IMPACTS OF A SEED PREDATOR ON SUNDIAL LUPINE
Jennifer Shimola
A Thesis
Submitted to the Graduate College of Bowling Green State University in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE
August 2013
Committee:
Dr. Helen Michaels, Advisor
Dr. Randy Mitchell
Dr. Daniel Pavuk
Dr. Daniel Wiegmann © 2013
Jennifer Shimola
All Rights Reserved iii
ABSTRACT
Helen Michaels, Advisor
Sundial lupine, Lupinus perennis (L.), is a perennial flowering legume integral to
maintaining reproductive populations of the federally endangered Karner Blue butterfly.
The seed coats of L. perennis are polymorphic, with variable amounts of dark speckling
ranging from light to heavily speckled against a white to gray background. In this legume
species, darker seeds may contain secondary metabolites like tannins and anthocyanins,
which may be associated with deterring seed predation. Other studies have found lighter
colored seeds to have weaker seed coats, be more digestible, and have higher nutritional
value than darker seeds. The chemical and physical traits of lighter seed coats could make
this seed type more desirable for seed predators as well as agriculturalists.
The seeds of L. perennis are a known food source of ant-mimicking alydid insects,
Megalotomus quinquespinosus and Alydus spp. This study explored whether alydids have
a preference for seed coat color variants of L. perennis and simultaneously determined the
relative abundance of this seed predator and the frequency of L. perennis seed colors in
field surveys of ten populations. We expected to find that 1) alydids would be found at
greater frequencies when lighter seeds were present and 2) the most amount of time
during a seed choice experiment will be spent interacting with lighter seeds. We also
measured environmental characteristics to determine whether other factors may be
playing a role in the distribution of L. perennis seed color or alydids. We expected soil characteristics to camouflage seeds, canopy cover and litter depth would influence site productivity, and that greater food availability would increase alydid abundance. iv
Populations of L. perennis were surveyed for seed color frequencies, alydid frequencies, environmental variables, and the presence of other legume species. Sites were confirmed to differ in the amount of speckling and proportion of seeds with speckling present. Seed speckling was impacted by soil characteristics, canopy cover, and pod abundance. Alydid abundance was impacted by food availability (pod abundance and the presence of other legumes), soil characteristics, and the amount of speckling. Fewer alydids were found in populations with darker speckled seeds.
A multiple choice behavioral experiment was performed on adult M. quinquespinosus to clarify findings of the population survey. There was no significant difference in alydid behavior with different seed colors, though we observed a trend of greater time interacting, a greater proportion of time, and more frequent visits to darker seeds. Based on the findings of our population survey and behavioral experiment, alydids did not have a strong preference for lighter seed colors in L. perennis. Some explanations for the differences in seed color abundance among sites are that seed color is random or undergoing selective pressures that we did not consider such as post-dispersal seed predators, the attraction of beneficial microbes, maintaining non-synchronous germination, or pleiotropic effects. Understanding whether this seed color polymorphism is important to L. perennis fitness may be important in future management decisions and should be further studied. v
ACKNOWLEDGMENTS
First and foremost, I wish to acknowledge Dr. Helen Michaels, as she has helped
me find a project to be enthusiastic about, helped me develop the skills to improve myself
in the field of ecology, and inspired me to pursue this research further. I would also like
to thank my committee, Dr. Randy Mitchell, Dr. Daniel Pavuk, and Dr. Daniel
Wiegmann, for their helpful input in designing this project. Thank you to Nancy
Boudreau and various graduate students in the statistical consulting office of Bowling
Green State University’s Business Administrative department who helped difficult data
become analyzable.
Furthermore, I would like to thank my lab mates (Mike Plenzler, Ryan Walsh,
Jacob Meier, Jacob Sublett, Paige Arnold, and Alyssa Dietz) for keeping the lab fun, answering endless inquiries, and helping me become both a better presenter and writer through many tedious lab meetings. A big thanks to the CURS program at BGSU for funding this project and giving me the opportunity to mentor an undergraduate field assistant, Jacob Sublett, whose hard work and enthusiasm for field and lab work is greatly appreciated
Finally, I would like to thank the management agencies that permitted us access to
our lupin populations: The Metroparks of the Toledo Area, Ohio Department of Natural
Resources, Division of Wildlife, and The Nature Conservancy.
vi
TABLE OF CONTENTS
Page
INTRODUCTION ...... 1
METHODS...... 6
Study Species and Study Area...... 6
Population Survey...... 8
Behavioral Experiment...... 9
Statistical Analysis...... 10
RESULTS...... 13
Population Survey...... 13
Population Means Analysis...... 16
Behavioral Experiment...... 17
DISCUSSION...... 19
REFERENCES...... 25
TABLES AND FIGURES...... 35
APPENDIX A. Correlation of Variables ...... 66
APPENDIX B. Parameter Estimates ...... 67
vii
LIST OF TABLES
Table Page
1 GPS coordinates and managing agencies of populations surveyed ...... 35
2 Percent of samples with white or speckled seeds by site ...... 36
3 Population means for environmental variables ...... 37
4 The relationship of site, alydid abundance, soil value, and the presence of two other
legumes on the amount of seed speckling...... 38
5 The relationship of environmental factors and seed characteristics on the presence of seed
speckling and alydid abundance ...... 39
6 Models of speckling and alydid abundance ...... 40
7 The time spent with seeds and proportion of time spent with seeds by seed color ... 41
8 Relationship of first visit and frequency of visits to seed color ...... 42 viii
LIST OF FIGURES
Figure Page
1 Distribution of amount of speckling by site ...... 43
2 Relationship between site and amount of speckling ...... 44
3 Relationship of soil value and amount of speckling ...... 45
4 Relationship of the presence of two other legumes and the amount of speckling ..... 46
5 Relationship of site and the presence of speckling ...... 47
6 Relationship of soil chroma and presence or absence of speckling ...... 48
7 Relationship of pod density and the proportion of seeds with speckling present ...... 49
8 Relationship of Lespedeza capitata presence and the proportion of seeds with speckling
present...... 50
9 Relationship of site and alydid abundance when considering the degree of speckling 51
10 Relationship of soil value and alydid abundance when considering the degree of
speckling...... 52
11 Relationship of canopy cover and alydid abundance when considering the degree of
speckling...... 53
12 Relationship of average amount of speckling and alydid abundance when considering
the degree of speckling...... 54
13 Relationship of site and alydid abundance when considering the presence of
speckling...... 55
14 Relationship of soil chroma and alydid abundance when considering the presence of
speckling...... 56 ix
15 Relationship of canopy cover and alydid abundance when considering the presence of
speckling...... 57
16 Relationship of site area and the amount of speckling for population means...... 58
17 Relationship of alydid abundance and the amount of speckling for population means.. 59
18 Relationship of pod abundance and proportion of samples at a site with speckling present
...... 60
19 Relationship of the amount of speckling and alydid abundance when considering the
amount of speckling...... 61
20 Relationship of the presence of Lespedeza capitata and alydid abundance when
considering the amount of speckling ...... 62
21 Relationship of the presence of Lespedeza capitata and alydid abundance when
considering the presence of speckling...... 63
22 Average amount of time, proportion of time, frequency of visits, and first visits to seeds
by seed color...... 64
1
INTRODUCTION
A variety of both vertebrate and invertebrate organisms exploit seeds as a source of food
(Hulme & Benkman 2002). Seed predation can have significant has direct effects on fecundity that affect both population dynamics and plant traits (Kolb et al. 2007). Seed predators can impact plant population dynamics by reducing both quantity and quality of emerging seedlings
(Janzen 1976; Hulme & Benkman 2002, Silman et al. 2003). Many granivores, especially rodents and birds, completely consume seeds, which lowers population fecundity via reducing the number of recruiting individuals (Kolb et al. 2007). Unlike typical herbivory, seed predation does not allow individuals to regenerate destroyed tissues or reproduce, even at a reduced rate
(Kolb et al. 2007). As a result, this type of seed predation is a strong selective pressure.
However, not all granivores completely destroy seeds.
In insects with sucking mouthparts or larvae that parasitize seeds, seeds are not always destroyed, but food reserves are merely depleted (Janzen 1971; Vallejo-Marín et al. 2006).
Depending on the degree of this partial predation, these seeds may still be viable and damage may facilitate germination (Karban & Lowenberg 1992, Takakura 2002, Vallejo-Marín et al.
2006; Lundbye & Johansson 2012). However, seedlings may grow more slowly (Pizo et al.
2006; Kuprewicz & García-Robledo 2010) than seeds that were not damaged. Some studies expect early germination to be a competitive advantage (Lundbye & Johansson 2012), while others suggest that depleted cotyledon size may slow initial growth (Pizo et al. 2006; Perea et al.
2011). These contrasting outcomes may depend on seed size, as larger seeds tend to have increased survival and establishment (Westoby et al. 1992), or on the amount of damage. Some studies have found that when partial seed predation results in low amounts of damage, seedlings have similar germination rates, establish successfully, and may even grow more quickly than 2
intact seeds (Branco et al. 2002; Perea et al. 2011). However, when the damage from partial seed predation is high, lower seedling quality may result as smaller food reserves commonly yield smaller seedlings (Westoby et al. 1992; Branco et al. 2002).
Due to the variety of seed predator-induced pressures, plants have developed many
strategies that deter seed predation (Hulme & Benkman 2002). These strategies range from
physical barriers such as hard seed coats to chemical deterrents, which can cause physical
damage to granivore hosts (Oigianbe & Onigbinde 1996; Hulme & Benkman 2002). In a review
by Strauss et al. (2002), the cost of maintaining chemical herbivory defenses was found to be
quite high, with a reduction in fitness that ranged from 8.7 – 73 % in natural populations. These
costs can be outweighed if the yield of plants increases when the pressure from seed predation is
high (Strauss et al. 2002).
Plants from the family Fabaceae are excellent resources for seed predators due to the high
nitrogen content of their seeds and other tissues (Wang et al. 2003). Interestingly, members of
this family also commonly contain alkaloids and phenolics, secondary metabolites that are
involved in deterring seed predators (Dixon & Sumner 2003). Phenolics like tannins have been
found to increase seed coat strength and decrease water permeability in some legume species
(Marbach & Mayer 1974; Werker et al. 1979; Legesse & Powell 1996), though other species do
not display this correlation (Slattery et al. 1982; Argel & Humphreys 1983). Stronger,
impermeable seeds will not only resist seed predation, but will also increase longevity of seed
banks by reducing germination synchronicity (Souza et al. 2001). Furthermore, seeds that lack
these secondary metabolites have been found to have weaker seed coats, increased digestibility,
and higher nutritional value than seeds with high concentrations of alkaloids or phenolics
(Duranti & Guis 1997; Moïse et al. 2005). 3
The chemical and physical traits of seeds that lack secondary metabolites are not only
more desirable for seed predators, but may also be important to agriculturalists. In fact, some
agricultural legumes have been selectively bred for lower alkaloids and tannins to reduce the
undesirable bitter flavor of seeds with high concentrations of secondary metabolites (Duranti &
Guis 1997). One of these agricultural lupins, Lupinus albus, has been recently explored for use as
a forage crop in the United States, though lupin has been used agriculturally as a source of green
manure in the United States since the 1940s (Bhardwaj et al. 2010). Agricultural lupins have
been bred for reduced alkaloids (Duranti & Guis 1997) and to have white seeds, though one may
not be a consequence of the other as both of these traits are desirable to agriculturalists
(Gustafsson & Gadd 1964). Legume seeds with high tannin content may also be less desirable
for agriculture due to their increased dormancy and, therefore, non-synchronous germination
(Woodstock 1988; Argel & Patton 1999). Alternatively, seeds with high phenolic compounds
may be favored by agriculturalists, as these compounds have antioxidant properties (Dixon &
Sumner 2003; Moïse et al. 2005).
