 

 

m LIBRARY
I Michigan State
r1000! f University

 

This is to certify that the
thesis entitled

THE EFFECTS OF POKEWEED DISPLAY CHARACTERISTICS
AND FRUGIVORE IDENTITY ON FRUIT REMOVAL

presented by

Eva J. Lewandowski

has been accepted towards fulfillment
of the requirements for the

MS. degree in Zoology

 

 

' Major Professor’s Signathe

IVA/0%

Date

MSU is an Affirmative Action/Equal Opportunity Employer

u. _.—.an.-.—o--.-n---—n-a-n—-.—

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K.lPrq/Acc&Pres/CIRC/Date0ue.indd

THE EFFECTS OF POKEWEED DISPLAY CHARACTERISTICS AND FRUGIVORE
IDENTITY ON FRUIT REMOVAL

By

Eva J. Lewandowski

A THESIS

Submitted to
Michigan State University
in partial ﬁJlﬁllment of the requirements
for the degree of

MASTERS OF SCIENCE
Zoology

2008

ABSTRACT

THE EFFECTS OF POKEWEED DISPLAY CHARACTERISTICS ON FRUGIVORE
IDENTITY AND FRUIT REMOVAL

By
Eva J. Lewandowski

Fruit color and contrast are important display traits that inﬂuence frugivory and
seed dispersal. Asynchronous ripening and the presences of unripe ﬁ'uits in mixed
displays can also influence frugivory and seed dispersal. We tested for effects of
ripening synchrony in pokeweed (Phytolacca americana) by studying ﬁ'uit removal on
plants with varying levels of synchrony. Over the course of our experiment, there was a
strong, but non-signiﬁcant trend for bushes with both ripe and unripe fruits to receive
more fruit removal than bushes that were fully ripe. There was also a trend for ﬁ'ugivory
to be positively correlated with the proportion of ripe fruit on the bush during the middle
and end of the fall fruiting season, but not the beginning. We hypothesize that the
presence of unripe fruits on a bush signal the freshness of the fruits that are ripe, and that
birds prefer to forage on bushes with freshly ripe fruits. However, the presence of unripe
ﬁ'uits could decrease foraging efﬁciency, making plants with a large proportion of ripe
fruits more attractive to frugivores than plants with a smaller proportion of ripe fruit. In
addition to the expected avian frugivory, our pokeweed plants also experienced a great
deal of white-tailed deer ﬁ'ugivory, leading to questions about the role deer play in seed

dispersal and plant reproduction for pokeweed.

ACKNOWLEDGEMENTS

I am grateful to many people for assisting me both with my research and with this thesis.
I would like to thank my committee: Jeff Conner, Fred Dyer, and Kay Holekamp, and
especially my adviser, Tom Getty. I would also like to thank my labmates: Katie
Wharton, Lindsey Walters, and Jean Johnson. Graduate students from the Zoology
Department, EEBB Program, and KBS also provided helpﬁil insight, and the K88 and
Zoology Department staffs have also been extremely helpful. Finally, I would like to
acknowledge the NSF IGERT in sequential decision-making and the Wallace Endowed

Scholarship Award, both of which provided me with ﬁnding for my research.

iii

TABLE OF CONTENTS

List of Tables .......................................................................................... v
List of Figures ........................................................................................ iv
Introduction ............................................................................................ 1
Materials and Methods ............................................................................... 5
Results .................................................................................................. 9
Discussion ............................................................................................ 15
References ............................................................................................ 21

LIST OF TABLES

Table 1. Fixed effects for a linear mixed model predicting total fruit removal per raceme.
Model: total ﬁ'uit removal ~ display treatment + frugivory type + fruit count + trial
duration, random effects of pair + pair/bush.
......................................................................................................... 10

Table 2. Fixed effects for a linear mixed model predicting total fruit removal per bush.
Model: total fruit removal ~ stem color treatment + trial duration + trial + ripeness index
+ ripeness indexrtrial, random effects of bush + bush/stem.
......................................................................................................... 14

LIST OF FIGURES

Figure 1. Difference in total removal between mixed and ﬁilly ripe treatments at each
site. Values above 0 indicate more removal on mixed displays; values below 0 indicate
more removal on fully ripe displays ............................................................ 1 1

Figure 2. Total number of fruits removed for racemes with and without deer frugivory.

Solid bars represent racemes with only bird frugivory. Slashed bars represent racemes
that had deer frugivory ............................................................................ 12

vi

Introduction

Frugivory is one of the most common and obvious types of mutualistic
relationships. While a great deal of work has been done on the interactions between
plants that produce ﬂeshy fruits and birds that eat them and disperse their seeds (Snow,
1971; Stiles, 1980; Gosper et al., 2005), it is still not entirely clear how plants attract
avian dispersers or to what extent a fruiting plant’s seed dispersal might be helped or
hindered by attracting other, non-avian, fruit—eaters.

