Bumblebee Response to Variation in Nectar Availability Author(s): John M. Pleasants Source: Ecology, Vol. 62, No. 6 (Dec., 1981), pp. 1648-1661 Published by: Ecological Society of America Stable URL: http://www.jstor.org/stable/1941519 Accessed: 21/08/2009 13:23 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=esa. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org. Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology. http://www.jstor.org Ecology, 62(6), 1981, pp. 1648-1661 © 1981 by the Ecological Society of America BUMBLEBEE RESPONSE TO VARIATION IN NECTAR AVAILABILITY1 JOHN M. PLEASANTS Department of Botany, Iowa State University, Ames, Iowa 50011 USA Abstract. I examined the response of bumblebees to two kinds of spatial variation and two kinds of temporal variation in nectar levels. The spatial variation involved differences in reward value among plant species and differences in nectar availability among patches of flowers of a single species. The temporal variation involved changes in nectar availability over a season and between years. In all cases the response of bees was measured by differences in bee density or abundance. The response of bees to different plant species was determined by the density of foraging bees on the flowers of each species at a given time (bee: flower ratio = B/F). I examined several groups of plant species whose members overlapped in their flowering periods and in their bumblebee visitors. Within a group, the relative magnitudes of the B/F's for species indicate the degree to which their flowers are preferred by bees. If bees are optimal foragers, preference should reflect the per-flower reward value of species. I characterized the potential reward value of species by their nectar production rates (NPR's). The bees' preference for species was found to be very similar to the relative magnitudes of their NPR's. I also considered whether including the time and energy costs of foraging on the different species would produce a better predictor of preference. While these factors did affect the absolute reward value of species they did not influence their reward value relative to one another. Foraging costs are most likely to have no effect on preference when species have similar floral morphologies. To assess the response of bees to differences in patch quality I bagged some of the inflorescences in an experimental patch. Following this reduction in the number of flowers, the number of bees in the patch decreased proportionately, producing a B/F very similar to that of a control patch. When the bagging was removed the number of bees increased such that the B/F was again similar to that of the control patch. In general these results provide further support for optimal foraging in bumblebees. I examined the response of bees to temporal variation in nectar levels to determine whether resource limitation for bumblebees changed over time. Seasonal changes in the abundances of three bumblebee species were compared with seasonal changes in the total nectar production by the plants each bee visited. Preference factors were used to convert floral production to nectar production. In general, seasonal production and utilization were closely matched, indicating no major change in nectar resource limitation. The cumulative numbers of bees seen during each of two summer seasons (1974 and 1975) were very similar. This indicates that the resource base, which had not changed between years, was capable of supporting only a limited number of bees. During 1975 a natural species removal experiment occurred which allowed an assessment of competition among bees. Apis mellifera, which had been well represented in 1974, was absent. The bumblebees of short and medium tongue length with which honeybees overlapped in resource use exhibited competitive release. Their abundances increased in 1975 resulting in near-perfect density compensation. Key wvords: bumblebees; competitive release; density compensation; nectar production rate; optimal foraging; pollinators; preference. INTRODUCTION A currently active area of inquiry in ecology is concerned with whether animals forage in a manner which optimizes their energy intake rate (see reviews by Schoener 1971 and Pyke et al. 1977). Three important aspects of this question are which types of food items to eat (optimal diet), whether to stay in one patch or move to another (optimal patch choice), and how to move between resource points (optimal movement patterns). Some answers to these questions can be obtained by observing animals in choice situations, i.e., where there is some variation in the types of food items, quality of patches, or reward value of resource points available. One animal group which has received a good deal of attention with regard to optimal foraging is bumblebees. This is because there is less difficulty Manuscript received 20 June 1980; revised and accepted 12 December 1980. with bees in quantifying the reward value and spatial distribution of the resources they use (nectar) and in observing their foraging behavior. Most of the work done on bumblebees has been directed at optimal movement patterns (Pyke 1978b, c, d, Heinrich 1979a, Zimmerman 1979), while little work has been done on the first two aspects of the question. In this paper I explore these two aspects of optimal foraging by examining the response of bumblebees to two types of variation in nectar resources: differences among plant species in nectar rewards and differences in nectar availability between patches of flowers of a single species. If bumblebees forage in an efficient, optimal manner we would expect them to forage more heavily on species which provide a greater energetic reward. The primary factor influencing the reward value of a species is its nectar production rate (NPR). Therefore, species whose flowers have higher NPR's should be more attractive to bees and should be preferred by December 1981 BEE RESPONSE TO NECTAR AVAILABILITY them. This preference will be manifested in a higher number of foraging bees relative to the available number of flowers (bee: flower ratio = B/F). We would also expect optimally foraging bees to concentrate their foraging activity in patches of flowers which provide the greatest reward. Rather than looking at the behavior of individual bees, which has been the most common approach in these kinds of studies, I have examined the collective response of bees by measuring differences in bee density on flowers. A second major question, also involving the response of bees to variation in resource levels, is whether the degree of resource limitation for bees changes over time. The degree of resource limitation will determine the strength of selection pressures for competitive displacement and resource partitioning among bumblebee species. It will also affect selection for adaptations to increase foraging efficiency. Resource limitation was studied by examining the response of bees to variation in total plant nectar production over a season and variation in nectar production and availability between years. Response was measured by changes in bee abundance. The plant species involved in this study grow in mountain meadow habitats. The pollinator fauna of these communities is dominated by bumblebees (responsible for >70% of all observed insect pollinator visits; Pleasants 1977). Several studies have shown that in bumblebee communities there is a partitioning of plant resources (Heinrich 1976b, Inouye 1976, Pleasants 1977, 1980). Partitioning results from different bumblebee species foraging on different plant species. In general, coexisting bumblebee species fall into different tongue length classes (short, medium, long) and the average corolla depth of flowers they visit is similar to worker bee tongue length (Inouye 1976, Pleasants 1980). These displacement patterns are good inferential evidence that competition among bumblebees for nectar resources is important, implying that nectar is a limiting resource. This is further corroborated by the increasing wealth of evidence on bee behavioral adaptations and foraging strategies which serve to maximize energetic returns. Because of the association between a particular bumblebee and a set of plant species, each meadow plant community can be divided into several guilds