EVOLUTIONARY AND APPLIED APPROACHES FOR THE MANAGEMENT OF
INVASIVE SWALLOW-WORT VINES (VINCETOXICUM SPP.): BIOLOGICAL CONTROL,
ECOSYSTEM RESTORATION AND CONSEQUENCES FOR MONARCH BUTTERFLIES
(DANAUS PLEXIPPUS)
By
Brianna Foster
A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Entomology – Doctor of Philosophy
Ecology, Evolution and Behavior – Dual Major
2024
ABSTRACT
Classical weed biological control uses natural enemies that have co-evolved with their host plants in
their native range and provides a sustainable and effective strategy for the long-term management of
invasive species. However, its successful implementation encounters several challenges that must be
addressed to maximize efficacy. Biological control agents, often reared extensively in quarantine, face
issues like adaptation to culture, reduced genetic diversity and inbreeding, which can affect their
establishment and efficacy. Upon implementation of biological control, site restoration may be limited
without the proper integration of supplemental restoration techniques. Restoration techniques combined
with biological control are not extensively studied, limiting their widespread adoption and application.
Furthermore, the impact of invasive species to native species of concern is important to optimize
management plans. This dissertation investigates the integration of applied and evolutionary approaches
in the control of invasive swallow-wort vines (Vincetoxicum spp.), emphasizing the importance of genetic
background when employing the biocontrol agent Hypena opulenta, the restoration of invaded forests,
and the implications for the monarch butterfly (Danaus plexippus). In Chapter 1, I research the
importance of genetic background by creating an outbred population by crossing a long-term laboratory-
reared population with individuals from an-established population of H. opulenta in the field. These
populations are used to understand the role of intraspecific outcrossing in enhancing the fitness and
neutral genetic diversity of Hypena opulenta with laboratory and field experiments alongside molecular
analysis. The results show that increases in neutral genetic variation can translate to tangible fitness
benefits in the outbred population measured as increases in production of pupae in the laboratory, and
larval feeding rates and second-generation adult production under field conditions. In Chapter 2, I
research how biological control, manual removal of swallow-wort, and native seeding, alone and
combined, can impact restoration success. The results show that when native seeding is combined with
biological control, swallow-wort cover is reduced, and species richness is enhanced. In Chapter 3, I
investigate the effects of genetic background and demography in the establishment success of Hypena
opulenta by conducting experimental field releases. However, I was not able to understand this
relationship because of establishment failure. In repeat releases two years later, where genetic background
was the sole focus, the results mirrored patterns seen in the field cage experiments from Chapter 1 with
the outbred population eating more plant tissue and producing more adults compared to lower genetic
diversity laboratory reared and field collected populations. Finally, in Chapter 4, I explore the oviposition
choices of the monarch butterfly in Michigan. I determine the frequency with which monarch butterflies
mistakenly lay eggs on swallow-worts under natural field conditions and use laboratory experiments to
uncover the underlying mechanisms of their oviposition decisions, aiming to clarify the factors
influencing monarch egg-laying behavior. Field results showed that monarchs can lay up to 17% of their
eggs on swallow-wort locally, and about 8% overall. Despite oviposition failure in the lab experiment,
monarch larvae exhibited 16% mortality in the lab directly attributed to feeding on swallow-wort. This
dissertation contributes to the discourse on ecological restoration, biological control, and provides
insights for future conservation and management endeavors in ecosystems threatened by invasive species.
This dissertation is dedicated to my family, pets, and friends.
Thank you for always being there for me.
iv
ACKNOWLEDGEMENTS
I would like to sincerely thank my major professor Marianna Szűcs for her support, guidance,
and insight throughout the duration of my Ph.D. degree. I have learned a great deal as a student
in her lab. I would also like to thank the members of my committee – Zsofia Szendrei, Lars
Brudvig, and Ruth Hufbauer – for their guidance and encouragement. I’d like to thank all the
technicians, undergraduate and graduate students that helped me – Anupreksha Jain, Mark
Hamlyn, Alyssa Guest, Thang Sawmtung, Morgan Atkins, Brianna Alred, Andrew Yoshimoto,
and Oscar Istas.
I would like to show appreciation for our funding sources, who we could not have completed
this work without. This work was supported by the Michigan Department of Agriculture
(MDNR) Michigan Invasive Species Grant Program (award #IS18-2006 and #IS21-0284), the
Michigan State University Mott Pre-Doctoral Fellowship, and the Sustainable Michigan
Endowed Project.
Additionally, I’d like to thank everyone with whom I worked with during my time at
Michigan State University, such as other professors, students, and administrative personnel. A
heartfelt thank you to all my friends, particularly those I've met in Michigan, whose unwavering
support was instrumental in my journey to completing my PhD. I have made many lifelong
friends and memories that I won’t ever forget. And lastly, I would like to thank my family who
has been incredibly supportive through this entire process.
v
TABLE OF CONTENTS
CHAPTER 1: Intraspecific hybridization improves performance of the biological control agent
Hypena opulenta…………………………………………….…………………………………….1
LITERATURE CITED………………………………………………………………….32
CHAPTER 2: Restoring swallow-wort (Vincetoxicum spp.) invaded woodlands: swallow-wort
control and native seeding diversifies plant communities…….……….…...………………...….39
LITERATURE CITED………………………………………………………………….71
APPENDIX……………………………………………………………………………...77
CHAPTER 3: Exploring the demographic and genetic factors for establishment of Hypena
opulenta for the biological control of invasive swallow-wort vines (Vincetoxicum spp.)..……...79
LITERATURE CITED………………………………………………………………...103
CHAPTER 4: Navigating the ecological trap: monarch butterfly (Danaus plexippus) behavior in
the presence of invasive swallow-wort vines (Vincetoxicum spp.)………………..……………107
LITERATURE CITED………………………………………………………………...121
vi
CHAPTER 1:
Intraspecific hybridization improves performance of the biological control agent Hypena
opulenta
Abstract
Evolutionary processes during species introduction are crucial for understanding population
establishment and persistence. Intraspecific hybridization can lead to varied fitness outcomes
depending on the genetic distance between parental populations. The link between neutral
genetic diversity and fitness in the literature is still uncertain, making it difficult to predict the
outcomes of intraspecific hybridization. We investigated the role of intraspecific outcrossing, in
enhancing the fitness and genetic diversity of Hypena opulenta, a biological control agent used
against invasive swallow-wort vines in North America. We created an outbred population by
crossing a long-term laboratory-reared population with a field-established population that was
derived from the same source as the laboratory population 12 generations prior. The fitness of
parental and outbred populations was tested in the laboratory by comparing the number of pupae
produced by individual females. In the field the levels of defoliation and second-generation adult
emergence by groups of five females was compared. Whole Genome Sequencing (WGS) was
conducted on 15-20 individuals from the parental and hybrid populations. Outbreeding increased
fitness both under laboratory and field conditions with the outbred females producing the most
pupae in the laboratory, the highest defoliation rates of the plants, and the most second-
generation adults in the field. Outbreeding increased genome-wide heterozygosity, and despite
how recently the parental populations were isolated from each other there was clear population
genetic structure between the parental and hybrid populations. These findings show that
increases in neutral genetic variation can translate to tangible fitness benefits both under
laboratory and field conditions. The results highlight the critical role genetic diversity and
1
intentional hybridization may play to improve the performance of biological control agents prior
to field releases in a novel environment.
1. Introduction
Evolutionary processes during species introduction are key to understanding population
establishment and persistence, particularly for organisms released into novel environments
(Sakai et al., 2001; Szucs et al., 2014). Such processes, like intraspecific hybridization, or the
crossing of individuals within the same species from different populations, can produce diverse
fitness outcomes influenced by the genetic distance between the populations involved effects
(Colautti et al., 2017; Frankham et al., 2011; Whiteley et al., 2015). While hybridization among
genetically similar populations may have negligible effects, increasing genetic distance can lead
to outbreeding depression due to the disruption of co-adapted gene complexes (Colautti et al.,
2017; Frankham et al., 2011; Templeton et al., 1986; Whiteley et al., 2015). This phenomenon
underscores the need for detailed genetic analysis to predict hybridization outcomes, especially
when historical data is scarce (Dlugosch et al., 2015). Furthermore, understanding intraspecific
hybridization requires examining genetic variation in neutral parts of the genome and clarifying
its link to fitness. In the context of biological control, understanding these dynamics is crucial as
control agents often face genetic bottlenecks and inbreeding during mass rearing (Bertin et al.,
2017; Szűcs et al., 2019). We seek to investigate the fitness effects of intentional hybridization in
the biological control agent Hypena opulenta Christoph (Lepidoptera: Erebidae), recently
introduced to combat invasive swallow-wort vines (Vincetoxicum spp.) in North America. By
comparing hybridized laboratory and field populations, we aim to elucidate the impact of genetic
diversity on fitness, ultimately contributing to the optimization of biological control strategies.
2
During intraspecific hybridization the genetic distance between the hybridizing populations
can mediate the outcome from neutral, to beneficial or negative fitness effects (Colautti et al.,
2017; Frankham et al., 2011; Whiteley et al., 2015). Hybridization between populations that are
genetically similar may not have any discernable effects on fitness, however, as the genetic
distance between hybridizing populations grow, the negative effects of outbreeding depressions
can occur (Colautti et al., 2017; Frankham et al., 2011; Templeton et al., 1986; Whiteley et al.,
2015). Outbreeding depression may arise as a result of disrupting co-adapted gene complexes
upon interbreeding that can vary even within the same environment (Frankham et al., 2011). The
degree of divergence between hybridized individuals can make the difference between heterosis
and outbreeding depression (Edmands, 1999; Wei & Zhang, 2018). The risk of outbreeding
depression is considered to be small if populations have the same karyotype, inhabit similar
environments, and have not experienced genetic isolation for more than 500 years (Frankham et
al., 2011). Yet it may not always be easy to discern if hybridization will have a positive, neutral,
or negative effect when historical data for populations is scarce, therefore genetic analysis to
understand population history and divergence is recommended (Dlugosch et al., 2015). In an
example by Audet et al. (2017), intraspecific crosses of salmon from two different geographic
locations resulted in neither heterosis nor outbreeding depression. In this case, assumptions on
population divergence were made based on population history and no molecular analysis was
done to determine the degree of population divergence and the genetic diversity of the resulting
crosses (Audet et al., 2017). In another example, research investigating the impact of population
divergence and geographical distance on the intertidal copepod Tigriopus californicus found that
F1 individuals exhibited similar or greater performance than their parental lines. However, all
but one divergent line – a hybrid line with less than 5% sequence divergence – saw F2 hybrid
3
breakdown indicating the importance of measuring the degree of divergence (Edmands, 1999).
Indeed, past studies have struggled to identify the optimal mating distance that might provide the
beneficial effects of heterosis and avoid the negative effects of outbreeding (Wei & Zhang,
2018). A meta-analysis by Wei and Zhang (2018), showed a hump shaped relationship between
intraspecific hybrid performance and mating distance across environments, whereas performance
boost is minimal (or nearly non-existent) at the shortest mating distances, improves with greater
mating distances up to a peak genetic distance, and then declines as mating distance continues to
increase. With advanced and increasingly affordable genetic techniques the risk of outbreeding
can be reduced, however, it is still unclear if hybridization between recently isolated and
genetically similar populations can bring any tangible benefits.
To enhance our understanding of intraspecific hybridization, it is essential to explore the
underappreciated significance of genetic variation in the neutral parts of the genome. Kimura's
neutral theory of molecular evolution (1968) posits that the genetic diversity observed within and
between species is primarily a product of non-selective forces like genetic drift, gene flow, and
mutation, rather than natural selection (Kimura, 1968; Neale & Wheeler, 2019). Numerous
studies have established a link between heterozygosity and fitness, particularly in highly inbred
populations, though these often rely on indirect methods such as breeding pedigrees (Arct et al.,
2017; Chapman et al., 2009; Hansson and Westerberg, 2002). Whether this relationship is due to
changes at neutral markers, linkage disequilibrium at neutral markers and fitness-related genes,
or if heterozygous markers signify overall genomic heterozygosity is unknown (Arct et al., 2017;
Hansson and Westerberg, 2002; Miller & Coltman, 2014). Despite this, the correlation between
heterozygosity and fitness is not always straightforward, and the role of neutral genetic variation
in this context is increasingly seen as potentially overemphasized (Chapman et al., 2009; Sudkya
4
et al., 2017; Teixeira & Huber, 2021). The effectiveness of using neutral genetic diversity as a
marker for conservation efforts is complicated by factors like the intricate relationship between
neutral genome-wide diversity and a population's survival or extinction risks, particularly when
considering outbreeding and other evolutionary processes like gene-environment interactions and
epistasis (Teixeira & Huber, 2021). The variable results from studies on heterozygosity-fitness
correlations, which depend on environmental conditions, highlight the complexities of this
relationship (Chapman et al., 2009). Presently, research on heterozygosity-fitness correlations is
primarily concentrated on populations that have naturally hybridized, using historical data and
indicators to decipher the influence of both neutral and non-neutral processes (Chapman et al.,
2009; Hansson and Westerberg, 2002).
Classical biological control, which offers a sustainable solution for managing invasive weeds
by introducing co-evolved natural enemies from the invader's native region, can improve our
understanding of the evolutionary processes that govern intentionally introduced populations and
provide insight into in their success rates. Globally, the success rate of established, intentionally
introduced weed biological control agents stands at 63.2%, and only 23.9% of these agents cause
significant damage to the target weed (Schwarzländer et al., 2018). These rates indicate that there
are fundamental factors affecting the success of classical biological control, potentially hindering
its broader implementation (Brodeur et al., 2018; Cock et al., 2016). Insects used in biological
control are often sourced from small native populations and mass-reared in quarantine for several
years before release because of the lengthy host-specificity testing and approval processes. This
practice subjects laboratory populations to phenomena such as population bottlenecks, founder
effects, cultural adaptation, and inbreeding (Bertin et al., 2017; Szűcs et al., 2019). Organisms
that are mass-reared often experience fluctuating and low population sizes that can lead to
5
genetic drift, reduced genetic diversity, the accumulation of recessive deleterious alleles and
inbreeding depression in sexually reproducing species (Bertin et al., 2017; Szűcs et al., 2019).
Releasing laboratory populations into new environments can impose stressors that exacerbate the
negative impacts of inbreeding (Fauvergue et al., 2012; Hufbauer et al., 2013; Szűcs et al., 2017;
Szűcs et al., 2019). Introducing new genetic material to founding populations can alleviate
inbreeding, enhance adaptive potential, and produce novel genotypes that may be more suited to
new environments (Barker et al., 2019; Sinclair et al., 2019; Stewart et al., 2017; Szűcs et al.,
2017). This approach is supported by laboratory studies, which have consistently shown that
admixed populations exhibit improved establishment success, higher population growth, and
greater long-term survival (reviewed in Szűcs et al., 2014). For instance, admixture in pheasant
populations has led to the 'catapult effect,' a temporary surge in fitness in initial generations due
to heterosis, or 'hybrid vigor', enhancing the population's persistence (Drake, 2006). Yet the
long-term effects of intentional intraspecific hybridization after release into the wild are still
unclear.
Biological control agents provide a unique opportunity to use intentional hybridization to
understand how neutral genetic variation affects fitness related traits including population
performance, establishment, and persistence. Here we explore the fitness effects of intraspecific
hybridization for a recently approved biological control agent Hypena opulenta that is being
released to control invasive swallow-wort vines (Vincetoxicum spp.) in North America. Hypena
opulenta had been reared in quarantine for over a decade prior to approval for field release in the
U.S. and our laboratory population was founded with few individuals (< 20 females). We
hybridized this likely inbred laboratory population with a population established in the field and
compared the fitness of the two parental (laboratory population vs. field population) and the
6
outbred populations in the field and in the laboratory. We also assessed the molecular genetic
diversity of each population and evaluate if they are genetically distinct. We expected relatively
low yet significant genetic divergence between the laboratory and field populations and that the
outbred population would exhibit higher heterozygosity and increased fitness compared to the
laboratory population. We hypothesized that the outbred and field populations may have similar
performance in the field but that the effects of outbreeding may be more apparent in laboratory
assays reflecting gene-environment interactions. Ultimately, we expected that intraspecific
hybridization would increase neutral genetic variation and be correlated with increased fitness
under different environmental conditions.
2. Methods
2.1 Study System
Invasive swallow-wort vines were introduced to North America in the mid to late 1800s
(Cappuccino, 2002; Weed & Casagrande, 2010). Two species were introduced: pale swallow-
wort (Vincetoxicum rossicum Kelopow) and black swallow-wort (Vincetoxicum nigrum L.) as
ornamentals from Ukraine and Mediterranean countries, respectively (Ditommaso et al., 2005).
Both have spread in southeastern Canada and the northeast United States These herbaceous,
perennial vines invade a variety of habitat types including forests, crop fields, pastures, road and
trail edges, and home gardens (Ditommaso et al., 2005; Weed & Casagrande, 2010). Swallow-
wort vines outcompete native vegetation and form dense monocultures because of their
aggressive growth habits and lack of natural enemies (Ditommaso et al., 2005; DiTommaso et
al., 2018). Swallow-worts have proven difficult to control by chemical and cultural methods,
they have adverse effects on arthropod diversity (Ernst & Cappuccino, 2005), reduce grassland
7
habitat for birds (Ditommaso et al., 2005), and are an oviposition trap for monarch butterflies
(Alred et al., 2022; Casagrande & Dacey, 2007).
The biological control agent Hypena opulenta was approved for release in the United States
in 2017 (Stewart, 2017). It feeds on the leaves of swallow-wort in all five of its larval stages
before it enters the pupal stage to undergo metamorphosis, then emerging asynchronously as
adults (Weed & Casagrande, 2010). H. opulenta is a bivoltine species in Michigan, meaning it
can complete two generations per year (Alred et al., 2022). However, like many other insects,
this agent enters diapause in the winter, a state of halted development that allows it to survive
extreme conditions (Tauber et al., 1986, Weed & Casagrande, 2010). However, diapause in H.
opulenta is facultative (occurring in the pupal stage) and is mainly regulated by the photoperiod
(Jones et al., 2020; Weed & Casagrande, 2010). The point at which 50% of H. opulenta enter
diapause is known as the critical photoperiod, which happens when daylight lasts 15h and 35 min
(Jones et al., 2020; Weed & Casagrande, 2010). In Michigan, release success can be maximized
by releasing shortly before the summer solstice to increase exposure to Michigan’s maximum
photoperiod conditions (15 h 20 min; East Lansing, MI) (Alred et al., 2022; Jones et al., 2020).
In another main release location – Canada – H. opulenta was approved for release in 2013 and
field releases started in September of 2013. Establishment of H. opulenta was confirmed in 2015
in Ontario (Bourchier et al., 2019). Overwintering success in Michigan wasn’t confirmed until
the summer of 2023 and monitoring for field establishment is still ongoing. There are several
theories to explain the delay in establishment of H. opulenta for Canada and the United States.
Firstly, mismatched photoperiod between the collection site in Ukraine and release latitudes in
North America could be causing early diapause (Jones et al., 2020). Secondly, H. opulenta
appear to be very vulnerable to predation in the field in Canada (Jones et al., 2022). Early
8
diapause can cause H. opulenta to spend longer periods of time in its pupal stage increasing its
risk of predation. Thirdly, we suspect that reduced genetic variation or inbreeding could be
causing performance issues in field populations causing delayed establishment and stagnant or
reduced population growth (Szűcs et al., 2019).
2.2 Moth populations and rearing
Population background
Before shipment to Michigan, the first lab population of Hypena opulenta was created from
wild caught individuals in Donetsk, Ukraine in 2006 from four pupae and 32 larvae (Weed et al.,
2011). Population maintenance started at CABI EU-CH (Delémont, Switzerland) before
shipment to the University of Rhode Island (URI) which then started quarantine and population
rearing in 2008 (Hazlehurst et al., 2012). URI provided a larval shipment to Michigan State
University (MSU; East Lansing, MI) in June of 2018 which founded the MSU population with
18 females and 22 males.
Additional H. opulenta individuals were shipped to MSU in 2020 that originated from a field
population in Canada (Bourchier et al., 2019). This Canadian population was founded from the
same base population as the MSU population from URI but six years (or 12 generations) earlier,
in 2012. Field releases in Canada started in 2013 and resulted in overwintering success and
establishment in Ontario by May 2015 (Bourchier et al., 2019). Individuals collected in the field
in Canada in 2019 were reared for one generation in quarantine at URI and then were shipped as
pupae to MSU in May of 2020. This population was founded with 15 females and 13 males.
Reciprocal crossings between the MSU laboratory and Canadian field population were
conducted in summer 2020. The original MSU population from 2018 will be referred to as the
laboratory adapted population, while the population shipped from Canada will be called the field
9
population, and the cross between the two will be called the outbred population. The outbred
population was created through reciprocal crosses of 5 males and 3 females from the field
population (shipped May 2020) and 5 males and 8 females of the lab population. Three separate
cages were set up for reciprocal crosses. The first cage allowed five females from the lab
population to mate freely with three males from the field population. The second cage allowed
three females from the field population to freely mate with five males from the lab population.
The third cage allowed three females from the lab population to freely mate with the two males
from the field population.
Rearing and population maintenance
Both species of swallow-wort were used for rearing as H. opulenta does not discriminate
between species for oviposition or feeding (Hazlehurst et al., 2012). However, by 2021 most
moths were reared on V. rossicum due to its availability. Field collected plants of both
Vincetoxicum spp. from infestations across Michigan including Grand Rapids and several towns
in Oakland County were transplanted into 1.14 L square plastic pots using SUREMIX® all-
purpose perlite mix (Michigan Grower Products Inc., United States). Potted swallow-wort was
grown in the greenhouse at 26.6±4.9C from March through September. In the fall the potted
plants were placed outside at the MSU Entomology Farm until the following spring. In
March/April the overwintering plants were brought back in the greenhouse and repotted prior to
use in rearing in subsequent seasons. Rearing was conducted in the laboratory in mesh cages (40
x 40 x 60 cm, Restcloud). In each cage 6 potted swallow-wort plants were placed on a tray with a
0.6-1.3 cm layer potting soil. Rearing cages were kept in a laboratory at 22-25˚C with ambient
humidity under 16:8h L:D photoperiod provided by supplemental lighting (4ft, 80 Watt;
Barrina®; Zhongshan, China). Adult moths were fed with a 1:5 ratio of honey to water in 2 oz
10
cups with a cotton wick. Honey water was changed for freshness when molding was detected,
and cages were misted with distilled water daily. Potted swallow-wort was watered as needed.
Between March and August of each year, pairs of adults were released into the mesh cage
setup with no more than 5 pairs per cage. Adults oviposited onto the potted swallow-wort plants
until death. Larvae were allowed to feed on the potted swallow-wort and were supplemented
with cut stems of swallow-wort in water picks (9 x 2.5 cm) in each cage. Cut-stem swallow-wort
was sourced from potted swallow-wort early in the season when swallow-wort was not yet
available in the field but was cut from field infestations starting in late May. Larvae fed in the
mesh cages until 3rd-5th instar when they were removed to a 2.4 L round plastic mixing container
(HDX®, Home Depot; United States) with a cut hole in the lid covered with mesh for ventilation.
Containers were supplemented with swallow-wort in water picks and provided paper towels for
pupation. Containers were placed in a laboratory under 16:8h L:D lights and checked daily to
every other day for pupae. Under high volume conditions, some larvae were allowed to pupate in
mesh cages.
Pupae were collected from plastic containers and mesh cages. They were then sexed and
placed, in groups of 8 or less into 0.35 L transparent round plastic containers with vermiculite
(Vigoro®; United States) moistened with distilled water. During the continuous rearing season,
these containers of pupae were placed on the benchtop under 16:8h L:D lights until emergence.
In August, pupae containers were placed in incubators under 12:12h L:D at 20°C to induce
diapause. The temperature was lowered by 5°C each month until 5°C was reached and
maintained. The containers with pupae were remoistened once a month until removal and
placement under 16:8h L:D and 22-25˚C in early spring. This rearing protocol has been used for
this laboratory population since 2018 and was used in subsequent population crosses.
11
2.3 Laboratory Experiment
To compare the performance of the three H. opulenta populations (laboratory, field and
outbred), we tested how many pupae are produced by individual females as a proxy for
performance. Counts from the pupal stage were chosen as an indicator of performance because
survival up to this point would incur the most impact to swallow-wort. Individual females were
allowed to mate and lay eggs for 72 hours in rearing cages. One to two males were used for
mating based on availability, but the numbers were kept consistent in each block. We exclusively
used V. rossicum for egg laying which was transplanted from field sites in Oakland County, MI
to pots in the greenhouse in the spring of 2021 and 2022. The larvae were removed from the
rearing cages at the 4th and 5th instar stage and transferred to rearing containers where they
pupated (see rearing methods above). For logistical reasons these fitness experiments were
conducted at two locations (laboratory and greenhouse) in staggered time frames (temporal
blocks) over a two-year period (2021 and 2022). Overall, the number of replications (individual
females tested) for each of the three treatments (laboratory, field and outbred populations) ranged
between 68-87. Two temporal blocks were setup both in 2021 and in 2022. In 2021, the first
block was setup in the greenhouse and the second block in the laboratory. In 2022, both temporal
blocks were tested in the laboratory. In the greenhouse experiment the temperature was
25.6±4.8°C and beyond natural daylight hours artificial lights were used to ensure 16:8 L:D
photoperiod. In the laboratory the cages were kept under ambient temperature and humidity with
artificial lighting set to 16:8h L:D. The number of replications for each year and temporal block
and the location of assays are summarized in Table 1.1.