Recent work has demonstrated that both color and the amount of contrast between
a ﬁ'uit and its background may have a significant impact on fruit choice by birds (Willson
& Whelan, 1989; Galetti et al., 2003; Schmidt et al., 2004). These results suggest that
selection has acted on ﬁ'uiting displays to attract avian ﬁ'ugivores, rather than to attract
other taxa such as mammals or to repel certain destructive taxa of insects. If this is true,
we might expect that some of the common visual characteristics of fruiting displays could
be explained by avian fruit choice.

The most basic example of a fruiting display is a ripe fruit presented against a
background of leaves. However, there are many species that have more complicated
displays than this. Fruiting displays with an additional colored component are often
referred to as bicolored displays. We avoid the term bicolored display because it implies
that two colors are involved in the display, when in fact there are three or more colors
present: that of the leaves, the ripe fruit, and an additional colored component. For
instance, plants can have temporally mixed displays, in which ripe and unripe fruits are
present simultaneously. Other plants have structurally mixed displays; these involve an

additional colored component of the plant besides the leaves and fruits. Structurally

mixed displays often involve colored stems. Some plants exhibit both structurally and
temporally mixed displays at the same time. Mixed displays are fairly common; one
study estimated that approximately one-third of the fruiting trees, shrubs, and bushes in
Illinois produce them; although only 10% of the fruiting herbs were categorized as mixed
displays (Willson & Thompson, 1982).

Many plants exhibit temporally mixed displays due to asynchronous ripening. It
could be argued that these mixed displays are a result of physiological constraints
preventing synchronous ripening. It is a necessity for many black fruits to change from
green to red and ﬁnally to black as their anthocyanin concentration increases. However,
there are species of plants in which the fruits mature from green to black without
transitioning through a red stage (Willson & Thompson, 1982), suggesting that red
coloration is not always a necessary physiological precursor to black coloration. In
addition, the degree of ripening asynchrony varies greatly between species, and for some
plants the asynchrony is negligible (Gorchov, 1990). While these facts do not completely
rule out physiological constraints as an explanation for asynchronous ripening for some
plant species, they do indicate that such constraints are not entirely limiting.
Furthermore, it is possible that avian frugivory could select for temporally mixed
displays, increasing asynchrony beyond any that might result from physiological
constraints.

A potential explanation for the prevalence of temporally mixed displays among
black-fruiting plants is that black ﬁ'uits accumulate more insect damage than do red fruits
(Mordenmoore & Willson, 1982). This might select for the plant to maintain only a few

ripe black fruits at any given time. This would decrease the attraction of the plant to

insects and restrict the amount of damage that insects can cause within a short timeframe.
Neither of these possible explanations addresses the prevalence of structurally mixed
displays.

An explanation that could encompass both temporally and structurally mixed
displays is the possibility that mixed displays act as either a signal or a cue to attract fruit-
eating birds. Sehaefer et a1. (2007) have shown that, to the avian visual system, the
reﬂectance spectra of secondary components of fruiting displays (unripe ﬁ'uit, colored
bracts, etc.) contrast more against the background leaves than do those of ripe hit. This
suggests that mixed displays render a ﬁ'uiting plant more visible to ﬁ'ugivorous birds than
would otherwise be so.

Many ﬁeld experiments have investigated the effect of mixed displays on avian
ﬁ'uit choice, but the results have been inconsistent. Several studies compared displays of
only ripe fruit to displays of ripe fruit plus an additional color and found that ﬁ'ugivory
was higher on the more complicated displays; however the absence of leaves in these
experiments makes it difﬁcult to extend the results to most situations in which fruits are
grown against a leafy background (Mordenmoore & Willson, 1982; Whelan & Willson,
1994; Burns & Dalen, 2002). Two studies experimented by adding colored material to
fully ripe displays, and in both cases the frugivory results varied with the type of color
that was added (Facelli, 1993; Cramer et al., 2003). Fuentes (1995) found either
increased or decreased ﬁ'ugivory depending on the manner in which the mixed displays
were presented. Wenny (2003) found increased ﬁ'ugivory with the addition of mixed
displays in the form of colored bracts to small shrubs, but Mordenmoore & Willson

(1982) found decreased ﬁugivory on the temporally mixed displays of whole trees. In

general, there is some limited evidence that mixed displays increase ﬁugivory, but a more
rigorous exploration of the topic is needed to conﬁrm this, as well as to determine why
birds might forage more on plants with mixed displays.

Here, we discuss the results of two experiments designed to explore the effects of
mixed displays on frugivory. To test for an effect of temporally mixed displays we
paired artiﬁcally manipulated fully ripe and partially ripe bushes of pokeweed
(Phytolacca americana) and recorded fruit removal. For this experiment, there are two
hypotheses. The ﬁrst hypothesis is that birds will preferentially forage on plants with
higher proportions of ripe fruits; this could lead to more efﬁcient foraging. If this
hypothesis is correct, we would predict that fruit removal would be higher on our ﬁilly
ripe plants than on our mixed bushes containing both ripe and unripe fruits. The second
hypothesis is that birds will forage more on bushes containing both ripe and unripe fruits.
This could be due to the presence of unripe fruit signaling the freshness and quality of the
ripe limit on the same plant, or it could be a result of unripe fruit adding to the visual
display of the plant and rendering it more conspicuous to avian dispersers. If this
hypothesis is correct, we would expect to ﬁnd higher ﬁ'uit removal on our mixed display
bushes than on our fully ripe bushes.

