PLANTÐINSECT INTERACTIONS

Pollen Selection by Feral Honey Bee (Hymenoptera: Apidae) Colonies in a Coastal Prairie Landscape KRISTEN A. BAUM,1, 2 WILLIAM L. RUBINK,3 ROBERT N. COULSON, VAUGHN M. BRYANT, JR.4

AND

Knowledge Engineering Laboratory, Department of Entomology, Texas A&M University, College Station, TX 77843Ð2475

Environ. Entomol. 33(3): 727Ð739 (2004)

ABSTRACT The collection of pollen by honey bees (Apis mellifera L.) provides valuable pollination services for many plants and the protein necessary for brood and young worker development. We collected and identiÞed pollen gathered by feral honey bee colonies living in tree cavities in a coastal prairie landscape over the duration of 1 yr. SpeciÞc objectives included evaluating overlap in pollen use between colonies throughout the year, examining the inßuence of the spatial locations of the colonies on overlap in pollen use, and describing general pollen collection patterns. The feral colonies collected a wide variety of pollen types. Anemophilous (wind pollinated) pollen types were important in the fall, but entomophilous (insect pollinated) pollen types were important for the remainder of the year. Herbaceous plants and shrubs provided pollen during the spring and early summer, trees in mid- to late summer, and herbaceous plants in the fall. The pollen sources used by the feral colonies also tended to be good nectar sources. Overlap in pollen use between colonies varied throughout the year. Pollen overlap was correlated with distance for some sampling periods and not others, probably because of the way colonies select resources and the ßowering phenology in the study area. KEY WORDS Apis mellifera, honey bee, pollen, feral colony, coastal prairie

POLLEN IS AN IMPORTANT protein source for honey bee (Apis mellifera L.) colonies and is required for brood and young worker development (Maurizio 1950, Haydak 1970). Pollen also provides lipids, carbohydrates, vitamins, and minerals (Roulston and Cane 2000). Honey bee colonies collect a wide variety of pollen types, including pollen from most angiosperm groups, as well as gymnosperms, ferns, and even non-nutritional particulate matter, such as coal dust and sawdust (Schmalzel 1980, Buchmann et al. 1992). This broad diet breadth is important for colony survival because honey bee colonies are perennial and active throughout much of the year. The collection of pollen by honey bees also provides valuable pollination services for many plants, including economically valuable crops, ornamentals, and native species (Buchmann and Nabhan 1996, Ingram et al. 1996, Allen-Wardell et al. 1998, Morse and Calderone 2000). Traps often are used to collect pollen loads for identiÞcation from colonies located in human-made hives (Synge 1947, Poulsen 1973, Adams et al. 1978, Severson and Parry 1981, OÕNeal and Waller 1 Current address: 206 Life Sciences Building, Department of Biological Sciences, Louisiana State University, Baton Rouge, LA 70803Ð 1715. 2 E-mail: [email protected]. 3 Current address: P.O. Box 2686, Edinburg, TX 78540. 4 Palynology Laboratory, Department of Anthropology, Texas A&M University, College Station, TX 77843Ð 4352.

1984, Pearson and Braiden 1990, Coffey and Breen 1997, Wilms and Wiechers 1997, Nagamitsu and Inoue 1999). In most cases, pollen traps are used for extended periods of time. However, this may alter foraging behavior by decreasing pollen ßow into the colony, causing the colony to collect larger quantities of pollen and potentially use different pollen sources. Other studies have recorded ßoral visitation by honey bees and other bees (de Menezes Pedro and de Camargo 1991). However, recording ßoral visitations does not provide an accurate measure of the cumulative foraging effort of a colony. No studies have documented pollen use by feral colonies under unmanipulated conditions in natural hive sites. Therefore, we documented pollen use by feral colonies in tree cavities in a coastal prairie landscape, with the goal of examining the foraging ecology of feral colonies under natural conditions. SpeciÞc objectives included evaluating overlap in pollen use between colonies throughout the year, examining the inßuence of the spatial locations of the colonies on overlap in pollen use, and describing general pollen collection patterns. Materials and Methods The study site was located on the Welder Wildlife Refuge, which is 3,157 ha of coastal prairie located in San Patricio County, TX. Feral colonies were found in

0046-225X/04/0727Ð0739$04.00/0 䉷 2004 Entomological Society of America

728

ENVIRONMENTAL ENTOMOLOGY

Fig. 1. Cavity and pollen trap locations. Cavities used by feral honey bee colonies are indicated by closed circles, cavities with pollen traps are indicated with stars, and pollen traps on Langstroth and Kenya top-bar hives are indicated with bulls-eyes.

the western one-quarter of the refuge in live oak mottes (clusters of woody vegetation formed around a live oak tree) and riparian woodland. Over a 12-yr period, 109 cavities were identiÞed that contained a feral colony during at least one point in time (Fig. 1). During the period of this study, 56 Ð 64 of those cavities were occupied by honey bees. Africanized honey bees, hybrids between European and African (A. m. scutellata) honey bees were Þrst recorded in the study area in 1993 and comprised a majority of the colonies during this study (Pinto et al. 2004). Dead bee traps modiÞed to hold removable pollen screens were placed on six feral colonies located in tree cavities. Ratcheting straps were used to attach the traps and polyurethane foam to seal off potential entrances. Traps also were placed on three feral-origin colonies located in Langstroth hives, and one feralorigin colony located in a Kenya top-bar hive (Berube 1989), none of which received any management. Colonies were selected based on their suitability for the use of a trap (cavity height, entrance size, and entrance shape) and their distance from other colonies with pollen collection traps (Fig. 1). The colonies located in the Langstroth hives and the Kenya top-bar hive were located in areas where occupied cavities were not available during the time period of this study or were not suitable for a pollen trap.