One legume, Lupinus perennis L. (sundial lupine), is an indicator species of oak savanna
ecosystems, which are declining in the mid-western United States (Ohio Department of Natural
Resources 1998, Pavlovic & Grundel 2009). L. perennis is an important species to preserve, as it
sustains rare butterfly species, including the federally endangered Karner Blue butterfly,
Lycaeides melissa samuelis (Clough 1992; Ohio Department of Natural Resources 1998).
Populations of L. perennis have seeds that range in color from cream to darkly mottled, but
individual plants produce one seed type (Grigore & Tramer 1996). The pigment in L. perennis
seeds has not been identified, but other species of legume with similar phenotypes contain
darkly-pigmented secondary metabolites (Moïse et al. 2005; Xu et al. 2007). Additionally, dark 4
seeded phenotypes have been studied to a small extent in other members of the genus Lupinus
(Horovitz & Harding 1983). Dark pigments in other legumes have been found to be heritable
(Horovitz & Harding 1992; Moïse et al. 2005; Yang et al. 2010), though there is great
environmental plasticity in pigment expression (Souza & Marcos-Filho 2001). Furthermore, this trait could be associated with other traits; For example, a pleiotropic effect of seed color and flower color exists in some species (Carlson & Holsinger 2013). Therefore, even if this trait is neutral, selection may be acting on another trait that is associated with seed color (Armbruster
2002; Strauss & Irwin 2004; Coberly & Rausher 2008). Conversely, the variable frequencies of this polymorphism could be due to factors other than selection, like non-random mating or genetic drift, as this trait may not be important to the fitness of L. perennis.
Although other factors may influence the frequency of seed color, seed predation is one pressure that may be influencing seed color in L. perennis. In a field seed removal experiment, mice removed 22% of experimentally offered L. perennis seeds (Kappler et al. 2012), while a preliminary study suggests that larger seed predators, like mice and birds, have a preference for darker L. perennis seeds (Wakeley et al. unpublished data). Vertebrate seed predators were
suspected to be responsible for the seed removal in this study, as L. perennis seeds, which are
large (8 - 41 mg in a study by Halpern (2005)) compared to alydids, were usually totally
consumed or removed.
One type of seed predator that was not previously studied, as they are incapable of
carrying seeds, is the alydid insects. The Alydidae is a family (Order Hemiptera: Suborder
Heteroptera) known to utilize grass and legume seeds as their main food source (Schaefer 1980).
The family Alydidae has three subfamilies, of which the Alydinae is the only subfamily known
to prefer legume hosts (Schaefer 1980). Some alydid species have been observed as agricultural 5
pests of legumes, both in the United States and abroad (Wilkinson & Daughtery 1967; Yonke &
Medler 1968; Panizzi 1987). These insects are known to feed pre-dispersally from pods, though,
like other Hemipterans, they likely rely on chemoreception and mechanoreception for
determining palatable food sources when at close distances (Hulme & Benkman 2002). Alydids
like Megalotomus quinquespinosus (Say) and Alydus spp. can be frequently observed preying upon the pods of L. perennis (Yonke & Medler 1964; personal observation). M. quinquespinosus
was not considered an agricultural pest by Yonke & Medler (1964) and relatively little work has
been done to further explore the impacts of this seed predator. Of the three Alydus spp., Alydus
eurinus and Alydus pilosulus have been found to prey on agricultural legumes (Underhill 1943).
However, experimentally damaged seeds of L. perennis showed a negative effect of partial seed
predation (Zhang & Maun 1991). In fact, percent emergence decreased, leaf area decreased, and
seedling dry weight decreased in damaged seeds when compared to intact seeds (Zhang & Maun
1991). Similar reductions in seedling size have been observed in Ohio L. perennis when seeds
with obvious puncture marks were grown alongside undamaged seeds (Abell et al., unpublished
data). Therefore, it is possible that partial predation by alydids could impact L. perennis fitness.
Alydid response to seed color is unknown, and their distribution and abundance in populations of
wild legumes have never been studied.
As there are potential impacts to L. perennis populations, butterfly conservation, and the
success of agricultural legume species, understanding the factors that influence seed color
frequencies and distributions should be explored. Because this polymorphism is maintained,
other seed predators appear to have a preference for darker seeds, and many anti-herbivore chemicals result in darker pigmentation, we hypothesized that alydids would prefer lighter L. perennis seeds. To determine whether the distribution of seed color types in L. perennis is related 6
to alydid abundance and whether alydids have a preferred seed color, we performed a field survey of ten L. perennis populations and a multiple-choice behavioral experiment based on seed color. We expected to find that 1) alydids would be found at greater frequencies when lighter seeds are present and 2) more time during the seed choice experiment will be spent interacting with lighter colored seeds (those with less speckling or pigmentation).
As alternatives to this hypothesis exist, we also measured environmental characteristics to determine whether other factors may be playing a role in the distribution of L. perennis seed color or alydids. We expected that seed color polymorphisms may be related to variation in soil properties that affect seed visibility within and among populations. Furthermore, canopy cover and litter depth may influence the abundance or foraging efficiency of other L. perennis seed predators (e.g. Peromyscus spp.; Kappler et al. 2012), indirectly affect soil color through incorporation of organic matter, or may affect L. perennis productivity (pod density). Finally, the productivity of sites as well as the presence of other legume species may increase the abundance of alydids by increasing food quantity as well as food availability throughout the year.
METHODS
Study Species and Study Area
Sundial lupine [Lupinus perennis (L.)] is a perennial legume that ranges from Maine to
Florida and west to Minnesota and is an important host to threatened and endangered butterflies
(Ohio Department of Natural Resources 1998). L. perennis is considered an indicator species of high quality oak savanna ecosystems (Grigore and Windus 1994), which are declining in the mid-western United States due to agriculture, fire suppression, and anthropomorphic encroachment (Nuzzo 1986). Flowering begins in May, fruits develop by June, and seeds are 7
balistically dispersed by July (Grigore & Tramer 1996). The hard seed coats of this species
require scarification for high rates of germination (Mackay et al. 1996, Kelly 1998), and as a
possible result, this species exhibits asynchronous germination (Grigore & Tramer 1996). L.
perennis seed color ranges from solid cream seeds to darkly speckled seeds, though seed color is
consistent within a single plant (Grigore & Tramer 1996).
Megalotomus quinquespinosus and Alydus spp. are Hemipteran insects in the family
Alydidae, which tends to have ant-mimicking nymphal stages. These alydid species are reported
to emerge in May and June, have five instars, complete two generations per year, and overwinter as eggs (Yonke & Medler 1964, 1968). Both M. quinquespinosus and Alydus spp. have been observed to utilize L. perennis seeds as a food source. These insects are known to be pre- dispersal predators. It is likely that these insects feed on L. perennis seeds post-dispersally as
well, but no direct observations of this behavior in the field have been made.
Study sites were distributed across northwest Ohio’s Oak Openings region. Ten
populations of L. perennis were used for seed and insect collection and a population survey
(Table 1). These populations were randomly selected among populations that did not contain
reintroduced Karner blue butterflies as access to occupied sites is prohibited to maximize
restoration success. Sites needed to have at least twenty, 12.56 m2 areas that did not overlap and contain L. perennis plants with pods in order to fit our experimental design with sufficient replication. Sites have been managed, though management practices differed between agencies.
One site, Mary’s Savanna, has had a recent reduction in canopy cover due to a tornado. Sites had a wide variation in canopy cover and litter depth.
8
Population Survey
Ten populations of L. perennis were surveyed from June 4-June 20, 2012 for seed color frequencies, alydid frequencies, and environmental variables. Within each population, thirty 4 meter diameter circles (12.56 m2 area) were sampled every 5 meters along transects. Transects were randomly selected and were at least five meters apart to maximize the chance of finding alydids without having sampling areas overlap between transects. Seed color, pod density, alydid sightings, soil color, litter depth, canopy cover, and population area were recorded for each sampling location. The presence or absence of other legume species (Desmodium canadense,
Lespedeza capitata, Tephrosia virginiana, and Baptisia tinctoria) was also recorded within each population.
At each sampling point one plant closest to each cardinal direction was chosen from which mature, dried pods and their seeds were collected to characterize seed color and score pre- dispersal predation damage. Seed color was quantified using ImageJ (US National Institutes of
Health, Bethesda, Maryland, USA) to determine the average area of dark speckling. Seeds were photographed in the lab with supplementary lighting. Both halves of all seeds were photographed, as slight variation in speckling does occur. Photographs were converted to 16-bit grayscale, the contrast threshold was altered to distinguish seed speckling from background color, and the seed area was selected to determine the percent of seed area covered by dark pigmentation.
Alydid sightings were recorded when temperatures were above 21 C between the hours of 10 AM and 3 PM, though alydid activity was observed at temperatures as low as 12 C. Soil color was obtained by using the Munsell Soil Chart (Munsell Color, NY, USA) to estimate 9
chroma, value, and hue on exposed soil. Litter depth was measured at the center of each
sampling location to the nearest 0.5 cm. Pod density was scored as the total number of pods that
occurred in each sampling area (12.56 m2).
Canopy photographs were taken at 45 cm from the ground, the approximate height of a L. perennis infructescence. Canopy cover photographs were also analyzed in ImageJ by converting to 16-bit grayscale and altering the contrast threshold to determine the area covered by canopy.
Population area was estimated by connecting waypoints taken every 10 meters around populations of L. perennis using a Trimble Geo XH (Trimble Navigation,CA, USA). Population
boundaries were defined as the area where L. perennis plants were more than 20 meters apart.
