Download beyond the ecological: biological invasions alter natural selection on

Ecology, 89(4), 2008, pp. 1023–1031
Ó 2008 by the Ecological Society of America
BEYOND THE ECOLOGICAL: BIOLOGICAL INVASIONS ALTER
NATURAL SELECTION ON A NATIVE PLANT SPECIES
JENNIFER A. LAU1
Center for Population Biology, One Shields Avenue, University of California, Davis, California 95616 USA
Abstract. Biological invasions can have strong ecological effects on native communities
by altering ecosystem functions, species interactions, and community composition. Even
though these ecological effects frequently impact the population dynamics and fitness of native
species, the evolutionary consequences of biological invasions have received relatively little
attention. Here, I show that invasions impose novel selective pressures on a native plant
species. By experimentally manipulating community composition, I found that the exotic plant
Medicago polymorpha and the exotic herbivore Hypera brunneipennis alter the strength and, in
some instances, the direction of natural selection on the competitive ability and anti-herbivore
defenses of the native plant Lotus wrangelianus. Furthermore, the community composition of
exotics influenced which traits were favored. For example, high densities of the exotic
herbivore Hypera selected for increased resistance to herbivores in the native Lotus; however,
when Medicago also was present, selection on this defense was eliminated. In contrast,
selection on tolerance, another plant defense trait, was highest when both Hypera and
Medicago were present at high densities. Thus, multiple exotic species may interact to influence
the evolutionary trajectories of native plant populations, and patterns of selection may change
as additional exotic species invade the community.
Key words: biological invasion; competition; diffuse selection; herbivory; Hypera brunneipennis; Lotus
wrangelianus; Medicago polymorpha; natural selection; plant defense; plant–insect interaction.
INTRODUCTION
Native communities continually face novel ecological
pressures. Many of these pressures arise from anthropogenic alteration of the environment and include such
widespread phenomena as global climate change,
changes in land use, and the spread of exotic species.
Because these environmental changes can influence
communities ecologically, affected natives may also
respond evolutionarily, provided that there is genetic
variation in the traits that influence fitness under the
novel environmental conditions (reviewed in Reznick
and Ghalambor 2001, Strauss et al. 2006). This adaptive
evolution may allow natives to persist in the face of
environmental change, but depends on levels of genetic
variation, the strength of selection, population demographics and gene flow, and correlations between traits
and/or between fitness in different environments (e.g.,
see Etterson 2004). Furthermore, even if natives are able
to adapt to novel ecological threats, evolutionary
responses may result in altered interactions with other
species, prompting a cascade of ecological and evolutionary changes that impact the larger community.
Manuscript received 30 November 2006; revised 14 May
2007; accepted 19 July 2007. Corresponding Editor: M. D.
Eubanks.
1 Present address: Kellogg Biological Station and Department of Plant Biology, Michigan State University, 3700 East
Gull Lake Drive, Hickory Corners, Michigan 49060 USA.
E-mail: [email protected]
Exotic species can directly affect native species via
competition, predation, or by altering abiotic conditions
(Gordon 1998, Mack et al. 2000, Ehrenfeld 2003, Levine
et al. 2003), and they can indirectly affect natives by
changing native species’ interactions with other organisms in the community (e.g., Schmidt and Whelan 1999,
Grosholz et al. 2000, Irwin et al. 2001, Suarez and Case
2002, Rand and Louda 2004, Lau and Strauss 2005).
Because biological invasions result in novel species
interactions, one would expect to see intensified selection
on traits relevant to the new interactions (Vermeij 1996,
Parker et al. 1999), whether these interactions are direct
or indirect. Studies using historical records or comparisons between invaded and uninvaded environments
have documented correlations between evolutionary
changes in native species and biological invasions (e.g.,
Zimmerman 1960, Carroll and Boyd 1992, Singer et al.
1993, Kiesecker and Blaustein 1997, Phillips and Shine
2004, Callaway et al. 2005; reviewed in Strauss et al.
2006). Little work, however, has experimentally quantified how exotic species influence the magnitude and
direction of natural selection on native community
members (but see Rose et al. 2005).
Interactions between pairs of species take place in the
context of a broader community, and interactions
between a focal native species and an abundant invader
could be affected by other co-occurring native and
exotic species alike. Thus, patterns of natural selection
on the native may depend not only on the presence of a
particular invasive but also on the presence and
1023
1024
JENNIFER A. LAU
abundance of other exotic (or native) community
members. For simplicity, here I focus on a native plant
species’ interactions with an exotic plant and an exotic
herbivore. The predicted evolutionary response of the
native species to one of the exotics will be affected by the
second exotic species (i.e., selection will be ‘‘diffuse,’’
sensu Janzen [1980], Fox [1981]) if (1) native traits
involved in interactions with the exotic plant are
genetically correlated with native traits that mediate
interactions with the exotic herbivore; (2) the presence of
one exotic species alters the strength or likelihood of
interactions between the native and the second exotic
species; or (3) the effects of the two exotics on the
relationships between traits and fitness in the native are
not additive (Iwao and Rausher 1997, see also Strauss
and Irwin 2004, Strauss et al. 2005). Several studies have
investigated such ‘‘diffuse’’ selection on plants resulting
from interactions between different herbivore species
(Pilson 1996, Stinchecombe and Rausher 2002), between
herbivores and pollinators (Gomez 2003), and between
herbivores and competitors (Tiffin 2002; see Strauss and
Irwin 2004 for a review). Because there is a dizzying
array of potential multispecies interactions in any
community, diffuse selection may be common, and the
effects of an exotic species on the evolution of a cooccurring native will likely differ across communities
that vary in composition. Similarly, this complexity may
be compounded over time, and selection pressures may
change as more exotic species enter a community.
