The Pennsylvania State University
The Graduate School
Intercollege Graduate Degree Program in Ecology
COMPETITIVE INTERACTIONS AND ASSOCIATIONS OF THE INVASIVE THISTLES CARDUUS NUTANS AND C. ACANTHOIDES
A Thesis in
Ecology
by
Emily Sofia Jálics Rauschert
© 2006 Emily Sofia Jálics Rauschert
Submitted in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy
August 2006
The thesis of Emily Sofia Jálics Rauschert was reviewed and approved* by the following:
Katriona Shea Associate Professor of Biology Thesis Advisor Chair of Committee
Ottar N. Bjørnstad Associate Professor of Entomology and Biology
Eric S. Post Associate Professor of Biology
David Mortensen Professor of Crop and Soil Sciences Chair of the Intercollege Graduate Degree Program in Ecology
*Signatures are on file in the Graduate School
iii ABSTRACT
As the number of biological invasions continues to increase, the degree to which communities can resist invasions and the interactions between multiple invasive species will become increasingly important. This will have important consequences for existing communities, as the combined impact of multiple invaders may not simply be additive.
Using experimental, observational and modeling approaches, we investigated the direct competitive interactions between Carduus nutans and C. acanthoides, two congeneric invasive thistle species. It was hypothesized that their observed spatial segregation was due to interspecific competition. Spatially-explicit simulation models of competitive interactions between these species were developed for both the landscape and the field levels, to explore the range of behaviors predicted. We examined direct interspecific competition using three response surface field experiments. Additionally, the thistle distribution patterns were quantified at two resolutions: the regional level in the area of overlap, and the field level in four fields of natural co-occurrence. Their interactions with the existing vegetation were examined in two ways: by quantifying their vegetative associations in four fields of co-occurrence, and by experimentally examining their germination and establishment response to microsite characteristics. The competition experiments did not demonstrate strong competition or density dependence. The simulation models do generate spatial segregation, but not for parameters corresponding to the results of the competition experiments; a gradual decay of the segregation pattern with eventual coexistence is more likely. The landscape-scale observational studies indicate negative associations between these species, but at the field scale, which is the
iv scale at which competition would occur, C. nutans and C. acanthoides are positively associated, implying that competition between them does not lead to exclusion. While there were significant differences in the plant communities in areas with and without thistles, there were no significant differences in communities associated with C. nutans versus C. acanthoides. Both species were found to be sensitive to microsite characteristics, with generally better establishment in larger gaps, and better survival in gaps with less disturbance. Our results reject the hypothesis that resource competition between these two species strongly influences their distributions; instead, our results suggest that interactions with other community members, combined with the spread history of these species, likely most strongly influence the current distribution. Although strong competitive interactions were not observed between these two congeneric invaders, interactions between invasive species will only continue to increase, and the combined empirical and theoretical approach presented here will be critical to assessing how such interactions influence the success and impact of invasive species.
v TABLE OF CONTENTS
LIST OF FIGURES ...... vii
LIST OF TABLES...... x
ACKNOWLEDGEMENTS...... xi
Chapter 1 Introduction ...... 1
References ...... 6
Chapter 2 Carduus nutans and Carduus acanthoides...... 10
Introduction ...... 10 Life cycle...... 11 Dispersal...... 17 Seedbank...... 19 Germination...... 21 Interactions between plant species ...... 24 Distribution...... 29 Economic Impact...... 31 Control and Management ...... 32 Conclusions ...... 39 Notes...... 41 References ...... 41
Chapter 3 Models of spatial segregation in Carduus nutans and C. acanthoides ...... 56
Abstract...... 56 Introduction ...... 56 Methods ...... 60 Results ...... 72 Discussion...... 77 Acknowledgements ...... 83 References ...... 83
Chapter 4 Competitive interactions between two invasive thistle species, Carduus nutans and Carduus acanthoides...... 104
Abstract...... 104 Introduction ...... 104 Methods ...... 108 Results ...... 116
vi Discussion...... 120 Acknowledgements ...... 126 References ...... 127
Chapter 5 Spatial coexistence patterns of two invasive thistle species, Carduus nutans and C. acanthoides...... 150
Abstract...... 150 Introduction ...... 151 Methods ...... 154 Results ...... 162 Discussion...... 164 Acknowledgements ...... 169 References ...... 170
Chapter 6 Plant community associations of the invasive thistles Carduus nutans and C. acanthoides...... 189
Abstract...... 189 Introduction ...... 189 Methods ...... 192 Results ...... 196 Discussion...... 198 Acknowledgements ...... 202 References ...... 202
Chapter 7 Effect of microsite characteristics on Carduus nutans and C. acanthoides germination and survival...... 218
Abstract...... 218 Introduction ...... 218 Methods ...... 222 Results ...... 226 Discussion...... 228 Acknowledgements ...... 234 References ...... 234
Chapter 8 Conclusions ...... 246
References ...... 252
vii LIST OF FIGURES
Figure 3-1: Four scenarios of the Lotka-Volterra model...... 93
Figure 3-2: Competition coefficient ranges for the landscape mode...... 94
Figure 3-3: Landscape model: overlap and distances traveled ...... 95
Figure 3-4: Landscape model: symmetric cases (i.e. interspecific equivalence)...... 96
Figure 3-5: Landscape model: outcomes of 3 different Lotka-Volterra scenarios ...... 97
Figure 3-6: Landscape model results with respect to Lotka-Volterra states...... 98
Figure 3-7: Landscape model: total thistle density relationship to the overlap width after 100 years ...... 99
Figure 3-8: Field model: outcomes under different interspecific competition coefficients...... 100
Figure 3-9: Field model: symmetric cases (i.e. interspecific equivalence)...... 101
Figure 3-10: Field model: Outcomes of 3 different Lotka-Volterra scenarios ...... 102
Figure 3-11: Field model results with respect to Lotka-Volterra states...... 103
Figure 4-1: Experimental design...... 140
Figure 4-2: Densities planted in the seedling placement experiment ...... 141
Figure 4-3: Seed placement experiment: germination response to density and the proportion of the other species present...... 142
Figure 4-4: Seed placement experiment: average days to germination response to density and the proportion of the other species planted ...... 143
Figure 4-5: Seedling placement experiment: flowering response to planting density and the proportion of the other species present...... 144
Figure 4-6: Seedling placement experiment: response in the number of flowers produced to total planting density and the proportion of other species present ...145
Figure 4-7: Germination trial: Germination response to the density and the proportion of the other species experienced by the mother plant...... 146
viii Figure 4-8: Seed scattering experiment: the response of germination and survival to July to density and the proportion of the other species planted...... 147
Figure 4-9: Seed scattering experiment: plot average longest leaf length relationship to density and the proportion of the other species planted ...... 148
Figure 4-10: Seed scattering experiment: average longest leaf length response to actual densities and actual proportions of the other species present ...... 149
Figure 5-1: 2002 distributions of C. nutans and C. acanthoides from Allen and Shea (2006)...... 178
Figure 5-2: Scales of study ...... 179
Figure 5-3: Regional survey maps from 2004 and 2005...... 180
Figure 5-4: Presence–absence correlograms from the regional survey ...... 181
Figure 5-5: Density index correlograms from the regional survey...... 182
Figure 5-6: P1 correlogram results ...... 183
Figure 5-7: P2 main patch correlogram results...... 184
Figure 5-8: P2 middle patch correlogram results...... 185
Figure 5-9: Site I correlogram results ...... 186
Figure 5-10: Site R correlogram results (shorter distances) ...... 187
Figure 5-11: Site R correlogram results (longer distances) ...... 188
Figure 6-1: Mantel correlograms using presence-absence data...... 211
Figure 6-2: Nonmetric multi-dimensional scaling of all plots: presence-absence data...... 212
Figure 6-3: NMDS of all plots: frequency data ...... 213
Figure 6-4: NMDS of all plots: percent cover data (2005 only)...... 214
Figure 6-5: NMDS of Carduus plots only: presence-absence data ...... 215
Figure 6-6: NMDS of Carduus plots only: frequency data ...... 216
Figure 6-7: NMDS of Carduus plots only: percent cover data (2005 only)...... 217
ix Figure 7-1: Overall germination ...... 242
Figure 7-2: Effect of microsite area...... 243
Figure 7-3: Effect of the watering treatment...... 244
Figure 7-4: Effect of disturbance type ...... 245
x LIST OF TABLES
Table 3-1: Landscape model parameters ...... 91
Table 3-2: Field model parameters ...... 92
Table 4-1: Summary of total response...... 134
Table 4-2: Summary of results...... 135
Table 4-3: Results from the seed placement experiment...... 136
Table 4-4: Results from the seedling placement experiment...... 137
Table 4-5: Results from the germination trial...... 138
Table 4-6: Results from the seed scattering experiment...... 139
Table 5-1: Regional survey summary...... 176
Table 5-2: Summary of the numbers of thistles in four fields of co-occurrence ...... 177
Table 6-1: Percentages of plots with thistles in the four fields of co-occurrence...... 207
Table 6-2: Plant species list ...... 208
Table 6-3: Differences in community in thistle versus non-thistle plots: Results of partial Mantel tests...... 209
Table 6-4: Differences in community in C. nutans vs. C. acanthoides plots: Results of partial Mantel tests...... 210
Table 7-1: Summary of germination and survival in fall 2004 and 2005...... 239
Table 7-2: Results of logistic regressions of germination and survival...... 240
Table 7-3: Analyses of average days to germination, fall size and herbivory...... 241
xi ACKNOWLEDGEMENTS
This thesis is the result of a five-year collaboration with my advisor, Katriona
Shea, who provided constant support and interest. My committee members have also been very encouraging. Eric Post was partially responsible for convincing me to come to
Penn State, and has always been very helpful and supportive. Ottar Bjørnstad provided considerable assistance, particularly with statistical issues, not just related to analyses used for this thesis but also how to approach statistical issues in general. Dave Mortensen has provided encouragement and an extremely useful biological perspective due to his extensive understanding of ecology and of agricultural systems. All three committee members made very useful suggestions for this thesis, which considerably improved the quality of the research.