Behavioral Experiment
To clarify findings of the population survey, we designed a behavioral experiment to
determine whether Megalotomus quinquespinosus has a preference for a given seed coat
phenotype. M. quinquespinosus adults were collected for use in behavioral choice experiments
from field sites that had high alydid abundance. Initally, funnel traps baited with ammonia were
used to collect M. quinquespinosus from L. perennis populations (Silva et al. 2010). However,
this method was not effective later in the season, so adults were collected with handheld tulle
nets to reduce impacts of sweep netting on vegetation and endangered butterflies. Insects were
maintained individually in one quart plastic deli containers, fed on a diet of unripe soybean seed,
and offered a moist sponge for water and humidity. Containers had tulle nets for lids and were
kept in a CONVIRON chamber (model MTR30; light levels 877 μEm-2sec-) at 24 C (following
methods in Yonke & Medler 1968 and Kristenová et al. 2011) and 61% relative humidity with a
12 hour light cycle. After collection, insects were offered soybean for a minimum of two days 10
prior to starvation. Insects were then starved for at least five days prior to choice trials to ensure willingness to feed (based on Pekár & Hrušková 2006; Kristenová et al. 2011).
Trials were performed in a 70 cm petri plate under the same growth chamber conditions as insects were housed. Individual bugs and seeds were used in a trial only once. Seeds from 3-6 randomly chosen plants were used in trials and each individual was offered seeds from different plants. These seeds were soaked for 24 hours to ensure seed coats were intact and that no seeds were damaged prior to trials. Three color categories were offered in trials: seeds with 0 % speckling, seeds with 10-30 % speckling, and seeds with greater than 50 % speckling. These categories were chosen as they spanned the typical distribution (range from 0-78.8 % speckling) of seed speckling found in our population survey and they were visually distinguishable for video analysis. Five seeds of each color category were placed evenly around the edge of the dish in an alternating pattern. Individual insects were placed in the center of the petri plate. Trials were video recorded for two hours. Videos were scored for first visited seed, number of visits to each color, time spent with each color, and the proportion of time (time spent on each color compared to an individual’s total time spent interacting with seeds during a trial). A visit to a seed was defined to have occurred when an insect touched a seed with its mouthparts, antennae, or forelegs for greater than 20 seconds.
Statistical Analysis
Analyses were performed using SAS 9.1 and JMP 10 (SAS Institute, California, USA).
As the distribution of speckling was non-normal and transformations did not improve fit, speckling was divided into two measures: the presence/absence of speckling and the amount of speckling. The presence or absence of speckling was a binomial measure of whether a seed had 11
any pigmentation. The amount of speckling was a continuous measure of the degree of speckling
when any pigment was present. The amount of speckling was logit transformed to normalize the
data for analysis. Models were initially constructed using all survey measures (pod density,
alydid abundance, soil value, soil chroma, canopy cover, litter depth, and the presence of other
legumes). Measures of soil value and chroma were used as ordinal variables as the difference
between each level was not necessarily equivalent. Selection regression using the GLM
procedure in SAS was utilized to determine which characteristics of L. perennis populations
were most important in predicting the amount of seed speckling. The best model was selected
based on having the lowest Akaike information criterion (AIC), greatest R2, and the first time that Mallow’s Cp +1 was greater than the number of factors in the model. Variables were then removed from the model if they were not significant (p > 0.05), did not result in an inflated AIC, and did not reduce R2.
For analysis of presence of speckling, a generalized linear model with a binomial distribution and Logit link function was used in JMP. Models were initially constructed using all survey measures (pod density, alydid abundance, soil value, soil chroma, canopy cover, litter depth, and the presence of other legumes). Non-significant (p > 0.05) factors in order of least significance were removed from the model until AIC increased.
To determine if alydid abundance was affected differently by either the degree of
speckling or the presence of speckling, seed color was scored as two variables: either the amount
of speckling when speckling was present or the presence/absence of speckling. To model alydid
abundance, two generalized linear models with Poisson distribution and a Log link function were
constructed in JMP using either the amount of speckling or the presence of speckling with survey 12
variables. Models initially contained all variables. Non-significant factors (p > 0.05) were removed from the models in order of least significance until AIC increased.
An analysis of population means was also performed to understand what site characteristics were associated with the amount of speckling, proportion of the presence of speckling, and alydid abundance. The proportion of samples in each site that had a particular species of legume present was used to estimate the average occurrence of each legume species
(L. capitata, T. virginiana, and B. tinctoria). Pod abundance, alydid abundance, litter depth, and canopy cover that occurred at each site were averaged. The most common soil value and soil chroma was taken to estimate a site’s typical soil color. All dependent variables were normally distributed. Three separate stepwise regressions in JMP were used to determine which variables had a significant relationship with the dependent variables. A mixed direction model with combined effects as a rule was used with a p-value threshold of 0.10 in forwards and backwards directions.
For behavioral analyses, total time and the proportion of total time spent feeding on each seed color was ranked as data were not normally distributed. A one-way ANOVA blocked by individual (to account for differences in individual insect activity during each two-hour trial) was used in JMP to analyze both ranked time and ranked proportion of time spent with seeds by seed color. As first visit was binomial data, a logistic regression was used in SAS using the
GENMOD procedure to evaluate whether the first visit to seeds differed by seed color. A
Poisson regression with individual modeled as a random effect was used in SAS using the
GENMOD procedure to evaluate whether the number of visits differed by seed color as number of visits was scored as count data. Individual was modeled as a subject effect in both the logistic regression and Poisson regression. 13
RESULTS
Population Survey
The amount of seed speckling ranged from 1.9 - 82.8 % of the seed surface showing pigmentation. Lou Campbell Preserve Turnpike (LCPT), Meilke Road Savanna, and Mary’s
Savanna had the highest amounts of speckling (Figure 1), with 54.6, 45.6, and 42.3 % speckling, respectively. South Piel had the least amount of speckling among pigmented seeds, with 17.9 % pigmentation.
Of the 496 plants sampled across all sites, most had some speckling: only 29 % of plants had seeds that lacked dark pigmentation (Table 2). Only two of the ten sites surveyed had more than half of their seeds completely lacking dark pigmentation. LCPT and Meilke Road Savanna had the highest frequency of white seeds with almost 60 % of plants without speckling (Table 2).
Mary’s Savanna and Lou Campbell Preserve Weckerly (LCPW) had the lowest frequency of white seeds, as less than 10 % of plants completely lacked pigmentation (Table 2).
The mean number of alydids per sampling area across sites ranged from 0 – 1.5 individuals per m2 (mean = 0.07, SD = 0.17). Julia’s Savanna and South Piel had the greatest alydid abundance per sampling area, with 0.18 per m2 (SD = 0.31) and 0.11 (SD = 0.20) alydids, respectively. LCPT and Mary’s Savanna had comparatively few, with 0.02 per m2 (SD = 0.04) and 0.02 alydids per m2 (SD = 0.06), respectively.
Sites varied somewhat in productivity, producing an average of 6.40 pods per m2 in the sampling area (SD = 7.15; Table 3). The greatest pod production in sampling areas occurred at
Mary’s Savanna and LCPW, which had an average of 10.29 (SD = 10.96) and 10.24 (SD = 6.97) pods per m2, respectively. Meilke Road Savanna and Reed Road had the lowest pod production, 14
with only 3.00 (SD = 3.42) and 3.18 (SD = 2.40) pods per m2 in sampling areas. Soil hue did not vary and was 10 yellow-red among all samples across all populations. Soil value ranged from 2-
7, with a value of six being most common (occurred 35% of the time). Soil chroma ranged from
1 to 4 (Table 3). The average amount of litter across sites was 1.16 cm (SD = 1.21). Average canopy cover was 26 % across sites (SD = 30.19). L. capitata was the most frequent legume species within L. perennis populations, occurring in almost 37 percent of samples (Table 3).
Desmodium canadense was the least common of the legume species that were tracked across populations, only occurring in 1.3 percent of samples.
Correlations were run between variables (amount of speckling, presence of speckling, soil value, soil chroma, litter depth, canopy cover, alydid abundance, pod abundance, presence of
Baptisia tinctoria, presence of Lespedeza capitata, and presence of Tephrosia virginiana; see
Appendix B). The strongest correlation between variables was that between soil value and soil chroma (r = 0.5244). Correlations between canopy cover and soil value (r = - 0.3680) and the amount of speckling with L. capitata (r = - 0.3046) followed.
Site, alydid abundance, soil value, the presence of L. capitata and T. virginiana were retained as predictors of the amount of speckling in the multiple regression (F15,68 = 7.67, p <
0.0001, R2 = 0.684; Table 4), while pod abundance, litter depth, canopy cover, soil chroma, and the presence of B. tinctoria were unimportant. L. capitata presence was retained in the model as removing this factor, though not significant, reduced R2 (from 0.68 to 0.66) and inflated AICc
(from 165.60 to 166.63). Of the included samples, high degrees of speckling were found at Lou
Campbell Preserve Central (LCPC), Weckerly, and Mary’s Savanna, while Reed Road had the least amount of speckling of any site (Figure 2). Sampling locations that had darker soils tended to have seeds with more speckling than lighter soils (p = 0.0239, Figure 3). Higher amounts of 15
speckling were associated with areas of fewer alydids (p = 0.0295; not shown), and Tephrosia virginiana presence (p = 0.0279; Figure 4), while the amount of speckling had a positive relationship with L. capitata presence (p = 0.0624, Figure 4). The parameter estimate for alydid abundance was low (see Appendix B) and in a univariate analysis only explained 0.484 % of the variation in speckled seeds.
Site (p < 0.0001), pod abundance (p = 0.0003), soil chroma (p = 0.0134), and the presence of L. capitata (p = 0.0162) were significant variables associated with the presence or absence of dark speckling (binomial generalized linear model with Firth Adjusted Maximum
Likelihood estimation; ChiSquare = 51.13, DF = 14, p < 0.0001; Table 5). Soil value, canopy cover, litter depth, alydid abundance, and the presence of B. tinctoria and T. virginiana were not retained in the final model. The presence of speckling was highest at Meilke Road, while other sites (Bond Tract, Lou Campbell Preserve Central, and Mary’s Savanna) had relatively few seeds with speckling present (Figure 5). As soils became more colorful (an increase in soil chroma), the presence of speckling tended to be greater (Figure 6). Speckling was present more frequently when pod abundance was low (Figure 7) and L. capitata was present (Figure 8).
When including the degree of speckling as a factor in a multivariate model (ChiSquare =
134.55, DF = 16, p < 0.0001; Table 5), variation in alydid abundance was associated with site (p
< 0.0001), pod abundance (p < 0.0001), soil value (p = 0.0093), canopy cover (p = 0.0410), and the amount of speckling (p = 0.0499) were related to alydid abundance. Soil chroma, litter depth, and the presence of other legumes did not influence alydid abundance. Alydid abundance was high at Julia’s Savanna, but low at Meilke Road, Reed Road, and Mary’s Savanna (Figure 9).
Soil value 7 had significantly less speckling than a slightly darker soil value, soil value 6 (Figure 16
10). No other soil values differed significantly. Alydid abundance increased with increased pod
abundance, lower canopy cover (Figure 11), and lower amounts of speckling (Figure 12).
