SEARCH

SEARCH BY CITATION

Keywords:

  • comparative methods;
  • correlated evolution;
  • hummingbirds;
  • independent contrasts;
  • nectar concentration;
  • nectar volume;
  • phylogeny;
  • sugar production

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We tested whether phylogeny, flower size and/or altitude were significant predictors of interspecific variation in nectar production of hummingbird-visited plants in an assembled database (289 species, in 22 orders, 56 families and 131 genera). Although the study is focused on hummingbird-pollinated plants (241 plant species), plants with different pollinator syndromes (48 species) are also included in the analyses. Nectar volume secreted in a given time period (usually 24 h) by a given flower, its sugar concentration and corolla length were compiled mainly from the literature. Altitude was also obtained from the original references. Sugar production was computed basically as the product of nectar secretion and sugar concentration, and expressed on a per 24-h basis. All nectar traits and corolla length (all log transformed), as well as altitude, showed statistically significant phylogenetic signal. Both nonphylogenetic and phylogenetically informed (independent contrasts) analyses indicated a highly significant positive correlation between corolla length and both nectar volume and sugar production. In addition, altitude (which is partially a surrogate for temperature) was significantly negatively correlated with both sugar concentration and production. Possible reasons for coadaptation of nectar production and sugar production with corolla length are discussed.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Nectar is the most ubiquitous floral reward among animal-pollinated angiosperms (Cruden et al., 1983; Simpson & Neff, 1983), and its provision within flowers influences components of pollination visitation, including plant choice, foraging time and foraging movements (e.g. Pyke, 1984; Zimmerman, 1988; Mitchell, 1994). Increased visitation rates might prove beneficial to plants in two ways: (1) the probabilities of pollen import and export may increase with visitation rates (Zimmerman, 1988; Mitchell, 1994); and (2) the plant can become selective as to which pollen grains fertilize the ovules, leading to higher seed quality (e.g. Mulcahy, 1979; Stanton et al., 1986; Stephenson et al., 1988; but see Klinkhamer et al., 1994; Mitchell, 1994). However, greater allocation of resources to increasing nectar production involves an energy investment (Pleasants & Chaplin, 1983; Southwick, 1984), which can entail a reproductive cost (Pyke, 1991; Ordano & Ornelas, 2005).

Hummingbird-visited flowers, which produce relatively large amounts of easily collected nectar, are particularly appropriate for a detailed quantitative analysis of interspecific variation in nectar production rates (hereafter NPR). The quantity of nectar varies widely among species (Cruden et al., 1983; Pleasants, 1983); standing crops have commonly between 0.1 and 10 μL, but nectar accumulation can reach > 650 μL in flowers of some columnar cacti and agaves commonly visited by hummingbirds during daylight hours (e.g. Cruden et al., 1983). Several authors have suggested that there is considerable opportunity for adaptive divergence of NPR through pollinator-mediated selection (e.g. Schemske & Bradshaw, 1999) and, accordingly, most workers have interpreted interspecific variation in NPR in terms of adaptations of plants to their pollinators (reviewed by Nicolson & Fleming, 2003; Pacini et al., 2003). This may be especially true with respect to hummingbirds, because they are obligate nectar feeders that distinguish between rich and poor nectar phenotypes over space (Sutherland & Gass, 1995; Irwin & Brody, 1999; Schemske & Bradshaw, 1999) and have exceptionally high energy requirements (Brice, 1992; Altshuler et al., 2004 and references therein).

The simplest selection scenario (involving rate of pollinator visitation) would suggest certain correlations between plant traits and between plant traits and environmental factors. Flowers that have presumably coevolved with pollination by hummingbirds provide both energy and water that could be balanced simultaneously by producing nectars of appropriate sugar concentrations, depending on the ambient temperature (Baker, 1975; Calder, 1979; Pyke & Waser, 1981). This might lead one to expect higher nectar concentration at higher elevation, but simultaneously balancing evaporative water loss with attainment of energy balance by a bird. Given that the rate of nectar uptake by hummingbirds is dependent on its viscosity (Kingsolver & Daniel, 1983), which increases with decreasing temperature (Calder, 1979), it is predicted that nectar concentration in hummingbird flowers will decrease with increasing altitude to compensate for the higher viscosity at lower temperatures (Calder, 1979). Because altitude is generally negatively correlated with temperature (higher is colder), this prediction can be indirectly tested by correlating nectar concentration with altitude. No such relationships between habitat (temperature or altitude) and nectar concentration have yet been found (Hainsworth, 1973; Calder, 1979; Pyke & Waser, 1981; but see Stiles & Freeman, 1993).

During the past three decades, evolutionary biologists have developed a large body of theory and empirical work for explaining the evolution of nectar rewards, particularly NPR, in terms of species differences in the selection of pollinators, in food preferences, or in response to such ecological conditions as competition for pollinators, their population densities and environmental factors (e.g. temperature and altitude: reviewed by Baker & Baker, 1983; Cruden et al., 1983; Stiles & Freeman, 1993; Nicolson & Fleming, 2003; Pacini et al., 2003; Lotz & Schondube, 2006). Following the development of evolutionary quantitative genetics in the 1980s (e.g. Roff, 1997), the focus of studies on the evolution of NPR shifted somewhat from traditional, ecological hypothesis with simple selection scenarios (e.g. involving rate of pollinator visitation) to other hypotheses and experimental work with more elaborate and realistic scenarios to elucidate proximate factors underlying the evolution of nectar. In this period, a number of workers explored natural selection in the wild, including pollinator-mediated selection on floral traits (e.g. Harder & Cruzan, 1990; Mitchell & Shaw, 1993; Stanton & Young, 1994; Campbell, 1996; Galen & Cuba, 2001), and erected adaptive hypotheses for such traits as flower size, NPR and observed correlations among such traits. Here, we examine patterns in trait variation and covariation among plant species by use of phylogenetically based statistical methods. Our general objective was to test both adaptive and nonadaptive hypotheses related to nectar among hummingbird-visited plants from various geographic regions of the New World. We contrast the adaptationist hypothesis that the correlation of corolla length with nectar is mediated by selection on the part of larger bodied pollinators for longer flowers and higher reward against competing hypotheses based only on flower allometry, genetic correlations between these traits and evolutionary history. Specifically, a hummingbird's use of a given pool of flower species may not result from active choice but rather may simply be a consequence of the scaling of corolla length (and/or nectary size) and nectar such that longer flowers are always richer in nectar (i.e. allometric relationships) and/or constrained by genetic correlations. Alternatively, the correlation between corolla length with nectar could be explained by variation in resource availability of different habitats (i.e. altitude) in which shifts into richer habitat could have allowed production of larger flowers and more nectar, rather than mediated by hummingbirds or differences in the evolutionary history of plant species.

Our focus in this paper was on variation of NPRs (nectar volume and sugar per flower) and sugar concentration because these nectar traits are of direct significance to pollinator energetics (e.g. Heinrich & Raven, 1972; Heinrich, 1975; Feinsinger, 1978; Rathcke, 1992) and have been shown to influence pollinator visits (Schemske & Bradshaw, 1999; Bradshaw & Schemske, 2003). In addition, NPRs are the most variable of functionally related floral attributes (Cresswell, 1998) and have been shown to be heritable to some degree in some plant species (reviewed in Mitchell, 2004; see also Leiss et al., 2004). Other things being equal (e.g. the number of flowers on offer), we predict a positive correlation between corolla length and both nectar production and sugar production rate as the result of selection by larger bodied pollinators for larger flowers and more energetically rewarding nectar. Correlated evolution of these flower characteristics might occur if hummingbird preference has indeed led to the evolution of higher NPR. This is the first attempt to distinguish among putative explanations for variation and covariation in nectar characteristics, considering flower size, environment (altitude) and phylogeny.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Data

We compiled a database for hummingbird-visited plants mainly from the literature (references listed in Appendix: Table A1), with some additions of our own work. A total of 289 species, in 22 orders, 56 families and 131 genera were included in the analysis (see Appendix: Table A2). Herbs (44.6%) dominated the species sample, with lesser representation of shrubs (18.7), trees (18.3), epiphytes (13.5) and lianas (6.3). Taxonomically, the sample was dominated by the orders Lamiales (77 species), Zingiberales (31), Poales (28), Fabales (26), Gentianales (17), Ericales (16), Asparagales (15) and Asterales (13). Most common families included Bromeliaceae (29), Gesneriaceae (27), Fabaceae (26), Heliconiaceae (24), Acanthaceae (15), Scrophulariaceae (14) and Rubiaceae (13). Most of the data come from geographic areas in the USA, Mexico, Costa Rica and Brazil.

