CALIFORNIA STATE UNIVERSITY, NORTHRIDGE
HUMMINGBIRD CHOICES AT ARTIFICIAL FLOWERS MADE TO RESEMBLE
ORNITHOPHILES VERSUS MELITTOPHILES
A thesis submitted in partial fulfillment of the requirements for
The degree of Masters of Science in Biology
By
Wyndee A. Haley
May 2010
The thesis of Wyndee A. Haley is approved:
____________________________________
_______________
Fritz Hertel, Ph.D.
Date
____________________________________
_______________
Mark Steele, Ph.D.
Date
____________________________________
_______________
Paul Wilson, Ph.D., Committee Chair
Date
California State University, Northridge
ii
DEDICATION
To my amazing family: Thomas and Alexis Haley for encouraging me throughout this
journey, the broader Haley and Baldwin clans for keeping me smiling, and to my
boyfriend, Nathan Guzman, for giving me strength when I needed it most.
iii
ACKNOWLEDGEMENTS
Thank you to Sigma Xi for funding. Thank you to Dr. Manuel Fernandez who
helped me order the proper sugars used in my study. Thank you also to Carly Ryan for
statistical help.
I am indebted to my advisor, Dr. Paul Wilson, who has given me the tools to
become a promising researcher. Without his unyielding patience, continuing guidance,
and dedication to my academic advancement, I would not have obtained the ability to
become the academic professional that I wish to be in the future. Thank you for your help
and guidance throughout this process. I thank my other committee members, Drs. Mark
Steele and Fritz Hertel, for their help with the formulation of experiments, as well as their
critique on the drafts of this thesis and helpful encouragement.
I am infinitely grateful to my loving, caring and one-of-a-kind parents, Thomas
and Alexis Haley, who have guided and fueled my education. Their never ending support
(monetary and emotional), constant dedication and unyielding love have shaped me into
the person that I am proud to be. Without them, I would have truly been lost. To the rest
of the Haley Family (Lucas and Estenia) and Baldwin Family (Andrew, Suzanne, Jacob
and Erik), who have given me the love, encouragement and guidance that I needed
throughout this journey. Thank you for always being there to keep me going. And to my
wonderful boyfriend, Nathan Guzman, who has supported me through difficult and
emotional times, as well as shared in my happiness. Thank you for driving up to my
secluded field site just to see me; for helping me set up experiments, when you had to
wake up at 5:00 a.m. to videotape birds and fill flowers with artificial nectar. Thank you
for your constant encouragement and unfailing love. Your light has kept me shining.
iv
TABLE OF CONTENTS
Page
SIGNATURE PAGE ............................................................................................... ii
DEDICATION......................................................................................................... iii
ACKNOWLEDGEMENTS..................................................................................... iv
LIST OF FIGURES AND TABLES........................................................................ vii
ABSTRACT............................................................................................................. viii
INTRODUCTION ................................................................................................... 1
FLORAL CHARACTERISTICS ................................................................ 1
HUMMINGBIRD FORAGING .......................................... ....................... 4
GOALS........................................................................................................ 8
GENERAL EXPERIMENTAL SET-UP................................................................. 10
MATERIALS AND METHODS ................................................................ 10
STATISTICS ............................................................................................... 14
EXPERIMENT 1 ..................................................................................................... 17
EXPERIMENT 2 ..................................................................................................... 23
EXPERIMENT 3 ..................................................................................................... 34
EXPERIMENT 4 ..................................................................................................... 44
EXPERIMENT 5 ..................................................................................................... 47
CRITIQUE OF EXPERIMENTAL DESIGN ......................................................... 50
WHY A FIELD STUDY?............................................................................ 50
REPLICATION AND RANDOMIZATION............................................... 51
DISCUSSION .......................................................................................................... 53
v
EVOLUTION OF FLOWERS OCCURS VIA MANY FACTORS ........... 54
BIRD BEHAVIORAL ECOLOGY............................................................. 62
LITERATURE CITED ............................................................................................ 68
vi
LIST OF FIGURES AND TABLES
Plate
Page
1. FIGURE 1 (Figures representing materials and methods used) ..................... 13
2. FIGURE 2: EXPERIMENT 1A RED VS. PURPLE (All variables).............. 19
3. FIGURE 3: EXPERIMENT 1B RED VS. YELLOW (All variables)............ 20
4. FIGURE 4: EXPERIMENT 1C PURPLE VS. YELLOW (All variables)..... 22
5. FIGURE 5: EXPERIMENT 2 NECTAR QUANTITY (graphs of categorical
variables)......................................................................................................... 30
6. TABLE 1: EXPERIMENT 2 NECTAR QUANTITY (statistics of categorical
variables)..........................................................................................................31
7. FIGURE 6: EXPERIMENT 2 GRAPHS (graphs of quantitative variables).. 32
8. TABLE 2: EXPERIMENT 2 STATS (statistics of quantitative variables).... 33
9. FIGURE 7: EXPERIMENT 3A 2 VS. 8 µL (All variables)........................... 37
10. FIGURE 8: EXPERIMENT 3B ORNITHOPHILOUS VS. MELITTOPHILOUS
(All variables) ................................................................................................. 39
11. FIGURE 9: EXPERIMENT 3C SUCROSE VS. HEXOSE (All variables)... 41
12. FIGURE 10: EXPERIMENT 3D 12% VS. 48% (All variables).................... 43
13. FIGURE 11: EXPERIMENT 4 LANDING PLATFORM VS. NO LANDING
PLATFORM (All variables) ........................................................................... 46
14. FIGURE 12: EXPERIMENT 5 USING PATTERN AS A CUE (All
variables)............................................................................................................... 49
vii
ABSTRACT
HUMMINGBIRD CHOICES AT ARTIFICIAL FLOWERS MADE TO RESEMBLE
ORNITHOPHILES VERSUS MELITTOPHILES
By
Wyndee Angelina Haley
Master of Science in Biology
Certain floral characteristics are associated with specific pollinators. Hummingbirdpollinated flowers are usually red, lack a landing platform, lack color patterns on the
perianth, and contain a high amount of dilute sucrose-rich nectar compared to beepollinated flowers. The goal of this study was to test hypotheses concerning the reasons
for these characters to the extent that they involve hummingbird behaviors. An array was
set up that contained 16 artificial inflorescences, each with five artificial flowers. In
Experiment 1, flowers were made that differed only in color, and birds showed very little
preference, slightly preferring red over other colors. In Experiment 2, color was made to
be associated with nectar offerings, and birds learned to visit flowers of the color that
provided much more nectar (6 versus 2 µL but not 4 versus 2 µL), but generally 2 µL was
above the threshold for inclusion in the diet. In Experiment 3, birds were offered birdnectar (8 µL of 12% sucrose) versus bee-nectar (2 µL of 48% hexose), and birds did not
prefer the nectars that were similar to natural bird-adapted flowers even though they
could extract bird nectar with less handling time. In Experiment 4, birds were offered
flowers with and without landing platforms, and birds preferred flowers that lacked
viii
landing platforms, which saved them time. In Experiment 5, birds were offered flowers
that were patterned or not, associated with differing nectar volumes, and birds did not
associate the higher nectar reward with either flower type. In general, the preferences of
birds fall far short of explaining the natural phenomenon of bird- versus bee-pollination
syndromes. Other factors, such as adaptation to discourage bees, are discussed as
additional causes of the differences between the syndromes. Furthermore, I delve into the
possibility that bird behaviors may be based on their drive to sample new and different
flowers, i.e., flowers unlike those that are optimally rewarding.
ix
INTRODUCTION
A suite of specific characteristics is associated with hummingbird-pollinated
flowers: compared to bee-pollinated relatives, most are red, often without much color
patterning, have high quantities of diluted sucrose-rich nectar, lack landing platforms,
have floral tubes that are long and narrow, and have outwardly exserted anthers and
stigmas (Faegri and van der Pijl 1979; Mitchell & Paton 1990; Thomson et al. 2000).
This thesis examined hummingbird behaviors at artificial flowers similar to those that
they pollinate, where I isolated individual floral characters in order to test bird responses.
Hummingbird-pollinated flowers are thought to have evolved from bee-pollinated
flowers (Thomson & Wilson 2008), and I tested bird behavior as a factor that influenced
the evolution of hummingbird-pollinated flowers. I studied birds’ preferences for floral
color, along with preferences for nectar volume, concentration, and landing platforms. I
also looked at how bird preferences can be altered if floral properties are changed, as well
as if they will use color patterning as a floral cue to find higher nectar volumes. I
recorded bout initiations, percentage of visits to inflorescences, number of probes at an
inflorescence, and handling time as behavioral responses by birds.
Floral characteristics
Flowers adapted to birds are called "ornithophiles." Flowers adapted to bees are
called "melittophiles." Ornithophiles are thought to have frequently evolved from
melittophiles (Wilson et al. 2006; Wilson et al. 2007). The syndromes of floral
differences are thought to be caused by a number of factors, namely pollinator efficiency,
1
floral costs, the deterrence of less beneficial visitors, and the behavioral proclivities of
pollinators.
Some floral characters facilitate efficient pollination by enabling animals to visit
flowers more easily, thereby preventing them from wasting much pollen. Melittophiles
tend to waste pollen by distributing pollen all over the bee’s body, and bees tend to
groom the pollen out of circulation. By comparison, ornithophiles have narrow floral
tubes and exserted anthers and stigmas, which may be adaptations to transfer pollen more
precisely onto the bird's forehead feathers (Thomson et al. 2000; Castellanos et al. 2003;
Castellanos et al. 2004). Floral color promises pollinators nectar rewards, and may also
help pollinators extract nectar more quickly (Baker 1961). Though ornithophiles are
usually red, hummingbirds do not only visit red flowers. They can be conditioned to visit
differently colored flowers, and when other colors are associated with nectar more to
their liking, they show little long-term preference to flower color (Bene 1941; Collias &
Collias 1968; Stiles 1976). Ornithophiles tend to lack landing platforms and to lack
patterns of lines and spots as compared with melittophiles (Faegri and van der Pijl 1979).
It has been suggested that landing platforms would obstruct hummingbirds from
extracting nectar quickly, and would contribute to less precise pollen placement
(Castellanos et al. 2004). Waser and Price (1985) presented evidence that in a
melittophile, patterning represents a nectar guide that speeds up handling time by bees.
The loss of such patterning in species that have become ornithophilous has not been
explained.
Nectar production can be costly for flowers. In one species (Blandfordia nobilis),
plants have been found to expend as much as 37% of their energy on nectar production
2
(Pyke 1991), as well as considerable water (Baker 1975; Pyke & Waser 1981). In
ornithophiles, nectar is the only floral reward. It is a sugar solution that is secreted by
small tissues near the base of the flower, called nectaries (Hale et al. 2008; Cnaani et al.
2006). Hummingbirds rely on cues that allow them to locate which flowers are rewarding
and which are not (Hurly & Healy 1996), and in order to ensure multiple visits from
pollinators and facilitate pollen dispersal, many flowers replenish their nectar throughout
the day (Castellanos et al. 2002).
Although pollinators are not the sole influence on floral evolution, they are widely
considered an important agent of selection (Waser & Price 1983; Waser & McRobert
1998; Irwin et al. 2004; Smith et al. 2008). In order to attract specific pollinators, a flower
is expected to offer nectar (or another reward) that is extremely attractive to its
pollinators. For instance, ornithophiles should have the nectar properties that attract
hummingbirds, and hummingbirds should visit flowers that meet their minimum nectar
preferences. Ornithophiles tend to possess large amounts of dilute, sucrose-rich nectar
(Stiles 1976; Faegri and van der Pijl 1979; Stiles & Freeman 1993; Baker et al. 1998;
Thomson et al. 2000; Castellanos et al. 2004). As examples, Ipomopsis aggregata flowers
produce between 1 and 5 µL of 20-25% sucrose-rich nectar (Pleasants 1983; Irwin &
Brody 2000), and ornithophilous penstemons average 7.76 µL of 26% sucrose-rich nectar
(Wilson et al. 2006). Pollinators should prefer floral characters that allow them to spend
less time while foraging and maximize energetic gain (Waser & Price 1983; Roberts
1995). If characters provide a larger energy gain per unit time for the principal
pollinators, then they should be preferred. If these nectar properties evolved in response
3
to hummingbirds, one should expect hummingbirds to prefer them over alternative
nectars.
Some characters could have evolved because they deter visitors that, in effect, are
floral parasites, whereas others may have evolved in response to pollinator preference.
Ornithophiles lack landing platforms and floral patterns (Faegri and van der Pijl 1979),
are usually red, and have a high amount of less concentrated (i.e., less viscous) nectar
(Roberts 1995). These floral characters may have evolved due to pollinator preference as
a way of reducing handling time (Waser 1982; Waser & Price 1985). Or, they may have
evolved because they deter bees from frequenting the flower, bees being less effective at
transferring pollen than birds (Castellanos et al. 2003). Rodriguez-Girones and
Santamaria (2004) suggest that bees are less likely to visit red flowers because they lack
red color receptors that allow them to easily distinguish between red flowers and green
foliage. If birds are more efficient pollinators than bees, then flowers adapted to birds
should select for characters that would deter pollen-wasteful bees from emptying the
nectaries and scattering the pollen. Under this theory, birds are constantly finding that
there is more nectar in red flowers, which have been under-visited by bees, so they come
to expect the rule to hold.
Bird behaviors are invoked in explaining the evolution of ornithophily, so the
experiments that follow tested the assumptions behind these explanations about how
hummingbirds responded to differences in the flowers they visit. My Discussion
considers other evolutionary causes as well.
Hummingbird foraging
4
Hummingbirds are thought to visit flowers based on the knowledge they have
obtained in previous visits (Hurly 1996), remembering which flowers (or patches of
flowers) have been more rewarding and which have been less. Hurly and Healy (1996)
found that hummingbirds tend to rely on the location of a nectar source rather than visual
cues associated with rewards when they are visiting a patch of four flowers. In larger
patches, one might see a different outcome. Hummingbirds are conditional learners and
may associate hidden floral rewards (i.e., nectar) with floral advertisements (i.e., petal
color). The experimental alteration of certain floral characters is expected to cause birds
to change preferences. For example, having a stronger preference for sugar properties
(e.g., sugar type, nectar concentration, nectar volume, etc.) would presumably override
any pre-existing color preference a bird might have toward a single color (Stiles 1976;
Melendez-Ackerman et al. 1997; Rodriguez-Girones & Santamaria 2004), conditioning
them to visit colors unlike what they experience in nature (Stiles 1976; Collias & Collias
1968).
Hummingbirds hover while foraging, and as they hover, they expend a high
amount of energy (Welch & Suarez 2008), calories that they must replenish quickly
(Tyrrell & Tyrrell 1985; Powers & Nagy 1988; Gill 1995). They sustain themselves
throughout the day by consuming nectar, a high-caloric food that is their primary energy
source (Baker & Baker 1983b), though they also eat insects for protein when flowers are
not available (Tyrell & Tyrell 1985). If nectar is limiting, hummingbirds will defend
nectar-rich territories, thereby expending even more energy (Kodric-Brown & Brown
1978; Tyrrell & Tyrrell 1985). There are usually flowers in a patch that act as mimics,
providing no nectar for pollination services (Brown & Kodric-Brown 1979), so one
5
should expect birds to visit flowers that they have learned provide “enough” food to make
visiting worthwhile. In other words, there ought to be a threshold in energy gain divided
by handling time above which hummingbirds should include a flower type in their diet
(Stephens & Krebs 1986).
Though birds remember cues related to recent visits that led them to rewarding
flowers (Hurly 1996; Hurly & Healy 1996), they also sample a variety of new flowers
unlike their normal flowers (Hurly 1996; Waser & McRobert 1998). When they find a
rewarding flower, they may change preferences in a matter of minutes, associating
different cues with preferred rewards, even cues that are contrary to the usual floral
syndrome. Sampling different flowers may cause them to find nectar in unlikely places
(i.e., in purple flowers that, contrary to syndrome rules, are offering substantial nectar
rewards), causing them to visit all types of flowers and constantly testing which flowers
are more rewarding and which flowers are less (Stiles 1976; Wells 1993; Hurly 1996;
Hurly & Healy 1996). In other words, it is thought that a pollinator’s willingness to visit
flowers of all types is a way to find nectar rewards that their own recent experience
would not have led them to expect (Smithson 2001).
