| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Behavioral Ecology Vol. 11 No. 5: 550-559
© 2000 International Society for Behavioral Ecology
Cue use affects resource subdivision among three coexisting hummingbird species
Department of Ecology and Evolutionary Biology, The University of Arizona, Tucson, AZ 85721, USA
Address correspondence to E. A. Sandlin at National Center for Ecological Analysis and Synthesis, 735 State St., Suite 300, Santa Barbara, CA 93101. E-mail: sandlin{at}nceas.ucsb.edu .
Received 2 May 1999; accepted 27 March 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Competition for food can influence the coexistence of species via habitat selection, and learned behavior can influence foraging decisions. I investigated whether learned behavior and competition act together to influence species interactions between three coexisting hummingbird species: black-chinned (Archilochus alexandri), blue-throated (Lampornis clemenciae), and magnificent (Eugenes fulgens) hummingbirds. I found that color cue use by individuals affects not only their foraging choices but also population-level responses to competition. I presented hummingbirds two types of habitats (rich and poor feeders). All birds shared a preference for the rich feeders, but shifted preference toward poor feeders in response to competition. I used color cues to manipulate the amount of information available to birds and examined the effects of two information states (complete or incomplete) on their foraging choices. I examined hummingbirds' preferences for the rich feeders when both competitor densities and information varied. To relate foraging choices to energetic intake, I also analyzed energy gained during a single foraging bout. Males of all species exhibited strong preferences for rich feeders when they foraged with complete information and low competitor densities. Without complete information, the two subordinate species (black-chinned and magnificent) shifted preference away from rich feeders in response to high densities of the dominant species (blue-throated). Each subordinate species shifted in a unique way: black-chinned hummingbirds reduced foraging efficiency, while magnificent hummingbirds reduced foraging time. Birds foraging with complete information remained selective on rich feeders even at high competitor densities. Thus, learned information affected competitive interactions (for rich feeders) among these species.
Key words: Archilochus alexandri, color cues, competition, Eugenes fulgens, foraging, ideal free distribution, hummingbirds, isoleg theory, Lampornis clemenciae, learning.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Competing species may coexist because they subdivide similar resources. This subdivision should depend on individuals' abilities to assess those resources. In nature, color cues or other environmental cues can allow animals to use resources more efficiently (e.g., Mitchell, 1989
;
Weiss, 1997). For example,
Weiss (1992) demonstrated that
many flowers change color after pollination. Often, a plant retains pollinated
flowers to increase the size of the display to attract pollinators from a
distance (Lewis, 1993).
However, when the pollinator gets closer to the display, it follows these
different color cues to choose only those flowers that contain nectar (because
they require pollination). Natural selection should favor an animal's ability
to learn cues that help it exploit its environment better (see
Stephens, 1993). Giraldeau
(1997) provides a
comprehensive review of the effects of information on ecology and
behavior.
Hummingbirds readily use several types of environmental cues such as
location
(Bené,
1941; Goldsmith and Goldsmith,
1979; Hurly and Healy,
1996; Miller and Miller,
1971; Stiles,
1975), patterns (Healy and
Hurly, 1995), and colors (Gass
and Sutherland, 1985;
Goldsmith and Goldsmith, 1979;
Mitchell, 1989;
Sandlin, 1999). Cues allow
individual hummingbirds to predict resource quality so they may increase their
foraging efficiency (Mitchell,
1989). Could cue use also affect resource subdivision among
species?
The term "information state" refers to the ability of an animal
to predict the quality of its environment (see
Mitchell, 1989). An animal
with complete information can use cues to predict the quality of its
environment with no additional sampling. An animal with incomplete information
can only predict the quality of its environment by sampling from it
(Stephens, 1993;
Stephens and Krebs, 1986) or
by watching other foragers (Bednekoff and
Balda, 1996; Fiorito and
Scotto, 1992; Kodric-Brown and
Nicoletto, 1993). I examined the effects of two information states
(complete or incomplete) on hummingbird foraging choices to elucidate a
possible mechanism by which learned information affects species
coexistence.
Theoretical framework
Isoleg theory helps us understand how habitat preferences can evolve to
facilitate stable coexistence among populations of competing species
(Rosenzweig, 1979,
1981,
1986,
1987). This theory, based on
ideal free distribution theory (IFD;
Fretwell and Lucas, 1969),
assumes that competition helps to determine foraging behavior and, thus,
habitat selection. It assumes that the competitors use two habitats and that
higher population densities depress resource availabilities to force
alterations in habitat use for at least one of the species. Asymmetry in
competitive abilities can force a subordinate species away from its primary
(preferred) habitat and into a secondary one.
Although isoleg theory predicts behavioral shifts of entire populations, it
necessarily depends on the behavior of the actual foragers (i.e., the
individuals within each population). Classical optimal foraging theory (OFT;
reviewed by Pyke et al., 1977)
predicts that individuals should feed exclusively from the most profitable
resource as long as that resource is in sufficient supply. Considerable
empirical evidence confirms that individuals of most hummingbird species
choose the richest sucrose solutions available from their environment
(Mitchell, 1989;
Montgomerie, 1984;
Pimm, 1978;
Pimm et al., 1985;
Pyke, 1978; this study).
Mitchell (1989) showed that
information state affects an individual hummingbird's selectivity for rich
(versus poor) feeders, especially when travel costs are high. Mitchell defined
selectivity as "the proportion of the diet comprising rich resource
items" (1989: 147). I
adopt his definition here. Pimm et al.
(1985) empirically tested
isoleg theory using the same three hummingbird species treated herein. They
discovered the predicted competitor-driven thresholds in selectivity for all
three species. However, they did not vary information state.