Information for most species includes altitude (metres above sea level), nectar production (μL day−1 per flower), sugar concentration (% w/w), sugar production (mg day−1 per flower) and corolla length (mm). We used nectar production and sugar production as the currency of pollinator energetics. Nectar production was calculated from the literature as the volume of nectar secreted in a given time period. We accounted for variation in how long nectar was measured by correcting all values to be per 24 hours. Total sugar production (mg sugar) was the product of nectar volume (mL) and concentration per unit volume (mg per mL). Values were first converted to mg sugar per 100 mL nectar with the table in the CRC Handbook of Chemistry and Physics (1978) according to Bolten et al. (1979), then multiplied by total nectar volume (see also Kearns & Inouye, 1993). Information on other traits influencing variation in nectar secretion patterns such as flower longevity, flower age, flower's sexual phase, floral morph, response to nectar removal and temporal patterns were rarely mentioned in the references. Nectar measurements were generally done on different flowers in which nectar accumulates over time; however, researchers sampled repeatedly the same flowers over time in some of the studies (< 10%). These two nectar-sampling methodologies might have introduced some bias in our study because some hummingbird-visited plants respond positively by replenishing nectar after repeated removal (Ordano & Ornelas, 2004 and references therein). Many plant species in our database have flowers that attract a broad spectrum of visitors, and hummingbirds sometimes visit flowers with a suite of characteristics that do not correspond to the typical floral syndrome of hummingbird pollination (Faegri & van der Pijl, 1979); overlap between floral syndromes exists in our data set. The entire data set is listed in Appendix (Table A2). It includes data generated during this study and data obtained from the literature. Species in our sample were categorized into five broad pollination groups: (1) hermit type, if mainly visited by hermit hummingbirds (Subfamily Phaethorninae, 82 species); (2) trochiline type, if mainly visited by trochiline hummingbirds (Subfamily Trochilinae, 156 species); (3) passerine type, if mainly visited by perching birds (10 species); (4) bat type, if mainly visited by bats (10 species); and (5) insect type of flower, if mainly visited by insects (23 species). Plant species were categorized by its habitat as xeric (including páramo and tropical dry forest), mesic (e.g. pine-oak forest) or hydric (tropical rain forest, cloud forest) and its elevation in 500-m intervals from sea level to above 3000 m above sea level.

Statistical analysis

Because closely related species tend to share similar trait values (i.e. exhibit ‘phylogenetic signal’sensuBlomberg et al., 2003), comparative data sets may violate the assumptions of conventional statistical methods (Felsenstein, 1985; Grafen, 1989; Martins & Garland, 1991; Purvis et al., 1994; Garland et al., 2005). We therefore analysed pairwise relations using both conventional nonphylogenetic Pearson's correlations and the equivalent with phylogenetically independent contrasts (PICs) (Felsenstein, 1985; Garland et al., 1992). Independent contrasts were calculated by use of the PDTREE program (Garland et al., 1999), as a module of Mesquite (Maddison & Maddison, 2004). Conventional statistical analyses were done using StatView and SuperAnova (Abacus Concepts, Inc., Berkeley, CA, USA), as well as PDTREE. In its simplest form, Felsenstein's (1985) method assumes that the evolution of continuous characters can be modelled as a random walk process in continuous time, i.e. Brownian motion. This provides a strong reason for logarithmic transformation of size data in comparative analyses (Purvis & Rambaut, 1995). Therefore, nectar volume, sugar production, sugar concentration and corolla length data were log10 transformed before all analyses. Because the phylogenetic tree used for analyses (see next paragraph) included many soft polytomies (indicating uncertainty about phylogenetic relationships), we bounded the degrees of freedom (d.f.) used for hypothesis testing as recommended by Purvis & Garland (1993) (see also Garland & Díaz-Uriarte, 1999).

We constructed a composite phylogeny that included all species represented in our data set based on the angiosperm higher-level topologies developed by Chase et al. (1993), Soltis et al. (2000) and Davies et al. (2004). We used these phylogenetic hypotheses to represent the backbone of the tree (deeper nodes) because their phylogenies achieve the most complete taxon sampling at higher taxonomic levels. For shallower taxonomic levels, we derived intrafamiliar and intrageneric relationships from more narrowly focused phylogenetic studies. We used the online software Phylomatic (http://www.phylodiversity.net/phylomatic/phylomatic.html) developed by Webb & Donoghue (2002) to rapidly assemble our phylogenetic tree from their mega-tree assembled ‘by hand’ in April 2004, rather than by an automated supertree algorithm (Sanderson et al., 1998). The tree created by Phylomatic used information from numerous published molecular phylogenies to create a tree containing all of the species in the data set. In the absence of detailed information on phylogenetic relationships within many of the 56 families and 131 genera found in our data set, we assumed most families and genera were monophyletic and polytomous when placed them on the tree. The resulting phylogenetic tree (Fig. 1; also see Appendix: Table A3) was used for all subsequent analyses of correlated evolution. Figure 1 illustrates the phylogeny used at the family level indicating the number of genera and number of species per family included in the study and whether they contain species pollinated by hummingbirds.

image

Figure 1.  Hypothesized phylogenetic relationships among hummingbird-visited plant species at the family level in the database. Numbers in the columns indicate number of genera per family (a), number of species per family (b), and number of species pollinated by hummingbirds per family (c). Numbers above the branches indicate the 22 orders: (1) Lamiales, (2) Solanales, (3) Gentianales, (4) Garriales, (5) Asterales, (6) Dipsicales, (7) Ericales, (8) Fabales, (9) Rosales, (10) Malpighiales, (11) Myrtales, (12) Malvales, (13) Sapindales, (14) Brassicales, (15) Saxifragales, (16) Caryophyllales, (17) Santalales, (18) Ranunculales, (19) Zingiberales, (20) Poales, (21) Asparagales and (22) Liliales.

Download figure to PowerPoint

Because the phylogenetic information came from multiple studies using different methods, consistent estimates of branch length were not available. Therefore, four different types of arbitrary branch lengths were considered: all = 1 (constant), Grafen (1989), Pagel (1992) and Nee (cited in Purvis et al., 1994, p. 299). Based on the diagnostic of Garland et al. (1992), which checks for patterns in the plots of absolute values of standardized contrasts vs. their standard deviations, constant branch lengths worked the best and were used in all analyses. Constant branch lengths correspond to a ‘punctuational’ mode of character evolution (Harvey & Pagel, 1991; Martins & Garland, 1991). The ‘punctuational’ model is simply a Brownian motion model on a phylogeny with all branch lengths set equal to unity in which changes in character states occur only associated with speciation events, i.e. change could occur in both daughter species, and is better referred to as a ‘speciational’ model (for further discussion see Díaz-Uriarte & Garland, 1996).

We used a randomization procedure to test whether each trait studied exhibited a significant tendency for related species to resemble each other (phylogenetic signal). As described in Blomberg et al. (2003) (MatLab program PHYSIG.M), this test asks whether a given phylogeny (topology and branch lengths) better fits a set of tip data when compared with the fit obtained when the data have been randomly permuted across the tips of the tree, thus destroying any phylogenetic signal that may have existed (we used 1000 randomized data sets for these tests).

Because NPR evolution is perhaps best viewed as the product of evolutionary change in multiple potentially independent traits with multiple constraints, we also conducted multiple regression analyses. However, in this case we had to ‘prune’ the tree down to include only those species for which all species have data for all variables in the analysis.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Group means (±SD) of nectar production, sugar concentration, sugar production and corolla length of all species in the data set sorted by: (a) pollinator type; (b) altitude; (c) habitat; (d) habit; and (e) flower colour are presented in Table 1. The data are clearly distorted by the inclusion of certain values. For example, high NPR values in xeric habitats are influenced by the inclusion of data from Erythrina oliviae, Pseudobombax ellipticum, agaves (Agave and Manfreda) and columnar cacti (Neobuxbaumia), whereas high NPR values in trees are influenced by the 15 Erythrina trees, P. ellipticum, Ceiba pentandra, Cordia rickseckeri and the inclusion of agaves and columnar cacti (13 species) in that category (see also Table A2).

Table 1.   Nectar and flower length data of all species in the data set sorted by (a) pollinator type, (b) altitude, (c) habitat, (d) habit and (e) flower colour.
 Nectar production (μL day−1 per flower)Sugar concentration (% w/w)Sugar production (mg day−1 per flower)Corolla length (mm)
n Mean ± SD n Mean ± SD n Mean ±SD n Mean ± SD
  1. Species in our sample were categorized into five broad pollination groups: (1) hermit type, if mainly visited by hermit hummingbirds (Subfamily Phaethorninae, 82 species); (2) trochiline type, if mainly visited by trochiline hummingbirds (Subfamily Trochilinae, 156 species); (3) passerine type, if mainly visited by perching birds (10 species); (4) bat type, if mainly visited by bats (10 species); and (5) insect type of flower, if mainly visited by insects (23 species). Plant species were categorized by its habitat as xeric (including páramo and tropical dry forest), mesic (e.g. pine-oak forest) or hydric (tropical rain forest and cloud forest), and its elevation in 500-m intervals from sea level to above 3000 m above sea level. Letters ‘a’ and ‘b’ denote differences among groups based on Tukey's HSD methods.