To examine the role of structurally mixed displays, we recorded fruit removal on
artiﬁcial stems of two different colors that had been placed on pokeweed bushes. Here
we hypothesize that the additional contrast presented by reddish-purple stems will attract
more birds than the contrast presented by green stems. If this is correct, we predict that
stems that we have colored reddish-purple will have more fruits removed than stems that

we have colored green.

Materials & Methods

We conducted this study in September through November of 2007 at the Lux
Arbor Reserve, part of Michigan State University’s Kellogg Biological Station, in
southwestern Michigan. The reserve consists of a mixture of secondary forest, wetland,
and agricultural habitats. It is home to a variety of frugivorous bird species, including
gray catbirds (Dumetella carolinensis), American robins (T urdus migratorius), and cedar
waxwings (Bombycilla cedrorum). Lux Arbor also has a large population of white-tailed
deer (Odocoileus virginianus). All experiments were done with the American pokeweed
(Phytolacca americana), a native herbaceous plant that produces ﬂeshy fruits that are
known to be eaten by birds (Smith & Riley, 1990). Pokeweed plants have bright reddish-
purple stems. The unripe green fruits turn reddish-purple when maturing and become
black when fully ripe. The ﬁuit may remain green for several months, but once they
begin to turn reddish-purple, they become fully black in a few days. Pokeweed seeds do
not require gut passage in order to germinate, (Armesto et al., 1983), but seeds that pass
through an avian gut are more likely to germinate and have faster germination rates

(Orrock, 2005).

Experiment 1

 

In our ﬁrst experiment, we investigated the impact of temporally mixed displays
on fruit removal. We experimentally manipulated bushes with both ripe and unripe fruits
to create pairs of bushes with one fully ripe and one partially ripe bush (n = 22 pairs). To
create a bush with only ripe ﬁ'uits, we randomly selected one bush within the pair and cut

off all its ﬁ'uits that were not ﬁilly ripe. We pruned our mixed displays to ensure they

contained only racemes with both ripe and unripe fruits. We also equalized the number
of racemes within each pair to control for the size of the display. The ﬁJIly ripe bush and
the partially ripe bush within each pair were between 2 and 8m apart. Pairs of bushes
were located throughout our ﬁeld site, and pairs were not placed within 20m of each
other. We marked three to ﬁve individual racemes on each bush and recorded the
number of unripe, ripe, and past ripe ﬁ'uits on each raceme every other day. Unripe fruits
were those that were either green or reddish-purple. Fruits were called ripe when they
were ﬁilly black. Past ripe fruits were shriveled and desiccated. We also categorized the
removal as bird or deer frugivory. Avian frugivory was characterized by the removal. of
individual fruits while leaving the stem intact. We labeled removal as deer frugivory
when all or part of the stern, along with the fruits, had been removed. In addition to the
evidence on the stems, many of our instances of deer frugivory were also accompanied
by deer prints and seat under and around the bushes, and we often observed deer among
our pokeweeds, thus strengthening our interpretation that deer were responsible for this
type of ﬁuit removal. We collected data for a total of 72 days; however, in many cases
the entire pokeweed clump collapsed or the individual racemes were completely stripped
of fruits prior to the completion of our study.

We conducted a census of fruits every other day and calculated ﬁ'uit removal for
each census. We summed the fruit removal for the entire study to determine total fruit

removal. Fruit removal for each census i was calculated as:

Removal; = ((31-1 + PM + Bi-1) - (G; + P; + 8;) — O + N

With 0 = (O; - OH) ifOi > Oi_1 and 0 ifOi S OH

And N=Gi—Gi_1iij>Gi_1 and OifGiSGi-1

G represents the number of green fruits, P represents the number of purple fruits, B
represents the number of black fruits, and 0 indicates the number of fruits that are past
ripe.

To test for variables with signiﬁcant effects on fruit removal, we used the
package nlme in R (Pinheiro et al., 2006; R Development Core Team, 2006) to create a
linear mixed effects model for our data with total fruit removal as the dependent variable.
We tested for the effects of display treatment, frugivory type, number of fruits on the
raceme at the beginning of the study, and the total number of days data were collected
(trial duration) by including them as ﬁxed effects in our model. We included pair and
bush nested within pair as random effects to account for our paired design. Due to
concern about the high degree of correlation between the display treatment and fruit
count variables, which could lead to issues of multicollinearity and potentially mask. the
signiﬁcance of display treatment, we also ran the model without fruit count.
Additionally, we tested for the presence of interactions between our terms; however, no
interactions were signiﬁcant, so we did not include them in our ﬁnal model. Analysis

with the package lme4 yielded similar results, so we do not report them here.