of plant species (Pleasants 1977, 1980). The members of a particular guild all receive the largest proportion of their pollinator visits from the same bumblebee species. In general, guild members avoid competition for pollinators by having blooming periods which show a regular temporal segregation over the summer season (Pleasants 1977, 1980). METHODS In this study I used data on plants and bumblebee pollinators collected primarily from two study sites. The sites, numbered 1 and 2, are located on the west- 1649 ern slope of the Rocky Mountains in the Gunnison National Forest near Crested Butte, Colorado and are at elevations of 2590 and 2743 m, respectively. The sites consist of 0.6 and 2.4 ha of meadow surrounded by aspen trees. A third site, Site 3 (3040 m), is a meadow surrounded by Engelmann spruce. To examine the response of bumblebees to variation among plant species in nectar production, I compared the nectar production rates of four groups of species with the degree of preference shown towards them by bumblebees. Nectar sugar production rates were determined for noncomposite species by bagging inflorescences and marking flowers that were just about to open. I then removed and measured the volume of nectar which had accumulated during the life-span of the flower using a capillary tube (5, 10, or 25 /xL). The sugar concentration of the nectar was measured with a hand-held Bausch and Lomb temperature-compensated refractometer. The total sugar produced per flower was divided by flower longevity so that nectar production rate (NPR) was expressed on a 24-h basis. There was no indication for any of the species studied that NPR was influenced by the presence of accumulated nectar (J. Pleasants, personal observation). I examined 20-50 flowers for each species. Nectar sugar production by composites (family Asteraceae) was measured in a different manner because I was unable to extract the small amount of nectar from the florets using a capillary tube. Composite florets remained open for 2 d. During day I they were in the male (pollen-shedding) phase and on day 2 they were in the female condition (style branches separated). Flower heads were bagged for I d and the nectar accumulation in both male and female phase florets was sampled (the nectar in female phase florets represents only I d of production because virtually all of the nectar production from the previous day was removed [J. Pleasants, personal observation]). The amount of nectar sugar in florets was measured by flushing them with 3 ,uL of distilled water using a 10- tL syringe equipped with a Chaney adapter (Hamilton Corporation, Reno, Nevada). The resulting solution was removed with either the syringe or a 5-,uL capillary tube and placed in a refractometer to determine percent sugar. The average of male and female floret sugar production was used to get a daily rate. The preference for a species was measured by the number of bee visitors per flower (BIF). BIF is equivalent to a snapshot census of bees on flowers. It is the number of bees seen on flowers during a 1-2 min scan of a patch of flowers divided by the number of flowers in the patch. Each bee in the patch is counted only once. The differences in preference shown towards two or more species within a group were expressed on a relative scale. The most preferred species (highest BIF in the group) was given a preference rating of 1.00. The preference rating for each of the other species was related to this standard by dividing its BIF 1650 JOHN M. PLEASANTS by that of the most preferred species. Relative preference was calculated at several times during the flowering period of each species and then averaged over all times. The following pairs of species were examined. At Site 2 there are two legume species, Lathyrus leucanthus and Vicia americana, which have extensive temporal and spatial overlap and receive the majority of their visits from Bombus appositus (Pleasants 1980). In 1975 the number of bumblebees on each of these species was sampled at four different times during their blooming period. On the same day a portion of the meadow was sampled by transect and quadrat to estimate the relative number of open flowers for each species. While this did not allow me to calculate the actual BIF for each species it did give an indication of the relative preference for each species. At Site 2, an 8 m x 8 m patch of Agastache urticifjlia (Lamiaceae) and a 9 m x 7 m patch of Polemonium foliosissimun (Polemoniaceae) were monitored over a 15d period in 1978. These species are visited by both Bombus bijfirius and B. flaviJfrons. On five different days during this period the number of bees in each of these patches was recorded at midmorning, late morning, and early afternoon. The number of flowers open on each census day in each patch was obtained by extrapolation from a flowering curve based on several flower censuses made during this period. At Site 3, two 8 m x 8 m patches containing both Delphinium barbevi (Ranunculaceae) and Corydalis caseana (Fumariaceae) were monitored over a 15-d period in 1978. These species are visited by B. flavifrons and B. appositus. During this period five bee censuses were made. Each census was made sometime between 0930 and 1200. The number of open flowers of each species was counted in the patches after each census. A group of five composite species was studied at Site 2 in 1978. These species have broadly overlapping blooming periods and are visited primarily by B. biiarius with B. flavlifJronsas a minor visitor to two of them. In four 10 m x 10 m patches the number of bees on each species was monitored two to three times during the day at four times over a 15-d period. On each census day the number of flower heads with open florets was counted and the number of open florets per flower head (from a sample of 20-40 flower heads) was estimated. To estimate the costs of foraging in these species I timed bees foraging on flowers. Two stopwatches were used, one to record the total period of observation and the other to record handling time. For species with inflorescences that required bees to fly between adjacent flowers, handling time was equal to the time spent in flowers. For species with inflorescences that allowed bees to walk between adjacent flowers, handling time included in-flower time and walking time. Travel time (either between flowers or inflorescences) was equal to the difference between total time and handling time. Handling and travel time were expressed on a per-flower basis by dividing by the num- Ecology, Vol. 62, No. 6 ber of flowers visited. Bees were observed for a minimum of 20 visits per foraging bout. Data on 200-1000 flower visits were obtained for each plant species. I conducted resource reduction experiments in the summer of 1978 on two species, Agastache urticiiolia (at Site 2) and Delphinium barbeyi (at Site 3). For each species I delineated two patches consisting of numerous inflorescences, within a large stand. Patches were 10 m x 10 m and were 3-5 m apart. The number of open flowers in each patch was counted. The two patches were designed to contain approximately equal numbers of flowers. On the day of the experiment I censused each patch for the number of bees visiting flowers of the subject species. The number of flowers was reduced in one patch (experimental) at = 1100 by placing nylon netting (bridal veil) over a number of inflorescences. The number of flowers under the bagging was counted. After 0.5-1 h I recensused the experimental and control patches. The bagging remained on the inflorescences for -24 h. Prior to removing the bagging I censused each patch. I then removed the bagging and censused again. The bee densities reported here are an average of several censuses that were made at different times shortly before and after treatments. In determining the response of bumblebees to seasonal changes in nectar availability I focused on Bornbus bifarius at Site I and B. bifirius and B. flavifrons at Site 2. These bees were chosen because of the large amount of data on their activity and on the floral resource production of the species they visited. The data were collected during the months of June through August 1975. The change in abundance of each bee species over the season was compared with the change in nectar