Vol. 33, no. 3

Pollen screens consisted of three layers of wire mesh with 0.45-cm openings between wires. Each layer was offset and separated from the other layers by ⬇0.45 cm. Honey bees passed through the screens to enter the hive, knocking off their corbicular pollen loads. Pollen fell through the metal slats at the bottom of the trap onto a removable tray, which prevented the bees from collecting the pollen pellets. Pollen was collected from each colony during an ⬃3-h sampling period once every 3 wk from July 2000 to July 2001. However, the length of the sampling period was extended under cool or adverse weather conditions to obtain a sample representative of a colonyÕs collective foraging effort. Pollen samples were processed using standard acetolysis procedures (Kapp 1969, Moore et al. 1991) and identiÞed by the Palynology Laboratory at Texas A&M University. Slides were made for each sample by thoroughly mixing the sample, placing several drops on a slide, spreading the drops over an area the size of the cover slip to assure adequate dispersion, and adding a cover slip. Previous studies of pollen extracted from honey have found that this procedure produces a subsample representative of the larger sample (Jones and Bryant 1998). Pollen types were identiÞed to the family or genus level using light microscopy. The level of identiÞcation depended on diagnostic characteristics of the pollen types in a given group of plants. A minimum of 200 pollen grains distributed among at least three transects was counted per sample to place the pollen types into frequency classes (Louveaux et al. 1978). We compared the percent overlap of pollen types between each pair of colonies using 100(1 ⫺ 0.5兺i Px,i ⫺ Py,i ), where Px,i and Py,i are the frequencies for colonies x and y of pollen type i (Schoener 1970). A Wilcoxon signed-rank test was used to compare overlap between sampling periods (Sokal and Rohlf 1995, SAS Institute 1998). Statistical signiÞcance was assessed by adjusting a signiÞcance level of ␣ ⫽ 0.05 for multiple comparisons using the Bonferroni technique. A Spearman rank correlation coefÞcient was used to test for correlation between overlap and distance between colonies throughout the year (Sokal and Rohlf 1995, SAS Institute 1998). To further evaluate potential spatial patterns in pollen use, the vegetation communities with predominant (⬎45% of the pollen from a single plant taxon in a sample from any colony during any sampling period) and cumulatively important pollen types (⬎8% of the combined foraging effort of all sampled colonies for a given sampling period) were identiÞed. The abundance of each plant in each vegetation community was obtained from the Welder plant list (unpublished data, Welder Wildlife Foundation). The Welder plant list contains commonly encountered plants on the Welder Wildlife Refuge based on cover data from point frame transects surveyed from 1975 to 1984. A landscape classiÞcation of the study area based on vegetation communities was used to examine spatially the distribution of collected pollen types during each sampling period (Fig. 2) (Baum 2003).

June 2004

BAUM ET AL.: POLLEN SELECTION BY FERAL HONEY BEE COLONIES

729

Fig. 2. A landscape classiÞcation showing the spatial distribution of vegetation communities in the study area.

In some cases, the frequency of a pollen type is not representative of its overall nutritional value, so the volume and protein content of a pollen type should be considered (da Silveira 1991, OÕRourke and Buchmann 1991). We made rough estimates of the nutritional contribution of each pollen type by placing each into general size categories and using protein content values reported in Roulston et al. (2000). General size (volume) categories included 0 Ð2, ⬎2Ð14, ⬎14 Ð50, ⬎50 Ð100, and ⬎100 ␮m3. Protein content values were generalized to the genus or family level to correspond to the level of identiÞcation, because Roulston et al. (2000) found that protein con-

tent was highly conserved within plant genera, families, and divisions. We used these estimates of nutritional contribution to identify pollen types overestimated or underestimated by only examining frequencies. Finally, we evaluated pollen collection patterns by examining the contribution of entomophilous versus anemophilous pollen types, herbaceous versus woody pollen sources, and nectariferous versus non-nectariferous pollen sources. These data provide generalized information about pollen collection patterns that allow for comparisons with other areas with different plant communities.