I report here the results of a series of experimental
removals of exotic species and comparisons between
invaded and uninvaded natural populations demonstrating that patterns of natural selection on native plant
traits depend on the presence and abundance of both an
exotic plant competitor and an exotic herbivore. A prior
study on this system (Lau and Strauss 2005) documented the ecological impacts of invasion by the exotic plant
Medicago polymorpha (hereafter Medicago) and the
exotic insect herbivore Hypera brunneipennis (hereafter
Hypera) on the native California annual Lotus wrangelianus (hereafter Lotus). This previous study found that
increased densities of Medicago and Hypera decreased
Lotus survival and seed production. These fitness effects
were often non-additive and largely resulted because the
presence of Medicago increased Hypera folivory on the
native Lotus (Lau and Strauss 2005). Here, rather than
focusing on the effects of these invasions on average
Lotus fitness, I assessed how the invasions influenced the
relative fitness of specific Lotus genotypes and how
Medicago and Hypera altered associations between
Lotus traits and relative fitness. Because nonadditive
ecological effects on mean fitness do not necessarily
result in non-additive evolutionary effects, it is necessary
to quantify the association between traits and relative
fitness in environments with different combinations of
interacting species to determine whether selection is
diffuse (Strauss et al. 2005). Thus, to assess how
Medicago and Hypera interact to influence evolutionary
Ecology, Vol. 89, No. 4
trajectories, I transplanted known Lotus genotypes into
(1) plots where I experimentally manipulated both
Medicago and Hypera density; and (2) plots in
environments that were naturally invaded or uninvaded
by Medicago where I experimentally manipulated
Hypera density. My results address the following
questions: (1) Does the invasive Medicago alter natural
selection on the competitive ability of the native Lotus?
(2) Does the exotic insect herbivore Hypera affect
natural selection on Lotus anti-herbivore defenses? and
(3) Does Medicago interact with Hypera to create novel
patterns of natural selection on Lotus traits? I predicted
that selection on Lotus anti-herbivore defenses would be
greater when Hypera was abundant and that selection
on Lotus competitive ability would be greater when
Medicago was present. Furthermore, because the presence of Medicago strengthened the ecological interaction
between Hypera and Lotus (Lau and Strauss 2005), I
predicted that the presence of Medicago would indirectly
increase selection on anti-herbivore defenses by increasing Hypera densities.
MATERIALS
AND
METHODS
Study system
Lotus wrangelianus Fischer and C. Meyer (Fabaceae)
is a native plant abundant in open grasslands throughout the McLaughlin Reserve in Napa County, California, USA. Medicago polymorpha L. (Fabaceae) likely
invaded California in the late 1800s from the Mediterranean region (de Haan and Barnes 1998) and now
reaches high densities in many Lotus populations. The
native Lotus and the exotic Medicago are both small,
primarily selfing, winter annuals that germinate with the
first rains in fall and flower in late spring.
Both Lotus and Medicago are attacked by Hypera
brunneipennis (Curculionidae: Coleoptera), the Egyptian
alfalfa weevil. Hypera is an exotic species that spread to
Northern California from North Africa in the 1960s,
approximately 80 years after the invasion by Medicago
(van den Bosch and Marble 1971). Hypera is the most
abundant herbivore in this system; while a weevil budgaller and a hymenopteran seed predator also feed on
Lotus, they damaged less than 26% of the buds and 3%
of the seeds over the duration of the experiment. Lotus
in areas invaded by Medicago typically receive twice as
much Hypera folivory than Lotus in uninvaded areas
(Lau and Strauss 2005). The strong effects of Medicago
and Hypera on Lotus are reflected in the fact that Lotus
seed production increased 10-fold (23.6 6 8.1 vs. 2.9 6
0.8 [mean 6 SE]) when both Medicago and insect
herbivore densities were reduced (Lau and Strauss
2005).
Experimental design
I tested whether Medicago and Hypera alter selection
on Lotus traits, either independently or interactively, by
measuring selection intensities on Lotus populations of
known genotypes transplanted into the replicated
April 2008
EVOLUTIONARY CONSEQUENCES OF INVASION
treatment plots used in Lau and Strauss (2005). In these
plots, Medicago and insect herbivore abundance (primarily Hypera) were manipulated in a four-treatment,
fully crossed factorial design (Medicago present/absent
by insecticide/no insecticide). Treatments were randomly
assigned to 3 3 3 m plots, and 11 replicates of each
treatment combination were established across three
naturally invaded Lotus populations. In these populations, Medicago and Lotus reached densities of 2100
individuals/m2, and Hypera densities were also high. At
one census, weevil larvae were observed feeding on
;25% of surveyed Lotus plants, and naturally occurring
Lotus in the control plots typically experienced herbivory levels where 56% of leaflets were damaged by
Hypera. This level of herbivory is substantial and
reduces Lotus seed production by approximately 80%
(Lau and Strauss 2005).
Manipulation of Medicago and Hypera abundance.—I
manipulated Medicago presence by removing all Medicago seedlings from the 22 Medicago-removal plots.
Weeding treatments were imposed in December 2002,
immediately after the first bout of Medicago germination
and again in January 2003 to remove later germinating
seedlings. Removing Medicago did not alter the growing
environment for Lotus substantially, since the Medicago
seedlings were so small at the time of removal (Lau and
Strauss 2005). Additionally, plots were weeded prior to
transplanting the genotypes used in this experiment. The
weeding treatments were very effective and reduced
Medicago aboveground biomass by 96% (Medicago
biomass [mean 6 SE]: Medicago-removal plots, 2.21 6
7.05 g/m2; control plots, 54.72 6 31.57 g/m2).