production by the species they visited. A measure of bumblebee abundance was obtained from 60-90 min censuses made every 5-6 d at each site. The census consisted of a slow walk by two observers 3-5 m apart along a fixed route through the meadow. Bees seen on flowers were identified (conspicuous differences in thoracic and abdominal markings made field identification of bumblebees possible) and the plant species was recorded as well as whether the bee was collecting pollen or nectar. To control for any diurnal changes in bumblebee activity, censuses were always conducted at the same time of day (between 1000 and 1200). The abundance results have been standardized to the average number of bees seen per 30 min of census time. To assess nectar availability I first determined the daily number of open flowers for each species by multiplying the proportion of its flowers open on a particular day (from flowering curves) by the total number of flowers it produced (per square metre) over its blooming period (see Pleasants 1980 for details). For species in the Asteraceae the number of flower heads with open florets was used instead of the number of open flowers. As discussed below, the preference of bumblebees for foraging on different December 1981 I ABLE BEE RESPONSETO NECTAR AVAILABILITY 1651 1. Comparison of relative reward value of plant species and degree to which they are preferred by bees. 1 / L 2 4 Average 5 6 (B/F) x 100 Relative reward value Relative attractiveness (preferencerating) no. per visitors per flower Sugar nectar concenflower-' tration Species 3 mg sugar per flower* (24 h)-' (%) 1.96 0.46 41 16 0.804 0.074 2.13t 0.16t 0.092 2.92 0.71 46 40 1.330 0.280 0.624 0.073 0.210 1.000 0.138 0.38 0.59 48 36 0.182 0.214 0.468 0.410 0.851 1.000 1.000 0.900 Group 1 Vicia americana Lathvrus leuc(anthus 1.000 1.000 0.075 Group2 Delphinium barbevi Corxvdalisca.seana 1.000 Group3 Agaastache urti(cifolia Poleimonium Joliosissimum per per per per per per florett head§ floret floret head floret head .301 .330 .106 .083 .024 15.77 9.04 2.27 .85 .27 .274 .328 .063 .085 .012 0.91 1.00 0.32 0.25 0.07 1.00 0.57 0.14 0.05 0.02 0.84 1.00 0.39 0.35 0.09 1.00 0.62 0.19 0.08 0.02 per Group4 Rudbeckia montana (R.m.) Helianthella quinquenervis (H.(q.) Helenium hoopesii (H.h.) Vig,iiera nmulltiflora(V.m.) Erigeron speciosus (E.s.) ' Column I x Column2. t Only an index. t For Rudbchkia montana and Helianthella quinquenervis this is an average of first-day (d phase) and second-day (D phase) NPR for florets. These are the only composites with nectar production on the second day and where second day florets are visited by bees. § Based on the following average number of florets per head: R. m. 52.4 (d + Y), H. q. 27.4 (d + Y), H. h. 21.4 (d), V. im. 10.2 (d), E. s. 11.1 (d). species seems to be a good indicator of their potential reward value (nectar production rate). To convert the number of open flowers for a species visited by a particular bumblebee into the amount of nectar it represented to that bee, the number of open flowers was multiplied by the species' preference rating. Preference ratings were calculated in the following manner: On each day that the pollinator census was made the number of bumblebees of a particular species seen on each plant species was recorded. This number was divided by the estimate of the number of open flowers of the species giving a BIF value. Standardized preference ratings were obtained from B/F's as described earlier. The preference rating for each species is an average of its rating for each census time. To estimate the total amount of nectar available to a particular bumblebee species on a given day I used the formula E summers of 1974 and 1975. As an estimate of the total number of bees seen during each summer I used the area under the curves of bee abundance plotted against time (see Figs. 6 and 7). Since I spent a shorter period of time at the study sites in 1975, I used the area under an equivalent portion of the 1974 curve (marked by arrows in Figs. 6 and 7) to make the 1975 bumblebee abundance estimates comparable to those for 1974. By knowing what species were in bloom and the stage of their blooming at the conclusion of my stay in 1975, I could determine the same point in time in 1974. This means that fewer calendar days were under consideration in 1975. This is because the flowering season in 1975, although delayed at the start, proceeded more rapidly than in 1974, with species' blooming periods being as much as 5-7 d ahead of 1974 after an equal number of calendar days. Pin;, where p, is the standardized preference shown shown by that bumblebee species for plant species i, ni is the number of flowers of species i open on that day, and s represents the entire set of plant species visited by that bumblebee. To determine the year-to-year response of bumblebees to differences in nectar availability I used data on bee abundances and floral production from the RESULTS Response AND DISCUSSION to differences among species I examined four groups of species to determine if the preference shown by bees towards the members of each group was related to their reward value. Each group was composed of species which had rather broad overlap in their time of flowering and species of 1652 JOHN M. PLEASANTS Ecology, Vol. 62, No. 6 TABLE2. Estimatesof foragingrewardand costs. 2 I Time per flower (s) Plantspecies Handling* Travelingt Group 2 Delphinium barbevi Corvdalis caseana Average interval between successive visits to a flower 4 5 6 Cost (Cost/ reward) Net energy intake rate 3 (s)t Average reward/ per flower (J)§ x100 (J/s)¶ .080 .193 31 33 .0632 .1113 (J) 1.86 1.01 1.72 444 4904 .254 .590 1.67 2.89 0.10 1.33 378 1032 .030 .017 .095 .059 56 57 .0075 .0096 .119 .0038 .0038 .0033 .0046 .0054 3.2 4.3 4.8 12.3 6.4 .0912 .0828 1.76 Group3 Agastache urticifolia Polemonium foliosissimum Group4 1.24 Rudbc'kia montana Helianthella quinquenervis Helenium hoopcsii Viguicra munltiflora Erigeron speciosus 0.96 0.86 0.79 0.81 .023 .044 .054 .118 .138 922 612 1451 1068 7900 .087 .066 .038 .082 .0690 .0368 .0803 * Th; for Agastache urticifoliaand all composites this includes nonflighttravel time between flowers on an inflorescence. t T,; based on average flighttime between flowers and between inflorescencesper flower (= total flighttime per inflorescence/averagenumberof flowers visited per inflorescence). t (FIB)(Th + T,). § SJ, = column 2 x NPR x J,; NPR (mg sugar/s)from Table 1, assumingthat 24-hrnectar accumulationwas produced over a 10-h daylight period; J, = 15.48 J/mg (IJ = .239 cal). ThJh + TtJ,; based on 0.034 J g- .s--' for handling and nonflight travel, 0.435 J-g- . s- for flight (Pyke 1978a); worker body mass B. appositus 0.28g, B. flavifrons 0.13g, B. bifarius0.06g (Pyke 1978a). The body mass used to calculate cost is a weightedaveragebased on frequencyof visits to a plant species by differentbumblebees.Thermoregulationcosts are not included. 11(Column3 - column 4)/(T,,+ T,). bumblebee visitors. The potential reward value of species was measured by their nectar production rates. Preference was measured by the number of bees per flower (B/F) on a species. Preference in this paper refers to how a population of foraging bees is dispersed over the available flowers of the various plant species at some point in time. It does not imply anything about the frequency of visitation by individual bees to various species during a foraging trip (cf. Heinrich 1979b). The results (Table 1) show a fairly close correspondence between the relative reward value of flowers (column 5) and the relative number of bees seen on them (column 6). Heinrich (1976b) presents similar evidence for five species in a Maine bog. I investigated whether the slight deviations between some of the values in columns 5 and 6 could be explained by considering other factors which might influence a species' reward value. The precise definition of reward value is the net energy intake rate a bee experiences while foraging on a species. The general expression for net energy intake rate is EIT = C T (1) where G = the energetic gain from a flower (energy from nectar sugar), C = the energetic cost involved in obtaining the nectar (energy expended in handling a flower and traveling to the next), and T = the time required to obtain the nectar (handling and traveling time per flower). NPR only affects G. Does a consideration of the time and energy costs of foraging change the reward value from that based on NPR alone'? A more detailed version of Eq. 1 can be written (simplified from Pyke 1978a) as follows: EIT= SJ, - (ThJh + T,J,) Th + Tt (2) where S = average standing crop of nectar sugar per flower in milligrams, J,, = joules per milligram nectar sugar, Th = average handling time per flower, T = average traveling time per flower (includes a withininflorescence and between-inflorescence component), Jh = cost of handling per unit time (joules per second), and Jt = cost of traveling per unit time (joules per second). S will be equal to the average time interval between successive visits to a flower times the rate of nectar production. It can be shown that the average intervisit interval is equal to the reciprocal of the number of bees foraging per flower (FIB) multiplied by the time per flower (Th + T,). Thus S = (F/B)(Th + T,)NPR. (3) When the energetic costs of foraging (ThJh + TtJt) are small relative to the energetic gain (SJ), the cost com- 1653 BEE RESPONSE TO NECTAR AVAILABILITY December 1981 5 *25 v- o 4 .,... 