Rufina N. Ward and Kenneth E. Ward

Department of Plant and Soil Science, Alabama A&M University


This 2-year study investigates the relationship between cotton production and proximity of supplemental honeybee sources in an irrigated Bt cotton field, using a similar field without bees for comparison. During year 1, cotton yield indicators declined with increasing distance from bee sources in the bee field; this corresponded with decreasing honeybee foraging activity with increasing distance from bee sources. There were no such trends between yield indicators or bee activity and distance from field corners in the non-bee field. Similar trends were found during year 2 of the study, when field status (bee or non-bee) was reversed. There were, however, significant numbers of honeybees in the non-bee fields both years. These results suggest a significant positive impact of supplemental honeybees on cotton yield.


Upland cotton, Gossypium hirsutum L., is usually referred to as a partially cross-pollinated crop, although many breeders have treated it as a completely self-fertile and self-pollinating crop, except for accidental cross-pollination via pollinating insects. Although isolated cotton flowers normally develop mature bolls with viable seeds, breeders have long noticed that such isolated blossoms usually don’t produce quite as many seed or as much lint as open-pollinated ones (Stephens1956). Several investigators have demonstrated increased yields in cotton that was artificially cross-pollinated (Meade 1918, Rose and Hughes 1955, Kohel 1968, Guseinov and Muktarov 1963). Increases in yield as a result of cross-pollination may be related to such factors as increased numbers of pollen over the entire surface of the stigma (within a flower, pollen-laden anthers contact only the base of the stigma) and increased rate of pollen tube formation in pollen grains from other flowers (Kearney 1926, Iyengar 1938, Arutiunova and Gubanov 1950, Arutiunova 1940). These factors may result in higher fertilization of ovules, resulting in higher levels of fruit setting and seed production.

Numerous authors have cited benefits derived by cotton from insect pollination. Shishikin (1946, 1952) demonstrated that saturation pollination in areas at the rate of 1/2 colony of honey bees per acre increased cotton production by 19.5% over areas depending only upon natural pollinators. In cage-grown cotton, which included honeybees but excluded natural pollinators, cotton production was increased 43% over caged cotton without pollinators. McGregor et al (1955) studied effects of bee pollination in cages, some of which contained a colony of honey bees. Pima cotton (Gossypium barbadense L.) production was increased 24.5% in cages with bees over cotton without bees; upland cotton did not show an increase in yield but set fruit earlier, which would have produced a greater yield in an area with a shorter growing season. In addition, there were fewer motes (cotton ovules that fail to develop into seeds with well-developed ginnable fibers) in bolls on cotton caged with bees; the presence of motes decreases efficiency of mechanical harvesting. If cotton had been mechanically harvested in this study, instead of being handpicked, yield differences between cotton caged with and without bees would have been even greater. Radoev and Bozhinov (1961) obtained an 11-24% greater yield from cotton flowers freely visited by bees compared with flowers tied to exclude insects; there were more seeds per boll, fewer motes and improved seed germination. Finally, Tanda (1984) found bee pollination to increase yield in two interplanted varieties of Asian cotton (Gossypium arboreum L.).

In spite of several studies indicating the potential benefits of pollinators to cotton production, there has been relatively little interest in this subject in recent years, (other than hybrid seed production), at least in the U.S. This is probably due to the decimation of populations of pollinators near and within cotton fields as a result of the heavy use of synthetic organic insecticides since the 1950s; in addition, the varroa mite has heavily impacted feral honey bee populations. However, boll weevil eradication and the use of transgenic Bt-cotton varieties are dramatically reducing insecticide use on cotton over much of the southeastern U.S. It may now be possible to establish and maintain populations of pollinators, especially honey bees, in many cotton-producing areas that are large enough to significantly improve cotton production. This could benefit cotton producers and beekeepers alike; cotton is an excellent source of nectar for honey production and, in some areas, is an important late-season nectar source when other sources are scarce.

We present here the results of a two-year field study designed to assess the effects of supplemental honey bees on production in an irrigated Bt-cotton field compared to a similar field without supplemental bees. We hypothesize that cotton production will decrease with increasing distance from bee sources.