As a starting point, suppose foraging involves some sampling but more
optimizing. A bird coming upon a mixed array of two types of flowers, both types with
rewards above the minimum threshold of acceptability, eventually samples both types. If
a flower is more rewarding than in the bird’s recent experience, then it spends more time
at that inflorescence assiduously probing its other flowers (Roberts 1996). Also enhanced
for a while is its willingness to visit other individuals of the same type (Bateson et al.
2003). As the bird works the patch, it gains knowledge of not only the rewards but also of
6
the time it takes to handle the flowers even if the energetic rewards are equal, and the bird
might adjust its preferences accordingly, visiting the two types of flowers in proportion to
their value.
It is unclear how much this proportional preference would carry through to a
subsequent bout after the bird had spent some time perching, preening, or chasing away
competitors. Birds might show differences between the preference they have when they
initiate a bout versus the flowers they visit as they proceed through their bout. Perhaps at
bout initiation, they are more likely to go to a flower that looks ornithophilous, then as
they work, they are more likely to sample widely and adjust their preference within a
bout to the kind of flower that is most rewarding (Waser & McRobert 1998). Or, perhaps
at bout initiation they are more likely to try something new, then once they get a search
image fixed in their mind, they might be consistent for the rest of the bout (Hurly &
Healy 1996).
Birds leave flower patches, either because most nectar has been depleted or they
are momentarily satiated, but their recent experiences tend to drive their preferences for
floral characters that will lead them to rewards in future visits (Hurly & Healy 1996;
Fenster et al. 2004). Birds, as well as bees, tend to spend more time probing each
individual flower on an inflorescence when they find rewarding flowers (Thomson et al.
1982; Thomson et al. 1987; Thomson 1988), which causes them to have a strong
attraction for specific cues (Melendez-Ackerman et al. 1997). After some time, birds
associate certain floral cues with how large rewards are, as well as how much energy they
expend during a visit (Bene 1941; Collias & Collias 1968; Blem et al. 2000). Bees use a
gradient scale associated with the simplicity of cues (Gould 1993). For example, Gould
7
(1993) found that bees tend to have a hierarchy of cues, where bees first rely on colors to
find food sources, then they will rely on patterns, especially those located on the edge of
flower petals, and will finally use floral shape to find nectar sources. Birds might have a
version of this hierarchy of cuing.
Birds should prefer floral characters that enable them to visit fewer flowers for the
same energetic reward. When birds begin to make quick observations of which flowers
provide higher rewards and more energy, they begin to have decreased handling time
from previous visits. When they begin to recognize those characters that provide them
with decreased handling time, they may show preferences for those characters. However,
these preferences may change quickly, because, as with color (Stiles 1976), they may
have a short holdover period for other characters, and they continuously sample various
flower types (Waser & McRobert 1998).
Goals
The following experiments tested not only bird preferences for certain flower
properties, they also tested how birds can alter their preferences, given certain associated
incentives. First, I simply tested bird preferences for flower color. Second, I tested bird
preference for nectar volumes. Third, I offered birds two nectar types, where one was
more similar to the nectar of melittophiles and the other was more similar to the nectar of
ornithophiles. When they showed a preference, I ran subsidiary experiments that tested
whether they preferred a certain concentration or sugar type. Fourth, I tested if birds
showed a preference for the presence or absence of landing platforms. Finally, I tested if
birds would use spots on the flower as a cue to find the higher nectar source. These
8
experiments tested if birds had preferences, and if they would alter their preferences
when other properties were changed.
9
GENERAL EXPERIMENTAL SET-UP
Materials and methods
The work involved hummingbirds of several species, sexes, and maturity levels:
male and female Black-chinned Hummingbirds (Archilochus alexandri), male and female
Anna’s Hummingbirds (Calypte anna), and female Rufous Hummingbirds (Selasphorus
rufus) visited the experimental array. Throughout experiments, species and, when
distinguishable, sex were recorded, but analyses were done pooling together all birds (see
"Critique of experimental design" below). Observations were done in the southern Sierra
Nevada of California (36o N, 118o W) during the summers of 2008 and 2009.
Artificial inflorescences were assembled in a 4 × 4 array (Figure 1a). In the array,
inflorescences were set 1 m apart from one another with the second row placed 1 m
diagonally from the front row. The array was set up as an alternating pattern where
inflorescences were arranged in a lattice of equilateral triangles. With an alternating
pattern, I ensured birds came into contact with two flower types in equal frequencies. If I
had used a randomized pattern, the two flower types may not have been located next to
each other. If this were the case, birds would not be exposed to differing flower types at
the same time. At the points of the lattice, metal pipes were hammered into the ground, so
the artificial inflorescences could be swapped out quickly by inserting a new
inflorescence into its pipe (Figure 1b).
Each artificial inflorescence had five flowers aligned vertically along a wooden
stake, each separated from the next by 7.62 cm (Figure 1c). Where each flower was
placed, a hole was drilled with a 15/64" bit. Into the holes, 200 µL PCR tubes were
inserted (Figure 1d), acting as nectar spurs in the center of an artificial flower. Artificial
10
nectar was pipetted into the nectar spurs. In each experiment (except those that varied
concentration) I used the same nectar concentration. Among experiments, the
concentration I used varied (unintentionally) between 20 and 25% sucrose (which is
naturally seen in ornithophiles: Pleasants 1983; Irwin & Brody 2000; Wilson et al. 2006).
Animal-pollinated flowers usually have brightly colored sepals and/or petals (collectively
called the perianth) that advertise a reward (Waser 1983). Each artificial flower had an
artificial perianth constructed from ribbon (Floral Satin 2 ½”) and cut in the form of a star
(Figure 1c). Several experiments contained differently-colored perianths (red, purple, or
yellow), and color was often a cue that was offered to the birds by which they might
associatively learn which kind of flower contained which kind of nectar. By giving birds
different choices and watching their decisions, I measured their preferences.
There were three reasons why I chose to have inflorescences with multiple
flowers. First, to attract a bird I presumed that I needed to offer a large number of
rewards. Second, by having five perianths, there was a degree of replication for each
inflorescence. When a bird arrived at an inflorescence, it had the choice of sampling from
zero to five flowers, or even revisiting some of the flowers. Lastly and most importantly,
the protocol was designed to offer birds many small rewards similar to what they would
find in real flowers, rather than having them choose between a few bonanza feeders
(Figure 1e & f). Gass and Roberts (1992) showed that birds have a higher licking rate
when visiting high-volume feeders over the lower-volume (more natural) flowers,
thereby changing natural behaviors when visiting unnatural food sources (i.e., highvolume bonanza feeders). The volume of a potential food source offered to foraging birds
determines the net energy gain of foraging birds as well as the amount of nectar that
11
adheres to the tongue with each lick (Roberts 1995). Unlimited food sources change
behaviors of foraging birds, causing them to linger around a feeder longer than they
would while foraging in natural conditions (Roberts 1996). Therefore, having bonanza
feeders would likely distort the behavior of the birds.
Observations were conducted for four hours in the morning (6:00 – 10:00 a.m.)
and two hours later in the day (5:00 – 7:00 or 6:00 – 8:00 p.m.). Each observation period
was 40 minutes long. At the end of 40 minutes, when I was moving on to a new
observation period, 16 new stakes with newly-filled PCR tubes replaced the stakes that
had just been observed (Figure 1b). Twenty minutes were taken to exchange the stakes
from one observation period to the next, making a unit of observation 1 hour. There are
two reasons for replacing 40-minute old flowers with fresh nectar tubes. First, this
ensured that all flowers had the same amount of nectar when starting each new hour. I
knew the exact amount of nectar in each nectar tube at the beginning of the hour. If the
stakes were not replaced, I would have had no way of figuring the amount of nectar in
each of the nectar tubes because not all of the tubes were emptied thoroughly. By
controlling the amount of nectar in each tube, I was able to assume that bird behavior was
based on the controlled experimental offerings. Second, as was noticed in the preliminary
experiments, hummingbirds are daring animals and will come up to a researcher with
very little hesitation. In order to minimize them drinking while flowers are being filled
and choices are unequal (i.e., interference from the researcher), it was important to
replace stakes quickly with already refilled nectar tubes.
The position of stakes with differing treatments (e.g., red versus purple flowers)
was alternated and repositioned at the beginning of every experimental day (not at each
12
observational hour). Positions of flower types were switched each day to ensure that birds
did not simply memorize the location of a nectar source, but instead based any difference
in behavior on the experimental difference: color, nectar amount, nectar concentration,
sugar type, corolla morphology, or color patterning.
b.
a.
d.
g.
c.
e.
f.
h.
i.
Figure 1 a. 4 × 4 array with alternating pattern; b. exchanging stakes between observation periods; c.
inflorescences with 5 artificial flowers made of ribbon; d. “nectar spur”: 200 µL PCR tube; e. unnatural
behavior of a bird at a bonanza feeder; f. natural behavior of a bird at an inflorescence; g. landing platform
flower (similar to melittophile); h. non-landing platform flower (similar to ornithophile); i. inflorescence
with 5 patterned flowers (similar to melittophile).
13
Statistics
There were four dependent variables recorded (Wilson and Jordan 2009):
1. The first inflorescence visited in a bout was recorded. Initial visits are important to
record because this shows bird preference when they first arrive at the array based
only on knowledge obtained from previous bouts (i.e., what they might choose in
nature). It also shows me that their preferences can be altered after a short period of
time, whether they show no preference or if they begin to show a preference for
what is offered.
2. The sequence of inflorescences visited after the initial visit was recorded; I later
simplified this to the percentage of inflorescences of each type visited. The
percentage of inflorescence type visited is important in understanding their overall
preferences and choices. Even though they might not have shown a preference in
terms of bout initiations, quick learning by birds could still amount to a big
response in their overall preferences for a flower type.
3. The number of probes at each inflorescence was recorded. Were birds more
thoroughly probing inflorescences that were more rewarding? Knowing the answer
could shed light on the reason for preferences at the inflorescence or bout-initiation
level.
4. The duration of stay at an inflorescence was recorded by video, and given the
number of probes and the amount of time each bird stayed at an inflorescence, I
calculated the average time spent per probe, which I call “handling time.” I use
handling time as a comparison between flower types, not as a direct caloric expense
measurement. For example, in my color experiment, I compared how long birds
14
took to extract nectar from red flowers versus purple flowers, though I do not know
how much energy was used while foraging. This allows me to compare importance
of flower type while foraging. Scrutiny of residuals compelled me to do analyses on
the log(handling time per probe).
I addressed five hypotheses. All had the same general form of a null hypothesis.
HO: Manipulations in flower characteristics (landing platform shape, spotting patterns,
nectar volume, concentration, and amount) do not elicit preferences by hummingbirds.
There would be a 1:1 visiting ratio in each experiment, both for bout initiations and for
visitation at all inflorescences, and birds would probe inflorescences of varying floral
types with equal frequencies and spend the same amount of time handling the flowers per
probe.
1. HA: However, based on the systematic differences between ornithophiles and
melittophiles, along with past experimental results, I would expect hummingbirds to visit
red flowers over other colored flowers. Even though hummingbirds can see all colors
well (Goldsmith and Goldsmith 1979; Stiles 1981), I would expect hummingbirds to
spend more time at flowers that they are used to finding rewarding (Grant & Grant 1968).
2. HA: After a short period of time, birds ought to learn to prefer any color
associated with higher volumes of nectar. Birds will spend more time at flowers that have
more nectar in them.
3. HA: When presented with high versus low concentrations of nectar, both
containing the same amounts of sugar, I predicted birds will show a preference for
flowers that contain higher volume (less concentrated) nectar. Flowers that have a higher
15
volume of (more dilute) nectar will have a decreased handling time because their nectar is
less viscous (Waser 1983; Harder 1986; Cnaani et al. 2006).
4. HA: Birds will show a preference for flowers that lack landing platforms. Birds
will have increased handling time when visiting flowers with landing platforms.
5. HO: In the final experiment on nectar guides, I predicted that the null
hypothesis would be accepted. Because ornithophilous flowers often lack nectar guides
compared to close relatives that are melittophilous, I expected hummingbirds to not
distinguish patterns on the perianth, even when doing so would allow them to prefer
rewarding flowers.
Bout initiations and % inflorescences are categorical variables, so for them I used
goodness-of-fit tests. Heterogeneity in goodness-of-fit was tested to see if days were
similar enough to be pooled (P>0.25), or if they had to be separated because they might
be different. When pooling was permissible, I report on the pooled data.
Number of probes and handling time are quantitative variables, so I used
ANOVA. Before doing the tests, I searched for outliers and winsorized them to the next
highest number. The ANOVAs often started out split-plot mixed-models because day was
a random variable whereas treatments such as colorcue, nectartype and flowertype were
fixed variables. Days were nested within color cue and pattern cue. When P>0.25, terms
were sequentially pooled, and the final model was presented.
16
EXPERIMENT 1: COLOR PREFERENCE
Methods: The color of a flower is often part of a pollination syndrome. Reds and
oranges are associated with ornithophily. In order to test a bird’s preference for a specific
flower color, I set up the 4 × 4 array with half of the stakes having red flowers, and the
other half of the stakes having purple flowers (Experiment 1A). Equal amounts of nectar
(2.5 µL of 23% sucrose solution) were pipetted into each of the artificial flowers prior to
the beginning of each hour of study. There was no difference in nectar volume,
concentration, or chemistry between flower colors. Flower type was alternated within the
array, with the first row being red-purple-red-purple, the second row being purple-redpurple-red, etc. On the next experimental day, the color of alternating stakes was
switched, with the first row being purple-red-purple-red, the second row being redpurple-red-purple, etc. Experiments were run for at least four consecutive days. Two
similar experiments were done with different color combinations. One used yellow versus
red perianths (Experiment 1B). The other used purple versus yellow perianths
(Experiment 1C). Experiment 1 was largely a test of what could complicate other
experiments. For instance, if I showed no color bias then the interpretation of other
experiments would be simplified.
Results for red versus purple: Figure 2 suggests that birds generally had a slight
preference for red. Bout initiations: On the birds’ first visit to the array during days 1-3,
they significantly visited red flowers over purple flowers (P<0.001; 0.013; 0.003). On
days 4-6, birds did not show a significant initial preference for either color. After running
G2 tests-of-heterogeneity, days could not be pooled. Inflorescences visited: On days 1 and
2, birds significantly preferred red inflorescences (Pday1;day2<0.001; =0.012), but after day
17
3, birds showed no significant preference for either flower color. G2 tests-ofheterogeneity indicated that days could not be pooled. Probes: The interaction of color ×
day yielded P=0.032. It seemed that on certain days when purple inflorescences attracted
fewer visits, the birds probed the purple flowers more thoroughly. Overall, however,
birds did not significantly probe purple more or less than red (Pcolor=0.208). Days did
vary in the amount of probing (Pday=0.007). Handling time: The interaction of color ×
day yielded P=0.161 and could not be pooled. Generally, birds spent significantly more
time visiting red over purple (Pcolor=0.009), and days varied in the amount of time birds
spent per flower (Pday<0.001).