1O I 3 3 ., I| I | |I a _1 1 E C E unbagged C bagged bagged DAY 1 C E E C unbagged DAY 2 FIG. 1. Agastache urticifolia resource reduction experiment. E = experimental patch, C = control patch. On day I some inflorescences were bagged in the experimental patch to reduce nectar availability. On day 2 the bagging was removed. Bee numbers were censused in control and experimental patches before and after each treatment on each day. ponent can be ignored, making the net energy intake rate approximately equal to the gross energy intake rate. E/T ers can be visited does not affect reward value (contra Heinrich 1979b). For example, consider a species with a given number of bees foraging on it and with a given time per flower (Th + T,). What would happen to the reward value of this species (E/T) if flowers could be visited twice as rapidly [(Th + T,)/2]? It would mean that the average intervisit interval would be halved and therefore the average standing crop of nectar would be half of what it was before. Bees would spend (F/B)(T, + T,)NPR J,, (T,, + T,) SJ,, T, + T, NPR J,. =FIB (4) Note that the handling and traveling time terms cancel out. What this indicates is that the rate at which flow- 3 30 [ 25 a) c 0 -0 2 ° 20 0 -o 0 In a) 15 0 a) -0 _0 > 0 1 ) 10 aQ a) -0 5 C E E unbagged bagged DAY 1 FIG. 2. C C E E bagged C unbagged DAY 2 Delphinium barbeyi resource reduction experiment. Details as in Fig. 1. 1654 JOHN M. PLEASANTS Ecology, Vol. 62, No. 6 TABLE3. Preference ratings for plant species. All plant species in the family Asteraceae except: Polemronium (Polemoniaceae). Hvdrophvylum (Hydrophyllaceae), Agastache (Lamiaceae), Geranium (Geraniaceae), and Vicia (Fabaceae). Preference factors for composites measured per head. Site 2 Site 1 Visited by: Bombus biflarius* Preference Plant species factor Rudbeckia montana Helenium hoopesii Vi,guier mulzIltif(ioru Geranium Jremontii HydrophylvlumfJndleri Erigeron speciiosus 1.00 0.50 0.10 0.08 0.03 0.02 Bombus bifarius Site 2 Bombus flavifrons Plant species Preference factor Rudbeckia montana Hvdrophyllumrfndleri Wyethia amplexicaulis Helianthella quinquenervis Helenium hoopesii Viguiera mulhltiflora Polemonium Joliosissimum Erigeron speciosus 1.00 0.66 0.49 0.46 0.21 0.10 0.03 0.02 Plant species Preference factor Hydrophyllum fendleri Wyethia amplexic'aulis Agastache urticifJlia Vicia americana Helianthella quinquenervis Polemonium foliosissimum Rudbeckia montana 1.00 0.59 0.09 0.04 0.03 0.03 0.01 ' See Pleasants (1980) for data on the extent to which plant species are shared with another bumblebees species. half as much time per flower as before and receive half as much reward; thus the E/T would not change. What then is the relationship between preference (BIF) and NPR? If bees are foraging optimally and basing their choice of forage species on energy intake rate, then, when all bees have made their choice, the reward value of all plant species should be equal (see General Discussion). For two species with equal reward values (F/B),NPR,J, = (F/B)NPRJ,,, (5) where subscripts refer to species 1 and 2, respectively. After rearrangement Eq. 4 becomes (BIF), (BIF)2 _ NPR1 NPR2 (Table 2) show that the net energy intake rate is very similar for the species within each group. This is what optimal foraging would predict and is what was assumed to occur so as to set up Eq. 5. The results also show that for the five composites in Group 4, foraging costs are a small percentage of the reward obtained, justifying the approximation in Eq. 4. For the pairs of species in Groups 3 and 4 foraging costs are more significant. In addition these costs are probably underestimates because energy expenditures for thermoregulation, which are difficult to estimate (Pyke 1980), are not included. However, in both groups energy expenditures are the same percentage of gain for each member. This is also true for the composites in Group 4, the species in Group 1, and the two species examined by Pyke (1980). Consequently, Eq. 1 can be rewritten (B F) /(B F)2 is the bees' relative preference for E/T = , where p is the proportionality constant T for NPR the relative is and I NPR,/NPR2 species relating cost to gain. For two species with the same species 1. Thus, theoretically, preference should only when bee foraging has equalized reward values (see p, be a function of NPR. General Discussion), This conclusion is based on the assumption that enG1 - pG1 _ G2 - pG2 ergy costs of foraging are negligible compared to enT, T, ergy gain. Is this the case? One would expect that the term for After rearrangement cancels, indicating p since bumblebee workers are not foraging only themselves, they would have to maintain a sizeable that once again the relative reward value of two profit margin. In the few studies where cost as well as species is a function of their gross energy intake rate gain was measured, the cost as a percentage of the and thus their NPR's alone, as in Eq. 6. Why should gain varied with time of day, season, and year, and the cost percentage be so similar within each group'? with ambient temperature (Heinrich 1975a, Pyke One way this can occur is if the floral morphologies 1978&, 1980). In some cases the costs were significant. of the species in the group are similar, such that hanFor the two species examined by Pyke (1980) costs dling and traveling times are similar. In fact it can be averaged =34% of the gain, a value too large to be shown that foraging costs can be ignored whenever ignored. I estimated gain and cost for the species in traveling (or handling) time per flower is the same proTable 1. 1 was not able to measure directly the gain portion of total time per flower for both species. Each for these species because I did not measure the stand- of the groups in Table 1 is composed of species with ing crops of nectar. Except for the species in Group similar floral morphologies except Group 3. A. urtiI standing crops were too small to be measured by cifolia has small flowers which are densely arranged extraction with capillary tubes. However I did esti- on a spike allowing bees to walk between them. P. mate standing crop using Eq. 3. Lathyrus and Vicia foliosissimum inflorescences are composed of numerwere not included in this analysis because BIF values ous clusters of two to three flowers which are not arfor them were only indices. First of all, the results ranged in any regular fashion. Bees must fly between BEE RESPONSE TO NECTAR AVAILABILITY Decenmber 1981 wards them. I then considered whether the rate at which flowers could be visited (Th + T,) and the costs of foraging should be included along with NPR to provide a more precise measure of reward value. Time and energy costs of foraging cancelled out when comparing species. Thus, although time and energy costs do affect the absolute reward value of a species, they do not affect its reward value relative to another species. Only NPR differences influence reward value and consequently determine preference. Variation in patch quality 5( LJ HZ 4r 0 rl(, LJ - - / G LU ·/ / LU 20/ Resource I 0- / 2 28 JUNE FIG. 3. / / l / / B biforius 1 1 8 18 