Experiment 2

To examine the effect of structurally mixed displays on fruit removal, we attached
artiﬁcial stems of two different colors to pokeweed bushes (n = 24 sites) and compared
fruit removal on the two types of stems. Each bush received three wire stems painted

reddish-purple to match the natural color of the pokeweed stem, and three wire stems

painted green to match the color ofpokeweed leaves and immature stems. We used an
Ocean Optics USB4000 spectrometer to match the color of our artiﬁcial stems to the
natural color of pokeweed stems and leaves. On each stem, we placed 20 freshly picked
pokeweed fruits. Every other day we recorded the number of fruits remaining on the
stem and replaced them with fresh fruits. While we cannot say conclusively that missing
fruits were removed during ﬁugivory, we did have a number of stems that did not lose
any ﬁ'uits, suggesting they were unlikely to simply fall off. Prior to the peak of the
frugivory season, we also conducted a small pilot study in which we left the fruits on the
artiﬁcial stems for 4 days; during that time we recorded no loss of mu. This suggests
that the fruit loss we recorded was due to fruits being removed from the artiﬁcial stems
rather than merely falling to the ground. We left the stems on the plants for a total of 15
days; however, some trials were shortened when the pokeweed collapsed or was pushed
to the ground.

We conducted three sets of trials throughout the course of our study. As a result,
each pokeweed bush had between one and three trials, depending on how long the bush
remained standing and retained fruits. This resulted in a total of 61 trials.

We analyzed our data using a linear mixed effects model in R. Using total fruit
removal per artiﬁcial stem as our dependent variable, we tested the ﬁxed effects of our
stem color treatment, the trial identity (1-3, indicative of the time in the season that we
conducted each trial), the trial duration (3-15 days), a ripeness index that we calculated as
the proportion of racemes on a bush bearing unripe fruit, and the interaction between the
ripeness index and the trial identity. We also included bush and individual stem nested

within bush as random effects.

Results
Experiment 1

Due to our experimental manipulation, racemes on fully ripe bushes held an
average of 47 fruits (46.87 :t 12.47). This was signiﬁcantly fewer fruits than racemes on
our mixed bushes, which had an average of 53 fruits (53.29 :‘c 11.87; t = 2.49, df= 102, p
< 0.02), of which approximately 30 (29.68 i 16.03) were ripe. The number of hits on a
raceme had a signiﬁcant positive effect on the total fruit removal on that raceme (t =
9.032, df = 99, p < 0.0001; Table 1). Over the course of the study, racemes on fully ripe
bushes lost an average of 42 fruits (41.60 5: 18.74), or approximately 89% of their initial
complement of fruits. On bushes with mixed displays, racemes lost an average of 52
ﬁ'uits (52.46 i 18.83), which amounted to around 98% of their him. This difference in
total ﬁ'uit removal between display treatments was not signiﬁcant; although there was a
very strong trend for mixed displays to have higher removal than fully ripe displays (t =
1.857, df = 99, p<0.07; Table 1). Racemes on ﬁilly ripe bushes lost an average of 2.66
ﬁ'uits (i: 8.78) every two days, while racemes on mixed bushes lost an average of 3.033
(:I: 9.57) every two days.

When we ran ran our model without including fruit count, the effect of display
treatment became signiﬁcant (t = 3.262, df= 100, p < 0.002). The correlation coefficient
of 0.765 for fruit count and display treatment suggests the possibility that the two
variables might be confounded in our full model. If this is the case, the true effect of
display treatment could be closer to what the simpler model reveals. This strengthens the

potential for display treatment to have a signiﬁcant effect on total fruit removal.

 

The type of frugivory also had a signiﬁcant effect on ﬁuit removal. Throughout
the course of our study, we recorded 840 instances of fruit removal. Of these only 42, or
around 5%, were categorized as deer removal. Despite this, deer frugivory was
responsible for 26% of the total fruit removal, indicating that individual instances of deer
ﬁ'ugivory removed more fruit than did instances of avian frugivory. Indeed, our analysis
of total removal per raceme revealed that deer frugivory resulted in signiﬁcantly larger
amounts of fruit removal than avian frugivory (t = 3.04, df = 99, p < 0.005; Table 1). The
average total number of berries removed ﬁom a raceme that experienced deer ﬁ'ugivory
was 53 (52.73 :t 12.77), compared to 44 (44.46 i 21.14) for racemes experiencing only
avian, ﬁ'ugivory (Figure 2). Total removal was not signiﬁcantly correlated with the

number of days that removal was recorded (t = 0.76, df = 99, p <0.5; Table 1).

Table 1. Fixed effects for a linear mixed model predicting total fruit removal per raceme.
Model: total ﬁ'uit removal ~ display treatment + ﬁ'ugivory type + fruit count + trial

duration, random effects of pair + pair/bush.