12
Table 1.1 Summary of the years, blocks, and replicates for the laboratory experiment.
Year
Block
Population
Number of
Location
Replicates
2021
2021
2021
2021
2021
2021
2022
2022
2022
2022
2022
2022
2022
2022
2022
1
1
1
2
2
2
1
1
1
2
2
2
3
3
3
Field
Lab
Outbred
Field
Lab
Outbred
Field
Lab
Outbred
Field
Lab
Outbred
Field
Lab
Outbred
4
4
4
4
4
4
39
33
37
15
20
23
6
13
19
Lab
Lab
Lab
Lab
Lab
Lab
Greenhouse
Greenhouse
Greenhouse
Lab
Lab
Lab
Lab
Lab
Lab
2.4 Field Experiment
To compare the performance of the three populations of H. opulenta under field conditions,
adults were released in cages setup in a natural infestation of pale swallow-wort at a field site in
Oakland County, MI (42.7087042, -83.5732006). Cages were square frames made of polyvinyl
13
chloride and covered with a fine white mesh that were staked to the ground over naturally
growing swallow-wort (60 x 60 x 60 cm; Gardeners Advantage). Cages were setup within ~500
square meters of infestation in early June of 2023. In each cage three types of annual flower
plugs were planted (one each of marigold, begonia, and impatiens) to provide nectar for adult
moths. There were 14 replicates for each population treatment (field, lab, and outbred) totaling
42 cages. A total of five female and five male adult moths were released in each cage on June
14th, 2023. Four of the pairs of H. opulenta adults were allowed to mate for 48 hours prior to
release under ambient laboratory conditions in small mesh cages (12 x 12 x 12, Restcloud) and
were given honey water. As more adults became available, another pair was added to each
replicate 24 hours prior to release. Adults ranged from 1 day old to 5 days old at the time of
release. Adults were allowed to lay eggs until death (1-3 weeks). Swallow-wort stem counts
within each cage were taken before release. Percent defoliation was assessed for five randomly
chosen stems in each cage once on July 10th, 2023, when fifth instar larvae were present, and
pupation had started. Percent defoliation is used as a proxy for fitness for each population
treatment. The number of 2nd generation adults emerged in each cage was checked weekly after
the start of pupation until the end of August. Adults were released from the cage upon each
monitoring date to avoid double counts later.
2.5 Molecular Genetic Diversity and Structure
Samples
Adult H. opulenta from the laboratory adapted population were saved in 2021 (n = 15), from
the field population in 2020 (n = 8) and in 2021 (n = 7) and from the outbred population in 2020
(n = 10) and 2021 (n = 10). For the outbred population the 2020 collections represent two
generations post-crossing, and the 2021 collections are third and fourth generation post-crossing.
14
The samples were stored at -80°C until shipment to the Novogene Corporation Inc. (Durham,
NC, USA).
DNA Extraction, Sequencing, and Bioinformatics
DNA extraction, library preparation, whole genome sequencing and sequence alignment
were performed by Novogene. Whole genome sequencing libraries were prepared using the
Illumina Standard 350bp DNA library. Samples were not mixed for library construction and
were checked with Qubit and real-time PCR for size detection and distribution. The libraries
were then pooled and sequenced by Novogene on NovaSeq 6000 (Paired-End 150bp).
Novogene completed a standard bioinformatic analysis including data quality control,
alignment with a reference genome and statistics of sequencing depth and coverage. The
sequencing error rate for all samples was 0.03% with the percent of Phred-scaled quality scores
greater than 30 all above 91%. Reads were aligned to the Hypena proboscidalis reference
genome available through NCBI (ilHypProb1.1; GCA_905147285.1; Boyes, et al., 2023). The
mapping rate ranged from 28.55% - 30.43% and average depth ranged from 16.10X – 21.46X.
Genome-wide Diversity
ANGSD (v.0.935) was used to calculate the global, genome-wide nuclear heterozygosity of
individual samples. This software uses the entire genome and incorporates genotype uncertainty
by using genotype likelihoods (GLs). This method is particularly useful for non-model
organisms with low to medium coverage (Sand Korneliussen et al., 2014).
For each sample a reference genome is provided as the ancestral state. We first took the site
allele frequency likelihood, which is based on individual genotype likelihoods assuming Hardy-
Weinberg equilibrium, by using the -doSaf 1 option. Filters were used since the input files are
raw alignments (BAM and FASTA files). The filter -C 50 was used to adjust mapping quality for
15
excessive mismatches. Then reads were removed if mapping quality fell below 30 (-minMapQ
30), minimum base quality score was below 20 (-minQ 20), they were flagged if above 255 (-
remove_bads), if location was not unique during mapping (-uniqueOnly), and filter for pairs of
reads with both mates mapped correctly (-only_proper_pairs). The regions used were filtered (-
rf) to include all 30 autosomes and exclude the sex chromosome. Genotype likelihoods were
calculated using -GL 1 (SAMtools method). The result was then used to obtain the maximum
likelihood estimate of the SFS by using the subprogram realSFS. Heterozygosity was then
calculated by using the data in the SFS by taking the number of heterozygous sites and dividing
it by the total number. This method was adapted from de Jager et al., 2021.
Population Structure
PCAngsd (v.1.21) was used to calculate individual allele frequencies and their covariance for
structured populations using principal component analysis (PCA) using genotype likelihoods
(Meisner and Albrechtsen, 2018). PCAngsd requires files to be input in a beagle format. BAM
files were converted to beagle files using ANGSD and the -doGlf 2, -GL 1 and -SNP_pval 1e-6
to work with sites with this p-value less than 1e-6. During conversion, the regions were again
filtered to include autosomes 1-30 for consistency. Individual allele frequencies were calculated
by simply running PCAngsd by inputting the beagle file.
2.6 Statistical Methods
All analyses were performed in R 4.3.0 (R Core Team, 2023). We used the DHARMa
package to calculate the residuals to assess best model fit for all non-parametric models. This
package assesses the model in several ways including QQ plot, the Kolmogorov-Smirnov test
(K.S. test), dispersion testing, and zero-inflation testing. Chi-squared values were provided for
mixed models using the Anova function from the car package (Zeileis & Hothorn, 2002). All
16
post-hoc pairwise comparisons were performed using the emmeans package and were adapted
for parametric tests by specifying for the Tukey adjustment (Lenth et al., 2021).
Laboratory Experiment
In this experiment, we measured if a female succeeded or failed in producing any pupae and
then the number of pupae produced by each female. Statistically, we compared the production of
pupae by population in two ways. First, we evaluated whether any pupae were produced
(hereafter “production success”) with a generalized linear mixed model with a binomial
distribution using the lme4 package (Bates et al., 2015). If no pupae were produced, then the
replicate was assigned a “0”. If one or more pupae were produced, then the replicate was
assigned a “1”. The model contained population (lab, field, outbred) as a fixed effect and
replicates nested within block nested within year as the random effect.
Second, we evaluated the number of pupae produced per female using a generalized linear
mixed model via the glmmTMB package with a Poisson distribution (Brooks et al., 2017). The
best model contained population (lab, field, outbred) as a fixed effect and replicates nested
within block nested within year as the random effect. The glmmTMB package allows for a zero-
inflation parameter to be applied to a model with a Poisson distribution. The zero-inflation
parameter was set to vary by population such that the probability that observations had structural
zeros depended on population.
Field Experiment
In this experiment, we measured if a female succeeded or failed in producing any larvae by
noting evidence of larval feeding and then measured the rate of defoliation averaged over each
stem. We also measured if any pupae succeeded or failed to produce a second generation and
how many adults were produced in the second generation. To compare performance between the
17
three treatments, we used evidence of larval feeding. If a cage exhibited no feeding, then the
replicate was assigned a “0”. If a cage exhibited any feeding, then the replicate was assigned a
“1”. A generalized linear mixed model was used with a binomial distribution with the glmmTMB
package in R (Brooks et al., 2017). We used population (field, lab, and outbred) as the fixed
effect and swallow-wort stem density as the random effect.
In addition, we compared total defoliation per stem as a proportion (between 0 and 1) using a
generalized linear mixed model with a beta distribution in the glmmTMB package in R (Brooks
et al., 2017). This package allows a zero-inflation parameter that permits zeros in the data
because beta distributions are bound between greater than zero and less than one yet is often zero
or one inflated. This bypass allowed us to set the zero-inflation parameter to vary by population.
We used population (field, lab, and outbred) as the fixed effect and swallow-wort stem density as
the random effect. One replicate in the field population was missed during data collection and
therefore n = 13 for this treatment while n = 14 for the other populations.
Adult emergence success by treatment was evaluated by assigning cages with no emergence
a “0” and one or more emergence a “1”. A generalized linear model was used, from the lme4
package, with a binomial distribution with population and emergence date as fixed effects (Bates
et al., 2015). The number of adults that emerged by each treatment with a generalized linear
model with a Poisson distribution, from the lme4 package (Bates et al., 2015). We used two fixed
effects, population and emergence date. Emergence was only recorded on two dates with only
one cage with one adult emergence at the later date. In a separate analysis, only population was
used as fixed effect and the number of emerging adults were summed for the two dates.
18
Genome-wide Diversity
Heterozygosity values were assessed for normality using a Shapiro-Wilk test then modeled
with a one-way ANOVA using the stats package (R Core Team, 2023). Population (lab, field,
outbred) was used as the fixed effect. A Q-Q plot and Levene’s test was used to assess model
assumptions from the car package (Fox and Weisberg, 2019).
3. Results
3.1 Laboratory Fitness Experiment
The probability that at least one pupa would be produced was not affected by population (χ2
= 3.55, df = 2, p =0.17). However, there was a trend of higher probability for the outbred
population producing at least one pupa (52%), compared to 40% probability for the field, and
36% probability for the lab population.
The number of pupae produced was affected by population (χ2 = 20.81, df = 2, p < 0.001)
(Fig. 1.1). The outbred population produced the most pupae (mean [95% confidence interval], n
= number of replicates, N = number of individual pupae; 34.01 [23.94, 48.31], n = 87, N = 2352)
compared to the field population (29.74 [22.85, 38.69], n = 68, N = 1160) and the lab population
(28.76 [20.03, 41.29], n = 74, N = 1369)) (Fig. 1). Effect sizes show a 14% increase in pupae
production by the outbred population when compared to the field population (pairwise
comparison: p < 0.01) and an 18% increase when compared to the lab population (pairwise
comparison: p = 0.0001). Whereas the lab and field population are not significantly different
from each other. This is reflected in the nearly doubling in pupal production in the outbred
population, despite slightly lower replication in the field and lab populations (Table 1).
19
Figure 1.1 Means and 95% confidence intervals of the number of pupae produced per female by
population (field, lab, outbred). Observations have been jittered to show individual points.
Asterisks indicate significant differences between treatments at the 0.05(*), 0.01(**), 0.001 (***)
levels.
20
3.2 Field Experiment
The probability that a cage would produce at least one larva (i.e., visible feeding damage)
was not significant by population (χ2 = 2.81, df = 2, p = 0.25) (field: p = 0.458; lab: p = 0.522;
outbred: p = 0.096). However, the outbred population had the highest probability of success
(96%), the lab population the second highest success probability (77%) and the field population
the lowest success probability (63%) in terms of replicated field cages showing any feeding
damage. Post-hoc pairwise comparisons were not significant.
Percent defoliation was affected by population (χ2 = 18.24, df = 2, p < 0.001) (Fig. 1.2). The
outbred population had the highest mean defoliation (0.17 [0.13, 0.21], mean [95% confidence
interval], n = 14) and the field population the second lowest mean (0.13 [0.08, 0.18], n = 13),
lacking significance from each other. This is likely due to the number of cages that initially
produced larvae (see results above), which was lower in the field population when compared to
the outbred. Ultimately, the field adapted population produced similar rates of defoliation to the
outbred population in cages that showed any larval feeding. The outbred population produced
significantly more defoliation than the lab population (0.04 [0.02, 0.06], n = 14) (pairwise
comparison: p < 0.001). And the field population had higher rates of defoliation when compared
to the lab population (pairwise comparison: p < 0.001). Ultimately, the number of cages that
produce larvae for each population and then the distribution of the pupal number influences the
significance of this non-linear model.
21
Figure 1.2. Mean and 95% confidence intervals of the proportion of leaf matter defoliated per
stem by population (field, lab, outbred). Observations have been jittered to show individual
points. Asterisks indicate significant differences between treatments at the 0.05(*), 0.01(**),
0.001 (***) levels.
The number of adults that emerged was affected by population (χ2 = 7.97, df = 2, p < 0.05)
(field: 0.012 [0.0006, 0.069]; mean [95% confidence interval], n = 13, N = 3) (lab: 1.24 [0.27,
22
6.29], n = 14, N = 4) (outbred: 4.02 [1.30, 17.55], n = 14, N = 13). The number of adults that
emerged depended on date (χ2 = 19.79, df = 1, p < 0.001). The outbred population produced 3.3
times more adults than the field population (pairwise comparison: p = 0.076) and 2.3 times more
adults than the lab population (pairwise comparison: p = 0.098) thus neither pairwise
comparisons were significant. The two dates that produced adults were significantly different as
97% of the adults emerged by August 8th, 2023, compared to August 16 (pairwise comparison: p
< 0.01). When dates were combined to assess the number of emerged adults the results for
population differences were similar, and the pairwise comparison between the outbred
population and the lab population became marginally significant (pairwise comparison: p =
0.057).
The probability that at least one adult would emerge was not affected by population (χ2 =
4.56, df = 2, p > 0.05). The probability of emergence depended on date (z = χ2 = 15.10, df = 1, p
< 0.001). The probability that a second generation of adults would emerge was 1% for the field
and the lab populations, and 5% for the outbred population (pairwise comparisons were not
significant). Adults were 22% more likely to emerge by August 8th than August 16th (pairwise
comparison: p < 0.01).
3.3 Molecular Genetic Diversity
Genome-wide Diversity
The mean global heterozygosity of the outbred population was significantly higher than the
field and the lab populations’ (F = 16.59, df = 2, p < 0.0001) (Fig. 1.3). The heterozygosity of the
outbred population was 12% higher than the field (pairwise comparison: p = 0.0001) and 13%
higher than the lab populations (pairwise comparison: p < 0.0001). Heterozygosity of the field,
and the lab population did not differ (p = 0.968).
23
Figure 1.3. Mean and 95% confidence interval of genome-wide heterozygosity by population
(field, lab, outbred). Observations have been jittered to show individual points. Asterisks indicate
significant differences between treatments at the 0.05(*), 0.01(**), 0.001 (***) levels.
Population Structure
The population structure of H. opulenta based on the principal component analysis of allele
frequencies displays three distinct populations (Fig. 1.4). Two individuals from the field
24
population deviate from the group by clustering more closely with the lab population. The
outbred population is inversely correlated with the field population. The lab population is more
closely correlated to the outbred population yet is still distinct as its first principal component is
positive. Populations seem to be primarily divided by the first principal component, however
some individuals in each population are more similar, falling closer to zero.
Figure 1.4. Population structure using individual allele frequencies of Hypena opulenta from the
field, the lab, and outbred. The principal component analysis (PCA) plot is the percentage of
variation explained by each principal component.
25
4. Discussion
This study shows that genetic structure can develop among recently isolated populations of
the same species and that outcrossing between populations separated by 5 years in the field, or
approximately 10 generations, can lead to fitness benefits in the outbred population. Notably,
intraspecific hybridization between the lab and field populations led to increased heterozygosity
and enhanced multiple fitness metrics in the hybrid offspring. Our findings indicate that neutral
genetic diversity is a reliable predictor of performance across various environments.
In our laboratory experiments, we observed that outbred females consistently produced more
pupae, indicating enhanced fecundity and/or better survival of larvae to the pupal stage. This
higher fecundity or larval survival in outbred populations could have been due to hybridization
increasing heterozygosity and the masking of deleterious alleles compared to the laboratory
population. The field population, despite its successful establishment in natural settings,
exhibited lower fitness than the laboratory adapted population when tested in the laboratory.
Considering heterozygosity alone these results may be unexpected. However, the rearing and
release history of the two populations can explain the observed better performance of the lab
population. The field population has likely undergone similar pre-release challenges as our lab
population, including adaptation to culture and heterozygosity loss during mass-rearing.
Moreover, this population faced novel biotic and abiotic conditions upon field release and novel
selection pressures. Adaptation to the field conditions could have led to further reductions in
heterozygosity and low population sizes if there was a hard selection event that resulted in high
mortality of suboptimal or unsuitable genotypes in the novel environment. In contrast, the lab
population has been reared continuously in the lab for over 45 generations and likely adapted to
the rearing conditions, experiencing genetic drift and inbreeding. Thus, it is not entirely
26
surprising that under laboratory conditions the lab population would outperform the field
population. In the lab experiment, we tested F3 to F7 hybrids and in the field experiment we
tested F8 to F9 hybrids. Despite testing later generation hybrids where the effects of heterosis are
less pronounced, we still saw clear fitness benefits of hybridization. This indicates that the
positive effects of hybridization can be lasting in this system, as seen in others (Szűcs et al.,
2014; Wagner et al., 2017). Over time, continuous lab conditions are likely to erode genetic
variation in the outbred population. This raises the question for biocontrol practitioners of what
the ideal timing for field releases of laboratory created hybrids may be.
In a field environment, the field population outperformed the lab population which is in line
with expectations and shows how testing long-term laboratory reared populations in the lab
environment could lead to artifacts that do not hold up under natural conditions. Overall, the
outbred population had the best performance in the field when looking at the metrics of
successful egg laying with subsequent larval feeding and the production of second-generation
adults. However, defoliation rates were similar between the outbred and field populations despite
lower heterozygosity of the latter suggesting that adaptation to field conditions can yield similar
benefits as hybridization in some metrics. These observations imply a greater potential for
sustained biocontrol effectiveness of the outbred population in field conditions. Limited literature
exists on field performance of hybrids that is coupled with genetic data (Abbott et al., 2016).
Most studies focus on naturally hybridizing populations, as seen in the case of the weed
biological control agent Longitarsus jacobaeae (Szűcs et al., 2019). This study reported
increased densities and impact of hybridized ragwort flea beetle populations on tansy ragwort
compared to parental populations. Conversely, much of the research in this area has concentrated
on invasive species, such as the naturally admixed populations of the Caribbean lizard Anolis
27
sagrei, where increased body size of the hybrids correlated with colonization success (Kolbe et
al., 2007). Similar patterns of genetic admixture contributing to the success of several invasive
plant species, including a salt marsh grass (Spartina alterniflora), a monkeyflower (Mimulus
guttatus), and the yellow star-thistle (Centaurea solstitialis), have been observed (Qiao et al.,
2019; Li et al., 2018; Barker et al., 2017). To our knowledge, no published studies have moved
beyond correlative experiments to intentionally perform intraspecific hybridization and assess
hybrid performance in the field. Our study fills this gap by experimentally releasing controlled
hybrid populations to discern performance differences, thus eliminating confounding factors like
founding size variability and number of introduction events (Baker et al., 2003; Dlugosch et al.,
2015; Szűcs et al., 2017). Genomic analysis further enhances our understanding of these
experiments.
Our heterozygosity analysis was instrumental in interpreting the laboratory and field
experiment results, demonstrating the benefits of increased heterozygosity in hybrids whose
parental generations were recently isolated. The extent of divergence between populations plays
a critical role in determining hybridization outcomes—either positive (heterosis) or negative
(outbreeding depression). In Drosophila littoralis studies, for instance, hybrids with low
divergence (seven generations) exhibited heterosis, which dissipated after another seven
generations of rearing. In contrast, outbreeding between more divergent lines (fifteen
generations) showed no advantage over inbred lines and even displayed lower fecundity than
control lines, likely indicating the presence of outbreeding depression (Pekkala et al., 2012). In
our case, however, we did not find such results suggesting that the level of divergence between
H. opulenta populations was not large enough to cause outbreeding depression, yet substantial
enough to induce higher performance. Variability in genetic variation within laboratory
28
populations also showed it can lead to differences in population structure. For example, Aedes
aegypti strains, isolated by being reared in different labs, showed significant genetic
differentiation, suggesting that minor environmental differences can prompt genetic divergence
(Gloria-Soria et al., 2019). In another example, hybridization of two inbred Drosophila
melanogaster lines resulted in breeding success improvements (Heber et al., 2012). Even though
the parental populations of our hybrid population experienced different environmental
conditions, they both had reduced heterozygosity and produced hybrids with increased
performance like in the previous example. Our findings contribute to the understanding that
hybridization of recently isolated populations can create distinct genetic structures and
heterozygosity changes, correlating with fitness differences. This has implications for managing
captive or inbred populations, which might be overlooked for intentional hybridization. Our
study also demonstrates that, despite similar genetic variation rates, recently isolated populations
can differ significantly. We found that neutral genetic variation effectively predicts performance,
underscoring the value of neutral genetic diversity. However, the predictive power of genome-
wide diversity for endangered species on the IUCN’s Red List is limited, possibly due to
complexities like gene-environment interactions and epistasis, suggesting that functional genetic
diversity might be more critical (Teixeira & Huber, 2021). DeWoody et al. (2021) emphasized
the longstanding association between neutral genetic diversity and fitness, noting reduced genetic
diversity in endangered species compared to non-threatened ones. They argue that neutral
diversity is unquestionably linked to functional diversity and improvements can be made by
integrating complex factors like demographic history and ecological relationships into these
considerations. Our study supports the notion that neutral genetic diversity significantly predicts
fitness, especially when gene-environment interactions are considered. Future studies should
29
further explore this by directly manipulating genetic variation in different organisms with varied
histories, contributing to the body of evidence supporting the role of neutral genetic diversity.
In the realms of conservation and restoration, the critical role of genetic diversity has been
well-established (DeWoody et al., 2021; Chapman et al., 2009; Hufford et al., 2003). However, it
is a relatively unexplored avenue in biological control. The success of biocontrol programs is
hindered by low establishment and limited control success that may stem from genetic problems
that biocontrol agents experience prior to field releases, as detailed above. While we demonstrate
here that crossing of recently isolated populations can give a fitness boost that also translates to
better field performance, the use of intentional hybridization is still rare in the practice of
biocontrol (Szucs et al. 2019). Previously, incidences of unintentional hybridization have led to
fitness boosts in biocontrol populations. For example, it is postulated post-release that population
admixture in the predatory ladybird Cryptolaemus montrouzieri increased establishment based
on later laboratory experiments (Li et al., 2018). This aversion to genetic manipulation of
biocontrol agents partly stems from fears that it might lead to changes in host specificity of
agents, compromising the safety of biocontrol programs (Hoffman et al., 2002; Mathenge et al.,
2010; Szűcs et al., 2019). However, there is no evidence that intraspecific hybridization would
result in either host shifts or host range expansion and less than 1% of species released for weed
biocontrol exhibit non-target effects (Clark et al., 2023; Dieckhoff et al., 2017; Hinz et al., 2019;
Szűcs et al., 2019). On the other hand, it can increase establishment success and persistence of
populations in the field, as shown here. These insights have significant implications not only for
biological control programs aimed at boosting performance of biocontrol agents but also for
other fields grappling with the challenges of captive breeding and release. Particularly,
populations that are endangered, extirpated, or extinct in the wild, often characterized by small
30
numbers and limited gene flow, could substantially benefit from strategic intraspecific
hybridization. Currently, species are encountering an array of increasing challenges, such as
biological invasions, habitat degradation, fragmentation, and climate change, all necessitating
multifaceted and interdisciplinary approaches for resolution. These challenges are leading to a
rise in isolated and diminishing native species populations (Dawson et al., 2011; Urban, 2015).
In scenarios with restricted gene flow, deliberate hybridization between populations could be a
vital strategy to avert extinction. The rapidly evolving climate conditions are likely to necessitate
more frequent species reintroductions and adaptation to new habitats. Therefore, ongoing
research into the role of genetic diversity under both laboratory and field conditions remains
crucial. Furthermore, these insights could shed light on the factors that underlie the success of
invasive species. From biocontrol to conservation and restoration, the continued study of the role
of genetic variation is invaluable.
31
LITERATURE CITED
Abbott, R. J., Barton, N. H., & Good, J. M. 2016. Genomics of hybridization and its evolutionary
consequences. Molecular Ecology, 25(11), 2325–2332.
https://doi.org/10.1111/mec.13685
Alred, B., Haan, N., Landis, D. A., & Szűcs, M. 2022. Does the Presence of the Biological
Control Agent, Hypena opulenta (Lepidoptera: Erebidae) on Swallow-Worts Deter
Monarch Oviposition? Environmental Entomology, 51(1), 77–82.
https://doi.org/10.1093/ee/nvab121
Arct, A., Sudyka, J., Podmokła, E., Drobniak, S. M., Gustafsson, L., & Cichoń, M. 2017.