I manipulated herbivore abundance by spraying the
insecticide treatment plots with the generalist insecticide
Sevin (Bayer CropScience, Research Triangle Park,
North Carolina, USA) and the no-insecticide plots with
a water control. Insecticide was applied every two to
four weeks from February to April, the period of most
intense Hypera folivory. Sevin does not directly affect
Lotus growth (Lau and Strauss 2005), but did reduce
Hypera damage to Lotus by 58%. Thus, damage from
Hypera was substantially reduced, but was not eliminated by the insecticide treatment. Neither the Medicago-removal nor insecticide treatments had strong
impacts on nontarget plant species (Lau and Strauss
2005), and because both Medicago and Lotus are highly
selfing, any non-target effects of insecticide on the
pollinator community likely had minimal fitness effects
on either species.
Comparisons between naturally invaded vs. uninvaded
Lotus populations.—To determine how selection on
Lotus traits differs between naturally invaded (Medicago
present) and naturally uninvaded (Medicago absent)
environments across substantial variation in Hypera
abundance, seven insecticide treatment plots and seven
no-insecticide plots were also established in two
naturally uninvaded Lotus populations, in addition to
the Medicago-presence plots in the invaded populations
1025
already described. Both invaded and uninvaded sites
were non-serpentine grasslands dominated by similar
plant communities; however, naturally invaded and
uninvaded populations could differ due to environmental differences that are correlated with invasion by
Medicago. For example, invaded populations tended to
be slightly wetter and less serpentine than uninvaded
populations (J. Lau, personal observation).
Estimates of Lotus traits.—I transplanted seedlings of
known full-sib families into a 0.25-m2 subplot in the
center of each treatment plot. Parental plants were
collected as seed from 10 different Lotus patches on the
McLaughlin Reserve. Patches were distinct but were
relatively close together (;0.25–1 km apart). All families
were reared in a greenhouse for one generation to
minimize maternal effects. Seeds were initially germinated in a growth chamber (in the absence of herbivores)
in December 2002, and one plant from each of 73 full-sib
families was transplanted into each treatment plot in
January 2003. Because of mortality prior to and within
two weeks after transplanting, 2659 seedlings were
included in the experiment. In March, I censused each
transplant for herbivory (measured as the proportion of
leaflets with chewing damage from Hypera). Because
leaflets are relatively small (;4–15 mm in length), this
measure of herbivory is likely a good correlate of the
amount of photosynthetic material removed, but may
underestimate the actual amounts of leaf area removed
at high compared to low damage levels. Survival to
flowering was monitored weekly from April to May, and
all fruits were collected at the end of the growing season
(June) to estimate fecundity (total seed production).
I measured the intensity of selection on three Lotus
traits: resistance to herbivory, tolerance to herbivory,
and competitive response to Medicago. Measuring
selection requires obtaining estimates of trait values
and fitness. Resistance is the ability of a plant to avoid
or minimize herbivory, and I estimated resistance in the
no-insecticide treatment plots as 1 (proportion of
leaflets damaged), a standard operational definition of
resistance (Simms and Rausher 1989). Tolerance refers
to a plant’s ability to regrow and reproduce after
herbivore attack (Strauss and Agrawal 1999), and
competitive response is a measure of the degree to
which competition with Medicago decreases Lotus
fitness. Both tolerance and competitive responses were
measured from reaction norms, which compare phenotypes of related individuals growing in different environments. Tolerance was estimated for each family as
the log response ratio of Lotus seed production in the
no-insecticide vs. insecticide treatment plots, i.e., ln([relative seed production in the no-insecticide treatment]/
[relative seed production in the insecticide treatment]).
Competitive response was estimated analogously as the
log response ratio of Lotus seed production in the
presence vs. absence of Medicago.
Calculation of selection intensities.—Selection was
measured in six different environments: naturally unin-
1026
JENNIFER A. LAU
vaded populations, Medicago-removal plots in naturally
invaded populations, and Medicago-presence plots in
naturally invaded populations, in both the presence and
absence of insecticide. Selection intensities measure the
composite selection acting on a trait from both direct
selection and from indirect selection acting on correlated
traits. Selection intensities for Lotus competitive ability,
tolerance to herbivory, and resistance were calculated in
each treatment plot, using logistic regressions with
survival as the fitness measure (Janzen and Stern
1998). To explain more fully, because the experiment
included 58 treatment plots (11 in each of the Medicago
removal 3 insecticide treatments in invaded populations,
plus an additional seven plots per insecticide treatment
in naturally uninvaded populations), I performed
separate regressions for each treatment plot and
estimated selection intensities on Lotus traits 58 times
independently. In each logistic regression, the standardized trait value for each family was included as a
continuous predictor variable; survival was included as a
binomial response variable (PROC GENMOD; SAS,
SAS Institute, Cary, North Carolina, USA). Survival
differed across treatments ranging from a low of 12% in
the Medicago present, no-insecticide treatment to a high
of 51% in naturally uninvaded populations.