Cue use and decision making
The word "choice" implies that individuals can discriminate
between available alternatives. How important is this discriminating ability
for habitat selectors, especially those that can learn cues from their
environment? I reasoned that the abilities of foragers to choose among patches
of habitat (e.g., rich and poor hummingbird feeders) should depend not only on
competitor pressure but also on information state. Do individuals that can
learn from their environment make better foraging (or habitat) choices in
different competitive situations? Could the competitor density that causes
population-level switches in selectivity (sensu
Pimm et al., 1985;
Rosenzweig, 1979) vary with
respect to the individuals' information state? Using IFD, OFT, and isoleg
theory, I predicted that: (1) Regardless of competitor density, foragers with
an information deficit should use poor feeders more readily than those with
complete information (i.e., incomplete information decreases selectivity). (2)
At high competitor densities, thresholds of foraging selectivity should exist
such that foragers switch from exclusive use of rich toward opportunistic use
of rich and poor feeders (i.e., high competition decreases selectivity). (3)
Individuals foraging at similar competitor densities, but with different
information states, should sometimes exhibit different degrees of selectivity
for rich feeders. (4) Subordinate species should exhibit a greater decrease in
selectivity in response to information deficit and competitor pressures than
the dominant species. Each of these predictions is based on theory; prediction
1 comes from OFT, prediction 2 comes from IFD, prediction 3 is a combination
of OFT and IFD, and prediction 4 comes from isoleg theory.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Study area
All research was conducted at the American Museum of Natural History's Southwestern Research Station (SWRS; 31°50' N; 109°15' W) in southeastern Arizona, USA. At an elevation of approximately 1650 m, the 36.4-ha SWRS is located along the Middle Fork of Cave Creek, one of the major drainages in the Chiricahua Mountains. At this elevation, the dominant plant communities are pine-oak-juniper scrub in xeric areas and sycamore-cottonwood-walnut forest along the mesic creek bottoms. My study site is a small (ca. 0.25 ha) meadow immediately south of Cave Creek on the SWRS property. Vegetation along the edges of the meadow provides shade and perching sites for hummingbirds. Dry grasses and some sedges cover the meadow during the dry season (March-July) when I conducted this study. Dry season conditions do not favor the growth of flowering plants normally visited by hummingbirds (e.g., Penstemon, Mimulus, and Aquilegia); I found none growing nearby. Arthropod resources (e.g., leafhoppers, flies, gnats, spiders) appropriate for hummingbirds were abundant throughout my study, as evidenced by the full crops of all specimens sacrificed for another study (Van Hook et al., manuscript in preparation). Even though hummingbirds were aggressive there, no single individual could defend this meadow, so all individuals foraged in it.
Overview of experimental design
I examined the effect of changes in information state on hummingbirds'
foraging choices. I superimposed this examination onto changes in foraging
choices generated by the competitive milieu. To manipulate information state,
I provided either color cues that allowed birds to distinguish between rich
and poor resources or color cues that did not allow such discrimination. While
I watched focal birds forage in each of these two information states, my
assistant counted and cataloged the numbers of birds, thus quantifying and
characterizing the competitive milieu.
Description of species
I studied breeding males (only) of black-chinned (Archilochus
alexandri, hereafter BC), blue-throated (Lampornis clemenciae,
hereafter BT), and magnificent hummingbirds (Eugenes fulgens,
hereafter MA). These birds migrate from Mexico and, possibly, Central America
to breed in the Chiricahua Mountains in the spring
(Johnsgard, 1997). Male BTs
(ca. 8.3 g) behaviorally dominate the other two species and hold territories
in rich riparian areas (Pimm,
1978; Powers and McKee,
1994). BC males (ca. 3.2 g) may hold territories, but only in
relatively xeric sites (Powers and McKee,
1994). Rather than hold territories, male MAs (ca. 7.9 g) trapline
[defined by Feinsinger and Colwell
(1978) as following "a
repeated foraging circuit among successive flowers or clumps"]
throughout the canyon (Johnsgard,
1997; Lyon et al.,
1977; Pimm, 1978;
Powers, 1996).
Bird identification
To identify individual males, several assistants and I trapped and marked
as many males of the three study species as possible (see
Sandlin, 1999, for details).
To distinguish between individuals, I painted (with Polly S nontoxic hobby
paint) the backs of those males with a unique combination of three color bars
between their shoulders. A bird foraging at a feeder board (see below)
presents its dorsal side, usually allowing the observer to see these color
bars.
Theoretical considerations
I presented patches of habitat (feeders) in such a way that I could easily
watch individuals select among them. To efficiently observe numerous
individuals as they made patch choices, I provided the birds with arrays of
closely spaced feeders. Putting feeders in close proximity reduced travel
costs for the birds, thus approximating an assumption of the simplest isoleg
theories (e.g., no cost to switching between habitats;
Rosenzweig, 1979).
Is a hummingbird feeder a patch or simply a modified prey item? I assert
that feeders are patches in the truest sense of the word. According to
Stephens and Krebs (1986:
14),
A prey yields a fixed amount of energy and requires a fixed amount of time to handle: the forager can control neither the energy gained nor the time spent in attacking a prey item. However, the forager controls the time spent in, and hence the energy gained from, a patch, because there is a well-defined relationship between time spent and energy gained. Any predator that sucks the juices out of its prey might be thought of as preying upon patches.
A tight positive relationship exists between the time a hummingbird spends
at a feeder and the energy it gains from that feeder
(Montgomerie, 1984; see
below). Therefore, I refer to hummingbird feeders in the text, but the reader
should remember that each feeder is a patch of habitat that can be selected by
individual foragers.
Feeder boards
A feeder board is an array of 10 (5 rich and 5 poor) acrylic 88-ml feeders
(Perky-Pet cat. no. 214) which each present one fountain covered by a
detachable bee guard. I used Velcro to attach feeders 15-20 cm apart in a
square pattern on a 0.61 m x 0.61 m pegboard
(Figure 1a). The pegboard
rested atop two PVC poles so that its bottom feeders were about 0.9 m above
the ground. This distance above ground assured that vigilance for ground
predators (e.g., roadrunners, Geococcyx californianus;
Lima, 1991) should not
influence a bird's decision to use feeders on the lower part of the board.
|
Information state
Manipulation of information state required that the birds first learn cues
so they could predict the location of rich (30% = 0.86 M sucrose) and poor
(15% = 0.43 M sucrose) feeders. All three species of hummingbird learned to
associate a color cue (via blue or yellow bee guards) with feeder quality
(Sandlin, 1999). Once birds
learned to associate bee guard color with feeder quality, I could deprive them
of these learned cues and the information that cues convey.
Reliable cues
Most individuals learned color cues in 3 days or less (unpublished data). I
let blue = rich and yellow = poor because I discovered that naive birds
(either juveniles or adults new to the area) may exhibit a bias for yellow
(Sandlin, 1999). I call these
blue or yellow bee guards "reliable cues." Birds can benefit from
reduced search costs after they learn to associate cue color with feeder
quality. With this knowledge, the bird enjoys complete information about
feeder quality (Stephens and Krebs,
1986) so it can predict a priori the quality of every feeder it
encounters.
Ambiguous cues
To deprive birds of information, I painted another set of bee guards with a
four-quarter pie-slice design with two opposing quarters yellow, and the
others blue (Figure 1b). I call
these bee guards "ambiguous cues." To make the birds suffer an
information deficit, I put ambiguous bee guards on both rich and poor feeders.