Heterozygosity–fitness correlations in blue tit nestlings (Cyanistis caeruleus) under
contrasting rearing conditions. Evolutionary Ecology, 31(5), 803–814.
https://doi.org/10.1007/s10682-017-9911-6
Audet, C. L., Wilson, C. C., & Pitcher, T. E. 2017. Effects of intraspecific hybridisation between
two hatchery-reared strains of Atlantic salmon, Salmo salar, on juvenile survival and
fitness-related traits. Fisheries Management and Ecology, 24(1), 1–9.
https://doi.org/10.1111/fme.12195
Baker, D. A., Loxdale, H. D., & Edwards, O. R. 2003. Genetic variation and founder effects in
the parasitoid wasp, Diaeretiella rapae (M’intosh) (Hymenoptera: Braconidae:
Aphidiidae), affecting its potential as a biological control agent. Molecular Ecology,
12(12), 3303–3311. https://doi.org/10.1046/j.1365-294X.2003.02010.x
Barker, B. S., Andonian, K., Swope, S. M., Luster, D. G., & Dlugosch, K. M. 2017. Population
genomic analyses reveal a history of range expansion and trait evolution across the native
and invaded range of yellow starthistle (Centaurea solstitialis). Molecular Ecology,
26(4), 1131–1147. https://doi.org/10.1111/mec.13998
Barker, B. S., Cocio, J. E., Anderson, S. R., Braasch, J. E., Cang, F. A., Gillette, H. D., &
Dlugosch, K. M. 2019. Potential limits to the benefits of admixture during biological
invasion. Molecular Ecology, 28(1), 100–113. https://doi.org/10.1111/mec.14958
Bates, D., Maechler, M., Bolker, B, Walker, S. 2015. Fitting Linear Mixed-Effects Models Using
lme4. Journal of Statistical Software, 67(1), 1-48. doi:10.18637/jss.v067.i01.
Bell, D. A., Robinson, Z. L., Funk, W. C., Fitzpatrick, S. W., Allendorf, F. W., Tallmon, D. A., &
Whiteley, A. R. 2019. The Exciting Potential and Remaining Uncertainties of Genetic
Rescue. Trends in Ecology and Evolution, 34(12), 1070–1079.
https://doi.org/10.1016/j.tree.2019.06.006
Bertin, A., Pavinato, V. A. C., & Parra, J. R. P. 2017. Fitness-related changes in laboratory
populations of the egg parasitoid Trichogramma galloi and the implications of rearing on
factitious hosts. BioControl, 62(4), 435–444. https://doi.org/10.1007/s10526-017-9795-y
32
Bjornson, S. 2018. The 2nd International Conference ‘Insects to Feed the World’ (IFW 2018).
Journal of Insects as Food and Feed, 4(5), 1–112. https://doi.org/10.3920/jiff2018.s1
Bourchier, R. S., Cappuccino, N., Rochette, A., des Rivières, J., Smith, S. M., Tewksbury, L., &
Casagrande, R. 2019. Establishment of Hypena opulenta (Lepidoptera: Erebidae) on
Vincetoxicum rossicum in Ontario, Canada. Biocontrol Science and Technology, 29(9),
917–923. https://doi.org/10.1080/09583157.2019.1608511
Brodeur, J., Abram, P. K., Heimpel, G. E., & Messing, R. H. 2018. Trends in biological control:
public interest, international networking and research direction. BioControl, 63(1), 11-26.
https://doi.org/10.1007/s10526-017-9850-8
Brooks, M. E., Kristensen, K., van Benthem, K. J., Magnusson, A., Berg, C. W., Nielsen, A.,
Skaug, H. J., Maechler, M., and Bolker, B. M. 2017. glmmTMB Balances Speed and
Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. The
R Journal, 9(2), 378-400. doi: 10.32614/RJ-2017-066.
Cappuccino, C. 2002. Spread of the Invasive Alien Vine Vincetoxicum rossicum: Tradeoffs
Between Seed Dispersability and Seed Quality. https://doi.org/10.1674/0003
Chapman, J. R., Nakagawa, S., Coltman, D. W., Slate, J., & Sheldon, B. C. 2009. A quantitative
review of heterozygosity-fitness correlations in animal populations. Molecular Ecology,
18(13), 2746–2765. https://doi.org/10.1111/j.1365-294X.2009.04247.x
Chen Y, Liu Z, Régnière J, Vasseur L, Lin J, Huang S, Ke F, Chen S, Li J, Huang J, Gurr GM,
You M, You S. 2021. Large-scale genome-wide study reveals climate adaptive variability
in a cosmopolitan pest. Nat Commun., 12(1):7206. doi: 10.1038/s41467-021-27510-2.
PMID: 34893609; PMCID: PMC8664911.
Clark, E. I., Bitume, E. v., Bean, D. W., Stahlke, A. R., Hohenlohe, P. A., & Hufbauer, R. A.
(2022). Evolution of reproductive life-history and dispersal traits during the range
expansion of a biological control agent. Evolutionary Applications.
https://doi.org/10.1111/eva.13502
Cock, M. J. W., Murphy, S. T., Kairo, M. T. K., Thompson, E., Murphy, R. J., & Francis, A. W.
2016. Trends in the classical biological control of insect pests by insects: an update of the
BIOCAT database. BioControl, 61(4), 349-363. https://doi.org/10.1007/s10526-016-
9726-3
Colautti, R. I., Alexander, J. M., Dlugosch, K. M., Keller, S. R., & Sultan, S. E. 2017. Invasions
and extinctions through the looking glass of evolutionary ecology. Philosophical
Transactions of the Royal Society B: Biological Sciences, 372(1712).
https://doi.org/10.1098/rstb.2016.0031
Dawson, Terence P., et al. 2011. Beyond Predictions: Biodiversity Conservation in a Changing
Climate. Science, 332, 53-58. DOI:10.1126/science.1200303
33
de Jager, D., Glanzmann, B., Möller, M., Hoal, E., van Helden, P., Harper, C., & Bloomer, P.
2021. High diversity, inbreeding and a dynamic Pleistocene demographic history revealed
by African buffalo genomes. Scientific Reports, 11(1). https://doi.org/10.1038/s41598-
021-83823-8
Délye C, Jasieniuk M, Le Corre V. 2013. Deciphering the evolution of herbicide resistance in
weeds. Trends Genet., 29(11):649-58. doi: 10.1016/j.tig.2013.06.001. Epub 2013 Jul 2.
PMID: 23830583.
DeWoody, J. A., Harder, A. M., Mathur, S., & Willoughby, J. R. 2021. The long-standing
significance of genetic diversity in conservation. Molecular Ecology, 30(17
), 4147-4154. https://doi.org/10.1111/mec.16051
Ditommaso, A., Lawlor, F. M., & Darbyshire, S. J. 2005. The Biology of Invasive Alien Plants in
Canada. 2. Cynanchum rossicum (Kleopow) Borhidi [= Vincetoxicum rossicum
(Kleopow) Barbar.] and Cynanchum louiseae (L.) Kartesz & Gandhi [= Vincetoxicum
nigrum (L.) Moench]. Canadian Journal of Plant Science, 243-263.
DiTommaso, A., Stokes, C. A., Cordeau, S., Milbrath, L. R., & Whitlow, T. H. 2018. Seed-
Dispersal Ability of the Invasive Perennial Vines Vincetoxicum nigrum and Vincetoxicum
rossicum. Invasive Plant Science and Management, 11(1), 10–19.
https://doi.org/10.1017/inp.2018.8
Dlugosch, K. M., Anderson, S. R., Braasch, J., Cang, F. A., & Gillette, H. D. 2015. The devil is
in the details: Genetic variation in introduced populations and its contributions to
invasion. Molecular Ecology, 24(9), 2095–2111. https://doi.org/10.1111/mec.13183
Drake, J. M. 2006. Heterosis, the catapult effect and establishment success of a colonizing bird.
Biol. Lett, 2, 304–307. https://doi.org/10.1098/rsbl.2006.0459
Edmands, S. 1999. Heterosis and outbreeding depression in interpopulation crosses spanning a
wide range of divergence. Evolution, 53(6), 1757–1768.
https://academic.oup.com/evolut/article/53/6/1757/6757678
Fauvergue, X., Vercken, E., Malausa, T., & Hufbauer, R. A. 2012. The biology of small,
introduced populations, with special reference to biological control. Evolutionary
Applications, 5(5), 424–443. https://doi.org/10.1111/j.1752-4571.2012.00272.x
Fox J, Weisberg S. 2019. _An R Companion to Applied Regression_, Third edition. Sage,
Thousand Oaks CA. .
Frankham, R., Ballou, J. D., Eldridge, M. D. B., Lacy, R. C., Ralls, K., Dudash, M. R., &
Fenster, C. B. 2011. Predicting the Probability of Outbreeding Depression. Conservation
Biology, 25(3), 465–475. https://doi.org/10.1111/j.1523-1739.2011.01662.x
34
Gloria-Soria, A., Soghigian, J., Kellner, D., Powell, J. R. 2019. Genetic Diversity of Laboratory
Strains and Implications for Research: The case of Aedes aegypti. PLoS Neglected
Tropical Diseases, 13(12), 1-17.
Hansson, B., & Westerberg, L. 2002. On the correlation between heterozygosity and fitness in
natural populations. Molecular Ecology, 11(12), 2467–2474.
https://doi.org/10.1046/j.1365-294X.2002.01644.x
Hartig F. 2022. _DHARMa: Residual Diagnostics for Hierarchical (Multi-Level /Mixed)
Regression Models_. R package version 0.4.6,b.
Hazlehurst, A. F., Weed, A. S., Tewksbury, L., & Casagrande, R. A. 2012. Host specificity of
Hypena opulenta: A potential biological control agent of Vincetoxicum in North America.
Environmental Entomology, 41(4), 841–848. https://doi.org/10.1603/EN12093
Heimpel, G.E. & M.J. Cock. 2018. Shifting paradigms in the history of classical biological
control. BioControl, 63:27-37
Hopper K, Roush R & Powell W. 1993. Management of genetics of biological-control
introductions. Annual Review of Entomology 38:27 – 51.
Hufbauer, R. A., Rutschmann, A., Serrate, B., Vermeil de Conchard, H., & Facon, B. 2013. Role
of propagule pressure in colonization success: Disentangling the relative importance of
demographic, genetic and habitat effects. Journal of Evolutionary Biology, 26(8), 1691–
1699. https://doi.org/10.1111/jeb.12167
Hufford, K. M., & Mazer, S. J. 2003. Plant ecotypes: Genetic differentiation in the age of
ecological restoration. Trends in Ecology and Evolution, 18(3), 147-155. Elsevier Ltd.
https://doi.org/10.1016/S0169-5347(03)00002-8
Jones, I. M., Seehausen, M. L., Bourchier, R. S., & Smith, S. M. 2020. The Effects of
Photoperiod on Diapause Induction in Hypena opulenta (Lepidoptera: Erebidae), a
Biological Control Agent Against Invasive Swallow-Worts in North America.
Environmental Entomology, 49(3), 580–585. https://doi.org/10.1093/ee/nvaa030
Jones, I. M., Smith, S. M., & Bourchier, R. S. 2022. Assessment of pupal mortality in Hypena
opulenta: An obstacle for establishment of a classical biological control agent against
invasive swallow-worts. Ecological Entomology, 47(4), 527–535.
https://doi.org/10.1111/een.13137
Kimura, M. 1968. Evolutionary Rate at the Molecular Level. Nature, 217, 624–626.
https://doi.org/10.1038/217624a0
35
Kolbe, J. J., Larson, A., & Losos, J. B. 2007. Differential admixture shapes morphological
variation among invasive populations of the lizard Anolis sagrei. Molecular Ecology,
16(8), 1579–1591. https://doi.org/10.1111/j.1365-294X.2006.03135.x
Korneliussen, T. S., Albrechtsen, A., & Nielsen, R. 2014. ANGSD: Analysis of Next Generation
Sequencing Data. BMC Bioinformatics, 15(356), 1–13.
http://www.biomedcentral.com/1471-2105/15/356
Lenth, R., Singmann, H., Love, J., Buerkner, P., & Herve, M. 2021. Emmeans: Estimated
marginal means, aka least-squares means. R package version 1 (2018).
Li, Y., Stift, M., & van Kleunen, M. 2018. Admixture increases performance of an invasive plant
beyond first-generation heterosis. Journal of Ecology, 106(4), 1595–1606.
https://doi.org/10.1111/1365-2745.12926
Liu, C., Jackson, L.V., Hutchings, SJ. et al. 2020. A holistic approach in herbicide resistance
research and management: from resistance detection to sustainable weed control. Sci Rep
10, 20741. https://doi.org/10.1038/s41598-020-77649-z
Meisner, J., & Albrechtsen, A. 2018. Inferring population structure and admixture proportions in
low-depth NGS data. Genetics, 210(2), 719–731.
https://doi.org/10.1534/genetics.118.301336
Messing, R., & Brodeur, J. 2018. Current challenges to the implementation of classical biological
control. BioControl, 63(1), 1–9. https://doi.org/10.1007/s10526-017-9862-4
Miller, J. M., & Coltman, D. W. 2014. Assessment of identity disequilibrium and its relation to
empirical heterozygosity fitness correlations: A meta-analysis. Molecular Ecology, 23(8),
1899–1909. https://doi.org/10.1111/mec.12707
Orr, H. A. 2009. Fitness and its role in evolutionary genetics. Nature Reviews Genetics, 10(8),
531–539. https://doi.org/10.1038/nrg2603
Pekkala, N., E. Knott, K., Kotiaho, J. S., Nissinen, K., & Puurtinen, M. 2012. The benefits of
interpopulation hybridization diminish with increasing divergence of small populations.
Journal of Evolutionary Biology, 25(11), 2181–2193. https://doi.org/10.1111/j.1420-
9101.2012.02594.x
Qiao, H., Lui, W., Zhang, Y., & Zhang Yuan-Ye. 2019. Genetic admixture accelerates invasion
via provisioning rapid adaptive evolution. Molecular Ecology, 28, 4012–4027.
R Core Team. 2023. _R: A Language and Environment for Statistical Computing_. R Foundation
for Statistical Computing, Vienna, Austria. .
Sakai, A. K., Allendorf, F. W., Holt, J. S., Lodge, D. M., Molofsky, J., With, K. A., Baughman,
S., Cabin, R. J., Cohen, J. E., Ellstrand, N. C., Mccauley, D. E., O’neil, P., Parker, I. M.,
36
Thompson, J. N., & Weller, S. G. (2001). The Population Biology of Invasive Species.
Annul. Rev. Ecol. Syst, 32, 305–332.
Schwarzländer, M., Hinz, H. L., Winston, R. L., & Day, M. D. 2018. Biological control of
weeds: an analysis of introductions, rates of establishment and estimates of success,
worldwide. BioControl, 63(3). https://doi.org/10.1007/s10526-018-9890-8
Sinclair, J. S., Arnott, S. E., Millette, K. L., & Cristescu, M. E. 2019. Benefits of increased
colonist quantity and genetic diversity for colonization depend on colonist identity.
Oikos, 128(12), 1761–1771. https://doi.org/10.1111/oik.06308
Stewart, C. D. 2017. Field release of the leaf-feeding moth, Hypena opulenta (Christoph)
(Lepidoptera: Noctuidae), for classical biological control of swallow-worts, Vincetoxicum
nigrum. USDA APHIS Final Environmental Assessment,
http://www.ascr.usda.gov/complaint_filing_file.html
Stewart, G. S., Morris, M. R., Genis, A. B., Szűcs, M., Melbourne, B. A., Tavener, S. J., &
Hufbauer, R. A. 2017. The power of evolutionary rescue is constrained by genetic load.
Evolutionary Applications, 10(7), 731–741. https://doi.org/10.1111/eva.12489
Neale, D. B., & Wheeler, N. C. 2019. The Conifers: Genomes, Variation and Evolution. Springer
Nature Switzerland.
Szűcs, M., Melbourne, B. A., Tuff, T., Weiss-Lehman, C., & Hufbauer, R. A. 2017. Genetic and
demographic founder effects have long-term fitness consequences for colonising
populations. Ecology Letters, 20(4), 436–444. https://doi.org/10.1111/ele.12743
Szűcs, M., Salerno, P. E., Teller, B. J., Schaffner, U., Littlefield, J. L., & Hufbauer, R. A. 2019.
The effects of agent hybridization on the efficacy of biological control of tansy ragwort at
high elevations. Evolutionary Applications, 12(3), 470–481.
https://doi.org/10.1111/eva.12726
Szűcs, M., Melbourne, B. A., Tuff, T., & Hufbauer, R. A. 2014. The roles of demography and
genetics in the early stages of colonization. Proceedings of the Royal Society B, 281.
https://doi.org/10.1098/rspb.2014.1073
Tauber, M. J., Tauber, C. A., & Masaki, S. (1986). Seasonal adaptations of insects. Oxford
University Press.
Teixeira, J. C., & Huber, C. D. 2021. The inflated significance of neutral genetic diversity in
conservation genetics. Proceedings of the National Academy of Sciences of the United
States of America, 118(10). https://doi.org/10.1073/PNAS.2015096118/-
/DCSUPPLEMENTAL
Urban, Mark C. 2015. Accelerating extinction risk from climate change. Science, 348, 571-573.
DOI:10.1126/science.aaa4984
37
Wagner, N. K., Ochocki, B. M., Crawford, K. M., Compagnoni, A., & Miller, T. E. X. 2017.
Genetic mixture of multiple source populations accelerates invasive range expansion.
Journal of Animal Ecology, 86(1), 21–34. https://doi.org/10.1111/1365-2656.12567
Weed, A. S., & Casagrande, R. A. 2010. Biology and larval feeding impact of Hypena opulenta
(Christoph) (Lepidoptera: Noctuidae): A potential biological control agent for
Vincetoxicum nigrum and V. rossicum. Biological Control, 53(2), 214–222.
https://doi.org/10.1016/j.biocontrol.2009.12.004
Weed, A. S., Gassmann, A., Leroux, A. M., & Casagrande, R. A. 2011. Performance of potential
European biological control agents of Vincetoxicum spp. with notes on their distribution.
Journal of Applied Entomology, 135(9), 700–713. https://doi.org/10.1111/j.1439-
0418.2010.01594.x
X. Wei, J. Zhang.2018. The optimal mating distance resulting from heterosis and genetic
incompatibility. Sci. Adv. 4, eaau5518.
Whiteley, A. R., Fitzpatrick, S. W., Funk, W. C., & Tallmon, D. A. 2015. Genetic rescue to the
rescue. Trends in Ecology and Evolution, 30(1), 42-49. Elsevier Ltd.
https://doi.org/10.1016/j.tree.2014.10.009
Zeileis A, Hothorn T. 2002. Diagnostic checking in regression relationships. R News, 2(3):7–10.
URL https://CRAN.R-project.org/doc/Rnews/.
38
CHAPTER 2:
Restoring swallow-wort (Vincetoxicum spp.) invaded woodlands: swallow-wort control and
native seeding diversifies plant communities
Abstract
Invasive plant species pose significant threats to ecosystem integrity and biodiversity,
particularly in forested regions. This study evaluates the efficacy of restoration strategies in
swallow-wort invaded woodlands in Michigan. Three different management strategies were
applied to three swallow-wort infested sites: defoliation of swallow-wort to simulate feeding by a
biological control agent, manual removal of swallow-wort, native seeding and some combination
of these treatments over a 3-year period. Plant cover, species richness, species diversity, and
swallow-wort stem densities were monitored for 3 years. While all treatments reduced swallow-
wort cover, the combination of manual removal and native seeding most significantly reduced
swallow-wort cover and density and enhanced native species richness and diversity, although
outcomes varied by site. Our results suggest that integrating defoliation with the addition of
native seeds emerges as the second most effective strategy, having an impact on swallow-wort
stem density. Given the laborious demands of manual removal, this discovery holds significant
relevance for the feasibility of applying biological control with native species reintroduction in
practical settings. At certain locations, treatments involving only the addition of native seeds
proved effective in curbing swallow-wort stem density. Defoliation by itself, that served to
mimic feeding by a moth biological control agent, did not enhance native species presence. For
effective swallow-wort suppression and to boost native species, proactive native revegetation is
essential. When combined with biological control, this approach offers a less labor-intensive
method for managing swallow-wort, particularly for large infestations.
39
1. Introduction
Invasive plant species present a formidable challenge to the integrity and resilience of
ecosystems globally, often outcompeting native flora, altering habitats, and diminishing
biodiversity (Huston et al., 1994; Levine et al., 2003; Vitousek et al., 1997). The traditional
arsenal against such invasions—mechanical removal, chemical applications, and prescribed
burns—while potentially effective, demands substantial labor, incurs high costs, and risks
adverse ecological impacts (Pile Knapp et al., 2023, Weidlich et al., 2019). Furthermore,
attempts to control invasive plants often result in mixed success (Kettenring & Adams, 2011).
Combining multiple management strategies might enhance efficacy and produce synergistic
effects while mitigating the negative consequences of individual techniques (reviewed in Lake &
Minteer, 2018). However, the success of these combinations likely varies, and the role of native
revegetation in conjunction with active removal of the invaders remains understudied
(Kettenring & Adams, 2011; Schuster et al., 2018).
The majority of literature in invasive species control primarily focuses on a few select
methods. According to a systematic global review by Weidlich et al. (2020), non-chemical
controls were employed in 57.7% of cases, although countries with higher HDI (Human
Development Index) reduced this number in preference for herbicide. The most prevalent
methods of non-chemical control globally are prescribed fire and mowing, while hand pulling
was done in just 12.3% of cases and cutting in 9.7% of cases. Insects and clipping were
employed at less than 2.5% each. This is contrasted with chemical controls, accounting for
42.3% of interventions globally, where glyphosate was the most used herbicide. In the biome of
interest for this study – eastern North American forests – herbicides are used in 64% of control
40
cases emphasizing the need for effective, alternative methods within this biome (Pile Knapp et
al., 2023).
The frequent use of non-chemical methods aligns with efforts to mitigate potential negative
impacts associated with chemical controls, such as environmental pollution and health risks to
humans (Weidlich et al., 2020). Such non-chemical methods include biological control, the
intentional introduction of natural enemies to support long term control of exotic pests (Heimpel
& Mills, 2017). However, the regulatory capabilities of biological control agents are frequently
surpassed due to the robustness of invasive species—a result of their swift growth, prolific
reproduction, and ability to drastically alter soil chemistry and hydrology—underscoring the
necessity for innovative management strategies (Blossey, 1999; Cuddington, 2011). Currently
published research underscores the value of integrating biological control with native plant
seeding to bolster native plant richness and diversity, but examples in different systems that span
multiple years are still limited (Cutting & Hough-Goldstein, 2013; Lake et al., 2014). According
to a review by Lake & Minteer (2018). The few studies (n = Only 55) that experiment with
biological control combined with other management techniques for invasive weed control mostly
use herbicides in combination with biological control (36.4%) and biological control together
with plant competition (23.6%). Mechanical tactics were the least studied (3.6%) and approaches
that deployed more than two control methods were minimal (16.4%). The efficacy of such
integrated interventions varies across systems and treatment. For instance, the integration of
biological control with native revegetation shows promise in enhancing native plant species
richness and diversity in mile-a-minute (Persicaria perfoliate) invaded systems (Cutting &
Hough-Goldstein, 2013, Lake et al., 2014). On the contrary, herbicides can alter the dynamics of
and be toxic to a biological control agent as was in the case of the biocontrol agent moth
41
Wheeleria spilodactylus for the invasive weed Marrubium vulgare (Ainsworth, 2003).
Understanding these nuances may be particularly important in scenarios where biological control
agents alone are insufficient to suppress invasive species significantly and local native
propagules are scarce.
Mechanical methods are often selected when herbicide use is inappropriate or undesirable,
although these can be labor-intensive and costly. The effectiveness of these treatments can
depend on timing, with certain phenological stages of the plant being more susceptible to control
measures (Pile Knapp et al., 2023). Manual and mechanical controls, such as stem girdling,
clipping, uprooting, and mowing, are frequently utilized due to their direct removal of biomass.
However, the absence of complete eradication of below-ground parts may lead to regrowth,
indicating a need for follow-up interventions (Pile Knapp et al., 2023). Mechanical treatments
targeting susceptible phenological stages could amplify their effectiveness, such as mowing after
midsummer to reduce seed production of invasive grasses (Pile Knapp et al., 2023). In pale
swallow-wort (Vincetoxicum rossicum (Kleopow) Barbar), manual removal showed that this
approach significantly reduced the target weed but did not increase native plant diversity or
cover, reaffirming the importance of native revegetation (Collings et al., 2023).