All data27841.6 ± 85.627223.8 ± 7.82639.4 ± 16.619732.6 ± 17.2
Pollinator type
 Trochilines15622.5 ± 33.9a15623.5 ± 6.9a1535.9 ± 8.5a10934.8 ± 37.2a
 Hermits8237.2 ± 25.0a7825.7 ± 6.4a7810.5 ± 8.4a6240.6 ± 12.8b
 Perching birds10196.9 ± 217.8b913.2 ± 2.8b930.5 ± 35.9b434.8 ± 37.2ab
 Bats10265.3 ± 247.2b916.1 ± 3.9b944.7 ± 58.9b739.3 ± 15.7ab
 Insects2319.5 ± 26.9a2026.1 ± 13.4a194.9 ± 6.7a1624.7 ± 14.5a
Altitude
 Low10137.9 ± 41.4a10026.2 ± 6.3a10010.8 ± 10.7a8034.1 ± 16.4a
 Medium8352.2 ± 112.2a7622.3 ± 8.6b7610.3 ± 24.3ab6336.2 ± 19.4a
 High4843.4 ± 97.6a4922.4 ± 6.7b488.7 ± 14.8ab2725.1 ± 13.1b
 Very high4426.3 ± 90.7a4422.4 ± 8.8b424.6 ± 11.7b2724.7 ± 13.9b
Habitat
 Xeric6276.9 ± 135.9a5622.9 ± 1.3a5414.2 ± 31.8a5032.6 ± 19.6a
 Mesic13826.9 ± 31.9b13722.9 ± 0.7a1356.6 ± 7.6b9330.1 ± 17.4a
 Wet7935.5 ± 31.5b7625.6 ± 0.6b7610.0 ± 9.3b5634.4 ± 12.7a
Habit
 Herb11629.3 ± 28.3a11025.1 ± 7.1a1098.2 ± 8.9a8234.3 ± 16.3a
 Shrub4935.8 ± 62.2a4721.9 ± 9.4ab468.1 ± 13.1a3222.3 ± 16.8b
 Tree46108.6 ± 181.0b4320.5 ± 7.2b4219.3 ± 34.5b3038.6 ± 23.0a
 Epiphyte3422.1 ± 19.8a3427.2 ± 6.1a347.2 ± 7.1a3233.3 ± 11.8a
 Vine1719.7 ± 15.9a1721.1 ± 13.1a174.1 ± 3.2a1433.9 ± 14.3a
Flower colour
 Red9736.1 ± 71.1ab9724.0 ± 7.9a968.4 ± 12.2a7932.1 ± 17.0a
 Orange1928.5 ± 51.6ab1822.1 ± 13.1a185.8 ± 7.2a1640.9 ± 26.1a
 Pink2721.5 ± 15.0b2724.5 ± 6.3a265.4 ± 3.8a2335.0 ± 16.4a
 White17103.4 ± 191.3a1425.6 ± 8.8a1425.1 ± 47.9b1527.6 ± 19.8a
 Yellow4875.6 ± 126ab4522.2 ± 6.2a4416.1 ± 20.8b4231.4 ± 14.6a
 Other1214.9 ± 17.2b1226.8 ± 4.9a124.4 ± 4.7a1125.2 ± 12.7a

Nectar production rate values ranged more widely within plant groups (Tables 2 and 3), yielding an average standard deviation of 82 μL for the 32 families in which multiple genera were sampled and 73 μL for the 19 orders in which multiple families were sampled. Among orders, the Asparagales (Agavaceae) and Caryophyllales (Cactaceae) were nectar ‘superproducers’, whereas Myrtales (e.g. Lythraceae and Onagraceae) and Lamiales (Acanthaceae, Bignoniaceae, Gesneriaceae, Lamiaceae and Scrophulariaceae) were relatively poor nectar producers (Table 2). The high NPR value for Malvales is due to the inclusion of two tree species in the Bombacaceae (C. pentandra and P. ellipticum). Relatively high NPR values (volume and sugar) are also observed among the Zingiberales (Heliconiaceae and Costaceae) and Fabales (mainly Calliandra and Erythrina species; see also Table A2). High corolla length values were observed for Amarylliaceae (Bomarea and Hippeastrum), Bignoniaceae and Loranthaceae (Gaiadendron and Psittacanthus) with moderate amounts of nectar (Table 3; see also Table A2).

Table 2.   Nectar and flower length data of all species in the data set sorted by order.
OrderNectar production (μL day−1 per flower)Sugar concentration (% w/w)Sugar production (mg day−1 per flower)Corolla length (mm)
n Mean±SD n Mean ± SD n Mean±SD n Mean ± SD
  1. Only orders with more than five species in the data set are shown.

Asparagales15156.2 ± 195.11515.3 ± 3.11524.5 ± 30.4734.0 ± 23.1
Asterales1322.1 ± 13.41320.3 ± 6.0135.2 ± 5.21036.0 ± 10.4
Caryophyllales9146.0 ± 258.4829.1 ± 7.4732.2 ± 68.71043.1 ± 17.0
Ericales1625.2 ± 35.71624.5 ± 14.1165.0 ± 4.9820.2 ± 11.5
Fabales2653.1 ± 61.92422.1 ± 7.22310.4 ± 9.11547.9 ± 25.9
Gentianales1326.4 ± 60.01320.0 ± 7.8137.2 ± 16.11126.2 ± 20.5
Lamiales7615.4 ± 14.27523.9 ± 5.7754.1 ± 3.74229.7 ± 13.0
Malvales8112.0 ± 200.1718.5 ± 3.9720.2 ± 28.3527.8 ± 7.3
Myrtales1110.4 ± 10.2921.7 ± 9.592.4 ± 1.6921.2 ± 16.8
Poales2823.9 ± 19.52827.7 ± 5.7287.6 ± 6.52633.6 ± 11.9
Santalales514.1 ± 12.6519.2 ± 1.343.4 ± 2.8350.5 ± 39.4
Solanales1017.9 ± 17.21026.9 ± 7.6106.2 ± 6.4829.0 ± 12.7
Zingiberales3163.1 ± 26.32727.2 ± 4.32719.7 ± 10.22740.2 ± 12.2
Table 3.   Nectar and flower length data of all species in the data set sorted by family.
OrderFamilyNectar production (μL day−1 per flower)Sugar concentration (% w/w)Sugar production (mg day−1 per flower)Corolla length (mm)
n Mean±SD n Mean ± SD n Mean±SD n Mean ± SD
  1. Only families with more than three species in the data set are shown.

AsparagalesAgavaceae7293.8 ± 214.9714.5 ± 2.1745.2 ± 34.2222.5 ± 10.6
Amarylliaceae637.7 ± 32.8615.2 ± 2.865.9 ± 5.0445.1 ± 25.1
AsteralesLobeliaceae1125.3 ± 11.91019.9 ± 6.7115.9 ± 5.4832.9 ± 9.2
CaryophyllalesCactaceae6216.9 ± 297.9531.4 ± 8.3455.3 ± 88.2750.9 ± 11.9
EricalesEricaceae1219.7 ± 13.71220.1 ± 3.6124.4 ± 3.2420.9 ± 16.5
FabalesFabaceae2554.1 ± 62.92422.1 ± 7.22310.4 ± 9.11448.7 ± 26.8
GentianalesRubiaceae1326.4 ± 60.01320.0 ± 7.8137.2 ± 16.11126.2 ± 20.5
LamialesAcanthaceae1520.8 ± 19.61521.5 ± 3.3154.8 ± 4.41131.9 ± 10.7
Bignoniaceae617.1 ± 9.8627.3 ± 4.565.1 ± 2.6641.5 ± 15.8
Gesneriaceae2617.1 ± 11.82625.7 ± 6.3265.1 ± 3.7930.8 ± 11.9
Lamiaceae1210.0 ± 7.31224.1 ± 4.2122.5 ± 1.7622.6 ± 7.4
Scrophulariaceae1413.0 ± 17.01322.2 ± 7.3133.1 ± 3.9924.8 ± 12.8
MalvalesMalvaceae534.3 ± 24.8519.3 ± 3.857.3 ± 6.0428.2 ± 8.3
MyrtalesLythraceae37.8 ± 6.4234.5 ± 12.423.9 ± 0.2122.9 ± 0.0
Onagraceae414.8 ± 14.4418.2 ± 6.142.4 ± 1.9428.0 ± 23.6
PoalesBromeliaceae2823.9 ± 19.52827.7 ± 5.7287.6 ± 6.52633.6 ± 11.9
SantalalesLoranthaceae514.1 ± 12.6519.2 ± 1.243.4 ± 2.8350.4 ± 39.4
SolanalesConvolvulaceae721.9 ± 18.9730.4 ± 5.177.9 ± 6.9635.5 ± 3.6
ZingiberalesCostaceae461.2 ± 22.9430.0 ± 3.4421.3 ± 9.6347.9 ± 12.5
Heliconiaceae2465.1 ± 25.92027.0 ± 4.62020.3 ± 10.42340.2 ± 11.3

Phylogenetic signal

As shown in Table 4, all traits showed statistically phylogenetic signal (all P < 0.001), with the descriptive K-statistic ranging in the region of 0.3–0.4. For all traits, the hierarchical phylogenetic tree gave a better fit to the tip data than did a star phylogeny, as indicated by lower MSE (Table 4). Therefore, the results of phylogenetically based statistical analyses should generally be more reliable. Note that transformations of the branch lengths would have yielded even better fits of the hierarchical trees (Blomberg et al., 2003), but we did not implement such transformations in the present study.