Even though the birds still see blue and yellow, ambiguous cues contain no
information about feeder quality because all feeders look the same. In this
case, search costs should increase because birds cannot predict feeder quality
without first sampling from the feeder. OFT predicts that selectivity, and
thus foraging efficiency, should decrease if a forager incurs higher search
costs (Pyke et al., 1977;
Stephens and Krebs, 1986).
Even with ambiguous cues, birds can still learn feeder locations
(Healy and Hurly, 1995;
Miller et al., 1985;
Sutherland and Gass, 1995).
Therefore, I randomized the locations of the five rich and five poor feeders
on feeder boards at least once a day, even on days that I did not collect
data. I also randomized locations immediately before data collection, making
location per se a relatively unreliable cue.
Data sessions
My assistant and I arranged three feeder boards facing inward from the
corners of an equilateral triangle with 10.5-m sides. This presented birds
with a constant supply of sugar water distributed among 30 feeders. On days we
collected data, we watched each feeder board (A, B, C) twice—once with
reliable cues and once with ambiguous cues—for a total of six 30-min
data sessions per day. I never mixed cue types during a data session; in each
session, all three boards provided the same type of cue. Over the course of
the field season, I also varied the time of day for sessions so that the data
sets include observations from just after dawn through dusk. However, I
generally avoided the hottest part of the day (e.g., 1030-1530 h) when bird
activities seemed lower. To avoid possible bias at certain times of day (e.g.,
always watching board A at sunrise or always beginning with reliable cues), I
randomized the order in which these six data sessions occurred (e.g., 5 June:
0622 h, B reliable; 0657 h, A ambiguous; 0837 h, B ambiguous; 1635 h, A
reliable; 1711 h, C ambiguous; 1848 h, C reliable). Immediately before data
collection, my assistant and I put the feeders on the focal feeder board
(e.g., C reliable) in new randomly determined locations.
With two stopwatches (one for rich feeders and one for poor), I measured
the cumulative amount of time a focal male spent feeding from each feeder type
during one visit to the board (a foraging bout). For 30 min, I recorded these
data for as many foraging bouts as possible. Occasionally, I observed the same
individual more than once during a data session. To avoid problems with
nonindependent estimates of foraging preferences, I used the average amount of
time that particular individual spent at rich and poor feeders during all his
visits within that data session. For each species, I treated all unpainted
males as the same individual in subsequent analyses. I did not purposely bias
sampling toward any particular species or attempt to get equal numbers of
observations from each species. Therefore, my observations should reflect the
relative activity densities (sensu Pimm et
al., 1985; see below) of males of these three species.
My assistant's censuses also measured activity densities. While I watched a single feeder board for 30 min, my assistant watched all three feeder boards (in a randomly-determined order) for 10 min each. During a 10-min segment, each bird was counted and identified as it arrived at that board. These counts estimate bird activity densities because they reflect the degree and type of competitor pressure (see below) birds encountered as they foraged. Differences in bird densities can vary by an order of magnitude over the course of 30 min (see below). I relied on this natural variation in bird density to vary competitor pressures.
Measurement of species-specific nectar extraction rates
Species-specific constraints on extraction rate place limits on how
efficiently birds may use their time. Therefore, I measured the average
extraction rate of each species separately using wild-caught birds that I put
into a large aviary (ca. 8 m x 4 m x 3 m). I placed one feeder
with a known volume of one of the sucrose solutions (rich or poor) in the
aviary and used a stopwatch to measure the cumulative amount of time all birds
(7-16 individuals) spent drinking from the feeder. They almost always drank
one at a time. Once the volume of solution in the feeder noticeably decreased,
I measured the remaining volume. I then calculated the change in volume per
time spent foraging to estimate extraction rate. Because I presented only one
feeder to several birds, this estimate represents the average extraction rate
for the population of each respective species. I followed this procedure for
each species and for each sucrose concentration. I used these extraction rates
to convert measurements of foraging time at feeder boards into energy gain
(kJ) per foraging bout (see below).
Data analysis
The data come from the field and are, therefore, inherently noisy and
non-normal. Because of the noisy nature of the data, I relaxed my requirements
for statistical significance and rejected the null hypothesis if p
< 0.07. I used JMP IN (SAS Institute,
Inc., 1996) to perform standard analysis of variance (ANOVA) on
these data because ANOVA is usually robust with respect to deviations from
normality (Sokal and Rohlf,
1981). The basic analysis was a three-way factorial design to
examine selectivity (proportion of total foraging time spent using rich
feeders; sensu Mitchell, 1989)
with the main treatment effects of information state, competitor density, and
bird species. My predictions require this multifactorial analysis, and I
expected complicated patterns to emerge from the data (e.g., predictions 3 and
4). Fortunately, most patterns in the data are clear.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Foraging behavior at feeder boards
Birds used feeder boards in the same way they would use an inflorescence of flowers: they hovered at the board, feeding or sampling from one, a few, or most of the feeders. Remember that, from a theoretical perspective, one feeder is a patch of undepletable resource. Thus, 1 feeder board contained 10 patches from which a hummingbird could forage. The birds need not switch between rich feeders, and they need not show aggression toward other birds feeding at other feeders. However, hummingbirds did not seem to perceive feeders as undepletable patches. Almost all birds were aggressive at feeder boards, suggesting that they viewed the boards, or feeders therein, as depletable patches that were profitable enough to defend. This is probably because high visitation rates of competitors (as many as 194 female BC visits and 16 male BT visits in 10 min; up to 273 individuals in 30 min) made access to feeders the limiting factor. Pimm (1978
) showed that competition
for time at feeders affected niche width, so it seems reasonable to think that
hummingbird aggression in my experiments reflected their perception of limited
access to feeders. This is the usual way that competitor pressure manifested
itself in these experiments.
Activity densities and estimates of competitor pressure
The estimates of activity density often contain counts of the same
individuals during a 30-min session. The relevant measure of competitor
pressure, activity density (sensu Pimm et
al., 1985), estimates how many birds were active and how often.
Activity density is not a count of population size; it is an estimate of the
intensity of competition experienced by individuals while foraging. Here,
activity density especially concerns the effects of male BTs, whose presence
probably influenced foraging choices of other birds (others often aborted
their visit; see Pimm et al.,
1985). It makes no difference if this competitor pressure resulted
from repeated activities from a few individuals or the same level of activity
from many individuals. The resulting effects of competition on subordinate
switches in feeder preference should be the same.