18
Bout initiations
Day G2
1 11.252
2 6.198
3 8.547
4 0.505
5 0.931
6 1.159
P
<0.001
0.013
0.003
0.477
0.335
0.282
Inflorescences visited
Day G2
P
1 12.498 <0.001
2 6.252
0.012
3 2.404 0.121
G2
P
df
4 0.091
0.763
5 1.944 0.163 Heterogeneity 6.893 0.229 5
6 2.994 0.084 All days pooled 19.290 <0.001 1
G2
P
df
Heterogeneity 13.048 0.023 5
All days pooled 15.544 <0.001 1
100%
100%
90%
90%
purple
% bout initiations
% inflorescences visited
80%
70%
60%
50%
40%
30%
20%
red
10%
0%
purple
80%
70%
60%
50%
40%
30%
20%
red
10%
day1
n=17
day2
n=9
day3
n=11
day4
n=8
day5
n=27
0%
day6
n=14
Number of probes at an inflorescence
Color
Day
Color x day
Error
5.5
probes
4.5
4
2.5
day4
n=44
day5
n=149
day6
n=86
Log (handling time per probe)
MS
P
0.111 0.009
0.315 <0.001
0.025 0.161
0.016
0.15
Day
2
3
0.1
6
4
1
5
purple
day3
n=42
df
Color
1
Day
5
Color x day 5
Error
430
3.5
3
day2
n=42
df
MS
P
1 17.767 0.208
5 11.294 0.007
5 8.517 0.032
430 3.457
log (handling time)
5
day1
n=79
0.05 Day
1
0 2
-0.05 5
4
-0.1 6
3
-0.15
red
purple
red
Figure 2. Red versus purple flowers. (●: red flowers; □: purple flowers)
Results for red versus yellow: Again, red was generally preferred (Figure 3). Bout
initiations: Heterogeneity G2 tests allowed all days to be pooled (P=0.415). Birds showed
a significant initial preference for red flowers over yellow flowers (P=0.004).
Inflorescences visited: The heterogeneity G2 test yielded P=0.004. On days 2 and 3, birds
significantly preferred red flowers, but they did not show a significant preference for
either color on days 1 or 4. Probes: One datum was winsorized from 12 probes to 9
probes. The interaction of color × day yielded P=0.594, so the term was pooled. Birds did
19
not significantly discriminate between colors (Pcolor=0.999). Different days yielded
different intensities of probing (Pday=0.048). Handling time: The interaction of color ×
day yielded a P=0.810. The simplified analysis shows that birds spent significantly more
time visiting red compared to yellow flowers (Pcolor=0.001), and days varied in how much
time birds spend visiting a flower (Pday=0.030).
Bout initiations
Day
G2
1
0
2 5.822
3 4.615
4 0.731
Inflorescences visited
Day G2
P
1 0.0323 0.857
G2
P df
2
9.537 0.002
Heterogeneity 13.188 0.004 3
3 11.453 0.001
4
0.43 0.508 All days pooled 8.274 0.004 1
P
1
G2
P df
0.016
Heterogeneity 2.850 0.415 3
0.032
0.392 All days pooled 8.317 0.004 1
100%
100%
90%
90%
yellow
% inflorescences visited
% bout initiations
80%
70%
60%
50%
40%
30%
red
20%
10%
yellow
80%
70%
60%
50%
40%
30%
red
20%
10%
0%
0%
all days pooled
n=65
day1
n=31
Number of probes at an inflorescence
Color
Day
Error
df
1
3
362
Color
Day
Error
log (handling time)
probes
df
MS
P
1
0.217 0.001
3
0.052 0.030
362 0.017
Day
-0.05 3
Day
4 1
2
4
3
2.5
2
day4
n=146
0
4.5
3
day3
n=128
Log (handling time per probe)
MS
P
0
0.999
7.817 0.048
2.931
5
3.5
day2
n=62
-0.1 4
2
1
-0.15
-0.2
-0.25
red
-0.3
yellow
red
yellow
Figure 3. Red versus yellow flowers. (●: red flowers; ∆: purple flowers)
Results for purple versus yellow: Birds had little preference for either color
(Figure 4). Bout initiations: On the birds’ first visit to the array during days 1 and 3, they
20
significantly visited purple flowers over yellow flowers (P=0.046; 0.008). On days 2 and
4, birds did not show a significant initial preference for either color. The G2 test-ofheterogeneity indicated that days could not be pooled (P=0.004). Inflorescences visited:
Birds showed no preference for flower color. The G2 test-of-heterogeneity indicated that
all days could be pooled (P=0.992). Probes: The interaction of color × day yielded a
P=0.972. The final analysis shows that birds did not significantly probe one color more
than the other (Pcolor=0.891), although days varied in assiduousness (Pday=0.007).
Handling time: The interaction of color × day yielded a P=0.263, so this term was
pooled. The analysis shows that birds did not spend significantly more time visiting one
flower color over the other (Pcolor=0.575), although handling times varied by day
(Pday=0.006).
21
Bout initiations
Day
1
2
3
4
G2
3.985
1.493
7.103
1.836
P
0.046
0.222
0.008
0.175
Inflorescences visited
Day
1
2
3
4
G2
P df
Heterogeneity 13.303 0.004 3
All days pooled 1.113 0.291 1
G2
0.046
0.006
0.017
0.237
100%
100%
yellow
90%
90%
% inflorescences visited
80%
% bout initiations
P
0.831
0.939
G2
P
df
0.896 Heterogeneity 0.096 0.992 3
0.627 All days pooled 0.209 0.648 1
70%
60%
50%
40%
30%
20%
purple
10%
0%
day1
n=21
day2
n=17
70%
60%
50%
40%
30%
20%
day3
n=25
0%
day4
n=27
df
Color 1
Day
3
Error 804
5
0
4.5
-0.05
log (handling time)
probes
all days pooled
n=809
Log (handling time per probe)
df
MS
P
Color 1
0.064 0.891
Day
3 13.916 0.007
Error 804 3.448
Day
2
3.5 4
3
1
3
2.5
2
purple
10%
Number of probes at an inflorescence
4
yellow
80%
MS
0.004
0.048
0.012
P
0.575
0.006
Day
-0.1 2
1
3
-0.15 4
-0.2
-0.25
purple
-0.3
yellow
purple
Figure 4. Purple versus yellow flowers. (●: red flowers; ∆: purple flowers)
22
yellow
EXPERIMENT 2: NECTAR VOLUME EXPERIMENT
Methods: This experiment was done in order to test a hummingbird’s preference
for different nectar amounts associated with different colors. Nectar of different amounts
(0 versus 2 µL; 0 versus 4 µL; 0 versus 6 µL; 2 versus 4 µL; 2 versus 6 µL; 4 versus 6
µL) was placed into differently colored artificial flowers. Differing amounts of 23%
sucrose solution were pipetted into each of the artificial flowers prior to the beginning of
each hour of study. In this experiment birds were using floral color as a cue to find
nectar. On half the days, red flowers contained one nectar volume and purple flowers
contained the other nectar volume, and on the other half of the days the color of the cue
was reversed. Each nectar volume combination was set up for a span of four experimental
days. As usual, each experimental day consisted of six forty-minute intervals. For
example, to compare 0 µL and 6 µL, I did as follows. On the first experimental day, 6 µL
was pipetted into red flowers, and 0 µL was in purple flowers, with the first row showing
a red-purple-red-purple arrangement. Measurements were then taken and the birds were
videotaped. On the second experimental day, 6 µL was pipetted into purple flowers and 0
µL was in red flowers, and the location of the colored flowers remained in the same
position (first row still red-purple-red-purple). The location of the nectar volume and the
type of flower the nectar volume occupied was switched. On the third experimental day,
6 µL was pipetted into purple flowers and 0 µL was in red flowers, but the location of
flower colors was switched (row 1 was purple-red-purple-red). On the fourth
experimental day, 6 µL was pipetted into red flowers and 0 µL was in purple flowers, but
the location of the colored flowers remained in the same position as on day three (first
row purple-red-purple-red). Because flower color was the conditioned cue, the color that
23
the birds choose predicted the amount of nectar in the artificial flower. A goodness-of-fit
test-of-heterogeneity was used to see if specific days (e.g., the two days of red@8 µL)
could be pooled. If both types of days could be pooled, then another goodness-of-fit testof-heterogeneity was used to see if days with opposing color cues could be pooled.
Results for 0 versus 2 µL: Birds generally preferred 2 µL over no nectar (Figures
5-6; Tables 1-2). Bout initiations: On first visits to the array during day 2, birds
significantly visited red flowers that had 2 µL over purple flowers that had 0 µL
(P=0.046). On days 1, 3, and 4, birds did not show a significant initial preference. After
running G2 tests-of-heterogeneity, red@2 could be pooled (P=0.795), but purple@2
could not be (P=0.077), hence I pooled neither. Inflorescences visited: Birds significantly
preferred 2 µL on days 2-4 (P<0.05 for all days), but showed no significant preference
for the nectar source on day 1 when it was in purple flowers (P=0.371). Birds showed a
stronger preference for the flower with nectar if it was located in red flowers. After
running G2 tests-of-heterogeneity, purple@2 could be pooled, but I was unable to pool
red@2. Neither type of day was pooled. Probes: There was an interaction between
nectartype and days, which suggests that birds significantly probed 2 µL over 0 µL
(Pnectartype=0.053), but the degree of probing differed by day (Pnectartype × day ⊂
colorcue=0.005).
Birds probed 2 µL over 0 µL, regardless of color (Pcolorcue × nectartype=0.467),
and they showed no preference for either color offered (Pcolorcue=0.583). Replicated days
did not vary by number of probes (Pday ⊂ colorcue=0.861). Handling time: Although there
was no interaction between the type of nectar offered and day, the interaction of
nectartype × day ⊂ colorcue yielded P=0.151, so this term could not be pooled.
Nectartype × colorcue yielded P=0.400. Birds did not spend more or less time at a nectar
24
type or a specific color (Pnectartype=0.714; Pcolorcue=0.792), although replicate days varied
in handling time (Pday ⊂ colorcue<0.001).
Results for 0 versus 4 µL: Birds preferred 4 µL over 0 µL when the nectar was in
red flowers (Figures 5-6; Tables 1-2). Bout initiations: There was no initial preference
observed by birds toward either flower type (P=0.099; 0.316). After running G2 tests-ofheterogeneity, replicated days were pooled (Pred@4;purple@4=0.372; 0.869). Inflorescences
visited: Birds significantly visited 4 µL over nectarless flowers, when the 4 µL was in red
flowers (P<0.05 for both days). Birds showed no preference for nectar when the nectar
was in purple flowers (P>0.05 for both days). After running G2 tests-of-heterogeneity,
purple@4 could be pooled (P=0.695), but red@4 could not (P=0.072), I pooled neither.
Probes: The interaction of nectartype × day ⊂ colorcue yielded P=0.516, so I pooled this
term. Then, nectartype × colorcue yielded P=0.254, so again I pooled. The simplified
analysis shows that birds significantly probed 4 µL over 0 µL (Pnectartype<0.001), and they
showed no preference for the colors offered (Pcolorcue=0.559). Days varied in the amount
birds probed flowers (Pday ⊂ colorcue=0.052). Handling time: The interaction of nectartype ×
day ⊂ colorcue yielded P=0.648. Then, nectartype × colorcue yielded P=0.390. After
pooling, birds significantly probed 4 µL over 0 µL (Pnectartype<0.001), and this was
without a color preference (Pcolorcue=0.511). Days varied (Pday ⊂ colorcue<0.001).
Results for 0 versus 6 µL: Birds generally preferred 6 µL over flowers with no
nectar (Figures 5-6; Tables 1-2). Bout initiations: On first visits to the array, birds only
showed a significant preference for 6 µL over 0 µL on one day when 6 µL was in red
flowers (P=0.001), but there was no significant preference shown on any other day.
Looking at G2 tests-of-heterogeneity, purple@6 could be pooled (P=0.291), but red@6
25
could not be (P=0.097). Inflorescences visited: On days 1, 2 and 4, birds significantly
visited 6 µL over 0 µL, regardless of color (P≤0.01 for all days). However, on day 3,
when purple had 6 µL, no preference was observed for either nectar type (P=0.850). The
G2 tests-of-heterogeneity showed that purple@6 could not be pooled (P<0.001) and
red@6 could not be pooled (P=0.058). Probes: The interaction between nectartype and
day was marginally significant (Pnectartype × day ⊂ colorcue=0.057). There was no interaction
between nectartype and colorcue (Pnectartype×colorcue =0.250). Days were also not very
different with respect to the number of probes (Pday ⊂ colorcue=0.389). Birds might have
probed flowers that contained 6 µL more than flowers with 0 µL (Pnectartype=0.060), and
there was no significant preference of flower color (Pcolorcue=0.758). Handling time: The
interaction of nectartype × day ⊂ colorcue yielded P=0.557, so this term was pooled.
Birds spent more time at flowers with 6 µL than at flowers with 0 µL (Pnectartype<0.001).
There was no significant preference for color (Pcolorcue=0.791), and birds spent more time
in red flowers with 6 µL than purple flowers with 6 µL (Pcolorcue × nectartype=0.003). Days
did not vary in handling time (P day ⊂ colorcue=0.133).
Results for 2 versus 4 µL: Figures 5-6 and Tables 1-2 show that birds had no
preference when offered volumes that differ by 2 µL. Bout initiations: Birds initially
showed no significant preference for 4 µL, though on day 1 birds showed marginal
preference for red flowers with 2 µL over purple flowers with 4 µL (P=0.056). After
running G2 tests-of-heterogeneity, neither purple@4 (P=0.079) nor red@4 (P=0.115)
could be pooled. Inflorescences visited: Birds did not significantly prefer 2 µL or 4 µL
(Pall days pooled=0.595). After running G2 tests-of-heterogeneity, purple@4 could be pooled
(P=0.591), red@4 could be pooled (P=0.416), then all days were poolable (P=0.650).
26
Probes: The interaction of nectartype × day ⊂ colorcue yielded P=0.928. Then,
nectartype × colorcue yielded P=0.883. After pooling, days varied when looking at
number of probes (Pday ⊂ colorcue<0.001), birds significantly probed 4 µL over 2 µL
(Pnectartype=0.025), and they did not probe a specific color more frequently
(Pcolorcue=0.586). Handling time: One datum was winsorized from 0.6186
log(seconds/flower) to 0.4401 log(seconds/flower). The interaction of nectartype × day ⊂
colorcue yielded P=0.957. Then, the interaction of nectartype × colorcue yielded
P=0.546. Day ⊂ colorcue yielded P=0.335. In the final analysis, birds spent significantly
more time at flowers with 4 µL than flowers with 2 µL (Pnectartype<0.001), though they did
not spend more or less time in a single color (Pcolorcue=0.674).
Results for 4 versus 6 µL: Again birds had no preference when offered volumes
that differ by 2 µL (Figures 5-6; Tables 1-2). Bout initiations: When looking at first
visits, bird significantly preferred flowers with 6 µL over ones with 4 µL on days 2 and 4,
(Pday2;4=0.025;0.038), but there was no initial preference shown on days 1 or 3. After
running G2 tests-of-heterogeneity purple@6 could not be pooled (P=0.132) and red@6
could not be pooled (P=0.011). Inflorescences visited: Running G2 tests-of-heterogeneity
allowed all days to be pooled (P=0.870). Birds showed no significant preference for
either nectar type (Ppooled=0.773), regardless of color. Probes: One datum was winsorized
from 14 probes to 10 probes. The interaction of nectartype × day ⊂ colorcue yielded
P=0.966. Then, nectartype × colorcue yielded P=0.591. Finally, day ⊂ colorcue yielded
P=0.820. After pooling, birds significantly probed more often into flowers with 6 µL over
flowers with 4 µL (Pnectartype=0.032), though they did not probe into a specific color more
often (Pcolorcue=0.453). Handling time: The interaction of nectartype × day ⊂ colorcue
27
yielded P=0.467, so this term was pooled. There was a marginal effect seen with
nectartype and colorcue (Pnectartype × colorcue=0.095), where the color affected how long
birds stayed at a specific nectar type. Birds generally spent significantly more time at
flowers with 6 µL over flowers with 4 µL (Pnectartype=0.041), though they did not spend
more time at a specific floral color (Pcolorcue=0.859). Days varied significantly in how
much time birds spent at a flower (Pday ⊂ colorcue<0.001).