 

ﬁxed effect value std. error df t-value p-value
Intercept -6.423 5.482 99 -l . 172 0.244
Display trt (mixed) 5.138 2.768 99 1.857 0.066
ﬁ'ugivory type (deer) 7.991 2.629 99 3.040 0.003
fruit count 0.932 0.103 99 9.032 <0.0001
trial duration 0.057 0.079 99 0.755 0.452

10

80 i
70 ~
60 i
50 ~

404 9

fully ripe bushes

30 ~
’ o
20 a

10—
O

0 I I I I I | . I 1
0 1O 20 30 40 50 60 70 8O 90

 

 

nixed bushes

Figure 1. Difference in total removal between mixed and fully ripe treatments at each
site. Data points below the line had higher ﬁuit removal on mixed bushes. Data points

above the line had higher ﬁ'uit removal on ﬁilly ripe bushes.

lI

4o-

35-

30-

254

20-

percentage of observations

    

 

 

 

 

 

 

 

 

 

 

15 -
a
10 —
\'
x
5‘ x
8 II as
0 T T I T T T \\\ I I 1
1-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99

total fruit removal

Figure 2. Total number of fruits removed for racemes with and without deer ﬁ'ugivory.
Clear bars represent racemes with only bird ﬁ'ugivory. Hatched bars represent racemes

that had deer frugivory.

12

Experiment 2

In our second experiment, we saw considerable ﬁ'uit removal on both our red and
green artiﬁcial stems, with each plants’ green stems experiencing an average loss of 2
fruits (2.00 321.60) every 2 days, and red stems experiencing an average loss of 2.1 fruits
(2.10 i 1.72) every 2 days. However, we found no signiﬁcant treatment effect (F = 0.56,
df = 1,23, p > 0.4). The trial duration, or total number of days the stems were on the
bush, was not signiﬁcantly related to total removal (F = 0.471, df =1, 68 , p < 0.005;
Table 2).

Trial identity had a signiﬁcant impact on total ﬁ'uit removal, with later trials
having more fruit removal than trials conducted at the beginning of the season (F = 24.31,
df= 1, 68, p<.0001). Ripeness index was not signiﬁcant (F = 0.032, df= 1, 68, p < 0.90).
However, there was a non-signiﬁcant but very strong trend for an interaction between the
ripeness index and trial (F = 2.95, df = 2, 68, p <0.06; Table 2). Trials that began early in
the season showed no clear correlation between ripeness index and fruit removal.
However, for trials conducted during the latter portion of our study, ﬁ'uit removal seemed
to increase with ripeness index, suggesting that bushes with higher proportions of ripe

fruits might have received higher frugivory.

l3

Table 2. Fixed effects for a linear mixed model predicting total fruit removal per bush.
Model: total fruit removal ~ stem color treatment + trial duration + trial + ripeness index

+ ripeness index:trial, random effects of bush + bush/stem.

 

ﬁxed effect num. df den. df F-value p-value
intercept 1 68 1 17.447 <.0001
stem color (pink) 1 23 0.562 0.461
trial duration 1 68 0.471 0.495
trial 2 68 24.313 <.0001
ripeness index 1 68 0.032 0.858
ripeness indexztrial 2 68 2.953 0.059

14

Discussion

The results of both our experiments suggest that the presence of both unripe and
ripe this on a pokeweed plant might increase ﬁ'ugivory; however we cannot say this
conclusively based on our data. In our ﬁrst experiment, bushes that contained both ripe
and unripe fruits showed a strong trend toward experiencing more frugivory than their
counterparts with only ripe ﬁuits. When we removed the potentially confounding
variable of fruit count, that trend became a signiﬁcant effect.

There are several possible explanations for why mixed displays would attract
more ﬁ'ugivores than ﬁilly ripe displays. It is possible that bushes with a combination of
ripe and unripe fruit are more easily seen by frugivores. If this is the case, increased
ﬁ'ugivory would not be a result of any direct preference for the fruits of one bush over
another. Rather, mixed displays would result in increased fruit removal due to a
detection bias on the part of the frugivores. Schaefer et al. reported that components of
ﬁuiting displays other than the leaf and ripe fruit, such as an unripe fruit, provide more
contrast against the display’s background than the ripe fruit (2007).

Another, not mutually exclusive, explanation for the higher levels of ﬁ'ugivory on
mixed displays is that the presence of unripe fruit might be providing valuable
information to the frugivores and attracting them to the plant. Greig-Smith (1986)
hypothesized that the presence of unripe fruits might indicate that ripe ﬁ'uits on the same
plant were fresh and unlikely to be damaged or desiccated.

