a collaboration with an insect/plant ecologist at the University of Vermont to further investigate the feasibility of developing a “bee friendly” hay mixture.



Participation Summary

Education & Outreach Activities and Participation Summary

Participation Summary:

Education/outreach description:

The were several outreach activities as part of this project.  A dedicated website for the project was developed in the first year.  It's main purpose is to post research updates and informational fact sheets. The URL is: http://pss.uvm.edu/beeclover/

The following publications were written and posted on the project website:

  • Developing a Bee Friendly Pasture (fact sheet)
  • Research Project Update - 2014
  • Research Project Update - 2015
  • Research Project Final Report

The following presentations were made about this project:

  • Jan. 2014 - Vermont Bee Keepers Association Annual Meeting - Project Overview
  • July 2014 - SARE Pollinator twilight meeting sponsored by John Hayden - presented an overview of this project, Jefforsonville, VT
  • Nov. 2015 - New Hampshire Pollinator Summit, Concord, NH
  • Dec. 2015 - Northeast Certified Crop Advisor Training, Syracuse, NY
  • Jan. 2016 - Vermont Grazing and Livestock Conference, Fairlee, VT
  • Jan. 2016 - Vermont Bee Keepers Association Annual Meeting - Project Summary

The following are media activities resulting from this project:

  • The Vermont Farm Bureau magazine “Vermont Fences” highlighted the project in their Spring 2014 edition.
  • An "Across The Fence" daily television program (UVM Extension and WCAX, Burlington, VT) highlighted the project featuring Sid Bosworth and Charles Mraz.

Figure 12




Project Outcomes

Project outcomes:

No economic analysis was conducted with this research.  Our goal was to introduce flowering clovers (white clover and alsike clover) without having a negative impact on forage yield or quality.  Based on our findings, we achieved that goal if the clover is seeded at no more than a two to three pound rate per acre with the alfalfa; therefore, we can conclude there was no negative economic impact.  We cannot make the assumption that other forage legume or clover species would have the same impact since we did not test them.

Farmer Adoption

We have no information on actual farmer adoption of these practices as a result of this project.  During the same time period of this project, UVM Extension in Middlebury had been providing a no-till drill to farmers for seeding legumes into hayfields and pastures; therefore, we had to opportunity to talk to farmers about the issues but did not have the results yet to make any specific recommendations.  Now that we have some information on suitable cultivars, we will be able to incorporation this into our general recommendations. 


Assessment of Project Approach and Areas of Further Study:

Areas needing additional study

  1. In order to formulate "bee friendly" hay mixtures, more information is needed on the flower abundance and seasonal distribution of white clover cultivars as well as other forage legumes.  
  2. Research is needed to study diverse mixtures of forage legume species and cultivars to test the resilience of the stand to extreme weather conditions.
  3. Even with the success of flower abundance in "hay friendly" mixtures, there needs to be further research evaluating the actual impact on honeybee and native bee health and success as pollinators.

Information Products

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.