Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Introduced species wikipedia , lookup
Biodiversity action plan wikipedia , lookup
Latitudinal gradients in species diversity wikipedia , lookup
Habitat conservation wikipedia , lookup
Island restoration wikipedia , lookup
Theoretical ecology wikipedia , lookup
Angraecum sesquipedale wikipedia , lookup
Convergence, Competition, and Mimicry in a Temperate Community of HummingbirdPollinated Flowers Author(s): James H. Brown and Astrid Kodric-Brown Source: Ecology, Vol. 60, No. 5 (Oct., 1979), pp. 1022-1035 Published by: Ecological Society of America Stable URL: http://www.jstor.org/stable/1936870 . Accessed: 21/06/2011 15:04 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 service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology. http://www.jstor.org Ecology, 60(5), 1979, pp. 1022-1035 ? 1979 by the Ecological Society of America CONVERGENCE, COMPETITION, AND MIMICRY IN A TEMPERATE COMMUNITY OF HUMMINGBIRD-POLLINATED FLOWERS' JAMES H. BROWN AND ASTRID KODRIC-BROWN Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, Arizona 85721 USA Abstract. We studied the pollination ecology of nine species of red, tubular flowers which bloom together in different combinations in the White Mountains of Arizona, USA. All species were strikingly convergent in floral color, size, and shape. Hummingbirds, the primary pollinators, usually did not visit flower species selectively, and individual birds often simultaneously carried four or more species of pollen. Flowers may have competed interspecifically for these shared pollinators, but competition was reduced because character displacement in orientation of anthers and stigma resulted in some species using different parts of the bird to transport their pollen. Most flower species secreted nectar at similar rates, particularly when they bloomed together in mixed stands. A population of Lobelia cardinalis secreted no nectar; it attracted hummingbirds by mimicing more abundant, nectarproducing species. This temperate flower community, which resembles some associations of convergent Mullerian and Batesian mimics, appears to have evolved its characteristic convergent structure because the advantages of using similar signals and rewards to share the same hummingbird pollinators outweigh the advantages of diverging to reduce interspecific competition. Key words: coevolution; coexistence; community; competition; convergence;flowers; hummingbirds; mimicry; nectar; pollination. INTRODUCTION Two processes produce the mutualistic relationship between plants and pollinators and determine the course of their coevolution. Transport of pollen between individual plants by foraging animals results in outcrossing, and production of nectar or pollen provides a food resource which attracts pollinators to flowers. In most natural situations, this mutualistic relationship is complicated by interspecific competition among plants for pollinators and among pollinators for nectar and pollen. Coexisting flower species often differ in morphology, odor, color, flowering time, and nectar and pollen rewards, and these patterns have been interpreted as adaptations to reduce competition for pollinators by promoting specific plant-pollinator associations (e.g., Robertson 1895, Free 1970, Levin and Anderson 1970, Mosquin 1971, Heithaus 1974, Heinrich 1976a, Lack 1976, Waser 1978). An apparent exception to this general pattern of competition, character displacement, and pollinator specificity was described by Grant and Grant (1968) in their monograph on hummingbirds and the flowers which they pollinate. The Grants noted that hummingbird-pollinated flowers in western North America are characterized by striking convergence in shape and color and are pollinated by one of several similar hummingbird species. They report that such convergence extends to those plant species which bloom simultaneously in the same habitats, producing (p. 92) i. . . a regional ecosystem consisting, on the one hand, of I Manuscript received 29 August 1978; accepted 22 January 1979. several hummingbird species able to feed equally well on any native species of hummingbird flower, and on the other hand, of numerous species of bird flowers which can be pollinated successfully by any species of hummingbird" and that this situation, ". . . with broad standardization and interchangeability of parts between different species of birds and plants, prevails in western North America." The Grants described morphological, biogeographic, and systematic attributes of temperate hummingbirds and flowers to document this convergent pattern, but they did not obtain sufficient data on nectar rewards and pollen transport to understand the ecological processes responsible for the coevolution and maintenance of these communities. This paper examines interactions among coexisting hummingbird-pollinated plant species in the mountains of eastern Arizona. Relationships among the hummingbirds have been reported in an earlier paper (Kodric-Brown and Brown 1978). In the present study, we have verified the striking convergence in floral shape and color described by Grant and Grant (1968), but in addition, we have obtained detailed quantitative data on spatial and temporal distribution of flowers, rates of nectar secretion, and patterns of pollen transport in order to determine how these flower species interact with each other and with their pollinators. Our results suggest that the flowers compete for pollinators, that floral adaptations reduce deleterious effects of this competition, and that the convergent organization of this community represents a stable condition typical of coevolved associations of hummingbirds and plants in temperate habitats. TEMPERATE HUMMINGBIRD FLOWERS October 1979 METHODS Study area We studied different associations of plant species at many specific localities, but all work was done within a 30-km radius of Alpine in the White Mountains of east central Arizona. Elevation was 1800-3000 m. Habitats include chaparral, riparian forest, and pinonjuniper woodland at the lowest elevations, through open forests of pine and oak at intermediate altitudes, to dense forests of mixed conifers at the highest elevations. At all altitudes, there were patches of open habitat with few if any trees; these included natural meadows, talus slopes and cliffs, as well as man-made meadows and roadsides. Artificially disturbed habitats often supported large populations of flowers and were convenient study sites. However, all disturbed areas studied were comparable in habitat structure, species composition, and flower density to natural sites nearby. Most of the study was conducted from June through August, 1973-1975, but we spent short periods on the study area in 1972, 1976, and 1977. In each of these years, we also visited similar habitats in the southern Rocky Mountains and the isolated mountain ranges of southern Arizona and New Mexico, where many of the same flower species also occur. Observations in these areas were an invaluable source of comparative information which enable us to be more confident of the generality of the results reported here. Characteristics of flowers, nectar, and pollen Our work centered around local habitats where red, tubular, hummingbird-pollinated flowers were in bloom. We concentrated on sites where each species was sufficiently abundant to measure conveniently floral and nectar characteristics, but we also searched over the range of habitats and elevations in the study area to document distributions and to observe plantpollinator interactions where densities were low. Densities of flowers of each species were recorded using an arbitrary scale which