However, when the presence or absence of speckling was included as a factor in a multivariate model of alydid abundance (ChiSquare = 144.36, DF = 14, p < 0.0001; Table 5), site (p <
0.0001), pod abundance (p < 0.0001), canopy cover (p = 0.0034), and soil chroma (p < 0.0001) were significant factors. Canopy cover, litter depth, soil value, and the presence of other legumes were not associated with alydid abundance. Sites significantly differed in alydid abundance
(Figure 13). Mary’s Savanna and LCPT had significantly fewer alydids/m2 than all other sites (p
< 0.0001 and p = 0.0281, respectively). Julia’s Savanna had the greatest alydid abundance, while
Meilke Road, Lou Campbell Preserve Turnpike, and Mary’s Savanna had relatively low alydid abundance. Alydid abundance generally increased with increases in soil chroma (more colorful soils), while soil chroma 3 had more alydids than the more colorful soil chroma 4 (p < 0.0001;
Figure 14). Alydid abundance was greater with high pod abundance and low canopy cover
(Figure 15).
Population Means Analysis
In the population means analysis, site area (p = 0.0080) and alydid abundance (p =
0.0345) were retained in a stepwise regression model of the amount of speckling on pigmented
2 seeds (F2,9 = 16.06, p = 0.0024, R = 0.821; Table 6). Pod density, soil value, soil chroma, litter
depth, canopy cover, and the presence of other legume species were not associated with the mean
amount of speckling. As site area increased, the amount of speckling decreased (Figure 16).
Conversely, the amount of speckling decreased as alydid abundance increased (Figure 17). Pod
density was maintained in a stepwise regression for the presence of speckling in an analysis of 17
2 population means (F1,9 = 6.91, p = 0.0303, R = 0.463; Table 6). Alydid abundance, soil value, soil chroma, litter depth, canopy cover, the presence of legume species, and site area were not associated with the mean amount of speckling. As pod abundance increased, the frequency of speckled seeds increased (Figure 18).
Amount of speckling (p = 0.0500) and the presence of L. capitata (p = 0.0072) were the only important variables retained in a stepwise regression of the response of alydid abundance in
2 samples with speckled seed (F2,9 = 16.67, p = 0.0022, R = 0.826; Table 6). Pod density, soil value, soil chroma, litter depth, canopy cover, the presence of T. virginiana, the presence of B. tinctoria, and site area were not associated with the mean amount of speckling per site. As speckling increased, alydid abundance decreased (Figure 19). One other legume species was important as alydid abundance increased in sites that had more L. capitata (Figure 20). The importance of L. capitata (p = 0.0030) persisted in the stepwise regression for alydid abundance
2 for the population averaged samples with speckling (F1,9 = 17.63, p = 0.0030, R = 0.69; Table
6). The presence of speckling, pod density, soil value, soil chroma, litter depth, canopy cover, the presence of T. virginiana and B. tinctoria, and site area were not associated with the mean amount of speckling. Once again, alydid abundance increased in sites that had more L. capitata
(Figure 21).
Behavioral Experiment
The total time interacting with seeds ranged from 0.5 - 119.05 minutes (with an average of 52.94 %) of time during a trial spent with seeds. Overall, insects visited 1 - 9 seeds during a trial with 76.92 % of insects tested visiting more than one seed. On average, individuals visited two seeds during a trial. Of these visits, 17 visits were to white seeds, 28 to light seeds, and 30 to 18
dark seeds. Five insects visited a white seed first, while 11 visited a light seed first and 10 visited
a dark seed first.
The amount of time spent feeding on each seed color was greatest for darkly speckled
seeds (29.1 minutes) compared to lightly speckled (20.8 minutes) and white seeds (13.6
minutes), but did not differ significantly between groups (F2,50 = 1.6334, p = 0.2055; Table 7;
Figure 22a).The proportion of time spent on each seed color did not differ significantly between groups (F2,50 = 2.325 , p = 0.1083; Table 7; Figure 22b). However, insects spent a greater
proportion of time on heavily speckled seeds compared to lightly speckled and white seeds.
Mean proportion of time spent on darkly speckled seeds was 50 % while proportion of time spent
on lightly speckled and white seeds was 29 % and 21 %, respectively. For both total time and
proportion of time, a power analysis revealed that n = 78 would be the required sample size to
achieve p = 0.05.
Visits to dark seeds were more frequent than to light seeds or white seeds (Figure 22c).
The difference in visit frequency between white seeds and light seeds was not significant
(ChiSquare = 0.07, DF = 1, p = 0.7929; Table 8). The difference in visit frequency between
white seeds and dark seeds was also not significant, but had a p-value much closer to 0.05
(ChiSquare = 3.50, DF= 1, p = 0.0613; Table 8).
Light seeds were the first seeds visited in 42.31 % of the trials, while dark seeds and
white seeds were the first visited 38.46 % and 19.23% of the time (Figure 22d). Though the
number of first visits to light seeds were twice as frequent as visits to white seeds, this finding
was not statistically significant (ChiSquare = 3.12, DF = 1, p = 0.0772; Table 8). The difference 19
between first visits to dark seeds as compared to white seeds was not significant (ChiSquare =
2.27, DF = 1, p = 0.1318; Table 8) though more visits were to dark seeds.
DISCUSSION
We did observe trends in seed speckling across populations. Lower amounts of speckling
were associated with increasing alydid abundance. In addition, pod abundance, soil color, and
the presence of L. capitata and T. virginana were associated with the distribution and abundance of seed speckling. These findings suggest that seed color distributions may not be random.
Although we expected that seed color distribution patterns may be associated with soil color traits that may help camouflage seeds (Porter 2013), we found a greater frequency of white seeds on darker soils. This would be consistent with evidence that some seed predators preferentially removed dark seeds in a previous study (Wakeley & Michaels, unpub. data) and have a preference for wooded environments with dark, organic soils (e.g. Peromyscus spp.;
Kappler et al. 2012). In addition, cryptic coloration may be unimportant if seed predators or other selective forces are not influenced by visual cues associated with seeds of different colors
(Nystrand & Granström 1997).
We confirmed that populations of L. perennis do differ in the rates of occurrence for speckled and non-speckled seeds. Interpopulation variation in seed color frequencies, first observed by Cartwright (1997), may be due to chance, genetic drift, or natural selection.
Heritability of this trait in L. perennis is not known, but similar seed color phenotypes in other species of lupin were heritable, with speckled seeds being dominant to white seeds in Lupinus pilosus (Horovitz & Harding 1983). However, the amount of speckling and the quantity of secondary compounds may be more complex traits (e.g. quantitative trait locus; Caldas & Blair 20
2009). Smaller populations had greater percentages of lightly speckled seeds. This observation could be due in part to non-random mating, founder effect, bottlenecks, or genetic drift .
Seedlings from smaller Oak Openings L. perennis populations have previously been shown to have lower fitness and higher selfing rates, suggesting that smaller lupine populations are prone to fitness declines through inbreeding depression, genetic drift, and the fixation of deleterious alleles (Michaels et al. 2008).
Though more alydids were found with lighter speckled seeds, alydid abundance was not associated with the presence or absence of speckling. The relationship between alydids and seed speckling could occur if the low pigmentation of lightly speckled seeds is not detectable or does not function as a strong herbivore deterrent. Darkly speckled seeds, alternatively, may have pigmentation that is easier to detect or is a stronger deterrent of alydid feeding. In analyses that consider both the amount of speckling and the presence of speckling, pod density was consistently associated with alydid abundance. This finding is expected, as concentrated resources lead to an increase in herbivores (Root 1973; Hambäck & Englund 2005).
Furthermore, the relationship between alydid abundance and canopy cover could also be related to food availabilty as greater sun exposure (lower canopy cover) could increase pod production.
These insects are not necessarily specialists as there is evidence these alydids feed on a variety of legumes, as well as some non-legume hosts (Ceanothus americanus and sumac; Yonke &
Medler 1964). Generalist species were found to increase in abundance when alternative host species were available (Östergård & Ehrlén 2005). Therefore, it is not unexpected that we found the presence of some host species, specifically L. capitata, associated with an increase in alydid abundance. As this legume species sets seed later than any other legume species encountered during our survey, it is possible that L. capitata is important to alydids completing multiple 21
generations per year and may increase egg production and overwintering success the next spring.
Because L. capitata was important to alydid abundance, further exploration into the distribution and abundance of agricultural legumes near L. perennis populations may also give insight into the alydid abundances we observed in this study.
Other environmental variables that were not considered could be responsible for some of the trends found in this study. For example, the relationship between the presence of L. capitata and T. virginiana and the amount of speckling was an unexpected finding that cannot easily be attributed to seed predation by alydids. Possible factors that may be correlated with the presence of these other legumes could include nutrient availability and rhizobia availability. For instance, rhizobia grow faster and are present in greater numbers when legumes are present (Bowen &
Rovira 1976; Dowling & Broughton 1986), which could allow for greater availability of rhizobia over time.
Our study suggests that alydids do not have a strong preference for seed color in L. perennis. In fact, we found that M. quinquespinosus will feed on seeds regardless of dark coloration. Contrary to our expectations, measures of behavior (frequency of visits, time spent feeding, and proportion time spent feeding) were not significantly different among seed colors, though a trend towards spending more time with darker seeds was observed.
Feeding time did not differ significantly between seed colors, though a trend towards spending more time on darker seeds was observed. Longer time spent on darker seeds could be due to seed pigments resulting in stronger seeds. As such, insects may need to spend more time penetrating seed coats. Some studies have found a trend towards legume species having physically stronger seeds when tannins or alkaloids are present in the seed coat (Marbach & 22
Mayer 1974; Werker et al. 1979; Legesse & Powell 1996; Oigianbe & Onigbinde 1996), though
other species of legumes do not show this relationship (Slattery et al. 1982; Argel & Humphreys
1983). Whether the pigments in L. perennis seed coats display a trend for harder seeds is not known. Conversely, the greater time spent interacting with darker seeds could be due to a preference for darker seeds. This idea is supported by the observation that darker seeds were also
the most common seed to be visited first and fed on more frequently than other seed types,
though this finding was not statistically significant. Furthermore, we did find a relationship
between alydid abundance and the amount of seed speckling during our population survey,
suggesting that alydids may prefer a particular seed color. Therefore, time spent on seeds may
not be wholly due to differences in seed coat hardness.
Another alternative explanation for more insect visits to dark seeds is that the background
color (white) used in trials allowed greater contrast for dark seeds. As a result, the first visit and
frequency of visits results could be attributable to dark seeds being easier to find. Pekár and
Hrušková (2006) found that all sensory cues (vision, chemoreception, and mechanoreception)
needed to be functioning for a phytophagous Hemipteran insect to identify its preferred food
sources. It is expected that visual cues are important at far distances to identify host plants or
seed pods, but chemoreception and mechanoreception are more important to identifying a
palatable food source at close proximity (Hulme & Benkman 2002). When all sensory organs are
functioning, it is probable that alydids and other Hemipteran phytophages rely on
chemoreception rather than sight at close distances.