Native revegetation efforts play a crucial role in the restoration process, particularly in the
context of controlling invasive species. These efforts not only aim to suppress the invasive
species but also to reestablish native plant communities that have been displaced or diminished
(Kettenring & Adams, 2011; Riebkes et al., 2018). Native revegetation contributes to the
restoration of ecosystem functions and biodiversity by providing habitat, food resources for
wildlife, and restoring soil health and water regulation (Lake et al., 2014). Moreover, the
introduction of native plants can create competitive pressures against invaders, reducing their
42
dominance and preventing reinvasion (Cutting & Hough-Goldstein, 2013; Pile Knapp et al.,
2023). In some cases, native plants selected for revegetation have specific traits that allow them
to effectively compete with invasive species, such as deep rooting systems that can access water
resources or specific allelopathic properties that inhibit the growth of invaders (Assis et al.,
2020). Competitive dynamics within plant communities can be leveraged, as demonstrated by an
experiment involving the over sowing of native species, which selectively suppressed the
invasive annual Rapistrum rugosum (Simmons, 2005). These strategies, when combined with
other management techniques like biological control or mechanical removal, might significantly
enhance the success of restoration efforts by creating more resilient and diverse ecosystems.
Restoration strategies must account for the temporal dynamics of invasive species control to
optimize recovery efforts. The timing of native revegetation appears to be important, with
previous literature suggesting that early integration with biological control can occupy the niche
that would otherwise be taken by reinvading weeds (Assis et al., 2020; Kettenring & Adams,
2011). In addition to the timing of control, the duration of control and monitoring has
significantly influenced the current knowledge in restoration ecology. Initial positive trends in
restoration can reverse over time, emphasizing the need for long-term monitoring and adaptive
management (Cordell et al., 2016). The timing of active control and monitoring varies, with 15%
of cases engaged in active control for a period of 2 to 5 years, and a smaller fraction, 7%,
extended beyond 5 years (Kettenring & Adams, 2011). Most studies conducted monitoring for a
duration of 1 to 2 years (22%), with monitoring for 2 to 5 years slightly less common (18%) and
significantly less frequent for periods exceeding 5 years (6%). The predominant approach to
implementing control methods was a single application, noted in 28% of studies (Kettentring &
Adams, 2011). This emphasizes the importance of diversifying the duration of control efforts and
43
enhancing the scope of monitoring to include multiple, adaptive management strategies. In
addition to such temporal effects, forests invaded by vines pose unique spatial challenges.
Globally, vines rank as the most detrimental invasive plants, given their ability to rapidly
colonize and dominate vertical and horizontal spaces that also complicate control efforts (David
& Lake, 2023). This often necessitates tailored management approaches such as targeted manual
removal and the use of vine-specific biocontrol agents (David & Lake, 2023).
In this research, we combined three different treatment methods, that were applied over a 3-
year period to control the invasive vine, pale swallow-wort (Vincetoxicum rossicum) in Michigan
woodlands. We wanted to incorporate the potential impact of a recently approved but not yet
established biocontrol agent and other environmentally friendly approaches. Swallow-wort has
shown resilience to mowing and cutting, and herbicide control takes repeat application which is
further complicated by the risk to non-target species (Biazzo & Milbrath, 2019; Collings et al.,
2023; DiTommaso et al., 2013). As a result of such challenges, we chose to include artificial
defoliation of swallow-wort, manual removal of swallow-wort and seeding of native plants. By
juxtaposing these methods both individually and collectively, we aimed to evaluate their
individual and combined efficacy. We hypothesized that combining treatments will prove more
effective than any single strategy, with manual removal and supplementary native seeding
anticipated as the most successful approach. To evaluate this hypothesis, we examined various
metrics including percent cover, species richness, diversity indices, and the density of swallow-
wort for the various treatments.
44
2. Methods
2.1 Study System
Swallow-wort vines – which primarily invade eastern North American forests – reduce
functional diversity, plant richness, and arthropod diversity (Carboni et al., 2021; Collings et al.,
2023; Ernst & Cappuccino, 2005; Sodhi et al., 2019). This experiment was conducted at three
such invaded sites in Oakland County, Michigan. According to the Michigan Natural Features
Inventory, these three sites are categorized as dry-mesic or mesic southern forest (Cohen et al.,
2020; Kost et al., 2007). The first site was located in a small recreational area within a residential
area in Oxford, Michigan (42° 49' 19.887"N, 83° 17' 14.0058"W). The 4.5 hectares infestation is
predominantly pale swallow-wort and the overstory is dominated by American elm (Ulmus
americana), bitternut hickory (Carya cordiformis), and sugar maple (Acer saccharum). The size
of the swallow-wort infestation at this site allowed us to include all treatments of interest. The
second site was located on private property in a rural area (42° 42' 22.1616"N, 83° 36'
28.0038"W). This site was predominantly pale swallow-wort but with patchier and a less dense
infestation than the first site. It was dominated by American elm (Ulmus americana), sugar
maple (Acer saccarum) and oak trees (Quercus spp.). The third site was located on private
property in a rural area near the previous site (42° 42' 18.8418"N, 83° 37' 32.3436"W). This site
had a more mixed understory than the previous two but with dense patches of swallow-wort in
certain areas. It was dominated heavily by black cherry (Prunus serotina) in the overstory. The
size of the infestations at these two sites required us to reduce the number of treatments we were
able to include. To differentiate the two reduced treatment sites, the former site which has a
moderate swallow-wort infestation will be called the reduced moderate site. The latter site which
has a patchy but dense infestation will be called the reduced dense site.
45
2.2 Experimental Design
Due to infestation size and site availability, sites varied in the number of treatments. We
conducted all treatments at one site and a reduced treatment design at two sites. At the full
treatment site, six treatments were executed in plots of 4m x 4m. At the reduced treatment sites,
four different treatments were applied within plots measuring 2m x 2m. Treatments were
implemented using a randomized complete block design and were replicated four times per site.
All plots had a minimum of 1m buffer between them and each plot contained a 1m x 1m subplot
positioned in the center of the plot. The restoration treatments included: (1) Control plots with no
intervention, (2) Artificial defoliation of pale swallow-wort, (3) Seeding and/or planting
seedlings of native species, (4) Defoliation of pale swallow-wort combined with seeding and/or
planting of native species, (5) Manual removal of pale swallow-wort, and (6) Manual removal of
pale swallow-wort followed by seeding and planting of native species. At the full treatment site,
all the above listed treatments were applied, and the seed addition treatments included the
planting of native seedlings, while only treatments 1 – 4 were applied at the reduced treatment
sites with addition of seeds but not of native plant seedlings. Planting of native seedling
happened at just one site due to availability of native plant plugs. Each site was established in the
fall of 2020.
A defoliating biological control agent – the moth Hypena opulenta Chrisoph (Lepidoptera:
Erebidae) – was introduced in the United States as a control method for swallow-wort vines
(Stewart, 2017). However, effectively implementing biological control measures can present
significant challenges in the field. The swallow-wort biocontrol agent had not yet successfully
overwintered in Michigan and had not exhibited sustainable populations at the time of this
experiment. To circumvent this issue, we used artificial defoliation to simulate feeding by H.
46
opulenta. Artificial defoliation represents a practical approach to simulating the impact of
biocontrol agents, providing insights into the invasive species' resilience to herbivory and the
compatibility with other restoration strategies to assess the potential of biocontrol agents before
their introduction and establishment. Several studies use artificial defoliation as a method to
evaluate the effects of herbivory on a target plant, with one study specifically investigating the
differences between simulated and insect herbivory and finding no difference (Kirby et al., 1997;
Milbrath, 2008; Milbrath et al., 2016; Watt et al., 2007; Wirf, 2006). This technique, while
experimental, complements traditional restoration methods by offering a proxy for the ecological
pressures exerted by enemies of invasive species. Therefore, for the defoliation only treatment
and defoliation plus seed addition treatment (all sites), feeding by the biocontrol agent was
mimicked by hand. Defoliation was conducted by ripping each leaf on a swallow-wort stem in
half. This level of defoliation was chosen because laboratory studies show that two larvae per
stem can cause up to a 50% reduction in leaf area on pale swallow-wort (Weed & Casagrande,
2010). At the full treatment site, in each year from 2021 to 2023, defoliation took three to four
weeks to implement due to the large size of the plots. Defoliation for this site was conducted in
the month of July. At the reduced treatment sites, defoliation took a week each and was
completed within the first two weeks of August.
The manual removal of swallow-wort was conducted at the full treatment site in treatments
five and six. The removal of swallow-wort plants was conducted in October of 2020 by digging
all swallow-wort stems out by the root and raking with a hard-tine rake to remove straggling root
balls and smooth the surface. Another manual removal was performed in spring of 2021 to
account for any roots or plants missed at the first removal.
47
For treatments three and four (all sites), and treatment six (full treatment site only) that
involved the addition of native plant seeds we used a native mesic woodland mix from Native
Connections (local Michigan genotypes; Kalamazoo, Michigan) supplemented with woodland
species from Michigan Wildflower Farm (Portland, Michigan). Seeding was conducted at the
recommended application rate by Native Connections of approximately 10 kg/ha of forbs and 25
kg/ha of grasses (30% forbs to 70% grasses). Core species were seeded yearly while some
species varied each year based on yearly availability (see a list of native plant included in the
seed mix in Appendix 1). At all sites in these treatments, seeding was conducted from fall 2020
to winter 2023 according to season specific recommendations. Each year, one seeding treatment
was conducted once in mid-October to January and one seeding treatment was conducted in mid-
April to mid-June for a total of six seed additions. Each fall/winter, seeding was performed by
raking the leaf layer from each plot. The soil was raked with a soft tine rake and seeds were hand
dispersed. Seeds were tamped into the soil by stomping and leaves were replaced on each plot.
We added native plantings to the full treatment site to take advantage of the increased size of
the plots and the additional treatments. Native plant plugs were grown from the same seed mix
used for the seed addition treatment by seeding the mix into a 51 x 25 cm plant growing tray
with drain holes using SUREMIX® all-purpose perlite mix (Michigan Grower Products Inc.,
United States). Trays were seeded in December of 2021 and placed outside for overwintering.
Trays were brought into the greenhouse in early April of 2021 and placed under 16:8h L:D lamps
at 22-25˚C. Species that grew in abundance were repotted into 1.14 L square plastic pots after
reaching approximately 2cm in height. Species that were abundant enough for planting were
Ageratina altissima (white snakeroot), Epilobium coloratum (cinnamon willowherb), Anemone
virginiana (tall thimbleweed), Oenothera biennis (common evening primrose), and Monarda
48
fistulosa (wild bergamot). Plugs were planted in any plot assigned to the seed addition treatment
outside of the 1m x 1m subplot but in the main plot, 70-80cm from the edge of the main plot.
Species were planted in a square, with at least one species per side (minimum 4 per plot). All
species except Epilobium coloratum and Monarda fistulosa were planted in replicates of 2 per
side (8 per plot). Transplantation occurred in June of 2021 and replacements were planted in July
2021 if transplantation failed. All plugs were watered immediately after planting.
Data Collection
Data collection took place in August to September from 2021 to 2023. Percent cover was
taken as a proportion of the subplot in each treatment allowing for overlap between species,
therefore allowing the total percent cover of all species to exceed 1. Species richness was taken
for the entire plot, including the subplot. For percent cover, each species was categorized into a
descriptive status as either ‘swallow-wort’, ‘invasive’, ‘native’, ‘non-native’, and ‘unknown’.
The statuses were classified using the University of Michigan Herbarium online flora search
(Reznicek & Walters, 2011). The ‘swallow-wort’ status only included the percent cover of
swallow-wort, whereas the percent cover of the ‘invasive’ status includes all other invasive
species other than swallow-wort. An invasive species is a non-native species introduced to an
ecosystem that causes or is likely to cause economic, environmental, or human health harm
(Higman et al., 2009). These species often outcompete native plants because they lack natural
population controls such as predators, competitors, and diseases from their native environment
(University of Michigan Environment, Health & Safety). The status ‘non-native’ included plant
species that would not be deemed invasive but did not originate in Michigan (U.S. Department of
the Interior). If a plant species could not be identified, often due to damage to the plant and/or
difficulty of identification in early life stages, it was placed in the ‘unknown’ status. Statuses
49
were altered for species richness so that swallow-wort was included in the ‘invasive’ category, as
there was only one species of swallow-wort at each site.
Likewise, a species was categorized as either seeded or not seeded, referring to species that
were actively seeded as part of the treatment or species that were not actively seeded, or
naturally occurring. In years 2022 and 2023, abundance was taken for each species in the subplot
for diversity indices calculations and swallow-wort stem density. Each species was identified at
minimum to family and at most to species. For consistency, if a plant was identified to the level
of genus or the level of species at first identification, it was kept at that level of identification for
the rest of the experiment. All forbs were identified to genus and often species. All trees were
identified to genus. Grasses, however, were only identified to family for percent cover as it was
difficult to disentangle species visually. Grasses in the full treatment site and one of the reduced
treatment sites were labeled as ‘native’ as there were no other grasses present at the site and
grass was only present where active seeding occurred. Grasses at the last reduced treatment site
could not be confidently identified and were left in the ‘unknown’ status category. If grasses
produced seed heads, they were identified to species for species richness data collection. Latin
names were formatted so that each taxonomic name was in a ‘family genus species’ format. If
genus or species was unknown, it was labeled as ‘spp’. Grasses were excluded from abundance
counts.
Native plantings at the full treatment site were checked for survival in 2022 and 2023. The
orientation of each plug was tracked by being assigned a side of the planting square which was
kept consistent for each plot. Additionally, the replicate per side was noted.
50
2.2 Statistical Analysis
All analyses were performed in R 4.3.0 (R Core Team, 2023). We constructed different
models for each response variable: percent cover, species richness, swallow-wort stem density,
and native planting survival. We provide chi-squared values for each model by using the Anova
functions from the lmtest and the car package (Fox & Weisberg, 2019; Zeileis & Hothorn, 2002).
2.2.1 Percent Cover
To assess if there were any differences in the percent cover by treatment, we chose to use a
beta family distribution which can be used for analyzing continuous proportions (Douma &
Weedon, 2019). Percentages were converted to proportions equal to or between 0 and 1 to use
this distribution. We used the glmmTMB package which allows us to implement a generalized
linear mixed model using the beta family distribution (Brooks et al., 2017). Fixed effects for this
model were additive and included treatment (control, defoliation, seed, defoliation + seed,
swallow-wort removal, swallow-wort removal + seed), year (2021, 2022, 2023), status
(‘swallow-wort’, ‘invasive’, ‘native’, ‘non-native’, or ‘unknown’), and site (full treatment site,
first reduced treatment site, second reduced treatment site). To account for repeated annual
sampling of each plot, plot replicate was used as a random effect (percent cover ~treatment +
year + status + site + (1|replicate)).
Additionally, we assessed if species that were actively seeded differed in their percent cover
by treatment (treatments three, four, and six). To do this, we used a nearly identical model as the
previous model but added the Latin name as a fixed effect (percent cover of seeded species ~
treatment + Latin name + year + site + (1|replicate)). We also added a zero-inflation parameter,
which was set with the same fixed and random effects as the conditional model. This was added
51
because beta family distributions cannot equal zero or one but a zero-inflation parameter allows
the model to include zeros.
To assess both models, we used the DHARMa package which provides residuals for
generalized linear mixed models (Hartig, 2022). Post-hoc comparisons were performed using the
emmeans package (Lenth et al., 2021).
2.2.2 Species Richness
To assess overall species richness, we first assessed the normality of species richness using a
Shapiro-Wilk’s test from the stats package (R Core Team, 2023). Once species richness was
found to fit the assumptions of normality, a linear mixed effects model was used for analysis
from the lme4 package (Bates et al., 2015). Fixed effects for this model were additive and
included treatment (control, defoliation, seed, defoliation + seed, swallow-wort removal,
swallow-wort removal + seed), year (2021, 2022, 2023), and site (full treatment site, first
reduced treatment site, second reduced treatment site). Plot replicate was used as a random effect
(percent cover ~ treatment + year + site + (1|replicate)). Model fit was assessed by plotting
residuals and creating a Q-Q plot from the stats package (R Core Team, 2023).
Species richness was also analyzed by status (‘native’, ‘invasive’, ‘non-native’, or
‘unknown). To do this, we again assessed the normality of species richness using a Shapiro-
Wilk’s test from the stats package after separating the number of species by status (R Core
Team, 2023). After separation, species richness did not follow the assumptions of normality,
therefore we implemented a generalized linear mixed effects model with a Poisson distribution,
which is often used to model count data. Fixed effects for this model included treatment (control,
defoliation, seed, defoliation + seed, swallow-wort removal, swallow-wort removal + seed), year
(2021, 2022, 2023), status (‘swallow-wort’, ‘invasive’, ‘native’, ‘non-native’, or ‘unknown’), and
52
site (full treatment site, first reduced treatment site, second reduced treatment site). We added an
interaction between treatment and status, but site and year were additive. Plot replicate was used
as a random effect (percent cover ~ treatment * status + year + site + (1|teplicate)). To assess this
model, we used the DHARMa package which provides residuals for generalized linear mixed
models (Hartig, 2022). Post-hoc comparisons for both models were performed using the
emmeans package (Lenth et al., 2021).
2.2.3 Swallow-wort Stem Density
To assess if swallow-wort stem density varied by treatment, we first assessed the normality
of the count of swallow-wort stems using a Shapiro-Wilk’s test from the stats package (R Core
Team, 2023). Fixed effects included treatment (control, defoliation, seed, defoliation + seed,
swallow-wort removal, swallow-wort removal + seed), year (2022, 2023), and site (full treatment
site, first reduced treatment site, second reduced treatment site). An interaction between
treatment and site was added. Plot replicate was used as a random effect (percent cover ~
treatment * site + year + (1|replicate)). Swallow-wort stem density did not meet the assumptions
of normality, so a generalized linear mixed effect model was performed using a Poisson
distribution. However, the model exhibited significant overdispersion after assessment for
overdispersion using the DHARMa package (Hartig, 2022). To address this issue, we created a
similar model using negative binomial distribution, which is often used to counteract
overdispersion. We reassessed dispersion of the negative binomial model and found it to be
satisfactory. Additionally, the Poisson model and the negative binomial model were compared
using alkaline information criteria (AIC) from the bbmle package (Bolker and Team, 2022). The
negative binomial model produced the lowest AIC, confirming our model choice.
53
2.2.5 Native Planting Survival
At the full treatment site, native plant plugs were individually assessed for survival and were
assigned a “0” if they did not survive and a “1” if they did survive. A generalized linear mixed
effects model was used with a binomial distribution using the lme4 package (Bates et al., 2015).
Treatment (seed, defoliation + seed, removal + seed), year (2021, 2022), and species planted
(Ageratina altissima, Monarda fistulosa, Epilobium coloratum, Oenthera biennis, and Anemone
virginiana) were all included in the fixed effects. Replicate was included as a random effect by
nesting the treatment replicate, the orientation (planting square side: 1-4) within the treatment
replicate and the replicate within orientation (survival ~ treatment + species planted + year +
(1|orientation replicate: orientation: treatment replicate)).
3. Results
3.1 Percent Cover
Percent plant cover varied by treatment (χ2 = 38.17, df = 5, p < 0.001), by year (χ2 = 9.28, df
= 2, p < 0.01), by plant status (χ2 = 684.49, df = 4, p < 0.001), and by site (χ2 = 25.15, df = 2, p
< 0.001) (Fig. 2.1). All treatments were significantly different from the control (pairwise
comparison: p < 0.03) except the seed addition treatment. The defoliation, seed addition,
defoliation plus seed addition treatments all increased percent plant cover when compared to the
removal treatment (pairwise comparisons: p < 0.01). Overall, the percent cover in 2022 was 1.2
times higher than in 2023 (pairwise comparison: p < 0.01) but other comparisons were not
significant. The total percent plant cover was dominated by swallow-wort (41.97%) and native
species (41.56%). A smaller proportion was covered by other invasive species than swallow-wort
(7.70%), non-native species (3.24%), and unknown species (5.56%). Native species were 0.2
times the percent cover of swallow-wort (pairwise comparison: p < 0.001) and 1.6 times the
54
percent cover of non-native species (pairwise comparison: p < 0.01). Additionally, comparisons
of the percent cover between swallow-wort and invasives, swallow-wort and non-natives, and
swallow-wort and unknown species were significant (pairwise comparisons: p < 0.001). All other
comparisons were not significant. The reduced dense treatment site exhibited significantly higher
plant cover from both the reduced moderate site and the full treatment site (pairwise
comparisons: p <0.001). However, the reduced moderate site and the full treatment site were not
significantly different from each other.
55
Figure 2.1. Mean and standard error of the percent plant cover by treatment (control, swallow-
wort defoliation, swallow-wort removal, native seed addition, swallow-wort defoliation + native
seed addition, swallow-wort removal + native seed addition) categorized by status (native or
swallow-wort). The categories of invasive, non-native, and unknown for plant invasion status
were not included for readability. Observations have been jittered to show individual points.
Asterisks indicate significant differences between treatments at the 0.05(*), 0.01(**), 0.001 (***)
levels.
56
Percent plant cover of species that were added with a seed mix varied by treatment (χ2 =
10.12, df = 2, p < 0.01), by year (χ2 = 24.14, df = 2, p < 0.001), by Latin name (χ2 = 60.61, df =
12, p < 0.001), but not by site (χ2 = 2.37, df = 2, p =0.31) (Fig. 2.2). The removal plus seed
addition treatment exhibited 1.7 times the seeded plant cover of the seed addition only treatment
(pairwise comparison: p < 0.01) and 1.5 times the defoliation plus seed treatment (pairwise
comparison: p < 0.05). However, the seed and the defoliation plus seed treatment were not
significant from each other. The year 2023 had higher seeded plant cover than both 2021 and
2022 (pairwise comparisons: p < 0.01), doubling the amount of seeded plant cover from 2021 to
2023. The species that dominated the seeded plant cover were native grasses and species in the
family Asteraceae, particularly Ageratina altissima. The cover of native grasses was significantly
higher than Zizia aurea, Rudbeckia spp., Monarda fistulosa, Oenothera biennis, and Thalictrum
dioicum (pairwise comparisons: p < 0.05).
57
Figure 2.2. Mean percent plant cover of actively seeded species by treatment (seed, def + seed,
removal + seed). The species are given by their Latin names in a ‘Family Genus species’ format.
3.2 Species Richness
Species richness varied by treatment (χ2 = 104.83, df = 5, p < 0.001), by year (χ2 = 42.24, df
= 2, p < 0.001), and by site (χ2 = 80.22, df = 2, p < 0.001). When compared to the control
treatment, all treatments except the defoliation only treatment had significantly higher species
58
richness (pairwise comparisons: p < 0.01). The defoliation plus seed treatment more than
doubled the species richness when compared to the defoliation only treatment (pairwise
comparison: p < 0.05). The removal treatment increased species richness when compared to the
defoliation treatment by 3.9 times (pairwise comparison: p < 0.01). And the swallow-wort
removal treatment plus seed addition outperformed all by increasing species richness by 8.5
times when compared to the defoliation treatment, by 6.8 times when compared to the seed
treatment, by 6.2 times when compared to the defoliation plus seed addition treatment, and by
4.5 times when compared to the swallow-wort removal treatment (pairwise comparisons: p <
0.01) (Fig. 2.3). There was a decrease in species richness from 2021 to 2022, but an increase in
species richness from and 2022 to 2023 and an overall increase from 2021 to 2023 (pairwise
comparisons: p < 0.05). The reduced dense site was significantly higher in species richness by
4.1 times from the reduced moderate site and by 6.8 times from the full treatment site (pairwise
comparisons: p < 0.001). The reduced moderate site was also 2.6 times more species rich than
the full treatment site (p < 0.05) (Fig. 2.4).
59
Figure 2.3. Mean and standard error for species richness by treatment (control, swallow-wort
defoliation, swallow-wort removal, native seed addition, swallow-wort defoliation + native seed
addition, swallow-wort removal + native seed addition). Observations have been jittered to show
individual points. Treatments not sharing any letter are significantly different from each other at
p < 0.05.
60
Figure 2.4. Mean and standard error for species richness by treatment (control, swallow-wort
defoliation, swallow-wort removal, native seed addition, swallow-wort defoliation + native seed
addition, swallow-wort removal + native seed addition). Panels are separated by site (reduced
treatment dense swallow-wort site, reduced treatment moderate swallow-wort site, full treatment
site). All sites were significantly different from each other (p < 0.05).
61
When species richness was separated by plant status, it varied by status (χ2 = 120.06, df = 3,
p < 0.001), by year (χ2 = 34.07, df = 2, p < 0.001), by site (χ2 = 95.00, df = 2, p < 0.001), and the
interaction between treatment and status (χ2 = 63.41, df = 15, p < 0.001) (Fig. 2.5). However,
species richness did not vary by treatment alone (χ2 = 8.46, df = 5, p = 0.13). All combinations of
status were significantly different from each other (pairwise comparisons: p < 0.001) within each
treatment with two exceptions. The number of non-native species and unknown species did not
significantly differ from each other across all treatments (pairwise comparisons: p > 0.05).
Additionally, the number of invasive species and native species were not significantly different
from each other within the control treatment (pairwise comparison: p = 0.06). Between each
treatment, all treatments except the defoliation only treatment were significantly different when
compared to the control (pairwise comparisons: p < 0.001). All other comparisons between
treatments were significant (p < 0.001), except for the defoliation plus seed addition and
removal, defoliation plus seed addition and seed addition, and the removal and seed addition
treatments. Similar to the previous model not including status, all pairwise comparisons were
significant for site and year (pairwise comparisons: p < 0.05).