ranged from one (<0.05 inflorescences/m2) to five (>5 inflorescences/m2). Specimens of each species were obtained to verify identification and to obtain reference pollen samples. Not all flower species were studied with equal intensity. We tried to obtain basic data on distribution, flowering time, floral morphology, and 24-h nectar secretion rates for all species. We made particularly detailed measurements on three species, Ipomopsis aggregata, Penstemon barbatus, and Castilleja integra, which were of particular interest because they often bloomed together in dense mixed stands. Most measurements of these species, as well as data on hummingbirds foraging on them, were obtained in the immediate vicinity of a single large meadow, 6 km northeast of Nutrioso, Arizona at 2300 m elevation. Several characteristics of flowers were measured, including length of the constricted floral tube (corolla 1023 of most species), which permits insertion of only the bill and tongue of a foraging hummingbird. Life of flowers was determined by removing all existing blossoms from a plant, marking individual flowers as they opened, and recording when these were dropped. Importance of animal pollination was assessed by covering individual plants with fine (3-mm) mesh nylon bags and leaving them in place for the entire blooming season. The mesh was sufficiently fine to exclude all animal visitors except the smallest insects. Adjacent plants of similar size, marked at the time of bagging but left uncovered, served as controls. Bagged and control plants were collected at the end of the flowering season when number of inflorescences, aborted flowers, fruits, and seeds were counted. Nectar secretion and availability were measured by collecting nectar from individual flowers in calibrated microcapillary tubes. Nearly all samples were taken at the same time of day (1400-1600) to standardize for daily variation in nectar secretion and hummingbird foraging. To measure rates of nectar secretion, plants were covered with nylon mesh bags to exclude nectar feeders. After flowers had been bagged for 24 h, accumulated nectar was measured, and amount of nectar in control flowers, measured at the time of bagging, was subtracted to calculate secretion rates. This method does not take into account those flowers which opened and began secreting nectar during the 24-h period of measurement. However, when the life of flowers is known, it is possible to correct measurements of nectar secretion rates for the proportion of newly opened flowers that have not been secreting for the full 24 h. Nectar concentration was measured with a hand-held, temperature-compensated refractometer. This measures nectar concentration in terms of equivalent percent sucrose and gives an accurate estimate of its caloric value to hummingbirds (Hainsworth and Wolf 1972). Because rates of nectar secretion per flower are small, it is difficult to measure accurately daily patterns of nectar secretion. However, we were able to estimate these patterns for I. aggregate and P. barbatus in 1974 when hummingbirds were so dense in relation to floral resources that their foraging kept the standing crop of available nectar per flower at very low levels. At standard times throughout the day (0600, 0900, 1200, 1500, 1800), we measured the volume of nectar in half of the flowers on sample plants, then immediately bagged the remaining flowers on sample plants, left the bags in place for 2 h, and then measured nectar volume in these flowers. Secretion of nectar during the 2-h period was determined by subtracting the mean initial volumes prior to bagging from the mean final volumes after bagging. By using the same technique, but leaving bags in place either from 0600-2000 or from 2000-0600, we also measured secretion during daytime and overnight periods, respectively. 1024 JAMES H. BROWN AND ASTRID KODRIC-BROWN Pollen samples were collected from mist-netted birds by pressing the sticky side of clear plastic tape against the bird and then sticking the tape to a clean white paper or glass slide. Separate samples were taken from the bill, crown, sides (cheeks), and chin. These samples were returned to the laboratory where number and species of pollen grains were counted under a microscope. If a sample contained fewer than 100 grains, all were counted; otherwise, the first 100 grains in arbitrarily selected fields were counted. Flower stigmas were collected, prepared under clear plastic tape, and the adhering pollen grains were later counted under a microscope. Pollen of P. barbatits collapsed and became difficult to detect when it had germinated on nonspecific stigmas, so counts of P. barbatus stigmas provide only a very conservative estimate of the quantity of nonspecific pollen present. Length and width of pollen grains were measured with an ocular micrometer. RESULTS AND DiSCUSSION Spatial anal temporal distribution of'fl() ers Nine species of red, tubular flowers, representing eight genera and seven families, bloomed within 30 km of Alpine, Arizona (Table 1). These plants are herbaceous perennials, except for Lonicera arizonica, which is a trailing, somewhat woody perennial vine, and Ipomopsis aggregate, which is a biennial herb. Distribution of the flower species varied spatially with habitat and elevation (Table 1). Penstenon barbatiis and I. aggregate were abundant and widespread. They occurred in many habitats and over a wide range of altitudes. They attained their highest densities in open or disturbed habitats such as meadows, cliffs, eroded hillsides, and banks of temporary watercourses, where they often bloomed profusely in mixed stands which sometimes exceeded densities of 20 inflorescences and 250 flowers/m2. The two species of Castilleja also were abundant, but they were segregated by habitat: C. integra occurred in dry meadows and open piflon-juniper woodland below 2500 m, whereas C. austromnontana was common in the herbaceous understory of aspen and conifer forests above 2500 m. Lonicera arizonica and Echinocerelis triglochidiatus were restricted to mesic talus slopes and bare rock outcrops, respectively, but sometimes they were dense in appropriate habitat. Aquilegia triternata and Si/ene laciniata were found only at low densities in a few mesic habitats at high elevations. Lobelia cardlinalis was known only from one population of 100 individuals growing along the outflow of a small spring at 2000 m. Distribution of these flowers also varied temporally, both within a growing season and from year to year. Most species bloomed for long periods (6-12 wk) wherever they occurred. Although there was some tendency for the peak flowering of some coexisting Ecology, Vol. 60, No. 5 species to be displaced (Table 1; Kodric-Brown and Brown 1978), the overall pattern of flowering phenology was such that there was much interspecific overlap and hummingbird flowers bloomed in most habitats throughout the summer. In arid habitats at low elevations, E. triglochidiatus and C. integra bloomed first, followed by P. barhatus and then by I. aggregata, but there was much overlap. In more mesic habitats higher in the mountains, there was somewhat more separation; A. triternata, Lonicera arizonica and E. triglochidiatus were replaced later in the season by P. barbatus, I. aggregate, C. austromontana, and S. laciniata. Lobelia (ardlinalis did not bloom until near the end of the season, overlapping the end of the flowering period of P. barbatus and I. aggregate, which bloomed in drier habitats nearby. Amount and seasonal pattern of precipitation varied greatly from year to year, and this had great effect on the local abundance and distribution of flowers. Normally precipitation falls primarily