Assuming seed coat pigments have evolved to deter seed predation, seed color may not
be important to alydid predation like it might be for other seed predators, as alydids may be able
to bypass the feeding deterrents with their sucking mouthparts. Therefore, these seed predators 23
may be able to exploit dark seeds differently than other seed predators. Though we do not know how this pigment is distributed through the seed, it is possible that the pigmentation in some L. perennis seeds that we expected to act as an herbivore deterrent may be found solely in the seed coat. This may explain why alydids seem to be unaffected by this deterrent in the seed coat, as they are merely penetrating this layer of the seed, not consuming it. Tannins and other polyphenolics are some secondary metabolites that may be responsible for the color observed in
L. perennis seeds as these compounds are commonly responsible for dark coloration in seed coats of other legume species (Moïse et al. 2005; Xu et al. 2007; Yang et al. 2010). In addition, tannins and other polyphenolics tend to be concentrated in seed coats as they are less water soluble and this phenomenon is especially prevalent in legumes (McKey 1979; Lundgren 2009).
Furthermore, there is no relationship between seed color and alkaloid content in other lupin species (Lattanzio et al. 2005; Lundgren 2009) and alkaloids tend to be distributed throughout seeds with concentrations increasing near the embryo (McKey 1979). Alternatively, because this trend of a preference for darker seeds was also observed in other seed predators of L. perennis, the pigment found in these seeds might not be harmful to seed predators. Thus, these seed pigments may not be a deterrent for seed predators, but may be associated with more desirable seed characteristics, or may be inconsequential to L. perennis fitness.
At this point, we have been unsuccessful in characterizing predation rates as visual scoring was expected to be biased on darker seeds and preliminary results on white seeds were inconsistent when compared to visual scoring. To detect if punctures had occurred, seeds were soaked for 48 hours in deionized water, which causes damaged seeds to imbibe. However, results of imbibed seeds were inconsistent with visually scored seeds. Future attempts at scoring 24
predation damage will involve staining seeds with fluorescent or histochemical dyes to characterize damage and germinating seeds to look for damage to cotyledons.
The findings of this study do not unequivocally support alydids having a preference for L. perennis seed color. It is possible that the patterns of seed color frequency observed across populations of L. perennis are random, undergoing genetic drift, or responding to other selective forces. If populations are diverging in response to selection, some additional selective pressures that may play a role in explaining the observed patterns of this seed color polymorphism include maintaining differential seed hardness to promote non-synchronous germination (Woodstock
1988; Argel & Patton 1999), attraction of rhizobia during germination (Hungria & Phillips
1993), and seed predation by other seed predators. These pressures may be occurring in conjunction with alydid seed predation if the predation rate by alydids is high. Further study into alydid seed predation, as well as these alternative hypotheses is necessary to understand if L. perennis seed color is being affected by natural selection and whether this polymorphism is important to the fitness of this species. The resulting findings may facilitate conservation decisions as maintaining multiple phenotypes could be important when varying selective pressures are present.
25
REFERENCES
Abell K, Tracey C, & Michaels HJ. Unpublished data. The effect of seed herbivory by
Megalotomus quinquespinosus on Lupinus perennis. Department of Biological Sciences.
Bowling Green State University.
Argel PJ & Humphreys LR. 1983. Environmental effects on seed development and
hardseededness in Stylosanthes hamate cv. Verano. I. Temperature. Australian Journal of
Agricultural Research. 34(3): 261-270.
Armbruster WS. 2002. Can indirect selection and genetic context contribute to trait
diversification? A transition-probability study of blossom-colour evolution in two genera.
Journal of Evolutionary Biology. 15: 468-485.
Argel PJ & Paton CJ. 1999. Overcoming legume hardseededness. Forage seed production. 2:
247-259.
Bhardwaj HL, Starner DE, & van Santen E. 2010. Preliminary evaluation of white lupin
(Lupinus albus L.) as a forage crop in the mid-atlantic region of the United States of
America. Journal of Agricultural Science. 2(4): 13.
Bowen GD & Rovira AD. 1976. Microbial colonization of plant roots. Annual Review of
Phytopathology. 14: 121-144.
Branco M, Branco C, Merouani H, & Almeida MH. 2002. Germination success, survival and
seedling vigour of Quercus suber acorns in relation to insect damage. Forest Ecology and
Management. 166: 159-164. 26
Caldas GV & Blair MW. Inheritance of seed condensed tannins and their relationship with seed-
coat color and pattern genes in common bean (Phaseolus vulgaris L.). Theoretical
Applied Genetics. 119: 131-142.
Carlson JE & Holsinger KE. 2013. Direct and indirect selection on floral pigmentation by
pollinators and seed predators in a color polymorphic South African shrub. Oecologia.
171: 905-919.
Cartwright CA. 1997. Interpopulation variation in Lupinus perennis, the wild lupine. M.S.
Thesis. Bowling Green State University. 87 p.
Clough MW. 1992. Endangered and threatened wildlife and plants: determination of endangered
status for the Karner blue butterfly. Federal Register. 57(240): 59236-59244.
Coberly LC & Rausher MD. 2008. Pleiotropic effects of an allele producing white flowers in
Ipomoea pupurea. Evolution. 62(5): 1076-1085.
Dixon RA & Sumner LW. 2003. Legume natural products: understanding and manipulation
complex pathways for human and animal health. Plant Physiology. 131: 878-885.
Dowling DN & Broughton WJ. 1986. Competition for nodulation of legumes. Annual Review of
Microbiology. 40: 131-157.
Duranti M & Guis C. 1997. Legume seeds: protein content and nutritional value. Field Crops
Research. 53: 31-45.
Grigore MT & Tramer EJ. 1996. The short-term effect of fire on Lupinus perennis (L.). Natural
Areas Journal. 16(1): 41-48. 27
Grigore MT & Windus J. 1994. Decline of the Karner Blue butterfly in the Oak Openings of
Northwest Ohio. In: Andow DA, Baker, RJ, Lane CP (eds) Karner Blue butterfly.
University of Minnesota. St. Paul, MN. pp. 135–142.
Gustafsson Å & Gadd I. 1964. Mutations and crop improvements: II. The genus Lupinus
(Leguminosae). Hereditas. 53 (1-2): 15-39.
Halpern SL. 2005. Sources and consequences of seed size variation in Lupinus perennis
(Fabaceae): Adaptive and non-adaptive hypothesis. American Journal of Botany. 92(2):
205-213.
Hambäck P & Englund G. 2005. Patch area, population density and the scaling of migration
rates: the resource concentration hypothesis revisited. Ecology Letters. 8: 1057-1065.
Horovitz A & Harding J. 1983. Genetics of Lupinus. XII. The mating system of Lupinus pilosus.
Botanical Gazette. 144(2): 276-279.
Hulme PE & Benkman CW. 2002. Granivory. Plant-animal interactions: an evolutionary
approach. 132-154.
Hungria M & Phillips DA. 1993. Effects of a seed color mutation on rhizobial nod-gene-
inducing flavonoids and nodulation in common bean. Molecular Plant-Microbe
Interactions. 6(4): 418-422.
Image J. US National Institutes of Health, Bethesda, Maryland, USA. 1997-2013.
Janzen DH. 1971. Escape of Cassia grandis L. beans from predators in time and space. Ecology.
52(6): 964-979. 28
Janzen DH. 1976. Reduction of Mucuna Andreana (Leguminosae) seedling fitness by artificial
seed damage. Ecology. 57(4): 826-828.
JMP, Version 10. SAS Institute Inc., Cary, NC, 1989-2013.
Kappler RH, Michaels HJ, and Root KV. 2012. Impacts of seed predation by mice on wild lupin
in and near oak savannas. American Midland Naturalist. 168(1): 18-29.
Karban R & Lowenberg G. 1992. Feeding by seed bugs and weevils enhances germination of
wild Gossypium species. Oecologia. 92: 196-200.
Kelly SD. 1998. Germination preferences and effects of micro-site variation on Lupinus perennis
establishment in an oak savanna community. Masters Thesis, Bowling Green State
University, Bowling Green.
Kolb A, Ehrlén J, & Eriksson O. 2007. Ecological and evolutionary consequences of spatial and
temporal variation in pre-dispersal seed predation. Perspectives in Plant Ecology,
Evolution and Systematics. 9: 79-100.
Kristenová M, Exnerová A, & Štys P. 2011. Seed preferences of Pyrrhocoris apterus
(Heteroptera: Pyrrhocoridae): Are there specialized trophic populations? European
Journal of Entomology. 108: 581-586.
Kuprewicz EK & García-Robledo C. 2010. Mammal and insect predation of chemically and
structurally defended Mucuna holtonii (Fabaceae) seeds in Costa Rican rain forest.
Journal of Tropical Ecology. 25(3): 263-269. 29
Lattanzio V, Terzano R, Cicco N, Cardinali A, Venere DD, & Linsalata V. 2005. Seed coat
tannins and bruchid resistance in stored cowpea seeds. Journal of the Science of Food and
Agriculture. 85: 839-846.
Legesse N & Powell AA. 1996. Relationship between the development of seed coat
pigmentation, seed coat adherence to the cotyledons and the rate of imbibition during the
maturation of grain legumes. Seed Science and Technology. 24(1): 23-32.
Lundbye H & Johansson DK. 2012. The effect of a seed-sucking bug on seed germination of an
Arctic cushion plant. Écoscience. 19(3): 209-212.
Lundgren JG. 2009. Seed nutrition and defense. Relationships of Natural Enemies and Non-Prey
Foods. 183-209.
Mackay WA, Davis TD, Sankhla D, & Riemenschneider DE. 1996. Factors influencing seed
germination of Lupinus perennis. Journal of Environmental Horticulture. 14(4): 167-169.
Marbach I & Mayer AM. 1974. Permeability of seed coats to water as related to dry conditions
and metabolism of phenolics. Plant Physiology. 54: 817-820.
Maron JL & Simms EL. 1997. Effect of seed predation on seed bank size and seedling
recruitment of bush lupine (Lupinus arboreus). Oecologia. 111: 76-83.
Moïse JA, Han S, Gudynaitę-Savitch L, Johnson DA, & Miki BL. 2005. Seed coats: structure,
development, composition, and biotechnology. In Vitro Cellular & Developmental
Biology. 41(5): 620-644.
Munsell Color. Munsell Soil Color Charts. 2000. Munsell Color. New Windsor, NY. 30
Nuzzo VA. 1986. Extent and status of midwest oak savanna: presettlement and 1985. Natural
Areas Journal. 6:6-36.
Nystrand O & Granström A. 1997. Post-dispersal predation on Pinus sylvestris seeds by
Fringilla spp: ground substrate affects selection for seed color. Oecologia. 110: 353-359
Ohio Department of Natrual Resources. 1998. Lupinus perennis L.: Wild Lupine.
Oigiangbe NO & Onigbinde AO. 1996. The association between some physico-chemical
characteristics and susceptibility to cowpea (Vigna unguiculata (L.) Walp) to
Callosobruchus maculatus (F.). Journal Stored Products Research. 32(1): 7-11.