62
Figure 2.5. Mean and standard error of species richness by treatment (control, swallow-wort
defoliation, swallow-wort removal, native seed addition, swallow-wort defoliation + native seed
addition, swallow-wort removal + native seed addition) categorized by status (native or swallow-
wort). The categories of invasive, nonnative, and unknown for plant invasion status were not
included for readability. Observations have been jittered to show individual points. Asterisks
indicate significant differences between treatments at the 0.05(*), 0.01(**), 0.001 (***) levels.
NS indicates there is not significant difference.
63
3.3 Swallow-wort Stem Density
Stem density of swallow-wort varied by treatment (χ2 = 116079.88, df = 5, p < 0.001), by
year (χ2 = 226.65, df = 1, p < 0.001), by site (χ2 = 133.72, df = 2, p < 0.001), and by the
interaction between treatment and site (χ2 = 14955.07, df = 6, p < 0.001) (Fig. 2.6). At the full
treatment site, all comparisons between treatments were significant (pairwise comparisons: p <
0.001), except between the defoliation and the defoliation plus seed addition treatment. At the
reduced treatment site with a moderate swallow-wort infestation, all comparisons between
treatments were significantly different (pairwise comparisons: p < 0.001). At the reduced
treatment site with a dense swallow-wort infestation, all comparisons between treatments were
significantly different (pairwise comparisons: p < 0.001), except between the control and the
defoliation plus seed treatment. The year 2022 had a higher swallow-wort stem density than the
year 2023 (pairwise comparison: p < 0.001)
64
Figure 2.6. Mean and standard error of swallow-wort stem density by treatment (control,
swallow-wort defoliation, swallow-wort removal, native seed addition, swallow-wort defoliation
+ native seed addition, swallow-wort removal + native seed addition). Observations have been
jittered to show individual points. Within each panel, treatments not sharing any letter are
significantly different from each other at p < 0.05.
65
3.4 Native Planting Survival
Overall, survival of native seedling plugs varied by treatment (χ2 = 16.35, df = 2, p < 0.001),
by year (χ2 = 6.15, df = 2, p < 0.05), and by the species planted (χ2 = 22.96, df = 4, p < 0.001).
Native plugs planted in the swallow-wort removal plus native seed addition treatment were 2.5
times more likely to survive than plugs planted in the seed addition treatment and 2.7 times more
likely to survive than plugs planted in the defoliation plus seed addition treatment (pairwise
comparisons: p < 0.01). Survival reduced from 2022 to 2023 (pairwise comparison: p = 0.06).
The plant species Ageratina altissima was more likely to survive than Anemone virginiana
(pairwise comparison: p < 0.001), however no other comparisons were significant.
4. Discussion
Overall, we demonstrated that combining any mechanical swallow-wort control with active
seeding of native plant species significantly enhanced species richness. Almost all actions,
whether single or combined treatments, positively impacted native species richness and reduced
swallow-wort cover and stem density. However, changes in percent plant cover were mainly
driven by treatments involving the active reduction of swallow-wort biomass. This suggests that
both dispersal limitation of native species and competitive pressures from invasive species must
be considered to promote native plant diversity in this system.
Our results build on previous work showing that the introduction of native plant species is a
critical factor in restoration success. Dispersal limitation can be a significant barrier to restoring
native plant communities (Suding et al., 2004). Native plants face challenges in naturally
recolonizing areas due to habitat reduction, habitat fragmentation, and population declines
(Biswas & Mallik, 2010; Cordeiro et al., 2009). This is well-documented in prairie plant
communities (Grman et al., 2015) and is more crucial for restoring forest communities than
66
understory light competition (Brudvig et al., 2011). In restoring a forest invaded by Euonymus
fortunei, dispersal limitation, rather than invasive species removal alone, was found to be an
important factor (Bauer & Reynolds, 2016). However, evidence for dispersal limitation in
invaded forests is sparse, particularly in North America (Pile Knapp et al., 2023). Our study adds
to this evidence, showing that in invaded forest communities, dispersal limitation is an essential
factor for restoration success. By manually introducing native plants to areas from which they
have been extirpated, land managers can bypass dispersal barriers and enhance the speed and
success of ecological restoration efforts.
Another key component of restoration success in invaded systems is the active control or
biomass reduction of the invader. Species richness was enhanced by swallow-wort removal with
or without seed addition, and defoliation with seed addition. All treatments that included a
swallow-wort biomass reduction method (defoliation or removal) reduced the percent cover of
swallow-wort, with removal having the most dramatic effect. This is consistent with previous
studies showing that swallow-wort removal can significantly reduce its coverage (Collings et al.,
2023). Defoliation, either alone or combined with seed addition, decreased swallow-wort
biomass and increased native species richness compared to the control. Biomass reduction of
swallow-wort by either method is likely beneficial by alleviating the competitive pressures that
invasive plants impose on the community. Invasive plants outcompete native plants for resources
such as light, water, and nutrients, hindering native plant establishment. Numerous studies have
demonstrated that invasive species can alter soil chemistry and disrupt mutualistic relationships,
sometimes continuing these processes even after removal (Vilà et al., 2011; Levine et al., 2003).
By implementing control measures to reduce the presence and spread of invasive species,
competitive pressure on native plants is alleviated, allowing them to reclaim their natural
67
habitats. Projects targeting vine invasions, such as those involving Vincetoxicum rossicum,
require such approaches due to vines' ability to rapidly dominate spaces and impede native plant
recovery (Ditommaso et al., 2005; Livingstone et al., 2020).
The combined effects of native seed addition and swallow-wort control methods were the
most effective in reducing swallow-wort and enhancing species richness simultaneously. The
removal of swallow-wort with native seed additions was the most effective treatment for all
measurements, indicating minimal legacy effects from swallow-wort infestation. The defoliation
plus seed addition treatment was not significantly different from defoliation treatment in percent
plant cover, nor from the removal treatment and seed addition treatment in species richness.
Despite this, the defoliation plus seed addition treatment outperforms both stand-alone treatments
when we consider both swallow-wort reduction and native species richness simultaneously. It
was more effective than defoliation alone, which did not significantly improve species richness,
and the seed-only treatment, which failed to substantially reduce swallow-wort cover compared
to the control. When we compare biological control to manual removal, biological control has
several advantages, although it is a less immediate solution. It is less labor-intensive, potentially
saving on labor costs. Additionally, biological control agents can naturally spread across barriers
that humans cannot, accessing difficult-to-reach areas through natural spread or nearby releases.
With this information, practitioners can weigh the pros and cons of both treatments to make a
practical decision fitting their goals and available resources.
The combination of native species establishment through active dispersal of seeds and
invasive species control benefits restoration through various mechanisms. Native plants may
occupy resources and space that would otherwise be taken by invading species, preventing what
is commonly called the 'treadmill effect', referring to the continual cycle of new invasive species
68
colonizing areas after the initial invasive species have been controlled (Lake et al., 2014; Pile
Knapp et al., 2023; Schuster et al., 2018). Established native plants can prevent reinvasion or
new invasions and potentially suppress the existing invasive plant population. This combined
approach can create a synergistic effect, leading to a more resilient and self-sustaining ecosystem
(Lake & Minteer, 2018). Invasive species often create feedback loops that further facilitate their
dominance, such as altering nutrient cycling in ways that disadvantage native plants (Ehrenfeld,
2003). By removing or reducing invasive species and reintroducing native plants, these negative
feedback loops can be disrupted, and positive interactions among native species can be restored,
leading to a more robust and dynamic ecosystem (Falk et al., 2006).
Additionally, considering which species of seeds were most successful is important. In seed
addition treatments, specific native plants succeeded, with native grasses being the earliest
established. The number of forbs in seed addition treatments increased each year, dominating the
percent cover in combination treatments but not in seed-only treatments. Species in the family
Asteraceae established early, similar to native grasses. To understand this success and enhance
resistance to invasions, it is beneficial to create native plant communities with functional
diversity, including species that share traits with invasive plants. This approach aligns with the
concept of "limiting similarity” " (Funk et al., 2008; Shea & Chesson, 2002; Pile Knapp et al.,
2023). Research into the traits of native forbs in the family Asteraceae and swallow-wort could
reveal if limiting similarity could be a factor in their success, allowing prioritization of such
species in swallow-wort restorations.
This experiment emphasizes that combining invasive plant control with native seed additions
can mitigate swallow-wort prevalence and bolster native species richness, despite some site-
specific exceptions. The consistent decrease in swallow-wort and increase in native species
69
richness across treatments points to the potential of combining management approaches. Our
results highlight the importance of addressing dispersal limitation and competition in restoration
approaches across various sites and the significance of conducting research that spans multiple
years of control and observation. We intend to track the long-term outcomes by continuing data
collection for at least an additional year, with aspirations for two more years of observation. This
extended observation period could gather supplementary information, including the effects of
treatments on arthropod diversity and the reproductive potential of swallow-wort. Future
research should explore combining various restoration tactics with biological control across
diverse environments and spatial arrangements, considering their effects on the biological control
agent. This approach could enhance the impact and efficacy of biocontrol agents.
70
LITERATURE CITED
Ainsworth, N. 2003. Integration of herbicides with arthropod biocontrol agents for weed control.
Biocontrol Science and Technology, 13(6), 547–570.
https://doi.org/10.1080/0958315031000151819
Alred, B., Haan, N., Landis, D. A., & Szucs, M. 2022. Does the Presence of the Biological
Control Agent, Hypena opulenta (Lepidoptera: Erebidae) on Swallow-Worts Deter
Monarch Oviposition? Environmental Entomology, 51(1), 77–82.
https://doi.org/10.1093/ee/nvab121
Anon. 2012. Black and pale swallow-worts. Invasive Species – Best Control Practices series.
Michigan Dept. of Natural Resources, Michigan Natural Feature Inventory. 7 p.
Assis, G. B., Pilon, N. A. L., Siqueira, M. F., & Durigan, G. 2021. Effectiveness and costs of
invasive species control using different techniques to restore cerrado grasslands.
Restoration Ecology, 29(S1). https://doi.org/10.1111/rec.13219
Bates, D., Maechler, M., Bolker, B, Walker, S. 2015. Fitting Linear Mixed-Effects Models Using
lme4. Journal of Statistical Software, 67(1), 1-48. doi:10.18637/jss.v067.i01.
Bauer, J. T., & Reynolds, H. L. (2016). Restoring native understory to a woodland invaded by
Euonymus fortunei: Multiple factors affect success. Restoration Ecology, 24(1), 45–52.
https://doi.org/10.1111/rec.12285
Biswas, S., & Mallik, A. (2010). Disturbance effects on species diversity and functional diversity
in riparian and upland plant communities. Ecology, 91, 28–35.
Blossey, B. and Nötzold, R. 1995. Evolution of increased competitive ability in invasive
nonindigenous plants: a hypothesis. J. Ecol. 83, 887-889
Blossey, B. 1999. Before, during and after: the need for long-term monitoring in invasive plant
species management. Biological Invasions 1: 301-311.
Bolker B, R Development Core Team. 2022. _bbmle: Tools for General Maximum Likelihood
Estimation_. R package version 1.0.25, .
Brooks, M. E., Kristensen, K., van Benthem, K. J., Magnusson, A., Berg, C. W., Nielsen, A.,
Skaug, H. J., Maechler, M., and Bolker, B. M. 2017. glmmTMB Balances Speed and
Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. The
R Journal, 9(2), 378-400. doi: 10.32614/RJ-2017-066.
Brudvig, L. A., Mabry, C. M., & Mottl, L. M. 2011. Dispersal, not Understory Light
Competition, Limits Restoration of Iowa Woodland Understory Herbs. Restoration
Ecology, 19(101), 24–31. https://doi.org/10.1111/j.1526-100X.2010.00675.x
71
Casagrande, R. A., & Dacey, J. E. 2007. Monarch butterfly oviposition on swallow-worts
(Vincetoxicum spp.). Environmental Entomology, 36(3), 631–636.
https://doi.org/10.1603/0046-225X(2007)36[631:MBOOSV]2.0.CO;2
Cohen, J.G., M.A. Kost, B.S. Slaughter, D.A. Albert, J.M. Lincoln, A.P. Kortenhoven, C.M.
Wilton, H.D. Enander, and K.M. Korroch. 2020. Michigan Natural Community
Classification [web application]. Michigan Natural Features Inventory, Michigan State
University Extension, Lansing, Michigan.
Available https://mnfi.anr.msu.edu/communities/classification. (Accessed: March 13,
2024).
Collings, J., Endriss, S. B., & Dávalos, A. 2023. Multiple stressors prevent gains in native plant
diversity following invasive species removal. Ecosphere, 14(3).
https://doi.org/10.1002/ecs2.4458
Corcoran D. 2023. _AICcPermanova: Model Selection of PERMANOVA Models Using AICc_.
R package version 0.0.2, .
Cordeiro, N. J., Ndangalasi, H. J., McEntee, J. P., & Howe, H. F. (2009). Disperser limitation
and recruitment of an endemic African tree in a fragmented landscape. Ecology, 90,
1030–1041.
Cribari-Neto F, Zeileis A. 2010. “Beta Regression in R.” _Journal of Statistical
Software,*34*(2), 1-24. doi:10.18637/jss.v034.i02
.
Cordell, S., Ostertag, R., Michaud, J., & Warman, L. 2016. Quandaries of a decade-long
restoration experiment trying to reduce invasive species: Beat them, join them, give up,
or start over? Restoration Ecology, 24(2), 139–144. https://doi.org/10.1111/rec.12321
Cuddington, K. 2011. Legacy Effects: The Persistent Impact of Ecological Interactions.
Biological Theory, 6(3), 203–210. https://doi.org/10.1007/s13752-012-0027-5
Cutting, K. J., & Hough-Goldstein, J. 2013. Integration of biological control and native seeding
to restore invaded plant communities. Restoration Ecology, 21(5), 648–655.
https://doi.org/10.1111/j.1526-100X.2012.00936.x
David, A. S., & Lake, E. C. 2023. Biological control of vines: A review of past efforts,
evaluation, and future directions. Biological Control, 183.
https://doi.org/10.1016/j.biocontrol.2023.105257
Ditommaso, A., Lawlor, F. M., & Darbyshire, S. J. 2005. The Biology of Invasive Alien Plants in
Canada. 2. Cynanchum rossicum (Kleopow) Borhidi [= Vincetoxicum rossicum
(Kleopow) Barbar.] and Cynanchum louiseae (L.) Kartesz & Gandhi [= Vincetoxicum
nigrum (L.) Moench]. Canadian Journal of Plant Science, 243-263.
72
Douglass, C. H., Weston, L. A., & Wolfe, D. 2011. Phytotoxicity and Potential Allelopathy in
Pale (Cynanchum rossicum) and Black swallowwort (C. nigrum) . Invasive Plant Science
and Management, 4(1), 133–141. https://doi.org/10.1614/ipsm-d-10-00021.1
Douma, J. C., & Weedon, J. T. (2019). Analysing continuous proportions in ecology and
evolution: A practical introduction to beta and Dirichlet regression. Methods in Ecology
and Evolution, 10(9), 1412–1430. https://doi.org/10.1111/2041-210X.13234
Ehrenfeld, J. G. (2003). Effects of Exotic Plant Invasions on Soil Nutrient Cycling Processes.
Ecosystems, 6(6), 503–523. https://doi.org/10.1007/s10021-002-0151-3
Falk, D. A., Palmer, M. A., & Zedler, J. B. 2006. Foundations of restoration ecology.
Washington, D.C., Island Press.
Fox J, Weisberg S. 2019. _An R Companion to Applied Regression_, Third edition. Sage,
Thousand Oaks CA. .
Funk, J.L., Cleland, E.E., Suding, K.N., Zavaleta, E.S. 2008. Restoration through reassembly:
plant traits and invasion resistance. Trends in Ecology and Evolution, 23, 695-703.
Grman, E., Bassett, T., Zirbel, C. R., & Brudvig, L. A. (2015). Dispersal and establishment
filters influence the assembly of restored prairie plant communities. Restoration Ecology,
23(6), 892–899. https://doi.org/10.1111/rec.12271
Hartig F. 2022. _DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed)
Regression Models_. R package version 0.4.6, .
Heimpel, G. E., & Mills, N. J. (2017). Definitions and Interactions. Biological Control, 1-18 .
Cambridge University Press. https://doi.org/10.1017/9781139029117.003
Higman, P., Campbell, S., Borland, K., & Schillo, R. (2009). A Field Identification Guide to
Invasive Plants in Michigan’s Natural Communities. https://mnfi.anr.msu.edu/invasive-
species/InvasivePlantsFieldGuide.pdf
Huston, M.A. 1994. Biological Diversity: the coexistence of species on changing landscapes.
Cambridge University Press. Journal of Tropical Ecology. 11(4):568-568.
doi:10.1017/S0266467400009135
Kettenring, K. M., & Adams, C. R. 2011. Lessons learned from invasive plant control
experiments: A systematic review and meta-analysis. Journal of Applied Ecology, 48(4),
970–979. https://doi.org/10.1111/j.1365-2664.2011.01979.x
Kirby, D. R., Hanson, T. P., Krabbenhoft, K. D., & Kirby, M. M. 1997. Effects of Stimulated
Defoliation on Leafy Spurge (Euphorbia esula)-Infested Rangeland. Weed Technology,
11(3), 586–590. https://www.jstor.org/stable/3988442
73
Kost, M.A., D.A. Albert, J.G. Cohen, B.S. Slaughter, R.K. Schillo, C.R. Weber, and K.A.
Chapman. 2007. Natural Communities of Michigan: Classification and Description.
Michigan Natural Features Inventory, Report No. 2007-21, Lansing, MI.
Lake, E. C., Hough-Goldstein, J., & D’Amico, V. 2014. Integrating management techniques to
restore sites invaded by mile-a-minute weed, Persicaria perfoliata. Restoration Ecology,
22(2), 127–133. https://doi.org/10.1111/rec.12035
Lake, E. C., & Minteer, C. R. 2018. A review of the integration of classical biological control
with other techniques to manage invasive weeds in natural areas and rangelands.
BioControl, 63(1), 71–86. https://doi.org/10.1007/s10526-017-9853-5
Lenth, R., Singmann, H., Love, J., Buerkner, P., & Herve, M. 2021. Emmeans: Estimated
marginal means, aka least-squares means. R package version 1 (2018).
Levine, J.M., Vila, M., D’Antonio, C.M., Dukes, J.S., Grigulis, K., Lavorel, S. 2003.
Mechanisms underlying the impacts of exotic plant invasions. Proceedings of the Royal
Society of London, Series B – Biological Sciences, 270, 775-781.
Livingstone, S. W., Isaac, M. E., & Cadotte, M. W. 2020. Invasive dominance and resident
diversity: unpacking the impact of plant invasion on biodiversity and ecosystem function.
Ecological Monographs. https://doi.org/10.1002/ecm.1425
Martinez Arbizu P. 2017. _pairwiseAdonis: Pairwise Multilevel Comparison using Adonis_. R
package version 0.4.1.
McKague, C. I., & Cappuccino, N. 2005. Response of Pale Swallow-wort, Vincetoxicum
rossicum, following aboveground tissue loss. Canadian Field Naturalist, 119(4), 525-
531.
MICHIGAN FLORA ONLINE. A. A. Reznicek, E. G. Voss, & B. S. Walters. February 2011.
University of Michigan. Web. June, 25, 2024. https://lsa.umich.edu/herbarium
Milbrath, L. R. 2008. Growth and reproduction of invasive Vincetoxicum rossicum and V.
nigrum under artificial defoliation and different light environments. Botany, 86(11),
1279–1290. https://doi.org/10.1139/B08-092
Milbrath, L. R., DiTommaso, A., Biazzo, J., & Morris, S. H. 2016. Tolerance of Swallowworts
(Vincetoxicum spp.) to Multiple Years of Artificial Defoliation and Clipping. Invasive
Plant Science and Management, 9(1), 1–11. https://doi.org/10.1614/ipsm-d-15-00042.1
Oksanen J, Simpson G, Blanchet F, Kindt R, Legendre P, Minchin P, O'Hara R, Solymos P,
Stevens M, Szoecs E, Wagner H, Barbour M, Bedward M, Bolker B, Borcard D,
Carvalho G, Chirico M, De Caceres M, Durand S, Evangelista H, FitzJohn R, Friendly
M, Furneaux B, Hannigan G, Hill M, Lahti L, McGlinn D, Ouellette M, Ribeiro Cunha E,
74
Smith T, Stier A, Ter Braak C, Weedon J. 2022. _vegan: Community Ecology Package_.
R package version 2.6-4, .
Oswalt, C.M., Fei, S., Guo, Q., Iannone III, B.V., Oswalt, S.N., Pijanowski, B.D., Potter, K.M.
2015. A subcontinental view of forest plant invasions. NeoBiota, 24, 49-54.
Pile Knapp, L. S., Coyle, D. R., Dey, D. C., Fraser, J. S., Hutchinson, T., Jenkins, M. A., Kern,
C. C., Knapp, B. O., Maddox, D., Pinchot, C., & Wang, G. G. 2023. Invasive plant
management in eastern North American Forests: A systematic review. Forest Ecology
and Management, 550. https://doi.org/10.1016/j.foreco.2023.121517
Prior, K. M., Adams, D. C., Klepzig, K. D., & Hulcr, J. 2018. When does invasive species
removal lead to ecological recovery? Implications for management success. Biological
Invasions, 20(2), 267-283. https://doi.org/10.1007/s10530-017-1542-x
R Core Team. 2023. _R: A Language and Environment for Statistical Computing_. R
Foundation for Statistical Computing, Vienna, Austria. .
Riebkes, J. L., Sherrard, M. E., & Jackson, L. L. 2018. Supplemental seed increases native
seedling establishment in roadside prairie restoration. Restoration Ecology, 26(6), 1149–
1156. https://doi.org/10.1111/rec.12699
Schuster, M. J., Wragg, P. D., & Reich, P. B. (2018). Using revegetation to suppress invasive
plants in grasslands and forests. Journal of Applied Ecology, 55(5), 2362–2373.
https://doi.org/10.1111/1365-2664.13195
Shea, K., Chesson, P. 2002. Community ecology theory as a framework for biological invasions.
Trends in Ecology & Evolution, 17, 170-176.
Simmons, M. T. 2005. Bullying the bullies: The selective control of an exotic, invasive annual
(Rapistrum rugosum) by oversowing with a competitive native species (Gaillardia
pulchella). Restoration Ecology, 13(4), 609–615. https://doi.org/10.1111/j.1526-
100X.2005.00078.x
Suding, K. N., Gross, K. L., & Houseman, G. R. (2004). Alternative states and positive
feedbacks in restoration ecology. Trends in Ecology & Evolution, 19(1), 46-53.
https://doi.org/10.1016/j.tree.2003.10.005
Thomas, M. B., & Reid, A. M. 2007. Are exotic natural enemies an effective way of controlling
invasive plants? Trends in Ecology and Evolution, 22(9), 447–453.
https://doi.org/10.1016/j.tree.2007.03.003
University of Michigan Environment, Health & Safety. (n.d.). Invasive species. Environmental.
https://ehs.umich.edu/environmental/invasive-species/
75
U.S. Department of the Interior. (n.d.). Invasive & Non-Native Species. National Parks Service.
https://www.nps.gov/subjects/invasive/learn.htm
Vilà, M., Espinar, J. L., Hejda, M., Hulme, P. E., Jarošík, V., Maron, J. L., Pergl, J., Schaffner,
U., Sun, Y., & Pyšek, P. (2011). Ecological impacts of invasive alien plants: A meta-
analysis of their effects on species, communities and ecosystems. Ecology Letters, 14(7),
702–708. https://doi.org/10.1111/j.1461-0248.2011.01628.x
Vitousek, P.M., D’Antonio, C.M., Loope, L.L., Rejmanek, M. & Westbrooks, R. 1997.
Introduced species: a significant component of human-caused global change. N.Z. J. Ecol
21, 1-16.
Watt, M. S., Whitehead, D., Kriticos, D. J., Gous, S. F., & Richardson, B.(2007. Using a process-
based model to analyse compensatory growth in response to defoliation: Simulating
herbivory by a biological control agent. Biological Control, 43(1), 119–129.
https://doi.org/10.1016/j.biocontrol.2007.06.011
Weed, A. S., & Casagrande, R. A. 2010. Biology and larval feeding impact of Hypena opulenta
(Christoph) (Lepidoptera: Noctuidae): A potential biological control agent for
Vincetoxicum nigrum and V. rossicum. Biological Control, 53(2), 214–222.
https://doi.org/10.1016/j.biocontrol.2009.12.004
Weidlich, E. W. A., Flórido, F. G., Sorrini, T. B., & Brancalion, P. H. S. 2020. Controlling
invasive plant species in ecological restoration: A global review. Journal of Applied
Ecology, 57(9), 1806-1817https://doi.org/10.1111/1365-2664.13656
Wickham H, François R, Henry L, Müller K, Vaughan D.(2023. _dplyr: A Grammar of Data
Manipulation_. R package version 1.1.2, .
Wirf, L. A. 2006. The effect of manual defoliation and Macaria pallidata (Geometridae)
herbivory on Mimosa pigra: Implications for biological control. Biological Control,
37(3), 346–353. https://doi.org/10.1016/j.biocontrol.2005.12.011
Zeileis, A., Hothorn, T. 2002. Diagnostic Checking in Regression Relationships. R News, 2(3), 7-
10. URL https://CRAN.R-project.org/doc/Rnews/
76
Table 2.1. A list of all native species in the seed mix from seed addition treatments.