Table 4.   Diagnostic test for adequacy of branch lengths (all set equal to unity) with phylogenetically independent contrasts (Garland et al., 1992), mean squared error of tree fitted to tip data (smaller indicates better fit), K statistic indicating amount of phylogenetic signal, and significance test for presence of phylogenetic signal (Blomberg et al., 2003).
PDI file n Internal branches of zero lengthTraitPhylogenyDiagnostic correlation*MSE K P signal from randomization test
  1. *Based on this diagnostic, none shows significant lack of fit to the tip data (all P > 0.05).

NEC283 L1.PDI283175log Nectar Production (μL day−1 per flower)Star0.0370.3100.433< 0.001
Tree0.244
CON266 L1.PDI266165log Sugar Concentration (% w/w)Star−0.0130.02010.284< 0.001
Tree0.0181
SUG263 L1.PDI263163log Sugar Production (mg day−1 per flower)Star0.0040.3140.367< 0.001
Tree0.268
COR201 L1.PDI201113log Corolla Length (mm)Star−0.084296.30.296< 0.001
Tree270.0
ALT272_1.PDI272165Altitude (metres above sea level)Star−0.1248.80*1050.345< 0.001
Tree7.94*105

Trait correlations

Based on conventional statistical analyses, all pairwise trait correlations were statistically significant, with the exception of sugar concentration and nectar production (Table 5). PICs tended to indicate somewhat weaker relations among traits, and in several cases nonsignificant ones if the minimum d.f. are considered. Considering the phylogenetic results (Table 5), altitude showed a weak but statistically significant negative relation with sugar concentration and also sugar production. Nectar production and corolla length were positively related, indicating that, as might be expected, larger flowers tend to produce more nectar. We found no significant linear correlation between nectar production and sugar concentration in our data. Corolla length was not correlated with sugar concentration, but was correlated with both nectar production and sugar production.

Table 5.   Pairwise correlations of nectar production, sugar production, corolla length, sugar concentration and altitude based on nonphylogenetic and phylogenetic (independent contrasts) analyses.
  n Conventional rConventional PPIC, rNo. of zero-length branchesd.f. min P, d.f. mind.f. max P, d.f. max
  1. VOL = nectar production (μL day−1 per flower), SUG = sugar production (mg day−1 per flower), COR = corolla length (mm), CON = sugar concentration (% w/w) and ALT = altitude (metres above sea level). Nectar, sugar, corolla length and concentration data were log10 transformed. All branch lengths = 1 for phylogenetically independent contrasts (PIC). For PIC, d.f. min are (n − 2 − number of zero-length branches) and d.f. max are (n − 2) (see text; Purvis & Garland, 1993). All P values are for two-tailed tests.

log VOL vs. log COR1930.472≪ 0.0010.39110982< 0.001191< 0.001
log VOL vs. ALT263−0.228< 0.001−0.118161100> 0.22610.08
log VOL vs. log SUG2630.960≪ 0.0010.95816398≪ 0.001261≪ 0.001
log SUG vs. log COR1800.481≪ 0.0010.39010078< 0.001178< 0.001
log SUG vs. ALT253−0.289≪ 0.001−0.175155960.092510.006
log COR vs. ALT188−0.259< 0.001−0.104103830.351860.15
log CON vs. log VOL262−0.063n.s.−0.02516397n.s.260n.s.
log CON vs. log COR1820.184< 0.020.05010179n.s.180n.s.
log CON vs. log SUG2630.211< 0.0010.254163980.01261< 0.001
log CON vs. ALT256−0.252< 0.001−0.20215797< 0.052540.001

Using species values, we found that a full regression model that included sugar concentration, corolla length and altitude (all log transformed) explained a small amount of the variation in nectar production (adjusted r2 = 0.19). The P values in the regression coefficients showed that corolla length is the only statistically significant (P < 0.05) variable predicting nectar volume (Table 6). In the case of sugar concentration, the adjusted r2 (0.05) indicates a very weak overall relationship. The P values in the regression coefficients showed that corolla length and altitude are useful in predicting sugar concentration (Table 6). For sugar production, we obtained a moderate overall relationship (r2 = 0.22). The P values of the regression coefficients showed that corolla length was the most useful variable in predicting sugar production and altitude, although significant, had a smaller contribution in predicting sugar production (Table 6).

Table 6.   Nonphylogenetic and phylogenetic multiple regressions with PICs of nectar traits (nectar production, sugar concentration and sugar production) with corolla length and altitude.
Nonphylogenetic (using species values)Phylogenetic (using independent contrasts)
DependentIndependent β F P β F P 95% Confidence Intervals for β
Lower boundUpper bound
  1. VOL = nectar production (μL day−1 per flower), SUG = sugar production (mg day−1 per flower), COR = corolla length (mm), CON = sugar concentration (% w/w), and ALT = altitude (metres above sea level). Nectar, sugar, corolla length and concentration data were log10 transformed. All branch lengths = 1 and linear regressions through the origin for phylogenetically independent contrasts (PIC). Also note that no d.f. have been subtracted for polytomies. Regression coefficients are standardized. n = 174 in all cases. All P values are for two-tailed tests.

Nectar productionCON−0.091.930.1666−0.081.370.243−0.8670.222
COR0.4337.21< 0.00010.3320.470.00010.4211.072
ALT−0.143.840.0519−0.060.680.4110.0000.000
Sugar concentrationVOL−0.111.930.1666−0.091.370.243−0.0660.017
COR0.206.200.01370.0570.500.479−0.0610.129
ALT−0.185.480.02070.1282.860.0930.0000.000
Sugar productionCOR0.416.71< 0.00010.3320.710.00010.4341.098
ALT−0.1836.720.0103−0.081.360.2450.0000.000

When using PICs, we found that both nectar production and sugar production were strongly correlated with corolla length, and altitude had no significant effect in explaining NPR (nectar and sugar production) (Table 6). Moreover, contrasts of sugar concentration were not significantly correlated with nectar production, corolla length or altitude (Table 6). As data showed phylogenetic signal, results from PICs are emphasized.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Both nonphylogenetic and phylogenetic analyses of the relation between corolla length and (1) nectar production and (2) sugar production for hummingbird-visited plant species indicated statistically significant positive relations. That the relations retained statistical significance after accounting for phylogeny suggests that these traits have undergone true correlated evolution. Possible mechanisms (not mutually exclusive) are positive genetic correlations or correlational selection. In general, barring a change in shape (becoming more bulbous), flowers with longer corollas will be able to hold more nectar or will have nectary tissue of larger size, and this simple physical relation should be reflected, to some extent, in a positive genetic correlation. We are not aware of any study showing the genetic correlation between corolla length and NPR. Under controlled conditions, a genetic correlation between corolla length and NPR in hummingbird-pollinated Penstemon centranthifolius (Scrophulariaceae) was positive but not significantly different from zero (see Table 3 in Mitchell & Shaw, 1993 and Table 1 in Mitchell et al., 1998). In bumblebee-pollinated Echium vulgare (Boraginaceae), no significant correlation between NPR and measures of floral dimensions was found by Klinkhamer & van der Veen-van Wijk (1999) (see also Leiss et al., 2004). Under natural conditions, no evidences for a genetic correlation between nectar volume and corolla length in hummingbird-pollinated Ipomopsis aggregata (Polemoniaceae) was found; however, their phenotypic correlation was moderately strong and positive, suggesting the possibility of common environmental effects on both of these traits (Campbell, 1996). That no one has shown a genetic correlation may simply reflect how few studies have looked for one. The idea that the correlation of corolla length with nectar could be produced by selection on the part of larger-bodied pollinators for larger flowers and higher reward is supported by phenotypic correlations. Indeed, phenotypic correlations between aspects of flower size (inflorescence size, flower depth, flower width, petal size, corolla length and perianth area) and nectar traits have been shown for some plant species (Harder & Cruzan, 1990; Cresswell & Galen, 1991; Mitchell & Shaw, 1993; Stanton & Young, 1994; Campbell, 1996; Mitchell et al., 1998; Worley & Barrett, 2000; but see Leiss et al., 2004). Genetic correlations may either constrain or facilitate response to selection (Garland, 1994; Mitchell, 2004). Correlational selection (sensuBrodie et al., 1995) may be acting through a relationship between tube length, tongue and/or bill lengths, pollinator size and energetic demands (Heinrich & Raven, 1972). In the following sections, we contrast adaptive and nonadaptive hypotheses related to variation in NPR.