Because fitness estimates were used to estimate
competitive ability and tolerance and were also used as
response variables in the selection analyses, the predictor
and response variables are not independent (Mauricio et
al. 1997, Tiffin and Rausher 1999). A common way of
dealing with this problem is to split the dataset, using
one part to estimate competitive ability and tolerance
and the other part to estimate fitness. Accordingly, I
used the individual within the focal plot (i.e., the plot for
which I was calculating selection differentials) to
estimate fitness (the response variable in the selection
analysis), while siblings in all other plots were used to
estimate tolerance and competitive ability (predictor
variables in the selection analyses). This process was
repeated for each plot. Thus, for the calculated selection
differentials in each plot, only one individual per family
was used to estimate fitness, as in a traditional
phenotypic selection analyses, but many individuals in
a given family were used to estimate the trait value, as in
a genotypic selection analysis. While estimating selection
based on the fitness of a single individual per family may
increase error, the selection estimates themselves are
replicated across the 7–11 plots within each treatment.
Selection also was measured as the covariance
between relative seed production and standardized trait
value (Price 1970, Lande and Arnold 1983, Rausher
1992). Patterns of selection estimated from the logistic
regressions (survivorship) were qualitatively similar and
are theoretically similar in magnitude and direction to
those estimated as the covariance between relative seed
production and the plant trait (Janzen and Stern 1998).
Because only a few plants set seed in some treatment
plots, the covariances between traits and seed produc-
Ecology, Vol. 89, No. 4
tion were calculated using the family’s relative fitness
averaged across all plots in a given treatment, and only
results from the selection analyses using survival are
presented in the main text. Results obtained from
selection analyses using seed production as the fitness
measure are presented in the Appendix. Transformed
logistic selection coefficients that can be used in
microevolutionary equations to predict evolutionary
responses (Janzen and Stern 1998) are also presented
in the Appendix.
Statistical analyses
Because selection intensities were calculated independently in each treatment plot, I can compare selection
across replicated environments that differ in the
presence of Medicago and the abundance of Hypera.
Other recent studies have also used replicated experimental manipulations to rigorously examine how
putative selective agents impact predicted evolutionary
trajectories (Bolnick 2004, Moeller and Geber 2005; see
also Wade and Kalisz 1990). To determine whether
selection on Lotus depends on the abundance of Hypera
and Medicago, I conducted ANOVAs for each native
plant trait (PROC MIXED, SAS) in which the selection
intensities for each plot were used as independent
response variables (Wade and Kalisz 1990, Bolnick
2004). Medicago-removal treatment, insecticide treatment, and the interaction were included as fixed factors,
and population was included as a random blocking
factor. When not significant, population and all
nonsignificant interactions were removed from the
analyses to increase power (see Table 1 for full model
results). I also used ANCOVA to further examine
interactions between herbivore abundance and Medicago presence on selection on Lotus traits: I replaced the
categorical insecticide factor with the mean herbivory
intensity (i.e., the mean proportion of leaflets damaged)
experienced by naturally occurring Lotus plants in each
plot. Because the insecticide treatments reduced, but did
not eliminate Hypera herbivory, this ANCOVA accounted for the variation in herbivory among treatment
plots (mean herbivory intensity per plot ranged from
0.08 to 0.52 in the insecticide treatment and from 0.17 to
0.98 in the no-insecticide treatment). The ANCOVA
also allowed me to determine whether the effects of
Medicago on selection on Lotus were primarily due to
Medicago boosting Hypera densities or whether other
mechanisms (i.e., genetic correlations between traits
involved in interactions with Hypera and Medicago or
non-additive effects of Medicago and Hypera on
patterns of selection) also contributed to any observed
diffuse selection. Two plots (both in the Medicago
present, no-insecticide treatment) were excluded from all
analyses because no test plants survived to flower. This
extreme mortality appeared to result from high amounts
of herbivore damage and highlights the potential for
these invaders to impose strong selection on the native
Lotus.
April 2008
EVOLUTIONARY CONSEQUENCES OF INVASION
1027
TABLE 1. ANOVA of the effects of Medicago presence, insecticide treatment, and random effects on the intensity of selection on
Lotus tolerance to herbivory, resistance to herbivory, and competitive response.
Tolerance
Source
df
F
A) Medicago removal ANOVA
Medicago removal
Insecticide
Medicago 3 insecticide
Random effects
Population
1, 37
1, 37
1, 37
0.01
6.26
1.11
B) Nested ANOVA
Medicago presence
Insecticide
Medicago 3 insecticide
Random effects
Population
1, 30
1, 30
1, 30
2
v
0.65
0.36
1.97
4.16
Resistance
P
df
F
0.92
0.02
0.30
1, 37
1, 37
1, 37
1.59
3.60
1.40
0.32
0.55
0.17
0.05
0.0
v
0.50
2
0.0
1, 29
1, 29
1, 29
0.55
0.58
0.01
Competitive response
P
df
F
0.21
0.07 0.24
1, 36
1, 36
1, 36
0.14
0.91
1.33
0.50
0.46
0.45
0.92
0.0
0.50
v2
0.71
0.35
0.26
1.4
1, 2
1, 28
1, 28
P
0.11
0.17
2.57
0.12
0.76
0.69
0.12
2.7
0.05
Notes: Chi-square statistics based on log-likelihood ratio tests are presented for random factors. Part (A) shows the effects of
manipulating Medicago presence by weeding in naturally invaded populations. Part (B) is a nested ANOVA, with population
nested within Medicago presence, and compares selection intensities in naturally invaded Lotus populations vs. naturally uninvaded
Lotus populations. Significant differences (P 0.05) are shown in bold; the dagger ( ) indicates a marginally insignificant effect.
To determine whether selection differs in naturally
invaded vs. uninvaded populations, I used nested
ANOVA. The selection differential in each plot was
the response variable, insecticide treatment and population invasion status were fixed factors, and population
nested within invasion status was a random factor. In all
analyses, degrees of freedom for fixed factors were
estimated with the Sattherthwaite approximation.