730

ENVIRONMENTAL ENTOMOLOGY Results

A total of 95 pollen types was collected by the feral colonies throughout the year, including 43 families, 66 genera, and 29 unknown taxa (Table 1; Fig. 3). The genera category included pollen types identiÞed to the genus level as well as those implied to be different genera at the family level. For example, pollen types identiÞed as Salvia (family Lamiaceae) and Lamiaceae were different pollen types, because those identiÞed as Lamiaceae were not Salvia yet belonged to a different, although unidentiÞed, genus. Unknown taxa typically comprised a very small portion of a sample, often only a few pollen grains. Only two unknown taxa comprised ⬎25% of a sample, and they occurred during the February and November sampling periods. The samples included 22 predominant (⬎45% of the pollen from a single plant taxon in a sample from any colony during any sampling period) and cumulatively important pollen types (⬎8% of the combined foraging effort of all sampled colonies for a given sampling period; Table 2). Four pollen collection periods were identiÞed: March through mid-May, late May through early August, mid-August through mid-December, and February, based on the pollen types collected throughout the year (Table 2). The spring collection period was dominated by Anacardiaceae Rhus type I and Anacardiaceae Rhus type II. Additional pollen types included other taxa in the Anacardiaceae, high spine Asteraceae (pollen with spines ⬎2.5 ␮m), low spine Asteraceae (pollen with spines ⬍2.5 ␮m), Lamiaceae, Lythraceae Lythrum, Papaveraceae Argemone, and Saliaceae Salix. The summer collection period consisted mainly of Fabaceae Prosopis, with additional input from Apiaceae, Arecaceae, Fabaceae Mimosa, and Poaceae. The fall collection period contained low spine Asteraceae, Euphorbiaceae Croton, and Ulmaceae Celtis, with some high spine Asteraceae, Cyrillaceae Cyrilla, Euphorbiaceae Cnidoscolus, Fabaceae Prosopis, Rhamnaceae, and Solanaceae. The winter collection period consisted of a dearth of pollen availability and little foraging activity from late December through January, with Lamiaceae, an unknown pollen type, Lamiaceae Salvia, and Anacardiaceae Rhus type I dominating in mid-February (Table 2). SigniÞcant differences existed in the amount of overlap between colonies among many of the sampling periods, but no clear patterns emerged. The amount of overlap and distance between colonies were signiÞcantly inversely correlated for some of the sampling periods (11 March, 21 April, 2 August, 18 August, 9 September, and 21 October), but not for others (18 February, 1 April, 13 May, 31 May, 22 June, 14 July, 30 September, 11 November, and 11 December; Table 3). The pollen types collected during a sampling period were typically located in multiple vegetation communities that were distributed throughout the study area (Table 4; Fig. 2). No distinguishable patterns emerged in relation to speciÞc spatial locations or foraging distances.

Table 1.

Vol. 33, no. 3 Identified pollen types by family and genus Taxa

Acanthaceae Ruellia Aceraceae Acer Alismataceae Alisma, Echinodorus, Sagittaria Amaryllidaceae Nothoscordum Anacardiaceae (family level), Rhus type I, Rhus type II Apiaceae Arecaceae Asteraceae Ambrosia, Artemesia, low spine,a high spine,b Ligulißoraec Berberidaceae Berberis Brassicaceae Chenopodiaceae Cheno-amd Cyperaceae Carex Cyrillaceae Cyrilla Euphorbiaceae Cnidoscolus, Croton Fabaceae (family level), Acacia, Dalea, Leucaena, Mimosa, Mimosa strigillosa, Prosopis, Trifolium Fagaceae Quercus Fumariaceae Corydalis Hydrophyllaceae Phacelia Juglandaceae Carya Lamiaceae (family level), Salvia Liliaceae (family level), Schoenocaulon drummondii Lythraceae Lythrum Magnoliaceae Liriodendron Malvaceae Nyctaginaceae Nymphaceae Nelumbo Oleaceae Fraxinus Onagraceae Oenothera Papaveraceae Argemone Plantaginaceae Plantago Poaceae (family level), Zea mays Polygonaceae Polygonum Portulaceae Portulaca Ranunculaceae Rhamnaceae Rosaceae (family level), Prunus Saliaceae Salix Scrophulariaceae (family level), Pedicularis Solanaceae Tiliaceae Tilia Ulmaceae Celtis, Ulmus Verbenaceae Phyla, Verbena Vitaceae Vitis Pollen types identiÞed at the family level that also have genus level identiÞcations are indicated as such. a Asteraceae (low spine) represents the ragweed group of anemophilous species. b Asteraceae (high spine) represents the sunßower group of entomophilous species. c Asteraceae Ligulißorae represents the dandelion group of entomophilous species. d Cheno-am represents the Chenopodiaceae and Amaranthus in the Amaranthaceae.

Protein content values were estimated for 17 of the 22 predominant and cumulatively important pollen types, with a mean and SE of 32.17 ⫾ 2.07% protein (Table 5). Twelve of these pollen types were overestimated, and nine were underestimated when volume was not considered. Furthermore, higher than average protein content values increased two underestimations and decreased four overestimations when con-

June 2004

BAUM ET AL.: POLLEN SELECTION BY FERAL HONEY BEE COLONIES

sidering the overall nutritional content of the pollen types. Lower than average protein content values decreased Þve underestimations and increased six overestimations (Table 5). Entomophilous pollen types were important for most of the year, but anemophilous pollen types were important from late September to December (Fig. 4). The use of herbaceous and woody pollen sources ßuctuated throughout the year, with herbs and shrubs being important pollen sources at the beginning of the year, trees from late June through early August, and herbs again for the remainder of the year (Fig. 5). In general, pollen sources came from nectariferous plants known to be good sources of honey (Fig. 6).