APPENDIX
Plant Type
Grass
Sedge
Sedge
Sedge
Grass
Grass
Grass
Grass
Grass
Rush
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Forb
Latin Name
Bromus purgans
Carex cristatella
Carex grayi
Carex sprengelii
Elymus canadensis
Elymus hystrix
Elymus villosus
Elymus virginicus
Glyceria striata
Juncus tenuis
Ageratina altissima
Allium tricoccum
Anemone canadensis
Anemone virginiana
Aquilegia canadensis
Arisaema triphyllum
Aster cordifolius
Blephilia hirsuta
Campanula americana
Caulophyllum thalictroides
Eupatorium purpureum
Geranium maculatum
Helianthus grosseserratus
Impatiens capensis
Lobelia siphilitica
Monarda fistulosa
Oenthera biennis
Penstemon digitalis
Polygonatum canaliculatum
Rudbeckia lacinata
Rudbeckia triloba
Smilacina racemosa
Solidago caesia
Solidago flexicaulis
Solidago rugosa
Thalictrum dioicum
Veronicastrum virginicum
Zizia aurea
77
Common Name
Hairy Wood Chess
Crested Sedge
Common Bur Sedge
Long-beaked Sedge
Canada Wild Rye
Bottlebrush Grass
Silky Wild Rye
Virginia Wild Rye
Fowl Manna Grass
Path Rush
White Snakeroot
Wild Leek
Canada Anemone
Tall Thimblweed
Wild Columbine
Jack-in-the-pulpit
Heart-leaved Aster
Hairy wood mint
Tall Bellflower
Blue Cohosh
Sweet Joe Pye Weed
Wild Geranium
Saw-thoothed Sunflower
Spotted Touch-me-not
Great Blue Lobelia
Wild Bergamot
Common evening-primrose
Foxglove Beardtongue
Great Solomon's Seal
Golden Glow
Brown-eyed Susan
False Solomon's Seal
Blue-stemmed Goldenrod
Zigzag Goldenrod
Rough Goldenrod
Early Meadow Rue
Culver's Root
Golden Alexander
Table 2.1. (cont’d)
Annual Nurse Crop
Annual Nurse Crop
Avena sativa
Lolium multiflorum
Seed Oats
Annual Rye
78
CHAPTER 3
Exploring demographic and genetic factors for the establishment of Hypena opulenta for the
biological control of invasive swallow-wort vines (Vincetoxicum spp.)
Abstract
About 40% of exotic herbivorous insects introduced for biological control of invasive weeds
do not establish. Demography, such as release size and the frequency of introductions is a well-
known factor in mediating establishment success in diverse taxa. Genetic diversity of the
released populations and their chances of establishment are also positively correlated in some
species. However, most evidence for the effects of demography and genetics in establishment
success comes from laboratory studies. First releases of newly approved biological control
agents can be used to test the importance of these factors for establishment and persistence in the
field. We conducted field releases of Hypena opulenta, a biological control agent against
invasive swallow-wort vines (Vincetoxicum spp.), to test the roles of demography and genetics in
establishment in the field. In 2020, releases were conducted at 18 field sites in southern
Michigan using H. opulenta populations with two distinct genetic backgrounds (a laboratory and
outbred population) and at two different release sizes (5 or 20 pairs of adults) using a factorial
design. Monitoring later in the season in 2020, the following two years in 2021 and in spring of
2022 indicated that these releases were not successful. Thus, we could not assess the combined
effects of demography and genetics in establishment. In 2022, we released replicated H. opulenta
populations with three distinct genetic backgrounds (lab, field, and outbred) at 15 field sites to
test how genetic background may affect establishment. For the 2022 releases, H. opulenta was
released all through the season, starting in early June until August using both adult and larval
stages for a total of 34-37 females, 32-35 males and 37 larvae per site. In the release cages, the
presence of any larval feeding and the extent of defoliation was measured. The outbred
79
population exhibited similar performance to the field population, although the outbred population
had the highest mean defoliation overall. As of spring 2024, establishment success could not be
confirmed. Given that the build-up of biocontrol agent populations to densities where they are
consistently detectable can take multiple years continued monitoring of the release sites will be
necessary to evaluate the importance of genetic background for establishment success.
Nevertheless, this study provides guidance how biological control agent releases could be used to
test various mechanisms that can mediate establishment with the ultimate goal of increasing the
success and impact of new biocontrol programs.
1. Introduction
The long-term management of invasive species is most efficiently achieved through the
introduction of natural enemies that have co-evolved with them in their native habitats. Despite
the long history of the use of classical biological control agents against invasive species, little is
known about what factors influence their successful establishment. Roughly 40% of weed
biocontrol agents fail to establish and only 24% of established agents have a high impact on the
target weed (Schwarzländer et al., 2018). Several factors can affect the successful establishment
of species colonizing novel environments including environmental conditions and random
stochastic events, as well as the genetic background and demography of introduced populations.
Propagule pressure—defined as the cumulative size and frequency of introductions—emerges as
the most reliable indicator of successful establishment (Blackburn et al., 2015; Lockwood et al.,
2005; Simberloff, 2009). While the influence of genetic background in establishment success in
the field is largely unknown.
Laboratory experiments using insects as model organisms have confirmed field observations
that larger initial populations enhance the likelihood of successful establishment (Hufbauer et al.
80
2013; Szűcs et al. 2014; Szűcs et al. 2017). Similarly, few field trials in weed biocontrol systems
have shown that larger releases of control agents tend to result in better establishment and
persistence (Grevstad, 1999; Grevstad, 2006; Memmott et al. 2005; Memmott et al. 1998). For
example, establishment success of a psyllid released for the control of Arytainilla spartiophila in
New Zealand generally increased with increasing release sizes that ranged from 2-270
individuals but some of the smallest releases led to establishment as well (Memmot et al., 2005).
This outcome may stem from several reasons, such as diminished Allee effects, and increased
resiliency to environmental stochasticity in larger populations (Boyce et al. 2006; Lande, 1993).
Moreover, a larger founding population often encompasses greater genetic diversity, potentially
mitigating inbreeding and enhancing adaptability to new environments, thereby promoting the
long-term survival of the population (Drake, 2006; Forsman, 2014; Rius and Darling, 2014).
Despite these trends, there are instances where small introductions succeed, as noted above in the
psyllid biocontrol agent, and where large-scale releases involving 65,000 individuals failed to
establish for a planthopper biocontrol agent (Grevstad et al., 2011). In instances of unsuccessful
establishment despite high release size, a critical, yet overlooked aspect of the habitat may affect
establishment, such as with planthoppers (Grevstad et al., 2011). Similarly, incidences of
establishment at low release sizes indicate that propagule pressure is not the only driving factor
behind establishment.
Genetic background and initial population release size of the colonizing populations stand
out as pivotal, yet underexplored, determinants of success (Szűcs et al., 2014; Szűcs et al., 2019).
Incorporating new genetic material into initial populations can mitigate the effects of inbreeding,
boost adaptive capabilities, and generate new genotypes better adapted to unfamiliar settings
(Barker et al., 2019; Sinclair et al., 2019; Stewart et al., 2017; Szűcs et al., 2017). One of the
81
mechanisms leading to heterosis is the masking of fixed recessive deleterious alleles by
dominant fitter alleles in sexually reproducing species (Dlugosch et al., 2015). Laboratory
research demonstrated that populations created from genetic admixture benefit from enhanced
establishment rates and increased survival (reviewed in Szűcs et al., 2014; Szűcs et al., 2017).
For example, laboratory experiments with Tribolium castaneum not only found a general positive
effect of outbreeding on the establishment and persistence of populations, but they also found
that small founding sizes reduced fitness (Szűcs et al., 2017). Additionally, genetic mixing in
pheasant populations has triggered a 'catapult effect', wherein a temporary boost in fitness in
early generations arises from heterosis or 'hybrid vigor', thereby aiding in the population's
endurance (Drake, 2006). Few studies investigated the significance of genetic variation on
establishment success in the field and there remains a substantial gap in the literature concerning
the combined and individual roles of demography and genetic background on establishment.
Biocontrol agents often suffer the negative effects of genetic drift and inbreeding that can
lead to reduced genetic diversity, lower fitness and limit adaptive potential (Hufbauer and
Roderick, 2005; Szűcs et al, 2019). These effects stem from the common practice of sourcing
biological control agents in small numbers from native populations and mass-rearing them for
release over tens or hundreds of generations, potentially diminishing their adaptive ability to new
environmental conditions (Bertin et al., 2017; Frankham, 2005; Szűcs et al., 2019). Empirical
evidence from laboratory studies supports the hypothesis that increased genetic diversity and
larger founding sizes can increase establishment and persistence, but these studies are not
conducted with any biocontrol agents (Hufbauer et la., 2013; Szűcs et al., 2014). Initial releases
of newly approved biocontrol agents lend themselves to experimentation to test mechanisms that
can mediate establishment success and persistence in novel environments. Insights obtained from
82
studies on biocontrol agents can inform our understanding of invasion dynamics, as well as
conservation and restoration efforts for other colonizing species (Szűcs et al, 2019).
We conducted releases of Hypena opulenta Christoph (Lepidoptera: Erebidae), a biological
control agent deployed against swallow-wort vines (Vincetoxicum spp.) in an explicitly
experimental way to test the roles demography and genetics may play in establishment success
and their impact on swallow-wort. We created genetically distinct populations by crossing
individuals from a long-term laboratory reared population (a lab population) and individuals
from a laboratory reared population that has been established in the field for 10 generations (a
field population) to create an outbred population. The genetic differentiation between the
parental and outbred populations and the increased heterozygosity of the outbred population was
confirmed through molecular analyses (see Chapter 1). In 2020, we released two genetically
distinct populations of H. opulenta (the lab population and the outbred population) in two
different release sizes (low and high) at 18 field sites in Michigan. We hypothesized that the
outbred population would exhibit higher levels of establishment than the lab population. And that
higher release sizes would result in higher levels of establishment than low release size. After we
failed to detect any signs of establishment in 2020 or 2021, we released again in 2022 using a
modified experimental design. In 2022, we omitted release size to solely focus on how genetic
background alone might influence establishment and impact. We expected that the outbred
population would have higher or at least similar establishment success as the field population and
that the lab population would have the lowest rates of establishment.
83
2. Methods
2.1 Study System
Invasive swallow-wort vines from the Vincetoxicum genus, including black swallow-wort
(Vincetoxicum nigrum (L.) Moench) and pale swallow-wort (Vincetoxicum rossicum (Kleopow)
Barber.), have emerged as significant ecological threats in their non-native ranges, especially in
the northeastern United States and southeastern Canada. Originating from Europe, these hardy
and adaptable perennial vines disrupt native ecosystems by forming dense mats that outcompete
native flora, reduce biodiversity, and alter habitats for local fauna. Their presence is notably
problematic for agricultural lands and poses risks to livestock due to their toxicity (Ditommaso et
al., 2005; Ditommaso et al., 2018; Ernst & Cappuccino, 2005; Weed & Casagrande, 2010).
Managing these invasive species is challenging due to their robust root systems and resistance to
herbicides, making mechanical removal and chemical treatments ineffective (Ditommaso et al.,
2005; Lawlor & Raynal, 2002; Weed & Casagrande, 2010). Introduced to North America with
regulatory approval and after comprehensive testing, the moth species H. opulenta has been
released as an alternative control method (Bourchier et al., 2019).
The biological control agent Hypena opulenta received approval from the United States
Department of Agriculture (USDA), Animal and Plant Health Inspection Service (APHIS) for
release in the US in 2017 (Stewart, 2017). This agent targets swallow-wort across all five of its
larval stages by feeding on the plant's leaves before transitioning to pupation either in the soil or
within leaf structures. Emergence is asynchronous and diapause is triggered primarily by
photoperiod (Weed & Casagrande, 2010). Host-specificity trials have confirmed that H. opulenta
poses no risk to economically significant or native North American species (Hazlehurst et al.,
2012). Research involving artificial defoliation demonstrated that repeated leaf removal
84
significantly diminishes biomass and seed output in both Vincetoxicum rossicum and
Vincetoxicum nigrum under low-light conditions within greenhouse settings (Milbrath, 2008).
Field studies over two years observed a direct correlation between the degree of leaf clipping on
swallow-wort and a reduction in seed production (Doubleday & Cappuccino, 2011). Further
evaluations revealed that as few as two larvae could substantially decrease above-ground
biomass and reproductive capability of V. rossicum, alongside diminished floral production in V.
nigrum, within controlled laboratory conditions (Weed & Casagrande, 2010). The capacity of H.
opulenta to undergo two generations annually, with evidence indicating generational overlap,
provides optimism that this agent could effectively impair the growth and reproductive success
of V. rossicum and V. nigrum over extended periods (Weed & Casagrande, 2010). However,
changes in photoperiod of the release environments from the natal environment and high
predation rates have emerged as barriers to establishment and persistence (Jones et al., 2020;
Jones et al., 2022).
Before shipment to Michigan, the first lab population of Hypena opulenta was created from
wild caught individuals in Donetsk, Ukraine in 2006 from four pupae and 32 larvae (Weed et al.,
2011). Population maintenance started at CABI EU-CH (Delémont, Switzerland) before
shipment to the University of Rhode Island (URI) which then started quarantine and population
rearing in 2008 (Hazlehurst et al., 2012). URI provided a larval shipment to Michigan State
University (MSU; East Lansing, MI) in June of 2018 which founded the MSU population with
18 females and 22 males.
In 2020, additional H. opulenta specimens from a Canadian field population, which was
established from the same initial stock as the MSU population but six years earlier in 2012, were
sent to MSU (Bourchier et al., 2019). After successful field releases in Canada in 2013 and
85
establishment by May 2015, specimens collected in 2019 underwent one generation of rearing at
URI before being shipped to MSU as pupae in May 2020, founding a population with 15 females
and 13 males. During the summer of 2020, reciprocal crosses were setup between the laboratory-
raised and the Canadian field-collected H. opulenta populations. The original MSU population
from 2018 is referred to as the "lab population," while the population from Canada is termed the
"field population." Crosses between these two groups is referred to as the "outbred population"
(see Chapter 1 for details).
For moth rearing, swallow-wort plants collected from various infestations in Michigan,
including Grand Rapids and Oakland County, were transplanted into 1.14 L pots with
SUREMIX® perlite mix and grown in a greenhouse at 26.6±4.9°C from March to September. In
the fall, these plants were moved outdoors at the MSU Entomology Farm and returned to the
greenhouse in March/April of the following year for repotting and use in rearing. Rearing
occurred in mesh cages (40 x 40 x 60 cm, Restcloud) in a lab maintained at 22-25˚C, with each
cage containing six potted plants on a tray of soil, under a 16:8h light-dark photoperiod with
supplemental lighting. Adult moths were fed a honey-water mixture, with the feeding solution
changed regularly and cages misted daily.
From March to August annually, up to five pairs of adult moths were introduced into each
cage for oviposition. Larvae fed on the potted plants and supplementary cut swallow-wort stems
inserted into water picks. Initially, these stems were sourced from potted plants, transitioning to
field-collected stems from late May. Larvae were moved to ventilated plastic containers for
pupation after reaching the 3rd-5th instar stage. Pupae collected from these containers and the
cages were sexed and placed in small containers with moist vermiculite under controlled
photoperiods. To induce diapause, pupae containers were kept in incubators where the initial 20
86
C rearing temperature was reduced by 5˚C each monthly between September and December and
kept at 5˚C from December to April/May until the rearing of the next generation started.
2.2 Experimental Field Releases in 2020
To evaluate the influence of genetic and demographic factors on the establishment success of
H. opulenta, we conducted field experiments in 2020 across 18 locations, utilizing populations of
H. opulenta with two genetic backgrounds: 1) a lab population, and 2) an outbred population.
The lab population had been reared for approximately eight generations at MSU before release
and the outbred population constituted second generation hybrids. We applied a factorial design,
pairing these two genetic backgrounds with two release sizes—either 5 or 20 pairs of adults—
resulting in 4-5 replications per treatment combination. The 18 field sites were divided into four
blocks based on geographic proximity. The distribution of treatments across the sites followed an
unbalanced block design. In the first three blocks, each of the four treatments (5-pair lab, 5-pair
outbred, 20-pair lab, 20-pair outbred) was implemented once. Block 4 lacked the 20-pair outbred
treatment, and block 5 omitted the 20-pair lab treatment due to site availability constraints. The
releases occurred within 1m2 quadrats enclosed by mesh cages to facilitate mating and egg-
laying, with cages removed after two weeks to allow larvae to disperse. Releases were conducted
on June 24th, 2020. The experiments took place in Oakland County with sites mostly located in
forested areas ranging from full shade to edge habitats on private properties or beside trails. All
locations were infested with pale swallow-wort (Vincetoxicum rossicum).
To monitor population level effects of H. opulenta, eight 0.5m2 permanent quadrat plots were
established at each site, with half marked to be treated with insecticide to serve as controls and
the other half marked to be left untreated to allow for H. opulenta feeding. These plots were
randomly arranged around the release areas within a 10m radius of the release site. To monitor
87
impact of H. opulenta on the individual plant level five swallow-wort plants were marked within
each plot. Within each plot the number of swallow-wort stems were counted at the start and end
of the growing season, and the percent defoliation was estimated on 10 stems. For the
individually marked plants the number of seed pods were counted at the end-of-season. Plots and
individual plants were monitored two times following the releases in 2020, monthly during 2021
and in May in 2022.
2.3 Experimental Field Releases in 2022
Field releases were repeated in a modified design by omitting the release size treatment in
2022 after establishment failed from the 2020 releases. To test the effect of genetic background
on establishment field releases took place at 15 locations, using three populations of H. opulenta:
1) the lab 2) the field, and 3) the outbred populations. Before release the lab population had been
reared for approximately 14 generations at MSU, the field population for eight generations, and
the outbred population constituted seventh generation hybrids. An unbalanced block design was
used whereas the first block contained three replicates of each treatment and the second block
contained two replicates of each treatment (n = 5) (Fig. 3.1). The 15 sites used for releases in
2022 were the same where the 2020 releases took place that were deemed unsuccessful. Changes
in treatment assignments at each location from 2020 to 2022 can be found in Table 3.1. In
general, an attempt was made to match the genetic background of releases at given sites between
the 2020 and 2022 releases to account for the possibility of establishment from the first release
that might have gone undetected. The outbred population was released at sites that were assigned
to the ‘low’ release size of 5 pairs of adults in 2020 because establishment from the first releases
were least likely at these sites that might have interfered with the later releases. The releases
occurred within 1m2 quadrats enclosed by mesh cages to facilitate mating and egg-laying. Both
88
adults and larval stages were released from June through August. Cages were kept on until
August 2nd. Release times and release sizes are detailed in Table 2. In sum, 37 adult females, 32
males and 37 larvae were released at each site for the field and outbred populations, and 34
females and 35 males for the lab population. We attempted to release equal numbers at the same
time for each population treatment but were restricted due to the availability and emergence time
of individuals, particularly from the lab population. For this reason, the timing and release sizes
of the field and outbred populations are in good agreement but there is a delay and initially lower
release sizes for the lab population (Table 3.2).
Defoliation by offspring of the released adults and from the larval releases were assessed
twice: once early in the season to capture defoliation from the first two releases and once late in
the season to capture defoliation from the last two releases. This was done by selecting ten
random stems within each release cage and estimating the average defoliation per stem. A
subsequent release occurred following the removal of the cage, with late-season defoliation
assessments carried out about one month after release in the same area previously enclosed by
the cage. The cages were removed due to the observation that intense feeding within some cages
had led to a decrease in available swallow-wort biomass for further feeding. Stem density counts
were taken in the release cages at the time of the first release to account for any effects in food
availability. To account for site differences, sites were categorized by their habitat type. Sites
were categorized as ‘roadside’ if they were within 5m of a road with motorized vehicles. If a site
was within 5m of a walking trail or campsite, it was categorized as an ‘edge’ site. Sites that were
more than 5m from any trail or roadside were categorized as ‘rural’.
89
Figure 3.1. Map of the 2022 release sites. Boxes are placed next to the pinned release locations
with three border colors to indicate the populations released at that site. Boxes in blue indicate
releases of the lab population. Boxes in gray indicate releases of the field population. Sites
within the circles belong to either of two spatial blocks.
90
Table 3.1. Sites used for field releases in 2020 and in 2022. The genetic background and release
size treatments are indicated for the 2020 releases. The new genetic background treatment
assignment is indicated for the 2022 releases. Habitat types are given for each site (edge,
roadside, or rural). If sites were dropped from 2020 to 2022, their habitat type is classified as Not
Applicable or NA.
Site
Location
2020
2020
2022
2022
Habitat
Block
Treatment
Block
Treatment
Type
Assign-
ment
Assign-
ment
Bald
42.785675,
4
Outbred,
2
Lab
Edge
Mountain
-83.196377
20
Burkhart
42.6742565,
3
Lab, 5
Dropped Dropped
NA
-83.6733341
Camp
42.736288,
3
Outbred, 5
1
Lab
Edge
Wathana
-83.61545
Clark Road
42.741395,
5
Outbred,
1
Lab
Rural
-83.485015
20
Davisburg
42.749908,
5
Outbred, 5
1
Outbred
Roadside
Road
-83.553234
Giroux Street 42.776174,
1
Lab, 20
1
Lab
Rural
-83.477617
91
Table 3.1 (cont’d)
Site
Location
2020
2020
2022
2022
Habitat
Block
Treatment
Block
Treatment
Type
Assign-
ment
Assign-
ment
Holly
42.823514,
1
Outbred, 5
1
Field
Edge
Recreational
-83.524798
Area
Indian Lake
42.797983,
4
Lab, 5
Road
-83.226714
Lahring
42.843221,
1
Lab, 5
Road
-83.596098
2
1
Outbred
Roadside
Field
Roadside
Munger
42.7052562,
3
Lab, 20
1
Outbred
Rural
Road
-83.6256498
Orion Road
42.743146,
4
Outbred, 5 Dropped Dropped
NA
-83.172950
Ormond
42.703515,
5
Lab, 5
1
Field
Rural
Road
-83.54906
Polly Ann
42.804251,
2
Outbred, 5
2
Lab
Edge
Trail North
-83.280229
92
Table 3.1 (cont’d)
Site
Location
2020
2020
2022
2022
Habitat
Block
Treatment
Block
Treatment
Type
Assign-
ment
Assign-
ment
Polly Ann
42.769441,
2
Outbred,
2
Outbred
Edge
Trail Center
-83.278723
20
Polly Ann
42.734927, -
2
Lab, 5
2
Field
Edge
Trail South
83.283541
Turkey
42.817365,
1
Outbred,
1
Outbred
Rural
Track
-83.495402
20
Roundabout
42.761100,
3
Outbred,
Dropped Dropped
NA
-83.567001
20
Spezia Drive
42.8219639,
2
Lab, 20
2
Field
Rural
-83.2867473
Table 3.2. The number and time of adult and larval releases at each of 15 sites for three
genetically distinct populations of H. opulenta in southern Michigan in 2022.
Release
Date
Life Stage
Lab
Field
Outbred
Female
Male
Female
Male
Female
Male
6/1/2022
Adult
6
5
19
10
19
10
6/9/2022
Larva
0
37
37
93
Table 3.2 (cont’d)
Release
Date
Life Stage
6/27/2022
Adult
8/25/2022
Adult
Total
Adult
Total
Larva
Total
2.4 Statistical Analyses
Lab
Field
Outbred
Female
Male
Female
Male
Female
Male
16
12
34
14
16
35
6
12
37
6
16
32
6
12
37
6
16
32
0
69
37
106
37
106
All analyses were performed in R 4.3.0 (R Core Team, 2023). We used the DHARMa
package to calculate the residuals to assess best model fit for all non-parametric models (Hartig,
2022). Models were also assessed using the Anova functions from the car package (Zeileis &
Hothorn, 2002). All post-hoc pairwise comparisons were performed using the emmeans package
(Lenth et al., 2021).
For the 2022 releases, feeding success of the different populations was evaluated by
categorizing release cages as"0" if no feeding was observed and “1" if feeding activity was
present. A generalized linear mixed model (GLMM) with a binomial distribution, implemented
using the lme4 package (Bates et al., 2015), was used for analysis. The model included an
interaction between genetic background (field, lab, and outbred) and season (early or late) plus
habitat type as fixed effects. Stem density nested within site was used as a random effect.
To compare feeding activity among populations with different genetic backgrounds for the
2022 releases we analyzed total defoliation per stem as a proportion (ranging from 0 to 1) using a
94
GLMM with a beta distribution in the glmmTMB package in R (Brooks et al., 2017). This
approach incorporates a zero-inflation parameter to accommodate data values of zero, which are
possible despite beta distributions being theoretically bound between values greater than zero
and less than one. The zero-inflation parameter was adjusted based on fixed and random effects.
The model included an interaction between genetic background (field, lab, and outbred) and
season (early or late) plus habitat type as fixed effects. Stem density nested within site was used
as a random effect.
