Discussion
The observation that the value of the Shannon Index increased with
rising number of food types indicates that the index accurately
estimates real changes in diet breadth. Moreover, the fact that the
values for Shannon Evenness did not significantly change with increasing
diet breadth also supports the utility of this approach to determine
non-random diet preferences.
The diversity approach for analyzing data collected from cafeteria
trials is useful in its simplicity. Statistically, there is increased
power by using a simple hypothesis test to determine if foragers are
consuming resources in the same proportion in which they encounter them.
One shortcoming is that this analysis will only detect that preferences
exist but cannot provide further information about specific preferences,
which requires additional testing. The diversity index can test for diet
differences among species, but can be misleading, as the data collected
for this study demonstrate. For example, although the mean Shannon
Evenness value for N. minimus (0.29) is statistically identical
to that for T. striatus (0.28), their foraging preferences are
actually highly divergent. The reason for the similarity in index values
is that the resources avoided by one species are the ones preferred by
the other (i.e., these species have complimentary foraging preferences).
Specifically, A. rubrum was avoided by T. striatus but
preferred by N. minimus , whereas A. saccharum and C.
cornuta was avoided by N. minimus but preferred by T.
striatus . One solution to this problem would be to compare species
using a different index, such as one for differentiation diversity (β
diversity; Maurer & McGill, 2011).
There are other potential applications of this approach to understand
foraging decisions. The greatest strength of using diversity indices to
quantify foraging choices is that these indices account for both number
of different food types consumed and the proportion in which they are
consumed simultaneously. In cases where foragers sample every food type
offered, investigators can focus solely on evenness. Another strength of
this approach is that it is unlimited in terms of loss of statistical
power by the number of different resources that can be tested
simultaneously. Moreover, the variation in index values can aid in
further testing. For example, the variability in index values was
greater for T. striatus , indicating that this species consumed
different amounts of resources relative to N. minimus . Another
application incorporates testing hypotheses when the total amount of
each resource is not equal. The amounts of each resource can be varied
without loss of discriminatory power. In other words, this approach
allows for behavioral titrations, in which amounts of each resource can
be modified to test hypotheses concerning how resource availability
affects foraging decisions. For example, future studies can test the
strength of preference for A. rubrum seeds by N. minimusby altering the amount of A. rubrum seeds relative to other seed
types.
Preferences for seeds by each chipmunk species clearly matched
differences in body size. The smaller-bodied N. minimus preferred
the small seeds of A. rubrum , whereas T. striatusconsumed more of the larger seeds of A. saccharum and C.
cornuta . This result is consistent with the concept of size assortment,
which has been documented in other speciose rodent communities (Brown &
Lieberman, 1973; Emmons, 1980; Brown & Bowers, 1984; Ivan & Swihart,
2000). Size assortment is a mechanism to reduce interspecific
competition pressures. Chipmunks are well-known to be aggressive toward
other species and are cited as a textbook example of how interspecific
competition can structure communities (Chappell, 1978). Thus, the
finding of size assortment between these species is consistent with
limiting similarity and is not surprising.
Preferences within each individual chipmunk species can be interpreted
using optimal foraging theory. The smaller N. minimus may avoid
larger seeds such as C. cornuta and A. saccharum not only
due to competition with T. striatus but also because large seeds
have a longer handling time relative to reward. This is a more likely
explanation than direct interference competition, as each individual was
able to forage freely during the trials. A. rubrum is a common
seed found in this system and is preferred by other rodents such as deer
mice (Peromyscus maniculatus gracilis ) presumably because their
seed coat is thin, thus reducing handling time (Cramer, 2014). P.
pensylvanica , also comparable in size, has a tough seed coat, and may
be avoided by N. minimus for this reason.
Both chipmunk species avoided consumption of A. balsamea , a
result consistent with studies that have shown that seed predators in
general and Peromyscus maniculatus and Myodes(=Clethrionomys ) gapperi specifically avoid consumption ofA. balsamea seeds (Abbott, 1962; Lobo, 2014). Consumption ofA. balsamea seeds by chipmunks has not been tested explicitly.
Presumably, seed predators avoid fir seeds because they contain a high
proportion of distasteful plant secondary compounds (Rubino et al.,
2012). Another potential reason for low consumption of A.
balsamea in this study centers on seed presentation. In nature,
sciurids are known to collect conifer seeds while they are still in
cones, for easier and more efficient transport to seed middens (Smith,
1970). In this study, fir seeds were presented without cones, and
foragers may have viewed individual seeds encountered in this manner as
not worth the effort.
Analysis of T. striatus foraging preferences uncovered an
interesting relationship. Overall, these large-bodied chipmunks
preferred the seeds of larger species, A. saccharum and C.
cornuta , in concordance with other studies of foraging preferences.
However, closer examination of individual chipmunk preferences exposed a
significant negative correlation between amount of C. cornutaconsumed compared to amount of A. saccharum consumed (Kendall’s
rank correlation: T = 26, P = 0.04, tau = -0.43).
This suggests that individuals decided to consume either hazelnut or
maple, but not both. Although this result is preliminary and based on a
relatively small sample size (n = 14), it could have serious
ramifications on the interpretation of foraging preferences because of
the suggestion that preferences can vary among individuals. More
detailed data collection will determine the ubiquity of this
observation. For example, it is possible that individuals that showed a
preference for hazelnut over maple are simply those that had encountered
hazelnut in the past, or even those that sampled the hazelnuts first at
the start of the trial. As total amounts eaten were measured, and not
the order in which seeds were sampled, the data reported herein cannot
definitively address this question.
Ultimately, chipmunk species composition is changing in the Great Lakes
region. Myers, Lundrigan, Hoffman, Haraminac, and Seto (2009) have
demonstrated a decrease in N . minimus and an increase inT . striatus populations in the Great Lakes. They attribute
this to a concomitant widespread increase in average minimum temperature
of 2.1°C. This potential shift in abundances between chipmunk species
could alter forest composition through differences in seed preferences
and caching behavior.