Results for 2 versus 6 µL: On half of the days, birds preferred flowers with 6 µL
over flowers with 2 µL, regardless of color (Figures 5-6; Tables 1-2). Bout initiations: On
day 2, where red contained 6 µL, birds showed an initial significant preference for 6 µL
(P=0.025), but there was no significant preference observed on other days. After running
G2 tests-of-heterogeneity, purple@6 could be pooled (P=0.680), but red@6 could not
(P=0.066). Inflorescences visited: On days 1 and 3, birds significantly visited 6 µL over 2
µL (Pday1;day3=0.032;0.008), but showed no significant preference on days 2 and 4. After
running G2 tests-of-heterogeneity, neither purple@6 (P=0.169) nor red@6 (P=0.206)
could be pooled. Probes: The interaction of nectartype × day ⊂ colorcue yielded P=0.246,
and the term was pooled even against recommendation. Nectartype × colorcue yielded a
P=0.714, so this term was pooled as well. However, day ⊂ colorcue yielded a P=0.106,
so I did not pool. Birds significantly probed flowers that contained 6 µL more than
flowers that contained 2 µL (Pnectartype<0.001), but they did not probe a specific color
more thoroughly (Pcolorcue=0.819). Handling time: Though there was no significant
interaction between nectartype and day (Pnectartype × day ⊂ colorcue=0.140), I could not pool.
Nectartype × colorcue yielded a P=0.400. Birds did not spend significantly more time at
28
any volume (Pnectartype=0.101) or flower color (Pcolorcue=0.361). However, the time birds
spent at flowers varied by day (Pday ⊂ colorcue =0.022).
29
100%
% inflorescences visited
90%
0 µL
0 µL
0 µL
2 µL
4 µL
6 µL
80%
70%
60%
50%
40%
30%
20%
10%
0%
100%
90%
0 µL
p@2 r@2 r@2 p@2
p@4 r@4 r@4 p@4
n=80 n=128 n=78 n=104 n=77 n=158n=165 n=181
100%
90%
% bout initiations
70%
60%
2 µL
50%
40%
30%
20%
2 µL
10%
0%
% inflorescences visited
0 µL
80%
2 µL
p@6 r@6 p@6 r@6
n=157 n=122n=112 n=46
2 µL
80%
70%
60%
50%
40%
30%
20%
10%
0%
p@2 r@2 r@2 p@2
n=19 n=21 n=9 n=20
4 µL
6 µL
all days pooled
n=908
p@6 r@6 r@6 p@6
n=126 n=89 n=137 n=69
100%
0 µL
90%
100%
2 µL
90%
70%
60%
4 µL
50%
40%
30%
20%
0%
4 µL
4 µL
10%
p@4
n=49
r@4
n=45
% inflorescences visited
% bout initiations
80%
0 µL
2 µL
4 µL
70%
60%
50%
40%
30%
20%
6 µL
10%
p@4 r@4 r@4 p@4
n=28 n=35 n=34 n=23
0%
100%
90%
80%
all days pooled
n=591
4 µL
% bout initiations
80%
70%
60%
6 µL
50%
40%
30%
20%
10%
0%
6 µL
6 µL
6 µL
p@6 r@6 p@6 r@6
n=21 n=21 n=18 n=19
p@6 r@6 r@6 p@6
n=25 n=17 n=21 n=16
p@6 r@6 r@6 p@6
n=15 n=25 n=27 n=24
Figure 5. Different volume offerings.
30
Inflorescences visited
0 vs. 2 µL
Day
1. purple@2
2. red@2
3. red@2
4. purple@2
0 µL
G2
1.331
3.985
1.019
1.828
Bout initiations
purple@4
heterogeneity
pooled
red@4
heterogeneity
pooled
Day
1. purple@4
2. red@4
3. red@4
4. purple@4
2 vs. 4 µL
P
0.413
0.654
0.069
0.548
G2
P
0.027 0.869
1.003 0.316
0.798 0.372
2.716 0.099
Day
1. purple@4
2. red@4
3. red@4
4. purple@4
P
0.056
0.397
0.168
0.531
G2
P
purple@4
heterogeneity
pooled
red@4
heterogeneity
pooled
0 vs. 6 µL
Day
1. purple@6
2. red@6
3. red@6
4. purple@6
0.291
0.261
0.097
0.001
G2
purple@6
heterogeneity
pooled
red@6
heterogeneity
pooled
G2
0.019
0.239
0.431
0.750
P
0.892
0.625
0.511
0.386
G2
P
0.289 0.591
0.481 0.488
0.662 0.416
0.008 0.929
0.207 0.650
0.282 0.595
Day
1. purple@6
2. red@6
3. red@6
4. purple@6
G2
0.067
5.010
1.836
4.297
G2
P
0.171 0.680
1.201 0.273
3.368 0.066
1.697 0.193
Table 1. Different volume offerings.
31
purple@6
heterogeneity
pooled
red@6
heterogeneity
pooled
G2
purple@6
heterogeneity 1.894
2.719
pooled
red@6
heterogeneity 1.598
pooled
5.759
P
0.796
0.025
0.175
0.038
P
2.267 0.132
2.096 0.148
6.537 0.011
0.308 0.579
P
0.032
0.596
0.008
0.904
P
0.169
0.099
0.206
0.016
4 vs. 6 µL
G2
0.008
0.118
0.008
0.030
G2
4 vs. 6 µL
P
0.548
0.025
0.827
0.315
G2
4.599
0.281
7.076
0.014
Day
1. purple@6
2. red@6
3. red@6
4. purple@6
Day
1. purple@6
2. red@6
3. red@6
4. purple@6
2.486 0.115
0.131 0.718
G2
0.361
5.017
0.048
1.011
P
2 vs. 6 µL
4 µL
3.080 0.079
0.964 0.326
2 vs. 6 µL
Day
P
G2
1. purple@6 2.379 0.123
2. red@6
11.887 0.001
3. purple@6
0
1
1.331 0.249
4. red@6
G2
purple@6
heterogeneity 1.115
pooled
1.263
red@6
heterogeneity 2.754
pooled
10.465
G2
3.652
0.717
1.900
0.392
G2
P
39.446 <0.001
30.830 <0.001
0.0357 0.850
9.289
0.002
G2
P
purple@6
0.154
0.695 heterogeneity 15.935 <0.001
23.546 <0.001
4.494
0.034 pooled
red@6
0.058
3.241
0.072
heterogeneity 3.584
36.535 <0.001
24.843 <0.001 pooled
purple@4
heterogeneity
pooled
red@4
heterogeneity
pooled
all days
heterogeneity
pooled
0.067 0.795
4.937 0.026
Day
1. purple@6
2. red@6
3. purple@6
4. red@6
P
G2
purple@4
0.447
heterogeneity
0.027
pooled
red@4
heterogeneity
0.222
<0.001 pooled
3.134 0.077
0.026 0.873
G2
0.670
0.200
3.314
0.361
G2
P
2.205
0.138
4.988
0.026
23.096 <0.001
2.442
0.118
P
2 µL
P
0 vs. 4 µL
Day
1. purple@4
2. red@4
3. red@4
4. purple@4
Day
1. purple@4
2. red@4
3. red@4
4. purple@4
0 vs. 6 µL
2 vs. 4 µL
P
0.249
0.046
0.313
0.176
G2
purple@2
heterogeneity
pooled
red@2
heterogeneity
pooled
P
0.371
<0.001
0.006
0.030
G2
0.801
29.258
7.506
4.689
G2
purple@2
heterogeneity 0.577
pooled
4.913
red@2
heterogeneity 1.493
pooled
35.271
0 vs. 2 µL
Day
1. purple@2
2. red@2
3. red@2
4. purple@2
0 vs. 4 µL
purple@6
heterogeneity
pooled
red@6
heterogeneity
pooled
all days
heterogeneity
pooled
P
0.927
0.731
0.930
0.862
P
0.035 0.852
0.004 0.950
0.020 0.887
0.106 0.745
0.027 0.870
0.083 0.773
6 µL
0 µL
log (handling time)
0
-0.05
-0.1
Day
2
-0.15
4
-0.2 3
-0.25
0 µL
2 µL
2 µL
2.75 Day
2
1
-0.05
4
3
1
-0.15
-0.2
Day
1
-0.1 4
3
0
-0.05
-0.15
-0.2 2
-0.1
-0.25
2 µL
probes
1
0 µL
4.5
0.05
Day
1
4 µL
4
3 3
2
5
2 µL
4 µL
4.5
3
4
4
0 µL
Day
3.5 2
2.5
6 µL
2
Day
1
-0.05 2
4
-0.1
3
-0.2
4.5
3
6 µL
0
5
3.5
Day
2.5 4
1
2
2
3
1.5
2 µL
-0.15
2.5
4 µL
-0.3
4 µL
0.1
3.5 2
Day
2.25 3
2
4
1.75
6 µL
-0.05
5
4
2.75
-0.25
0 µL
0
4 µL
log (handling time)
2 µL
3.25
1
-0.1
-0.3
0 µL
0.1
-0.2
0 µL
3.75
probes
Day
2
-0.15
3
4
4.25
1.25
4
log (handling time)
probes
3.25
1.25
0
0.05
3.75
1.75
-0.05
-0.25
4.25
2.25
0.05
-0.1 Day
3
4
-0.15
2
1
-0.2
1
-0.3
0
2 µL
6 µL
6 µL
3.5
1
4
3
4 µL
3
2.5
6 µL
2
4 µL
6 µL
Figure 6. Different volume offerings. (●: red flowers; □: purple flowers; ◊: pooled days; —: days that were
red contained higher nectar volume; ─: days that were purple contained higher nectar volume)
32
Number of probes at an inflorescence
0 vs. 2 µL
0 vs. 4 µL
0 vs. 6 µL
2 vs. 4 µL
4 vs. 6 µL
2 vs. 6 µL
df
MS
Colorcue
1
0.162
Day ⊂ colorcue
2
0.384
Nectartype
1 247.359
Colorcue x nectartype
1
1.364
Nectartype x day ⊂ colorcue 2 14.106
Error
382 2.575
Colorcue
Nectartype
Day ⊂ colorcue
Error
P
0.583
0.861
0.053
0.467
0.005
df
MS
P
1
7.395 0.559
1 552.805 <0.001
2
15.335 0.052
576
5.171
df
MS
Colorcue
1 0.442
Nectartype
1 164.441
Nectartype x colorcue
1 4.983
Day ⊂ colorcue
2 3.553
Nectartype x day ⊂ colorcue 2 10.813
Error
509 3.753
P
0.758
0.060
0.250
0.389
0.057
MS
P
0.015 0.792
0.005 0.714
0.010 0.400
0.166 <0.001
0.028 0.151
0.015
df
Colorcue
1
Nectartype
1
Nectartype x colorcue
1
Day ⊂ colorcue
2
Nectartype x day ⊂ colorcue 2
Error
382
Colorcue
Nectartype
Day ⊂ colorcue
Error
df
MS
P
Colorcue
1 17.644 0.586
Nectartype
1 23.629 0.025
Day ⊂ colorcue 2 42.616 <0.001
Error
903 4.678
df
1
1
2
576
MS
P
0.086 0.511
0.433 <0.001
0.137 <0.001
0.015
Colorcue
Nectartype
Colorcue x nectartype
Day ⊂ colorcue
Error
df
1
1
1
2
511
MS
P
0.004 0.791
1.113 <0.001
0.195 0.003
0.044 0.133
0.022
df
MS
P
Colorcue
1 0.003 0.674
Nectartype 1 0.415 <0.001
Error
905 0.018
df
Colorcue
1
Nectartype
1
Colorcue x nectartype 1
Day ⊂ colorcue
2
Error
585
df
MS
P
Colorcue
1
2.101 0.453
Nectartype
1 17.282 0.032
Error
588 3.729
df
Colorcue
1
Nectartype
1
Day ⊂ colorcue 2
Error
416
Log (handling time per probe)
MS
P
0.008 0.859
0.081 0.041
0.054 0.095
0.196 <0.001
0.019
df
Colorcue
1
Nectartype
1
Nectartype x colorcue
1
Day ⊂ colorcue
2
Nectartype x day ⊂ colorcue 2
Error
413
MS
P
0.484 0.819
58.907 <0.001
7.172 0.106
3.177
Table 2. Different volume offerings.
33
MS
0.080
0.253
0.038
0.058
0.030
0.015
P
0.361
0.101
0.400
0.022
0.140
EXPERIMENT 3: EFFECTS OF NECTAR TRAITS
Methods: This experiment was done in order to test bird preferences for or against
nectar concentrations and the size of sugar molecules, both of which affect nectar
viscosity. There were four parts to this experiment, each of which consisted of a
difference in nectar. Four solutions were made to accommodate the four sub-experiments
that were run: I used three sucrose solutions, 12% (Solution 1), 25% (Solution 2), and
48% (Solution 3) weight to volume; and a single 48% solution (Solution 4) that was half
glucose and half fructose. Solution 2, as well as the lower-concentrated sucrose-rich
solution (Solution 1) resembles the nectar type seen in ornithophiles (Baker et al. 1998),
whereas the higher-concentrated hexose-rich solutions (3 and 4) resembles the nectar
type seen in melittophiles (Baker & Baker 1983).
Each of the four separate sub-experiments consisted of four days:
In Experiment 3A, I tested the preference for nectar volumes that differed by 6 µL
(i.e., 2 versus 8 µL). The procedure used in this experiment was similar to the one
described in Experiment 2, except I used 8 µL and 2 µL, with red containing 8 µL first.
This method allowed me to ensure that each volume was associated with each flower
color and each flower color was located in each position of the array. Flower color was
the conditioned cue, so the color birds chose predicted the amount of nectar in the
artificial flower. Solution 2 was used.
In Experiment 3B, I used nectars differing in volume, concentration, and sugar
molecule size to see if hummingbirds showed a preference for nectar type that was more
similar to what they found in nature (i.e., a higher volume of lower concentrated sucrose
nectar). I used Solutions 1 and 4 in this experiment. The flower colors and solutions were
34
switched to ensure that flower colors were in every position, solutions were in every
position, and solutions were combined with every flower color in every position. The
goal in this experiment was to have two solutions with equal amounts of physiologically
available energy, but differing amounts of water. Using the different sugar types to
compare in the same experiment confounds the concentration with the chemical
differences, but this difference in sugars (as seen between melittophiles and
ornithophiles) is natural and was deemed the most likely to have an effect.
In Experiment 3C, birds were offered equal amounts of the same concentration of
differing sugar types. 5 µL of Solutions 3 and 4 were used.
In Experiment 3D, birds were given the choice between two solutions that only
differed by concentration, and had equal volumes of the same sugar. 8 µL of each
Solutions 1 and 3 were used.
Results for 2 versus 8 µL: Birds preferred 8 µL on red@8 days (Figure 7). Bout
initiations: On day 2, birds initially preferred 8 µL over 2 µL when 8 µL was in red
flowers (P=0.005). However, birds showed no significant initial preference for 8 µL on
days 1, 3 or 4. After running G2 tests-of-heterogeneity, red@8 could not be pooled
(P=0.209) and purple@8 could not be pooled (P=0.109). Inflorescences visited: Looking
at overall preference, birds preferred 8 µL on red@8 days (Pday1<0.001;Pday2=0.068).
However, even though 8 µL was visited more on purple@8 days, the preference was not
significant (P=0.230; 0.130). After running G2 tests-of-heterogeneity, purple@8 could be
pooled (P=0.658), but red@8 could not be pooled (P=0.006), so neither day was pooled.
Probes: One datum was winsorized from 15 probes to 12 probes. There was a marginally
significant suggestion that birds probed 8 µL over 2 µL more often (Pnectartype=0.089), but
35
the degree of preference marginally varied by day (Pnectartype × day ⊂ colorcue=0.088). There
was no interaction of nectartype with colorcue (Pnectartype × colorcue=0.353). Days varied
with reference to the number of probes (Pday ⊂ colorcue=0.024), but birds probed colors
similarly (Pcolorcue=0.664). Handling time: One datum was winsorized from 0.5185
log(seconds/flower) to 0.3863 log(seconds/flower). Though there was no interaction
between the amount of time spent at nectartype and day ⊂ colorcue, this term could not
be pooled (Pnectartype × day ⊂ colorcue=0.112). Nectartype was also not affected by the color it
was located in (Pnectartype × colorcue=0.287). Birds did not spend significantly more time at
either nectartype or flower color (Pnectartype=0.104; Pcolorcue=0.542). Only days varied in
how much time birds spent at flowers (Pday ⊂ colorcue<0.001).