This project would not have been possible without the help of many
undergraduate field helpers. Special thanks to Elizabeth Dlugosz, John Mellon and
Gabby Hrsychyn for particular biological interest combined with valuable logistical
assistance. Stephen Selgo, Adam Reese and Jeff Butterbaugh also worked very hard on
this project, as well as an army of other Shea lab undergraduates. My fellow graduate
students Emily Phillips, Carrie Schwarz, Dietmar Schwarz, Jessica Peterson, Tiffany
Bogich and Danielle Garneau provided helpful support. During this project, I learned
how useful post-docs can be: Olav Skarpaas has provided valuable assistance throughout
this project, as has Eelke Jongejans.
The Intercollege Graduate Degree Program in Ecology has been a lively,
interactive setting for me as well as for many students in the program. Dave Mortensen
xii has done an excellent job as Chair in fostering the student-driven nature of the program.
Thanks to Chuck Fisher from the Biology Department for allowing IGDP students to really get the best of both worlds. Kathryn McClintock and Paula Farwell from the
Biology Department have provided valuable logistical assistance throughout this project.
Most importantly, I am extremely grateful to all of the emotional support I have received during these five years. My parents Paul and Kristi Jálics have provided constant support, even if my dad has always made me jealous with his Ph.D. story! I am also grateful to my aunt and godmother Isabella H. Jálics for her lifelong academic encouragement; she is the first person who told me to get a Ph.D., when I was around eight years old. My sister Alice Claus and my brother Andy Jálics have listened to me
(over and over again) as has Diana Sitt, who never seemed to get bored trying to make sense of what was going on in my life. Two people deserve particular mention, as they have had to put up with the most. My academic twin, Zeynep Sezen, provided support and by-the-minute advice for five years; through her, I have learned a tremendous amount about science. Finally, I wish to thank my husband and best friend Ingmar for constant and untiring Unterstutzung.
Chapter 1
Introduction
Classical Lotka-Volterra theory predicts that similar species competing for the same limited resources cannot coexist unless intraspecific interactions are stronger than interspecific interactions. The competitive exclusion principle, sometimes called Gause’s principle, was formalized by Hardin (1960), who stated that “complete competitors cannot coexist.” Hardin (1960) deliberately chose four ambiguous words when stating this principle to emphasize the fact that the limits of this principle are not clear. The fact that many species that require the same resources are in fact able to coexist implies the existence of coexistence-promoting mechanisms.
Competition has been defined in many different ways; several researchers have tried to provide a definition that is comprehensive but useful. Grime (1973), in an attempt to emphasize that competition is only one facet of the struggle for existence, defined competition as “the tendency of neighboring plants to utilize the same quanta of light, ions of mineral nutrients, molecules of water or volume of space.” In a review on competition, interactions were defined more broadly by Gibson (1999) as “any effect one species has on another.” Keddy (2001) uses a more mechanistic approach by defining competition as the “negative effects that one organism has upon another by consuming, or controlling access to, a resource that is limited in availability.” Harper (1977) emphasizes density dependence when discussing competition, as do classic competition models. Some definitions emphasize population dynamics by requiring competition to
2 have an effect on the net reproductive rates of species (Silvertown and Lovett-Doust
1993). It is important to recognize that a resource must be limiting in order for it to strongly influence population dynamics.
Many researchers assume that the experimental study of competition began with
Gause’s (1934) experiments; in fact, plant competition has a much longer history
(Jackson 1981). For example, Tansley (1917) found that the outcome of competition between two Gallium spp. depended on the type of soil present. Gause (1934) himself refers to the previous work of many foresters on competition.
The idea that competition is a major factor influencing species distributions has received criticism in the past. For example, Connell (1980) strongly criticized the idea of
using segregated patterns and /or niche differences as evidence that competition had led
to such differences (also referred to as the ghost of competition past), as such patterns can
also arise from species evolving independently in the absence of any interactions.
Nonetheless, there is compelling evidence that competitive exclusion occurs, much of
which involves invasive species. In a certain sense, invasive species are a type of “natural
experiment” where it is possible to observe the effect of the presence of a novel species
on resident community members. For example, the grey squirrel Sciurus carolinensis has largely replaced the native red squirrel Sciurus vulgaris in Britain since the grey squirrel’s introduction in 1876 (MacKinnon 1978). Many studies have focused on the negative impacts of a particular exotic species on a native species; for example, Byers
(2000) found that the exotic snail Batillaria attramentaria competitively excluded the native snail Cerithidea california because it was more efficient at resource conversion.
3 Much of this body of literature is focused on interactions between native and invasive species; interactions between invaders have not been as well studied. If the current trends in biological invasions continue, interactions between invasive species will become increasingly common. When such interactions occur they can be very dramatic.
For example, the parasitoid wasp Aphytis lingnanensis, introduced in California to control the citrus red scale Aonidiellia aurantii, was rapidly replaced by Aphytis melius on its introduction (Murdoch et al. 1996). There are also examples of invasive species, which initially displaced native species, themselves being replaced by subsequent invaders. In Bermuda, Iridomyrmex humilis, the Argentine ant, replaced populations of
Pheidole megacephala, another non-native ant which had previously displaced native ant populations (Crowell 1968). In the St. Lawrence River, non-native quagga mussels,
Dreissena burgensis, appear to be replacing non-native zebra mussels, Dreissena polymorpha, which had previously replaced the native mussels (Ricciardi and Whoriskey
2004).
Two important related themes in ecology are the development and maintenance of species patterns in space and time and the consequences of spatiotemporal patterns for population and ecosystem dynamics (Levin 1992). Generally, abiotic factors are considered to determine larger scale patterns(Levin 1989, Wiens 1989), although it is known that biotic factors can also be important, as recognized by Hutchinson (1957), who developed the terms fundamental and realized niche in the context of competition. In this thesis I examine some of the competitive interactions that can arise between invaders, and the consequences of such interactions at different spatial scales.
4 Carduus nutans L. and C. acanthoides L., which are native to Europe and Asia, have become pest species in North and South America, Australia and New Zealand
(Holm et al. 1979). C. nutans and C. acanthoides have both spread throughout the US
(Dunn 1976), and Skinner et al. (2000) list C. nutans and C. acanthoides as the second
and fifteenth most commonly listed noxious weeds in the US. Both occur in
Pennsylvania; however, Allen and Shea (2006) documented a strikingly segregated
distribution in central Pennsylvania, which cannot be readily explained by abiotic factors.
Possible explanations for such a pattern include spread history, competition between
these species, indirect interactions between these species, abiotic differences in the two
areas and different management practices in different areas. Although both of these
thistles are often subject to intense management in pastures by farmers and along roads
by road maintenance authorities, management practice appear to be uniform across this
area and are unlikely to favor one species over the other. This dissertation examines
competitive interactions between C. nutans and C. acanthoides and the communities in
which they invade. This work is used to address the hypothesis that competition between
C. nutans and C. acanthoides influences their distributional patterns, leading to spatial
segregation. It is very difficult to rule out other possible explanations for species
distributional patterns, such as spread history; however, it is possible to examine whether
or not competition between these two species is strong enough to cause the observed
distributional patterns.