3. Results
No sign of feeding was detected from the 2020 release experiment. For the 2022 field
releases, feeding success varied significantly across populations (χ2 = 20.83, df = 2, p < 0.001),
seasons (χ2 = 24.39, df = 1, p < 0.001), and by habitat type (χ2 = 10.93, df = 2, p < 0.01) (Fig.
3.2). Early in the season, feeding damage was more likely to be present at sites where the field
and outbred populations had been released compared to sites with the lab population (pairwise
comparisons: p < 0.001). This effect can be attributed to the variation in release numbers for the
lab population. However, later in the season there were not significant differences in the presence
of feeding among populations (pairwise comparison: p > 0.05). The different populations showed
different feeding success across the season (population*season interaction: χ2 = 15.13, df = 2, p <
0.001). The lab population increased feeding success from the early to late season (pairwise
comparison: p < 0.001), likely due to release number variation. The outbred population also
increased feeding success from early to late season (pairwise comparison: p < 0.01) but the field
population did not show significant differences in feeding between seasons (pairwise
comparison: p = 0.521). Across all populations and seasons, releases at rural sites were more
95
successful than releases at either edge sites (pairwise comparison: p < 0.01) and roadside sites
(pairwise comparison: p < 0.05) in terms of showing any feeding success.
Figure 3.2. The frequency at which stems were recorded as having some defoliation (present) or
no defoliation (absent) for each population (lab, field, outbred) and season (early or late) in 2022.
Each bar is divided by habitat type (edge, roadside, or rural). Please see Table 3.1 for a list of the
total number of sites per habitat and population, which varied. At each site, ten stems were
recorded.
Defoliation levels did not differ significantly between populations (χ2 = 4.60, df = 2, p =
0.10) but did vary by season (χ2 = 13.60, df = 1, p < 0.001) and habitat type (χ2 = 10.74, df = 2, p
< 0.01) (Fig. 3.3). Early in the season, the field and outbred population had nearly identical
96
feeding rates. However, the lab population had significantly lower defoliation rates compared to
the field population (pairwise comparison: p < 0.05) and the outbred population (pairwise
comparison: p < 0.05). Again, this can be attributed to variations in release size of the lab
population. These differences disappeared later in the season. The different populations showed
marginally different feeding patterns across the season (population*season interaction: χ2 = 5.67,
df = 2, p = 0.059). By season's end, both the lab and outbred populations had doubled their total
defoliation rates (pairwise comparisons: p < 0.05), whereas the field population did not show
significant variation in feeding from early to late season (pairwise comparison: p = 279). Across
all populations and seasons, releases at rural sites had higher defoliation rates than releases at
edge sites (pairwise comparison: p < 0.01) but only negligibly so at roadside sites (pairwise
comparison: p = 0.07).
97
Figure 3.3. Defoliation of stems from each of the three populations (lab, field, and outbred) over
two seasons (early and late). For each season and population, separate boxes are given for habitat
type (edge, roadside, or rural). Note that the lab population had no sites that were categorized as
roadside. Large dots indicate outlier observations and small dots indicate the individual values.
The horizontal line indicates the median with the box representing the interquartile range, and
vertical lines are 1.5 times the interquartile range.
4. Discussion
Our conclusions are constrained due to the inability to verify the overwintering success
following the releases in both 2020 and 2022. To date, H. opulenta establishment has only been
confirmed at one site in Ottawa, Ontario, Canada where approximately 2000 larvae were
released between 2013 and 2014 (Bourchier et al., 2019). Even with these numbers, it took until
2015 to successfully confirm establishment (Bourchier et al., 2019) so it is unsurprising that we
98
did not detect overwintering success from our relatively small releases within 3 years. The lack
of establishment to date may stem from a complex interplay of factors such as predation, climate,
and photoperiod, all of which significantly impact the survival and efficacy of H. opulenta (Jones
et al., 2020; Jones et al., 2022; Weed & Casagrande, 2010). The impact of predation on H.
opulenta is an important consideration for its successful establishment and effectiveness in target
areas. Predation can significantly influence the survival rates of H. opulenta at various life
stages, from eggs to adults, potentially affecting its population density and control efficiency
(Jones et al., 2022). Studies on biological control agents often highlight the role of native
predators in shaping the outcomes of biocontrol introductions (Louda et al., 2003). For instance,
the presence of generalist predators, such as certain species of ants and beetles, can lead to higher
mortality rates in biocontrol agents, thereby requiring careful assessment of the ecological
context prior to release (Van Driesche et al., 2008). Predation presents a considerable challenge,
as evidenced by Jones et al. (2022), who found that H. opulenta pupal mortality rates soared to
70% during summer due to predation, contrasting with the lower winter predation mortality.
However, non-predation mortality was notably high (around 66%), especially during colder
winters. This variance underscores the vulnerability of H. opulenta to predation and climate
which can adversely affect its population density and control effectiveness.
Photoperiod plays a pivotal role in the life cycle of H. opulenta by triggering diapause (Weed
& Casagrande, 2010). The critical photoperiod, or critical threshold, for H. opulenta is 15 h 35
min, triggering 50% of individuals to enter diapause. In the introduced range, photoperiod does
not reach the same maximum length as it does in Donesk, Ukraine (16 h 2 min) where H.
opulenta was originally collected from (Jones et al., 2022). However, where H. opulenta has
established in Ottawa, Ontario the photoperiod is still above the critical threshold (15 h 40 min)
99
(Jones et al., 2022). Whereas the photoperiod in East Lansing, Michigan is below the critical
threshold (15 h 20 min) (Jones et al., 2022). The probability of diapause at the maximum day
length in Michigan is still approximately 75% (Jones et al., 2022). In Michigan, two generations
of H. opulenta are feasible, with releases conducted prior to the summer solstice being more
advantageous than those made post-solstice to minimize the early entry of H. opulenta into
diapause (Alred et al., 2022). Minimizing early diapause is two-fold. Firstly, premature diapause
can heighten vulnerability to predation and environmental conditions, highlighting the
importance of understanding these mechanisms to improve the utilization of H. opulenta as a
biological control method (Jones et al., 2022; Louda et al., 2003; Van Driesche et al., 2008).
Secondly, high diapause frequency detracts from the impact of the biocontrol agent (Jones et al.,
2020). In 2020, we released very close to the summer solstice. While adults in this release were
exposed to the maximum photoperiod in Michigan, larvae from those releases were exposed to
decreasing daylight which could have induced early diapause. In 2022 we amended this and
conducted releases early in June so that first generation larvae would be exposed to the most
daylight possible. Releases later in the season were made to attempt to supplement a second
generation if most of the H. opulenta from the first release went into diapause. Although we have
not detected any signs of establishment from the 2022 releases, continued monitoring might still
reveal establishment if 2023 levels were below detection.
The second experimental release provided limited data to how genetic background may have
influenced the presence and rates of feeding. The lower feeding rates of the lab population early
in the season could have been due to their later emergence in the laboratory rearing that resulted
in lower initial release size. However, by the end of the season the release sizes of the three
different H. opulenta populations were similar and at that time point there was no difference in
100
their feeding rates. The field and outbred populations had similar feeding rates, which was
generally consistent with results from a field cage experiment (see Chapter 1). Although the field
population exhibits significantly lower genetic variation than the outbred population, it appears
that environmental effects, that is pre-adaptation to the field environment, masks any negative
effects of lower genetic diversity may have had on performance. Seasonal defoliation patterns,
particularly the increase from early to late season in the outbred population, suggest a cumulative
effect, possibly driven by the emergence of a second generation that went undetected in our
experiments. Similar outcomes have been observed in past field cage studies involving H.
opulenta, where the emergence of a second generation resulted in significant increases in
defoliation rates later in the season (Alred et al., 2022). Consistent with previous studies, we
found that H. opulenta were more likely to succeed at sites that were primarily shaded and more
isolated (rural sites) (Alred et al., 2022). This is an important factor to consider when attempting
to minimize site effects.
The endeavor to establish Hypena opulenta as a biological control agent for invasive
swallow-wort vines illuminates the complex interplay of ecological, environmental, and
biological factors essential for its success. Predation, climate conditions, and the critical timing
of photoperiods are significant challenges, impacting the survival and efficiency of H. opulenta.
Further monitoring may help elucidate the genetic factors that drive the successful establishment
and persistence of colonizing populations. Despite facing setbacks in initial experiments, these
studies contribute to a growing body of knowledge, reinforcing the necessity of a multifaceted
approach in biological control strategies. They also highlight the gaps in our understanding,
particularly regarding the interplay of genetic and demographic factors that affect establishment.
By refining our strategies to account for these factors, we can enhance the potential of H.
101
opulenta to serve as an effective biological control agent, paving the way for more sustainable
management of invasive species.
102
LITERATURE CITED
Alred B, Hufbauer RA, Szűcs M. 2022. Potential impact and phenology of the biological control
agent, Hypena opulenta on Vincetoxicum nigrum in Michigan. Biocontrol Science and
Technology, 32:671–684. https://doi.org/10.1080/09583157.2022.2040950.
Barker BS, Cocio JE, Anderson SR, Braasch JE, Cang FA, Gillette HD, Dlugosch KM. 2019.
Potential limits to the benefits of admixture during biological invasion. Molecular
Ecology, 28:100–113. https://doi.org/10.1111/mec.14958.
Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4.
Journal of Statistical Software, 67:1–48. doi:10.18637/jss.v067.i01.
Bertin A, Pavinato VAC, Parra JRP. 2017. Fitness-related changes in laboratory populations of
the egg parasitoid Trichogramma galloi and the implications of rearing on factitious
hosts. BioControl, 62:435–444. https://doi.org/10.1007/s10526-017-9795-y.
Blackburn, T. M., Lockwood, J. L., & Cassey, P. 2015. The influence of numbers on invasion
success. Molecular Ecology, 24(9), 1942–1953. https://doi.org/10.1111/mec.13075
Bourchier RS, Cappuccino N, Rochette A, des Rivières J, Smith SM, Tewksbury L, Casagrande
RA. 2019. Establishment of Hypena opulenta (Lepidoptera: Erebidae) on Vincetoxicum
rossicum in Ontario, Canada. Biocontrol Science and Technology, 29:917–923.
https://doi.org/10.1080/09583157.2019.1608511.
Boyce, M. S., Haridas, C. v., Lee, C. T., Boggs, C. L., Bruna, E. M., Coulson, T., Doak, D.,
Drake, J. M., Gaillard, J. M., Horvitz, C. C., Kalisz, S., Kendall, B. E., Knight, T.,
Mastrandrea, M., Menges, E. S., Morris, W. F., Pfister, C. A., & Tuljapurkar, S. D. 2006.
Demography in an increasingly variable world. In Trends in Ecology and Evolution,
21(3), 141-148. https://doi.org/10.1016/j.tree.2005.11.018
Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, Skaug HJ,
Maechler M, Bolker BM. 2017. glmmTMB balances speed and flexibility among
packages for zero-inflated generalized linear mixed modeling. The R Journal, 9:378–400.
doi: 10.32614/RJ-2017-066.
Ditommaso, A., Lawlor, F. M., & Darbyshire, S. J. 2005. The Biology of Invasive Alien Plants in
Canada. 2. Cynanchum rossicum (Kleopow) Borhidi [= Vincetoxicum rossicum
(Kleopow) Barbar.] and Cynanchum louiseae (L.) Kartesz & Gandhi [= Vincetoxicum
nigrum (L.) Moench]. Canadian Journal of Plant Science, 243-263.
DiTommaso, A., Stokes, C. A., Cordeau, S., Milbrath, L. R., & Whitlow, T. H. 2018. Seed-
Dispersal Ability of the Invasive Perennial Vines Vincetoxicum nigrum and Vincetoxicum
rossicum. Invasive Plant Science and Management, 11(1), 10–19.
https://doi.org/10.1017/inp.2018.8
103
Dlugosch, K. M., Anderson, S. R., Braasch, J., Cang, F. A., & Gillette, H. D. 2015. The devil is
in the details: Genetic variation in introduced populations and its contributions to
invasion. Molecular Ecology, 24(9), 2095–2111. https://doi.org/10.1111/mec.13183
Drake, J. M. 2006. Heterosis, the catapult effect and establishment success of a colonizing bird.
Biol. Lett, 2, 304–307. https://doi.org/10.1098/rsbl.2006.0459
Ernst, C. M., & Cappuccino, N. 2005. The effect of an invasive alien vine, Vincetoxicum
rossicum (Asclepiadaceae), on arthropod populations in Ontario old fields. Biological
Invasions, 7(3), 417–425. https://doi.org/10.1007/s10530-004-4062-4
Forsman, A. 2014. Effects of genotypic and phenotypic variation on establishment are important
for conservation, invasion, and infection biology. Proceedings of the National Academy
of Sciences of the United States of America, 111(1), 302–307.
https://doi.org/10.1073/pnas.1317745111
Frankham R. 2005. Genetics and extinction. Biological Conservation, 126:131-140. DOI:
10.1016/j.biocon.2005.05.002.
Grevstad, F. S. 1999. Factors influencing the chance of population establishment: Implications
for release strategies in biocontrol. Ecological Applications, 9(4), 1439–1447.
https://doi.org/10.1890/1051-0761(1999)009[1439:FITCOP]2.0.CO;2
Grevstad, F. S. 2006. Ten-year impacts of the biological control agents Galerucella pusilla and
G. calmariensis (Coleoptera: Chrysomelidae) on purple loosestrife (Lythrum salicaria) in
Central New York State. Biological Control, 39(1), 1–8.
https://doi.org/10.1016/j.biocontrol.2006.03.007
Grevstad, F. S., Coombs, E. M., & Mcevoy, P. B. 2011. Revisiting Release Strategies in
Biological Control of Weeds: Are We Using Enough Releases? XIII International
Symposium on Biological Control of Weeds, 368–376.
Hartig F. 2022. _DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed)
Regression Models_. R package version 0.4.6, .
Hazlehurst AF, Weed AS, Tewksbury L, Casagrande RA. 2012. Host specificity of Hypena
opulenta: A potential biological control agent of Vincetoxicum in North America.
Environmental Entomology, 41:841–848. https://doi.org/10.1603/EN12093.
Hufbauer RA, Roderick GK. 2005. Microevolution in biological control: Mechanisms, patterns,
and processes. Biological Control, 35:227–239.
https://doi.org/10.1016/j.biocontrol.2005.04.004.
Hufbauer RA, Rutschmann A, Serrate B, Vermeil de Conchard H, Facon B. 2013. Role of
propagule pressure in colonization success: Disentangling the relative importance of
104
demographic, genetic and habitat effects. Journal of Evolutionary Biology, 26:1691–
1699. https://doi.org/10.1111/jeb.12167.
Jones, I. M., Seehausen, M. L., Bourchier, R. S., & Smith, S. M. 2020. The Effects of
Photoperiod on Diapause Induction in Hypena opulenta (Lepidoptera: Erebidae), a
Biological Control Agent Against Invasive Swallow-Worts in North America.
Environmental Entomology, 49(3), 580–585. https://doi.org/10.1093/ee/nvaa030
Jones IM, Smith SM, Bourchier RS. 2022. Assessment of pupal mortality in Hypena opulenta:
An obstacle for establishment of a classical biological control agent against invasive
swallow-worts. Ecological Entomology, 47:527–535. https://doi.org/10.1111/een.13137.
Lande, R. 1993. Risks of population extinction from demographic and environmental
stochasticity and random catastrophes. American Naturalist. 142(6):911-927.
Lawlor, F. M., & Raynal, D. J. 2002. Response of Swallow-Wort to Herbicides. Weed Science,
50(2), 179–185. https://about.jstor.org/terms
Lenth, R., Singmann, H., Love, J., Buerkner, P., & Herve, M. 2021. Emmeans: Estimated
marginal means, aka least-squares means. R package version 1 (2018).
Louda SM, Pemberton RW, Johnson MT, Follett PA. 2003. Nontarget effects – the Achilles' heel
of biological control? Retrospective analyses to reduce risk associated with biocontrol
introductions. Annual Review of Entomology, 48:365–396. DOI:
10.1146/annurev.ento.48.060402.102800.
Memmott, J., Fowler, S.V. 7 Hill, R.L. 1998. The effect of release size on the probability of
establishment of biological control agenets: forse thrips (Sericothrips staphylinus)
released against gorse (Ulex europaeus) in New Zealand. Biocontrol Science and
Technology, 8, 103-115.
Memmott, J., Craze, P. G., Harman, H. M., Syrett, P., & Fowler, S. v. 2005. The effect of
propagule size on the invasion of an alien insect. Journal of Animal Ecology, 74(1), 50–
62. https://doi.org/10.1111/j.1365-2656.2004.00896.x
Milbrath, L. R. 2008. Growth and reproduction of invasive Vincetoxicum rossicum and V. nigrum
under artificial defoliation and different light environments. Botany, 86(11), 1279–1290.
https://doi.org/10.1139/B08-092
R Core Team. 2023. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. https://www.R-project.org/.
Rius, M., & Darling, J. A. 2014. How important is intraspecific genetic admixture to the success
of colonising populations? Trends in Ecology and Evolution 29(4), 233-242. Elsevier Ltd.
https://doi.org/10.1016/j.tree.2014.02.003
105
Simberloff, D. 2009. The role of propagule pressure in biological invasions. Annual Review of
Ecology, Evolution, and Systematics, 40, 81–102.
https://doi.org/10.1146/annurev.ecolsys.110308.120304
Sinclair JS, Arnott SE, Millette KL, Cristescu ME. 2019. Benefits of increased colonist quantity
and genetic diversity for colonization depend on colonist identity. Oikos, 128:1761–
1771.https://doi.org/10.1111/oik.06308.
Stewart, G. S., Morris, M. R., Genis, A. B., Szűcs, M., Melbourne, B. A., Tavener, S. J., &
Hufbauer, R. A. 2017. The power of evolutionary rescue is constrained by genetic load.
Evolutionary Applications, 10(7), 731–741. https://doi.org/10.1111/eva.12489
Schwarzländer, M., Hinz, H. L., Winston, R. L., & Day, M. D. 2018. Biological control of
weeds: an analysis of introductions, rates of establishment and estimates of success,
worldwide. BioControl, 63(3). https://doi.org/10.1007/s10526-018-9890-8
Szűcs M, Melbourne BA, Tuff T, Hufbauer RA. 2014. The roles of demography and genetics in
the early stages of colonization. Proceedings of the Royal Society B: Biological Sciences,
281. https://doi.org/10.1098/rspb.2014.1073.
Szűcs, M., Melbourne, B. A., Tuff, T., Weiss-Lehman, C., & Hufbauer, R. A. 2017. Genetic and
demographic founder effects have long-term fitness consequences for colonising
populations. Ecology Letters, 20(4), 436–444. https://doi.org/10.1111/ele.12743
Szűcs M, Vercken E, Bitume EV, Hufbauer RA. 2019. The implications of rapid eco-
evolutionary processes for biological control - a review. Entomologia Experimentalis et
Applicata, 167:598–615. https://doi.org/10.1111/eea.12807.Van Driesche R, Hoddle M,
Center T. 2008. Control of Pests and Weeds by Natural Enemies: An Introduction to
Biological Control. Blackwell Publishing.
Weed AS, Casagrande RA. 2010. Biology and larval feeding impact of Hypena opulenta
(Christoph) (Lepidoptera: Noctuidae): A potential biological control agent for
Vincetoxicum nigrum and V. rossicum. Biological Control, 53:214–222.
https://doi.org/10.1016/j.biocontrol.2009.12.004.
Weed AS, Gassmann A, Leroux AM, Casagrande RA. 2011. Performance of potential European
biological control agents of Vincetoxicum spp. with notes on their distribution. Journal of
Applied Entomology, 135:700–713. https://doi.org/10.1111/j.1439-0418.2010.01594.x.
Zeileis A, Hothorn T. 2002. Diagnostic checking in regression relationships. R News, 2(3):7–10.
URL https://CRAN.R-project.org/doc/Rnews/.
106
CHAPTER 4
Navigating the ecological trap: monarch butterfly (Danaus plexippus) behavior in the presence of
invasive swallow-wort vines (Vincetoxicum spp.)
Abstract
The monarch butterfly (Danaus plexippus) is faced with declining populations of its
milkweed host plants across its range. At the same time, populations of invasive species, such as
swallow-wort (Vincetoxicum spp.) vines are increasing that are related to milkweeds. Oviposition
choices of herbivorous insects, including those of monarchs can be maladaptive, especially when
closely related plants are available en masse, and may not reflect survival probability of their
offspring. We investigated oviposition choices of the monarch butterfly in Michigan, focusing on
its preference for common milkweed over swallow-wort in the field. In addition, we attempted to
test in a laboratory experiment how larval exposure to swallow-worts during development may
affect the oviposition choice of adult monarchs, and whether the presence of the biocontrol
agent, Hypena opulenta, may deter oviposition. Field surveys found a marked preference for
milkweed, however, monarchs laid 8% of their eggs on swallow-wort across sites. There were
marked site differences with up to 17% of monarch eggs were laid on swallow-wort locally. The
laboratory experiments that aimed to test some underlying mechanisms for oviposition were not
successful. However, we documented a 16% mortality rate among larvae that consumed
swallow-wort alongside of common milkweed, suggesting that non-host chemical signals could
misleadingly attract monarchs and elicit feeding. This reveals an ecological trap for monarch
butterflies. Our findings contribute to a growing body of knowledge about the interplay between
monarch reproductive strategies, invasive plant species, and biocontrol agents, emphasizing the
need for comprehensive conservation strategies that address these complex ecological
interactions.
107
1. Introduction
The iconic monarch butterfly (Danaus plexippus), celebrated for its remarkable migration
from Mexican wintering grounds to breeding sites across North America, is currently facing a
significant conservation crisis. Recent population studies have highlighted a dramatic decline,
especially in the Midwest, where numbers have decreased by up to 58% (Pleasants &
Oberhauser, 2013). This decline is attributed to habitat destruction, increased application of
herbicides, and the rise of glyphosate-resistant crops, compounded by the invasion of non-native
species, all of which substantially reduce the availability of milkweed, the monarch's host plant
(Pleasants & Oberhauser, 2013; Zaya et al., 2017). Invasive swallow-wort vines (Vincetoxicum
spp.) compete with native flora including milkweeds but more importantly they are also related
to milkweeds and monarchs mistakenly lay eggs on them (Alred et al., 2022; Casagrande &
Dacey, 2007). In response to this urgent conservation challenge and the complex interplay
between monarchs and invasive plants, there has been a growing focus on understanding the
monarch's oviposition behavior to enhance habitat support along its migratory route.
The preference-performance hypothesis is thought to explain oviposition behavior, where
females select host plants for oviposition based on the maximization of survival and fitness of
their offspring (Greenstein et al., 2022; Thompson, 1988). However, plants and insects are
involved in a complex evolutionary arms race, attempting to improve survival and reproduction
(Ferrer-Paris et al., 2013; Labandeira, 1998; van der Linden, 2021; Whitfield et al., 2012). A
result of this dynamic is an ecological trap set for monarch butterflies (Alred et al., 2022;
Casagrande & Dacey, 2007). Oviposition behaviors are shaped by evolutionary and ecological
mechanisms and can lead monarchs to choose suboptimal oviposition sites. The monarch
butterfly has a propensity to deposit eggs on non-host plants, notably on swallow-wort species,
108
with as much as 15-25% of eggs being laid on these unsuitable hosts (Alred et al., 2022;
Casagrande & Dacey, 2007). Despite both swallow-worts and milkweeds belonging to the
Apocynaceae family and sharing morphological traits, swallow-worts fail to support monarch
larvae to maturity, suggesting that shared oviposition cues may misguide monarchs (Endress &
Bruyns, 2000; Mattila & Otis, 2003). These chemical cues can cause egg laying on less than
suitable or non-host plants leading to reduced fitness or mortality after hatching. The pipevine
swallowtail (Battus philenor) is known to oviposit on non-host plants that emit similar volatile
compounds as its host pipevine (Aristolochia), leading to larval mortality upon hatching
(Pellmyr, 1986). Non-host plants may share chemical cues with the host plants of butterflies,
leading to mistaken oviposition, due to a complex interplay of evolutionary strategies and
ecological interactions. This phenomenon can be understood through the lens of mimicry,
convergent evolution, and co-evolutionary dynamics, where plants and insects engage in an
ongoing arms race for survival and reproductive success. Mimicry, for instance, is a strategy
where non-host plants evolve to emit chemical signatures that resemble those of genuine host
plants. This adaptation serves as a defensive mechanism, reducing the likelihood of being
consumed by herbivorous larvae (Berenbaum & Zangerl, 1996). Convergent evolution also plays
a role, where unrelated plants independently develop similar chemical compounds as a result of
evolving under analogous environmental pressures. These similarities can inadvertently attract
butterflies that rely on specific chemical cues to identify suitable oviposition sites (van der
Linden, 2021; Whitfield et al., 2012). Additionally, closely related plant species may share
chemical profiles due to their shared evolutionary history, potentially confusing butterflies that
have evolved to cue into specific chemical signals for host plant identification (Ferrer-Paris et al.,
2013; Labandeira, 1998). These strategies highlight the delicate balance of co-evolution, where
109
plants and insects continually adapt to each other's defenses and counter defenses. Understanding
these interactions is crucial for conservation efforts, especially for species like the monarch
butterfly, whose populations are in decline.