Discussion The colonies collected a wide variety of pollen types throughout the year, comprising ⬇30% of the species on the Welder plant list (unpublished data, Welder Wildlife Foundation). Other research also suggests honey bees typically collect pollen from 20 to 30% of angiosperms within their foraging range, although few species are used intensively (Wills et al. 1990, Roubik 1991, Buchmann et al. 1992). Overall, the most important pollen sources included low spine Asteraceae, Prosopis, Rhus type I, Croton, high spine Asteraceae, Celtis, Rhus type II, Lamiaceae, Apiaceae, Lythrum, and an unknown pollen type. However, the timing of pollen collection patterns may be as important or more important than the amount any given pollen type contributes to the annual pollen harvest of a colony (OÕNeal and Waller 1984). Pollen types collected during periods of low pollen availability and/or early in the foraging season when brood rearing begins may be functionally more important than those collected at other times when pollen availability is high and brood rearing is well established. Characteristics of the pollen types varied in terms of length of collection periods, importance within a sampling period, and protein content (as estimated from Roulston et al. 2000). Some pollen types were collected throughout much of year. For example, low spine Asteraceae comprised 21.4% of all samples combined (all colonies and all sampling periods), more than twice as much as the next most common pollen type. Low spine Asteraceae also occurred in all but two of the sampling periods (18 February and 18 August) and was a predominant pollen type in November and December when pollen availability was relatively low (Table 2). However, low spine Asteraceae pollen also contains less protein than many of the other pollen types (Table 5). In contrast, Prosopis was collected in all but three sampling periods (18 February, 11 March, and 11 December), represented a predominant pollen type in Þve sampling periods (the most of any pollen type), and contained a relatively high protein content (Tables 2, 6). Other pollen types were collected for only brief periods of time. Lythrum was collected during a 2-mo period from mid-April to mid-June and was a predominant pollen

731

type during only one of the sampling periods (Table 2). Solanaceae, Argemone, and Mimosa contained the highest protein contents, followed by Salix, Rhamnaceae, and Prosopis (Table 5). High spine Asteraceae contained the lowest protein content, followed by Lamiaceae, Salvia, low spine Asteraceae, and Poaceae. Pollen types with low and high protein contents may be more overestimated or underestimated when overall nutritional contribution is considered. Although evaluating the nutritional content of different pollen types is beyond the scope of this study, considering general characteristics of pollen volume and protein content may provide additional insight into the observed patterns of pollen collection. The normal procedure for estimating pollen volume involves measuring numerous pollen grains of each species using reference slides made from pollen collected from vouchered plant specimens (OÕRourke and Buchmann 1991). However, this was not possible because pollen types were only identiÞed to the genus or family level, with a few exceptions. Therefore, our estimates of pollen volume are based on placing each pollen type into a general size category. The value of several of the pollen types was misrepresented based on abundance alone. Most notably, the values for Cnidoscolus and Croton were underestimated and the values for Rhamnaceae and Salix were overestimated when volume was considered (Table 5). Protein content values were not available for Cnidoscolus and Croton, but Rhamnaceae and Salix had higher than average protein contents (Table 5), which probably would increase their nutritional value to a honey bee colony. The protein content and volume of a pollen type may change its overall nutritional contribution to the diet of a feral honey bee colony compared with the frequency of the pollen type (da Silveira 1991, OÕRourke and Buchmann 1991). For example, the contribution of a very abundant, but also very small pollen grain may be overestimated by evaluating only the percentage that pollen type comprises of a sample. At the other extreme, the value of an uncommon, but extremely large pollen grain may be underestimated by only examining frequencies. In general, four main foraging periods were identiÞed from the pollen collection patterns of feral colonies on the Welder Wildlife Refuge. The spring, summer, and winter periods consisted mainly of entomophilous pollen types, but anemophilous pollen types were important from late September to December (Fig. 4). Other studies have also identiÞed brief periods when anemophilous pollen types were important to honey bee colonies (reviewed in OÕNeal and Waller 1984, Pearson and Braiden 1990). Depending on geographic location and local plant communities, anemophilous pollen can represent 0.5Ð71% of the annual pollen harvest of a colony (Percival 1947, OÕNeal and Waller 1984). This periodic focus on anemophilous pollen types is interesting because these plants do not possess characteristics that attract insects for the dispersal of pol-

732

ENVIRONMENTAL ENTOMOLOGY

Vol. 33, no. 3

Fig. 3. Light micrographs of pollen types collected by honey bees. The distance between any two adjacent numbers on the scale in the micrograph is 25 ␮m.

len but instead rely on the wind to provide pollination services. In terms of honey bee colonies, the collection of anemophilous pollens requires nectar or honey to

provide energy for foraging and to pack pollen grains into corbicular pellets (OÕNeal and Waller 1984). However, protein content does not differ between

June 2004

BAUM ET AL.: POLLEN SELECTION BY FERAL HONEY BEE COLONIES

733

Fig. 3. Continued

zoophilous and anemophilous pollen types when phylogeny is considered (Roulston et al. 2000), although the authors did not evaluate honey bee development based on these different pollen types. Furthermore, Roulston et al. (2000) found that bees do not seem to select pollen types based on protein content. Therefore, honey bees may collect a variety of pollen types to meet their pollen needs, increasing the probability of obtaining nutritionally suitable pollens (Gary et al. 1972). Collecting numerous pollen types also should decrease the chance of poisoning by the consumption of a toxic pollen type or nutritional imbalances from

consumption of only one pollen type (OÕNeal and Waller 1984). The use of herbs, shrubs, and trees as pollen sources ßuctuated throughout the year. Herbs and shrubs were important during the spring and early summer, trees were important during the mid- to late summer, and herbs were important during the fall, roughly corresponding to the identiÞed foraging periods (Fig. 5). Different patterns are observed in different locations (Severson and Parry 1981, Coffey and Breen 1997). For example, trees provided the initial ßow of pollen in the spring, but herbs provided the majority