in winter (November to April) and summer (July to August); the spring and fall are dry. In 1974 and 1977, winter precipitation was far below average, and the light snow cover melted early in the spring. Because of the ensuing drought most flower species suffered widespread mortality and failure to bloom. The effect was most pronounced at lower elevations. At 2300 m, fewer than 1% of the flowers blooming in 1972, 1973, 1975, and 1976 were present in 1974 and 1977. In places where hillsides had been covered with dense stands (>5 inflorescences/ mi2) during normal years, it was impossible to find a single flower during the droughts. During the drought years, hummingbirds competed for nectar with incredible intensity. Their frequent frequent foraging resulted in extremely low levels of nectar (standing crops) available in individual flowers (Table 2). During 1974 and 1977, only a small fraction of the birds which normally bred on or migrated through the study area could be supported there; many birds emigrated, but many others undoubtedly died of starvation (Kodric-Brown and Brown 1978). Our earlier paper (Kodric-Brown and Brown 1978) gave detailed data on seasonal flowering phenology and the exact number of flowers of each species within hummingbird territories at the primary study site. While that paper should be consulted for details (see especially Fig. 1 and Table 2), the major patterns will be summarized briefly here. In 1975, P. barbatus and C. integra flowered profusely from early July to midAugust. Ipomnopsisaggregate began blooming in midJuly, reached a peak in mid-August, and probably continued to flower well into September. We counted flowers in 47 mapped hummingbird territories, 23 in the period from July 20-31, 1975, and 24 from August 4-6, 1976. Number of flowers per territory ranged from 302 to 1699. Relative proportions of the three species varied widely from virtually pure stands of P. barbatus and L. aggregate to approximately evenly TE'FMPERAT'E HUMI\IINGBIRI) Ocvtober 19)79 tt tt tn t :Zo >. 102'5 FlO(WEIRS tt E< X tS s~~ct OnX _tv; q: t_ J ur = a~ C s _ X O XCv, .XC Ct iv - . ri - l "3 v, x0Nt xO0 - Z < Q c c, O a _ baxO >E tt-3>_s: - 3?~'t oc =o. < ttv nE -t~a -vX ' ~ ~ ~ -> 9 vr J L CE, _ IC "I ,,crOC C rlc1,rl r1 FlJ c_, ~~~r IC r _~ J~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ _ X e) X Q e) E.: _c ~~~~~~~~~~~~~~~_-_c. - > t>-M U~~~~~~~~~~~~~~~~~~~~~~~~~~7 ? ) ~~~~~~~~~~~~~~~~~~~~~~~ _ C ct Cr m D D C: v c 1026 JAMES H. BROWN AND ASTRID KODRIC-BROWN Ecology, Vol. 60, No. 5 2. Year-to-year variation in quantity of nectar available in two flowers on our study area in late July and early August. Note that in the drought year of 1974, flowers contained a small fraction of the amount in other years. Since the flowers produced nectar at similar rates in all years, this difference reflects increased rate of hummingbird visitation in 1974 when the number of flowers blooming was < 1% of the number in flower in the other years. Values in parentheses are sample sizes. TABLE x Available nectar/flower (gl) Flower species 1973 1974 1975 1976 Penstemon barbatus Ipomopsis aggregate 1.57 (159) 1.61 (283) 0.26 (150) 0.20 (300) 1.43 (1000) 2.14 (800) 1.14 (401) 1.60 (828) mixed two- and three-species stands. We did not map any territories which contained only C. integra, but there were several such territories within 0.5 km. These broad interspecific overlaps in space and time are characteristic of hummingbird-pollinated flowers throughout our entire study area. Floral morphology and nectar The nine species had convergent floral morphologies (Fig. 1). All were predominantly bright red, had generally tubular shapes, and in all species except E. triglochidiatus the tube was constricted so as to admit only the bill and tongue of a foraging hummingbird. In seven species the tube was of similar length (20-25 mm, Table 1). Since most species belong to different genera and families and have close relatives which are strikingly different in appearance and are insect pollinated (Grant and Grant 1968), the similarities must be attributed to convergence toward a common hummingbird-pollinated form. As is often true in evolutionary convergence, these species have achieved similar appearance and function by different means. For example, most species had petals fused to form a thickened corolla tube, but S. laciniata had thin, sep- arate petals held in a functional tube by a tubular calyx, and A. triternata had long, tubular spurs on the petals which contained nectaries at the bases. In the two species of Castilleja, most of the red color was owing to coloration of the stout bracts which surrounded the rather fragile, colorless perianth. In some species the convergent similarity involved not only the individual flower, but also extended to the arrangement of blossoms on the inflorescence. The most striking example was provided by I. aggregate, P. barbatus, and Lobelia cardinalis; these species displayed their flowers on loose spikes having similar numbers and spacing of open blossoms (Table 3). We might expect these flower species also to have converged to provide similar nectar rewards; otherwise, hummingbirds might learn to discriminate against those flowers which secrete less sugar than others (Hainsworth and Wolf 1976). Unfortunately the problem is more complicated, because the costs of hummingbird foraging vary with flower density, environmental temperature, intensity of competition with other birds, and other factors. In order to be similarly attractive, flowers should provide similar net, rather than gross, rewards, and we might expect sig- 3. Characteristics of the three species of hummingbird-pollinated flowers which were studied in detail where they bloomed in mixed stands, 6 km northeast of Nutrioso, Arizona. Values of 24-h nectar accumulation in bagged flowers measured in 1975 were used to calculate nectar secretion, corrected for differences between species in the proportion of new flowers opening each day. Note that this method results in all three species showing higher and more similar rates of nectar secretion than the values presented in Table 1. Sample sizes are given in parentheses. TABLE Species Arrangement of flowers on plant Mean nectar Mean life of concentrations flower (d) (% sucrose) Mean 24-h nectar secretion* (pA) (mg sucrose) Ipomopsis aggregate Loose, branched inflorescence; usually 1-4 inflorescences/plant and 5-15 open flowers/inflorescence 1.81 (104) 27.54 (52) 8.59 (100) 2.37 Penstemon barbatus Loose, branched inflorescence; usually 1-6 inflorescences/plant and 4-12 open flowers/inflorescence 2.96 (89) 33.89 (41) 7.55 (100) 2.56 Castilleja integra Tight, unbranched inflorescence; usually 2-7 inflorescences/plant and 2 receptive flowers/inflorescence 2.00 (33) 41.75 (31) 4.98 (96) 2.08 * 24-h nectar secretion (y) was calculated from the equation A py/2 + (I - p)(x + y), where A = 24-h accumulation in bagged flower, x = amount of nectar in control flower at time of bagging, and p = proportion of new flowers opening each day. Values of p used were 0.5, 0.33 and 0.5 for I. aggregate, P. barbatus and C. integra. respectively. October 1979 1027 TEMPERATE HUMMINGBIRD FLOWERS Ipomopsis uggregatf Penstemon burbafus Cast//fe/u infegru Aqul/egia triternato Lonicera arizonica Castif/e/u austromontana Silence Eclinocereus Lobe li triglochidiatuls card/na//s Gciftiatf FIG. 1. Scale drawings of the nine species of red, hummingbird-pollinated flowers which bloomed on the study area. Note the convergence in size and form. nificant local variation in rates of sugar secretion. Over our entire study area, most flowers secreted 1-8 gl of nectar, which varied in concentration