Because both species are considered to be economically important pests, their
biology and control has been the subject of considerable research since the 1950s.
Chapter 2 presents a literature review of the biology of these two species, with a
5 particular focus on their life cycles and what is known of their interactions with other
species.
Modeling is an important heuristic technique to explore the potential outcomes of
various hypotheses. The coexistence patterns created by various levels of competition are
explored through the use of spatially-explicit simulation models in Chapter 3. A
landscape level model explores competition within a series of patches in a landscape that
are linked by dispersal. A field level model examines the effect of spatially explicit
competition and dispersal. A range of potential competition coefficients are explored in
both models, and the results of these models are compared to the classic Lotka-Volterra
models, which predict coexistence only when intraspecific interactions are stronger than
interspecific interactions (Chesson 2000).
It is not possible to merely assume that a certain process took place based on
current distributions (Connell 1980). In order to determine what the effects of direct
competition between these two species actually are, I carried out a series of competition
experiments, described in Chapter 4. All three field experiments manipulate density and
relative composition of C. nutans and C. acanthoides. Two of the experiments track the
fate of individuals starting from seeds or from seedlings through to flowering. The third
experiment examines the outcome of competition at higher densities.
The aforementioned Allen and Shea (2006) survey, which documented the large
scale distribution of these two species, implied potential negative interactions, such as
competition, between these two species, leading to their spatial segregation. However,
different patterns may be observed at different scales (Wiens 1989). In order to examine
whether or not this same pattern of negative associations is also present at finer scales, I
6 investigated the spatial distributions of individuals in four locations of natural co- occurrence (Chapter 5). In this chapter, I also present the results of further large-scale monitoring of the area of overlap, and compare the associations between these two species at different spatial scales.
The models and experiments assume that space is nearly entirely occupied by thistles. Focusing only on pair-wise interactions between species can lead to incorrect conclusions about the results of competition due to the presence of indirect interactions
(Tilman 1987) or through non-additive effects (Dormann and Roxburgh 2005). The
interactions of C. nutans and C. acanthoides with other plant species are investigated in
Chapters 6 and 7. In Chapter 6, I examine whether vegetation associated with Carduus
thistles differs from uninfested vegetation in four locations of natural co-occurrence.
Additionally, I explore whether C. nutans is associated with different vegetation
communities than is C. acanthoides. In Chapter 7, I examine differences in the
germination and establishment response of these species to microsite characteristics,
focusing on water, microsite size and the relative importance of below and aboveground
disturbance.
In Chapter 8, the conclusions from the various studies are summarized, and
potential future directions for research on these invaders are discussed.
References
Allen, M., and K. Shea. 2006. Spatial segregation of congeneric invaders in central
Pennsylvania, USA. Biological Invasions 8:509-521.
7 Byers, J. E. 2000. Competition between two estuarine snails: Implications for invasions
of exotic species. Ecology 81:1225-1239.
Chesson, P. 2000. Mechanisms of maintenance of species diversity. Annual Review of
Ecology and Systematics 31:343-+.
Connell, J. H. 1980. Diversity and the Coevolution of Competitors, or the Ghost of
Competition Past. Oikos 35:131-138.
Crowell, K. L. 1968. Rates of Competitive Exclusion by Argentine Ant in Bermuda.
Ecology 49:551-555.
Dormann, C. F., and S. H. Roxburgh. 2005. Experimental evidence rejects pairwise
modelling approach to coexistence in plant communities. Proceedings of the
Royal Society B-Biological Sciences 272:1279-1285.
Dunn, P. H. 1976. Distribution of Carduus nutans, C. acanthoides, C. pycnocephalus and
C. crispus, in the United States. Weed Science 24:518-524.
Gause, G. F. 1934. The Struggle for Existence. Hafter Publishing Company, New York.
Gibson, D. J., J. Connolly, D. C. Hartnett, and J. D. Weidenhamer. 1999. Designs for
greenhouse studies of interactions between plants. Journal of Ecology 87:1-16.
Grime, J. P. 1973. Competition and Diversity in Herbaceous Vegetation - Reply. Nature
244:311-311.
Hardin, G. 1960. Competitive Exclusion Principle. Science 131:1292-1297.
Harper, J. L. 1977. Population Biology of Plants. Academic Press, New York.
Holm, L., J. V. Pancho, J. P. Herberger, and D. L. Plucknett. 1979. A Geographical Atlas
of World Weeds. John Wiley and Sons, New York.
8 Hutchinson, G. E. 1957. Concluding remarks. Cold Spring Harbor Symposium on
Quantitative Biology 22:415-427.
Jackson, J. B. C. 1981. Interspecific Competition and Species Distributions - the Ghosts
of Theories and Data Past. American Zoologist 21:889-901.
Keddy, P. 2001. Competition, Second edition. Kluver, Boston.
Levin, S. A. 1989. Challenges in the development of a theory of ecosystem structure and
function. in J. Roughgarden, R. M. May, and S. A. Levin, editors. Perspectives in
Ecological Theory. Princeton Univeristy Press, Princeton, N.J.
Levin, S. A. 1992. The Problem of Pattern and Scale in Ecology. Ecology 73:1943-1967.
MacKinnon, K. 1978. Competition between Red and Grey Squirrels. Mammal Review
8:185-190.
Murdoch, W. W., C. J. Briggs, and R. M. Nisbet. 1996. Competitive displacement and
biological control in parasitoids: A model. American Naturalist 148:807-826.
Ricciardi, A., and F. G. Whoriskey. 2004. Exotic species replacement: shifting
dominance of dreissenid mussels in the Soulanges Canal, upper St. Lawrence
River, Canada. Journal of the North American Benthological Society 23:507-514.
Silvertown, J. W., and J. Lovett-Doust. 1993. Introduction to plant population biology.
Blackwell Scientific Publications.
Skinner, K., L. Smith, and P. Rice. 2000. Using noxious weed lists to prioritize targets for
developing weed management strategies. Weed Science 48:640-644.
Tansley, A. G. 1917. On competition between Galium saxatil L. (G-Hercynicum Weig.)
and Galium sylvestre Poll. (G-asperum Schreb.) On different types of soil. Journal
of Ecology 5:173-179.
9 Tilman, D. 1987. The Importance of the Mechanisms of Interspecific Competition.
American Naturalist 129:769-774.
Wiens, J. A. 1989. Spatial Scaling in Ecology. Functional Ecology 3:385-397.
Chapter 2
Carduus nutans and Carduus acanthoides
Introduction
Carduus nutans L. and Carduus acanthoides L. (Asteraceae) are congeneric
thistle species that have spread from their native ranges in Europe, Asia and North Africa
to become invasive in many parts of the world. Both species are commonly found along
roadsides and in pastures (Batra 1978), and can cause significant economic damage in
pastures (Thompson et al. 1987). For this reason, both have been the focus of
considerable research.
In the US and Europe, Carduus nutans is referred to as musk thistle whereas in
New Zealand it is more commonly known as nodding thistle. C. acanthoides is referred to as plumeless thistle, although actually the entire Carduus genus is plumeless (McCarty
1985). In Flora Europaea, the nutans group is referred to as “a difficult group in need of
further study” (Tutin et al. 1964). This has been best studied by Desrochers et al. (1987)
in Canada, using morphology and flavenoid profiles. At various times, Carduus nutans
has been divided into different subspecies: Carduus nutans spp. nutans, spp. leiophyllus, spp. macrolepis, and spp. macrocephalus. Carduus nutans spp. leiophyllus has also been
referred to as Carduus theormeri. Desrochers et al. (1987) conclude that the entire C. nutans group in Canada is best considered as one species, which most usefully divided into two subspecies, spp. nutans and spp. leiophyllus.
11 This review first introduces the thistle life cycle, and then focuses on their interactions with other plants and with herbivores. As both of these can be strongly influenced by management practices, we also discuss human attempts to control them, as well as the economic impacts of these thistles. We summarize what has been studied so far, and attempt to point out some areas in need of further research.
Life cycle
Both Carduus nutans and C. acanthoides are monocarpic perennials: they live for
an indeterminate number of years as rosettes, bolt, and then die after flowering. While the
C. nutans life cycle has been well studied in its invaded ranges in New Zealand, Australia
and the US, the C. acanthoides life cycle has received less attention.
Different C. nutans sites have been found to have different proportions of the
population experiencing annual, biennial or triennial life cycles (Kelly and McCallum
1990). In New Zealand, most C. nutans behaved as winter annuals (Popay and Thompson
1979). Occasionally C. nutans can be a summer annual, presumably when temperatures
are low enough initially for vernalization. However, biennials produce more seed than
winter annuals, which produce more seed than summer annuals (Popay and Medd 1990).