36
Bout initiations
G2
0.532
7.953
0.532
2.306
Day
1. red@8
2. red@8
3. purple@8
4. purple@8
P
0.466
0.005
0.466
0.129
Inflorescences visited
G2
red@8
heterogeneity 1.577
6.908
pooled
purple@8
heterogeneity 2.565
0.273
pooled
Day
1. red@8
2. red@8
3. purple@8
4. purple@8
P
0.209
0.009
0.109
0.601
100%
100%
90%
2 µL
90%
2 µL
80%
% inflorescences visited
% bout initiations
80%
70%
70%
60%
60%
50%
50%
40%
40%
30%
30%
20%
20%
8 µL
8 µL
10%
10%
0%
G2
P
G2
P
3.323 0.068 red@8
0.006
heterogeneity 7.543
33.550 <0.001
pooled
29.330 <0.001
1.440 0.230
2.294 0.130 purple@8
heterogeneity 0.195
0.658
3.538
0.060
pooled
1. red@8
n=17
2. red@8
n=26
0%
3. purple@8 4. purple@8
n=17
n=16
Number of probes at an inflorescence
Colorcue
Nectartype
Nectartype x colorcue
Day ⊂ colorcue
df
MS
P
1
5.122 0.664
1 127.755 0.089
1
4.629 0.353
2 20.116 0.024
0
4
3
2
4
3.5
3
2.5
-0.05
Day
-0.1 1
2
4
-0.15
-0.2
2 µL
MS
P
0.170 0.542
0.414 0.104
0.026 0.287
13.746 <0.001
2.200 0.112
0.023
0.05
log (handling time)
probes
4.5
3. purple@8 4. purple@8
n=178
n=112
Log (handling time per probe)
Day
1
5
2. red@8
n=221
df
Colorcue
1
Nectartype
1
Nectartype x colorcue
1
Day ⊂ colorcue
2
Nectartype x day ⊂ colorcue 2
Error
707
Nectartype x day ⊂ colorcue 2 13.071 0.088
Error
707
5.355
5.5
1. red@8
n=204
-0.25
8 µL
3
2 µL
8 µL
Figure 7. 8 µL versus 2 µL. (●: red flowers; □: purple flowers; —: days that were red contained 8 µL; ─:
days that were purple contained 8 µL)
Results for 2 µL 48% hexose versus 8 µL 12% sucrose: Figure 8 shows that birds
preferred the 2 µL solution in red flowers on one day. Bout initiations: On the 4th day of
experimentation, birds initially preferred red flowers with 8 µL 12% sucrose (P=0.028).
However no initial preference was observed on any other day. After running G2 tests-ofheterogeneity, purple@8 days could be pooled (P=0.724), but red@8 days could not be
37
pooled (P=0.201), so I pooled neither. Inflorescences visited: Birds significantly
preferred 2 µL of 48% hexose solution on day 1 (i.e., red@8: P=0.001), but showed no
significant preference for either nectar type on any other day. After running G2 tests-ofheterogeneity, red@8 could be pooled (P=0.863), but purple@8 could not be pooled
(P=0.100). Probes: The interaction of nectartype × day ⊂ colorcue yielded a P=0.491.
Then, nectartype × colorcue yielded a P=0.917. Finally, day ⊂ colorcue yielded P=0.544.
After pooling, birds significantly probed flowers more that contained 2 µL of 48% hexose
solution (Pnectartype<0.001), regardless of color (Pcolorcue=0.655). Handling time: The
interaction of nectartype × day ⊂ colorcue a yielded P=0.250. Then, nectartype ×
colorcue yielded a P=0.843. Finally, day ⊂ colorcue yielded a P=0.351. On three out of
four days, birds spent significantly more time at 2 µL 48% hexose, but on day 2, birds
spent more time at 8 µL 12% sucrose (Pnectartype=0.002), and this nectar preference
occurred regardless of color (Pcolorcue=0.286). Generally, these results are consistent with
the view that birds have to work harder to extract higher concentrations of nectar, but this
doesn’t seem to make them subsequently avoid flower colors associated with such high
viscosity.
38
Bout initiations
Day
1. purple@8
2. purple@8
3. red@8
4. red@8
G2
0.532
1.493
0.111
4.818
P
0.466
0.222
0.739
0.028
Inflorescences visited
G2
purple@8
heterogeneity 0.125
1.900
pooled
red@8
heterogeneity 1.638
3.291
pooled
G2
Day
P
G2
1. purple@8 10.113 0.001 purple@8
heterogeneity 2.709
2. purple@8 0.718 0.397
pooled
8.122
3. red@8
0.964 0.326
4. red@8
1.871 0.171 red@8
heterogeneity 0.030
pooled
2.804
P
0.724
0.168
0.201
0.070
90%
90%
2 µL 48% hexose
80%
0.863
0.094
2 µL 48% hexose
80%
% inflorescences visited
% bout initiations
0.100
0.004
100%
100%
70%
70%
60%
60%
50%
50%
40%
40%
30%
30%
20%
20%
8 µL 12% sucrose
10%
0%
P
8 µL 12% sucrose
10%
1. purple@8 2. purple@8
n=17
n=17
3. red@8
n=9
0%
4. red@8
n=11
1. purple@8 2. purple@8
n=116
n=113
3. red@8
n=51
4. red@8
n=65
Number of probes at an inflorescence
Log (handling time per probe)
df
MS
P
Colorcue
1
0.711 0.655
Nectartype 1 48.431 <0.001
Error
342 3.563
df
MS
P
Colorcue
1 0.090 0.286
Nectartype 1 0.783 0.002
Error
342 0.079
0.1
5
0.05
log (handling time)
4.5
probes
4
3.5
3
0
Day
2
-0.05
4
-0.1
3
1
-0.15
2.5
-0.2
2
2 µL
8 µL
48% hexose 12% sucrose
2 µL
8 µL
48% hexose 12% sucrose
Figure 8. 2 µL 48% hexose versus 8 µL 12% sucrose. This experiment tested preferences for bee- versus
bird-type nectar. (●: red flowers; □: purple flowers; ◊: pooled days; —: days that were red contained 2 µL
48% hexose; ─: days that were purple contained 2 µL 48% hexose )
Results for sucrose versus hexose: Birds seemed unaffected by sugar type (Figure
9). Bout initiations: After running G2 tests-of-heterogeneity, redS (sucrose in red) could
be pooled (P=0.672), purpleS could be pooled P=0.682), and I was able to pool all days
(P=0.463). The pooled data indicate that birds did not initially prefer either sugar (Pall days
pooled=0.884).
Inflorescences visited: Birds showed no significant preference for either
sugar type (P>0.05 for all days). After running G2 tests-of-heterogeneity, redS could not
39
be pooled (P=0.147) and purpleS could not be pooled (P=0.241). Probes: The interaction
of nectartype × day ⊂ colorcue yielded a P=0.947. Then, nectartype × colorcue yielded a
P=0.581. Finally, day ⊂ colorcue yielded a P=0.507. Following pooling, I see that birds
probed flowers with sucrose more than flowers with hexose, but this was a marginally
significant preference (Pnectartype=0.086). Their choices were not affected by color
(Pcolorcue=0.408). Handling time: One datum was winsorized from 0.4293
log(log(seconds/flower) to 0.2975 log(seconds/flower). The interaction of nectartype ×
day ⊂ colorcue yielded a P=0.197, so this term could not be pooled, even though there
was no interaction between the amount of time spent at a specific sugar and day.
Nectartype × colorcue yielded a P=0.685. There was variation of how much time was
spent at sugar types on different days (Pday ⊂ colorcue<0.001), but birds did not significantly
spend more time at either sugar (Pnectartype=0.544), or color (Pcolorcue=0.984).
40
Bout initiations
Day
G2
1. purpleS 0.335
2. purpleS
0
3. redS
0.067
4. redS
0.505
100%
90%
G2
P
0.563 purpleS
heterogeneity 0.168
1
0.796
pooled
0.167
0.477 redS
heterogeneity 0.180
pooled
0.392
all days
heterogeneity 0.538
0.021
pooled
60%
50%
40%
5
48% sucrose
5 µL
µL 40%
Sucrose
G2
purpleS
heterogeneity 2.099
pooled
0.061
redS
heterogeneity 1.377
pooled
1.309
P
0.147
0.805
0.241
0.253
100%
0.463
0.884
90%
% inflorescences visited
% bout initiations
0.672
0.531
5 µL 48% hexose
80%
70%
20%
0.682
0.683
5 µL 48% hexose
80%
30%
Inflorescences visited
Day
G2
P
1. purpleS 0.591 0.442
2. purpleS 1.569 0.210
3. redS
0.087 0.768
4. redS
2.599 0.107
P
70%
60%
50%
40%
30%
20%
5 µL 48% sucrose
10%
10%
0%
0%
all days pooled
n=47
Number of probes at an inflorescence
Colorcue
Nectartype
Error
df
MS
P
1 2.143 0.408
1 9.283 0.086
297 3.127
1. purpleS
n=83
2. purpleS
n=64
df
1
1
1
2
Nectartype x day ⊂ colorcue 2
Error
292
4.5
4. redS
n=47
Log (handling time per probe)
Colorcue
Nectartype
Nectartype x colorcue
Day ⊂ colorcue
5
3. redS
n=103
MS
P
<0.001 0.984
0.015 0.544
0.165
0.685
0.284 <0.001
1.635
0.017
0.197
0.2
0.15
3.5
log (handling time)
probes
4
3
2.5
2
sucrose
hexose
0.1
Day
2
0.05 4
3
0
1
-0.05
-0.1
sucrose
hexose
Figure 9. Sucrose versus hexose. This experiment tested preferences for sugar type. (●: red flowers; □:
purple flowers; ◊: pooled days; —: days that were red contained hexose; ─: days that were purple contained
sucrose)
Results for 12% versus 48% sucrose: Figure 10 shows that birds generally
preferred 48% over 12% sucrose. Bout initiations: After running G2 tests-ofheterogeneity, red@48 could be pooled (P=0.883), purple@48 could be pooled
(P=0.428), and all days could be pooled (P=0.879). On first visits to the array, birds
showed no preference for the higher or lower concentration (Pall days pooled=0.881).
41
Inflorescences visited: Overall, birds significantly preferred 48% sucrose on two days,
and showed marginal significance on one day. They showed no significant preference for
either concentration on day 1, a red@48 day (P=0.163). After running G2 tests-ofheterogeneity, purple@48 could be pooled (P=0.644), but red@48 could not be pooled
(P=0.167), so I pooled neither. Probes: One datum was winsorized from 18 probes to 10
probes. The interaction of nectartype × day ⊂ colorcue yielded P=0.698. Then, nectartype
× colorcue yielded a P=0.929. Finally, day ⊂ colorcue yielded P=0.345. The simplified
analysis shows that birds significantly probed more flowers that contained higher
concentrated sugar (Pnectartype<0.001), and their choices were not affected by color
(Pcolorcue=0.241). Handling time: The interaction of nectartype × day ⊂ colorcue yielded
P=0.741. Then, nectartype × colorcue yielded P=0.839. Finally, day ⊂ colorcue yielded
P=0.830. After pooling, the ANOVA showed that birds spent significantly more time
visiting higher concentrated nectar (Pnectartype<0.001), but they also spent more time
depending on the color of the cue (Pcolorcue=0.017).
42
Bout initiations
G2
Day
1. purple@48 0.077
2. purple@48 1.646
1.159
3. red@48
0.699
4. red@48
P
0.781
0.200
0.282
0.403
100%
90%
8 µL 12% sucrose
Inflorescences visited
G2
purple@48
heterogeneity
pooled
red@48
heterogeneity
pooled
all days
heterogeneity
pooled
P
Day
1. purple@48
2. purple@48
3. red@48
4. red@48
0.627 0.428
1.096 0.295
0.218 0.883
1.836 0.175
% inflorescences visited
% bout initiations
60%
50%
40%
30%
20%
8 µL 48% sucrose
purple@48
heterogeneity 0.213
pooled
8.445
red@48
heterogeneity 1.912
pooled
9.979
P
0.644
0.004
0.167
0.002
80%
70%
60%
50%
40%
30%
20%
8 µL 48% sucrose
10%
0%
all days pooled
n=50
Number of probes at an inflorescence
1. purple@48 2. purple@48 3. red@48
n=53
n=48
n=87
4. red@48
n=75
Log (handling time per probe)
df
MS
P
Colorcue
1
4.902 0.241
Nectartype 1 148.850 <0.001
Error
264 3.546
df
MS
P
Colorcue
1
0.111 0.017
Nectartype 1
1.647 <0.001
Error
264 0.019
0.15
4
0.1
log (handling time)
4.5
3.5
probes
G2
8 µL 12% sucrose
90%
70%
0%
P
0.073
0.020
0.163
0.002
100%
0.023 0.879
2.908 0.881
80%
10%
G2
3.221
5.437
1.950
9.942
3
2.5
2
0.05
0
-0.05
-0.1
1.5
-0.15
12% sucrose 48% sucrose
12% sucrose 48% sucrose
Figure 10. 12% versus 48%. This experiment tested preferences for nectar concentrations. (●: red flowers;
□: purple flowers; —: days that were red contained 12% sucrose; ─: days that were purple 12% sucrose)
43
EXPERIMENT 4: PREFERENCE AGAINST LANDING PLATFORMS
Methods: This experiment was done in order to test bird preferences for or against
landing platforms. An array was set up so that half of the stakes consisted of artificial
flowers that had landing platforms (i.e., an elongated tube that was cut to resemble a
flower with five petals), whereas the other half of the stakes consisted of artificial flowers
without landing platforms (i.e., elongated tubes with the bottom two petals cut out)
(Figure 1g & h). Landing platforms are generally present in melittophilous flowers, but
they are often absent in ornithophilous flowers. The goal of this experiment was to see if
birds showed any kind of behavioral response to landing platforms. The dimensions of
the landing platform were as follows: diameter: 24.241 ± 1.365 mm (mean ± standard
deviation); length: 32.625 ± 1.004 mm (mean ± standard deviation). These dimensions
were constructed after Castellanos et al.’s (2004) measurements of Penstemon barbatus,
a hummingbird-pollinated flower. Each artificial flower in the array contained 5 µL of
25% sucrose solution. Both types of artificial perianths were red. This experiment was
replicated for twenty-four time blocks of forty minutes (i.e., 4 six-hour days).
Inflorescences were set up in an alternation pattern (i.e., first row no platform-platformno platform-platform). Positions were rotated each day.
Results for platformed versus non-platformed: As expected, Figure 11 shows that
birds preferred flowers lacking landing platforms. Bout initiations: After running G2
tests-of-heterogeneity, experimental days could be pooled (P =0.382). On first visits to
the array, birds significantly preferred flowers without landing platforms over flowers
with landing platforms (Pall days pooled=0.038). Inflorescences visited: Again, after running
G2 tests-of-heterogeneity, days could be pooled (P=0.496). Overall, birds significantly
44
visited flowers without landing platforms over flowers with landing platforms (Pall days
pooled<0.001).
Probes: The interaction of flowertype × day yielded a P=0.291, so the term
was pooled. Days did not significantly vary with respect to number of probes
(Pdays=0.799), so days were pooled. Birds significantly probed flowers that lacked landing
platforms more often than flowers with landing platforms (Pflowertype=0.002). Handling
time: The interaction of color × day yielded a P=0.466, so this term was pooled. Days
could not be pooled (Pday=0.166). Birds spent significantly less time visiting flowers
lacking landing platforms than flowers with landing platforms (Pflowertype<0.001).