from 20-45% sucrose; this amounts to secretion rates of 0.5-2.5 mg of sucrose/d (Table 1). A conspicuous exception was Lobelia cardinalis, which secreted no nectar and appears to be mimetic. Although the five-fold variation in nectar secretion rates may not appear to make a convincing case for convergence, it is much less than the 40-fold variation recorded for nonconvergent hummingbird-pollinated flowers in Puerto Rico (Brown and Kodric-Brown, personal observation) and the almost 1000-fold variation among morphologically dissimilar bumblebee-pollinated flowers which bloom together in the same habitats in Maine (Heinrich 1975). A better test for convergence in nectar secretion rates is degree of similarity when flowers bloom simultaneously in mixed species stands, because costs 6 --60 4. Effects of bagging flowers to exclude pollinators on fruit and seed set in two species of hummingbird-pollinated plants. Note that bagged plants produced approximately as many inflorescences and flowers, but far fewer fruits and seeds than unbagged controls. TABLE PENSTEMON BARBATUS 4 - c?I -40 0 2~~~~~~~0 2 2 2 z~~~~~~~~~ 0~~~~~~~~~ C) Ecology, Vol. 60, No. 5 JAMES H. BROWN AND ASTRID KODRIC-BROWN 1028 00< 0~~~~~~~~~ LAJ~~~~~~~~~~~~~~~~~~~~~~~~~~L z~~~~~~~~~ Inflorescences/ Plants plant Bagged Control 10 10 Bagged Control 8 8 Fruits/ plant Seeds/ fruit Penstemon harhatus 6.40 50.10 2.60 35.00 58.10 2.70 6.48 30.35 Ipoinopsis 2.75 2.50 Flowers/ plant aggregat(1 236.13 188.00 13.50 105.75 3.76 5.97 I. aggregate and P. barbattus, but these do not suggest any strong displacement in daily patterns of nectar secretion (Fig. 2). Although there is some tendency for P. barbatus to secrete more nectar at night and in z the early morning and for I. aggregate to secrete most O ?QOQO HbJI a Z 0 ~~~~~~~~0 of its nectar in the afternoon, in general the two =? Nspecies show broadly overlapping patterns of secreP o 0.0 tion throughout the daylight hours. Furthermore, comparisons of standing crops of nectar in flowers measured at the same time of day (Table 2) suggest that aggregatea during the day, measured by bagging flowers for these two species make nectar available to humming2-h periods during the day, as well as during the entire day- birds at comparable rates throughout the day. Obserlight period (0600-2000). overnight (2000-0600). and for 24 hummingbirds foraging in mixed species h. Values are means of accumulated nectar minus amounts vation of stands suggested that they visited flower species in the in control flowers measured at the time of bagging. Nectar order and proportions that they encountered them concentrations as percent sucrose are also given. (Kodric-Brown and Brown 1978). An unexpected result of this study was the discovto the hummingbirds should be similar, particularly ery of an apparently mimetic population of hummingwhen the flowers have similar densities and spatial bird-pollinated flowers. Whereas the Lobelia cardiarrangements on inflorescence. WhereI. <ggrgata, nalis on our study area conformed in other respects P. barbatuls, and C. integra bloomed together on the to the convergent pattern of the other hummingbird primary study area, they secreted sucrose at very sim- flowers, they secreted no nectar (Table 1). Although ilar rates (Table 3). Ipomopsis aggregate and P. bar- we bagged large numbers of plants in two different batuls were particularly similar in floral morphology summers, we were unable to obtain nectar from a sinand display, and foraging hummingbirds often visited gle flower. Plants transplanted to our garden near Tucthese species indiscriminately in the sequence and son bloomed profusely, but also secreted no nectar. proportions in which the flowers were encountered We observed several hummingbirds visiting these (Kodric-Brown and Brown 1978). It is perhaps addi- flowers, both in their natural habitat and in our garden, tional evidence of convergence that these two species and we found pollen of L. (cardinalis on birds captured provided nectar to the hummingbirds at similar rates, in the immediate vicinity of this population in its nateven though they differed in floral lifespan and hence ural habitat. in the total amount of sucrose secreted by a flower Distribution o( pollen during its life. What matters to the hummingbirds is the rate at which sucrose can be harvested, and we Animal pollinators were required for fruit and seed take the near identity of secretion rates (Table 3) and set in these flowers, and hummingbirds were the only quantities of available nectar per flower (Kodricimportant pollinators. Plants of Ipoinopsis aggregate Brown and Brown 1978) as evidence thatgthes species and P. barbatus. which were bagged to exclude all have converged to offer similar rewards at least when animal visitors, produced as many inflorescences and they bloom together in mixed stands. flowers as unbagged control plants, but set less than It is possible that flowers which are otherwise sim- 10%,as many seeds (Table 4). Very few insects were ilar in morphology and nectar rewards might secrete seen visiting the flowers. Butterflies (Papilio glaucius), nectar at different times of day. We have data only for hawkmoths (Celerio lineata), and small solitary bees 4- o --40 - o Z> 0 October 1979 1029 TEMPERATE HUMMINGBIRD FLOWERS 5. Distribution of pollen on three species of hummingbirds on the study area in 1975. Values are mean numbers of pollen grains per bird, counting only those birds with pollen on some part of the head. Note that Pensternon and Castilleja pollen is transported primarily on the crown and bill, whereas Ipomopsis pollen is carried mainly on the chin. TABLE Hummingbird species Selasphorus rufus (N =63) Pollen species Penstemon harbatus Castelleja sp. Ipomopsis aggregata Pollen location Bill Crown Sides Chin 34.4 74.2 29.5 28.2 4.6 14.0 5.9 5.9 0.8 1.0 9.1 18.9 Archilochus alexandri (N =2) Selasphorus platycercus (N 9) Penstemon Other harbatus 4.2 2.7 1.6 2.0 Castilleja sp.* Ipomopsis aggregata 5.0 11.3 2.6 3.3 0.1 0.4 1.3 4.0 51.9 79.4 19.4 17.9 Penstemon Other harbatus 2.6 3.1 0.8 4.5 15.0 77.5 16.0 3.0 Castilleja sp.* Ipomopsis aggregata Other 2.5 21.5 8.5 0.0 0.0 0.0 0.0 0.0 0.0 1.0 2.0 0.0 * Most of this pollen undoubtedly is from Castilleja integra, the most abundant species of Castilleja where these birds were captured. Pollen from different species of Castilleja could not be distinguished. visited flowers in a manner that might have resulted in some pollination, but these accounted for much less than 5% of all animal visits which we observed. Bumblebees (Bomnbussp.) robbed nectar from flowers of I. aggregate and P. barbatus in some localities, but they cut through the corolla at the base without contacting the anthers and stigma. We observed literally hundreds of hummingbirds foraging at thousands of flowers on the study area, and we recorded hummingbird visits to all flower species (Table 1). At least two species of hummingbirds visited each flower species, and we are confident that resident hummingbirds visited all flower species within their territories. In addition to territorial residents, nonterritorial individuals moved through the study area, sneaking into territories and visiting flowers defended by resident birds (Kodric-Brown and Brown 1978). From data on numbers of flowers and amounts of nectar per flower on hummingbird territories, it is clear that hummingbirds were the only animals which re- moved significant amounts of nectar from I. aggregate and P. barbatiis on our primary study area (KodricBrown and Brown 1978). Some coexisting flower species differed conspicuously in the placement of anthers and stigma (Table 1). This suggested that flowers might use different parts of the hummingbird head to transport pollen. This was confirmed by analysis of pollen collected from hummingbirds (Table 5). Penstemon barbatuts and C. integra, which have dorsally located reproductive structures, placed most of their pollen on the crown and top of the bill, whereas I. aggregate, which has ventrally located anthers and stigma, deposited its pollen primarily on the chin. The distribution of pollen of P. barbatus and C. integra might have been more segregated than is apparent from our data. Pollen of both species was white and could not be distinguished with the naked eye. In the San Francisco peaks about 200 km west of our study site, Castille a conflisa bloomed together with P. barbatus and had a distinc- 6. Distribution of pollen of different plant species on stigmas of three species of hummingbird-pollinated flowers on the study area in 1975. Note that half or more of the stigmas have some grains of heterospecific pollen and about 20% of all stigmas had more heterospecific than conspecific pollen. TABLE Stigmas of With pollen of Penstetnon barbatus Castilleja sp.