The variation in demography of C. nutans may be attributable to phenotypic plasticity
(Lee and Hamrick 1983).
Most C. nutans seedlings emerge in the fall in New Zealand (Popay and Kelly
1986, Kelly and McCallum 1990), which is when maximum population sizes are seen
(Popay and Thompson 1979). Timing of germination affects subsequent plant
12 characteristics for C. nutans. For example, in Australia, fall germinating C. nutans do not
generally flower the following year (Doing et al. 1969). Individuals that germinate later
in the fall have higher mortality as well as delayed and reduced seed production (Lee and
Hamrick 1983). The period of greatest seedling mortality for C. nutans is in the summer
and late spring (Lee and Hamrick 1983). C. nutans survivorship, flowering probability
and seed production are positively related to rosette size (Lee and Hamrick 1983).
Differences in the biotic environment of C. nutans can lead to different mortality patterns (Lee and Hamrick 1983). C. nutans is sensitive to temperature and moisture variation during spring and early summer (Doing et al. 1969). Summer drought followed by overgrazing can lead to pastures with many bare patches, which increases thistle germination in the fall (Popay and Thompson 1979). Alternation of dry and wet summers is favorable for C. nutans, as drought increases gaps necessary for germination, but wet periods are necessary for life cycle completion (Doing et al. 1969). Differences in density and soil fertility do not necessarily lead to differences in C. nutans development (Medd
and Lovett 1978 II). In Ontario, C. acanthoides tends to be found on drier gravel soils
than C. nutans (Moore and Mulligan 1956).
C. nutans requires vernalization for flowering, although experiencing short days
before vernalization reduces the need for vernalization and the need for increased day
period after vernalization (Medd and Lovett 1978). In Pennsylvania, C. nutans flowers
from May to August, and C. acanthoides flowers from June to October (Rhodes and
Block 2000). C. nutans which flower outside the main flowering time in New Zealand
(December through January) produce considerably fewer seeds (Jessep 1990). In New
Zealand, Kelly and McCallum (1990) estimated mean C. nutans seed production per
13 flowering plant to be between 1230 and 8960. Jessep (1990) report a maximum of approximately 15,000 seeds, with a mean of only approximately 6000 seeds, for C. nutans in New Zealand. In the US, McCarty (1982) estimate mean seed production to be between 3535 and 7098 seeds per individual. For C. acanthoides, average flower production is between approximately 1000 and 11,000 and in Argentina (Feldman and
Lewis 1990).
C. nutans has been reported to exhibit some degree of density dependence. C. nutans survival to flowering is density-dependent and is typically low (less than 6%)
(Kelly and McCallum 1990). Rosettes were generally smaller and seed production was lower per individual in more dense populations (Lee and Hamrick 1983). The variation in
C. nutans rosette size also appeared to be smaller in more dense populations (Lee and
Hamrick 1983). The seed to seedling transition may be very important in the C. nutans life cycle in New Zealand (Kelly and McCallum 1990, Shea and Kelly 1998).
Reproductive biology
Both species are pollinated primarily by Bombus spp., Apis spp. and a variety of
Lepidoptera (Smyth and Hamrick 1987, Giurfa and Nunez 1993). The movement of
pollen has been studied only for C. nutans. Pollinator behavior affects the pattern of
pollen movement in C. nutans (Smyth and Hamrick 1987). The mean pollinator flight distance observed in populations of C. nutans in the United States was 1.2 m (Holscher and Hamrick, unpubl., cited in Smyth and Hamrick 1987); however, pollinators typically do not deposit the entire pollen load on the next flower visited. The mean pollen movement distance is 5 m, although the tail of the pollen movement distribution was
14 longer than the maximum distances (18 m) they measured (Smyth and Hamrick 1987).
Most C. nutans heads are pollinated by a few pollinators (Smyth and Hamrick 1984).
Gene flow in C. nutans populations may be greater at higher population densities (Smyth and Hamrick 1987).
The foraging behavior of pollinators in response to potential rewards has only been studied for C. acanthoides. C. acanthoides flower heads have been shown to be visited by Apis mellifera liguistica in proportion to the sugar reward of each flower; second day capitula have the largest reward (Giurfa and Nunez 1992). A. mellifera presence in patches of C. acanthoides was positively correlated with the total available sugar in the patch (Giurfa and Nunez 1992). In a related study of C. acanthoides, although no evidence of systematic inspection of florets by A. mellifera was found, the number of floret visits observed correlated well with the number of depleted florets, indicating high foraging efficiency perhaps due to the marking of depleted flowers with scent (Giurfa and Nunez 1993). The lifespan of C. acanthoides inflorescences is three days (Giurfa and Nunez 1992). Scarcity of pollinators, pollen-feeding Coleoptera and water stress led to an approximately 25 % abortion rate of capitula (Feldman and Lewis
1990).
Both species are capable of self fertilization (Warwick 1987). Estimates of seed set in self-fertilized individuals ranges from 0-10% to 8-50% in C. acanthoides and 0-
42% to 8-60% in C. nutans (Moore and Mulligan 1964, Warwick and Thompson 1989).
Outcrossing rates in C. nutans are variable (but generally less than 50%) and dependent
upon pollinator behavior, tending to be more outcrossed when pollinators are available
(Smyth and Hamrick 1984). A reduction in C. nutans seed set of 97% was observed when
15 pollinators were excluded, suggesting that perhaps pollinators may be involved even in self-fertilization (Jessep 1990). Outcrossing rates are higher in dense stands of C. acanthoides than in isolated stands (Warwick and Thompson 1989). In more dense stands, there are still significant departures from random outcrossing, presumably due to self-fertilization (Warwick and Thompson 1989).
Hybridization
C. nutans and C. acanthoides are known to hybridize in their native ranges (Stace
1975, Hegi 1987). The hybrids are also known as C. orthocephalus (Moore and Mulligan
1956). Hybridization in invaded ranges has mostly been studied in Ontario, Canada
(Moore and Mulligan 1956, 1964, Warwick et al. 1989, 1992). Hybridization between C.
acanthoides (2n=22) and C. nutans (2n=16) was first noticed in 1951 (Moore and
Mulligan 1964). Hybrids are almost completely sterile (Warwick et al. 1989). These
hybrids have intermediate numbers of chromosomes; individuals with odd chromosome
numbers mostly produced nonviable seeds (Moore and Mulligan 1956).
The hybrid scoring method of Moore and Mulligan (1956) involved the following
characteristics: head solitary or clustered, head nodding or erect, peduncles spiny winged
or not, phyllaries reflexed or not, phyllaries marked or unmarked and phyllaries
contracted at the base or not. Of the 41 apparently hybrid individuals selected for study,
several individuals appeared to be backcrosses with parental species (Moore and
Mulligan 1956). Warwick et al. (1992) caution against using morphological features to
identify hybrids because considerable overlap exists, and not all hybrids exhibit
intermediate morphology. This is in contrast to growth responses: Warwick et al. (1990)
16 found that hybrid growth patterns were similar to C. nutans and were sometimes superior to C. nutans.
Warwick et al. (1989) also found evidence of hybridization and backcrossing using more sophisticated techniques including allozymes and molecular data in the same area as Mulligan and Moore thirty years later. In the hybrid swarms they studied, they found the complete range of morphological variation of both species as well as intermediate morphologies. They also found some evidence of bi-directional introgression, although they conclude that there is a stable zone of hybridization in
Ontario which has changed little in thirty years, and that the introgression is of little importance (Warwick et al. 1989).
Moore and Mulligan (1956, 1964) speculated that there is selection for an increased chromosome number, since the seedling offspring of F1 hybrids tend to have higher chromosome numbers. The relatively better performance of C. acanthoides in this environment is a possible explanation for this (Moore and Mulligan 1964). Dense infestations of hybrids generally had lower hybrid indices, indicating closeness to C.
acanthoides. No evidence of selection during mating was found by Warwick and
Thompson (1989). In artificial hybridization studies, C. nutans was found to be the better
maternal parent (22% of the offspring were F1 hybrids, versus 0.6% for C. acanthoides),
although both species can serve as backcross parents (Warwick et al. 1989).
McCarty and Scifres (1969) were not able to find hybrids in Nebraska using the
Mulligan and Moore (1956) criteria (cited in Dunn 1976).
17 Dispersal
C. nutans and C. acanthoides both reproduce by seeds; there is no evidence of
asexual reproduction in either species (Desrochers et al. 1988). The wind-dispersed seeds
are also moved by water, birds, farm animals and vehicles (Medd and Smith 1978). In
addition, seed can be moved in contaminated hay bales (D. Mortensen, pers. comm.);
contaminated agricultural seed is also a source for spread (Medd and Smith 1978). The
dispersal of these species has been estimated by both seed trapping and seed tracking (i.e.
by following the dispersal of individual seeds) methods.