Östergård H & Ehrlén J. 2005. Among population variation in specialist and generalist seed
predation – the importance of host plant distribution, alternative hosts and environmental
variation. OIKOS. 111: 39-46.
Panizzi AR. 1987. Nutritional ecology of seed-sucking insects of soybean and their management.
Memórias do Instituto Oswaldo Cruz. 82: 161-175.
Pavlovic NB & Grundel R. 2009. Reintroduction of wild lupine (Lupinus perennis L.) depends
on variation in canopy, vegetation, and litter cover. Restoration Ecology. 17(6): 807-817.
Pekár S & Hrušková M. 2006. How granivorous Coreus marginatus (Heteroptera: Coreidae)
recognizes its food. Acta Ethologica. 9: 26-30.
Perea R, San Miguel A, & Gil L. 2011. Leftovers in seed dispersal: ecological implications of
partial seed consumption for oak regeneration. Journal of Ecology. 99: 194-201. 31
Pizo MA, Von Allmen C, & Morellato LPC. 2006. Seed size variation in the palm Euterpe edulis
and the effects of seed predators on germination and seedling survival. Acta Oecologia.
29: 311-315.
Porter SS. 2013. Adaptive divergence in seed color camouflage in contrasting soil environments.
New Phytologist. 197: 1311-1320.
Putnam DH, Oplinger ES, Hardman LL, & Doll JD. 1989. Lupine. Alternative Field Crops
Manual. Purdue University.
Root RB. 1973. Organization of a plant-arthropod association in simple and diverse habitats: the
fauna of collards (Brassica Oleracea). Ecological monographs. 43(1): 95-124.
Schaefer CW. 1980. The host plants of the Alydinae, with a note on heterotypic feeding
aggregations (Hemiptera: Coreoidea: Alydidae). Journal of the Kansas Entomological
Society. 53(1): 115-122.
Slattery HD, Atwell BJ, & Kuo J. 1982. Relationship between colour, phenolic content, and
impermeability in the seed coat of various Trifolium subterraneum L. genotypes. Annals
of Botany. 50: 373-378.
Silman MR, Terborgh JW, & Kiltie RA. 2003. Population regulation of a dominant rain forest
tree by a major seed predator. Ecology. 84(2): 431-438.
Silva JJD, Arruda-Gatti ICD, Mikami AY, Pissinati A, Panizzi AR, & Ventura MU. 2010.
Attraction of Neomegalotomus parvus (Westwood) (Heteroptera: alydidae) to cow urine
and ammonia. Scientia Agricola. 67(1): 84-86. 32
Souza FHDD & Marcos-Filho J. 2001. The seed coat as a modulator of seed-environment
relationships in Fabaceae. Revista Brasileira de Botânica. 24(4): 265-375.
Strauss SY & Irwin RE. 2004. Ecological and evolutionary consequences of multispecies plant-
animal interactions. Annual Review of Ecology, Evolution, and Systematics. 35: 435-
466.
Strauss SY, Rudgers JA, Lau JA, & Irwin RE. 2002. Direct and ecological costs of resistance to
herbivory. Trends in Ecology & Evolution. 17(6): 278-285.
Takakura K. 2002. The specialist seed predator Bruchidius dorsalis (Coleoptera: Bruchidae)
plays a crucial role in the seed germination of its host plant, Gleditsia japonica
(Leguminosae). Functional Ecology. 16: 252-257.
Trimble Geo XH. Trimble Navigation. Sunnyvale, CA, USA. 2006-2008.
Underhill GW. 1943. Two pests of legumes: Alydus eurinus Say, and A. pilosulus Herrick-
Schaeffer. Journal of Economic Entomology. 36(2): 289-293.
Vallejo-Marín M, Domínguez CA, & Dirzo R. 2006. Simulated seed predation reveals a variety
of germination responses of neotropical rain forest species. American Journal of Botany.
93(3): 369-376.
Wakeley EF, Michaels HJ, and Gorsevski PV. Unpublished data. Lupine seed predation: the
effects of seed color and substrate. Department of Biological Sciences and School of
Earth, Environment, and Society. Bowling Green State University.
Wang TL, Domoney C, Hedley CL, Casey R, & Grusak MA. 2003. Can we improve the
nutritional quality of legume seeds? Plant Physiology. 131: 886-891. 33
Werker E, Marbach I, & Mayer AM. 1979. Relation between the anatomy of the testa, water
permeability and the presence of phenolics in the genus Pisum. Annals of Botany. 43(6):
765-771.
Westoby M, Jurado E, & Leishman M. 1992. Comparative evolutionary ecology of seed size.
Trends in Ecology & Evolution. 7(11): 368-372.
Wilkinson JD & Daugherty DM. 1967. Biology of the Broadheaded Bug Alydus pilosulus
(Hemiptera: Alydidae). Annals of the Entomological Society of America. 60(5): 1018-
1021.
Woodstock LW. 1988. Seed imbibition: a critical period for successful germination. Journal of
Seed Technology. 12: 1-15.
Xu BJ, Yuan SH, & Chang SKC. 2007. Comparative analyses of phenolic composition,
antioxidant capacity, and color of cool season legumes and other selected food legumes.
Journal of Food Science. 72(2): 167-177.
Yang K, Jeong N, Moon J-K, Lee Y-H, Lee S-H, Kim HM, Hwang CH, Back K, Palmer RG, &
Jeong S-C. 2010. Genetic analysis of genes controlling natural variation of seed coat and
flower colors in soybean.
Yonke TR & Medler JT. 1964. Biology of Megalotomus quinquespinosus (Hemiptera:
Alydidae). Annals of the Entomological Society of America. 58(2): 222-224.
Yonke TR & Medler JT. 1968. Biologies of three species of Alydus in Wisconsin. Annals of the
Entomological Society of America. 61(2): 526-531. 34
Zhang J & Maun MA. 1991. Effects of partial removal of seed reserves on some aspects of
seedling ecology of seven dune species. Canadian Journal of Botany. 69(7): 1457-1462.
35
TABLES AND FIGURES
Table 1. GPS coordinates and managing agencies of populations surveyed. Waypoints were taken at the center of each population using a Trimble GeoXH.
Management Site Latitude Longitude Agency
Meilke Road 41°38’28.40012”N 83°45’59.64683”W Division of Wildlife
The Nature Bond Tract 41°37’15.55728”N 83°47’37.89633”W Conservancy The Nature Oak Dune 41°37’10.21334”N 83°47’32.65176”W Conservancy The Nature Julia’s Savanna 41°37’14.82735”N 83°47’15.67219”W Conservancy The Nature South Piel 41°37’06.74435”N 83°47’06.96547”W Conservancy Ohio Department of LCPT 41°35’37.91079”N 83°46’35.76500”W Natural Resources Ohio Department of LCPC 41°35’30.22548”N 83°46’39.68826”W Natural Resources Ohio Department of LCPW 41°35’23.53135”N 83°46’45.59970”W Natural Resources The Metroparks of Reed Road 41°32’59.23970”N 83°50’10.86403”W the Toledo Area The Metroparks of Mary’s Savanna 41°32’14.40066”N 83°50’42.96746”W the Toledo Area
36
Table 2. Percent of samples that were white or speckled seeds by site. Only Meilke Road and Lou Campbell Preserve Turpike had more than 50 % of their seeds without speckling (white). N represents the number of plants sampled from each site.
Site Frequency White Frequency Speckled N
Meilke Road 58 % 42 % 36
29 % 71% 56 Bond Tract
Oak Dune 24 % 76 % 51
Julia’s Savanna 22 % 78 % 54
South Piel 33 % 67 % 46
Turnpike 59 % 41 % 56
Central 11% 89 % 65
Weckerly 9 % 91 % 22
Reed Road 49 % 51 % 42
Mary’s Savanna 7 % 93 % 67
37
Table 3. Population means for environmental variables. Soil value ranged from 2-7 across sites and is a measurement of soil darkness (a smaller value is darker than a larger value). Soil chroma ranged from 1-4 across sites and is a measurement of soil color (a smaller value is less colorful while a darker value is more colorful. Litter depth was measured to the nearest 0.5 cm. Canopy cover is the percent area covered by canopy. Population area is the km2 a population of L. perennis spanned. L. capitata, T. virginiana, and B. tinctoria were reported as present or absent. The values given for each of these legume species is the proportion of samples at each site that had each legume species present. Ranges are included in parentheses.
Site Soil Soil Litter Canopy Pop. Area L. T. B. Value Chroma Depth Cover capitata virginiana tinctoria 59.54 Meilke 1.59 (0- 4 (2-6) 2 (1-3) (10.29- 0.37 0.06 0 0.03 Road 6) 91.85)
Bond 5,6 (3- 26.36 (0- 2 (2-3) 0.5 (0-3) 0.15 0.43 0.16 0.30 Tract 7) 87.5)
Oak 2.14 (0- 50.37 (0- 4 (3-6) 2 (2-3) 0.32 0.16 0.3 0.23 Dune 4) 80.04)
Julia’s 0.97 (0- 7.98 (0- 6 (5-7) 3 (3-4) 0.26 0.96 0.06 0.00 Savanna 3) 67.8)
South 0.53 (0- 4.41 (0- 6 (5-7) 3 (3-4) 0.06 0.33 0 0 Piel 2) 87.07)
2.07 31.99 (0- LCPT 3 (3-5) 2 (2-3) 0.49 0.37 0 0 (0.5-4) 78.55) 0.97 (0- 11.58 (0- LCPC 4 (3-6) 2 (2-3) 0.64 0.16 0.35 0.03 3.5) 60.49) 42.11 3,4 (3- 0.61 (0- LCPW 2 (1-2) (0.15- 0.51 0.59 0 0 6) 2) 86.72)
Reed 1.68 (0- 20.27 (0- 6 (5-6) 3 (2-3) 0.05 0.43 0 0 Road 4) 88.57)
Mary’s 0.61 (0- 14.2 (0- 5 (2-7) 1 (1-3) 0.46 0.06 0.5 0.13 Savanna 2.5) 56.68)
Average 6 (2-7) 2 (1-4) 1.16 25.9 - 0.37 0.14 0.07 38
Table 4. The relationship of site, alydid abundance, soil value, and the presence of two other legumes (Tephrosia virginiana and Lespedeza capitata) on the amount of seed speckling (R2 = 0.68).
Source DF Sum of Mean F Prob>F Direction
Squares Square Ratio Site 8 25.38 7.36 < 0.0001* Alydid 1 2.16 5.01 0.0295* - Abundance Soil Value 4 5.30 3.07 0.0239* Tephrosia 1 2.21 5.11 0.0279* + Amount of virginiana Speckling (%) Lespedeza 1 1.56 5.11 0.0624 - capitata Model 15 49.6 3.31 7.67 < 0.0001* Error 53 22.9 0.43 Total 68 72.5
39
Table 5. The relationship of environmental factors and seed characteristics on the presence of seed speckling, alydid abundance when considering the amount of speckling, and alydid abundance when considering the presence of seed speckling. Direction indicates whether a continuous variable was negatively or positively related to the response variable. Whole model test for presence of speckling χ2 = 51.13, DF = 14, p < 0.0001; alydid abundance (amount of speckling) χ2 = 134.55, DF = 16, p < 0.0001; alydid abundance (presence of speckling) χ2 = 144.36, DF = 14, p < 0.0001.