45
Bout initiations
2
Day G
P
1 4.818 0.028
2 1.699 0.192
Inflorescences visited
2
G
P
Heterogeneity 0.763 0.382
All days pooled 6.517 0.038
2
G
Day
P
1 18.536 <0.001
2 12.172 0.001
100%
90%
100%
platformed
% inflorescences visited
% bout initiations
70%
60%
50%
40%
30%
non-platformed
10%
0%
80%
70%
60%
50%
40%
30%
20%
non-platformed
10%
0%
all days pooled
n=26
Number of probes at an inflorescence
all days pooled
n=158
Log (handling time per probe)
df
MS
P
Flowertype 1
0.356 <0.001
Day
1
0.039 0.166
Error
155
0.02
df
MS
P
Flowertype 1 40.610 0.002
Day
1
0.274 0.799
Error
155 4.227
5
log (handling time)
0.3
4.5
4
probes
platformed
90%
80%
20%
3.5
0.2
2.5
0.05
0
platformed
1
0.15
0.1
nonplatformed
Day
2
0.25
3
2
G2
P
Heterogeneity 0.464 0.496
All days pooled 30.244 <0.001
nonplatformed
platformed
Figure 11. Platformed versus non-platformed. This experiment tested preferences against landing
platforms. (◊: pooled days)
46
EXPERIMENT 5: PATTERNING ON THE PERIANTH
Methods: The purpose of this experiment was to test if birds could use another
flower cue, besides the color of the perianth as a whole, to associate with a higher nectar
source. An array was set up, where half of the inflorescences contained a pattern on the
perianth – the pattern was two black spots positioned in the space above two lobes of the
perianth (Figure 1i) – whereas the other half of the stakes had unmarked flowers. This
was an associative-learning experiment in which varying nectar reward was associated
with presence or absence of pattern. All perianths were red. This experiment was
replicated for forty-eight time blocks of forty minutes (i.e., 8 six-hour days) – four days
in August of 2008, and four days in April of 2009. Stakes were set up in an alternating
pattern (first row marked-unmarked-marked-unmarked), where marked flowers contained
8 µL and unmarked flowers contained 2 µL. The flower types and nectar amounts were
switched to ensure that flower types were in every position, solutions were in every
position and solutions were combined with every flower type in every position. The
volumes used were chosen from Experiments 2 and 3A. In 2008, 21% sucrose solution
was used. In 2009, 23% sucrose solution was used.
Results for patterned versus unpatterned: Birds did not use patterns as a cue to
find the higher nectar source, although they definitely paid attention to higher nectar
volumes once they were at an inflorescence (Figure 12). Bout initiations: On day 2, birds
showed initial significant preference for non-patterned flowers with 2 µL over patterned
flowers with 8 µL (P=0.031). However, on day 7, birds showed a significant initial
preference for 8 µL in non-patterned flowers over 2 µL in patterned flowers (P=0.016).
The former (i.e., day 2) is a highly unexpected result. After running G2 tests-of-
47
heterogeneity, pattern@8 could not be pooled (P=0.092) nor could nopattern@8
(P=0.249). Inflorescences visited: After running G2 tests-of-heterogeneity, I could pool
pattern@8 (P=0.718) and nopattern@8 (P=0.755). Furthermore, both pattern@8 and
nopattern@8 could be pooled together (Pheterogeneity=0.680). Overall, birds showed no
significant preference for either flower type, regardless of which volume was presented
(Pall days pooled=0.933). Probes: One datum was winsorized from 18 probes to 14 probes.
Though there was no interaction between nectartype and day, this term could not be
pooled (Pnectartype × day ⊂ patterncue=0.172). Patterncue did not interact with nectartype in
regards to the number of probes (Pnectartype × patterncue=0.490). The final analysis shows that
birds significantly probed flowers with 8 µL over flowers with 2 µL (Pnectartype=0.011),
and did not discriminate based on pattern (Ppatterncue=0.633). Days varied in how many
probes inflorescences received (Pday ⊂ patterncue<0.001). Handling time: One datum was
winsorized from 0.9841 to 0.3971 log(seconds/flower). None of the terms could be
pooled. Birds spent significantly more time at 8 µL than 2 µL, but the amount of time
spent varied by day (Pnectartype × day ⊂ patterncue<0.001). Furthermore, there was no interaction
between nectartype and patterncue (Pnectartype × patterncue=0.740), though replicate days
varied greatly by the amount of time spent at a pattern type (P day ⊂ patterncue<0.001). Birds
spent more time at flowers with 8 µL over those with 2 µL (Pnectartype=0.007), though they
did not spend significantly more time at patterned or not patterned flowers
(Ppatterncue=0.479).
48
Bout initiations
Day
1. pattern@8
2. pattern@8
3. nopattern@8
4. nopattern@8
5. pattern@8
6. pattern@8
7. nopattern@8
8. nopattern@8
2
G
3.104
4.636
0.087
0.016
0.251
0.133
5.782
0
Inflorescences visited
P
0.078
0.031
0.768
0.898 pattern@8
heterogeneity
0.617
pooled
0.715 nopattern@8
0.016
heterogeneity
1
pooled
G2
P
Day
1. pattern@8
2. pattern@8
3. nopattern@8
4. nopattern@8
5. pattern@8
6. pattern@8
7. nopattern@8
8. nopattern@8
df
7.981 0.092 3
0.143 0.705 1
5.393 0.249 3
0.493 0.483 1
100%
% inflorescences visited
% bout initiations
90%
2 µL
80%
70%
60%
50%
40%
30%
8 µL
20%
10%
patt
patt nopatt nopatt patt
@8
@8
@8
@8 @8
n=73 n=56 n=46 n=61 n=16
df
0.718 3
0.741 1
0.755 3
0.794 1
0.680 2
0.933 1
2 µL
80%
70%
60%
50%
40%
30%
20%
0%
patt nopatt nopatt
@8
@8
@8
n=30 n=15 n=8
8 µL
4.5
5.5
3.5
Day
1 5
4
2 4.5
3
4
4
2.5
8 µL
6
5
3
0.1
Day
8
3.5
2 µL
MS
0.057
1.68
0.002
0.099
P
0.479
0.007
0.740
<0.001
Nectartype x day ⊂ pattercue 6 0.103 <0.001
Error
1253 0.020
7
3
df
1
1
1
6
Patterncue
Nectartype
Nectartype x pattercue
Day ⊂ patterncue
Day
8
7
0.05
0
Day
1
-0.05
4
2
-0.1
3
log (handling time)
Nectartype x day ⊂ pattercue 6
7.290 0.172
Error
1253 4.839
5
6
all days pooled
n=1269
Log (handling time per probe)
df
MS
P
1
8.537 0.633
1 94.955 0.011
1
2.308 0.490
6 33.772 <0.001
Patterncue
Nectartype
Nectartype x pattercue
Day ⊂ patterncue
probes
P
10%
Number of probes at an inflorescence
2
P
G2
0.295 pattern@8
0.591 heterogeneity 1.348
1.110
0.692 pooled
0.364 nopattern@8
1.191
heterogeneity
0.799
pooled
0.068
0.939
all days
0.714 heterogeneity 0.170
0.705 pooled
0.007
100%
90%
0%
G2
1.099
0.288
0.157
0.825
0.065
0.006
0.134
0.143
5
6
-0.15
2 µL
8 µL
-0.2
2 µL
8 µL 2 µL
8 µL
Figure 12. Patterned versus non-patterned flowers. This experiment tested whether hummingbirds would
use pattern cues to find more rewarding nectar. (●: non-patterned flowers; □: patterned flowers; —: days
that were patterned@ higher volume; ─: days that were non-patterned@ higher volume)
49
CRITIQUE OF EXPERIMENTAL DESIGN
Why a field study?
In developing this project, there was some discussion of capturing birds, but I
decided that the cons outweighed the pros. If I was able to catch and mark each
individual bird with colored ink, then perhaps I could have used individual birds as a unit
of replication. I could have recorded the behavior of a single individual and, therefore, its
response to the experimental treatments. However, this would probably have led to trying
to address more complicated questions than I have considered here: if a bird is chased off
before it is able to complete its bout, then the behavior of the visiting bird as a replicate is
not only, or even primarily, about the floral manipulations. I narrowed my time focus to
at most a whole bout. This is also important because each experiment assumed that there
was only one bird emptying flowers during a bout. If I had marked birds, then I would be
able to distinguish birds that visit the array frequently and those that visit periodically. It
would have allowed me to distinguish among the birds that were dominating the array
and those that were more submissive. This would have required vigilant trapping for
extended periods to keep all the birds marked individually. Without worrying about such
marks, more experimental runs were possible.
It also occurred to me that if I caught birds, I might use them in captivity one at a
time rather than mark and release them. This would have eliminated the complication of
birds interacting agonistically, but this would also have changed the nature of the study to
one in which a captive bird was making choices of what flowers to visit in a small array
with no opportunity to fly away and otherwise occupy itself. A captured bird might have
realized it was the only bird with access to the array and could have then changed its
50
foraging patterns. Occasionally, birds react poorly to being captured (Tyrrell & Tyrrell
1985), making it dangerous to keep them in captivity for longer than a day.
Hummingbirds that react poorly would have been set free. Females are particularly hard
to hold because if I captured a female, there was a chance that she might have a brood. If
a captured female is unable to return to her young frequently, then I could have been
endangering her young. In addition, I may not have been able to keep more than one
hummingbird in a cage at a time, considering that they are extremely territorial and nonsocial animals (Williamson 2001) and that interactions between them would have
nullified the benefit of working with captive birds.
The trapping of birds was not necessary. In most cases, birds visited the array one
at a time. When there was more than one bird in the location of the array, birds fought for
possession of the array and one would not allow others to feed. This did not happen very
often. For the 0.78% of observation periods where birds were actively being aggressive to
one another, the data were excluded from the above analyses.
Replication and randomization
There were various replicates taken into consideration: each observation day can
be considered a block. Bouts were replicates for bout initiation. For the other three
dependent variables, a visit to an inflorescence was the unit of replication. There are a
number of other units of replication that one might wish to use, for example, individual
bird’s responses at the array. However, as discussed above, I did not attempt this. I may
be measuring the preference for single birds in a location, instead of a population
51
preference. Though I attempted to run my experiments in a different location, I failed at
this due to time constraints.
Stakes were arranged within a 4 × 4 array representing an alternating pattern (i.e.,
on every side of one flower type, there was a different flower type). In other words,
flower types were not randomly located within the array. Although the array was of small
size, randomization would have meant that any given run would have presented birds
with a particular patchiness. The alternating pattern forced birds to come into contact
with the two differing flower types; if they wanted to be constant to a flower type, they
had to fly over the contrasting flower type. The use of this alternating pattern also
allowed me to replace stakes quickly before birds returned to the array. In some
instances, as I was replacing stakes, birds would attempt to visit, but actual visits to the
stakes were eliminated because I was able to replace stakes quickly, rather than trying to
lookup a random pattern between hourly trials.
52
DISCUSSION
In my experiments, birds sometimes showed a slight preference for red over
purple. Birds generally preferred flowers with nectar over nectarless flowers. When
nectar volumes were 2 µL or more, regarding number of inflorescences visited, birds
were not consistently discerning between nectar volumes. For example, birds visited 8 µL
when also offered 2 µL, but the difference was not significant for all days. The birds did
probe more when there was more nectar, and often spent more handling time per probe,
but their proportional preference while working the mixed patch was little affected. When
nectar was energetically equivalent, birds generally visited the two types of flowers
similarly with a slight bias for red cues. However, they showed a significant preference
for melittophilous nectar in red flowers on Day 1 of the experiment. Birds visited sucrose
and hexose in equal frequencies when both sugar types were offered in equal amounts
and concentrations. Birds generally visited 48% sucrose nectar over 12% sucrose nectar.
Birds preferred flowers without landing platforms, and had decreased handling time when
visiting flowers without landing platforms. Regardless of the presence or absence of spots
on the flowers, birds visited both types of flowers equally, seeming to make no
association with the difference in nectar volume, or else being willing to visit both nectar
volumes equally. The formation of these specific hummingbird-flower characteristics
(i.e., flower color, a higher nectar volume, and the lack of landing platforms) seems to be,
in part, due to the preferences of hummingbirds for the flowers they pollinate. The main
inconsistency with the pollination syndromes is the preference birds have for highconcentration small-quantity nectar.
53
Evolution of flowers occurs via many factors
How did the suite of traits associated with ornithophily evolve? Floral characters
are likely under selection, but what are the agents of selection? Pollinators should attempt
to maximize their own profit (i.e., energy intake versus energy expense) while foraging,
so they should be expected to prefer floral characters that enable them to do so.
Alternatively, in ornithophiles, bees will take pollen and nectar while transferring less
pollen, or even bypassing the stigma(s) and anthers all together (Castellanos et al. 2003).
In some cases, as with Penstemon newberryi (a species half way between the two
syndromes) flowers are frequented by bees more often than hummingbirds, of which the
former pollinates wastefully (Wilson et al. 2006). Therefore, ornithophiles should reduce
bee interaction, so as to reduce the loss of pollen and nectar without a pollination service,
and hence, there should be strong selection for characters that will keep bees away.
Finally, characters may be adaptive that increase pollen transfer efficiency, so selection
might also favor flowers that channel their specific pollinator’s probing head, thereby
possibly increasing visitation rate by decreasing handling time (Wilson 1995).
Color. Floral color is a cue, or advertisement, that pollinators use to find nectar,
which is the reward. Ornithophiles tend to be red, though hummingbirds can see all
colors of the visible spectrum (Tyrrell & Tyrrell 1985), whereas melittophiles tend to be
yellow, purple, or blue, and bees tend to visit red flowers less often, thereby failing to
deplete nectar in them (Wilson and Jordan 2009). Hummingbirds are associative learners
(Collias & Collias 1968; Stiles 1976; Healy and Hurly 2001), often making associations
between floral advertisements and floral rewards after only one visit. They have learned
that red floral cues lead to more nectar, so initial visits to a patch of mixed flowers are
54
more commonly red (Stiles 1976). As birds learn about the rewards from other flower
colors by sampling various flowers in a patch, they learn to visit those flowers (Waser &
McRobert 1998), though birds have brief holdover periods for color preferences.
Therefore, as ornithophiles may be evolving in response to other factors, hummingbirds
would be able to respond to floral changes without harming the plant-animal mutualism;
this benefits both actors in the mutualism. Hummingbird’s propensity for red is supported
from this study, but as also shown, after a short period of time, they recognized that red,
purple, and yellow flowers had equal volumes, so they began to visit these colors in
nearly equal frequencies. Though bees can see red, they lack the photoreceptors to
distinguish red petals from green foliage (Giurfa et al. 1996; Chittka & Waser 1997). My
experiments suggest that even if birds showed a slight preference for red, this preference
can be altered, supporting the presence of a weak holdover period for color preference. If
red flowers were assumed to be under-visited by bees, it is more likely that nectar would
be found in red flowers compared to the depleted nectaries of purple or yellow
inflorescences. Therefore, birds should visit purple and yellow artificial inflorescences at
a lower percentage, which I found they did slightly. This may be complicated by floral
constancy. Studying constancy, Wilson and Stine (1996) found bees were more likely to
move between flowers that differed in shape with similar colors than between flowers
that differed in color but had similar shapes. This is what I observed with reference to
hummingbirds. Both flower types offered had the same shape but different colors. Birds
spent less time at flowers that differed in color (i.e., more time at red versus less time at
yellow or purple) and spent more time at flowers they knew would have nectar – flowers
they learned contained nectar that would sustain them.