RESULTS
AND
DISCUSSION
Direct effects of Medicago and
Hypera on selection on Lotus
As predicted, insect herbivores selected for higher
levels of Lotus defenses (Fig. 1A, B, Table 1A).
Averaged across both Medicago treatments in naturally
invaded populations, selection for increased resistance
and tolerance was more intense when insect herbivores
were abundant than in the insecticide treatment plots
(selection differential [mean 6 SE] for resistance, no
insecticide 0.15 6 0.14, insecticide 0.24 6 0.14; F1,37 ¼
4.0, P ¼ 0.05; for tolerance, no insecticide 0.20 6 0.13,
insecticide 0.18 6 0.12; F1,38 ¼ 6.13, P ¼ 0.02).
Additionally, selection tended to favor lower levels of
resistance and tolerance when herbivory was reduced
with insecticide, suggesting that anti-herbivore defenses
are costly for Lotus (Fig. 1A, B). Hypera likely was
responsible for the insecticide treatment effects; while
other insect herbivores were relatively rare in 2003,
Hypera was very abundant and caused much damage in
the control (Medicago-present, no-insecticide) plots.
I failed to detect evidence that Medicago altered
selection on Lotus competitive response. While selection
for increased competitive response tended to be more
intense in the presence of Medicago than in Medicagoremoval plots or naturally uninvaded populations, these
differences were not statistically significant (Fig. 1C,
Table 1).
Interactions between Medicago and
Hypera on selection on Lotus
I detected few statistically significant interactions
between Medicago-presence and insecticide treatments
on selection on Lotus traits (Table 1A, B). In the
comparison of naturally invaded vs. uninvaded populations, however, there was a significant interaction
between Medicago-presence and the insecticide treatment on selection on Lotus tolerance (Table 1B); in the
absence of insecticide, selection on tolerance tended to
be more intense in naturally invaded populations than
uninvaded populations (selection differential: invaded
0.33 6 0.15, uninvaded 0.10 6 0.17; t1,27 ¼ 1.83, P ,
0.08). In addition, the difference in selection on Lotus
tolerance between the no-insecticide and insecticide
treatments tended to be greatest, and was only
significant, in the presence of Medicago (Fig. 1A). In
the absence of insecticide, there was also a trend for
more intense selection on resistance in invaded populations when Medicago was experimentally removed
compared to plots where Medicago remained present
(selection differential: Medicago present 0.11 6 0.21,
Medicago removed 0.37 6 0.19; t1,35 ¼ 1.70, P , 0.1).
Interestingly, in the insecticide treatments where Hypera
herbivory was reduced; selection differentials on both
defense traits were very similar in magnitude across all
Medicago treatments. This result implies that any effects
of Medicago on patterns of selection on Lotus antiherbivore defenses are likely indirect and mediated by
the herbivore community. The direction of the Medicago
effect, however, differed depending on the type of
defense. For tolerance, the trend was in the direction
predicted based on the ecological effect of Medicago
increasing herbivory on Lotus, with Medicago increasing
selection. For resistance, the trend was opposite that
predicted, and Medicago tended to reduce selection on
this defense trait (Fig. 1A, B).
1028
JENNIFER A. LAU
FIG. 1. Selection differentials for (A) Lotus tolerance to
herbivory, (B) resistance to herbivores, and (C) competitive
response in control plots in invaded patches (Medicago
present), Medicago-removal plots in invaded patches (Medicago
absent), and naturally uninvaded patches (Medicago absent) in
insecticide (insects reduced) and no-insecticide (insects present)
treatments. Selection differentials measure the total selection on
a trait from both direct selection and from indirect selection
acting on correlated traits. Positive differentials indicate that
genotypes with higher trait values are selectively favored;
negative differentials indicate that genotypes with lower trait
values are favored. See Materials and methods: Calculation of
selection intensities for more detail. LS mean values (6SE) with
the same letter are not significantly different (P . 0.05). Only
selection on resistance in the Medicago-removal treatment in
the absence of insecticide differed significantly from zero.
Ecology, Vol. 89, No. 4
The high amount of variation in herbivory across
plots within insecticide treatments could have limited my
ability to detect significant insecticide treatment effects
and interactions between Medicago-presence and insecticide treatments. The ANCOVA explicitly examines
how Medicago presence impacts patterns of selection
across this continuous range of herbivory. For all traits,
I detected significant interactions between the Medicagoremoval treatment and the intensity of herbivory (Table
2); however, it should be noted that when the two plots
receiving the highest amounts of herbivory were
excluded, these interactions were no longer significant,
although the same trends remained. These results
suggest that the strength of selection exerted on Lotus
by one exotic species (Medicago) may depend on the
abundance of a second exotic species (Hypera). For
example, high levels of herbivory tended to increase
selection on Lotus tolerance, as might be expected;
however, this positive relationship only occurred in the
presence of Medicago (Fig. 2A). In contrast, increasing
herbivory levels only resulted in selection for higher
resistance in the absence of Medicago (Fig. 2B). Thus,
Medicago appeared to increase selection on tolerance at
high damage levels but may decrease selection on
resistance. These results suggest that the presence of
Medicago may shift the type of plant defense (resistance
vs. tolerance) that is favored in the native Lotus. Recent
studies in other systems also provide convincing
evidence that different defenses (resistance vs. tolerance)
may be favored under different environmental conditions (Fornoni et al. 2004).