JULY 1 7 AUGUST 28 1655 Comparison of seasonal change in total available nectarproductionby plant species visited by B. bifariuswith seasonal change in B. bifjrius abundance. clusters. Because flying costs are 13 times greater than nonflight costs (see footnote", Table 2), the lower energetic cost of foraging on A. urticifjolia flowers may explain why it is slightly more preferred than expected based on NPR alone (Table 1). Despite the difference in floral morphology for these two species the cost is the same proportion of the gain for both (Table 2), indicating that NPR alone is still the dominant influence on preference. Whether the similarity in cost percentage among species visited by the same bees is a general pattern, regardless of floral morphology, remains to be seen. To summarize, in trying to characterize the reward values of species I initially considered only the relative magnitude of their NPR's, which is related to the gross energy intake rate. This was found to be closely correlated with the relative preference shown by bees to- To understand how bees respond to patches of flowers that differ in reward expectations I experimentally reduced the number of flowers within one of two patches of Agastache urticijolia at Site 2 and of Delphinium barbeyi at Site 3. The numbers of bees in the experimental patch and in the control patch were censused before and after the reduction. The results are shown in Figs. 1 and 2. Immediately after the resource reduction the bees in the experimental patch experienced a reduced average reward per flower because of the now higher BIF ratio. According to optimal foraging theory (Charnov 1976, Pyke et al. 1977) a bee should leave a patch when the reward expectation falls below the expectation from other patches. Whether or not bees are responding in this manner will be most apparent when the BIF ratio is examined. Under the null hypothesis we would expect no change in the bee density in the patch following bagging of inflorescences. This would be manifested by a higher ratio of bees to available (unbagged) flowers relative to the control patch. The percent increase in BIF would be equal to the percent decrease in the number of flowers in the patch. If bees forage optimally we would expect bee density to decrease in the experimental patch such that the BIF remains similar to that for the control patch. To evaluate the effect of bagging I have com- 50 (I-) SITE 2 uJ SITE 2 D z 40- 40- 2 0 r1r) (I-) 30- 30- / \ cnI L-J D m 20- / 20/ B tlavifrons biforius *14 w 0 0Resource LLi 10. / I D z 30 JUNE I-.' _ I 10 20 JULY I 1 30 9 AUGUST -_- · 30 JUNE ~l mw 10 I I I I 20 30 9 JULY AUGUST FI;. 4. Comparisons of seasonal change in total available nectar production by the plant species visited by B. hifjrius or B. flaviiJrons and the seasonal change in the abundance of B. bifjrius or B. flavifrons. JOHN M. PLEASANTS 1656 Ecology, Vol. 62, No. 6 the behavior of individual bees in the patch. They stayed longer in flowers and on inflorescences (J. * Pleasants, personal observation; also see Heinrich ·S u 1979a and Inouye 1978 for similar observations). z 30< One feature of Fig. 2 that deserves comment is the markedly higher BIF for both experimental and con* zr D~ < *trol patches on the 2nd d of the experiment. This was 20probably due to the fact that the experiment was con' near the end of the flowering season (18 and 19 eeducted ,, when the blooming of most species was endcnJ *August) Q had nowhere else to go. In fact, the numand bees ing 10ber of open flowers in the control patch declined by 50% from the 1st d of the experiment to the ·~*~~ ~~~~almost J ^~~~second. LJ[~ ~~ Despite the overall higher bee densities and thus the lower average profit margin on the 2nd d, the 10 30 20 40 BF's for both patches remained similar. This does not RESOURCE ABUNDANCE imply that the bees' method of choosing acceptable FIG. 5. Correlationbetween amount of nectar resource foraging locations has changed, but only that their noavailableand bee abundance.Data points from census times tion of what was an acceptable reward had been low40- in Figs. 3 and 4. r = .92, P < .001. ered. Seasonal variation in nectar availability pared the BIF for the experimental patch after treatment with the BIF for the control patch at that time. I have not compared the after-treatment BIF with the before-treatment B/F because some of the differences between these two could be due to diurnal fluctuations in bee density unrelated to the experiment. For Agastache urti('iolia the available resource was reduced by 31.7% in the experimental patch. As predicted, the BIF ratio remained similar to that for the control patch, being only 4.7% higher. For Delphinium barbeyi the results were somewhat equivocal. The number of flowers was reduced by 48.2% in the experimental patch. However, in this case the BIF increased, being 22.4% higher than the control patch. This is still a smaller increase than would be expected under the null hypothesis, indicating some movement of bees from the patch. The experimental design also made it possible to test the response of bees to resource augmentation. When the bagging was removed, the nectar which had accumulated in the bagged inflorescences over the previous day became available. Under the null hypothesis we would expect no change in bee density following removal of the bagging, which would result in a lower BIF ratio. For Agastache urticijolia the available number of flowers increased by 31.7% as a result of removing the bagging. Thirty minutes later, when the patch was censused, the BIF was already back to within control levels (only 8.9% lower than the control). For Delphinium barbeyi the available flowers were increased by 48.2% but the BIF (after 30 min) was within 4.3% of control levels. There was no indication that the accumulated nectar caused the BIF to overshoot control levels. This suggests that the accumulated nectar was rather quickly depleted, returning the system to a steady state. Although the BIF was unaffected, the accumulated nectar did have a noticeable effect on To examine the response of bumblebees to seasonal changes in nectar availability I focused on individual bumblebee species and the nectar production by the guild of plant species each visited. The response of a bumblebee species was measured by the change in its abundance over the summer (utilization curve). The seasonal change in the amount of resource available to that bumblebee (production curve) was obtained by summing the daily number of open flowers of each species weighted by its nectar production rate per flower. The results of a previous section indicated that the degree to which plant species are preferred is a fairly accurate predictor of their relative nectar production rates. Therefore, I employed preference as a weighting factor to convert the floral production of a species into its nectar resource equivalent. Although this would appear to be a very indirect way of determining the reward value of flowers there are two reasons why it might be better than using nectar production rates. First, as mentioned above, the reward value of a species may be influenced by its floral arrangement and each guild includes species which differ in this regard. The effect of this variable is difficult to measure but it appears that preference may take it into account. Second, the actual reward value of a plant species to one particular bumblebee species is influenced by the number of bees of other species foraging on it. Although most plant species have only one major bumblebee visitor, many of them are visited to a small extent (<30% of all visits), and a few are visited to a larger extent (>30% but <50%) by a second bumblebee species (Pleasants 1980). Using NPR's alone to estimate reward ignores the losses to other bees but preference would be expected to incorporate this factor. One other important point must be made to justify using preference to determine nectar production. At BEE RESPONSETO NECTAR AVAILABILITY December 1981 1974 & biforius B. flavifrons Apis mellifera -- - B. occidentalis 30- (Curve) Total bees 1 SITESITE 1657 (Uppe (Curve) 1/) LU D z 20- oC) cn LJ LJ 10- 30 20 June 10 20 July 30 9 19 August 20 June July August FIG.6. Changein the abundanceof bee species over time in 1974.The