* Ip(mopsis aggregate Other With some heterospecific pollen Penstemnon barbatist (N = 111) Castilleja integra (N 75) Ipomopsis aggregate (N 80) Number 107 28 4 36 (96.4) (25.2) (3.6) (32.4) Number 19 65 34 18 (%) (25.3) (86.7) (45.3) (24.0) Number 21 18 77 3 (S) (26.3) (22.5) (96.3) (3.8) 55 (49.5) 48 (64.0) 46 (57.5) (Cc) With more grains of heterospecific than nonspecific pollen 21 (18.9) 17 (22.7) 18 (22.5) * Most of this pollen undoubtedly is from Castilleja integra, but pollen from different species of Castilleja could not be distinguished. t Quantity of P. harbatus pollen on nonspecific stigmas is underestimated because grains of this species collapse. and are difficult to detect once they have germinated on compatible stigmas. 1030 JAMES H. BROWN AND ASTRID KODRIC-BROWN Ecology, Vol. 60, No. 5 tive yellow-orange pollen; its pollen was carried by TABLE 7. Means and standarddeviations of body weights and bill lengths of the three common species of humminghummingbirds at the base of the bill and the front of birds on the study area. Sample sizes are given in parenthe crown, farther forward on the head than most poltheses. len of P. barbatus. Despite displacement of reproductive structures Bill length Body weight* (mm) (g) among some species so that pollen tends to be trans- Species ported on different parts of the bird, much interspe- Selasphorus rufus cific pollen transfer occurred. About half of all flower 15.7 ? 0.6 3.48 + 0.41 (118) (13) stigmas sampled had some heterospecific pollen, and almost 20W had more heterospecific than nonspecific 16.7 ? 0.8 3.45 + 0.35 (97) (13) pollen (Table 6). This is not surprising in view of the indiscriminate foraging of hummingbirds among these Selasphorus platycercus flower species. Of 119 hummingbirds sampled, the av16.9 ? 0.7 3.44 + 0.32 (12) erage bird was carrying 3.2 species, and 34% of the (39) birds had four or more species of pollen. Since pollen 17.1 ? 1.1 3.76 + 0.32 9$ (31) (21) of all species is generally similar in size, shape, and consistency, competition for pollen transport is poten- Archilochus alexandri tially severe and appears to favor coexisting species 17.8 ? 0.5 ? 0.21 dSd3.12 (17) which deposit pollen on different parts of the bird. The (15) 20.2 + 0.6 best evidence for character displacement in placement 9 $? 3.34 + 0.21 (3) (16) of reproductive structures comes from I. aggregate. Ipomopsis and the closely related genus Gilia have * Datafrom sources indicatedin Kodric-Brownand Brown basically radially symmetrical flowers (Grant and (1978, Table 3). t Length of exposed culmen measuredon birds captured Grant 1965), but some populations have secondarily in this study. evolved either dorsal or ventral placement of anthers and stigmas. Ventral orientation in our population of I. aggregate is achieved by downward curvature of GENERAL DISCUSSION the style and differential exertion of the lower anthers. Convergence, competition, and This condition contrasts with the symmetrical oriencharacter displacement tation of reproductive structures in Ipomopsis tenuiWe have documented convergence of the hummingtuba (Brown and Kodric-Brown, personal observations). a hawkmoth-pollinated form which occurs bird-pollinated flowers on our study area. Several cotogether and hybridizes with I. aggregate in the existing species are convergent not only in size, shape, mountains of Arizona. Ventral displacement of repro- and color, but also in nectar rewards and arrangement ductive structures in I. aggregate causes pollen to be of flowers on the inflorescence. The convergence is transported primarily on the chin of hummingbirds and particularly marked in those species which bloom toreduces competition with coexisting P. barbatus and gether in mixed stands. The similar appearance and Castilleja species which place their pollen on the bill nectar rewards result in generally nonselective foraging by all of the hummingbird species on the study and crown (Table 5). Although we quantified pollen loads collected from area on all of the flower species. These hummingbirds many hummingbirds and flower stigmas, we could de- are all extremely similar in body size and bill length tect no systematic relationship between species com- (Table 7; Kodric-Brown and Brown 1978). Undoubtposition of these pollen samples and species compo- edly floral convergence increases the probability that a flower will be visited by a hummingbird and insures sition of the flowering plants in the immediate vicinity. There was a general correspondence on a large scale; that any hummingbird can serve as a pollinator, but it also increases the likelihood and intensity of interfor example, hummingbirds captured long distances from the nearest flowers of a particular species rarely specific competition for pollinators. A rapidly increasing literature presents evidence of carried that kind of pollen. However, pollen loads from hummingbirds or stigmas collected in a mixed competition for pollinators among coexisting plant species stand were highly variable and did not corre- species (e.g., Robertson 1895, Free 1968, 1970, Grant spond closely to the species composition of flowers in and Grant 1968, Levin 1968, 1972, Macior 1970, Mosthe stand. We suspect that this is because pollen carry- quin 1971, Heithaus 1974, Heinrich 1975, 1976, Stiles over from one plant to the next is limited, so that 1975, Lack 1976, Feinsinger 1978, Waser 1978). pollen loads reflect the pattern of immediately pre- Much of the evidence consists of: (1) character displacement in habitat or flowering phonology so that vious pollinator visits and these are highly stochastic. Detailed investigations of pollinator movements and the same kinds of pollinators are shared but not utipollen dispersal in stands of varying species compo- lized simultaneously, (2) divergence in floral characteristics which promote species-specific pollinator forsition clearly are warranted. October 1979 TEMPERATE HUMMINGBIRD FLOWERS 1031 aging, and (3) character displacement in orientation of be limited by energy and/or other resources, but not anthers and stigma so that pollen is transported on by availability of pollen or pollinators. However, male different parts of the body of shared pollinators. Most function (pollen dispersal) should be limited by comstudies infer the role of competition from circumstan- petition to fertilize the limited number of ovules that tial evidence and do not attempt to describe the un- will ultimately develop into seeds, and hence should derlying mechanisms or to quantify their effects on be limited by availability of