By placing our fully ripe and partially ripe bushes side by side, we hoped to
eliminate or drastically decrease any opportunity for our results to be based on detection

bias. If unripe fruits do increase the conspicuousness of a pokeweed’s display, once a

15

frugivore has been drawn to the bush, the neighboring plant with only ripe fruits should
be readily apparent as well. Furthermore, unripe pokeweed fruits turn from green to a
reddish-purple before ripening to black. The green is similar in color to the plant’s
leaves, and the reddish-purple fruits are very similar in color to pokeweed’s stems. This
suggests that the unripe fruits of pokeweed are not adding any additional colors to the
pokeweed display, and while they could be increasing the amount of each color present,
we do not believe that their presence signiﬁcantly alters the visual contrast or
conspicuousness of the display Therefore, we suggest that if there is increased ﬁ'uit
removal on mixed displays of pokeweed, it would be due to a preference for their fruits
rather than an inability to easily notice the fully ripe bushes.

However, if there is no real difference between fruit removal on mixed and ﬁilly
ripe displays, we must look for other explanations for the asynchronous ripening that
pokeweeds undergo. Since unripe hits are less likely to be attacked by insects,
asynchronous ripening could serve to protect a plant’s reproductive investment from
being destroyed (Mordenmoore & Willson, 1982). Asychronous ripening could also
serve to increase the number and type of birds acting as seed dispersers by ensuring that
not all fruits are removed at one time. This change in the quantity and type of seed
dispersers could inﬂuence the seed shadow and reproductive success of the plant.

In our second experiment, we did not see a difference in fruit removal between
our green and red stems. This indicates that birds are not using the color of our stems to
make foraging decisions on an individual bush. It is likely that the additional color and
contrast gained ﬁom a structurally mixed display like pokeweed’s could play a role in

attracting ﬁ'ugivores from a distance. The added conspicuousness of pokeweed’s bright

l6

red stems could easily catch the eye of a bird and draw it to the plant. Once the birds are
there, the stems might have little to no inﬂuence on fruit choice. This is contradictory to
the results of several laboratory studies in which birds have shown differences in feeding
based on the conspicuousness of the food against its background, even when the food is
placed directly in front of them (Osorio et al., 1999; Schmidt et al., 2004). However,
Honkavaara et al.’s study on fruit preference in redwings found no evidence of a role for
contrast in short-range fruit choice (2004), which is consistent with our own experimental
results.

While not ﬁnding a signiﬁcant effect of structurally mixed displays, our second
experiment did indicate that there is potentially an interaction between ripeness index and
trial identity. This suggested that later in the season, fruit removal increased as the
proportion of ripe fruits on the bush increased. Only one of our bushes had a ripeness
index of 1, meaning it was fully ripe. The rest held both ripe and unripe fruits. As such,
our results are most appropriately interpreted within our range of mixed displays. With
this in mind, our data indicate that in the middle and end of the season, pokeweed mixed
displays with high proportions of ripe fruit will receive more frugivory than bushes with
mostly unripe fruit.

Birds might prefer bushes with large proportions of ripe fruit because it makes it
easier for them to locate and eat the ripe ﬁ'uit. Fully ripe fruit are generally more
palatable and contain more nutrients than unripe fruit (Foster, 1977; Schaefer & Schaefer,
2006). Fruits may also contain more water per fruit when ﬁJlly ripe (Foster, 1977; Poston
& Middendorf, 1988). These characteristics make it more proﬁtable for birds to choose

ripe fruits over unripe fruits, and bushes with high numbers of ripe fruits would likely

17

have increased foraging efﬁciency compared to bushes with few ripe fruits. Accessibility
to ripe fruit has been shown to inﬂuence foraging choices and efﬁciency (Whelan &
Willson, 1994; Schaefer & Schaefer, 2006).

If this trend, as well as the trend from our ﬁrst experiment that mixed displays had
higher fruit removal, are both biologically relevant and not simply a result of chance, it
would suggest the possibility of a trade-off between ripening asynchronously in order to
maintain fruit freshness and having a large proportion of ripe fruits in order to serve as an
efﬁcient foraging destination. Further work would be needed, ﬁrst to determine if these
trends are in fact true relationships, and second to accurately determine the number or
percentage of unripe fruits necessary to signal the freshness of other links on the raceme
while still allowing enough ripe fruits to allow efﬁcient frugivory. Fuentes (1995) found
that unripe hits on the shrub Pistacia terebinthus decreased a bird’s ability to access
ripe ﬁ'uits. Yet Pistacia terebinthus unripe ﬁ'uit also increase the conspicuousness of the
display, which suggests that, like the pokeweed, the presence of unripe fruit on this plant
may inﬂuence frugivory in more than one way.