Medicago also interacted with Hypera abundance to
alter selection on Lotus competitive response. Over all
herbivory intensities, I did not detect selection on
competitive response in the absence of Medicago.
However, there was a positive relationship between the
intensity of herbivory and selection for increased Lotus
competitive response when Medicago was present (Fig.
2C). These results provide suggestive evidence that
herbivore abundance may mediate the strength of
selection on Lotus competitive ability.
In this system, diffuse selection may result, in part,
from the indirect ecological effect of Medicago increasing Hypera herbivory on Lotus; these types of interactive
effects of exotics on natives may be common (e.g., Rand
and Louda 2004, see also White et al. 2006). If diffuse
selection results entirely from Medicago boosting
densities of Hypera, however, then the mean selection
intensity on Lotus defense traits should be greater in the
presence of Medicago than in the absence of Medicago,
as was observed for tolerance, but the relationship
between selection and intensity of herbivory should
remain constant across Medicago treatments. Instead, I
observed significant interactions between Medicago
presence and intensity of Hypera herbivory on selection
on all traits (Fig. 2). These interactions imply that
diffuse selection also results through other mechanisms,
such as genetic correlations between Lotus traits
April 2008
EVOLUTIONARY CONSEQUENCES OF INVASION
1029
TABLE 2. ANCOVA of the effects of Medicago removal, level of Hypera herbivory, and random effects on the intensity of
selection on Lotus tolerance to herbivory, resistance to herbivory, and competitive response.
Tolerance
Source
df
F
Medicago removal
Herbivory level
Medicago 3 herbivory
Random effects
Population
1, 38
1, 36
1, 36
3.30
0.22
4.52
v
2
0.0
Resistance
P
df
F
0.08
0.64
0.04
1, 37
1, 36
1, 37
2.08
5.07
6.93
0.50
v
2
1.1
Competitive response
P
df
F
0.16
0.03
0.01
1, 38
1, 36
1, 37
2.69
7.23
4.49
0.14
v2
P
0.11
0.01
0.04
0.5
0.24
Note: Significant differences (P 0.05) are shown in bold.
mediating interactions with Medicago and traits mediating interactions with Hypera or nonadditive effects of
Medicago and Hypera on the relationship between Lotus
traits and fitness (Iwao and Rausher’s criteria 1 and 3
for diffuse selection described previously). For example,
Lotus tolerance and competitive ability tend to be
positively correlated (r ¼ 0.31, P , 0.07). While the
mechanistic basis of Lotus tolerance and competitive
ability are unknown, it is plausible that traits conferring
high tolerance also are associated with increased
competitive ability (e.g., high growth rates). Regardless
of the underlying mechanism, however, the positive
correlation between tolerance and competitive ability
could result in indirect selection for increased competitive ability in plots that experience high amounts of
herbivore damage and could contribute to the observed
positive relationship between intensity of herbivory and
selection on competitive ability. Thus, while some of the
evolutionary consequences of these invasions might be
predictable responses to the novel ecological interactions
(e.g., selection on tolerance was most intense in areas
where Medicago increased herbivory on Lotus), some
evolutionary outcomes that result from interactions
between exotic competitors and exotic herbivores cannot
be predicted based on ecological effects alone, making
the long-term consequences of invasion difficult to
foresee. These data also illustrate that diffuse selection
cannot be identified by only measuring mean fitness
effects and ecological outcomes and demonstrate that
more trait-based approaches are needed to gain a full
understanding of how multiple species interact to
influence evolutionary trajectories (Strauss et al. 2005).
While similar interactions easily could occur between
native species, this example highlights the potential
complexities that arise when multiple novel species enter
communities and may be particularly applicable to cases
where a second strongly interacting exotic species is
introduced.
Prior work demonstrated that Lotus genotypes
collected from heavily invaded sites where both Hypera
and Medicago were present at high density tended to
have lower competitive abilities, lower tolerance, and
lower resistance than Lotus genotypes collected from
uninvaded sites, where densities of both invaders were
low (Lau 2006). While none of these differences in trait
means were statistically significant, the trends for
FIG. 2. Relationship between mean herbivory intensity and
selection on Lotus (A) tolerance to herbivory, (B) resistance,
and (C) competitive response in the presence of Medicago (solid
line; solid symbols) and in Medicago-removal plots (dashed
line; open symbols). Each symbol represents the selection
differential and herbivory intensity in a single treatment plot.
Regression coefficients for each treatment are as follows.
Tolerance: Medicago present, b ¼ 0.98, P , 0.12; Medicago
removed, b ¼0.63, P , 0.18. Competitive response: Medicago
present, b ¼ 2.1, P , 0.005; Medicago removed, b ¼ 0.34, P ,
0.51. Resistance: Medicago present, b ¼ 0.14, P , 0.83;
Medicago removed, b ¼ 1.77, P , 0.005.
1030
JENNIFER A. LAU
reduced mean tolerance and competitive ability in
invaded relative to uninvaded populations are in the
opposite direction of predicted differences between
invaded and uninvaded populations based on the
selection analyses described here. When both Hypera
and Medicago are present at high density (as in invaded
populations since at least 2001 and, likely, for several
decades), the selection analyses predict that populations
should evolve higher tolerance and, potentially, higher
competitive ability than uninvaded populations. The
discrepancy between the observed and predicted differences in trait values between invaded and uninvaded
populations could result because of the high error
surrounding estimates of selection or because selection
is temporally variable. However, one other possible
explanation is that while Lotus genotypes may have
adapted to the initial Medicago invasion, they have not
yet responded to invasion from the later arriving Hypera
(Lau 2006). Temporal variation in Hypera abundance,
constraints on selection due to other herbivore community members that were rare in 2003 but that sometimes
reach high densities (Lau and Strauss 2005), or
inadequate time for adaptive changes to occur could
all contribute to the lack of observed evolutionary
response to the combined presence of Hypera and
Medicago.