Objectively measuring competitor pressure proved difficult because this pressure was manifested in many ways. Interference competition from aggressive male BTs was often particularly fierce. In addition, high densities of female BCs provided pressure via exploitative competition. On average, we counted more female BCs in 30 min than all other birds combined (number of female BCs = 93.16 ± 42.71, number of all others = 54.3 ± 21.22; z = 6.81; p <.0001). Often, a focal male would leave a feeder—or the entire board—not because of active aggression but because of passive displacement by another bird, usually a female BC. Therefore, I considered exploitative competition from female BCs an important component of the competitive milieu. MAs also tended to displace others, rather than exhibit overt aggression at feeder boards.
Before data analysis, I separated each data session into one of two
categories. I considered competitor (activity) densities to be high if counts
showed more than 40 female BCs or more than 10 male BTs during at least one
10-min surveys (n = 27 data sessions). All other data sessions
(n = 29) represented low competitor densities. The total number of
birds varied greatly between one data session and the next (range: 65-273;
mean: 147.46 ± 44.71, coefficient of variation = 30.32). These bird
counts were not correlated between one data session and the next
(Sandlin, 2000). Therefore, I
am confident I sampled foraging behavior under a wide variety of competitive
regimes.
Estimates of selectivity
I calculated s, a measure of each individual's preference, or
selectivity, for rich feeders, as
 |
This index ranges from 0 to 1; s = 1 indicates exclusive use of rich feeders, s = 0.5 indicates equal use of rich and poor feeders, and s = 0 indicates exclusive use of poor feeders. In the context of isoleg theory, s measures a forager's preference for rich patches of habitat. Here, I am most concerned with population-level changes in s for the two subordinate species. A population's average s should equal 1.0 as long as foragers can withstand competitor pressures in the rich habitat. As competition forces the birds to accept trade-offs and follow an IFD, the average s should decrease to the point that it approximately equals 0.5. This value of s indicates that the subordinates have suffered sufficient costs from competition to switch to opportunistic use of both habitats. A positive relationship exists between s and the energy gained per unit of foraging effort (e.g., kJ/s). Therefore, s can also be interpreted as a measure of relative foraging efficiency. Because s is a proportion, I subjected it to a standard arcsine-square-root transformation before subsequent data analysis.
Predictions revisited
Prediction 1 requires s in all ambiguous cue treatments to be less
than s in reliable cue treatments. Prediction 2 requires s
at high competitor densities to be less than s at low densities, all
other things being equal. It predicts that s should change from being
indistinguishable from 1.0 to being indistinguishable from 0.5. Prediction 3
is an interaction of predictions 1 and 2, and says that, when competitor
pressures are the same, s for foragers with an information deficit
should be lower than that of foragers with complete information. Prediction 4
requires the two subordinate species to exhibit greater changes in s
than the dominant species.
Figure 2 shows species-specific differences in s with respect to cue type and competitor density. Notice three patterns. First, birds of all species foraging with reliable cues and low competitor densities (LD.) were most selective (Figure 2). The values for s in this case are 0.97 for BC, 0.94 for BT, and 0.96 for MA (and are indistinguishable from 1.0). All individuals foraged almost perfectly with reliable cues. This result is exactly what OFT predicts for foragers with complete information, immediately renewing patches, and no travel costs between patches.
|
Second, in all cases, birds of each species were less selective with ambiguous cues than with reliable cues (Figure 2). This pattern of response to incomplete information is predicted by OFT. I outlined this expectation in predictions 1 and 3.
Third, for BC at high competitor densities with ambiguous cues, s
= 0.571. This value is indistinguishable from s = 0.5
(Ho: s = 0.5: Wilcoxon signed-ranks test:
p =.70) and shows the predicted shift in BC selectivity. This result,
that competitively driven thresholds in selectivity exist, is expected from
prediction 2 and corroborates those of Pimm et al.
(1985). However, MAs exhibited
no such shift to opportunistic foraging. Because the pattern exists for one
subordinate species but not the other, prediction 4 is equally supported and
refuted.
ANOVA results for individual preference
Patterns revealed in Figure
2 show that information state (cue type), competitor pressure, and
the species of forager all affected selectivity. Predictions 3 and 4 lead us
to expect a three-way interaction between these three treatment factors.
Specifically, I expected all three species to reduce s in response to
high competition and an information deficit, but not necessarily to the same
degree. All birds should suffer from an information deficit; BCs and MAs
should suffer from high competitor pressures more than BTs. Contrary to these
expectations, the three-way interaction term was not significant
(Table 1). However, all three
two-way interactions and the three main treatment effects were significant,
consistent with predictions 1 and 2.
|
I separated the data into smaller logical segments to more closely examine these patterns. First, to examine species-specific differences in response to competitor pressure and cue type, I analyzed s for each species separately. Each species responded differently to these two main factors (Table 2). Selectivity by BCs was negatively affected by both factors (Figure 2). Selectivity by BTs was affected by these main factors in combination and by cue type alone; Figure 2 shows the synergistic negative effect of both factors on s for male BTs. Selectivity by MAs was affected only by cue type and not by competitor pressure. This last result contradicts prediction 4 for a subordinate species and may reflect the transient nature of individual MAs (see below).
|
Second, I let competitor density remain constant. Two-way ANOVA of s in the low competitor density data revealed a cue type x species interaction (p =.01). This interaction indicates that, when competition is relatively low, the species responded to the information treatments in different ways. Inspection of Figure 2 reveals the source of the interaction term: with ambiguous cues, BCs suffered a greater decrease in s than the other two species. A two-way analysis of the high competitor density data revealed no significant interaction term (p =.69). Instead, both main treatment effects were significant (cue type: p <.0001; species: p <.0001). Therefore, when competitor densities were high, the species exhibited different selectivities (BC s was lower than BT and MA), and s depended on information state, just as prediction 4 requires.
Third, I examined s when information state remained constant. Analysis of the ambiguous selectivities with species and competitor density as treatment factors revealed no significant interaction term (species x competitor density: p =.55), but the main treatment effects were significant (competitor density: p =.007; species: p =.001). Analysis of the reliable-cued selectivities revealed that s differed, depending on competitor pressure and the bird species (species x competitor density: p <.0001). Thus, even when birds had complete information, the species responded differently to competition, a pattern expected from isoleg theory and prediction 4.