arrow marksthe point in the 1974 season that is equivalentto the end of the 1975study period. first glance it would appear that using preference as a weighting factor would necessarily cause a match between resource abundance and bee abundance because preference is based on the number of bees per flower. However, circularity is avoided because preference is based on the relative B/F's among species. Although BIF itself is clearly influenced by the abundance of bees, this influence is removed when the B/F for each species is standardized by dividing it by the largest BIF value among species. For example, suppose we found that the floral abundances of two species did not change from one census time to the next. Suppose also that the overall abundance of bees on these two species doubled over this same period. Then the absolute BIF for each species would be twice as large for the second census. However, the relative BIF will not change. When the number of flowers is weighted by preference it will indicate equal amounts of resource for the two census periods despite a doubling of bee abundance. Thus bee abundance and the estimate of nectar production are independent of one another. All that is assumed here is that the degree to which one species is preferred over the other is not influenced by total bee abundance (see General Discussion). The preference ratings for the species visited by B. bifarius at Site 1, by B. bifarius at Site 2, and by B. flavifrons at Site 2 are shown in Table 3. Utilization and production curves are shown in Figs. 3 and 4. What is important is the shape of the production curve relative to the shape of the utilization curve. There is no a priori criterion for determining the scale of these two curves. This is because the curves are not absolute measures of production or utilization but merely indicate change over time. A scaling factor was chosen for the production curve which facilitated comparison with the utilization curve. There is rather close agreement between seasonal production and utilization for all three cases. There is a significant correlation (r = .92) between the number of bumblebees present on census days and the estimate of the amount of nectar available to those bees (Fig. 5). The fact that the bee: resource ratio is relatively constant throughout the summer indicates that, in general, the degree of resource limitation for bees does not change over time. Thus the intensity of inter- and intraspecific competition among bumblebees for nectar resources changed little throughout the season. Despite the constancy of resource limitation within a site over time, sites can differ in the degree of their resource limitation. Comparing plant species found at both Sites 1 and 2, the average BIF was 2.6 times greater at Site 1, indicating that bees there were receiving less than half the profits per flower. A possible explanation for this is discussed below. The value of using preference ratings to take into account the nectar production lost to other bumblebee species is illustrated by the ratings for the plants visited by B. bifarius at Site 2. The factors for the members of the Asteraceae can be compared with the preference factors (per head) in Table 1 for these same species based on visits by all bees. The values for Helenium, Viguiera, and Erigeron are higher in Table 3. This is because the standard species, Rudbeckia, unlike these three species, is also visited by B. flavifrons. Thus its reward value to B. bifarius is lowered, making the value of the three species relatively higher. The value for Helianthella, on the other hand, is lower in Table 3. This species is also visited by B. flavifrons, but more so than Rudbeckia (Pleasants 1980), which lowers its relative value to B. bifarius. Variation between years If bumblebees are resource limited we would expect that the cumulative seasonal production of nectar would support only a limited number of bees. When total nectar production is similar for two different years we would expect that the total number of bees 1658 JOHN M. PLEASANTS ...... bifarius 1975 1975B.B j flavifrons Apis mellifera 50 SITE 1 Ecology, Vol. 62, No. 6 - -. Total bees Projected 40 LJ I- 7z O 30 - 20 - if) o LU LU r' 10 - II 1 10 20 July FIG. 7. I 30 II· I 9 August July August Change in the abundance of bee species over time, 1975. seen in the 2 yr would also be similar. The abundances had the greatest diet overlap were B. bifarius and B. of each plant species were very similar for the sum- flavifrons. B. flavifrons is a medium-tongued bumblemers of 1974 and 1975 (r = .88, P < .001) and hence bee while the other three species are short tongued the amount of available resource was constant for (Macior 1974). At Site 1, 8 out of the 10 plant species those 2 yr. In comparing bumblebee abundances for visited by B. bifarius and 4 out of 10 visited by B. these 2 yr I used the combined seasonal abundances flavifrons were also visited by honeybees. At Site 2, of species with short and medium tongue length. These all 11 species visited by B. bifarius and 7 out of the 10 visited by B. flavifrons were also visited by honspecies are B. bifarius, B. flavifrons, B. occidentalis, and the honeybee, Apis mellifera. Season totals for eybees. B. occidentalis visited 9 out of the 10 species bee abundance were obtained from the area under the used by B. bifarius and 3 out of the 10 used by B. curves in Figs. 6 and 7. Fig. 8 shows that the total flavifrons at Site 1. B. occidentalis also used all 11 number of bees seen at each site did not change from species visited by B. bifarius and 5 out of the 10 used 1974 to 1975. This supports the idea that a similar by B. flavifrons at Site 2. As Fig. 8 shows, B. bifarius amount of resource supports a similar number of bees. and B. flavifrons appear to have undergone competiA particularly significant feature of Fig. 8 is that the tive release in 1975. Increases in numbers such as this, contribution of certain species to the total bee abun- termed density compensation, are often observed for dance changed markedly between 1974 and 1975. The species on islands where competitors are missing numbers of B. bifarius and B. flavifrons seen in 1975 (Mac Arthur et al. 1972, Yeaton 1974). The presence were greater than for 1974 and compensated for the of density compensation in this pollinator system inlow numbers of B. occidentalis and Apis mellifera dicates that bees do have a competitive effect on one seen in 1975. The reason these latter two species had another by depleting a limited supply of resources. such low numbers was probably because they were Competitive release between bumblebee species has more vulnerable to the severe winter of 1974-1975. also been observed in a short-term species-removal For B. occidentalis the reduction might have been be- experiment (Inouye 1978). cause the queens of this species have much shallower overwintering hibernaculae than other bumblebees GENERALDISCUSSION (Hobbs 1968). The fall of 1974 was characterized both Nature of the response to variation by freezing temperatures and little insulating snowfall The is not which For native, previous sections showed the responses by Apis mellifera, (Inouye 1976). the reduction may have been due to inability to main- bumblebees to four kinds of variation in nectar availtain hives over the winter. In 1975, with honeybees ability: among species, among patches of flowers withand B. occidentalis virtually absent, the floral re- in a population, over a season, and between years. sources usually taken by these two species were avail- What are the mechanisms that produce these reable to B. bifarius and B. flavifrons. This situation sponses? thus provided a natural species removal experiment to The dynamics of the response of bees to variation test whether pollinators that use similar resources are in nectar availability among species can be better in competition with one another. The pollinator understood by referring to a conceptual model. The model is a more formal development and a graphic species with which B. occidentalis and A. mellifera BEE RESPONSETO NECTAR AVAILABILITY December 1981 SITE 2 SITE 1 700- B. occidentalis ::::-l Apis mellifera K'\\ B.Ibifarius B. flavifrons 1659 RA t 600 - 500 - ILJ ULJ . 