We did not see a correlation between ripeness index and ﬁ'ugivory at the
beginning of our experiment. Early in the season, for instance late August or early
September, juvenile birds would have little foraging experience. Both ﬁeld and
laboratory studies have shown differences in fruit choice and foraging between juvenile
and adult birds. Juvenile American robins (T urdus migratorius) and redwings (T urdus
iliacus) exhibit different color preferences than adults and may not have formed search
images for ripe fruit (Willson, 1994; Honkavaara et al., 2004). American robin juveniles

may not be able ﬁilly differentiate between ripe and unripe hit, and they are less

18

successful at picking up individual fruits and less skilled at fruit-foraging in general,
compared to their adult counterparts (Vanderhoff & Eason, 2007). If juvenile frugivores
begin the season with relatively low foraging skills and are less likely to reliably
differentiate unripe and ripe fruits, they could forage indiscriminately on plants of
varying ripeness. This could lead to a lack of correlation between plant ripeness and
frugivory early in the season, such as we saw on our pokeweed plants. As the season
progresses and juveniles gain more experience choosing and eating fruits, they might
begin to prefer plants with higher ripeness indices, leading to a positive correlation
between ripeness index and fruit removal.

In addition to our tentative findings regarding mixed displays, we also saw an
unexpectedly large amount of deer frugivory on our pokeweed bushes, with deer
responsible for 26% of the fruits removed. The majority of studies done on ﬁ'ugivory in
temperate ecosystems focus on avian ﬁ'ugivores (Thompson & Willson, 1979; Sorensen,
1981; Willson, 1986; Gosper et al., 2005). While some work touches on the possibility
of mammalian ﬁ'ugivory (Stiles, 1980; Willson & Melampy, 1983), surprisingly little
work has been done on the proclivity of white-tailed deer to serve as seed-dispersers for
plants that produce ﬂeshy ﬁ'uits (Lay, 1965; Myers et al., 2004; Williams & Ward, 2006).

We do not know if white-tailed deer can act as effective seed dispersers for
pokeweed or if they render the seeds inviable during gut passage. Several studies have
indicated that deer eat the fruit and disperse the seeds of the invasive honeysuckle
(Lonicera spp.) (Vellend, 2002; Myers et al., 2004) and the invasive autumn olive
(Elaeagnus umbellata) (Williams & Ward, 2006). There is some evidence that the

likelihood of deer being able to disperse seeds increases as seed size decreases (Willson,

l9

1993). The seeds of both autumn olive and honeysuckle are larger than those of
pokeweed, suggesting that deer might be able to successfully disperse the seeds of
pokeweed.

The months leading up to our study were unusually dry. While rainfall during the
month of August was not low, the total precipitation for April through July at the Kellogg
Biological Station Long-Term Ecological Research (LTER) site, which is close to our
study site, was the lowest for a twenty year period. Dry conditions for much of the
growing season could have signiﬁcantly inﬂuenced the availability of food sources for
white-tailed deer. If this is the case, deer might have consumed more pokeweed fruits
than would be expected under normal conditions.

Future work is needed to determine exactly what role white-tailed deer play in the
reproductive success and dispersal of pokeweed, as well as other ﬂeshy fruit-producing
plants. Our results indicate the potential for deer to have a signiﬁcant impact either as a
seed disperser or as a seed predator. The results of our study also suggest a possibility for
avian ﬁ'ugivory on pokeweed to be inﬂuenced by temporally mixed displays, but in the

context of our work, not by structurally mixed displays.

20

REFERENCES

Armesto JJ, Cheplick GP, and McDonnell MJ. 1983. Observations on the
Reproductive-Biology of Phytolacca-Americana (Phytolaccaceae). Bulletin of the
Torrey Botanical Club 110: 380-383.

Burns KC, and Dalen JL. 2002. Foliage color contrasts and adaptive fruit color
variation in a bird-dispersed plant community. Oikos 96: 463-469.

Cramer JM, Cloud ML, Muchhala NC, Ware AE, Smith BH, and Williamson GB.
2003. A test of the bicolored fruit display hypothesis: Berry removal with
artiﬁcial ﬁ'uit ﬂags. Journal of the Torrey Botanical Society 130: 30-33.

F acelli JM. 1993. Experimental Evaluation of the Foliar Flag Hypothesis Using Fruits of
Rhus-Glabra (L). Oecologia 93: 70-72.

Foster MS. 1977. Ecological and Nutritional Effects of Food Scarcity on 3 Tropical
Frugivorous Bird and Its Fruit Source. Ecology 58: 73-85.

Fuentes M. 1995. The Effect of Unripe Fruits on Ripe Fruit Removal by Birds in
Pistacia-Terebinthus - Flag or Handicap. Oecologia 101: 55-58.

Galetti M, Alves-Costa CP, and Cazetta E. 2003. Effects of forest ﬁagmentation,
anthropogenic edges and fruit colour on the consumption of ornithocoric fruits.
Biological Conservation 111: 269-273.

Gorchov DL. 1990. Pattern, adaptation, and constraint in fruiting synchrony within
vertebrate-dispersed woody plants. Oikos 58: 169-180.

Gosper CR, Stansbury CD, and Vivian-Smith G. 2005. Seed dispersal of ﬂeshy-
ﬁ'uited invasive plants by birds: contributing factors and management options.
Diversity and Distributions 11: 549-558.

Greig-Smith PW. 1986. Bicolored ﬁ'uit displays and frugivorous birds: The importance
of link quality to dispersers and seed predators. American Naturalist 127: 246-
251.