Seed tracking results from a study performed in fields suggest mean estimates of
dispersal ranging from 0.99-3.43 m for C. nutans and 1.32-4.22 m for C. acanthoides,
with maximum distance observed ranging from 22.33-35 m for C. nutans and 26.25-
52.55 m for C. acanthoides, depending on the vegetation characteristics (Skarpaas and
Shea, in revision). Dispersal distances of both species are negatively affected by the
height of surrounding vegetation and positively affected by release heights and horizontal
wind speeds, although updrafts do not seem to play a role in determining dispersal
distances (Skarpaas and Shea, in revision). Skarpaas and Shea (in revision) develop
mechanistic models using wind speeds, release heights and environmental roughness that
are good predictors of observed dispersal distances. Under artificial laboratory
conditions, Feldman and Lewis (1988) found that wind speed only affected the distance
C. acanthoides seeds traveled once they had landed; however, this is unlikely to be of
importance in the field, because the roughness of both bare ground and vegetation is
much greater than a laboratory floor. In the course of a demography study, Lee and
18 Hamrick (1983) released about 1200 seeds C. nutans seeds, although they do not mention how far the seeds moved in their results.
Seed trapping experiments were conducted for C. nutans by Smith and Kok
(1984) and Kelly et al. (1988), for C. acanthoides by Feldman and Lewis (1990) and for both C. nutans and C. acanthoides by Skarpaas and Shea (unpublished manuscript).
Smith and Kok (1984) placed thistle traps in paved areas, limiting the value of their study
for understanding dispersal in pastures, where surrounding vegetation can impact
dispersal distances (Skarpaas and Shea, in revision). Additionally, the heating of the
pavement at low wind speeds caused increased turbulence (Smith and Kok 1984), which
can also influence dispersal distances (Skarpaas et al. 2006). Smith and Kok (1984)
artificially dispersed heads collected in the field at a constant height of 1.5m. Their
mechanistic model, which includes wind speed, terminal velocity and turbulence, agrees
qualitatively with observed seed dispersal. Although they do not report mean or
maximum dispersal estimates, it appears that some seeds traveled further than 80 m.
Seeds in flowerheads that were attacked by R. conicus tended to remain within the thistle
heads and not be dispersed (Smith and Kok 1984).
Both Kelly et al. (1988) and Skarpaas and Shea (unpublished manuscript) used
actual thistle stands as a seed source. Kelly et al. (1988) set traps at different distances
and heights from a source. Kelly et al. (1988) conclude that wind dispersal is not
effective because seeds easily detach from pappi. While most seeds were generally found
in the closest (10 m) nets, some seeds were caught at 100 m; however, mean dispersal
distances are not given in Kelly et al. (1988). Feldman and Lewis (1990), in a seed
19 trapping study of C. acanthoides in Argentina, found most seeds were dispersed less than
2 m, although they found many seeds at their furthest trapping distance of 8 m.
Both Feldman and Lewis (1990) and Skarpaas and Shea (unpublished manuscript) found that wind direction affected dispersal for both species. The distribution of the distances was found to be highly skewed (Skarpaas and Shea , unpublished manuscript).
The longest dispersal distances recorded were 96 m for C. nutans and 16 m for C. acanthoides, and mean dispersal distances were 2.1 and 1.8 m for C. nutans in 2003 and
2004, and 1.6 and 2 m for C. acanthoides in 2003 and 2004 (Skarpaas and Shea , unpublished manuscript). Skarpaas and Shea (unpublished manuscript) use models to estimate an annual spread rate of 3 m per year for C. acanthoides and 10 m per year for
C. nutans. Stuckey and Forsyth (1971) report in detail the spread of one infestation of C. nutans and suggest that local spread of C. nutans may be extremely slow as most seeds fall below the parent plant.
Wind turbulence is an important component of seed release for C. nutans and C. acanthoides (Skarpaas et al. 2006). Seed release occurs at lower wind speeds in turbulent as compared to laminar wind flow (Skarpaas et al. 2006). Wind speeds necessary for release were higher in C. acanthoides than in C. nutans (Skarpaas et al. 2006).
Seedbank
Feldman and Lewis (1990) claim that in Argentina, the seedbanks of Carduus spp. (specifically C. acanthoides and C. nutans) are small and contain mostly non-viable seeds, which they attribute to predation by birds and rodents and a lack of dormancy. The
20 distribution of C. acanthoides seeds in the seedbank may not be random (Feldman and
Lewis 1990). Estimates of the number of C. acanthoides seeds in an Argentinean pasture were 99 seeds/m2 in fall and 40 seeds/m2 in spring, which may imply low persistence of
C. acanthoides in the seedbank (Feldman and Lewis 1990). In contrast to these low seedbank estimates, in New Zealand, Popay and Thompson (1980) recorded C. nutans seedbanks of between 3600-5300 (mean 3225) seeds/m2 in August and between 300-
1500 (mean 925) seeds/m2 in January, before seed shed.
Long-term seedbank studies have only been conducted for C. nutans. James et al.
(1998) found that burial depth had the greatest influence on seed survival: the highest survival was 32%, which was found at the deepest depth of 20 cm. Viability of seed buried in the top 2 cm of soil was found to decline quickly, with only 1-16% viability after 1 year. They estimate that it could take up to 80 years for seed viability to decline to
1% for seeds that get buried at greater depths. They do not discuss the likelihood of this occurring, although they suggest a few mechanisms for burial, such as cracks, cultivations or earthworms.
Popay and Thompson (1979) conducted a similar seedbank experiment with C. nutans in Australia, studying depths of 0-2 cm, 4-6 cm and 19-21 cm. Most of the seeds at the soil surface had disappeared or germinated by three years. Seeds buried 19-21 cm showed little reduction in viability in four years (Popay and Thompson 1979). They also conclude that burial depth had the most influence on seed survival (Popay and Thompson
1979). After 10 years, seeds buried at 5 and 20 cm were still viable, and it is projected that seeds buried at 20 cm would take 34-77 years to be eliminated (Popay et al. 1987).
21 In a study of C. nutans seedling emergence from the seedbank in pastures in New
Zealand, most seedlings were found to emerge in tilled plots, and least in grazed plots.
Seedlings continued to emerge for several years after no more seeds entered the seedbank
(Popay et al. 1987).
Germination
Germination of both C. nutans and C. acanthoides was well studied in a laboratory setting by McCarty et al. (1969). These authors manipulated temperature, planting depth, moisture, achene coat disturbance (scarification or acid treatment), growth regulators, pH, salinity, light and stratification. C. nutans achenes are larger (2-4 mm) and heavier (4 mg) than C. acanthoides (1-3 mm, 2 mg). Moist chilling of achenes led to greater decline in subsequent C. acanthoides germination than C. nutans. C. acanthoides achenes appeared to be slightly more heat tolerant than C. nutans achenes, but C. nutans achenes better tolerated brief cold periods. Higher germination was seen under alternating light levels. Emergence of both species was best at a depth of 0.5-1.0
cm in the soil, although C. nutans was able to germinate from greater depths than C.
acanthoides, which did not emerge from depths greater than 4 cm. Disturbance of the
achene coat generally reduced germination. A pH range of 3-9 led to good germination in
both species. Both species were sensitive to moisture stress, although C. acanthoides
appeared to be more sensitive. As this moisture response had the most pronounced effect
on germination of all factors studied, they suggest that this may explain why C.
acanthoides is less widely distributed. Light did not have a large impact on germination,
22 although somewhat lower germination was seen in the dark (McCarty et al. 1969). In contrast to this, Jessep (1990) found that shading strongly reduced germination. Higher and fluctuating temperatures have been shown to accelerate C. nutans germination
(Ahmed and Wardle 1991).
Although it has been claimed that there is no period of dormancy for C. nutans
(Doing et al. 1969, Lee and Hamrick 1983), Popay et al. (1987) suggest that C. nutans does have some innate dormancy which can be ended by four months dry storage. They suggest that dormancy allows some achenes to become buried before autumn rains cause germination (Popay et al. 1987). Jessep (1990) found highest germination rates at intermediate time since development of seeds. They note differences in germination rates of pale and dark seeds and suggest that there may be seed polymorphism in this species.
Microsite effects
Both species are strongly affected by the quality of the germination sites
available, particularly as some degree of release from competitors appears to be
necessary for germination (K. Shea, unpublished data). Both species generally germinate
well in larger gaps (Chapter 7, Ruggiero 2004).