Pollen type

P 54, P 8, P 9

15, P

73, P 8, P 62, P P 57, P

11

20, P

22, P

9

37, P 9, P

33, P

11, P

11

37, P

15, P

40, P 9, P

60, P 9, P 9, P

38, P 34, P

8 12

P

8

18 Feb

11 Mar

1 April

21 April

13 May

31 May

34, P

22 June

21, P

14 July

76, P

2 Aug

8 33, P

18 Aug

14, P

9 Sept

82, P

30 Sept

30, P

21 Oct

15 58, P P

11 Nov

11 74, P

11 Dec

ENVIRONMENTAL ENTOMOLOGY

Anacardiaceae Anacardiaceae Rhus type I Anacardiaceae Rhus type II Apiaceae Arecaceae Asteraceae (high spine) Asteraceae (low spine) Cyrillaceae Cyrilla Euphorbiaceae Cnidoscolus Euphorbiaceae Croton Fabaceae Mimosa Fabaceae Prosopis Lamiaceae Lamiaceae Salvia Lythraceae Lythrum Papaveraceae Argemone Poaceae Rhamnaceae Saliaceae Salix Solanaceae Ulmaceae Celtis Unknown

Table 2. Predominant pollen types (>45% of the pollen from a single plant taxon in a sample from any colony during any sampling period) are indicated by “P” and numbers represent the percentage of a cumulatively important pollen type (>8% of the combined foraging effort of all sampled colonies for a given sampling period) for a given time period

734

Vol. 33, no. 3

of pollen for the remainder of the foraging season in Wisconsin (Severson and Parry 1981). Pollen sources were also good honey plants for most of the year, although the summer collection period contained the least correspondence between nectar and pollen sources (Fig. 6). The foraging range of a colony varies depending on resource availability and colony status. Foraging distances reported in the literature range from a few meters to over 10,000 m (Gary et al. 1972, Visscher and Seeley 1982, Schneider 1989, Schneider and McNally 1993, Waddington et al. 1994, Schneider and Hall 1997, Beekman and Ratnieks 2000). Visscher and Seeley (1982) recorded a mean foraging distance of 2,260 m for a colony located in a temperate forest in New York, with distances ranging from 50 to 10,100 m. Gary et al. (1972) reported mean foraging distances of 557 and 1,663 m in an agricultural setting, with distances ranging from 41 to 6,117 m. Therefore, all the colonies within the 2,500 by 2,500-m area used in this study were within the potential foraging ranges of all the other colonies, and all the ßoral patches within the study area were available to all the colonies. However, no clear patterns emerged from the overlap in pollen use between sampling periods. The colonies did not use the same resources in the same percentages, or exploit resources in a consistent manner throughout the year. This emphasizes both the complexity of honey bee foraging behavior and the ßowering phenology in the study area. Honey bee colonies select pollen to collect from a variety of potential sources based on incomplete knowledge of the resources available within their foraging range. The extent of a colonyÕs knowledge probably varies with resource availability, with the colonies obtaining more complete knowledge about resources within their foraging range when resource availability is low (Waddington et al. 1994). The Welder Wildlife Refuge provided a wide variety of pollen sources that were typically abundant within the study area for most of the year. Thus, the colonies would be expected to have less complete knowledge of the available resources and potentially use resources in different ways based on their incomplete knowledge. Other studies have documented colonies concentrating their foraging effort on a small number of large patches (Visscher and Seeley 1982, Schneider 1989) or a relatively large number of small, rich patches (Waddington et al. 1994), depending on the distribution and availability of ßoral resources in the area. Although no information was available on the number of patches visited in this study, the number of collected pollen types was known (Table 3). The mean number of pollen types collected showed that the colonies probably focused on the largest number of patches in late March and early April, and the fewest patches from mid-June to mid-July. Furthermore, the number of colonies without predominant pollen types indicated the colonies concentrated their foraging effort on several different sources in late March and early April, late May and early June, and late July and early August.

June 2004 Table 3.

18 Feb 11 Mar 1 April 21 April 13 May 31 May 22 June 14 July 2 Aug 18 Aug 9 Sept 30 Sept 21 Oct 11 Nov 11 Dec

BAUM ET AL.: POLLEN SELECTION BY FERAL HONEY BEE COLONIES

735

Spearman rank test results, means, and SEs for the correlation between pollen overlap and distance between colonies Overlap (mean ⫾ SE)

␳

P value

42.53 ⫾ 3.37 43.05 ⫾ 4.02 48.29 ⫾ 2.25 34.74 ⫾ 2.87 50.21 ⫾ 3.08 29.27 ⫾ 2.70 50.86 ⫾ 3.65 55.93 ⫾ 3.17 69.40 ⫾ 2.05 31.87 ⫾ 2.36 65.12 ⫾ 3.20 75.85 ⫾ 3.13 56.23 ⫾ 3.48 53.42 ⫾ 4.62 75.31 ⫾ 1.87

⫺0.038 ⫺0.455 0.241 ⫺0.353 ⫺0.271 0.091 ⫺0.03 ⫺0.405 ⫺0.551 ⫺0.327 ⫺0.465 0.277 ⫺0.549 0.096 0.027

0.803 0.003 0.153 0.037 0.110 0.591 0.895 0.145 ⬍0.001 0.03 0.002 0.066 ⬍0.001 0.522 0.855

Pollen types Mean number

Total number

Number predominant

Number of colonies without a predominant

6.4 6.4 12 7 8.1 6.6 3.1 2.3 5.6 8.2 7.1 6.3 6.8 6.8 5

17 21 24 19 27 20 10 8 16 23 25 16 14 22 15

3 3 1 5 2 2 3 2 1 2 2 2 2 2 2

3 2 6 2 0 6 0 0 6 0 0 0 1 2 0

The mean number of pollen types (averaged across all colonies), total number of pollen types (summed across all colonies), number of predominant pollen types (⬎45% of the pollen from a single plant taxon in a sample from any colony during any sampling period), and number of colonies without a predominant pollen type are provided for each sampling period.