49 ....... :: ::6: 1::' ::::6 ::::::: LU \ 66 :110 '372 400 - 406 :44 7 0 o -*u 300 - :::: :::::: D z c Qr D \\ 301< d 200 - 196 100- =176 ... 136 70 1974 1975 1974 1975 FIG. 8. Comparison of total number of bees and composition of bee fauna for species of short and medium tongue length in 1974 and 1975. representation of an explanation given by Heinrich (1976a). The form of the model is borrowed from Fretwell (1972), who used it to explain the successive colonization of optimal and suboptimal habitats by birds. Consider four plant species, A, B, C, and D, in a simplified system where nectar secretion begins in the morning and the rate remains constant throughout the day. Points RA, RB, Rc, and RI (see Fig. 9) refer to the potential reward value (nectar sugar intake rate) of each species as foraging begins in the morning, and are indicative of each species' NPR. The curves extending downward from each of these points indicate the decline in the actual reward value of each species as the number of bees foraging on it increases. I will assume, as discussed above, that the costs of foraging do not influence reward value. I will also assume, although this is not essential, that bees remain constant once they have chosen which species to forage on. As the first bumblebees begin to visit flowers in this meadow they should choose to forage on the species which provides the highest reward (species A). As more bees arrive in the meadow the number of bees on species A increases and consequently its expected reward value decreases. Eventually the expected reward from species A will decrease to a level equal to that of a second, as yet unvisited species (species B; point RB in Fig. 9). Additional bees entering the meadow will find both species equally profitable and will be just as likely to choose to forage on one as the other. With the continued influx of bees to the system, the number of visitors on both A and B increases and the reward c b a NUMBEROF VISITORS PERFLOWER FIG.9. Model showing the relationshipbetween the nectar productionrates of species A, B, C, D and the number of bees foragingon them per flower. RA, RB, RC, and RD are the potential reward values of species A, B, C, and D beforebees beginforaging,andare indicativeof each species' nectar production rate. The curves A, B, C, D show the decline in each species' rewardvalue as the numberof bees foragingon it increases. The intersectionsof the horizontal lines and curves A, B, C, D indicatepoints where the reward value of species is the same. Points a, b, c are the B/F values when the numberof bees in the meadow has stabilized;the reward value of all three species is then the same (=R). Species D does not receive visits in this scheme. expectation of both of these species becomes reduced until a third species (species C) has equal reward value (point RC) and begins to attract visitors. When no more bees enter the meadow the system reaches a steady state. At this equilibrium point the flowers of all species will have the same reward value (R) and standing crops of nectar will not change. It should be noted however, that each species may have a different standing crop of nectar at this point. The standing crop will depend upon how rapidly flowers can be visited (see earlier discussion). A species with a greater traveling and handling time per flower will have a larger standing crop. The shapes of the curves in Fig. 9 come from assuming that the lag time between successive visits to a flower is a function of the reciprocal of the number of bees foraging per flower, as discussed earlier. With curves of this shape the relative attractiveness of species (based on their B/F's, a, b, c, and d in Fig. 9) does not change with overall bee densities (i.e. with position of the equilibrium value R) and always corresponds to the relative magnitude of NPR's (RA, RB, RC, and RD). In Fig. 9, RD is lower than R, meaning that species D will not be visited. Actually, because of the accumulation of nectar in its flowers, the reward value of species D may later in the day equal or exceed that of other species, at which point it will receive visits. To incorporate this effect into the model would require a third axis, a time axis, showing 1660 JOHN M. PLEASANTS the rise in a species' reward value prior to visitation. This would lower the BIF when a higher ranking species would be perceived by bees as being equal to a lower ranking one. However, it will not affect the eventual relative B/F's among species. Although Fig. 9 is primarily of heuristic value, it does have some testable features. It predicts that at equilibrium, species should have a similar reward value. Heinrich (1976a, 1979b) and Pyke (1980) provide some support for this as do the estimates of reward value for the pairs or groups of species in Table 2. It also predicts that when R is sufficiently lowered, due to a high overall abundance of bees, some species, not previously visited, should begin to receive visits. This is illustrated by bumblebee utilization of Geranium fremontii (Geraniaceae), a species found at both Sites 1 and 2. Geranium produces a small amount of dilute nectar (Watt et al. 1974) and is mainly visited by flies. At Site 1, where bumblebee density is higher and flowers receive two to three times as many visits, Geranium is visited by B. bifarius and B. flavifrons. At Site 2, where the species is equally abundant, I have never observed bumblebees foraging on it. Another feature of the model is the successive colonization of species with different NPR. This has been observed in a field experiment (Heinrich 1979b). However, it will probably not be observed under natural conditions since it appears that a bee's decision about which species to forage on is not made on a daily basis. A bee's specialization for a certain species is often remembered and carried over to successive days (Free 1970, Heinrich 1976a). However, there is evidence that despite this constancy there is low-level visitation to other species (Heinrich's majoring and minoring, 1976a, 1979b). This has been interpreted in part as a periodic assessment of the floral environment to determine if the species currently being visited is the most profitable. In contrast to the relatively long-lasting decision as to which species to forage on, a bee's choice of whether to remain in a patch (or on an inflorescence) appears to be part of a continual decision-making process. This is some evidence that a decision is made after each flower visited (Pyke 1978d). How this decision is made is not precisely known but it is thought to involve an assessment of the average reward expectation for a species in the area in which the bee is foraging. When flowers in a patch are encountered which have rewards less than this standard, bees will leave the patch. Such behavior will be optimal when the reward values of neighboring inflorescences in a patch are correlated (hot or cold spots, Pleasants and Zimmerman 1979). This does appear to be the case in the few species examined to date (Pleasants and Zimmerman 1979, M. Zimmerman, personal communication). Considering the inflorescence as a patch, there also appears to be a correlation between the reward value of neighboring flowers for many species (Pyke 1978c, J. Pleasants and M. Zimmerman, personal observation). Ecology, Vol. 62, No. 6 There are two possible explanations for the seasonal tracking of resource levels by bumblebees as seen in Figs. 3 and 4. One, it could represent a population size (numerical) response to changes in resource levels. Two, it could be a facultative response in foraging activity to changes in resource availability. Explanation 1 implies that increases in nectar production bring about worker production. While this is probably generally true, the appearance of new workers is not instantaneous. It takes 21-30 d for an adult worker to develop from an egg (Hobbs 1968). It is possible that the appearance of new workers could be timed to coincide with anticipated future resource levels. However, the shape of the bumblebee abundance curve is very different at Sites 1 and 2 for B. bifarius. If the numerical response is a programmed feature of the biology of this species we would expect similar responses. Also, the production of workers can only account for increases in bumblebee abundance and cannot explain the