Taken together, these two-way ANOVA results strongly suggest a trade-off between competitor pressure and information. Specifically, low competitor pressures can compensate birds for the added cost of finding rich feeders in the ambiguous cue situation. Alternatively, reliable cues can compensate birds for the added cost of high competitor pressures. However, because of the drastic drop in s for BCs in the high density ambiguous situation, we also know that the species respond to these trade-offs in different ways (cf. prediction 4 above). These two-way ANOVA results also conform to the other three predictions and may offer better understanding of the complex patterns than the three-way ANOVA.
Relating selectivity to energetics
My analysis of s is based on theoretical predictions that arise
from ecological and evolutionary arguments. However, analysis of selectivity
per se may not provide a rigorous enough understanding of the constraints that
animals actually endure. Natural selection often acts on factors that affect
survivorship. Furthermore, OFT arguments often assume the forager optimizes
its rate of energy intake. I chose to study hummingbirds largely because they
must meet their short-term (ca. 60 min) energy needs or they lose the ability
to fly and may die (Sandlin, personal observation; see
Diamond, 1990;
Hixon et al., 1983;
Tiebout, 1991;
Wolf and Hainsworth, 1971).
This energy demand closely ties a hummingbird's foraging choices to its
survival and reproduction. Therefore, I converted s into a currency
that might make the interpretation of such constraint-bounded behaviors
easier.
Nectar extraction rates
My index of selectivity is based on the time that birds apportion to
feeding at rich feeders. For hummingbirds, however, nectar extraction rate is
a function of bill length (Montgomerie,
1984), and these species have different bill lengths (BC: 19.3 mm,
BT: 22.6 mm, MA: 26.9 mm). Species-specific constraints on extraction rate
place limits on how efficiently birds may use their time. My results (from the
aviary) confirm Montgomerie's
(1984) report that extraction
rate is a function of bill length (y = 3.441x - 53.609,
R2 =.77, p <.0001). Furthermore, extraction
rates differed between species (BC: 8.71 µl/s; BT: 30.37 µl/s; MA: 35.66
µl/s; one-way ANOVA: n = 19, df = 2, F = 111.17,
p <.0001). This means that the time the different species spent
feeding at rich feeders yielded different energetic rewards.
Montgomerie (1984) also
showed that sucrose concentration can affect hummingbird extraction rate.
However, I found no significant differences between extraction rates for my
two sucrose concentrations (one-way ANOVA: BC: p = 0.65; BT:
p = 0.75; MA: p = 0.74). For subsequent data analysis, I
used the overall average extraction rate for each species.
Converting time into energy
How much energy does an average bird of each species acquire during an
average foraging bout at a feeder board? I examined the amount of energy an
individual gained during a single foraging bout by converting my time
measurements (that contributed to s) into energy gain measurements.
First, I determined that every milliliter of 30% sucrose contains 5.16 kJ
(Weast et al., 1983). I used
this equality to calculate the energy content of each microliter of rich and
poor solutions (each aliquot of the rich solution contained twice the energy
as the same volume of the poor solution). I then used these quantities to
calculate energy gain. I know how much time each bird spent feeding from rich
and poor feeders, so it was easy to convert time spent feeding into energy
gain. The resulting measure, e, is simply the total energy (kJ) an
individual ingested during a single foraging bout. The measure e
accounts for species-specific differences in nectar extraction rates.
An individual also spends energy while it forages. Unfortunately, I could
not directly calculate each bird's foraging costs because I do not know the
total amount of time the birds spent hovering at a feeder board (they hover
even when not feeding). Considerable variability existed in the proportion of
time birds spent feeding while hovering at a feeder board (Sandlin, personal
observation), so I cannot make an informed estimate of hovering cost. However,
general metabolic costs scale with the mass of the individual
(Calder, 1996;
West et al., 1997), and flight
costs relate to both mass and wing area
(Calder, 1996;
Carpenter et al., 1991;
Feinsinger et al., 1979). BTs
are significantly larger than MAs (BT: 8.28 ± 0.44 g; MA: 7.9 ±
0.49 g; t = 8.06, df = 86, p <.0001), and both species
are considerably larger than BCs (3.18 ± 0.22 g) (mass measurements
from Powers DR, unpublished data). In addition, the species differ in wing
morphologies. For example, BTs and MAs differ in wing chord length (BT >
MA), but not wing depth (chord: p <.0001; depth: p =.24).
We might be tempted to incorporate wing disc loadings here (see
Epting and Casey, 1973;
Feinsinger and Chaplin, 1975);
however, power output can change while a bird hovers, as can its metabolic
rate (see Feinsinger and Chaplin,
1975, and references therein), making the effect of wing disc
loading on estimates of energy expenditure somewhat uncertain. Moreover,
recent work on this problem by Wells
(1993) showed that flight
costs do not greatly differ between hummingbirds of similar body mass but
dissimilar wing areas. Therefore, I assumed that a hummingbird's costs while
foraging relate most closely to its body mass.
I adjusted each bird's calculated energy gain by dividing e by the average mass for males of its species. This new measure is the energy gain per gram of bird (kJ/g) during an average foraging bout (Figure 3). It reflects an estimate of the birds' flight costs as well as their immediate energy benefits from foraging. I use e/g as shorthand notation for this estimate.
|
ANOVA results for energy gain per gram
Three-way ANOVA (same design as for s above) of e/g
revealed the predicted significant three-way interaction
(Table 1). Therefore, in terms
of e/g per foraging bout, each species responded to competition
differently, depending on cue type. Two-way ANOVAs (as above) also showed
different responses between the species
(Table 2). In addition, each
species' pattern of response to competitor density and cue type for
e/g was different from its pattern of response for s. For
BCs, e/g decreased only in response to high competitor densities and
not cue type (one-way ANOVA for BC competitor density: p =.0004;
Figure 3). There were no
differences in BT energy intake per foraging bout, regardless of cue type or
competitor density. MAs suffered reduced e/g only when they
experienced high competitor density coupled with an information deficit
(Table 2,
Figure 3).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Information and selectivity
Foraging information strongly influences an animal's foraging efficiency (Bateson and Kacelnik, 1996
;
Krebs and Inman, 1992;
Mitchell, 1989; Valone,
1991,
1992). I have confirmed that
competitor pressures also affect hummingbird foraging decisions (see
Pimm et al., 1985;
Rosenzweig, 1986). Competition
must exert a strong force on hummingbird behaviors because, in the present
study, their responses occurred at a short time scale (ca. 3-8 s) and a small
spatial scale (ca. 10 cm). But competitor pressures were much more severe when
these animals foraged with incomplete information (ambiguous cues). Selectivity, the proportion of rich food in the diet, indicates an animal's degree of preference for rich food. Changes in selectivity (s) reflect trade-offs between foraging efficiency and other constraints. Hummingbirds that enjoyed low competitor densities (LD) used reliable cues to forage almost perfectly (Figure 2). Therefore, a decrease in s reflects, at least in part, the energetic price a bird paid to forage with either high competitor densities (HD), an information deficit, or a combination of these two factors. All three species showed a decrease in s with a decrease in their information state. Although they responded further to competitor pressures, only the BC exhibited the predicted (by isoleg theory) drastic switch in s in the HD ambiguous situation.