Interestingly, the other sampling periods from midMay through late September were marked with all colonies collecting at least one predominant pollen type. The signiÞcant inverse correlation between overlap in collected pollen types and distance between colonies for some sampling periods suggested localized patterns of pollen availability were important at certain times of the year (Table 3). At other times, larger scale patterns (at a broader spatial extent) appeared to drive pollen collection, and no correlation was found between overlap and distance. However, these patterns could be created by a number of different conditions. For example, no correlation may exist when resources are scarce and the colonies must use

the same resources, regardless of distance from the hive. At the other extreme, there may be no correlation between overlap and distance when important pollen sources are abundant and widely distributed. Another interesting observation is that colonies located only a few meters apart often collect very different pollen types (Waddington et al. 1994, K.A.B., W.L.R., and R.N.C., unpublished data). Again, this can be attributed to the incomplete sampling effort of the colony and the way in which a colony chooses among available resources. In conclusion, feral honey bee colonies collected a wide variety of pollen types in a coastal prairie landscape. Four main foraging periods were identiÞed, with anemophilous pollen types being important in

Table 4. The abundance of predominant and cumulatively important pollen types in different vegetation communities within the study area based on data from point frame transects provided in the Welder plant list (unpublished data, Welder Wildlife Foundation) Pollen type Anacardiaceae Anacardiaceae Rhus type I Anacardiaceae Rhus type II Apiaceae Arecaceae Asteraceae (high spine) Asteraceae (low spine) Cyrillaceae Cyrilla Euphorbiaceae Cnidoscolus Euphorbiaceae Croton Fabaceae Mimosa Fabaceae Prosopis Lamiaceae Lamiaceae Salvia Lythraceae Lythrum Papaveraceae Argemone Poaceae Rhamnaceae Saliaceae Salix Solanaceae Ulmaceae Celtis

Brushland

Live oak

Woodland O F

O

O O O O

A F

A O

O F A O O

O A A O O

A O

O A F

O F

F O

Aquatic

O

R A R O F O

A F O O F

Disturbed

A O F

A

O A

O

R

Abundance values correspond to cover data, with abundant species (A) ⱖ 5% cover, frequent species (F) ⱖ 1Ð5% cover, occasional species (O) ⫽ 0 Ð1% cover, and rare species (R) ⫽ not encountered during sampling, but seen along sampling transect.

736

ENVIRONMENTAL ENTOMOLOGY

Vol. 33, no. 3

Table 5. Predominant (>45% of the pollen from a single plant taxon in a sample from any colony during any sampling period) and cumulatively important (>8% of the combined foraging effort of all sampled colonies for a given sampling period) pollen types with overestimated and underestimated contributions based on volume Pollen type

Proportion by volume

Proportion by abundance

Proportional difference

Result

Protein contenta

Difference from meanb

Anacardiaceae Anacardiaceae Rhus type I Anacardiaceae Rhus type II Apiaceae Arecaceae Asteraceae (high spine) Asteraceae (low spine) Cyrillaceae Cyrilla Euphorbiaceae Cnidoscolus Euphorbiaceae Croton Fabaceae Mimosa Fabaceae Prosopis Lamiaceae Lamiaceae Salvia Lythraceae Lythrum Papaveraceae Argemone Poaceae Rhamnaceae Saliaceae Salix Solanaceae Ulmaceae Celtis

3.00 2.38 3.62 1.26 0.90 2.04 7.95 0.20 2.59 36.26 0.04 11.83 6.41 1.44 1.22 1.26 0.75 0.04 0.05 0.40 3.17

4.13 5.58 2.68 2.96 0.67 4.79 18.66 0.47 0.51 7.19 0.03 27.77 4.75 1.07 2.87 0.93 0.55 0.62 0.74 0.94 7.45

0.73 0.43 1.35 0.43 1.35 0.43 0.43 0.43 5.04 5.04 1.35 0.43 1.35 1.35 0.43 1.35 1.35 0.07 0.07 0.43 0.43

Overestimated Overestimated Underestimated Overestimated Underestimated Overestimated Overestimated Overestimated Underestimated Underestimated Underestimated Overestimated Underestimated Underestimated Overestimated Underestimated Underestimated Overestimated Overestimated Overestimated Overestimated

28.93 28.93 28.93 29.00 31.40 20.85 24.74 NA NA NA 43.34 39.00 22.80 22.80 NA 45.30 25.26 40.40 40.84 46.71 27.65

⫺3.24 ⫺3.24 ⫺3.24 ⫺3.17 ⫺0.77 ⫺11.32 ⫺7.43 NA NA NA 11.18 6.83 ⫺9.37 ⫺9.37 NA 13.13 ⫺6.91 8.23 8.67 14.54 ⫺4.52

The proportion by volume, proportion by abundance, and proportional difference are based on the contribution of a pollen type to the overall sample (summed across all sampling periods and all colonies). The result (overestimated or underestimated), protein content, and difference from the mean protein content are provided. NA, not available. a Protein content values estimated from Roulston et al. (2000). b Difference from mean protein content of 32.17.