In a study of artificially created gaps in pastures in New Zealand, C. nutans
emergence was highest in 10 cm gaps (gap sizes studied were 2, 6 and 10 cm in diameter,
as well as 3 x 1 m bare plots), although the opposite result was found in a paired
greenhouse experiment (Panetta and Wardle 1992). Trampling in the field killed all
seedlings in the large bare ground plots (Panetta and Wardle 1992). Most gaps in a New
23 Zealand pasture were less than 10 cm in size, and the largest gap diameter was 30 cm
(Panetta and Wardle 1992).
C. nutans seedlings are more likely to be in gaps from dead C. nutans than in areas which were not occupied by C. nutans in the previous year (Wardle et al. 1994a), although this may be due to the presence of more propagules. A lack of pasture cover may lead to more spring germinations (Popay Thomson and Bell 1987). Germination of
C. nutans is higher in sites with no litter, and in sites with the least topographic heterogeneity (Hamrick and Lee 1987). Light litter cover best promotes C. nutans establishment and growth (species composition of the litter was not specified; Hamrick and Lee 1987). Interestingly, in a study focusing on germination and early growth, the effects of intraspecific competition were largely obscured by microtopography, litter cover, moisture and their interactions (Hamrick and Lee 1987).
Wardle et al. (1995) claim that some pasture cover may be needed for a suitable microclimate for C. nutans germination, since seeds often had lowest germination in bare
ground plots. Perhaps desiccation is important in determining the optimal size of gaps
needed, as C. nutans was found to germinate best under conditions which reduced
evaporation (Hamrick and Lee 1987). Moisture conditions can also affect variation in the
timing of germination (Hamrick and Lee 1987).
The effect of microsite characteristics on C. acanthoides was studied in Argentina
by Feldman et al. (1994). Germination and establishment were studied in 5, 10 and 30 cm
gaps with and without litter cover; gaps of 30 cm were best for germination. Germination
of C. acanthoides was found to be higher in rough soil than smooth, and higher with less
24 light (Feldman et al. 1994). C. acanthoides is sensitive to desiccation during germination and early growth (Feldman et al. 1994).
The optimal conditions for germination are likely not the same as the optimal conditions for growth for either species. More C. nutans in bare plots flowered in the first
year (in contrast to a rate of 3 to 16% in grass and legume sown plots) and grew larger
than in plots with other potential competitors (Wardle et al. 1995), which is opposite to
the observed germination response. Feldman et al. (1994) point out that different
conditions are optimal for different stages of C. acanthoides establishment: closed
canopy or litter cover gave the best protection from predation, but gaps in litter cover
improve germination. In a study comparing the germination and survival of C. nutans and
C. acanthoides along a disturbance gradient (i.e. pasture, pasture edge and forest habitat),
although the highest germination rates were in the forest habitat, there was subsequently
100% mortality when plots were covered by thick leaf litter in autumn (Peterson-Smith
and Shea , unpublished manuscript).
Interactions between plant species
Presumably, the benefit of larger microsite size is related to lower competition
with other plant species. Competition with other species may cause the decline of
Carduus populations (Desrochers et al. 1998). In a study of six thistle species, C. nutans
was found to be very sensitive to competition during rosette stages (Austin et al. 1985).
This is perhaps not surprising as C. nutans also had the smallest achene weight of the
thistles studied, which may be particularly important in competition during the seedling
25 stages. C. nutans germination can depend on the height of surrounding vegetation (Kelly and McCallum 1990). C. nutans had higher mortality when grown with grasses than when grown in bare plots or with legumes, and thus may be more vulnerable to grasses than legumes (Wardle et al. 1995).
Daucus carota (Apiaceae) has been suggested to have contributed to the decline of Carduus populations in Ontario (Moore and Mulligan 1964). The abundance of C. acanthoides increased after Lotus tenuis (Fabaceae) was removed, and C. acanthoides was found less often than expected near more dense L. tenuis patches in a rangeland study in Argentina (Laterra 1997).
The use of other species’ allelopathic effects on C. nutans has been suggested as a management strategy (Wardle et al. 1992, 1996). C. acanthoides has been shown to be vulnerable to allelopathic affects of Lotus tenuis (Laterra and Bazzalo 1999), and C.
nutans is vulnerable to the effects of other pasture species (Wardle et al. 1992, 1996).
Speed of emergence, root and shoot growth of C. nutans has been shown to be negatively impacted by ten common pasture species in greenhouse experiments; grasses, particularly
Lolium perenne (Poaceae) and Holcus lanatus (Poaceae), seemed to be particularly inhibitory (Wardle et al. 1992). Greenhouse and bioassay results for the effects of these ten species on C. nutans were found to be consistent with field results (Wardle et al.
1996). C. nutans and Cirsium vulgare emergence had different responses to different pasture species, implying that it may not be possible to simultaneously manage emergence of both species through allelopathic interactions with other species, although their growth responses to other species were similar (Wardle et al. 1992). Additionally,
26 the species that most effectively suppress thistle seedling growth were not always the same as those that most inhibit germination (Wardle et al. 1992).
When considering two-way inhibitory effects, Lolium perenne was found to
exhibit inhibitory effects on C. nutans, but C. nutans did not negatively affect L. perenne.
Trifolium repens (Fabaceae) and C. nutans seemed to have reciprocal inhibitory effects
(Nicholson et al. 1990) . Wardle et al. (1991b) suggest planting L. perenne and clover
pastures to suppress C. nutans. Allelopathic effects of other pasture species seem to slow
C. nutans growth rather than stop it (Wardle et al. 1991b).
Allelopathic effects of C. nutans and C. acanthoides
Both C. nutans (Wardle et al. 1991a, Wardle et al. 1998) and C. acanthoides
(Woodward and Glenn 1983) have been reported to be allelopathic. C. nutans has been
found to be allelopathic at two stages: newly bolting adults (decomposition of old rosette
leaves release soluble inhibitors) and senescing adults (releasing insoluble inhibitors;
Wardle et al. 1993). The allelopathic potential of C. nutans may have evolved to
stimulate seedling growth (Wardle et al. 1991a). C. nutans has been found to be slightly
autoallelopathic when achenes were touching, but this also appeared to stimulate
germination at intermediate densities (Wardle et al. 1991a, Wardle et al. 1993). Wardle et
al. (1998) suggest that the allelopathic properties of C. nutans may be unique among
broadleaf weeds in pastures.
Species that have been tested for susceptibility (mostly using bioassays) to C.
nutans include the grasses Dactylis glomerata, Phalaris tuberose, Bromus wildenowii,
Lolium perenne, Festuca arundinacea and Holcus lanatus; the clovers Medicago sativa,
27 Trifolium pratense, T. subterraneaum and T. repens, and C. nutans itself (Wardle et al.
1991a, 1993). Grasses are more generally tolerant of allelopathy than legumes (Ahmed
and Wardle 1994). C. nutans reduced the amount and speed of germination in L. perenne
and in C. nutans itself only when adjacent achenes were touching, and T. repens
germination appeared to be enhanced by touching C. nutans achenes (Wardle et al. 1991).
Speed of germination of L. perenne and T. subterraneaum were inhibited by high
densities of C. nutans (Wardle et al. 1991). Flowering C. nutans severely decreased
acetylene reduction (and thus nitrogen fixation) in T. repens even when the thistle plant
had already died (Wardle et al. 1994a). The presence of vegetative C. nutans led to
increased root production in L. perenne, while flowering plants had the opposite effect
(Wardle and Nicholson 1996).
There may be stimulatory and inhibitory compounds in the thistle achene coat that
different species respond to differently. For example, Wardle et al. (Wardle et al. 1991a)
suggest that C. nutans may have a slight impact on the germination of some pasture
species but a possibly larger impact on radicle elongation.
Woodward and Glenn (1983) examined C. nutans, C. acanthoides and Cirsium
arvense for allelopathic affects on the germination and radicle elongation of crop species
(corn, turnip and soybeans) using root and foliar extractions. Foliar and root extracts
inhibited germination of turnip and soybean and radicle length in corn. Interestingly, C.
acanthoides was the most toxic to corn in this greenhouse study.
Wardle’s work on C. nutans does not meet the criterion of Inderjit and Calloway
(2003), who claim that in studies of allelopathy, resource effects and allelopathy should
be separated, exudates should be manipulated (for example by using carbon additions)
28 and studies should involve some understanding of natural concentrations of the allelochemicals in the soil. In the absence of all three of these, it is not possible to be sure that other factors such as an increase in biomass observed by Wardle (1994) might not have caused the patterns observed (Inderjit and Callaway 2003).