Information modulates density-dependent habitat selection in
hummingbirds
The measure of energetic intake per gram per foraging bout, e/g
(Figure 3), revealed different
aspects of hummingbird decision making than did the measure of selectivity,
s (Figure 2). To
understand these differences, let us compare and contrast the two measures
(see Table 2). BCs reduced
their e/g in response to high competitor densities. However, at
either competitor pressure, e/g was the same for both cue treatments
(Figure 3). This result is
predicted by IFD theory and usually assumed in experimental studies. We rarely
see empirical confirmation that animals trade-off two currencies to gain the
same fitness. In this case, male BCs gained the same total amount of energy
during a single foraging bout, regardless of cue type
(Figure 3, Table 2). This means that the
birds compensated for their reduced foraging efficiency (because of an
information deficit; Figure 2)
by spending more time feeding at poor feeders than they did when they had
complete information.
Spending additional time hovering at poor feeders caused birds to suffer
proportionally higher flight costs than they would if feeding at rich feeders.
The costs of spending more time at feeder boards are at least fourfold: (1)
added flight cost for hovering longer at the feeder board, (2) missed
opportunity costs for not feeding at rich feeders and not doing other
activities, (3) predation risk (probably very low during the day), and (4)
added risk of conflict with another hummingbird. This last cost must be real
for BCs because they experienced reduced e/g in the HD situation
relative to the LD situation (Figure
3). These results show the true nature of BCs' subordinate status
in this guild. In other hummingbird guilds, small-bodied species sometimes
show flexibility in their roles within the community
(Feinsinger and Colwell,
1978). Perhaps BCs exhibited such flexibility by switching to
opportunistic foraging.
BT males gained the same e/g regardless of cue type or competitor
density (Figure 3). We know
that BTs suffered reduced foraging efficiency in the ambiguous cue x HD
situation (Table 2,
Figure 2). But, just as BCs
did, male BTs paid the extra costs associated with foraging longer (relative
to other situations) at ambiguously cued poor feeders to obtain the same
e/g (Figure 3). Notice
that, even though BT s decreased, it stayed as high or higher than
that for the other two species in every case
(Figure 2). Therefore, via
aggressive behavior, the BTs dominated the others and controlled distribution
of the food resources (see Lyon et al.,
1977; Pimm et al.,
1985), or they controlled the amount of time available to use rich
feeders (Pimm, 1978;
Sandlin, 2000).
Like BTs and BCs with low competitor pressures, MAs compensated for an information deficit by feeding longer at poor feeders (than they did with reliable cues) to get the same amount of energy they would enjoy with complete information (Figure 3). Unlike the BCs and BTs, however, male MAs did not spend a greater proportion of their time feeding at poor feeders in the HD situation. They spent less time feeding and, therefore, got significantly less energy per foraging bout in that case.
Magnificent hummingbirds
Magnificent hummingbirds have puzzled many scientists because they do not
seem to fit nicely into any given category in hummingbird communities (see
Colwell, 1973;
Feinsinger and Colwell, 1978;
Mitchell, 1989;
Pimm, 1978;
Powers, 1996;
Sandlin, 2000; Van Hook et
al., in preparation; Wolf et al.,
1976). Why might MAs respond differently to competitor pressures
than the other two species? This species, called an "interstitial
species" by Colwell
(1973), behaves differently in
different competitive situations. It holds territories in the highlands of
central Mexico, although it remains subordinate to BTs there
(Johnsgard, 1997;
Lyon, 1976;
Wolf et al., 1976). In the
Chiricahua Mountains of southeastern Arizona, though, this species is a
"high-reward trapliner"
(Feinsinger and Colwell, 1978;
Powers, 1996) and not a
territorialist. This traplining strategy seems to allow MAs a degree of
foraging flexibility not available to its competitors in southeastern Arizona
(Van Hook et al., in preparation).
Lyon (1976) found high
turnover among nonterritorial individuals (of several species) in Mexico.
Examination of my focal observations of marked males suggests different trends
in the visitation rates of individuals differ between the species. Individual
MAs tended to be less likely than BCs or BTs to contribute multiple times to
my data (mean number of visits per individual in entire data set: BC: 7.28
± 8.75, BT: 4.08 ± 5.58, MA: 3.81 ± 4.83). In addition,
more focal MAs were unmarked than the other two species (25.5% of MA
observations were on unmarked birds, while 10.5% of BC and 13.8% of BT were
unmarked birds). My results seem to punctuate the truly ephemeral nature of
any individual MA's membership in the competitive milieu.
Because individual MAs probably range widely throughout the canyons, individuals suffering from high competitor pressures and incomplete information at feeder boards may have abandoned the area to continue along their trapline. However, during the dry season, my feeder boards were likely the major sources of quick carbohydrate energy. Therefore, I propose another interpretation of the observed MA response to high competitor densities and ambiguous cues.
Until now, I have restricted the discussion to behavioral switches between
two alternatives (rich or poor feeders). Yet MAs may switch to another source
of energy: arthropods (Powers,
1996; Van Hook et al., in preparation). In fact, I speculate that
MAs might eat more arthropods than either of the other two hummingbird species
(see Feinsinger and Colwell,
1978, for characteristics of high-reward trapliners;
Powers, 1996, and references
therein). The secondary habitat for MAs may contain patches of arthropods,
rather than patches of low-reward sucrose. Results in Figures
2 and
3 are consistent with my
assertion that MAs may switch to some other habitat. This scenario is similar
to Rosenzweig and Abramsky's
(1986) discovery that two
gerbil species share a primary habitat, but each uses a unique secondary
habitat. It also agrees with Feinsinger and Colwell's
(1978) assertion that MAs
(although in another multispecies assemblage) restrict themselves to the
highest nectar concentrations available. Finally, if this species exhibits a
differentially high use of arthropods, that difference may explain why it
enjoys the freedom of a trapliner, rather than the confinement and energy
demands of territorial defense of nectar resources
(Lyon, 1976;
Lyon et al., 1977;
Stiles and Wolf, 1970;
Wolf et al., 1976). I discuss
this idea in more detail elsewhere
(Sandlin, 2000).