midseason decrease in bee abundance observed at Site 1. The response suggested by explanation 2 may be of two forms. Bumblebee hives have a division of labor such that some bees are engaged in foraging while others are involved in nest maintenance and brood care (Free and Butler 1959). This suggests the possibility that bees normally involved in nest maintenance might be pressed into foraging service when available resources increase or foragers might remain in the hive when resources decrease. The available evidence actually indicates that bumblebee foraging activity might operate contrary to what I have just described. When nectar in the hive honeypot is augmented experimentally (simulating increased resource availability) fewer workers leave the nest to forage (Brian 1954). On the other hand, when colony resources are reduced experimentally (Brian 1954) or when nectar input to the hive is reduced because foragers are removed (Inouye 1978), more bees, including house bees, begin to forage. These experimental conditions are rather drastic but do suggest that the changes in bee abundance seen in Figs. 3 and 4 are not regulated entirely by the number of foragers sent from the hive. It must be remembered that bumblebees are solitary foragers and do not recruit other individuals to nectar sources as do honeybees. A second, more likely form of facultative response involves changes in the dispersion of bees over the meadow. My census route covered only the main part of the meadows where flowers were most dense. There were also flowers near the margins of the meadow and scattered under trees. All else being equal, bees should prefer to forage in the central area. As the density of bees increases, their average profit will decrease and additional bees may begin to visit more peripheral areas. The successive colonization would be similar to that modeled by Fig. 9. Plants in marginal areas may therefore act as a buffer zone, receiving visits when resource levels get too low in the central area, but receiving few visits when resource levels are high. BEE RESPONSE TO NECTAR AVAILABILITY December 1981 This will tend to keep the ratio of bees to available resource in the central area relatively constant over time. The higher overall bee: resource ratio I observed at Site 1 may be due to the fact that the meadow is much smaller than Site 2 (0.6 vs. 2.4 ha) and more isolated. At Site 1 bees may be forced to continue foraging in the central area because they have nowhere else to go. A numerical response is the most plausible explanation for the competitive release and density compensation observed when comparing bumblebee abundances over 2 yr (Fig. 8). In this case the additional nectar resources available in 1975, due to the absence of Apis Imtellijera,allowed the populations of B. bifarius and B. flavifrons to increase over 1974 levels. In general the fine-scale match-up between bee abundance and nectar level differences among species or patches would appear to involve foraging adaptations which maximize net energy returns. The coarse tuning of bumblebee abundance to seasonal or yearly nectar production levels probably involves both a foraging activity and a bee population growth response. ACKNOWLEDGMENTS Part of this research comes from my Ph.D. work at the University of California, Los Angeles (UCLA). I would like to thank my thesis readers, Henry Hespenheide and Lynn Carpenter, and especially my advisor Martin Cody. This part of the research was supported by a grant from Sigma Xi and from the Stephen A. Vavra fund at UCLA for botanical research. The rest of the work was done while at Washington University in St. Louis. I would like to thank the following people for comments on earlier drafts or for advice and suggestions: Mike Zimmerman, David Inouye, Graham Pyke, Peter Raven, and Owen Sexton. I would particularly like to thank my wife, Barbara, for assistance in the field and suggestions on improving the manuscript. LITERATURE CITED Beattie, A. J., D. E. Breedlove, and P. R. Ehrlich. 1973. The ecology of the pollinators and predators of Frasera speciosa. Ecology 54:81-91. Brian, A. D. 1954. The foraging of bumblebees. Bee World 35:61-91. Butler, C. G. 1945. The influence of various physical and biological factors of the environment on honeybee activity. An examination of the relationship between activity and nectar concentration and abundance. Journal of Experimental Biology 21:5-12. Charnov, E. L. 1976. Optimal foraging, the marginal value theorem. Theoretical Population Biology 9:129-136. Free, J. B. 1970. The flower constancy of bumblebees. Journal of Animal Ecology 39:395-402. Free, J. B., and C. G. Butler. 1959. Bumblebees. Collins, London, England. Fretwell, S. D. 1972. Populations in a seasonal environment. Princeton University Press, Princeton, New Jersey, USA. Heinrich, B. 1975a. The role of energetics in bumblebeeflower interrelationships. Pages 141-158 in L. E. Gilbert and P. H. Raven, editors. Coevolution of animals and plants. University of Texas Press, Austin, Texas, USA. .1975b. Energetics of pollination. Annual Review of Ecology and Systematics 6:139-170. 1661 . 1976a. The foraging specializations of individual bumblebees. Ecological Monographs 46:105-128. . 1976b. Resource partitioning among some eusocial insects: bumblebees. Ecology 57:874-889. 1979a. Resource heterogeneity and patterns of movement in foraging bumblebees. Oecologia 40:235-245. . 1979b. "Majoring" and "minoring" by foraging bumblebees, Bombus vagans: an experimental analysis. Ecology 60:245-255. Heinrich, B., and P. Raven. 1972. Energetics and pollination ecology. Science 176:597-602. Hobbs, G. A. 1968. Ecology of species of Bombus in southern Alberta VII subgenus Bombus. Canadian Entomologist 100:156-164. Inouye, D. 1976. Resource partitioning and community structure: a study of bumblebees in the Colorado Rocky Mountains. Dissertation. University of North Carolina, Raleigh, North Carolina, USA. . 1978. Resource partitioning in bumblebees: experimental studies of foraging behavior. Ecology 59:672-678. Kauffeld, N. M., and E. L. Sorenson. 1971. Interrelations of honeybee preference of alfalfa clones and flower color, aroma, nectar volume, and sugar concentration. Research Publication 163, Kansas Agricultural Experiment Station, Manhattan, Kansas, USA. Mac Arthur, R. H., J. M. Diamond, and J. Karr. 1972. Density compensation in island faunas. Ecology 53:330-342. Macior, L. W. 1974. Pollination ecology of the Front Range of the Colorado Rocky Mountains. Melanderia 15, Washington State Entomological Society, Washington State University, Pullman, Washington, USA. Mosquin, T. 1971. Competition for pollinators as a stimulus for the evolution of flowering time. Oikos 22:398-402. Pleasants, J. M. 1977. Competition in plant-pollinator systems: an analysis of meadow communities in the Colorado Rocky Mountains. Dissertation. University of California, Los Angeles, California, USA. 1980. Competition for bumblebee pollinators in Rocky Mountain plant communities. Ecology 61(6):14461459. Pleasants, J. M., and M. Zimmerman. 1979. Patchiness in the dispersion of nectar resources: evidence for hot and cold spots. Oecologia 41:283-288. Pyke, G. 1978a. Optimal body size in bumblebees. Oecologia 34:255-266. . 1978b. Are animals efficient harvesters? Animal Behavior 26:241-250. . 1978c. Optimal foraging in bumblebees and coevolution with their plants. Oecologia 36:281-293. 1978d. Optimal foraging: movement patterns of bumblebees between inflorescences. Theoretical Population Biology 13:72-98. 1980. Optimal foraging in bumblebees: calculation and net rate of energy intake and optimal patch choice. Theoretical Population Biology 17:232-246. Pyke, G. H., H. P. Pulliam, and E. L. Charnov. 1977. Optimal foraging: a selective review of theory and tests. Quarterly Review of Biology 52:137-154. Schoener, T. W. 1971. Theory of feeding strategies. Annual Review of Ecology and Systematics 2:369-404. Watt, W. B., P. G. Hoch, and S. G. Mills. 1974. Nectar resource used by Colias butterflies. Oecologia 14:353-374. Yeaton, R. I. 1974. An ecological analysis of chaparral and pine forest bird communities on Santa Cruz Island and mainland California. Ecology 55:959-973. Zimmerman, M. 1979. Optimal foraging: a case for random movement. Oecologia 43:261-267.