Competition between C. nutans and C. acanthoides
Competition between C. nutans and C. acanthoides has not been well studied,
despite the fact that they are known to occur together. Warwick et al. (1990), using a
target-neighbor design for C. nutans, C. acanthoides and their hybrids, found evidence
that C. nutans is a more aggressive competitor than C. acanthoides, and hybrids were
found to be potentially better competitors than C. nutans. In an observational study of
central Pennsylvania, the presence of C. nutans and C. acanthoides were found to be
negatively correlated (Allen and Shea 2006); however, this observational study is not
conclusive with respect to whether competition caused this pattern. In a series of response
surface competition experiments between the two species, strong competition was not
detected, nor was significant density dependence (Chapter 4). Additionally, a series of
fine-scale studies of the distribution of C. nutans and C. acanthoides demonstrated
positive correlations between the two species, which does not support the idea of strong
competition between them (Chapter 5).
Indirect interactions
There is some evidence that C. nutans may have both negative and positive
effects, mediated through shared predation, on other plant species. Interestingly, some
29 species appear to benefit from being near C. nutans because they are not grazed (Seefeldt et al. 2005).The presence of C. nutans has been shown to negatively affect seed
production in populations of the native thistles Cirsium undulatum in Nebraska indirectly
through the weevil Rhinocyllus conicus. Attack of C. undulatum by R. conicus increased with increasing proximity to C. nutans populations (Rand et al. 2004). More than half of the variation in attack by R. conicus on Cirsium undulatum could be predicted by C. nutans density (Rand and Louda 2004). This interaction appears to be asymmetric: it does not appear that C. nutans is harmed by proximity to C. undulatum.
Indirect interactions between C. nutans and C. acanthoides have not been well
studied. C. acanthoides may be less attacked by R. conicus in the presence of C. nutans
(E. Rauschert unpublished data).
Distribution
C. nutans and C. acanthoides have spread from their native ranges in Europe,
Asia and North Africa (Moore and Frankton 1974) to become pest species in North
America and South America, Australia and New Zealand (Holm et al. 1979). C. nutans
appears to have been introduced to North America in ballast dumps (Stuckey and Forsyth
1971). In Australia, C. nutans was a frequent contaminant of grass seed from New
Zealand and the United Kingdom (Doing et al. 1969).
The distribution of these species in their invaded ranges has been best studied in
the US. The first record of C. nutans in North America was in Harrisburg, Pennsylvania
(Stuckey and Forsyth 1971). Dunn (1976) surveyed botanists, weed biologists and
30 agronomists in the US about county levels of infestations of Carduus thistles. Over 10%
of the counties surveyed had potential economic infestations of C. nutans. Overlaying the
distributional maps of C. nutans and C. acanthoides (Dunn 1976), it appears that most of
the highest infestations of C. acanthoides were also in areas of high C. nutans infestation.
Interestingly, central Pennsylvania is not listed as having C. nutans infestations on
Dunn’s (1976) maps. In an intensive roadside survey of central Pennsylvania, Allen and
Shea (2006) found spatial segregation of C. nutans and C. acanthoides, which was not
explained by environmental covariates. In a study in Maryland, Tipping (1992) found that
western counties had higher densities of Carduus thistles in pastures and fields. Of the
three counties studied, the eastern county had much higher C. nutans densities than C.
acanthoides, whereas the western county had higher C. acanthoides densities (Tipping
1992). In Virginia, Kok and Roberts (1987) found a mostly segregated distribution of C.
nutans and C. acanthoides.
Doing et al. (1969) compared climatic and community characteristics of C. nutans
in its native ranges to potential habitat in Australia. Medd and Smith (1978) developed
physiological and phenological models to predict future infestations of C. nutans in
Australia, and suggest that these could be used to focus management efforts.
Both of these thistles are commonly found in pastures and on roadsides (Batra
1978). In Ohio, Stuckey and Forsyth (1971) commonly found C. nutans in moderately
grazed pastures, hayfields, on hilly areas and along roadsides, and rarely on cultivated
ground, in flatter terrain, and along railroad tracks. C. acanthoides is occasionally a weed
in crop systems, where its abundance was found to be higher in no-tilling treatments
(Tuesca et al. 2001). C. acanthoides was more commonly found on areas disturbed by
31 common voles, which had smaller areas, than those disturbed by wild boars (Milton et al.
1997). C. nutans and C. acanthoides grow particularly well on fertile soils over limestone
(Batra 1978), and C. nutans is also known to grow well over dolomite bedrock (Stuckey
and Forsyth 1971). Batra (1978) suggest that C. nutans and C. acanthoides are economic
problems due to their detrimental effects on agriculture in some of the most productive
areas. In the Canadian prairie, neither pH, Ca or Mg is associated with their distribution
(Harris unpublished, cited in Desrochers et al. 1998). In Italy, which is part of C. nutans’
native range, heavy infestations of C. nutans are often found in well-drained soil, loamy
sand in texture with low nutrients in the soil (Boldt 1976).
Economic Impact
Most estimates of the economic impact of C. nutans on pastures come from New
Zealand. Densities of 1000 C. nutans plants per hectare led to an 8% reduction in pasture
production between October and May in New Zealand (Thompson et al. 1987). In
another study of two New Zealand pastures, Kelly and Popay (1985) found that about 6%
of pasture was lost to thistles, with more than 3/4 of the losses due to C. nutans. The
adverse effects can be larger than the area occupied by the thistle itself; the impacts of
thistles may be 50% to 100% larger than plant diameters (Thompson et al. 1987).
Because grazing animals may avoid heavily infested areas in pastures, studies in
the absence of grazing may underestimate economic damage from thistle infestations. C.
nutans caused a 72% reduction in the amount of forage used by grazing animals in a
study in a New Zealand pasture (Seefeldt et al. 2005). Bolting individuals have the largest
32 impact (Thompson et al. 1987, Seefeldt et al. 2005). Thistles may contribute to the invasion of other species, as other pasture weeds may be more numerous in areas where thistles died (Thompson et al. 1987).
Control and Management
Biocontrol
Several species of natural enemies of the musk (C. nutans, C. thoermeri) and
plumeless (C. acanthoides) thistles have been used for biocontrol in the Northern US. In
some studies, they have been used together, as they target different life stages and parts
of the plant, and also in conjunction with other mechanical and chemical control methods
such as mowing and herbicides. Efforts have focused primarily on Rhinocyllus conicus,
Trichosirocalus horridus (formerly known as Ceuthorryncidius horridus), and Puccinia
carduorum.
Rhinocyllus conicus is a receptacle feeding weevil that reduces seed production. It
was first released in Virginia in 1969 for biological control of C. nutans and C.
acanthoides (Surles et al. 1974). There were 23 initial release sites, 12 of which were
reported to have sustained a population of weevils in the following years. Timing of
releases, number of weevils released and the availability of host plants were the most
important factors contributing to the success. Success of this receptacle feeding weevil
was reported in 1975 in Pulaski VA, with a 90% reduction in thistle densities in all but
one of 11 plots surveyed (Kok and Surles 1975). Successful establishment of R. conicus
was reported in 21 C. nutans sites in northern Georgia as well (Buntin et al. 1993). The
33 redistribution of the weevil in the Northern US continued until 2000 when its interstate
movement was banned by USDA APHIS Plant Protection and Quarantine (Louda et al.
2003), presumably due to concerns about non-target effects. Julien and Griffiths (1999)
list the states where R. conicus was released for the control of C. nutans and C.
acanthoides and where establishment was successful.
R. conicus has been reported to have had variable success worldwide (Shea and
Kelly 1998, Shea et al. 2005). Although in New Zealand seed destruction is similar to
that in North America, R. conicus has had little effect in controlling C. nutans (Kelly and
McCallum 1992), as it does not reduce seed production enough to limit population
growth rates (Shea and Kelly 1998). R. conicus has also been released in Argentina and
Australia (Louda et al. 2003). Although it has been observed to spread relatively well in
its introduced habitat, it has also expanded its host range (Louda et al. 2003) and is
negatively impacting the population dynamics of native thistles and native insects (Louda
et al. 1997).
Ceuthorryncidius horridus, a rosette feeding weevil, was renamed as
Trichosirocalus horridus by Colonnelli (1979 in (Alonso-Zarazaga and Sanchez-Ruiz
2002). Host specificity studies were carried out on T. horridus imported from Italy for the
biocontrol of thistles (Kok 1975). This weevil is native to western, central and southern
Europe and attacks thistles in the rosette stage before any seed production occurs in the
thistles. After confirming T. horridus feeding preference for thistles (Kok 1975), it was
reared in the laboratory and field releases were carried out in Virginia in 1974 and 1975
(Trumble and Kok 1979).