Theory revisited
Ideal free distribution theory predicts that animals will trade-off various
costs so they obtain the same fitness in two different habitats. Rich and poor
feeders provide patches of different habitats
(Stephens and Krebs, 1986).
For hummingbirds, average energy gain seems to serve as an excellent surrogate
for fitness. Results here confirm that hummingbirds follow ideal free
distributions with respect to feeder quality
(Figure 3).
Rosenzweig and Abramsky
(1997) discuss
shared-preference systems in which there is asymmetric competition between the
species. In shared-preference systems, both species prefer one habitat but can
profitably use another (Rosenzweig,
1979). One species, the dominant one, cannot tolerate energy
losses (due to use of the second habitat) as well as the subordinate species
can (see also Rosenzweig,
1987). Therefore, Rosenzweig and Abramsky
(1997) call the dominant
species the "intolerant" one, while the subordinate species is
labeled the "tolerant" one. These labels apply especially well to
hummingbirds. Dominant BTs reduced s in response to ambiguous cues
and high competitor densities but never tolerated a reduction in e/g.
Subordinate BCs tolerated a reduction in both s and e/g in
response to high competition and incomplete information. Subordinate MAs
reduced s only in response to incomplete information. The MAs
tolerated reduced e/g (from nectar feeding) in response to a
combination of ambiguous cues and high competitor densities. Each species'
response reflects its place in the hummingbird guild. As discussed above, the
results also suggest that the secondary habitat for MAs may differ from that
for BCs and BTs (see Feinsinger and
Colwell, 1978; Rosenzweig and Abramsky, 1996;
Sandlin, 2000; Van Hook et
al., in preparation).
In summary, the observed differences in hummingbird foraging responses to
competition, cue type, and resource subdivision suggest that information may
be important in regulating not only individual foraging efficiency and
competitive interactions but also species coexistence for these three
hummingbird species. Foraging behavior provides a good indicator of niche
breadth because habitat selection usually reflects the animal's food
requirements (see Perry and Pianka,
1997; Schmitz,
1997). An animal's experiences (e.g., via learning) can also
strongly influence its foraging behavior (e.g.,
Mitchell, 1989;
Sandlin, 1999; Valone,
1991,
1992). Until now, no one has
attempted to carry out an integrative study that links learned behavior to
interactive phenomena involving several species that coexist via
density-dependent habitat selection. Complete information allowed hummingbirds
to better withstand the negative effects of competition. This allowed the
subordinate species to forage more efficiently or gain more energy than they
would without information. These results should be general for species that
use information to choose among resources. Therefore, the ability to learn and
use information from the environment may allow more species to coexist than
could otherwise exploit a single type of resource.
|
ACKNOWLEDGEMENTS |
|---|
Much of the field work was supported by the American Museum of Natural History, the Frank M. Chapman Memorial Fund, and the Southwestern Research Station Student Support Fund. Additional funding was provided by the Faculty of Sciences, the Department of Ecology and Evolutionary Biology, and the Robert S. Flinn Foundation at the University of Arizona. Mitch Erickson of Perky-Pet Products donated feeders and extra bee guards. I thank Graeme Bell, Rick Ulstad, Geoff Parks, and Joannie Lortet for invaluable field assistance. I thank Michael Rosenzweig, Bill Mitchell, Dan Papaj, Don Powers, Judith Bronstein, Steve Russell, Ruth Russell, Gretchen LeBuhn, Yaron Ziv, Wade Leitner, Cynthia Riginos, Brian McGill, Cathy Collins, and many other friends for encouragement, constructive criticism, and support throughout this study. I thank two anonymous reviewers for their insightful comments and careful reading of the manuscript. Mike Willig provided office space at Texas Tech University during manuscript revision. This study is part of my dissertation at the University of Arizona.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bateson M, Kacelnik A, 1996. Rate currencies and the foraging starling: the fallacy of the averages revisited. Behav Ecol 7: 341-352.
Bednekoff PA, Balda RP, 1996. Observational spatial memory in Clark's nutcrackers and Mexican jays. Anim Behav 52: 833-839.[ISI]
Bené F. 1941. Experiments on the color preference of black-chinned hummingbirds. Condor 43: 237-242.
Calder WA III, 1996. Size, function, and life history. Mineola, Minnesota: Dover Press.
Carpenter FL, Hixon MA, Hunt A, Russell RW, 1991. Why hummingbirds have such large crops. Evol Ecol 5: 405-414.
Colwell RK, 1973. Competition and coexistence in a simple tropical community. Am Nat 107: 737-760.
Diamond JM, 1990. How to fuel a hummingbird. Nature 348: 392.
Epting RJ, Casey TM, 1973. Power output and wing disc loading in hovering hummingbirds. Am Nat 107: 761-765.
Feinsinger P, Chaplin SB, 1975. On the relationship between wing disc loading and foraging strategy in hummingbirds. Am Nat 109: 217-224.
Feinsinger P, Colwell RK, 1978. Community organization among neotropical nectar-feeding birds. Am Zool 18: 779-795.
Feinsinger P, Colwell RK, Terborgh J, Chaplin SB, 1979. Elevation and the morphology, flight energetics, and foraging ecology of tropical hummingbirds. Am Nat 113: 481-497.[ISI]
Fiorito G, Scotto P, 1992. Observational learning in
Octopus vulgaris. Science 256:
545-547.
Fretwell SD, Lucas HL, 1969. On territorial behavior and other factors influencing habitat distribution in birds. I. Theoretical development. Acta Biotheor 19: 16-36.
Gass CL, Sutherland GD, 1985. Specialization by territorial hummingbirds on experimentally enriched patches of flowers: energetic profitability and learning. Can J Zool 63: 2125-2133.
Giraldeau L-A, 1997. The ecology of information use. In: Behavioural ecology: an evolutionary approach, 4th ed. (Krebs JR, Davies NB eds). Oxford: Blackwell; 42-68.
Goldsmith TH, Goldsmith KM, 1979. Discrimination of colors by black-chinned hummingbird, Archilochus alexandri. J Comp Physiol A 130: 209-220.
Healy SD, Hurly TA, 1995. Spatial memory in rufous hummingbirds (Selasphorus rufus): a field test. Anim Learn Behav 23: 63-68.
Hixon MA, Carpenter FL, Paton DC, 1983. Territory area, flower density, and time budgeting in hummingbirds: an experimental and theoretical analysis. Am Nat 122: 366-391.