pollinators. If this view is individual fitness and population size. The problem of correct, it suggests that virtually all competition estimating the magnitude and mechanism of competi- among plants for pollinators must occur through limtion for pollinators is complicated by the fact that itation of male function. It also provides another reaplants do not compete for pollinators in the same way son why competition for pollinators may affect plant that animal and plant populations compete for most fitness and result in coevolution without significantly other limited resources. Unlike competition for light, affecting population size. The only reliable data that water, and nutrients among plants or for food and we have bearing on Charnov's hypothesis are for Ipospace among animals, competition for pollinators does mopsis aggregate. This species set 56% of its flowers not use up the limited resource and make it absolutely as fruit and an average of six seeds per fruit (Table 4), unavailable to other individuals. In fact outcrossing is but 47 (59%) of 80 stigmas had six or more nonspecific accomplished only because the potentially limited pol- pollen grains. The latter probably underestimates the linators are shared at least among some intraspecific real situation, because we could not see all of the stigcompetitors. A consequence of this difference in the matic surface when counting grains and we may have mechanism of competition is that complete interspe- collected some stigmas before the plants had been cific overlap in utilization of pollinators need not result completely pollinated. On the other hand, some of the in competitive exclusion or even significant reductions grains may have been incompatible pollen from the in population density among competing plant species. same plant. The observation that seveval stigmas had We suspect that coexistence among plant species, par- more than 40 grains suggests that interference comticularly among distantly related taxa such as the hum- petition for space on the stigmatic surface rarely, if mingbird-pollinated flowers studied here, depends pri- ever, limits seed set. While these data are not definimarily on subdivision of edaphic requirements. While tive and experimental tests of Charnov's hypothesis competition for pollinators may not be sufficiently in- are urgently needed, we suspect that these plants comtense to have large, direct effects on population size pete for pollinators primarily, if not exclusively, and coexistence, it apparently reduces fitness suffi- through male function. If this is true, competition ciently to act as an important selective agent in plant- among species for shared pollinators is probably a pollinator coevolution, and thereby to influence the weak force in influencing plant-community structure structure of plant communities. in ecological time, although it can lead to significant Coexisting plant species might compete for shared adaptive change over evolutionary time. pollinators in two ways. One is by interfering with the Interspecific interference with pollen transport aptransfer of pollen to nonspecific stigmas. Limited stig- pears to be important in the coevolution of commumatic surfaces may be occluded with heterospecific nities of plants and pollinators which share each othpollen, preventing fertilization of female gametes. er's services. Significant quantities of pollen are Also pollen deposited on heterospecific stigmas rep- actually deposited on heterospecific stigmas. Flower resents loss of male gametes capable of fertilizing species have evolved to reduce this form of competiovules. Coexisting plant species might also compete tion by character displacement in orientation of the simply to attract pollinators. Even in the absence of anthers and stigmas so that pollen is transported on interference with pollen transport, it would be advan- different parts of the bird (see also Grant and Grant tageous to have a species-specific pollinator, which 1968). Apparently once interference competition would visit more nonspecific flowers and produce among these flower species has been reduced to some more outcrossing than a nonselective one. Plants tolerable level, the advantages of converging to use potentially can avoid both interference with pollen the same pollinators outweigh the disadvantages of transport and exploitative competition for shared pol- diverging to coevolve species-specific associations. In linators by coevolving specific plant-pollinator asso- this respect temperate communities of hummingbirds ciations, either directly or indirectly by flowering at and flowers differ from two other plant-pollinator asdifferent times and in different places. Interference sociations that have been well studied. Not only are with pollen transport potentially can be avoided by temperate hummingbird-pollinated flowers convercharacter displacement in location of reproductive gent, but also the hummingbird species which utilize structures, but plants which diverge only in this re- them are strikingly similar in body size and bill length spect might still compete to attract shared pollinators. (Table 7). Tropical communities of hummingbird flowRecently Charnov (in press) has suggested that in ers typically vary greatly in size, shape, and nectar outcrossed, hermaphroditic (with perfect flowers) rewards and are visited selectively by hummingbird plants such as these, female function (seed set) should species which differ in body size and in length and 1032 JAMES H. BROWN AND ASTRID KODRIC-BROWN pomopsts aggregate ,--' Archi/ochus co/ubrys ? ftL1 Archi/ochus Ecology, Vol. 60, No. 5 alexandri *enslemon arba/us . Calyple0cosine =3 Calypte y Se/osphorus * - ' Ste//u/n .- - .b p/otycercus Se/osphorus ru/us Se/osphorus ,s+ - sos/n calliope E1 .r" Coste/le/a in/egro Coste//eta austromontona Lan/cern arizoni-co tri'terno:to ~~~~~~~~~~~~~~~~~~~~~Aquil Si/ene Inciniato EchlOnocereus trig/och/dioltus Lobe/ia cardinal/s FiG. 3. Approximate geographic ranges of the eight hummingbird species which are widely distributed north of Mexico (left), and of the nine flower species which coexist on our study site (0) in eastern Arizona (right). Note that there is no correspondence between the distributions of particular hummingbird and plant species. shape of the bill (Snow and Snow 1972, Stiles 1975, Feinsinger 1976, Wolf et al. 1976, Feinsinger and Colwell 1978). Temperate bumblebee-pollinated flowers are diverse in morphology and color, and individual species tend to be visited specifically by bee species with appropriate tongue lengths and by individual bees that have learned to forage efficiently from them (Heinrich 1975, 1976a, 1976b, Inouye 1976). Why are temperate hummingbird-plant communities characterized by convergence and nonspecific plant-pollinator associations? We suggest that there is no simple answer to this question. Because competition for pollinators need not result either in divergence in utilization or in competitive exclusion, sharing of pollinators represents only a relative disadvantage which can be outweighed by net benefits of convergence. The situation is complicated because the optimal strategy of the plants depends not only on inherent characteristics of their pollinators, but also on the capacity of these resources to coevolve. The following characteristics of temperate hummingbirds and hummingbird-pollinated plants all appear to be important in the coevolutionary convergence in community structure: 1) Geographic ranges of individual hummingbird and plant species appear to be independent of each oth- er (Fig. 3). Temperate hummingbirds are migratory, and their extensive breeding, wintering, and migratory ranges include many flower species. Ranges of hummingbird flowers appear to be limited primarily by edaphic conditions and competition from nonhummingbird-pollinated plant species. Thus it is advantageous for any hummingbird species to recognize and forage from any sufficiently rewarding flowers within its range, and it is likewise advantageous for hummingbird-pollinated plants over all of western North America to employ similar signals and rewards to attract whatever hummingbirds are locally available (Grant 1966). 