Strong ecological perturbations, such as global
climate change and biological invasions, frequently
result in evolutionary changes in natives (e.g., Zimmerman 1960, Carroll and Boyd 1992, Singer et al. 1993,
Kiesecker and Blaustein 1997, Reale et al. 2003, Phillips
and Shine 2004, Callaway et al. 2005). Others have used
comparisons between invaded and uninvaded environments and temporal changes in the phenotypes of natives
to document evolutionary responses to invasion by
exotic species. The experimental approach detailed here
allows for investigation of one of the mechanisms
(altered patterns of natural selection) potentially underlying these evolutionary responses. Adaptation by
natives to exotics could minimize the impact of exotic
species on the persistence of natives, by reducing strong
ecological effects of invasion. However, even if evolutionary responses minimize the demographic impact, the
resulting native populations may be very different
(genetically, and as a result, phenotypically) than they
were prior to the invasion. Thus, it is plausible that these
evolutionary responses could affect how the native
species interacts with other community members,
resulting in further perturbations to the community.
Understanding how exotic species influence patterns of
natural selection on natives is the first step to determining how native species will respond evolutionarily to
biological invasions and to comprehending the longterm impacts of exotic species on native communities.
ACKNOWLEDGMENTS
R. Hutchinson, M. Lau, P. Lau, E. Renquist, and C.
Simonson assisted with fieldwork, and suggestions from S.
Strauss, K. Rice, M. Stanton, A. Agrawal, D. Bolnick, M.
Ecology, Vol. 89, No. 4
Brock, M. Katz, A. McCall, E. Preisser, P. Tiffin, L. Yang, and
two anonymous reviewers greatly improved the manuscript.
Work was performed at the University of California Natural
Reserve System McLaughlin Reserve and was funded by Center
for Population Biology research awards and the Biological
Invasions IGERT NSF-DGE 0114432 (UC–Davis), and a
National Science Foundation Dissertation Improvement Grant
IBN-0206601 to J. A. Lau and S. Strauss.
LITERATURE CITED
Bolnick, D. I. 2004. Can intraspecific competition drive
disruptive selection? An experimental test in natural populations of sticklebacks. Evolution 58:608–618.
Callaway, R. M., W. M. Ridenour, T. Laboski, T. Weir, and
J. M. Vivanco. 2005. Natural selection for resistance to the
allelopathic effects of invasive plants. Journal of Ecology 93:
576–583.
Carroll, S. P., and C. Boyd. 1992. Host race radiation in the
soapberry bug: natural history with the history. Evolution 46:
1052–1069.
de Haan, R. L., and D. K. Barnes. 1998. Inheritance of pod
type, stem color, and dwarf growth habit in Medicago
polymorpha. Crop Science 38:1558–1561.
Ehrenfeld, J. G. 2003. Effects of exotic plant invasions on soil
nutrient cycling processes. Ecosystems 6:503–523.
Etterson, J. R. 2004. Evolutionary potential of Chamaecrista
fasciculata in relation to climate change. II. Genetic
architecture of three populations reciprocally planted along
an environmental gradient in the Great Plains. Evolution 58:
1459–1471.
Fornoni, J., P. L. Valverde, and J. Núñez-Farfán. 2004.
Population variation in the cost and benefit of tolerance
and resistance against herbivory in Datura stramonium.
Evolution 58:1696–1704.
Fox, L. R. 1981. Defense and dynamics in plant–herbivore
systems. American Zoologist 21:853–864.
Gomez, J. M. 2003. Herbivory reduces the strength of
pollinator-mediated selection in the Mediterranean herb
Erysimum mediohispanicum: consequences for plant specialization. American Naturalist 162:242–256.
Gordon, D. R. 1998. Effects of invasive, non-indigenous plant
species on ecosystem processes: lessons from Florida.
Ecological Applications 8:975–989.
Grosholz, E. D., G. M. Ruiz, C. A. Dean, K. A. Shirley, J. L.
Maron, and P. G. Connors. 2000. The impacts of a
nonindigenous marine predator in a California bay. Ecology
81:1206–1224.
Irwin, R. E., A. K. Brody, and N. M. Waser. 2001. The impact
of floral larceny on individuals, populations, and communities. Oecologia 129:161–168.
Iwao, K., and M. D. Rausher. 1997. Evolution of plant
resistance to multiple herbivores: quantifying diffuse coevolution. American Naturalist 149:316–335.
Janzen, D. H. 1980. When is it coevolution? Evolution 34:611–
612.
Janzen, F. J., and H. S. Stern. 1998. Logistic regression for
empirical studies of multivariate selection. Evolution 52:
1564–1571.
Kiesecker, J. M., and A. R. Blaustein. 1997. Population
differences in responses of red-legged frogs (Rana aurora)
to introduced bullfrogs. Ecology 78:1752–1760.
Lande, R., and S. J. Arnold. 1983. The measurement of
selection on correlated characters. Evolution 37:1210–1226.
Lau, J. A. 2006. Evolutionary responses of native plants to
novel community members. Evolution 60:56–63.
Lau, J. A., and S. Y. Strauss. 2005. Insect herbivores drive
important indirect effects of exotic plants on native
communities. Ecology 86:2990–2997.