Response Variable DF L-R χ2 Prob> χ2 Direction
Presence of Speckling Site 9 36.64 < 0.0001* Pod Abundance 1 13.32 0.0003* - Soil Chroma 3 10.70 0.0134* Lespedeza 1 5.78 0.0162* + capitata Alydid Abundance Site 9 73.86 < 0.0001* (Amount of Speckling) Pod Abundance 1 24.45 < 0.0001* + Soil Value 4 13.45 0.0093* Canopy Cover 1 4.18 0.0410* - Amount Speckling 1 3.84 0.0499* - Alydid Abundance Site 9 93.59 < 0.0001* (Presence of Speckling) Pod Abundance 1 25.84 < 0.0001* + Soil Chroma 3 22.43 <0.0001* Canopy Cover 1 8.55 0.0034* -
40
Table 6. Models of speckling and alydid abundance. The relationship of site area and alydid abundance on the amount of seed speckling (R2 = 0.82); pod abundance on the proportion of speckled seeds present at a site (R2 = 0.46); the amount of speckling as well as the presence of L. capitata on alydid abundance given the amount of speckling (R2 = 0.83); and the presence of L. capitata on alydid abundance given the presence/absence of speckling (R2 = 0.69) in an analysis of population means.
Source DF Sum of Mean F Prob>F Direction
Squares Square Ratio Site Area 1 432.17 13.43 0.0080* + Alydid Abundance 1 216.13 6.86 0.0345* - Amount of Model 2 1012.48 506.24 16.06 0.0024* Speckling (%) Error 7 220.59 31.51
Pod Abundance 1 1587.67 6.91 0.0303* + Proportion of Model 1 1587.67 1587.6 6.91 0.0303* Speckled Error 8 1839.23 229.9 Seeds Present Total 9 3426.90 Amount of 1 0.0031 5.59 0.0500* - Alydid Speckling Abundance L. capitata 1 0.0078 14.06 0.0072* + (Amount of Model 2 0.0185 0.0092 Speckling) Error 7 0.0039 0.0055 Total 9 0.0223 L. capitata 1 0.0154 17.63 0.0030* + Alydid Model 1 0.0154 0.0154 17.63 0.0030* Abundance Error 8 0.0070 0.0009 (Presence of Total 9 0.0223 Speckling)
41
Table 7. The time spent with seeds (R2 = 0.168) and proportion time spent with seeds (R2 = 0.107) by color.
Source DF Sum of Mean F Prob>F
Squares Square Ratio Color 2 1985.9 993.0 1.63 0.2055 Individual 25 4160.2 166.4 0.27 0.9996 Time Spent with Error 50 30395.4 607.9 Seeds Total 77 36541.5
Color 2 2912.8 1456.4 2.33 0.1083 Proportion Time Individual 25 848.2 33.93 0.05 1.0000 Spent with Error 50 31320.5 626.41 Seeds Total 77 35081.5
42
Table 8. Relationship of first visit and frequency of visits to seed color. First visit (r = -0.49) and frequency of visits (r = -0.02) by color with individual as a repeated measure.
Variable DF SE 95 % CI χ2 Prob> χ2
Dark 1 0.6404 (-2.22 – 0.29) 2.27 0.1319 Light 1 0.6365 (-2.37 – 0.12) 3.12 0.0772 First Visit White 0 - - - -
Dark 0 - - - - Frequency Light 1 0.2628 (-0.58 – 0.45) 0.07 0.7929 of Visits White 1 0.3036 (-1.16 – 0.03) 3.50 0.0613
43
Amount of Speckling (%) Figure 1. Distribution of amount of speckling by site. Distribution does not include seeds with no speckling. Reed road had the lowest average amount of speckling across sites. Meilke Road, Lou Campbell Preserve Turnpike, and Mary’s Savanna had the highest average amount of seed speckling. 44
60
A 50 A AB A 40 AB
30 BC
20
CD CD D 10 Amount of Speckling (%)
0
Site Figure 2. Relationship of site and amount of speckling. Sites differed in the amount of speckling with Mary’s Savanna having the greatest amount of speckling and Reed Road having the least amount of speckling (F8,68 = 7.36, p < 0.0001). Sites not connected by the same letter were significantly different from one another in a standard least squares multiple regression analysis with Tukey-Kramer comparison of means. Meilke Road was not included in this analysis as it did not have enough replicates with all variables retained in the model. Least square means are shown. Error bars represent standard error.
45
A 60
50
40 AB AB 30 AB
B 20 Amount of Speckling (%)
10
0 3 4 5 6 7 Soil Value
Figure 3. Relationship of soil value and amount of speckling. Soil value was significantly related to the amount of speckling of a seed (F4,68 = 3.07, p = 0.0239). Seeds from soil value 7 (lighter soil) showed significantly more speckling than soil value 4 (darker soil). Soil values 3, 5, and 6 did not differ significantly in the amount of speckling compared to any other soil values. Soil values not connected by the same letter were significantly different from one another in a standard least squares multiple regression analysis with a Tukey-Kramer comparison of means. Least square means are shown.
46
60
50
40 Tephrosia virginiana 30
20 Lespedeza
Amount of Speckling (%) capitata 10
0 0 1 Presence of Legumes
Figure 4. Relationship of the presence of two other legumes and the amount of speckling, Tephrosia virginiana and Lespedeza capitata, on the amount of speckling. When Tephrosia virginiana was present, the amount of speckling was found to be significantly greater than when this species was absent (F1,68 = 5.11 ,p = 0.0279). There was a trend towards a lower degree of speckling when Lespedeza capitata was absent, though this was not a significant effect (F1,68 = 3.62, p = 0.0624). Least square means are shown.
47
1.2
1
0.8
0.6
0.4
0.2
0 Presence/Absence ofPresence/Absence Speckling
Site Figure 5. Relationship of site and the presence of speckling. Site was significantly associated with the presence of speckling in a binomial generalized linear model (ChiSquare = 36.64, DF = 9, p < 0.0001).Meilke Road had a high presence of speckling in this model while Bond Tract, Lou Campbell Preserve Central, and Mary’s Savanna had relatively low presence of speckling. Predicted means are shown. Error bars represent standard error.
48
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Presence/Absence ofPresence/Absence Speckling 0 1 2 3 4 Soil Chroma
Figure 6. Relationship of soil chroma and presence or absence of speckling. Soil chroma was found to be associated with the presence or absence of speckling in a binomial generalized linear model (ChiSquare = 10.70, DF = 3, p = 0.0134). More colorful soils tended to have greater presence of speckling than less colorful soils. Predicted means are shown. Error bars represent standard error.
49
Speckling Present Proportion of Seeds with of Seeds Proportion
Pod Density (pods/m2)
Figure 7. Relationship of pod density and the proportion of seeds with speckling present. In a binomial generalized linear model, it was found that total pods had a negative relationship with the presence of speckling (ChiSquare = 13.32, DF = 1, p = 0.0003).
50
0.35
0.3
0.25
0.2
0.15
0.1 Speckling Present
Proportion of Seeds with of Seeds Proportion 0.05
0 0 1 Presence of Lespedeza capitata
Figure 8. Relationship of Lespedeza capitata presence and the proportion of seeds with speckling present. Seeds with speckling present were more common when Lespedeza capitata was present (ChiSquare = 5.78, DF = 1, p = 0.0162). Predicted means are shown. Error bars represent standard error.
51
) 2 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 Meilke Bond Oak Julia's South LCPT LCPC LCPW Reed Mary's
Alydid Abundance (individuals/m Abundance Alydid Road Tract Dune Savanna Piel Road Savanna
Site
Figure 9. Relationship of site and alydid abundance when considering the degree of speckling. Alydid abundance was high at Julia’s Savanna compared to other sites when an analysis was performed with the amount of speckling included in the model (ChiSquare = 73.86, DF = 9, p < 0.0001). Meilke Road, Reed Road, and Mary’s Savanna had relatively low alydid abundance. Predicted means are shown. Error bars represent standard error.
52
)
2 0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0 Alydid Abundance (individuals/m Abundance Alydid 3 4 5 6 7 Soil Value
Figure 10. Relationship of soil value and alydid abundance when considering the degree of speckling. Overall, an association between soil value (soil darkness) and the predicted alydid abundance was observed when considering the degree of speckling (ChiSquare = 13.45, DF = 4, p = 0.0093). Alydid abundance tended to be higher in darker soils compared to lighter soils. Predicted means are shown. Error bars represent standard error.
53
) 2 ls/m Alydid Abundance (individua Abundance Alydid
Canopy Cover (%)
Figure 11. Relationship of canopy cover and alydid abundance when considering the degree of speckling. When considering the amount of speckling, predicted alydid abundance decreased as canopy cover increased (ChiSquare = 4.18, DF = 1, p = 0.0410). 54
) 2 Alydid Abundance (individuals/m Abundance Alydid
Average Amount of Speckling (%)
Figure 12. Relationship of the average amount of speckling and alydid abundance when considering the degree of speckling. As the amount of speckling increased, predicted alydid abundance increased (ChiSquare = 3.84, DF = 1, p = 0.0499).
55
) 0.5 2 0.45 0.4 0.35 0.3 0.25 0.2 (individuals/m Alydid Abundance Abundance Alydid 0.15 0.1 0.05 0
Site Figure 13. Relationship of site and alydid abundance when considering the presence of speckling. Sites differed in alydid abundance when an analysis was performed with presence of speckling as a possible factor (ChisSquare = 93.59, DF = 9, p < 0.0001). Mary’s Savanna, Reed Road, Meilke Road, and Lou Campbell Preserve Turnpike had relatively low alydid abundance while Julia’s Savanna had high alydid abundance. Predicted means are shown. Error bars represent standard error. 56
) 0.1 2 0.09 0.08 0.07 0.06 0.05
(individuals/m 0.04 Alydid Abundance Abundance Alydid 0.03 0.02 0.01 0 1 2 3 4 Soil Chroma
Figure 14. Relationship of soil chroma and alydid abundance when considering the presence of speckling. Soil chroma was significantly associated with alydid abundance in an analysis that included presence of speckling as a possible factor (ChiSquare = 22.43, DF = 3, p < 0.0001). More colorful soils tended to be associated with greater alydid abundance. Predicted means are shown.
57
) 2 Alydid Abundance (individuals/m Abundance Alydid
Canopy Cover (%) ‘
Figure 15. Relationship of canopy cover and alydid abundance when considering the presence of speckling. As canopy cover increased, predicted alydid abundance decreased in an analysis that considered presence of speckling as a possible factor (ChiSquare = 8.55, DF = 1, p = 0.0034).