21

Honkavaara J, Siitari H, and Viitala J. 2004. Fruit colour preferences of redwings
(Turdus iliacus): Experiments with hand-raised juveniles and wild-caught adults.
Ethology 110: 445-457.

Lay DW. 1965. Fruit Utilization by Deer in Southern Forests. The Journal of Wildlife
Management 29: 370-375.

Mordenmoore AL, and Willson MF. 1982. On the Ecological Signiﬁcance of Fruit
Color in Prunus-Serotina and Rubus-Occidentalis - Field Experiments. Canadian
Journal of Botany-Revue Canadienne De Botanique 60: 1554-1560.

Myers JA, Vellend M, Gardescu S, and Marks PL. 2004. Seed dispersal by white-
tailed deer: implications for long-distance dispersal, invasion, and migration of
plants in eastern North America. 0ecologia 139: 35-44.

Orrock JL. 2005. The effect of gut passage by two species of avian frugivore on seeds of
pokeweed, Phytolacca americana. Canadian Journal of Botany-Revue
Canadienne De Botanique 83: 427-43 1.

Osorio D, Miklosi A, and Gouda Z. 1999. Visual ecology and perception of coloration
patterns by domestic chicks. Evolutionary Ecology 13: 673-689.

Pinheiro J, Bates D, DebRoy S, and Sarkar D. 2006. nlme: Linear and nonlinear mixed
effects models. R package version 3.1-78.

Poston ME, and Middendorf GA. 1988. Maturation Characteristics of Rubus-
Pennsylvanicus Fruit - Are Black and Red the Same. 0ecologia 77: 69-72.

R Development Core Team. 2006. R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria. 3-900051-
07-0, http://www.R-project.org.

Schaefer HM, and Schaefer V. 2006. The fruits of selectivity: How birds forage on
Goupia glabra fruits of different ripeness. Journal of Ornithology 147: 638-643.

Schaefer HM, Schaefer V, and Vorobyev M. 2007. Are fruit colors adapted to
consumer vision and birds equally efﬁcient in detecting colorful signals?
American Naturalist 169: $159-$169.

22

Schmidt V, Schaefer HM, and Winkler H. 2004. Conspicuousness, not colour as
foraging cue in plant-animal signalling. Oikos 106: 551-557.

Smith KG, and Riley CM. 1990. Avian Removal of Fruits from a Pokeweed in
Northwestern Arkansas. Wilson Bulletin 102: 163-166.

Snow DW. 1971. Evolutionary Aspects of Fruit-Eating by Birds. [his 113: l94-&.

Sorensen AE. 1981. Interactions between Birds and Fruit in a Temperate Woodland.
0ecologia 50: 242-249.

Stiles EW. 1980. Patterns of Fruit Presentation and Seed Dispersal in Bird-Disseminated

Woody-Plants in the Eastern Deciduous Forest. American Naturalist 116: 670-
688.

Thompson JN, and Willson MF. 1979. Evolution of Temperate Fruit-Bird Interactions -
Phenological Strategies. Evolution 33: 973-982.

Vanderhoff EN, and Eason PK. 2007. Disparity between adultand juvenile American
robins Turdus migratorius foraging for ground invertebrates and cherry fruits.
Ethology 113: 1212-1218.

Vellend M. 2002. A pest and an invader: White-tailed deer (Odocoileus virginianus
Zimm.) as a seed dispersal agent for honeysuckle shrubs (Lonicera L.). Natural
Areas Journal 22: 230-234.

Wenny DG. 2003. Short-term fruit removal from a neotropical shrub with experimental
bicoloured displays. Journal of Tropical Ecology 19: 469-472.

Whelan CJ, and Willson MF. 1994. Fruit Choice in Migrating North-American Birds -
Field and Aviary Experiments. Oikos 71: 137-151.

Williams SC, and Ward JS. 2006. Exotic seed dispersal by white-tailed deer in southern
Connecticut. Natural Areas Journal 26: 383-390.

Willson MF. 1986. Avian ﬁ'ugivory and seed dispersal in eastern North America.
Current Ornithology 3.

23

Willson MF. 1993. Mammals as Seed-Dispersal Mutualists in North-America. Oikos 67:
159-176.

Willson MF. 1994. Fruit Choices by Captive American Robins. Condor 96: 494-502.

Willson MF, and Melampy MN. 1983. The Effect of Bicolored Fruit Displays on Fruit
Removal by Avian Frugivores. Oikos 41: 27-31.

Willson MF, and Thompson J N. 1982. Phenology and Ecology of Color in Bird-
Dispersed Fruits, or Why Some Fruits Are Red When They Are Green. Canadian
Journal of Botany-Revue Canadienne De Botanique 60: 701-713.

Willson MF, and Whelan CJ. 1989. Ultraviolet Reﬂectance of Fruits of Vertebrate-
Dispersed Plants. Oikos 55: 341-348.

24

M'IllllljlllljljlllllLiljlljljlllljllljllllllfs