In the evolutionary arms race, not only do plants engage in complex strategies, but insects
partake as well. The concept of phenotypic plasticity—where distinct genotypes yield varied
phenotypes in different environmental contexts (Pigluicci et al., 2006)—plays a pivotal role in
shaping insect behavior. This adaptability is illustrated in fruit flies (Drosophila melanogaster),
which adjust their behaviors based on environmental signals and past experiences, exemplifying
the essence of phenotypic plasticity (Dukas, 2008). Similarly, research on monarch butterflies
has shed light on how their varying preferences for different milkweed species are a result of
phenotypic plasticity and learning, enabling them to navigate the diverse habitats along their
migratory paths (Jones & Agrawal, 2019; Snell-Rood & Papaj, 2009). This adaptability,
however, can come with trade-offs in fitness, influenced by the female's previous experiences
with host plants. While it allows monarchs to lay eggs on a variety of milkweed species—
beneficial in times of scarcity—it also poses the risk of choosing host plants unsuitable for larval
development (Jones & Agrawal, 2019). Contrasting outcomes in earlier experiments, with some
finding monarch oviposition on swallow-wort species and others not, underscore the complexity
of such behaviors (Alred et al., 2022; DiTommaso & Losey, 2003; Mattila & Otis, 2003;
Casagrande & Dacey, 2007). Variations between studies could be attributed to differences in the
origin of the monarch populations examined or their prior exposure to swallow-worts, hinting at
a learned component to their oviposition choices (Alred et al., 2022; DiTommaso & Losey, 2003;
Mattila & Otis, 2003; Casagrande & Dacey, 2007). This learning behavior suggests that
monarchs encountering swallow-worts alongside milkweeds might change their oviposition
110
strategy, a hypothesis supported by the observed plasticity in host preference (Casagrande &
Dacey, 2007; Jones & Agrawal, 2019).
We investigated oviposition behavior of monarch butterflies in Michigan under natural
conditions at field sites where milkweeds and swallow-worts co-occur. In a lab study, we aimed
to test how exposure to swallow-worts during larval development may influence the oviposition
behavior of adult monarchs on swallow-wort plants that either had larvae of the biocontrol agent,
Hypena opulenta Christoph (Lepidoptera: Erebidae) larvae and feeding present or not. We
hypothesized that adult monarchs that had been exposed to swallow-worts during development
would be more likely to lay eggs on swallow-worts. Furthermore, we expected that the presence
of another defoliating caterpillar in the form of the biocontrol agent H. opulenta on swallow-
worts may deter oviposition to avoid larval competition and egg cannibalism (Alred et al. 2022,
Jones & Agrawal 2019).
2. Methods
2.1 Field Survey
To evaluate the frequency with which monarch butterflies deposit their eggs on swallow-wort
vines in Michigan, an egg survey was carried out across seven locations in Oakland County
(Table 4.1). These sites featured both pale swallow-wort (Vincetoxicum rossicum) and varieties
of milkweed, specifically common milkweed (Asclepias syriaca L.) and swamp milkweed
(Asclepias incarnata L.). The survey involved inspecting 50 randomly chosen stems each of
milkweed and pale swallow-wort, examining every part of the plant from the top down,
including flowers, seed pods, stems, and both surfaces of leaves. This detailed egg scouting took
place at the end of July in 2023, spanning four days. The chosen field sites encompassed a
111
diverse range of habitats, including forest peripheries, trail borders, and open areas alongside
roads.
Table 4.1. Monarch egg survey locations given by the site name and the associated latitude and
longitude.
Site
Latitude
Indian Lake Road
42.797983
Davisburg Road
42.749908
Maureen Lane
42.7087042
Munger Road
42.7052562
Clark Road
Turkey Track
Wood Trail
42.741395
42.817365
42.804251
Longitude
-83.226714
-83.553234
-83.5732006
-83.6256498
-83.485015
-83.495402
-83.280229
2.2. Lab Experiment
In 2023, a series of experiments were conducted to determine how exposure of larvae to
swallow-wort during development may affect monarch oviposition and to examine whether the
presence of H. opulenta larvae and feeding on these plants influenced their oviposition decisions.
In brief, monarch eggs and early instar larvae were collected in the field and were subjected to
two rearing treatments during development to adulthood. The two treatments involved
milkweed-only or the combination of milkweed and swallow-wort plants next to each other in
the rearing cages. Then the adults emerging from these two rearing treatments were offered two
plant treatments for oviposition: 1) a fresh, undamaged swallow-wort plant alone, or 2) a
swallow-wort plant with feeding damage by H. opulenta and with H. opulenta larvae present
112
alongside of a fresh, undamaged swallow-wort plant. The adult monarchs were allowed to
oviposit on either of these swallow-wort plant treatments if they chose to and after 72 h of
exposure, they were moved onto a milkweed plant to check egg laying capability to validate if
the lack of position on swallow-worts was due to choice or physiological constraints.
For the rearing experiment. monarch eggs and 1st-2nd instar larvae were collected around East
Lansing and in Oakland CO., Michigan starting on June 20, 2023, with subsequent collections
made as necessary to compensate for hatching failures or early larval losses. Collection involved
detaching milkweed leaves directly from stems in the field and placing them in water picks for
hydration. Field-collected larvae, sourced from areas without swallow-wort (East Lansing area),
were designated for the milkweed-only rearing process and those collected from mixed stands of
swallow-wort and milkweeds (Oakland CO.) were assigned to the combined plant treatment to
ensure that larvae that had hatched before entering the lab had not had any exposure to swallow-
wort. All experiments were conducted at ambient temperature and humidity in the laboratory
with supplemental lighting set at 16:8h L:D photoperiod (4ft, 80 Watt; Barrina®; Zhongshan,
Chin). Each treatment initially received twenty-five replicates, undergoing replacements due to
hatch failures or larval demise, with three replacements in the milkweed-only group and five in
the combined plant treatment. Upon hatching, larvae were individually housed in small mesh
cages (30.5 x 30.5 x 30.5 cm, Restcloud) equipped with a water pick containing a milkweed leaf.
For those in the combined plant treatment, cages also contained a water pick with a fresh pale
swallow-wort stem alongside milkweed. The water picks were stabilized in 1.14 L square plastic
pots filled with SUREMIX® all-purpose perlite mix (Michigan Grower Products Inc., United
States). In the combined plant treatment, milkweed and swallow-wort were placed adjacently
within the same pot. Milkweed leaves were changed daily, with additional leaves provided as
113
larval appetite increased in later stages. Swallow-wort stems were similarly maintained, with
new daily cuttings to ensure freshness and replacements were made upon signs of wilting.
Within their respective enclosures, larvae transitioned into the chrysalis stage, with the cut
stems preserved until the adult monarchs emerged. Throughout their development, adults were
provided with Gatorade dispensed to 2 oz cups with cotton wicks, refreshed every other day to
ensure hydration. Upon emergence, adults were grouped into large mesh cages (61 x 40 x 40 cm,
Restcloud) equipped with a potted mum plant for resting to facilitate mating. Group sizes were
carefully managed to include no more than two females and three males, or at minimum, one
female and two males. Mating was initiated as soon as an adequate number of adults were
available, allowing a 48-hour period for mating. Following this mating period, females were
relocated to separate tall mesh cages. Here, they were presented with the following setup: one
fresh cut stem of swallow-wort (at least 30 cm tall) and another from swallow-wort with H.
opulenta feeding damage, alongside two 4th to 5th instar H. opulenta larvae. Preparation for the
feeding damage involved allowing two H. opulenta larvae to feed on a swallow-wort stem in a
cage for 24 hours, after which the stem, along with the larvae, was introduced to the cage
containing the adult female monarch. Swallow-wort stems were each placed into individual
potting containers filled with soil, positioned at opposite ends of the cage. Females were given a
72-hour window to lay eggs before being transferred to a milkweed-exclusive environment to
verify egg-laying capability. Egg-laying observation cages were set up similarly, featuring a cut
milkweed stem of at least 30 cm in a water pick placed into a soil-filled pot for stability. Egg-
laying activity was monitored every 24 hours during these 72-hour periods. Due to high
mortality, just ten females from the larva exposed to the milkweed only rearing treatment
survived to the egg laying phase, four of which died before the 72-hour egg laying period
114
finished. Only three females from the combined plant treatment survived to the egg laying phase,
and two of those died before the 72-hour egg laying period had finished. In summary, six
monarch adults from the milkweed only treatment and one monarch adult from the combined
treatment were able to be assessed.
2.2 Statistical Analysis
To assess the prevalence of monarch oviposition on common milkweed and swallow-wort in
the field, a generalized linear model was used with a negative binomial distribution from the
lme4 package in R (Bates et al., 2015; R Core Team, 2023). Plant species (either milkweed or
swallow-wort) and site were used as fixed effects. After assessing that common milkweed and
swamp-milkweed had no significant differences in monarch oviposition, they were batched
together as ‘milkweed’ for comparison to swallow-wort. Model evaluations were completed
using Anova from the car package (Zeileis & Hothorn, 2002) and residual testing from the
DHARMa package (Hartig, 2022). Post-hoc pairwise comparisons were completed using
estimated-marginal means in the emmeans package with a Tukey adjustment (Lenth et al., 2021).
No formal statistical models were used on the lab experiment data as adult monarchs did not lay
any eggs.
3. Results
3.1 Field Survey
Monarch oviposition varied by plant species (χ2 = 59.01, df =1, p < 0.001) and site (χ2 =
38.85, df = 6, p < 0.001). On the 350 common milkweed and swamp milkweed stems examined
across 7 sites, 78 monarch eggs were found or a 22% oviposition rate (z = -3.95, p < 0.001). On
the 350 swallow-wort stems assessed, 7 monarch eggs were found, or a 2% oviposition rate (z =
-2.43, p < 0.001). Monarchs were 11 times more likely to lay eggs on milkweed than swallow-
115
wort (pairwise comparison: p < 0.001) but were shown to lay up to 8% of their eggs on swallow-
wort across all sites. However, there were notable site-specific effects. At only three sites did
monarchs lay eggs on swallow-wort, two of which had the highest combined egg counts (29 and
20 eggs). The other site had just 6 eggs on both swallow-wort and milkweed combined. Other
sites where monarchs laid eggs on only milkweed had moderate to low egg counts (equal to or
below 12). The site with the highest egg count on swallow-wort contributed to 71% of all eggs
found on swallow-wort across all surveyed sites and monarchs laid 17% of their eggs on
swallow-wort at this site.
In the laboratory experiment, adult female monarchs did not lay eggs on either swallow-wort
of milkweed. Despite this setback, we observed one behavior of interest in monarch larvae.
There was no mortality in larvae reared in the milkweed only treatment. However, there were 6
larvae that died in the swallow-wort exposure treatment (24% mortality rate). Four of which
were observed having actively fed from the swallow-wort leaves in their last few instars, despite
being provided with sufficient fresh milkweed (16% mortality rate) (Figure 4.1). Five adult
monarchs from the milkweed only treatment failed to emerge from their chrysalis or died during
or shortly after emergence (20% mortality rate). Just ten of the nineteen monarchs in the
swallow-wort exposure treatment emerged (47% mortality rate). At the time of emergence, the
cumulative mortality rate in the swallow-wort exposure treatment was 40%, double that of the
milkweed only treatment. Since this behavior was not anticipated, we believe that some feeding
on swallow-wort went undetected initially and it is possible that larvae that fed on swallow-wort
made it to the chrysalis stage before dying.
116
Figure 4.1. The left image shows a monarch larva resting on a fresh milkweed cutting after
feeding on a swallow-wort leaf (left). The right image shows a monarch larva resting underneath
a swallow-wort leaf with defoliation present on the opposite leaf.
4. Discussion
The field study revealed a marked preference of monarch butterflies for common
milkweed compared to pale swallow-wort. Observations from natural swallow-wort infestations
in Michigan suggest that the proportion of eggs laid on swallow-wort by monarchs could be as
high as 17% at certain locations. Site-specific variables, however, suggest that additional factors
influence the oviposition choices of monarchs. Although our lab experiment did not yield the
expected outcomes to elucidate these behaviors further, we uncovered a previously unknown
larval behavior. Specifically, we observed a 16% mortality rate among monarch larvae attributed
to consuming swallow-wort, a figure that is presumably underestimated since monitoring larval
consumption of swallow-wort was not an initial focus of the study.
117
Previous field research using potted swallow-wort indicated that up to 25% of monarch eggs
might be laid on swallow-wort in Michigan (Alred et al., 2022) and at approximately 6% in
pasture surveys in Rhode Island (Casagrande and Dacey 2007). Our results suggest overall lower
rates of oviposition across swallow-wort infestations in nature at 2%, yet strong special
differences exist with one site reaching 17% of egg laying on swallow-worts. Monarch
butterflies do exhibit spatial patterns in their oviposition behavior, largely influenced by the
distribution and availability of their primary host plant, milkweed (Pleasants & Oberhauser,
2013). Studies have shown that monarchs tend to lay eggs more frequently on milkweed plants
situated in open or edge habitats rather than in densely vegetated or shaded areas (Zalucki &
Lammers, 2010). This preference is thought to maximize the survival chances of their offspring,
as these habitats often offer higher quality host plants and reduced competition and predation
(Jones & Agrawal, 2019; Kessler & Baldwin, 2001; Zalucki & Lammers, 2010). Furthermore,
within a suitable habitat, monarchs may exhibit selectivity towards individual plants based on
plant size, age, and health, which can influence the nutritional quality of the plant for the
developing larvae (Ladner & Altizer, 2005). This approach to monarch oviposition is assumed to
maximize the reproductive success of their offspring by choosing sites that maximize survival.
Studies have shown that factors at the landscape level, such as habitat fragmentation and the
existence of agricultural areas, can influence where monarch butterflies choose to lay their eggs
by impacting the availability and accessibility of milkweed (Flockhart et al., 2015; Oberhauser et
al., 2017). Future research could benefit from expanding the diversity and number of sites
surveyed to further understand the site-specific factors that drive monarch oviposition patterns in
Michigan. Pale swallow-wort, which prefers shaded or semi-shaded areas and creates dense
thickets, presents an environment that is not typically favored by monarchs for egg-laying
118
(Ditommaso et al., 2005; Ditommaso et al., 2018). Moreover, the low oviposition rates observed
at many of our sites suggest that the number of stems examined may have been insufficient to
accurately capture monarch egg-laying behavior on swallow-wort. It is unclear from the field
survey the extent of role of chemical cues and learned behavior play in monarch oviposition,
except that they must play some role since there is prevalence of mistaken oviposition in
monarchs in Michigan. Understanding this aspect could be crucial in deciphering the nuances of
monarch butterfly oviposition decision-making process.
Our experiment did not shed light on the learning behavior in monarch butterflies due to the
absence of egg-laying activity among the female monarchs involved in the study. This outcome
suggests that the adult butterflies might have belonged to the migratory generation, which does
not reproduce, or that an environmental variable during laboratory rearing inhibited their
oviposition. A preliminary trial involving a single adult female monarch, which successfully laid
eggs on milkweed, leads us to speculate that the generation of monarchs used in our main
experiment was not inclined to lay eggs under natural conditions either. However, we were able
to observe an undocumented behavior in monarch larvae. About 16% monarch larvae mortality
in our experiment can be attributed to direct consumption of swallow-wort. This finding supports
the idea that swallow-wort might produce chemical signals that attract monarchs, a theory
previously considered primarily in the context of egg-laying behavior and larval death due to the
larvae's inability to travel far from where they hatch in their early stages (Oberhauser &
Solensky, 2004). Yet, our observations indicate that larvae might also be misled by the chemical
signals of non-host plants, even when their host plants are available. Research by Greenstein et
al. (2022) showed that up to 62% of monarch larvae might attempt to consume a low-
performance host (plants that considerably decrease survival), leading to a mere 3.5% survival
119
rate. These findings, from no-choice experiments, did not clarify whether nutritional deficiencies
or toxins were responsible for the high mortality rates. Despite having the option between
suitable and non-suitable plants in our study, larvae still opted for the latter, hinting that chemical
signals could be a significant factor in their erroneous feeding behavior. Both species of
Asclepias and Vincetoxicum, along with other plants like foxgloves (Digitalis), produce cardiac
glycosides, a compound monarch larvae accumulate for their defense (Groen & Whiteman, 2021;
Sliumpaite et al., 2013). The specific makeup of cardenolides can differ based on the plant
source, potentially leading to variations in toxicity upon ingestion (Greenstein et al., 2022).
Investigating how cardiac glycosides influence the feeding behavior of monarch larvae on
swallow-wort could be a valuable direction for future research.
This study contributes to our knowledge on the oviposition behaviors of monarch butterflies,
confirming a pronounced preference for common milkweed over pale swallow-wort, albeit with
a notable oviposition rate on swallow-wort in natural settings. Despite the absence of egg-laying
in our laboratory experiments that could have further illuminated the mechanisms behind these
behaviors, our field observations offer a glimpse into the complex interplay of environmental,
spatial, and chemical factors influencing monarch choices. Particularly, the documentation of a
significant mortality rate among larvae due to swallow-wort consumption underlines the
potential deceptive role of chemical cues emanating from non-host plants. These insights
underscore the importance of considering both habitat characteristics and plant chemical
properties in understanding and supporting monarch conservation efforts, highlighting the need
for expanded research to capture the full spectrum of factors impacting monarch reproductive
success in diverse environments.
120
LITERATURE CITED
Alred, B., Haan, N., Landis, D. A., & Szucs, M. 2022. Does the Presence of the Biological
Control Agent, Hypena opulenta (Lepidoptera: Erebidae) on Swallow-Worts Deter
Monarch Oviposition? Environmental Entomology, 51(1), 77–82.
https://doi.org/10.1093/ee/nvab121
Bates, D., Maechler, M., Bolker, B, Walker, S. 2015. Fitting Linear Mixed-Effects Models Using
lme4. Journal of Statistical Software, 67(1), 1-48. doi:10.18637/jss.v067.i01.
Berenbaum, M.R., & Zangerl, A.R. 1996. Phytochemical diversity: adaptation or random
variation? Evolutionary Applications, 9(1), 113-129.
Casagrande, R. A., & Dacey, J. E. 2007. Monarch butterfly oviposition on swallow-worts
(Vincetoxicum spp.). Environmental Entomology, 36(3), 631–636.
https://doi.org/10.1603/0046-225X(2007)36[631:MBOOSV]2.0.CO;2
DiTommaso, A., & Losey, J.E. 2003. Oviposition preference and larval performance of the
monarch butterfly (Danaus plexippus) under laboratory conditions. Journal of Insect
Conservation, 7(4), 209-220. https://doi.org/10.1023/B:JICO.0000004893.02008.8e
Dukas, R. 2008. Evolutionary biology of insect learning. Annual Review of Entomology, 53, 145-
160. https://doi.org/10.1146/annurev.ento.52.110405.091200
Endress, M. E., & Bruyns, P. v. 2000. A Revised Classification of the Apocynaceae s.l. The
Botanical Review, 66(1), 1–56. http://www.inform.umd.edu/PBIO
Ferrer-Paris, J.R., Sánchez-Mercado, A., Viloria, Á.L., Donaldson, J. 2013. Congruence and
diversity of butterfly-host plant associations at higher taxonomic levels. Plos One, 8(5),
e63570. https://doi.org/10.1371/journal.pone.0063570
Flockhart, D. T. T., Pichancourt, J. B., Norris, D. R., & Martin, T. G. 2015. Unravelling the
annual cycle in a migratory animal: Breeding-season habitat loss drives population
declines of monarch butterflies. Journal of Animal Ecology, 84(1), 155–165.
https://doi.org/10.1111/1365-2656.12253
Greenstein, L., Steele, C., & Taylor, C. M. 2022. Host plant specificity of the monarch butterfly
Danaus plexippus: A systematic review and meta-analysis. Plos One, 17(6).
https://doi.org/10.1371/journal.pone.0269701
Groen, S. C., & Whiteman, N. K. 2021. Convergent evolution of cardiac-glycoside resistance in
predators and parasites of milkweed herbivores. Current Biology, 31(22), R1465–R1466.
https://doi.org/10.1016/j.cub.2021.10.025
121
Hartig F (2022). _DHARMa: Residual Diagnostics for Hierarchical (Multi-Level /Mixed)
Regression Models_. R package version 0.4.6,b.
Jones, P. L., & Agrawal, A. A. 2019. Beyond preference and performance: host plant selection by
monarch butterflies, Danaus plexippus. Oikos, 128, 1092-1102
https://doi.org/10.1111/oik.06001
Kessler, A., & Baldwin, I. T. 2001. Defensive function of herbivore-induced plant volatile
emissions in Nature. Science, 291, 2141–2144.
Labandeira, C.C. 1998. Early history of arthropod and vascular plant associations. Annual
Review of Earth and Planetary Sciences, 26, 329-377.
https://doi.org/10.1146/annurev.earth.26.1.329
Ladner, D. T., & Altizer, S. 2005. Oviposition preference and larval performance of North
American monarch butterflies on four Asclepias species. Entomologia Experimentalis et
Applicata, 116(1), 9–20. https://doi.org/10.1111/j.1570-7458.2005.00308.x
Lenth, R., Singmann, H., Love, J., Buerkner, P., & Herve, M. 2021. Emmeans: Estimated
marginal means, aka least-squares means. R package version 1 (2018).
Mattila, H. R., & Otis, G. W. 2003. A comparison of the host preference of monarch butterflies
(Danaus plexippus) for milkweed (Asclepias syriaca) over dog-strangler vine
(Vincetoxicum rossicum). Entomologia Experimentalis et Applicata, 107(3), 193–199.
https://doi.org/10.1046/j.1570-7458.2003.00049.x
Oberhauser, K., Wiederholt, R., Diffendorfer, J. E., Semmens, D., Ries, L., Thogmartin, W. E.,
Lopez-Hoffman, L., & Semmens, B. 2017. A trans-national monarch butterfly population
model and implications for regional conservation priorities. Ecological Entomology,
42(1), 51–60. https://doi.org/10.1111/een.12351
Pleasants, J. M., & Oberhauser, K. S. 2013. Milkweed loss in agricultural fields because of
herbicide use: Effect on the monarch butterfly population. Insect Conservation and
Diversity, 6(2), 135–144. https://doi.org/10.1111/j.1752-4598.2012.00196.x
Oberhauser, K.S. and Solensky, M.J. 2004. The monarch butterfly. Biology and conservation.
Cornell University Press, Ithaca.
Pellmyr O. 1986. Three pollination morphs in Cimicifuga simplex; incipient speciation due to
inferiority in competition. Oecologia, 68(3), 304-307.
https://doi.org/10.1007/BF01036741
Pigluicci, M., Murren, C.J., & Schlichting, C.D. (2006). Phenotypic plasticity and evolution by
genetic assimilation. Journal of Experimental Biology, 209(12), 2362-2367.
https://doi.org/10.1242/jeb.02070
122
R Core Team (2023). _R: A Language and Environment for Statistical Computing_. R
Foundation for Statistical Computing, Vienna, Austria. .
Šliumpaite, I., Murkovic, M., Zeb, A., & Venskutonis, P. R. 2013. Antioxidant properties and
phenolic composition of swallow-wort (Vincetoxicum lutea L.) leaves. Industrial Crops
and Products, 45, 74–82. https://doi.org/10.1016/j.indcrop.2012.11.036
Snell-Rood, E.C., & Papaj, D.R. 2009. Patterns of phenotypic plasticity in common and rare
environments: A study of host use and color learning in the cabbage white butterfly,
Pieris rapae. American Naturalist, 174(6), 729-743. https://doi.org/10.1086/606024
Thompson JN. 1988. Evolutionary ecology of the relationship between oviposition preference
and performance of offspring in phytophagous insects. Entomologia Experimentalis et
Applicata. 47: 3–14. https://doi.org/10.1111/j.1570-7458.1988.tb02275.x
van der Linden, C. F. H., WallisDeVries, M. F., & Simon, S. 2021. Great chemistry between us:
The link between plant chemical defenses and butterfly evolution. Ecology and
Evolution, 11(13), 8595–8613. https://doi.org/10.1002/ece3.7673
Whitfield, T. J. S., Novotny, V., Miller, S. E., Hrcek, J., Klimes, P., & Weiblen, G. D. 2012.
Predicting tropical insect herbivore abundance from host plant traits and phylogeny.
Ecology, 93, S211–S222. https://doi.org/10.1890/11-0503.1
Zalucki, M. P., & Lammers, J. H. 2010. Dispersal and egg shortfall in Monarch butterflies: What
happens when the matrix is cleaned up? Ecological Entomology, 35(1), 84–91.
https://doi.org/10.1111/j.1365-2311.2009.01160.x
Zaya, D. N., Pearse, I. S., & Spyreas, G. 2017. Long-term trends in midwestern milkweed
abundances and their relevance to monarch butterfly declines. BioScience, 67(4), 343-
356. Oxford University Press. https://doi.org/10.1093/biosci/biw186
Zeileis A, Hothorn T. 2002. Diagnostic checking in regression relationships. R News, 2(3):7–10.
URL https://CRAN.R-project.org/doc/Rnews/.
123