In the present study, we have emphasized results of phylogenetically informed statistical analyses. This is generally justified because all of the traits studied showed highly significant phylogenetic signal, based on the randomization test of Blomberg et al. (2003) (see Table 4). Although highly statistically significant, the amount of phylogenetic signal in the traits studied, as indicated by K statistics ranging between 0.30 and 0.43, was substantially less than if the phylogenetic trees (and assumed Brownian motion model of character evolution) provided a ‘perfect’ fit to the tip data, in which case K = 1.00. However, the present K values are similar to those that have been reported previously in a broad survey of traits (see Fig. 6 in Blomberg et al., 2003; also Rezende et al., 2004). Phylogenetic signal lower than unity can result from several things, including adaptive evolution in response to natural selection, evolution in response to sexual selection, and various types of ‘measurement error’ (Blomberg et al., 2003; Ives et al., 2007). Measurement error is likely to be fairly large for our trait data because they were collected mainly from plants growing in nature rather than under common-garden conditions (Garland & Adolph, 1994; Mitchell, 2004), by a variety of workers, using somewhat varied techniques. It is possible that the type of measurement employed in our study (accumulated nectar) poorly represents the real variation in nectar production dynamics because of among-species variation in post-secretory processes (e.g. reabsorption, evaporation, dilution; Southwick, 1984; Zimmerman, 1988; Búrquez & Corbet, 1991) and sources of unaccounted for variance among species (e.g. flower longevity and daily secretion rate; Devlin & Stephenson, 1985; Delph & Lively, 1992). Also, dynamic regulation of nectar offerings, by refilling nectar reservoirs after nectar has been removed or by readjusting the concentration of sugar as water evaporates, has been shown for several species (e.g. Castellanos et al., 2002 and references therein), and plant responses to intense nectar removal seem strongest for those pollinated by hummingbirds (reviewed by Ordano & Ornelas, 2004). The contribution of these factors to the variation in nectar traits cannot be comparatively evaluated because data on accumulated nectar is more widely and easily available for many species. In addition, the phylogenetic trees used for analyses included a large number of soft polytomies (see Appendix: Table A3) and employed arbitrary branch lengths. Therefore, the true amount of phylogenetic signal for these traits in these species is likely to be larger than indicated by the present analyses.

Nonadaptive hypotheses

Among the nonadaptive explanations for the origin of nectar is the old idea that nectar is a by-product of a ‘leaky phloem’ and a ‘sugar-excreting’ mechanism (De la Barrera & Nobel, 2004); nectar could have originated as a leakage of the phloem solution resulting from the structural weakness of developing tissue exposed to the elevated hydrostatic pressure in the phloem (for experiments contra the position that NPR is in part a ‘passive’ result of phloem and xylem characteristics see Reed et al., 1971 and Fahn, 1979). From a developmental perspective, the ‘sugar excretion’ hypothesis might also contribute to our understanding of why long-tubed flowers produce copious nectar (Table 5). On one hand, the surface area for the tubular portion of a flower changes in the same proportion as its length, e.g. a 10-cm-long flower would have a surface area four times as large as that for a 5-cm-long flower. Considering the relatively high transpiration rates of flowers (Galen et al., 1999; Chapotin et al., 2003), the amount of water loss through transpiration would be substantially higher for long-corolla flowers than for short-corolla flowers, leading to a greater accumulation of phloem-supplied solutes (mainly sugar) needing to be excreted for maintaining a relatively stable water potential in floral tissues. Corolla length may correlate with the storage capacity of nectar and particular solutes (e.g. amino acids and toxic compounds) that influence pollinator visitation. However, nectar itself does not explain how far the corolla tube can grow because it is an aqueous solution presented in glands in the flowers and often deeply concealed at the base of tubular flowers.

After nectar originated, subsequent evolution of nectar volume and its concentration might have been influenced by other factors, including the origin, size and anatomical structure of nectaries (Galetto & Bernardello, 2004 and references therein). This would then lead to a purely allometric relationship whereby scaling up the corolla tube also scales up the size of nectary tissue, and nectar characteristics can be constrained by allometric relationships between the size of the floral corolla tube and the size of the nectaries (Galetto & Bernardello, 2004). Whether or not these allometric associations are maintained by selection from pollinators or nectar robbers (Heinrich & Raven, 1972; Lara & Ornelas, 2001; Irwin et al., 2004), and/or by plant allocation or by genetic correlations (Schemske & Bradshaw, 1999; Bradshaw & Schemske, 2003), is totally unknown. However, it must be remembered that allometric relations themselves can evolve in response to selection (Garland & Carter, 1994). Given that nectar extraction rates are a positive function of nectar volume and a negative function of corolla length, hummingbirds should prefer flowers with the highest nectar volumes and the shortest corollas because longer corollas make it harder for hummingbirds to get the nectar (Hainsworth, 1973; Montgomerie, 1984). Therefore, if flowers were evolving so as to enhance their attractiveness to hummingbirds, then we might expect to find a negative evolutionary correlation between nectar production and corolla length. Conversely, we found positive correlated evolution between corolla length and nectar production (Table 5), which is consistent with nonadaptive allometric expectations.

A plant–pollinator relationship based on predictions from one or two morphological or behavioural characters may be inappropriate and additional more complex multidimensional variables are needed to provide a more accurate description of evolutionary patterns. For example, Hurlbert et al. (1996) have suggested that flexible pedicels of Impatiens capensis flowers appear to facilitate hummingbird pollination (a ‘mobile’ pollination mechanism). Although a hummingbird's efficiency to feed from flowers increases with increasing corolla diameter (Temeles, 1996) and that positive allometric relationships between corolla length and corolla entrance width have been reported for a number of plant species, some of which are hummingbird pollinated (e.g. Campbell, 1996; Temeles et al., 2002), the occurrence of inverse relationships between lengths and widths in some hummingbird-pollinated genera indicates that plant species with long flowers do not always have wider flowers (Temeles et al., 2002; see also Grant & Temeles, 1992). Furthermore, Castellanos et al. (2004) have shown experimentally in Penstemon species that hummingbirds removed almost twice as many pollen grains and efficiently extracted nectar from narrow flowers than from nonmodified flowers. They concluded that the narrow corollas of hummingbird-pollinated penstemons may function to improve the morphological fit of flowers to birds. All these evidences suggest that the positive correlation between corolla length and NPRs studied here may not result exclusively from floral allometry.

Adaptive hypotheses

We found a positive association between nectar volume and corolla length, and between sugar production and corolla length after statistically controlling for covariation related to phylogenetic inheritance (Tables 5 and 6). The strength of the relationships (ca. r = 0.4) is perhaps remarkable considering the multifactorial nature of floral-trait evolution (see Stanton, 2003). Above, we note that simple geometry and allometry should yield a positive correlation between flower size (e.g. corolla length) and amount of nectar per flower. The simplest explanation for the significant relationships between NPRs and corolla length (i.e. allometry) do not necessarily mean that hummingbirds are actively choosing flower species on the basis of such relationships. These relations could arise either as a result of selection to maintain pollinator attractiveness or simply as a consequence of proximate allocation constraints (see Harder & Cruzan, 1990). As larger flowers and higher NPRs are both likely more costly to make (e.g. Pyke, 1991; Galen et al., 1999), the correlations of corolla length and NPRs could be explained by variation in resource availability of different habitats. For instance, shifts into richer habitats could have allowed production of larger flowers and more nectar. However, differences in NPR values among habitat types (see Table 1 and Appendix: Table A2) are surely influenced by the inclusion of certain taxa inhabiting xeric conditions (a situation taken into account in our phylogenetic analyses) and the nonphylogenetic relationships between corolla length and NPRs still hold separately by habitat type (results not shown). The observed increase in corolla length and amount of sugar of hummingbird-pollinated flowers when compared with that in other flowers, along with other floral characters (e.g. Galen & Cuba, 2001; Castellanos et al., 2004; Wilson et al., 2004), is potentially associated with the evolution of hummingbird pollination. Although the relationship holds up within broad habitat type, NPRs can be very sensitive to microhabitat as well, sometimes varying several folds depending on conditions. One way to test that the association is due to selection by hummingbirds would be to analyse future data sets that also include plant species pollinated by other animals for comparison.