Study Area, Pollinator Sources- The study area was located in northern Madison County, Alabama, at Tate Farms, approximately eight miles north of the campus of Alabama A&M University. The study was conducted in two fields (Field 1= 140 acres; Field 2= 150 acres) 1.5 miles apart which have been used exclusively for irrigated cotton production for the past 10 years; both fields have center-pivot irrigation systems. Yearly records on yield and other significant data are available for these two fields. .The Alabama Beekeepers Association provided the honey bee colonies used in this study.

Study Design and Data Collection- The study was initiated July 16, 1999, when all honeybee hives had been set up and cotton had started flowering. One hundred forty colonies (1 colony per acre) of honey bees were divided between the NW and NE corners of the bee field outside the irrigated area.. A transect sampling line was established perpendicular to and running diagonally away from each group of hives to the opposite corner of the field (see experimental setup diagram). Fifteen sampling points were established at intervals (100, 200, 400 feet) along each transect line at increasing distances from honey bee colonies. There were 5 sampling points for each distance interval in each transect. A 6-ft high bamboo stake marked each sampling point. A similar set-up was used for the non-bee field, except that no bees were placed in field corners. Twenty 1-2 day-old flowers (2 flowers/plant) were tagged at each sampling point in both fields for four consecutive weeks beginning in mid-July, approximately one week after initiation of flowering. Plants with blooms were randomly chosen prior to tagging the flowers with aluminum tags. All flower tagging was conducted between 8:00 and 11:30 a.m.. Observations of foraging honey bees were also made during the 4 weeks of flower tagging, between 10:00 a.m. and 12:00 noon. Bee activity at selected sampling points (every other point along one transect line in each field) was measured by counting the number of bees foraging within a 20-ft radius of the sampling point stake for 3 minutes. All boll samples from tagged flowers were collected just prior to harvest (post-defoliation). Number of bolls per 80-tagged flowers was determined for all samples, followed by ginning and determination of lint and seed weights.

Study design and data collection for year 2 of the study were similar except for the following:

Bee treatments were reversed for the two study fields.

Flower tagging was started approximately two weeks after initiation of flowering (due in part to later placement of bee colonies)

An additional small boll collection was made (denoted herein as "early season") to account for possible loss of important information due to the later initiation of flower tagging. Just prior to harvest, twenty pairs of bolls were collected from the third node, first position of twenty randomly selected plants at each sampling point.

Data Analysis- Data were analyzed via linear regression analysis (SAS-Proc Reg). Two regression equations were developed to determine (1) relationship between cotton yield indicators (dependent variables) and distance from bee colonies/field corners (independent variable), and (2) relationship between bee activity estimates and distance from bee colonies/field corners. Yield indicator variables were weight per boll (seed/boll + lint/boll), and boll number for each sample. Data points receiving no or partial irrigation were eliminated from the analysis. Separate equations were developed for each field each year.



Cotton production and proximity of bee source. Regression analysis indicates a significant relationship between per boll cotton yield indicators and distance from bee source in bee fields. Per boll (lint + seed) weights for tagged flower samples (weeks pooled) declined with increasing distance from bee sources during both years of the study; this negative trend was significant for year 1 (0.01<p<0.05), but not year 2 (0.15<p<0.20; Fig. 1). Per boll weight for early season samples (year 2 only) showed a strong decline with increasing distance from bee source (p<0.01; Fig. 1). In contrast, there were no trends detected between cotton yield indicators and distance from field corners in non-bee fields during either year of the study (p>0.20) (Fig. 1). This suggests that the effect seen in bee fields was not related to proximity to field edges. Negative trends between yield and distance from bee sources in bee fields were strongest during mid- to latter- stages of flowering during both years of the study

The relationship between boll numbers per 80 tagged flowers and distance from bee source or field corners was not consistent between years. A negative trend, consistent with other yield indicators, was detected in the bee field during year 1 (0.01<p<0.05) (Fig. 2), with no trend for the non-bee field (Fig. 2). However, significant positive trends (p<0.05) between boll number and distance from bee source/field corners were detected for both bee and non-bee fields in year 2 (Fig. 2). Positive trends in year 2 may have been related to the timing of data collection, which was initiated one week later relative to advent of flowering, combined with differences in flowering time between field edges and field centers.


Bee activity and proximity of bee source. As expected, bee activity was significantly greater in bee fields than in non-bee fields (p<0.05). Additionally, bee activity data in bee fields showed a significant declining trend with increasing distance from bee sources (p<0.05; data pooled for weeks and years; Fig. 3). Bee activity levels in bee fields declined to levels roughly similar to those of non-bee fields ca. 1500- 2000 ft from bee sources. We did not find any trends between bee activity and distance from field corners in non-bee fields; nevertheless there was significant honeybee activity observed in non-bee fields (Fig. 3).