2) Temperate hummingbirds tend to be intra- and interspecifically territorial (Pitelka 1942, Armitage 1955, Cody 1968, Dunford and Dunford 1972, Stiles 1973, Gas et al. 1976, Yeaton and Laughrin 1976, Kodric-Brown and Brown 1978), so that often only one resident avian nectarivore is reliably available to serve as a pollinator within a local area. This makes it highly advantageous for all bird-pollinated plant species to be able to share the same individual pollinator. 3) Abundance and species composition of hummingbird-pollinated plants vary greatly between local areas and between seasons (Table 3; Kodric-Brown and Brown 1978). This variation in the competitive TEMPERATE HUMMINGBIRD FLOWERS October 1979 environment means that only some local populations in some seasons may compete intensely for pollinators. 4) Plants which usually grow at low densities potentially have much to gain from hummingbird pollination, because the birds' high-energy requirements and vagility result in foraging patterns which promote long-distance pollen transport. However, these rare species may have difficulty attracting and sustaining avian pollinators unless they share them with other plant species which will supply some of the foraging and maintenance costs. 5) Because hummingbirds hover while foraging, their position relative to flowers is highly predictable, and it is possible to avoid much interference competition for pollen transport by displacement in orientation of anthers and stigmas. Once such interference is reduced, deleterious effects of competing to attract pollinators may be small. The convergent structure of temperate communities of hummingbirds and flowers is a stable condition which has evolved because of mutual advantages to both plants and pollinators. The combination of factors which selects for this convergence in temperate hummingbird-plant associations does not occur in either tropical hummingbird-flower or temperate bumblebee-flower associations. Tropical hummingbirds are nonmigratory, and some species have different strategies for using and defending space which permit local coexistence (Feinsinger and Colwell 1978). At least one of these strategies, "traplining," appears to represent a coevolutionary adaptation to forage from and pollinate flower species which occur at low densities. In addition there probably is less year-toyear variation in local availability of flowers in most tropical habitats. Bumblebees are neither migratory nor territorial. Since several species and many individuals may have broadly overlapping foraging areas, it is possible for coexisting flower species to develop species-specific pollinator relationships with particular bumblebee species and individuals (Heinrich 1975, 1976a, 1976b, Inouye 1976). Also, because bees are relatively small, have well-developed pollen-gathering and grooming behavior, and crawl into flowers, it may be difficult for bumblebee-pollinated flowers to avoid interference competition by orienting reproductive structures to use different parts of the bee to transport pollen. Nectar secretion and mimicry Hummingbird flowers attract pollinators because the birds find it profitable to forage for the sugars contained in floral nectar. Thus it would seem that morphological convergence of flower species to attract and use the same pollinators must be accompanied by coevolution of convergent nectar rewards. Otherwise, birds would discriminate against less rewarding 1033 species and select for distinctive signals by which the more productive species would advertise their superior rewards. The virtually identical secretion rates of the three species which bloomed together profusely at the primary study site (Table 2) indicate that these coexisting, morphologically convergent flower species apparently have also converged to offer similar nectar rewards. Nectar secretion rates should vary in response to foraging and maintenance costs of their pollinators. Thus bird-pollinated flowers typically secrete more sugars than coexisting or closely related insectpollinated species (Heinrich and Raven 1972, Brown et al. 1978). We would also expect variation in secretion rates among hummingbird-pollinated flowers as, for example, colder climate increased the metabolic cost of thermoregulation, or more sparsely distributed flowers required the pollinator to fly greater distances. Our data indicate significant intra- and interspecific variation in nectar secretion rates between local areas (Table 1). Although data are too few to make a rigorous case, it is our impression that secretion rates increase with increasing elevation (colder climate) and decreasing flower density, as we would expect from increasing pollinator costs. The population of Lobelia cardinalis on our study area appears to be mimetic. Flowers of these plants do not secrete nectar to reward their pollinators, but instead, obtain sufficient hummingbird visits because of their resemblance to other hummingbird-pollinated flowers which produce significant quantities of nectar. This population of L. cardinalis possesses several attributes which are consistent with its apparent role as a mimic: (1) It is rare. We know of only one population, numbering perhaps 100 plants, on our study area. (2) It closely resembles I. aggregate and P. barbatus, which are probably the most important models because they are common in adjacent habitats. There is close resemblance among these species not only in floral color, size, and shape, but also in arrangement of flowers on loose, spike-like inflorescences. (3) This population of L. cardinalis blooms in late August and September when most of the hummingbirds present are juvenile migrant Selasphorus rufus. These birds should be generally inexperienced and, when they first arrive from their breeding grounds in the northern Rocky Mountains and Coast Ranges, totally unfamiliar with the local flora. (4) Lobelia cardinalis has a very large stigmatic surface. Not only is the receptive surface of the stigma large, but it is densely covered with long sticky papillae which serve further to increase surface area and to which many pollen grains adhere. Such a large surface may be advantageous in avoiding competition for space on the stigma from pollen grains of more numerous model species. Of 16 Lobellia stigmas, 8 had heterospecific pollen (primarily of P. barbatus), but all 16 had some L. cardinalis pollen and many had hundreds of grains. Lobelia cardinalis is probably the most widely dis- 1034 JAMES H. BROWN AND ASTRID KODRIC-BROWN tributed hummingbird-pollinated flower in North America, and most populations probably are not mimetic. We have discovered (Brown and KodricBrown, personal observation) one other population of this species (near Montezuma Well National Monument about 250 km west of our study area) which appears to secrete no nectar. On the other hand, Baker (1975) records values of nectar concentration for a population grown in Berkeley, California (original source not indicated). We have obtained (Brown and Kodric-Brown, personal observation) amounts and concentrations of nectar typical of other hummingbird flowers from a population of L. cardinalis in the Chiricahua