As predicted, we found a significant negative relation between sugar concentration and altitude (see also Stiles & Freeman, 1993) and between sugar production and altitude (Table 5), which suggests that nectar concentration in hummingbird flowers decreases to compensate for the higher viscosity at lower temperatures, and/or the result of differences in water and carbon resources, i.e. at high altitudes sugar production may be more costly. Stiles & Freeman (1993) observed that this negative relation is consistent with the fact that hummingbird-pollinated plants at high elevations are pollinated by smaller, less-energy demanding, short-billed trochiline species. However, the significant negative correlations between sugar concentration and altitude and between sugar production and altitude are removed when using multiple regression with PICs (Table 6). Although NPRs tend to decrease with increased altitude in some plant species (e.g. Cruden et al., 1983), in some other hummingbird-pollinated plant species no significant changes occurred along altitudinal gradients (e.g. Scobell & Scott, 2002). Also, it is not known whether nectar is more costly to produce and/or hummingbird-pollinated plant species are more resource limited at higher elevations than at lower elevations. Clearly, comparative studies of nectar production costs along altitudinal gradients are needed to tease apart among these hypotheses. Surprisingly, the commonly reported inverse relationship between volume and concentration in nectar was not found in our study (Table 5). Although hummingbirds choose concentrated sucrose solutions in preference to dilute ones (Van Riper, 1958; Stiles, 1976; Pyke & Waser, 1981; Roberts, 1996), experimental studies on sugar preferences have shown that they take nectar of a wide sugar concentration range (e.g. Martínez del Rio et al., 2001) and that hummingbirds do not prefer sucrose to hexose nectars, and even prefer hexose nectars at low concentrations (Fleming et al., 2004). These findings suggest that preferences for different sugar compositions are concentration dependent (see also Lotz & Schondube, 2006).

The simplest scenario for pollinator-mediated selection on NPR is directional selection for high NPR. Once secreted, nectar can be manipulated in its concentration and constituents, and even reabsorbed to ensure visits by a given pollinator (or by more than one type of pollinator) and/or to recover some of the energy cost (e.g. Búrquez & Corbet, 1991; Nicolson, 1995; Nepi et al., 2001; Ordano & Ornelas, 2004, 2005). But this scenario may be too simple (even including the cost of nectar), because plant fitness is a more complex function of pollinator behaviour. The best situation for the plant is to attract pollinators that bring conspecific pollen and remove their own pollen to carry it to conspecifics, in a pattern that produces matings of high fitness (e.g. that avoids self-pollination in many species) – all at minimal cost. What trait values achieve this, and the mechanisms of selection, may sometimes not be the ones that seem obvious at first thought (Campbell et al., 1996; Hurlbert et al., 1996; also Mitchell & Waser, 1992; Mitchell, 1993). Wilson et al. (2006) have postulated that for a shift from bee pollination toward hummingbird pollination to proceed among Penstemon species, there has to be evolution of the nectar characteristics of the plants. They emphasized that a penstemon must offer voluminous nectar, probably dilute and sucrose rich, to entrain hummingbird visitors. This first change during a shift from bee to bird pollination may be followed by changes in the floral dimensions, specifically from a globular to narrower corolla tubes that increase pollen transfer by hummingbirds (e.g. Castellanos et al., 2003, 2004). Lastly, the evolution of floral traits needs to be understood in terms of all class of pollinators; plants may exhibit specializations for the most effective pollinators and in some cases for the less effective pollinators (Aigner, 2001). Differences in effectiveness certainly can foster floral specialization to the most effective pollinator in some cases, but when adaptation to less effective pollinators requires little loss in the fitness contribution of a more effective pollinator, plants may specialize for the less effective pollinator (Aigner, 2001, 2004). The amount of nectar produced by hummingbird-pollinated plants has probably been under selection simultaneously exerted by multiple interactions with pollinating and nonpollinating fauna (Irwin et al., 2004 and references therein). The strong positive relations between nectar traits and corolla length may have evolved in response to selection by pollinators. Yet, these same (or correlated) traits that lure pollinators may also increase the susceptibility of hummingbird-pollinated plants to antagonists (Grant & Temeles, 1992; Mitchell et al., 1998; Lara & Ornelas, 2001; Castellanos et al., 2004). We found a significant positive correlation between nectar concentration and corolla length in the conventional analysis but not in the phylogenetic analysis (see Table 5). The result from the phylogenetic analysis (Table 5) is inconsistent with the hypothesis posed by Bolten & Feinsinger (1978) that flowers with short-corolla tubes have low-concentrated nectar (dilute nectar) to deter nectar robbers (see also Pyke & Waser, 1981).