1. This study suggests a positive impact of supplemental honey bees on cotton yield indicators in Bt cotton, as indicated by decreasing per boll weights with increasing distance from bee sources. Trends were consistent over both years of the study, as bees were switched from one study field to the other. No such trends were detectable in non-bee fields during either year of the study.

2. Bee activity patterns were consistent with those of yield indicators, significantly decreasing with increasing distance from bee sources, with no trends in non-bee fields. Activity levels in bee fields decreased to approximate those in non-bee fields ca. 1500-2000 ft from bee sources. These results were also consistent over both years of the study.

3. These trends were detected in producer fields, with limited control over experimental conditions. Other sources of variation included trends toward greater growth rates of cotton and later flowering with increasing distance from field edges, variation in weather conditions between years, etc. None of these sources of variation are consistent with the patterns observed and may have partially masked a stronger impact of honeybees.

4. Our ability to detect an apparent honeybee impact on cotton yield in such a small study suggests the need for additional larger studies designed to quantify such an effect.


Arutionova, L. G. and G. I. A. Gubanov. 1950. In regard to the biology of cotton fertilization. Agrobiologiya 6: 94-99.

Guseinov, A. M. and Z.M. Mukhtarov. 1963. New data on effectiveness of intravarietal hybridization of cotton. Agrobiologiya 4: 490-496.

Iyengar, N. K. 1938. Pollen tube studies in Gossypium. Jour. Genetics 37: 69-106.

Kohel, R. J. 1968. Effect of genotype and heterozygosity of pollen source and method of application on seed set and seed and fiber development in cotton. Crop Science 8: 293-295.

Mahadevan, V. and K. C. Chandy.1959. Preliminary studies on the increase in cotton yield due to honeybee pollination. Madras Agr. Jour. 47: 23-26.

McGregor, S. E. 1958. Relation of honeybees to pollination of male-sterile cotton. In Eleventh Cotton Impr. Conf. Proc.:93-94. Houston.

Meade, R. M. 1918. Bee keeping may increase the cotton crop. Jour. Hered. 9:282-285.

Radoev, L. 1963. Studies on bee pollination and honey productivity of cotton. In 19th International Apic. Cong. Proc., Liblice, Czechoslovakia. p.99.

Rose, M. F. and L. C. Hughes. 1955. Progress report of the cotton breeding station, season of 1954-55. Cesira Research Station, Wad Medani: 2-12.

Shishikin, E. A. 1946. Honey bees in the service of cotton pollination. Pchelovodstvo 23: 31-32.

Shishikin, E. A. 1952. Effect of pollination by honeybees on increasing the productivity of cotton. In Krishchunas, I.V. and A. F. Gubin (eds.) Pollination of Agricultural Plants. Moska, Gos. Izd-vo Selkhoz Lit-ry, pp. 95-103.

Stephens, S. G. 1956. The composition of an open pollinated segregating cotton population. Amer. Nat. 90: 25-39.

Tanda, A. S. 1984. Bee pollination increases yield of 2 interplanted varieties of Asiatic cotton (Gossypium arboreum L. ). The Amer. Bee Jour. 124: 539-540.


We would like to thank Tate Farms for providing the study fields and making it convenient for us to complete this project; the Madison County Beekeepers Association and the Alabama Beekeepers Association (including especially Mr. Bill Mullins and Mr. Bob Fanning) for their support and for providing honey bee colonies for this study; Mr. Bill Mullins for coordinating the provision and placement of honey bee colonies, helping with field work and cotton ginning and providing essential guidance and support; Ms. A’ishah Farid and Mr. Cedric Hudson for their hard work in the field and in processing data; and Ms. Beth Dunson and her daughters, Michelle and Kim, for helping with the cotton ginning and field work.


            Beefield.GIF (5879 bytes)(Bee Source)

T1 = Transect 1 T2 = Transect 2





Figure 1. Per boll weights as related to distance from bee source/field corners, bee and non-bee fields.






Figure 2. Boll numbers as related to distance from bee source/field corners, bee and non-bee fields.


Figure 3. Bee activity (weekly samples pooled) as related to distance from bee source/field corners.