58
Average Amount of Speckling (%)
Site Area (km2)
Figure 16. Relationship of site area and amount of speckling for population means. Site area was associated with an increase in the predicted amount of speckling (F1,9 = 13.43, p = 0.0080).
59
Average Amount of Speckling (%)
Figure 17. Relationship of alydid abundance and amount of speckling for population means. An increase in alydids was associated with an increase in the predicted amount of speckling (F1,9 = 6.86, p = 0.0345).
60
Figure 18. Relationship of pod abundance and proportion of samples at a site with speckling present. An increase in the proportion of samples with speckling was associated with an increase in pod abundance (F1,9 = 6.91, p = 0.0303). 61
) 2 Alydid Abundance (individuals/m Abundance Alydid
Figure 19. Relationship of the amount of speckling and alydid abundance when considering the amount of speckling. An increase in the amount of speckling given speckling was associated with a
decrease in alydid abundance (F2,9 = 5.59, p = 0.0500).
62
) 2 e (individuals/m Alydid Abundanc Alydid
Figure 20. Relationship of the presence of Lespedeza capitata and alydid abundance when considering the amount of speckling. Alydid abundance increased when L. capitata was present
(F2,9 = 14.06, p = 0.0072).
63
) 2 (individuals/m Alydid Abundance Abundance Alydid
Figure 21. Relationship of the presence of Lespedeza capitata and alydid abundance when considering the presence of speckling. Alydid abundance increased when L. capitata was present
(F1,9 = 17.63, p = 0.0030).
64
A.
B.
C.
D.
65
Figure 22. Average amount of time, proportion of time, frequency of visits, and first visits to seeds by seed color. A. Average amount of time spent with each seed color during trials. No difference between ranked time spent with seeds was observed (F2,50 = 1.6334, p = 0.2055). Analysis performed on rank data and controlled for individual as a block effect. B. Proportion of time spent with each seed color compared to the total amount of time interacting with seeds. No significant difference between the ranked proportion of time spent with different seed colors were observed (F2,50 = 2.325 , p = 0.1083), though a trend shows that a greater proportion of time was spent with dark seeds. Color was blocked by individual. C. Frequency of visits to each seed color. No significant difference between the frequency of visits to different seed colors were observed between dark and light seeds was observed (ChiSquare = 0.07, DF = 1, p = 0.7929) or between white and dark seeds (ChiSquare = 3.50, DF= 1, p = 0.0613). However, a trend shows that the greatest frequency of visits was to dark seeds. Analysis was blocked by individual. D. Proportion of an insect’s first visit by seed color. The number of first visits to light seeds did not differ significantly from the number of visits to white seeds (ChiSquare = 3.12, DF = 1, p = 0.0772). The difference between first visits to dark seeds as compared to white seeds was also not significant (ChiSquare = 2.27, Df = 1, p = 0.1318). Analysis was blocked by individual.
66
APPENDIX A.
Correlation of Variables
Correlation between alydid abundance, presence/absence of speckling, amount of speckling, pod density, T. virginiana, B. tinctoria, L. capitata, canopy cover, litter depth, soil chroma, and soil value.
Presence/ Litter Alydid Pod Amount of Canopy Soil Absence Tephrosia Baptisia Lespedeza Depth Soil Value Abundance Density Speckling, Cover Chroma of Dark (cm)
Alydid Abundance 1 0.147 -0.147 0.1397 0.0237 -0.0404 0.1161 -0.0594 0.0401 0.1112 0.0658
Pod Density 0.147 1 0.0542 0.2088 0.2269 0.0203 0.0288 -0.0705 0 -0.0627 -0.0852
Amount of Speckling, No -0.147 0.0542 1 0 0.1467 0.038 -0.3046 0.0051 -0.1136 -0.2229 -0.2235 white
Presence/Absenc 0.1397 0.2088 0 1 0.1235 0.1009 0.0105 -0.0374 -0.2414 0.0178 0.1272 e of Dark
Tephrosia 0.0237 0.2269 0.1467 0.1235 1 0.243 -0.1427 0.1377 0 -0.0961 -0.0329 Baptisia -0.0404 0.0203 0.038 0.1009 0.243 1 -0.0652 0.1741 0.0764 -0.1402 -0.0601
Lespedeza 0.1161 0.0288 -0.3046 0.0105 -0.1427 -0.0652 1 -0.1037 -0.0528 0.187 0.2169
Canopy Cover -0.0594 -0.0705 0.0051 -0.0374 0.1377 0.1741 -0.1037 1 0.1303 -0.187 -0.368
Litter Depth (cm) 0.0401 0 -0.1136 -0.2414 0 0.0764 -0.0528 0.1303 1 -0.0817 -0.1749
Soil Chroma 0.1112 -0.0627 -0.2229 0.0178 -0.0961 -0.1402 0.187 -0.187 -0.0817 1 0.5244 Soil Value 0.0658 -0.0852 -0.2235 0.1272 -0.0329 -0.0601 0.2169 -0.368 -0.1749 0.5244 1
67
APPENDIX B.
Parameter Estimates
Parameter estimates of sites, alydid abundance, soil value, the presence of T. virginiana, and the presence of L. capitata from the standard least squares analysis of the amount of speckling.
Term Estimate Std Error t Ratio Prob>|t| Intercept -0.962044 0.268878 -3.58 0.0007* Site[Bond] -0.343583 0.200963 -1.71 0.0932 Site[Central] 1.2540482 0.256814 4.88 <.0001* Site[Julia'sSavanna] -1.307673 0.409896 -3.19 0.0024* Site[Mary's Savanna] 0.8734408 0.272646 3.20 0.0023* Site[Oak Dune] 0.6369693 0.29341 2.17 0.0344* Site[Reed Road] -1.734088 0.324807 -5.34 <.0001* Site[South Piel] -1.430015 0.329097 -4.35 <.0001* Site[Turnpike] 0.9990021 0.356623 2.80 0.0071* Alydid Abundance -0.154997 0.06927 -2.24 0.0295* Soil Value[4-3] -0.355571 0.316417 -1.12 0.2662 Soil Value[5-4] 0.5863309 0.305831 1.92 0.0606 Soil Value[6-5] 0.2908099 0.279962 1.04 0.3036 Soil Value[7-6] 0.8428315 0.444567 1.90 0.0634 T. virginiana[1-0] -0.540236 0.238882 -2.26 0.0279* L. capitata [1-0] -0.415127 0.21807 -1.90 0.0624
68
APPENDIX B.
Parameter Estimates
Parameter estimates of sites, pod abundance, soil chroma, and the presence of L. capitata from the generalized linear model of presence/absence of speckling.
Term Estimate Std Error L-R Prob>ChiSq ChiSquare Intercept -3.72827 2.3728631 4.7317315 0.0296* L. capitata 1.8878591 0.9730089 5.7831757 0.0162* Soil Chroma[2-1] 2.3207018 2.4487242 1.6931976 0.1932 Soil Chroma[3-2] 2.0740784 1.1089056 6.3298893 0.0119* Soil Chroma[4-3] 4.7362964 2.5871137 4.1466417 0.0417* Pod Abundance -0.022226 0.0090111 13.319652 0.0003* Site[Bond] -2.78414 1.6295545 0 1.0000 Site[LCPC] -1.116218 1.5473978 0.4370912 0.5085 Site[Julia's Savanna] -3.250647 1.8592675 7.6856695 0.0056* Site[Mary's Savanna] -0.921251 1.6028427 0.7108461 0.3992 Site[Meilke Road] 6.9629597 3.1516488 24.201839 <.0001* Site[Oak Dune] 0.5848132 1.039509 0 1.0000 Site[Reed Road] 0.4846888 0.9501557 0 1.0000 Site[South Piel] -3.375193 2.1231634 6.5008183 0.0108* Site[LCPT] 1.4925904 0.9790983 1.5717018 0.2100
69
APPENDIX B.
Parameter Estimates
Parameter estimates of sites, pod abundance, soil value, canopy cover and the amount of speckling from the generalized linear model of alydid abundance of samples that had some amount of speckling.
Term Estimate Std Error L-R Prob>ChiSq ChiSquare Intercept -1.96074 948.25859 0.6637464 0.4152 Site[Bond] 1.1921451 948.25851 0.0033119 0.9541 Site[Central] 2.1212036 948.2585 2.0169454 0.1556 Site[Julia's Savanna] 3.4334904 948.25853 19.218425 <.0001* Site[Mary's Savanna] -0.666246 948.2587 0 1.0000 Site[Meilke Road] -15.30489 8534.326 0.5209186 0.4705 Site[Oak Dune] 2.4190203 948.25853 3.4511459 0.0632 Site[Reed Road] 0.6676074 948.25871 1.762e-5 0.9967 Site[South Piel] 2.5382252 948.25852 5.0864116 0.0241* Site[Turnpike] 0.8912926 948.25865 0.0001649 0.9898 Pod Abundance 0.0077166 0.001623 24.454231 <.0001* Soil Value[4-3] -0.637725 0.4492647 2.0671452 0.1505 Soil Value[5-4] 0.5720243 0.4436431 1.7274951 0.1887 Soil Value[6-5] -0.074651 0.4303533 0.0300246 0.8624 Soil Value[7-6] -2.085581 0.820511 9.0116636 0.0027* Canopy Cover -0.012801 0.0065774 4.1775253 0.0410* Amount of Speckling, -0.01718 0.0090626 3.8441067 0.0499* No white
70
APPENDIX B.
Parameter Estimates
Parameter estimates of sites, pod abundance, soil chroma, and canopy cover from the generalized linear model of alydid abundance that considered the presence/absence of speckling as a possible effect.
Term Estimate Std Error L-R Prob>ChiSq ChiSquare Intercept -0.986473 0.5257671 4.1421422 0.0418 Soil Chroma[2-1] 0.6119728 0.5118679 1.5781228 0.2090 Soil Chroma[3-2] -0.381665 0.3423977 1.3115989 0.2521 Soil Chroma[4-3] -2.218375 0.6287328 19.907518 <.0001 Total Pods 0.0057433 0.0010652 25.840949 <.0001 Site[Bond] -0.393178 0.3263928 1.5991651 0.2060 Site[Central] 0.1075376 0.2884818 0.1366095 0.7117 Site[Julia's Savannah] 2.2647026 0.3547815 43.649286 <.0001 Site[Mary's Savannah] -2.076575 0.6006863 19.537502 <.0001 Site[Meilke Road] -0.491423 0.5615871 0.874055 0.3498 Site[Oak Dune] 0.6656934 0.3690696 2.9979027 0.0834 Site[Reed Road] -0.618457 0.43753 2.3335693 0.1266 Site[South Piel] 0.5296519 0.3504788 2.2959974 0.1297 Site[Turnpike] -0.90994 0.4745015 4.823513 0.0281 Canopy Cover -0.016206 0.0058691 8.5541614 0.0034