Levine, J. M., M. Vila, C. M. D’Antonio, J. S. Dukes, K.
Grigulis, and S. Lavorel. 2003. Mechanisms underlying the
April 2008
EVOLUTIONARY CONSEQUENCES OF INVASION
impacts of exotic plant invasions. Proceedings of the Royal
Society B 270:775–781.
Mack, R. N., D. Simberloff, W. M. Lonsdale, H. Evans, M.
Clout, and F. A. Bazzaz. 2000. Biotic invasions: causes,
epidemiology, global consequences, and control. Ecological
Applications 10:689–710.
Mauricio, R., M. D. Rausher, and D. S. Burdick. 1997.
Variation in the defense strategies of plants: Are resistance
and tolerance mutually exclusive? Ecology 78:1301–1311.
Moeller, D. A., and M. A. Geber. 2005. Ecological context of
the evolution of self-pollination in Clarkia xantiana: population size, plant communities, and reproductive assurance.
Evolution 59:786–799.
Parker, I. M., D. Simberloff, W. M. Lonsdale, K. Goodell, M.
Wonham, P. M. Kareiva, M. H. Williamson, B. Von Holle,
P. B. Moyle, J. E. Byers, and L. Goldwasser. 1999. Impact:
toward a framework for understanding the ecological effects
of invaders. Biological Invasions 1:3–19.
Phillips, B. L., and R. Shine. 2004. Adapting to an invasive
species: Toxic cane toads induce morphological change in
Australian snakes. Proceedings of the National Academy of
Sciences (USA) 101:17150–17155.
Pilson, D. 1996. Two herbivores and constraints on selection
for resistance in Brassica rapa. Evolution 50:1492–1500.
Price, G. R. 1970. Selection and covariance. Nature 227:520–
521.
Rand, T. A., and S. M. Louda. 2004. Exotic weed invasion
increases the susceptibility of native plants to attack by a
biocontrol herbivore. Ecology 85:1548–1554.
Rausher, M. D. 1992. The measurement of selection on
quantitative traits: biases due to environmental covariances
between traits and fitness. Evolution 46:616–626.
Reale, D., A. G. McAdam, S. Boutin, and D. Berteaux. 2003.
Genetic and plastic responses of a northern mammal to
climate change. Proceedings of the Royal Society B 270:591–
596.
Reznick, D. N., and C. K. Ghalambor. 2001. The population
ecology of contemporary adaptations: what empirical studies
reveal about the conditions that promote adaptive evolution.
Genetica 112:183–198.
Rose, K. E., S. Louda, and M. Rees. 2005. Demographic and
evolutionary impacts of native and invasive insect herbivores:
a case study with Platte thistle, Cirsium canescens. Ecology
86:453–465.
Schmidt, K. A., and C. J. Whelan. 1999. Effects of exotic
Lonicera and Rhamnus on songbird nest predation. Conservation Biology 13:1502–1506.
1031
Simms, E. L., and M. D. Rausher. 1989. The evolution of
resistance to herbivory in Ipomoea purpurea. II. Natural
selection by insects and costs of resistance. Evolution 43:573–
585.
Singer, M. C., C. D. Thomas, and C. Parmesan. 1993. Rapid
human-induced evolution of insect–host associations. Nature
366:681–683.
Stinchecombe, J. R., and M. D. Rausher. 2002. The evolution
of tolerance to deer herbivory: modifications caused by the
abundance of insect herbivores. Proceedings of the Royal
Society Biological Sciences Series B 269:1241–1246.
Strauss, S. Y., and A. A. Agrawal. 1999. The ecology and
evolution of plant tolerance to herbivory. Trends in Ecology
and Evolution 14:179–185.
Strauss, S. Y., and R. E. Irwin. 2004. Ecological and
evolutionary consequences of multispecies plant–animal
interactions. Annual Review of Ecology, Evolution, and
Systematics 35:435–466.
Strauss, S. Y., J. A. Lau, and S. P. Carroll. 2006. Evolutionary
responses of natives to introduced species: What do
introductions tell us about natural communities? Ecology
Letters 9:357–374.
Strauss, S. Y., H. Sahli, and J. K. Conner. 2005. Toward a more
trait-centered approach to diffuse (co)evolution. New Phytologist 165:81–90.
Suarez, A. V., and T. J. Case. 2002. Bottom-up effects on
persistence of a specialist predator: ant invasions and horned
lizards. Ecological Applications 12:291–298.
Tiffin, P. 2002. Competition and time of damage affect the
pattern of selection acting on plant defense against herbivores. Ecology 83:1981–1990.
Tiffin, P., and M. D. Rausher. 1999. Genetic constraints and
selection acting on tolerance to herbivory in the common
morning glory Ipomoea purpurea. American Naturalist 154:
700–716.
van den Bosch, R., and V. L. Marble. 1971. Egyptian alfalfa
weevil: the threat to California alfalfa. California Agriculture
25:3–4.
Vermeij, G. J. 1996. An agenda for invasion biology. Biological
Conservation 78:3–9.
Wade, M. J., and S. Kalisz. 1990. The causes of natural
selection. Evolution 44:1947–1955.
White, E. M., J. C. Wilson, and A. R. Clarke. 2006. Biotic
indirect effects: a neglected concept in invasion biology.
Diversity and Distributions 12:443–455.
Zimmerman, E. C. 1960. Possible evidence of rapid evolution in
Hawaiian moths. Evolution 14:137–138.
APPENDIX
Estimates of selection differentials for Lotus tolerance, resistance, and competitive ability in each environment (Ecological
Archives E089-061-A1).