In summary, our results are consistent with the hypothesis of pollinator-mediated selection for larger flowers and high NPR (other things being equal). An alternative scenario is that pollinator preference has indeed led to the evolution of higher NPR but that the target of selection has been flower size, to which NPR is positively correlated phylogenetically.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Comments and suggestions on the manuscript by Erick de la Barrera, Steve Heard, Olle Pellmyr, Nick Waser and two anonymous reviewers greatly improved the manuscript. Many thanks to Nick Waser for discussion and data interpretation that got JFO thinking, and Carlos Lara and José Luis Martínez y Pérez for assistance in data gathering. J. F. Ornelas was partially funded by a doctoral scholarship (No. 56254) from the Consejo Nacional de Ciencia y Tecnología (CONACyT-México), the Department of Ecology and Evolutionary Biology at the University of Arizona, and by the Departamento de Biología Evolutiva at the Instituto de Ecología, A.C. (Ref. 902-12-563). M. Ordano was supported with a scholarship granted by the Mexican Government through the Instituto Mexicano de Cooperación Internacional (IMEXCI) of the Secretaría de Relaciones Exteriores (SRE), and a fellowship from the Organization of American States Fellowship Program (LASPAU) of the Academic and Professional Programs for the Americas. M. E. Quintero and A. de Nova were supported with doctoral scholarships (145847 and 164478) granted by CONACyT-Mexico.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information
  • Aigner, P.A. 2001. Optimality modelling and fitness trade-offs: when should plants become pollinator specialists? Oikos 85: 177184.
  • Aigner, P.A. 2004. Floral specialization without trade-offs: optimal corolla flare in contrasting pollination environments. Ecology 85: 25602569.
  • Altshuler, D.L., Dudley, R. & McGuire, J.A. 2004. Resolution of a paradox: hummingbird flight at high elevation does not come without a cost. Proc. Natl Acad. Sci. U.S.A. 101: 1773117736.
  • Baker, H.G. 1975. Sugar concentration in nectars from hummingbird flowers. Biotropica 7: 3741.
  • Baker, H.G. & Baker, I. 1983. Floral nectar constituents in relation to pollinator type. In: Handbook of Experimental Pollination Biology (C. E. Jones & R. J. Little, eds), pp. 117141. Scientific and Academic Editions, New York.
  • Blomberg, S.P., Garland, T. Jr & Ives, A.R. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution 57: 717745.
  • Bolten, A.B. & Feinsinger, P. 1978. Why do hummingbirds flowers secrete dilute nectar? Biotropica 10: 307309.
  • Bolten, A.B., Feinsinger, P., Baker, H.G. & Baker, I. 1979. On the calculation of sugar concentration in flower nectar. Oecologia 41: 301304.
  • Bradshaw, H.D. & Schemske, D.W. 2003. Allele substitution at a flower colour locus produces a pollinator shift in monkeyflowers. Nature 426: 176178.
  • Brice, A.T. 1992. The essentiality of nectar and arthropods in the diet of the Anna's hummingbird (Calypte anna). Comp. Biochem. Physiol. A 101: 151155.
  • Brodie, E.D. III, Moore, A.J. & Janzen, F.J. 1995. Visualizing and quantifying natural selection. Trends Ecol. Evol. 10: 313318.
  • Búrquez, A. & Corbet, S.A. 1991. Do flowers reabsorb nectar? Funct. Ecol. 5: 369379.
  • Calder, W.A. 1979. On the temperature-dependency of optimal nectar concentration for birds. J. Theor. Biol. 78: 185196.
  • Campbell, D.R. 1996. Evolution of floral traits in a hermaphroditic plant: field measurements of heritabilities and genetic correlations. Evolution 50: 14421453.
  • Campbell, D.R., Waser, N.M. & Price, M.V. 1996. Mechanisms of hummingbird-mediated selection for flower width in Ipomopsis aggregata. Ecology 77: 14631472.
  • Castellanos, M.C., Wilson, P. & Thomson, J.D. 2002. Dynamic nectar replenishment in flowers of Penstemon (Scrophulariaceae). Am. J. Bot. 89: 111118.
  • Castellanos, M.C., Wilson, P. & Thomson, J.D. 2003. Pollen transfer by hummingbirds and bumblebees, and the divergence of pollination modes in Penstemon. Evolution 57: 27422752.
  • Castellanos, M.C., Wilson, P. & Thomson, J.D. 2004. ‘‘Anti-bee’’ and ‘‘pro-bird’’ changes during the evolution of hummingbird pollination in Penstemon flowers. J. Evol. Biol. 17: 876885.
  • Chapotin, S.M., Holbrook, N.M., Morse, S.R. & Gutiérrez, M.V. 2003. Water relations of tropical dry forest flowers: pathways for water entry and the role of extracellular polysaccharides. Plant Cell Environ. 26: 623630.
  • Chase, M.W., Soltis, D.E., Olmstead, R.G. et al. 1993. Phylogenetics of seed plants: an analysis of nucleotide sequences from the plastid gene rbcL. Ann. Miss. Bot. Gard. 80: 528580.
  • Cresswell, J.E. 1998. Stabilizing selection and the structural variability of flowers within species. Ann. Bot. 81: 463473.
  • Cresswell, J.E. & Galen, C. 1991. Frequency-dependent selection and adaptive surfaces for floral character combinations: the pollination of Polemonium viscosum. Am. Nat. 138: 13421353.
  • Cruden, R.W., Hermann, S.M. & Peterson, S. 1983. Patterns of nectar production and plant animal coevolution. In: The Biology of Nectaries (B. Bentley & T. Elias, eds), pp. 126152. Columbia University Press, New York.
  • Díaz-Uriarte, R. & Garland, T. Jr 1996. Testing hypotheses of correlated evolution using phylogenetically independent contrasts: sensitivity to deviations from Brownian motion. Syst. Biol. 45: 2747.
  • Davies, T.J., Barraclough, T.G., Chase, M.W., Soltis, P.S., Soltis, D.E. & Savolainen, V. 2004. Darwin's abominable mystery: insights from a supertree of the angiosperms. Proc. Natl Acad. Sci. U.S.A. 101: 19041909.
  • De la Barrera, E. & Nobel, P.S. 2004. Nectar: properties, floral aspects, and speculations on origin. Trends Plant Sci. 9: 6569.
  • Delph, L.F. & Lively, M. 1992. Pollinator visitation, floral display, and nectar production of the sexual morphs of a gynodioecious shrub. Oikos 63: 161170.
  • Devlin, B. & Stephenson, A.G. 1985. Sex differential floral longevity, nectar secretion, and pollinator foraging in a protandrous species. Am. J. Bot. 72: 303310.
  • Faegri, K. & van der Pijl, L. 1979. The Principles of Pollination Ecology, 3rd edn. Pergamon Press, Oxford.
  • Fahn, A. 1979. Ultrastructure of nectaries in relation to nectar secretion. Am. J. Bot. 66: 977985.
  • Feinsinger, P. 1978. Ecological interactions between plants and hummingbirds in a successional tropical community. Ecol. Monogr. 48: 269287.
  • Felsenstein, J. 1985. Phylogenies and the comparative data. Am. Nat. 125: 115.
  • Fleming, P.A., Hartman-Bakken, B., Lotz, C.N. & Nicolson, S.W. 2004. Concentration and temperature effects on sugar intake and preferences in a sunbird and a hummingbird. Funct. Ecol. 18: 223232.
  • Galen, C. & Cuba, J. 2001. Down the tube: pollinators, predators, and the evolution of flower shape in the alpine skypilot, Polemonium viscosum. Evolution 55: 19631971.
  • Galen, C., Sherry, R.A. & Carroll, A.B. 1999. Are flowers physiological sinks or faucets? Costs and correlates of water use by flowers of Polemonium viscosum. Oecologia 118: 461470.
  • Galetto, L. & Bernardello, G. 2004. Floral nectaries, nectar production dynamics and chemical composition in six Ipomoea species (Convolvulaceae) in relation to pollinators. Ann. Bot. 94: 269280.
  • Garland, T. Jr 1994. Quantitative genetics of locomotor behavior and physiology in a garter snake. In: Quantitative Genetic Studies of Behavioral Evolution (C. R. B. Boake, ed.), pp. 251277. University of Chicago Press, Chicago, IL.
  • Garland, T. Jr & Adolph, S.C. 1994. Why not to do two-species comparative studies: limitations on inferring adaptation. Physiol. Zool. 67: 797828.
  • Garland, T. Jr & Carter, P.A. 1994. Evolutionary physiology. Annu. Rev. Physiol. 56: 579621.
  • Garland, T. Jr & Díaz-Uriarte, R. 1999. Polytomies and phylogenetically independent contrasts: an examination of the bounded degrees of freedom approach. Syst. Biol. 48: 547558.
  • Garland, T. Jr, Harvey, P.H. & Ives, A.R. 1992. Procedures for the analysis of comparative data using phylogenetically independent contrasts. Syst. Biol. 41: 1832.
  • Garland, T. Jr, Midford, P.E. & Ives, A.R. 1999. An introduction to phylogenetically based statistical methods, with a new method for confidence intervals on ancestral states. Am. Zool. 39: 374388.
  • Garland, T. Jr, Bennett, A.F. & Rezende, E.L. 2005. Phylogenetic approaches in comparative physiology. J. Exp. Biol. 208: 30153035.
  • Grafen, A. 1989. The phylogenetic regression. Philos. Trans. R. Soc. Lond. B 326: 119157.
  • Grant, V. & Temeles, E.J. 1992. Foraging ability of rufous hummingbirds on hummingbird flowers and hawkmoth flowers. Proc. Natl Acad. Sci. U.S.A. 89: 94009404.
  • Hainsworth, F.R. 1973. On the tongue of a hummingbird: its role in the rate and energetics of feeding. Comp. Biochem. Physiol. 46: 6578.
  • Harder, L.D. & Cruzan, M.B. 1990. An evaluation of the physiological and evolutionary influences of inflorescence size and flower depth on nectar production. Funct. Ecol. 4: 559572.
  • Harvey, P.H. & Pagel, M. 1991. The Comparative Method in Evolutionary Biology. Oxford University Press, Oxford.
  • Heinrich, B. 1975. Energetics of pollination. Annu. Rev. Ecol. Syst. 6: 139170.
  • Heinrich, B. & Raven, P.H. 1972. Energetics and pollination ecology. Science 176: 597602.
  • Hurlbert, A.H., Hosoi, S.A., Temeles, E.J. & Ewald, P.W. 1996. Mobility of Impatiens capensis flowers: effect on pollen deposition and hummingbird foraging. Oecologia 105: 243246.
  • Irwin, R.E. & Brody, A.K. 1999. Nectar-robbing bumblebees reduce the fitness of Ipomopsis aggregata (Polemoniaceae). Ecology 80: 17031712.