55
Nectar quantity. Nectar is an important component of flowers – the only reward
offered to pollinators in my experiments – and nectar type is associated with flower
syndrome type. Melittophiles produce small amounts of concentrated nectar that is a
hexose mixture (i.e., fructose and glucose with relatively small amounts of sucrose),
whereas ornithophiles produce high volumes of dilute sucrose-rich nectar (Baker &
Baker 1983; Thomson et al. 2000; Wilson et al. 2004). These syndrome differences in
nectar properties might have diverged because of differing preferences of the animals that
pollinate those two kinds of flowers. Pollinators will not visit empty nectaries after they
have learned to evaluate them as empty (Hurly & Healy 1996). Flowers, like those in the
genus Penstemon, have evolved a way to replenish their nectar once it has been emptied
(Castellanos et al. 2002; Ornelas & Carlos 2009), and both bees (Cnaani et al. 2006) and
hummingbirds (Hurly & Healy 1996) learn to expect when flowers have refilled their
nectaries. In Experiment 2, birds generally preferred flowers with nectar over those
without nectar; on some days they showed no preference. Any threshold below which
birds would distain flowers of a certain appearance would seem to be less than 2 µL. The
choice of volumes may have confounded my efforts to test bird preferences for their
natural volume: 1 µL would have been melittophilous, and 2 µL or more would have
been ornithophilous (Roberts 1996), but instead, I used all nectar volumes that were more
similar to ornithophilous flowers (i.e., 2 µL and above). However, Gass and Roberts
(1992) suggested that some flowers produce large volumes of nectar, whereas others
produced very small volumes (sometimes from 1-4 µL). This inconsistency of nectar
volumes in ornithophiles made it difficult for me to decide which nectar volumes might
have been appropriate prior to testing bird preferences on volumes that were similar to
56
melittophiles versus ornithophiles. Birds showed no preferences for higher nectar
volumes when flowers differed by 2 µL. They continually visited both nectar amounts,
regardless of the volume difference. Finally, when nectar differed by 4 µL, birds showed
a stronger preference for 6 µL when it was in red flowers, but again, they only preferred
the higher nectar volume on one day. Nectar volume may be the most important property
that affects hummingbird preferences during pollination (Blem et al. 2000), and for that
reason, I would expect there to be some volume at which the flower and the pollinator
come to a “compromise” (i.e., a high enough volume where birds will visit, but a low
enough volume where flowers will not expend too much energy and water). If
melittophiles and ornithophiles offer equal amounts of large volumes of nectar, the latter
may be in danger of limited pollination. If ornithophiles do not present a high enough
nectar volume, birds may avoid these flowers completely. Therefore, flowers should offer
enough nectar that will attract hummingbirds, and hummingbirds should visit flowers that
meet their minimum nectar volume requirement. This study suggests that a nectar volume
of 2 µL may be more than enough to keep birds coming back to flowers. Regardless of
what natural flowers offer (approximately 1 to 5 µL of 20-25% sucrose-rich nectar:
Pleasants 1983; Irwin & Brody 2000; or an average of 7.76 µL of 26% sucrose-rich
nectar in Penstemon: Wilson et al. 2006), when birds are presented with two differing
nectar volumes, they prefer nectar sources that have considerably higher nectar amounts
(Hurly & Oseen 1999) and nectar concentrations (Stiles 1976; Roberts 1995; Roberts
1996; Lopez-Calleja et al. 1997).
Nectar viscosity and its components. Melittophiles tend to have a much higher
sugar concentration than ornithophiles: about 35% (Baker 1975; Pyke & Waser 1981),
57
but possibly as high as 50-80% sugar (Baker 1975). A possible explanation for the
difference in nectar concentration is that bees avoid flowers with dilute nectar, having a
maximum net energy gain when consuming nectar that is 40% sugar or higher (Harder
1986; Cnaani et al. 2006). The extra water in more dilute nectar takes longer for bees to
get rid of. Contrary to this, birds have extremely efficient kidneys that excrete water
quickly and easily (Tyrrell & Tyrrell 1985). Floral characters that decrease handling time
and increase net energy gain while foraging should be favored by selection. As nectar
concentration increases, the amount of nectar extracted per lick decreases and licking
takes longer (Roberts 1995). This causes hummingbirds to spend more time foraging at
flowers with highly concentrated nectar. It has been suggested that if birds are able to
extract all floral nectar in a single lick (1.9 µL for a Rufus Hummingbird), then they
would prefer lower concentrated nectar (Kingsolver and Daniel 1983); however, Roberts’
(1995, 1996) data cast doubt on whether energy intake rate would actually be higher at
lower concentrations for flowers with 4-8 µL (i.e., a higher nectar volume) and calculated
over the duration of a visit to a flower. As mentioned previously, hovering is extremely
expensive, so birds should be expected to minimize the amount of energy expended while
foraging, hence visit less viscous nectar (Baker & Baker 1983b; Powers & Nagy 1988;
Gill 1995; Wilson et al. 2006). My findings were not particularly consistent with this
theory from the literature, rather more consistent with past data (e.g., Stiles 1976; Roberts
1996). In Experiment 3, when birds showed a significant preference for flower type, it
was for melittophilous-type nectar when in red flowers, even though they also had an
increased handling time. Stiles (1976), with the use of high-volume bonanza feeders,
showed that hummingbirds in the laboratory generally have a significant preference for
58
sucrose-containing nectar, rather than solutions that contained glucose or any fructose.
Contrary to Stiles (1976), in my experiment birds had no preference for sugar type, but
similar to Stiles, they had a significant preference for higher concentrated nectar. Even
when both nectar types contained the same caloric value, birds may have chosen higher
concentrated nectar because it “tastes sweeter” (Stiles 1976 “taste conditioning” effect).
Though hummingbirds preferred melittophilous nectar, they had increased handling time.
Furthermore, the presence of large amounts of dilute, sucrose-rich nectar in ornithophiles
may be a consequence of flower costs. If flowers expend less energy to produce the
minimum nectar properties hummingbirds will visit, then there are fewer costs to the
flower to increase pollinator efficiency. Ornithophiles should be expected to only expend
enough energy required to attract hummingbirds (Zimmerman 1988). Therefore, because
these flowers are their primary food source, birds have little choice but to visit
ornithophiles, which contain nectar that is lower than the amount they prefer. However, it
may be possible that the reason for dilute nectar is to deter bees, because bees tend to
have a lower net energy gain when extracting dilute nectar (Harder 1986). If bees refuse
to visit flowers with low concentrated nectar, or are physically excluded from
ornithophiles, then those flowers possibly have more nectar. Therefore, the discrepancies
I see with hummingbird behavior on nectar concentration may be an artifact of bee
preference, not bird preference. I would expect to see that nectar volume and
concentration in flowers should allow birds to visit flowers quickly and provide them
with enough energy to carry out other tasks (Roberts 1996), though the cost that plants
pay for making nectar (sometimes as much as 37% energy) should not be larger than the
benefit of pollination (Southwick 1984; Pyke 1991; Aigner 2001). Therefore, a moderate
59
nectar volume (approximately 2 to 4 µL) with minimal concentrations (~25 % sucrose)
could be sufficient to convince hummingbirds to keep visiting flowers.
Landing platforms. Understanding the evolution of how flowers eliminated
landing platforms is complicated. There are a number of possible explanations. First,
landing platforms may obstruct hummingbirds from extracting nectar (Sutherland &
Vickery 1993; Castellanos et al. 2004), thereby influencing them to prefer flowers
without landing platforms, as they did in Experiment 4. Consistent with this, I also found
that birds took significantly more time extracting nectar when there were landing
platforms than when landing platforms were absent, thereby expending more energy
while foraging. Alternatively, the absence of landing platforms may not be a response to
bird preference, but instead to deterring nectar and pollen-robbers. In other words, the
absence of landing platforms may be common in ornithophiles as an adaptation to deter
bees from visiting flowers, thereby decreasing competition between hummingbirds and
bees for nectar (Valido et al. 2002; Rodriguez-Girones & Santamaria 2004) and reducing
wasteful bees from taking pollen from flowers without providing a pollination service
(Wilson et al. 2006). A third possibility for the absence of landing platforms might be
with the loading and the pickup of pollen from the bird's head. Landing platforms make
contact with the anthers and stigmas less precise (Castellanos et al 2003; Fenster et al.
2004; Armbruster et al. 2009), so selection for increased pollen transfer efficiency would
eliminate the landing platform. The direct factor that influenced the absence of landing
platforms in ornithophiles is unclear, but it is understood that, if given the choice, birds
avoid flowers with landing platforms, possibly because it is more difficult and it takes
longer to extract nectar.
60
Nectar guides. In shifts from melittophily to ornithophily, flowers sometimes lose
color patterning. Waser (1982) suggested that the presence of patterns on perianths in
melittophiles may be used as nectar guides to lead bees to rewards, thereby improving the
bees’ handling time and making the flowers more attractive. Further investigation by
Waser and Price (1985) supports this, suggesting that foragers have increased handling
times when visiting albino flowers without nectar guides. They expected hummingbirds
to show preferences for flowers with nectar guides, but that is not what is seen in my
study. Perhaps birds are not so guided. For bees, Gould (1985) showed the ability to
discern sources based on associated patterning. In Experiment 5, hummingbirds did not
use patterns to find rewarding nectar, though when they visited flowers with 8 µL, they
not only probed them more often, they spent more time at them. Limited or no patterning
could be a result of hummingbird negligence – not recognizing patterns that might help
them find rewarding nectar. Nectar guides may be at a scale that birds are insensitive to.
Hence, the color of the whole flower, or plant, or even patch, and not detailed floral
patterns, may be the best cue to aid hummingbirds to find rewarding nectar.
What I didn’t study. There are many environmental factors, besides pollinator
preference, that affect floral evolution, such as herbivory, temperature stress, and length
of reproductive season (Elliott & Irwin 2009; Gomez et al. 2009). I discussed how
pollinator preference, efficiency and parasite deterrence can influence floral evolution,
but herbivores can affect plants in a different way. The model presented by Kang et al.
(2008) suggested that plants may increase leaf growth if herbivory is high. Their model
also explained that if growth rate increases, plants may expend less energy in nectar
production, thereby causing pollinators to settle for nectar they may not have preferred if
61
given the choice. This would cause plants to produce the minimum amount of nectar that
would still persuade pollinators to visit. In addition, temperature may affect the shape of
flowers – shapes that could keep nectar from evaporating might be favored (Tyrrell &
Tyrrell 1985). Certain floral shapes could hinder visits by some pollinators, though it
holds nectar safely. Finally, how long flowers are available for reproduction is also a
factor that influences floral evolution. Flowers should maximize pollination during the
time they are dehiscing anthers and making stigmas sticky, otherwise the flower is
rendered useless. Therefore, absence of a character like landing platforms allows
hummingbirds to extract nectar quickly and position pollen on specific parts of the bird’s
head (Tyrell & Tyrell 1985).
Bird behavioral ecology
Hummingbird behavior is plastic, responding to cues that they can associate with
nectar rewards, though these cues are not always the same. They rely on their memory for
finding previously rewarding nectar sources (Hurly 1996), taking into account their
ability to extract nectar quickly (Tyrrell &Tyrrell 1985). They change their mind
constantly, sampling flowers that possess cues that may be dissimilar to ornithophilous
flowers (Tyrell & Tyrell 1985; Wells 1993; Fenster et al. 2004). This behavior of
spending a bit of time sampling, or “tracking,” flowers that at the moment are underpreferred, is called “minoring” (terminology coined by Heinrich 1979 who studied
bumblebees). Applied to hummingbirds, I would say that they “major” on red flowers
with copious nectar but “minor” on purple flowers with scant nectar. This unfaithfulness
62
to specific flower types enables the birds to find new and rewarding flowers as they
appear, which might help them increase their net energy gain over the longer term.
Their behavioral plasticity may cause some preferences for floral characters to be
altered by other preferences as the birds gain experience. For example, though birds
preferred higher volumes when all other factors were held equal (Experiment 2), their
preference shifted when flowers differing in nectar concentration were presented
(Experiment 3). On the first day of Experiment 3B, birds had an overall preference for 2
µL of 48% hexose (cued with red) over 8 µL of 12% sucrose (cued with purple) and on
the last experimental day they showed an initial preference for 8 µL of 12% sucrose (cued
with red) compared to 2 µL of 48% hexose (cued with purple). In addition, birds not only
probed flowers with the higher concentration more, regardless of color, they also spent
more time at those flowers. This is also consistent with Experiment 3D, where they
probed higher concentration nectar more often and spent more time foraging at higher
concentration flowers. Birds initially showed a strong overall preference for flowers that
provide them with a longer handling time (i.e., 2 µL 48% hexose). However, on later
days, birds preferred flowers that provided them with a shorter handling time (i.e., 8 µL
12% sucrose). When birds are visiting higher concentrated nectar for longer periods of
time, there is a possibility that they are expending more energy (Roberts 1996). If they
are expending more energy while foraging longer on higher concentrated nectar, these
differences in handling times should be learned. However, differences in handling times
and amount of energy expended may not be learned as quickly as sugar concentrations –
evaluating efficiency takes continual feeding from flowers by the bird to see how certain
nectars affect their daily activity. The change in preference from a source that provides
63
longer handling time to one that provides shorter handling time may be a prolonged and
on-going assessment by the bird attempting to choose flowers that maximize their net
energy gain. Their eventual choice for ornithophilous nectar is an integrated response to
the type of nectar they are feeding on, rather than an immediate response associated with
the taste of a nectar type (see Cnaani et al.’s 2006 concentration-before-volume
explanation).
Birds are initially lured into patches by flowers that possess characters similar to
rewarding flowers they have experienced in recent visits (Hurly 1996; Hurly & Healy
1996; Fenster et al. 2004). Bout initiations are a bird’s first response to an array, but, as
seen in this study, the results for bout initiation mirror the results for visitation to all
inflorescences. This similarity seen between their initial preference and their overall
preference suggests that how birds are drawn into a mixed patch of flowers is not a very
different behavioral process from how they proceed throughout the bout. Therefore,
simply looking at birds’ overall preferences rather than looking at initial preferences
separately is an accurate analysis of bird behavior.
Unlike other nectar-eating birds, such as Australian honeyeaters that perch while
foraging at flowers, hummingbirds expend a large amount of energy while hovering to
forage (Mitchell & Paton 1990). True hovering flight, only done by hummingbirds, is
extremely expensive, being more expensive than any other mode of flight (Powers &
Nagy 1988; Suarez 1992; Gill 1995; Welch & Suarez 2008). In order to sustain their
daily activity, hummingbirds consume as much as half of their body weight in nectar
(Baker & Baker 1983b; Tyrrell & Tyrrell 1985). If an available nectar source is present,
any hummingbird that claims a site will defend that site aggressively (Baker & Baker
64
1983b). However, once a flower patch is depleted, hummingbirds will be less aggressive
(Kodric-Brown & Brown 1978). Although hummingbirds expend massive amounts of
energy just by moving around and defending territories, it is expected that they would
prefer characters that would reduce the amount of time spent probing flowers. However,
this is not the case: though they preferred red flowers – a character that likely led them to
rewards more quickly – they not only probed flowers with higher concentrations more
often, they spent more time at higher-concentration flowers. Other characters, such as
landing platforms, are avoided by hummingbirds who seem to assess that they cannot
extract nectar as quickly. Furthermore, floral patterns could be used to find nectar more
quickly, but it seems that hummingbirds, unlike bees (Waser 1983; Gould 1985), may not
use as fine-tuned senses for the perception of patterns. When they visit an inflorescence
with more nectar, whether intentionally or not, they tend to probe its flowers more
thoroughly, realizing that they have come into contact with a large nectar source.
Therefore, even though hummingbirds expend massive amounts of energy moving from
place to place, they still do not seem to behave in a way that would reduce energy
expense. Their constantly-changing foraging behavior and unwillingness to associate a
few certain cues with nectar rewards causes them to readily sample nectar sources that
may not be worthwhile.
Foraging theory is built around assuming that animals reduce the amount of time
foraging so as to expend less energy and increase net energy gain. I expected
hummingbirds to visit characters associated with ornithophilous flowers because these
characters seem to be the ones that reduce handling time. However, when given choices
between two artificial nectars with the same caloric values but different concentrations,
65
hummingbirds preferred the higher concentrated nectar, though they had increased
handling time, even though ornithophiles have lower concentrated nectar. Hummingbirds
are able to recognize a low-volume and high-concentrated nectar source (similar to
melittophiles) more quickly than a high-volume low-concentrated nectar source (similar
to ornithophiles), and will therefore prefer the lower volume and higher concentrated
nectar (Roberts 1996). Natural flowers do not necessarily offer birds what they might
most prefer, as suggested by my results. Although birds showed an initial preference for
melittophilous nectar on one day, their preference for melittophilous nectar may have
been caused by what Stiles (1976) calls “taste conditioning,” where they choose nectar
directly based on how it tastes, not on the overall net energy gain. This idea is also
supported by Roberts (1996). Alternatively, the syndrome rule may be a result of flowers
evolving in response to other factors. Sugar preference results are less consistent with
previous studies: Stiles (1976) and Martinez del Rio (1990) both showed that in most
cases Anna’s Hummingbirds and various larger-bodied Mexican hummingbirds prefer
sucrose solutions similar to ornithophilous nectar – preferring sucrose over a hexose
mixture, or glucose and fructose individually. However, my study showed that birds had
no preference for sugar type. Stiles (1976) suggested that hummingbirds show a
“preference hierarchy,” where birds make choices based on what is important to them.