Hurly TA, Healy SD, 1996. Memory for flowers in rufous hummingbirds: location or local visual cues? Anim Behav 51: 1149-1157.
Johnsgard PA, 1997. The hummingbirds of North America, 2nd ed. Washington, DC: Smithsonian Institution Press.
Kodric-Brown A, Nicoletto PF, 1993. The relationship between physical condition and social status in pupfish Cyprinodon pecosensis. Anim Behav 46: 1234-1236.
Krebs JR, Inman AJ, 1992. Learning and foraging: individuals, groups, and populations. Am Nat 140:S63-S84.
Lewis AC, 1993. Learning and the evolution of resources: pollinators and flower morphology. In: Insect learning: ecological and evolutionary perspectives (Papaj DR, Lewis AC eds). New York: Chapman and Hall; 219-242.
Lima SL, 1991. Energy, predators and the behaviour of feeding hummingbirds. Evol Ecol 5: 220-230.
Lyon DL, 1976. A montane hummingbird territorial system in Oaxaca, Mexico. Wilson Bull 88: 280-299.
Lyon DL, Crandall J, McKone M, 1977. A test of the adaptiveness of interspecific territoriality in the blue-throated hummingbird. Auk 94: 448-454.
Miller RS, Miller RE, 1971. Feeding activity and color preference of ruby-throated hummingbirds. Condor 73: 309-313.
Miller RS, Tamm S, Sutherland GD, Gass CL, 1985. Cues for orientation in hummingbird foraging: color and position. Can J Zool 63: 18-21.
Mitchell WA, 1989. Informational constraints on optimally foraging hummingbirds. Oikos 55: 145-154.
Montgomerie RD, 1984. Nectar extraction by hummingbirds: response to different floral characters. Oecologia 63: 229-236.
Perry G, Pianka ER, 1997. Animal foraging: past, present and future. Trends Ecol Evol 12: 360-364.
Pimm SL, 1978. An experimental approach to the effects of predictability on community structure. Am Zool 18: 797-808.
Pimm SL, Rosenzweig ML, Mitchell WA, 1985. Competition and food selection: field tests of a theory. Ecology 66: 798-807.[ISI]
Powers DR, 1996. Magnificent Hummingbird (Eugenes fulgens). In: The birds of North America, no. 221 (Poole A, Gill F eds). Washington, DC: Academy of Natural Sciences.
Powers DR, McKee T, 1994. The effect of food availability on time and energy expenditures of territorial and non-territorial hummingbirds. Condor 96: 1064-1075.
Pyke GH, 1978. Optimal foraging in hummingbirds: testing the marginal value theorem. Am Zool 18: 739-752.
Pyke GH, Pulliam HR, Charnov ER, 1977. Optimal foraging: a selective review of theory and tests. Q Rev Biol 52: 137-154.
Rosenzweig ML, 1979. Optimal habitat selection in two-species competitive systems. Fortschr Zool 25(2/3): 283-293.
Rosenzweig ML, 1981. A theory of habitat selection. Ecology 62: 327-335.
Rosenzweig ML, 1986. Hummingbird isolegs in an experimental system. Behav Ecol Sociobiol 19: 313-322.
Rosenzweig ML, 1987. Habitat selection as a source of biological diversity. Evol Ecol 1: 315-330.
Rosenzweig ML, Abramsky Z, 1986. Centrifugal community organization. Oikos 46: 339-348.
Rosenzweig ML, Abramsky Z, 1997. Two gerbils of the Negev: a long-term investigation of optimal habitat selection and its consequences. Evol Ecol 11: 733-756.
SAS Institute, 1996. JMP IN version 3 for Macintosh. Cary, North Carolina: SAS Institute Inc.
Sandlin EA, 1999. Information use and species interactions in a hummingbird guild (PhD dissertation). Tucson: University of Arizona.
Sandlin EA, 2000. Foraging information affects the nature of competitive interactions. Oikos (in press).
Schmitz OJ, 1997. Commemorating 30 years of optimal foraging theory. Evol Ecol 6: 631-632.
Sokal RR, Rohlf FJ, 1981. Biometry, 2nd ed. New York: WH Freeman.
Stephens DW, 1993. Learning and behavioral ecology: incomplete information and environmental predictability. In: Insect learning: ecological and evolutionary perspectives (Papaj DR, Lewis AC, eds). New York: Chapman and Hall; 195-218.
Stephens DW, Krebs JR, 1986. Foraging theory, Princeton, New Jersey: Princeton University Press.
Stiles FG, 1975. Ecology, flowering phenology, and hummingbird pollination of some Costa Rican Heliconia species. Ecology 56: 285-301.[ISI]
Stiles FG, Wolf LL, 1970. Hummingbird territoriality at a tropical flowering tree. Auk 87: 467-491.[ISI]
Sutherland GD, Gass CL, 1995. Learning and remembering of spatial patterns by hummingbirds. Anim Behav 50: 1273-1286.
Tiebout HM III, 1991. Daytime energy management by tropical hummingbirds: responses to foraging constraint. Ecology 72: 839-851.
Valone TJ, 1991. Bayesian and prescient assessment: foraging with preharvest information. Anim Behav 41: 569-577.
Valone TJ, 1992. Information for patch assessment: a
field investigation with black-chinned hummingbirds. Behav Ecol
3: 211-222.
Weast RC, Astle MJ, Beyer WH (eds), 1983. CRC handbook of chemistry and physics, 64th ed. Boca Raton, Florida: CRC Press.
Weiss MR, 1992. Floral colour changes as cues for pollinators. Nature 354: 227-229.
Weiss MR, 1997. Innate colour preferences and flexible colour learning in the pipevine swallowtail. Anim Behav 53: 1043-1052.
Wells DJ, 1993. Muscle performance in hovering hummingbirds. J Exp Biol 178: 39-57.[Abstract]
West GB, Brown JH, Enquist BJ, 1997. A general model
for the origin of allometric scaling laws in biology. Science
276: 122-126.
Wolf LL, Hainsworth FR, 1971. Time and energy budgets of territorial hummingbirds. Ecology 52: 980-988.[ISI]
Wolf LL, Stiles FG, Hainsworth FR, 1976. Ecological organization of a tropical, highland hummingbird community. J Anim Ecol 45: 349-379.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. L. Altshuler and R. Dudley The ecological and evolutionary interface of hummingbird flight physiology J. Exp. Biol., August 15, 2002; 205(16): 2325 - 2336. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||