Mountains of southeastern Arizona, about 200 km south of our study site. Presumably the mimetic populations save sufficient energy by not secreting nectar to compensate for visits lost by not rewarding the pollinators. Such a strategy should be advantageous only for rare flowers which occur where there are sufficient inexperienced pollinators and nectarproducing model flowers. A number of other flower species, particularly orchids, appear to be mimetic and to achieve pollination by deceit without offering rewards (van der Pijl and Dodson 1967, Proctor and Yeo 1972). Our observations on hummingbird-pollinated flowers suggest that analogs to both Mullerian and Batesian mimicry may be found in plant-pollinator interactions (also see Grant 1966, Macior 1974, Wiens 1978). In some cases flower species may find it mutually beneficial to resemble each other and be treated identically by shared pollinators. Then they should converge to present similar signals and attractants in much the same way that Mullerian mimics benefit by adopting similar signals to advertise their distastefulness to shared predators. If a plant species is sufficiently rare, it can adopt an alternative strategy, dispense with nectar secretion, and parasitize the coevolved relationship between more common, nectar-producing flowers and their pollinators. In much the same way, Batesian mimics take advantage of the warning signals of more common distasteful species to avoid predators. Wickler (1968) describes sets of coexisting insect species which have superficially similar, convergent morphologies and color patterns, and which consist of a mixture of Mullerian and Batesian mimics. Similarly the present community of hummingbird-pollinated flowers might be viewed as consisting of nine convergent species, eight "Mullerian" mimics and one "Batesian" mimic. ACKNOWLEDGMENTS We thank R. Vestal and C. L. Smith for assistance with pollen counts and data analysis, and L. Arnow for checking flower identifications.A. C. Gibson, D. Inouye and P. Feinsinger criticallyreviewed the manuscript.Kevin and Karen Brown provided help and companionshipin the field. Numerousstudentsand colleagues suppliedencouragementand critical discussion, but W. M. Schaffer and N. M. Waser Ecology, Vol. 60, No. 5 deserve special mention. The work was supported in part by National Science Foundation Grants GB 39260 and DEB 7609499. LITERATURE CITED Armitage, K. 1955. Territorial behavior in fall migrant Rufous hummingbirds. Condor 57:239-240. Baker, H. G. 1975. Sugar concentrations in nectars from hummingbird flowers. Biotropica 7:37-41. Brown, J. H., W. A. Calder, and A. Kodric-Brown. 1978. Correlates and consequences of body size in nectar-feeding birds. American Zoologist 18:687-700. Charnov, E. L. In press. Simultaneous hermaphroditism: sperm competition and the cost of sex. Proceedings of the National Academy of Sciences of the United States of America. Cody, M. L. 1968. Interspecific territoriality among hummingbird species. Condor 70:270-271. Dunford, C., and E. Dunford. 1972. Interspecific aggression of resident Broad-tailed and migrant Rufous hummingbirds. Condor 74:479. Feinsinger, P. 1976. Organization of a tropical guild of nectarivorous birds. Ecological Monographs 46:257-291. 1978. Ecological interactions between plants and hummingbirds in a tropical successional community. Ecological Monographs 48:269-287. Feinsinger, P., and R. K. Colwell. 1978. Community organization among neotropical nectar-feeding birds. American Zoologist 18:779-796. Free, J. B. 1968. Dandelion as a competitor to fruit trees for bee visits. Journal of Applied Ecology 5:169-178. . 1970. Insect pollination of crops. Academic Press, London, England. Gas, C. L., G. Angeher, and J. Centa. 1976. Regulation of food supply by feeding territoriality in the Rufous hummingbirds. Canadian Journal of Zoology 54:2046-2054. Grant, K. A. 1966. A hypothesis concerning the prevalence of red coloration in California hummingbird flowers. American Naturalist 100:85-98. Grant, K. A., and V. Grant. 1968. Hummingbirds and their flowers. Columbia University Press, New York, USA. Grant, V., and K. A. Grant. 1965. Flower pollination in the Phlox family. Columbia University Press, New York, USA. Hainsworth, F. R., and L. L. Wolf. 1972. Crop volume, nectar concentration, and hummingbird energetics. Comparative Biochemistry and Physiology 42:359-366. . 1976. Nectar characteristics and food selection by hummingbirds. Oecologia 25:101-113. Heinrich, B. 1975. Bee flowers: a hypothesis on flower variety and blooming times. Evolution 29:325-334. . 1976a. The foraging specializations of individual bumblebees. Ecological Monographs 46: 105-128. . 1976b. Resource partitioning among some ensocial insects: bumblebees. Ecology 57:874-889. Heinrich, B., and P. H. Raven. 1972. Energetics and pollination ecology. Science 176:597-602. Heithaus, E. R. 1974. The role of plant-pollinator interactions in determining community structure. Annals of the Missouri Botanical Garden 61:675-691. Inouye, D. W. 1976. Resource partitioning and community structure: a study of bumblebees in the Colorado Rocky Mountains. Dissertation. University of North Carolina, Chapel Hill, North Carolina, USA. Kodric-Brown, A., and J. H. Brown. 1978. Influence of economics, interspecific competition, and sexual dimorphism on territoriality of migrant Rufous hummingbirds. Ecology 59:285-296. Lack, A. 1976. Competition for pollinators and evolution in Centaurea. New Phytologist 77:787-792. October 1979 TEMPERATE HUMMINGBIRD FLOWERS Levin, D. A. 1968. Effect of corolla color and outline on interspecific pollen flow in Phlox. Evolution 23:444-445. . 1972. Competition for pollinator service: a stimulus for the evolution of autogamy. Evolution 26:668-669. Levin, D. A., and W. W. Anderson. 1970. Competition for pollinators between simultaneously flowering plant species. American Naturalist 104:455-467. Macior, L. W. 1970. The pollination ecology of Pedicularis in Colorado. American Journal of Botany 57:716-728. . 1974. Behavioral aspects of coadaptations between flowers and insect pollinators. Annals of the Missouri Botanical Garden 61:760-769. Mosquin, T. 1971. Competition for pollinators as a stimulus for the evolution of flowering time. Oikos 22:398-402. Pitelka, F. A. 1942. Territoriality and related problems in North American hummingbirds. Condor 44:189-204. Proctor, M., and P. Yeo. 1972. The pollination of flowers. Toplinger, New York. Robertson, C. 1895. The philosophy of flower seasons, and the phaenological relations of the entomophilous flora and the anthophilous insect flora. American Naturalist 29: 97-117. Snow, B. K., and D. W. Snow. 1972. Feeding niches of 1035 hummingbirds in a Trinidad valley. Journal of Animal Ecology 41:471-485. Stiles, F. G. 1973. Food supply and annual cycle of the Anna hummingbird. University of California Publications in Zoology 97: 1-1 10. 1975. Ecology, flowering phenology, and hummingbird pollination of some Costa Rican Heliconia species. Ecology 56:285-301. van der Pijl, L., and C. H. Dodson. 1967. Orchid Flowers. University of Miami Press, Miami, Florida, USA. Waser, N. M. 1978. Competition for pollination and sequential flowering in two Colorado. wild flowers. Ecology 59: 934-944. Wickler, W. 1968. Mimicry in plants and animals. World University Library, London. Wiens, D. 1978. Mimicry in plants. Evolutionary Biology 11:365-403. Wolf, L. L., F. G. Stiles, and F. R. Hainsworth. 1976. The ecological organization of a tropical highland hummingbird community. Journal of Animal Ecology 45:349-379. Yeaton, R. I., and L. Laughrin. 1976. Fall resource division in Santa Cruz Island hummingbirds. Wilson Bulletin 88: 272-279.