  • Irwin, R.E., Adler, L.S. & Brody, A.K. 2004. The dual role of floral traits: pollinator attraction and plant defense. Ecology 85: 15031511.
  • Ives, A.R., Midford, P.E. & Garland, T. Jr 2007. Within-species variation and measurement error in phylogenetic comparative methods. Syst. Biol. 56: 252270.
  • Kearns, C.A. & Inouye, D.W. 1993. Techniques for Pollination Biologists. University Press of Colorado, Niwot, CO.
  • Kingsolver, J.G. & Daniel, T.L. 1983. Mechanical determinants of nectar feeding strategy in hummingbirds: energetics, tongue morphology and licking behavior. Oecologia 60: 214226.
  • Klinkhamer, P.G.L. & van der Veen-van Wijk, C.A.M. 1999. Genetic variation in floral traits of Echium vulgare. Oikos 85: 515522.
  • Klinkhamer, P.G.L., de Jong, T.J. & Metz, A.J. 1994. Why plants can be too attractive: a discussion of measures to estimate male fitness. J. Ecol. 82: 191194.
  • Lara, C. & Ornelas, J.F. 2001. Preferential nectar robbing of flowers with long corollas: experimental studies of two hummingbird species visiting three plant species. Oecologia 128: 263273.
  • Leiss, K.A., Vrieling, K. & Klinkhamer, P.G.L. 2004. Heritability of nectar production in Echium vulgare. Heredity 92: 446451.
  • Lotz, C.N. & Schondube, J.E. 2006. Sugar preferences in nectar- and fruit-eating birds: behavioural patterns and physiological causes. Biotropica 38: 315.
  • Maddison, W.P. & Maddison, D.R. 2004. Mesquite: A Modular System for Evolutionary Analysis. Version 1.01 [on line]. Available at: http://mesquiteproject.org.
  • Martínez del Rio, C., Schondube, J.E., McWhorter, T.J. & Herrera, L.G. 2001. Intake responses in nectar feeding birds: digestive and metabolic causes, osmoregulatory consequences, and coevolutionary effects. Am. Zool. 41: 902915.
  • Martins, E.P. & Garland, T. Jr 1991. Phylogenetic analyses of the correlated evolution of continuous characters: a simulation study. Evolution 45: 534557.
  • Mitchell, R.J. 1993. Adaptive significance of Ipomopsis aggregata nectar production: observation and experiment in the field. Evolution 47: 2535.
  • Mitchell, R.J. 1994. Effects of floral traits, pollination visitation, and plant size on Ipomopsis aggregata fruit production. Am. Nat. 143: 870889.
  • Mitchell, R.J. 2004. Heritability of nectar traits: why we know so little? Ecology 85: 15271533.
  • Mitchell, R.J. & Shaw, R.G. 1993. Heritability of floral traits for the perennial wild flower Penstemon centranthifolius (Scrophulariaceae): clones and crosses. Heredity 71: 185192.
  • Mitchell, R.J. & Waser, N.M. 1992. Adaptive significance of Ipomopsis aggregata nectar production: pollination success of single flowers. Ecology 73: 633638.
  • Mitchell, R.J., Shaw, R.G. & Waser, N.M. 1998. Pollinator selection, quantitative genetics, and predicted evolutionary responses of floral traits in Penstemon centranthifolius (Scrophulariaceae). Int. J. Plant Sci. 159: 331337.
  • Montgomerie, R.D. 1984. Nectar extraction by hummingbirds: response to different floral characters. Oecologia 63: 229236.
  • Mulcahy, D.L. 1979. The rise of the angiosperms: a genecological factor. Science 206: 2023.
  • Nepi, M., Guarnieri, M. & Pacini, E. 2001. Nectar secretion, reabsorption, and sugar composition in male and female flowers of Cucurbita pepo. Int. J. Plant Sci. 162: 353358.
  • Nicolson, S.W. 1995. Direct demonstration of nectar reabsorption in the flowers of Grevillia robusta (Proteaceae). Funct. Ecol. 9: 584588.
  • Nicolson, S.W. & Fleming, P.A. 2003. Nectar as food for birds: the physiological consequences of drinking dilute sugar solutions. Pl. Syst. Evol. 238: 139153.
  • Ordano, M. & Ornelas, J.F. 2004. Generous-like flowers: nectar production in two epiphytic bromeliads and a meta-analysis of removal effects. Oecologia 140: 495505.
  • Ordano, M. & Ornelas, J.F. 2005. The cost of nectar replenishment in two epiphytic bromeliads. J. Trop. Ecol. 21: 541547.
  • Pacini, E., Nepi, M. & Vesprini, J.L. 2003. Nectar biodiversity: a short review. Pl. Syst. Evol. 238: 721.
  • Pagel, M.D. 1992. A method for the analysis of comparative data. J. Theor. Biol. 156: 431442.
  • Pleasants, J.M. 1983. Nectar production in Ipomopsis aggregata (Polemoniaceae). Am. J. Bot. 70: 14681475.
  • Pleasants, J.M. & Chaplin, S.J. 1983. Nectar production rates of Asclepias quadrifolia: causes and consequences of individual variation. Oecologia 59: 232238.
  • Purvis, A. & Garland, T. Jr 1993. Polytomies in comparative analyses of continuous characters. Syst. Biol. 42: 569575.
  • Purvis, A. & Rambaut, A. 1995. Comparative analysis by independent contrasts (CAIC): an Apple application for analyzing comparative data. Comp. Appl. Biosci. 11: 247251.
  • Purvis, A., Gittleman, J.L. & Luh, H.K. 1994. Truth or consequences: effects of phylogenetic accuracy on two comparative methods. J. Theor. Biol. 167: 293300.
  • Pyke, G.H. 1984. Optimal foraging theory: a critical review. Annu. Rev. Ecol. Syst. 15: 523575.
  • Pyke, G.H. 1991. What does it cost a plant to produce floral nectar? Nature 350: 5859.
  • Pyke, G.H. & Waser, N.M. 1981. The production of dilute nectars by hummingbird and honeyeaters flowers. Biotropica 13: 260270.
  • Rathcke, B.J. 1992. Nectar distributions, pollinator behavior, and plant reproductive success. In: Effects of Resource Distribution on Animal–Plant Interactions (M. D. Hunter, T. Ohgushi & P. W. Price, eds), pp. 113138. Academic Press, New York.
  • Reed, M.L., Findlay, N. & Mercer, F.V. 1971. Nectar production in Abutilon. IV. Water and solute relations. Aust. J. Biol. Sci. 24: 677688.
  • Rezende, E.L., Bozinovic, F. & Garland, T. Jr 2004. Climatic adaptation and the evolution of basal and maximum rates of metabolism in rodents. Evolution 58: 13611374.
  • Roberts, W.M. 1996. Hummingbirds’ nectar concentration preferences at low volume: the importance of time scale. Anim. Behav. 52: 361370.
  • Roff, D.A. 1997. Evolutionary Quantitative Genetics. Chapman & Hall, New York.
  • Sanderson, M.J., Purvis, A. & Henze, C. 1998. Phylogenetic supertrees: assembling the trees of life. Trends Ecol. Evol. 13: 105109.
  • Schemske, D.W. & Bradshaw, H.D. 1999. Pollinator preference and the evolution of floral traits in monkey flowers (Mimulus). Proc. Natl Acad. Sci. U.S.A. 96: 1191011915.
  • Scobell, S.A. & Scott, P.E. 2002. Visitors and floral traits of a hummingbird-adapted cactus (Echinocereus coccineus) show only minor variation along an elevational gradient. Am. Midl. Nat. 147: 115.
  • Simpson, B.B. & Neff, J.L. 1983. Evolution and diversity of floral rewards. In: Handbook of Experimental Pollination Biology, Scientific and Academic Editions (C. E. Jones & R. J. Little, eds), pp. 142159. Van Norstrand, New York.
  • Soltis, D.E., Soltis, P.S., Chase, M.W., Mort, M.E., Albach, D.C., Zanis, M., Savolainen, V., Hahn, W.H., Hoot, S.B., Fay, M.F., Axtell, M., Swensen, S.M., Nixon, K.C. & Farris, J.S. 2000. Angiosperm phylogeny inferred from a combined data set of 18S rDNA, rbcl, and atpB sequences. Bot. J. Linn. Soc. 133: 381461.
  • Southwick, E.E. 1984. Photosynthate allocation to floral nectar: a neglected energy investment. Ecology 65: 17751779.
  • Stanton, M.L. 2003. Interacting guilds: moving beyond the pairwise perspective on mutualisms. Am. Nat. 162: S10S23.
  • Stanton, M.L. & Young, H.J. 1994. Selecting for floral character associations in wild radish, Raphanus sativus L. J. Evol. Biol. 7: 271285.
  • Stanton, M.L., Snow, A.A. & Handel, S.N. 1986. Floral evolution: attractiveness to pollinators influences male fitness. Science 232: 16251627.
  • Stephenson, A.G., Winsor, J.A., Schlichting, C.D. & Davis, L.E. 1988. Pollen competition, nonrandom fertilization, and progeny fitness: a reply to Charlesworth. Am. Nat. 132: 303308.
  • Stiles, F.G. 1976. Taste preferences, color preferences, and flower choice in hummingbirds. Condor 78: 1026.
  • Stiles, F.G. & Freeman, C.E. 1993. Patterns in floral nectar characteristics of some bird-visited plant species from Costa Rica. Biotropica 25: 191205.
  • Sutherland, G.D. & Gass, C.L. 1995. Learning and remembering of spatial patterns by hummingbirds. Anim. Behav. 50: 12731286.
  • Temeles, E.J. 1996. A new dimension to hummingbird–flower relationships. Oecologia 105: 517523.
  • Temeles, E.J., Linhart, Y.B., Masonjones, M. & Masonjones, H.D. 2002. The role of flower width in hummingbird bill length–flower length relationships. Biotropica 34: 6880.
  • Van Riper, W. 1958. Hummingbird feeding preferences. Auk 75: 100101.
  • Weast, R.C. & Astie, M.J. 1978. CRC Handbook of Chemistry and Physics, 59th (1978–1979) ed. CRC Press Inc., West Palm Beach, FL.
  • Webb, C.O. & Donoghue, M.J. 2002. Phylomatic: A Database for Applied Phylogenetics [on line]. Available at: http://www.phylodiversity.net/phylomatic.
  • Wilson, P.W., Castellanos, M.C., Hogue, J.N., Thomson, J.N. & Armbruster, W.S. 2004. A multivariate search for pollination syndromes among penstemons. Oikos 104: 345361.
  • Wilson, P., Castellanos, M.C., Wolfe, A.D. & Thomson, J.D. 2006. Shifts between bee and bird pollination in penstemons. In: Plant–pollinator Interactions, from Specialization to Generalization (N. M. Waser & J. Ollerton, eds), pp. 4768. The University of Chicago Press, Chicago, IL.
  • Worley, A.C. & Barrett, S.C.H. 2000. Evolution of floral display in Eichhornia paniculata (Pontederiaceae): direct and correlated responses to selection on flower size and number. Evolution 54: 15331545.
  • Zimmerman, M. 1988. Nectar production, flowering phenology, and strategies for pollination. In: Plant Reproductive Ecology: Patterns and Strategies (J. Lovett-Doust & L. Lovett-Doust, eds), pp. 157178. Oxford University Press, New York.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
JEB1374ST1to3.doc650KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.