the fall. Herbaceous plants and shrubs provided pollen during the spring and early summer, trees in mid- to late summer, and herbaceous plants in the fall. The pollen sources used by the feral colonies also tended

to be good nectar sources. Overlap in pollen use between colonies varied throughout the year and pollen overlap was correlated with distance for some sampling periods but not others, probably because of the

Fig. 4. Percent of entomophilous, anemophilous, both (entomophilous and anemophilous), and unidentiÞed pollen types collected each sampling period.

June 2004

BAUM ET AL.: POLLEN SELECTION BY FERAL HONEY BEE COLONIES

737

Fig. 5. Percent of pollen types from herbs, trees, shrubs, combination (pollen types that include both herbaceous and woody plants at the level of identiÞcation), and unidentiÞed pollen sources collected each sampling period.

way colonies select resources and the ßowering phenology in the study area. Acknowledgments We thank L. Lavold and D. Marshall for assistance with pollen processing and identiÞcation. L. Drawe, T. Blankenship, S. Glasscock, A. Kresta, A. Bunting, and M. Tchakerian

assisted with various aspects of this project. Funding was provided by the Welder Wildlife Foundation, the USDAARS, BeneÞcial Insects Research Unit, Honey Bee Laboratory (Weslaco, TX), and the Texas Legislative Initiative: Protection and Management of Honey BeesÐPollinators of Agricultural Crops, Orchards, and Natural Landscapes. K. A. Baum was supported by a Welder Wildlife Foundation fellowship and the Texas Legislative Initiative: Protection and

Fig. 6. Percent of pollen types from nectariferous, non-nectariferous, and unidentiÞed pollen sources collected each sampling period.

738

ENVIRONMENTAL ENTOMOLOGY

Management of Honey BeesÐPollinators of Agricultural Crops, Orchards, and Natural Landscapes. This is contribution 612 of the Welder Wildlife Foundation.

References Cited Adams, R. J., G. C. Manville, and J. H. McAndrews. 1978. Comparison of pollen collected by a honey bee colony with a modern wind-dispersed pollen assemblage. Can. Field Nat. 92: 359 Ð368. Allen-Wardell, G., P. Bernhardt, R. Bitner, A. Burquez, S. Buchmann, J. Cane, V. Dalton, P. Feinsinger, M. Ingram, D. Inouye, C. E. Jones, K. Kennedy, P. Kevan, H. Koopowitz, R. Medellin, S. MedellinMorales, G. P. Nabhan, B. Pavlik, V. Tepedino, P. Torchio, and S. Walker. 1998. The potential consequences of pollinator declines on the conservation of biodiversity and stability of food crop yields. Conserv. Biol. 12: 8 Ð17. Baum, K. A. 2003. Feral Africanized honey bee ecology in a coastal prairie landscape. Ph.D. dissertation. Texas A&M University, College Station, TX. Beekman, M., and F.L.W. Ratnieks. 2000. Long-range foraging by the honey-bee, Apis mellifera L. Funct. Ecol. 14: 490 Ð 496. Berube, C. 1989. The Kenya top-bar hive as a better hive in developing-countries. Am. Bee J. 129: 525Ð528. Buchmann, S. L., and G. P. Nabhan. 1996. The forgotten pollinators. Island Press, Washington, DC. Buchmann, S. L., M. K. O’Rourke, C. W. Shipman, S. C. Thoenes, and J. O. Schmidt. 1992. Pollen harvest by honey bees in Saguaro National Monument: potential effects on plant reproduction, pp. 149 Ð156. In C. P. Stone and E. S. Bellatoni (eds.), Proceedings of the symposium on research in Saguaro National Monument: 23Ð24 January 1991. U. S. National Park Service Rincon Institute and the Southwest Parks and Monuments Association, Tucson, AZ. Coffey, M. F., and J. Breen. 1997. Seasonal variation in pollen and nectar sources of honey bees in Ireland. J. Apic. Res. 36: 63Ð76. da Silveira, F. A. 1991. Inßuence of pollen grain volume on the estimation of the relative importance of its source to bees. Apidologie. 22: 495Ð501. de Menezes Pedro, S. R., and J.M.F. de Camargo. 1991. Interactions on ßoral resources between the Africanized honey bee Apis mellifera L and the native bee community (Hymenoptera: Apoidea) in a natural “cerrado” ecosystem in southeast Brazil. Apidologie. 22: 397Ð 415. Gary, N. E., P. C. Witherell, and J. Marston. 1972. Foraging range and distribution of honey bees used for carrot and onion pollination. Environ. Entomol. 1: 71Ð78. Haydak, M. H. 1970. Honey bee nutrition. Annu. Rev. Entomol. 15: 143Ð156. Ingram, M., G. P. Nabhan, and S. L. Buchmann. 1996. Ten essential reasons to protect the birds and the bees: how an impending pollination crisis threatens plants and the food on your table. The Forgotten Pollinators Campaign, Arizona-Sonora Desert Museum, Tucson, AZ. Jones, G. D., and V. M. Bryant, Jr. 1998. Are all counts created equal, pp. 115Ð120. In V. M. Bryant, Jr., and J. H. Wrenn (eds.) New developments in palynomorph sampling, extraction, and analysis. American Association of Stratigraphic Palynologists Foundation, Dallas, TX. Kapp, R. O. 1969. How to know pollen and spores. Wm. C. Brown, Dubuque, IA. Louveaux, J., A. Maurizio, and G. Vorwohl. 1978. Methods of melissopalynology. Bee World. 59: 139 Ð157.