Smith et al. (2008) supported this hypothesis by showing hummingbirds consider nectar
reward and floral display (i.e., number of flowers per inflorescence) as more important
than other cues (i.e., floral color). My study further supports this idea, suggesting that
sugar content is not as important as concentration, or even water content (i.e., nectar
volume) because birds in my study showed no significant preference for sugar type,
66
visiting sugar types in equal frequencies (P>0.05 for all days). It may be that
hummingbirds have a preference for certain nectar properties, but flowers have not
evolved to accommodate hummingbirds fully, considering flowers have other costs:
nectar production, parasite deterrence, as well as pollinator efficiency and preference.
The discrepancies I see between bird preferences and ornithophilous offerings may be
due to how flowers have evolved, rather than direct preferences from hummingbirds.
67
LITERATURE CITED
Aigner, P.A. 2001. Optimality modeling and fitness trade-offs: when should plants
become pollinator specialists? Oikos 95:177-184.
Armbruster, W.S., T.F. Hansen, C. Pelabon and R. Perez-Barrales. 2009. The adaptation
of flowers: measurement and microevolutionary patterns. Annals of Botany
103:1529-1545.
Baker, H.G. 1961. The adaptation of flowering plants to nocturnal and crepuscular
pollinators. Quarterly Review of Biology 36:64-73.
Baker, H.G. 1975. Sugar concentrations in nectar from hummingbird flowers. BioTropica
7:37-41.
Baker, H.G. and I. Baker. 1983. Floral nectar sugar constituents in relation to pollinator
type; chapter 5. Handbook of Experimental Pollination Biology. Page 117-141.
Van Nostrand Reinhold, New York.
Baker, H.G. and I. Baker. 1983b. A brief historical review of the chemistry of floral
nectar. In B. Bently and T. Elias (Eds). The biology of nectaries, page 126-152.
Columbia University Press, New York, New York.
Baker, H.G., I. Baker and S.A. Hodges. 1998. Sugar composition of nectars and fruits
consumed by birds and bats in the tropics and subtropics. BioTropica 30:559-586.
Bateson, M, S.D. Healy and T.A. Hurly. 2003. Context-dependent foraging decisions in
rufous hummingbirds. Proceedings: Biological Sciences 270:1271-1276.
Bene, F. 1941. Experiments on the color preference of black-chinned hummingbirds. The
Condor 43:237-242.
Blem, C.R., L.B. Blem, J. Felix and J. van Gelder. 2000. Rufous hummingbird sucrose
preference: precision of selection varies with concentration. The Condor 102:235238.
Brown, J.H. and A. Kodric-Brown. 1979. Convergence, competition, and mimicry in a
temperate community of hummingbird-pollinated flowers. Ecology 60:10221035.
Castellanos, M.C., P. Wilson and J.D. Thomson. 2002. Dynamic nectar replenishment in
flowers of Penstemon (Scrophulariaceae). American Journal of Botany 89:111118.
Castellanos, M.C., P. Wilson and J.D. Thomson. 2003. Pollen transfer by hummingbirds
and bumblebees, and the divergence of pollination modes in Penstemon.
Evolution 57:2742-2752.
Castellanos, M.C., P. Wilson and J.D. Thomson. 2004. ‘Anti-bee’ and ‘pro-bird’ changes
during the evolution of hummingbird pollination in Penstemon flowers. Journal
of Evolutionary Biology 17:876-885.
Cnaani, J., J.D. Thomson and D.R. Papaj. 2006. Flower choice and learning in foraging
bumblebees: effects of variation in nectar volume and concentration. Ethology
112:278-285.
Chittka, L. and N. Waser. 1997. Why red flowers are not invisible to bees. Israel Journal
of Plant Sciences 45:169-183.
Collias, N. E. and E. C. Collias.1968. Anna’s hummingbirds trained to select different
colors in feeding. The Condor 70:273-274.
68
Elliott, S. and R.I. Irwin. 2009. Effects of flowering plant density on pollinator visitation,
pollen receipt, and seed production in Dephinium barbeyi (Ranunculaceae).
American Journal of Botany 96:912-919.
Faegri, K. and L. van der Pijl. 1979. The Principles of Pollination Ecology, 2nd edn.
Pergamon Press, New York.
Fenster, C.B., W.S. Armbruster, P. Wilson, M.R. Dudash and J.D. Thomson. 2004.
Pollination syndromes and floral specialization. Annual Review of Ecology and
Systematics 35:375-403.
Gass, C. L. and W. M. Roberts. 1992. The problem of temporal scale in optimization:
three contrasting views of hummingbird visits to flowers. The American
Naturalist 5:829-853.
Gill, F.B. 1995. Ornithology 2nd ed. Page 98-109. W. H. Freeman and Company, New
York NY.
Giurfa, M., M. Vorobyev, P. Kevan and R. Menzel. 1996. Detection of colored stimuli by
honeybees: minimum visual angles and receptor specific contrasts. Journal of
Comparative Physiology A 178:699-709.
Goldsmith, T.H. and K.M. Goldsmith. 1979. Discrimination of colors by the blackchinned hummingbird. Journal of Comparative Physiology A 130:209-220.
Gomez, J.M., F. Perfectti, J. Bosch and J.P.M. Camacho. 2009. A geographic selection
mosaic in a generalized plant-pollinator-herbivore system. Ecological
Monographs 79:245-263.
Gould, J.L. How bees remember flower shapes. 1985. Science 227:1492-1494.
Gould, J.L. 1993. Insect learning: ecological and evolutionary perspectives; Chapter 2.
Pages 18-50. Eds. Papaj, D.R. and A.C. Lewis. 1993. Chapman and Hall; New
York and London.
Grant, K.A. and V. Grant. 1968. Hummingbirds and their flowers. New York, Columbia
University Press.
Hale, W.G., V.A. Saunders and J.P.Margham. 2008. Page 329. Collins Web-linked
dictionary – Biology. Harper Collins Publishers, New York, NY.
Harder, L.D. 1986. Effects of nectar concentration and flower depth on flower handling
efficiency of bumble bees. Oecologia (Berlin) 69:309-315.
Healy, S.D. and T.A. Hurly. 2001. Foraging and spatial learning in hummingbirds;
Chapter 7. Page 127. Eds. Chittka, L. and J. Thomson. 2001. Cambridge
University Press. Cognitive ecology of pollination.
Heinrich, B. 1979. “Majoring” and “minoring” by foraging bumblebees, Bombus vagans:
and experimental analysis. Ecology 60:245-255.
Hurly, T.A. 1996. Spatial memory in rufous hummingbirds: memory for rewarded and
non-rewarded sites. Animal Behaviour 51:177-183.
Hurly, T.A. and S.D. Healy. 1996. Memory for flowers in rufous hummingbirds: location
or local visual cues? Animal Behaviour 51:1149-1157.
Hurly, T.A. and M.D. Oseen. 1999. Context-dependent, risk-sensitive foraging
preferences in wild rufous hummingbirds. Animal Behaviour 58:59-66.
Irwin, R.E. and A.K. Brody. 2000. Consequences of nectar robbing for realized male
function in a hummingbird-pollinated plant. Ecology 81:2637-2643.
Irwin, R.E., L.S. Adler and A.K. Brody. 2004. The dual role of floral traits: pollinator
attraction and plant defense. Ecology 86:1503-1511.
69
Kang, Y., D. Armbruster and Y. Kuang. 2008. Dynamics of a plant-herbivore model.
Journal of Biological Dynamics 2:89-101.
Kingsolver, J.G. and T.L. Daniel. 1983. Mechanical determinants of nectar feeding
strategy in hummingbirds: energetics, tongue morphology, and licking behavior.
Oecologia (Berlin) 60:214-226.
Kodric-Brown, A. and J.H. Brown. 1978. Influence of economics, interspecific
competition, and sexual dimorphism on territoriality of migrant rufous
hummingbirds. Ecology 59:285-296.
Lopez-Calleja, M.V., F. Bozinovic and C. Martinez del Rio. 1997. Effects of sugar
concentration on hummingbird feeding and energy use. Comparative
Biochemistry and Physiology 118A:1291-1299.
Martinez del Rio, C. 1990. Sugar preferences in hummingbirds: the influence of subtle
chemical differences on food choice. The Condor 92:1022-1030.
Melendez-Ackerman, E., D.R. Campbell and N.M. Waser. 1997. Hummingbird behavior
and mechanisms of selection on flower color in Ipomopsis. Ecology 18:25322541.
Mitchell, R.J. and D.C. Paton. 1990. Effects of nectar volume and concentration on sugar
intake rates of Australian honeyeaters (Meliphagidae). Oecologia 83:238-246.
Ornelas, J.F. and L. Carlos. 2009. Nectar replenishment and pollen receipt interact in
their effects on seed production of Penstemon roseus. Oecologia (Berlin)
160:675-685.
Pleasants, J.M. 1983. Nectar production in Ipomopsis aggregata (Polemoniaceae).
American Journal of Botany 70:1468-1475.
Powers, D.R. and K.A. Nagy. 1988. Field metabolic rate and food consumption by freeliving Anna’s Hummingbirds (Calypte anna). Physiological Zoology 6:500-506.
Pyke, G.H. 1991. What does it cost a plant to produce floral nectar? Nature 350:58-59.
Pyke, G.H. and N.M. Waser. 1981. The production of dilute nectar by hummingbird and
honeyeater flowers. Biotropica 13:260-270.
Roberts, W.M. 1995. Hummingbird licking behavior and the energetics of nectar feeding.
The Auk 112:456-463.
Roberts, W.M. 1996. Hummingbirds’ nectar concentration preferences at low volume:
the importance of time scale. Animal Behaviour 52:361-370.
Rodriguez-Girones, M.A. and L. Santamaria. 2004. Unsolved mystery: Why are so many
bird flowers red? Public Library of Science Biology 2:1515-1519.
Smith, S.D., C. Ané and D.A. Baum. 2008. The role of pollinator shifts in the floral
diversification of Iochroma (Solanaceae). Evolution 62:793-806.
Smithson, A. 2001. Pollinator preference, frequency dependence, and floral evolution.
Pp. 237-258 in L. Chitka and J.D. Thomson [eds], Cognitive Ecology of
Pollination. Cambridge University Press.
Southwick, E.E. 1984. Photosynthate allocation to floral nectar: a neglected energy
investment. Ecology 65:1775-1779.
Stephens, D.W. and J.R. Krebs. 1986. Pgs. i-xiv, 1-247. Princeton University Press, New
Jersey.
Stiles, F.G. 1976.Taste preferences, color preferences, and flower choice in
hummingbirds. The Condor 78:10-26.
70
Stiles, F.G. 1981. Geographic aspects of bird-flower coevolution, with particular
reference to Central America. Annals of the Missouri Botanical Garden 68:323351.
Stiles, F.G. and C.E. Freeman. 1993. Patterns in floral nectar characteristics of some birdvisited plant species from Costa Rica. Biotropica 25:191-205.
Suarez, R.K. 1992. Hummingbird flight: sustaining the highest mass-specific metabolic
rates among vertebrates. Experientia 48:565-570.
Sutherland, S. and R.K. Vickery, Jr. 1993. On the relative importance of floral color,
shape, and nectar rewards in attracting pollinators to Mimulus. Great Basin
Naturalist 53:107-117.
Thomson, J.D., W.P. Maddison and R.C. Plowright. 1982. Behavior of bumblebee
pollinators of Aralia hispida Vent. (Araliaceae). Oecologia (Berlin) 54:326-336.
Thomson, J.D., S.C. Peterson and L.D. Harder. 1987. Response of traplining bumble bees
to competition experiments: shifts in feeding location and efficiency. Oecologia
(Berlin) 71:295-300.
Thomson, J.D. 1988. Effects of variation in inflorescence size and floral rewards on the
visitation rates of traplining pollinators of Aralia hispida. Evolutionary Ecology
2:65-76.
Thomson, J.D., P. Wilson, M. Valenzuela and M. Malzone. 2000. Pollen presentation and
pollination syndromes, with special reference to Penstemon. Plant Species
Biology 15:11-29.
Thomson J.D. and Wilson P. 2008. Explaining evolutionary shifts between bee and
hummingbird pollination: Convergence, divergence, and directionality.
International Journal of Plant Sciences 169: 23-38.
Tyrrell, E.Q. and R.A. Tyrrell. 1985. Hummingbirds, Their Life and Behavior a
Photographic Study of the North American Species. 1st Ed. ed. New York:
Crown 135-139; 173-199.
Valido, A., Y.L. Dupont and D.M. Hansen. 2002. Native birds and insects, and
introduced honeybees visiting Echium wildpretii (Boraginaceae) in the Canary
Islands. Acta Oecologia 6:413-419.
Waser, N.M. 1982. A comparison of distances flown by different visitors to flowers of
the same species. Oecologia 55:251-257.
Waser, N.M. 1983. Competition for pollination and floral character differences among
sympatric plant species; a review of evidence. Pp. 277-293 in C.E. Jones and R.J.
Little [eds], Handbook of experimental pollination biology. Van NostrandReinhold, New York.
Waser, N.M. and M.W. Price. 1983. Pollinator behaviour and natural selection for flower
colour in Delphinium nelsonii. Nature 302:422-424.
Waser, N.M. and M.V. Price. 1985. The effect of nectar guides on pollinator preference:
experimental studies with a montane herb. Oecologia (Berlin) 67:121-126.
Waser, N.M. and J.A. McRobert. 1998. Hummingbird foraging at experimental patches
of flowers: evidence for weak risk-aversion. Journal of Avian Biology 29:305313.
Welch, K.C. and R.K. Suarez. 2008. Altitude and temperature effects on the energetic
cost of hover-feeding migratory rufous hummingbirds, Selasphorus rufus.
Canadian Journal of Zoology 86:161-169.
71
Wells, D.J. 1993. Ecological correlates of hovering flight of hummingbirds. Journal of
Experimental Biology 178:39-57.
Williamson, S.L. 2001. Peterson Field Guides: Hummingbirds of North America. Page
30.
Wilson, P. 1995. Selection for pollination success and the mechanical fit of Impatiens
flowers around bumble bee bodies. Biolological Journal of the Linnean Society
55: 355-383.
Wilson, P. and M. Stine. 1996. Floral constancy in bumble bees: handling efficiency or
perceptual conditioning? Oecologia 106:493-499.
Wilson, P., Castellanos, M.C., Hogue, J.N., Thomson, J.D. and Armbruster, W.S. 2004.
A multivariate search for pollination syndromes among penstemon. Oikos
104:345-361.
Wilson, P., M.C. Castellanos, A. Wolfe and J.D. Thomson. 2006. Shifts between bee- and
bird-pollination among penstemons. Pp. 47-68 in N.M. Waser and J. Ollerton
(eds.) Plant-pollinator interactions: from specialization to generalization.
University of Chicago Press.
Wilson, P., A.D. Wolfe, W.S. Arbruster and J.D. Thomson. 2007. Constrained lability in
floral evolution: counting convergent origins of hummingbird pollination in
Penstemon and Keckiella. New Phytologist 176:883-890.
Wilson, P. and E.A. Jordan. 2009. Hybrid intermediacy between pollination syndromes in
Penstemon, and the role of nectar in affecting hummingbird visitation. Botany
87:272-282.
Zimmerman, M. 1988. Nectar production, flowering phenology and strategies for
pollination. In J. L. Doust and L. L. Doust [eds.], Plant reproductive ecology,
157-178. Oxford University Press, Oxford, UK.
72