Vol. 33, no. 3

Maurizio, A. 1950. The inßuence of pollen feeding and brood rearing on the length of life and physiological condition of the honeybee: preliminary report. Bee World. 31: 9 Ð12. Moore, P. D., J. A. Webb, and M. E. Collinson. 1991. Pollen analysis, 2nd ed. Blackwell ScientiÞc Publications, Oxford, UK. Morse, R. A., and N. W. Calderone. 2000. The value of honey bees as pollinators of U. S. crops in 2000. Bee Culture 128. Nagamitsu, T., and T. Inoue. 1999. Differences in pollen sources of Apis cerana and Apis mellifera at a primary beech forest in central Japan. J. Apic. Res. 38: 71Ð78. O’Neal, R. J., and Waller, G. D. 1984. On the pollen harvest by the honey bee (Apis mellifera L.) near Tucson, Arizona (1976 Ð1981). Desert Plants. 6: 81Ð109. O’Rourke, M. K., and S. L. Buchmann. 1991. Standardized analytical techniques for bee-collected pollen. Environ. Entomol. 20: 507Ð513. Pearson, W. D., and V. Braiden. 1990. Seasonal pollen collection by honeybees from grass/shrub highlands in Canterbury, New Zealand. J. Apic. Res. 29: 206 Ð213. Percival, M. S. 1947. Pollen collection by Apis mellifera. New Phytol. 54: 353Ð368. Pinto, M. A., W. L. Rubink, R. N. Coulson, J. C. Patton, and J. S. Johnston. 2004. Temporal pattern of Africanization in a feral honey bee population from Texas inferred from mitochondrial DNA. Evolution. (In press). Poulsen, M. H. 1973. Diurnal and inter-colony variation in pollen collection from different seed crops by honeybees. Entomol. Scand. 4: 100 Ð104. Roubik, D. W. 1991. Aspects of Africanized honey bee ecology in tropical America, pp. 259 Ð281. In M. Spivak, D.J.C. Fletcher, and M. D. Breed (eds.), The “African” honey bee. Westview Press, Boulder, CO. Roulston, T. H., and J. H. Cane. 2000. Pollen nutritional content and digestibility for animals. Plant Syst. Evol. 222: 187Ð209. Roulston, T. H., J. H. Cane, and S. L. Buchmann. 2000. What governs protein content of pollen: pollinator preferences, pollen-pistil interactions, or phylogeny? Ecol. Monogr. 70: 617Ð 643. SAS Institute 1998. StatView: StatView reference, 2nd ed. SAS Institute, Cary, NC. Schmalzel, R. J. 1980. The diet breadth of Apis (Hymenoptera: Apida). Thesis. University of Arizona, Tucson, AZ. Schneider, S. S. 1989. Spatial foraging patterns of the African honeybee, Apis mellifera scutellata. J. Insect Behav. 2: 505Ð521. Schneider, S. S., and L. C. McNally. 1993. Spatial foraging patterns and colony energy status in the African honey bee, Apis mellifera scutellata. J. Insect Behav. 6: 195Ð210. Schneider, S. S., and H. G. Hall. 1997. Diet selection and foraging distances of African and European-African hybrid honey bee colonies in Costa Rica. Insect. Soc. 44: 171Ð187. Schoener, T. W. 1970. Nonsynchronous spatial overlap of lizards in patchy habitats. Ecology. 51: 408 Ð 418. Severson, D. W., and J. E. Parry. 1981. A chronology of pollen collection by honeybees. J. Apic. Res. 20: 97Ð103. Sokal, R. R., and F. J. Rohlf. 1995. Biometry, 3rd ed. W. H. Freeman, New York. Synge, A. D. 1947. Pollen collection by honeybees (Apis mellifera). J. Anim. Ecol. 16: 122Ð138. Visscher, P. K., and T. D. Seeley. 1982. Foraging strategy of honeybee colonies in a temperate deciduous forest. Ecology. 63: 1790 Ð1801. Waddington, K. D., P. K. Visscher, T. J. Herbert, and M. Raveret Richter. 1994. Comparisons of forager dis-

June 2004

BAUM ET AL.: POLLEN SELECTION BY FERAL HONEY BEE COLONIES

tributions from matched honey bee colonies in suburban environments. Behav. Ecol. Sociobiol. 35: 423Ð 429. Wills, R. T., M. N. Lyons, and D. T. Bell. 1990. The European honey bee in western Australian kwongan: foraging preferences and some implications for management. Proc. Ecol. Soc. Aust. 16: 167Ð176. Wilms, W., and B. Wiechers. 1997. Floral resource parti-

739

tioning between native Melipona bees and the introduced Africanized honey bee in the Brazilian Atlantic rain forest. Apidologie. 28: 339 Ð355.

Received 7 November 2003; accepted 23 February 2004.