34 Successful establishment, defined as the presence of larvae in spring rosettes for 2 years after release and presence of adults in the second year, was confirmed for 5 out of
10 initial releases in Virginia (Kok and Trumble 1979). Three of these sites with successful establishment were musk thistle infestations, one was a plumeless thistle infestation and the fifth was a mixture of musk and plumeless thistles. The ovipositional preference of T. horridus was later evaluated in two fields infested with both C. nutans and C. acanthoides thistles in Virginia in 1980 (Sieburth and Kok 1982). There was a clear preference for larger rosettes of both species and C. nutans was the preferred species when the ratio of weevils to thistle rosettes was low. Successful biocontrol of C. nutans (Kok 1986 in Kok and Mays 1991) and C. acanthoides (Kok and Mays 1991b) thistles by T. horridus in the Northern U.S. was confirmed in Virginia.
It was suggested that R. conicus and T. horridus would not interfere with each other, as they attack different life-history stages of the thistles (Kok and Trumble 1979).
This assumption appears to be incorrect: T. horridus negatively impacts the development of R. conicus, possibly through its impacts on C. nutans plant quality, such as a reduction in the number of flowerheads produced (Milbrath and Nechols 2004a). Lower oviposition rates by R. conicus were observed on thistles infested by T. horridus (Milbrath and
Nechols 2004a). In Northeast Kansas, the combined effects of both species are not
enough to limit seed production by C. nutans (Milbrath and Nechols 2004b).
Puccinia carduorum is a rust fungus from the Mediterranean, which decreases
thistle biomass and seed production and hastens senescence (Kok 2001). Greenhouse
studies on the susceptibility of Cirsium, Cynara and Carduus species to P. carduorum
were carried out by Bruckart (1996) as part of a pre-release risk assessment study. Study
35 results showed that P. carduorum was not a good pathogen on species other than the musk thistle C. thoermeri and did not pose any risk of non-target effects. However, this study emphasized the fact that Carduus species or subspecies are distinct as C. nutans, contrary to C. thoermeri, had very low incidence of the disease. As specificity was
confirmed, this fungus was released in Virginia (Baudoin et al. 1993). It was also located
in Wyoming in 1996 (Julien and Griffiths 1999).
P. carduorum is naturally wind-dispersed, but Kok and Abad (1994) reported that
the three biological control agents of the musk thistle established in Virginia, R. conicus,
T. horridus and Cassida rubiginosa, were also potential vectors for its dispersal. The
effects of P. carduorum on these three musk thistle (C. thoermeri) herbivores are
reported in Kok et al. (1996).
There have also been isolated biocontrol attempts with other agents. The seed-
feeding tephritid fly Urophora solstitialis, was released in Maryland for the control of plumeless thistle, but did not become established. It was also released in Montana for the control of musk thistle but its establishment was not confirmed (Julien and Griffiths
1999). Cheilosia corydon, the root-crown fly was released for the control of musk thistle in a few states but its establishment has not been confirmed in any (Julien and Griffiths
1999).
Additional seed predation
Pre- and post - dispersal seed predation of C. nutans by rodents and birds was studied by McCallum and Kelly (1990) on a naturally infested farm in New Zealand. The majority of the pre-dispersal seed predation was attributed to goldfinches whereas seed
36 destruction by R. conicus was very low (3%). In a seed pre-dispersal seed predation study on C. nutans in its native range in France, R. conicus, Larinus spp. (Curculionidae) and the seed-feeding tephritid fly Urophora solstitialis were found to be the most damaging capitulum feeding insects (Sheppard et al. 1994).
C. acanthoides achenes on the soil experience heavy predation from rodents and insects (Feldman et al. 1994). Silverman (2006) found that most seeds of both species were removed by insects, rather than by small mammals, and that C. acanthoides seeds were removed at higher rates than C. nutans seeds.
Herbicides
Two of the most widely used herbicides in New Zealand pastures are 2, 4-
D/picloram mix, which is only lethal to broadleaf species, and glyphosate, which is lethal to both broadleaf species and grasses (Wardle et al. 1994b). The timing of herbicide relative to the thistle life cycle is of primary importance. In a study of two herbicides,
MCPA (potassium salt) and 2, 4-D, on C. nutans in New Zealand, the choice of herbicide
had little impact, but the timing of herbicide application was the most important factor
affecting the percentage of thistles killed (Popay et al. 1989). In Maryland, although the
choice of herbicide did not have a significant effect, the stage of musk thistle (C.
thoermeri) development when the herbicide treatment was applied influenced the
production and viability of seeds produced (Tipping 1991). Because there is inter-annual
as well as within-population variation in the phenology of C. nutans, differences were observed in the effect of various herbicides tested on the production of germinable and viable seeds in thistle infestations Nebraska (McCarty and Hatting 1975).
37 When these two herbicides (2, 4-D/picloram and glyphosate) were applied to
Carduus nutans leaf litter, it stimulated decomposition, in contrast to their temporary inhibitory effect on other species such as Trifolium repens and Lolium perenne. Root litter decomposition of C. nutans on the other hand was observed to be slower than corresponding shoot tissue. Results suggest that effects of herbicides on litter persistence and microbial colonization in pastoral systems are only short term (Wardle et al. 1994b).
Mechanical control
C. nutans and C. acanthoides are primarily pasture weeds; it may be possible to
control some of their impacts through selective grazing. Animal grazing as a weed
management strategy was reviewed by Popay and Field (1996). Grazing by goats has
been suggested as an efficient pre-dispersal management strategy for C. nutans in
Australia (Holst et al. 2004). Thistles in their vegetative state are not palatable to either
sheep or goats, but goats have been found to prefer and consume more thistle capitula
than sheep (Holst et al. 2004). The number of viable seeds excreted by goats that were
fed C. nutans seeds was negligible. Holst et al. (2004) propose stocking infested pastures
with goats in addition to other control methods (such as spray – grazing) to reduce
numbers of C. nutans.
Mowing is an efficient means to eliminate flowering capitula and prevent viable
seed production. The effect of clean mowing was a reduction in the number of flowers in
capitula and in the dry weight of capitula and flowers, as well as a delay in reproduction
and a reduction in number of capitula in C. acanthoides (Feldman and Lewis 1990). In
C. nutans, a single mowing event for a thistle infestation did not give effective results due
38 to the variation in timing of flowering (McCarty and Hatting 1975). In our experience, individuals will still produce capitula even if mowing has occurred several times (K.
Shea, unpublished data).
It is important to consider potential interactions when combining control methods, as well as their relative timing. Tipping (1991) emphasize that the timing of mowing and herbicide application are critical for achieving effective control, especially in the presence of a biocontrol agent. The stage of development at which musk thistles were mowed impacted the survival of the biocontrol agent, the thistle head weevil R. conicus
(Tipping 1991). Stoyer and Kok (1989) suggest reduced dosages of the herbicide 2, 4-D combined with T. horridus as an efficient means of biocontrol of Carduus thistles. Kok
(1980) proposes the combined use of 2, 4-D, T. horridus and R. conicus for biocontrol of
Carduus thistles in Virginia and contends that these biocontrol agents are compatible and that there are no adverse effects of the herbicide 2, 4- D on the growth and survival of the two biocontrol agents.
Few studies have compared the effectiveness of different management strategies, or Integrated Weed Management (IWM), but the recent field experiments (Huwer et al.
2005) and models (Shea et al. 2005) rank the outcomes of various management strategies for C. nutans. The best single strategy (lethal spraying) was not the same as the best combined strategy (non-lethal spraying and grazing in spring; Shea et al. 2005).
39 Conclusions
Although C. nutans and C. acanthoides have been the focus of considerable research, certain areas stand out as in need of further exploration. For example, more research including C. acanthoides is needed. C. nutans is generally better studied, perhaps because it is more widely distributed. In our experience, C. acanthoides infestations can be as detrimental to pasture quality as C. nutans infestations.
Remarkably, C. acanthoides does not appear to have reached Australia yet (Doing et al.
1969); if it does establish there, it may have the potential to become as large a problem as
C. nutans. Biocontrol agents of these species, such as R. conicus, are generally better synchronized phenologically with C. nutans, as flowering plants are in the appropriate stage (bud formation) when the weevils are active. The phenology of both species relative to biocontrol agents specifically and management practices in general would be a useful area of study.
Remarkably, little modeling work has been published on these species (but see
Shea and Kelly 1998, Shea et al. 2005, Shea et al. 2006). Particularly when considering management, it may be beneficial to use models to consider what potential impact various management options might have, in order to improve management efforts.
Management might also be improved by a better understanding of what prevents these species from being major pests in their native ranges. Both of these species have been predominantly studied in their invaded ranges, primarily in New Zealand, Australia, the US, Canada, and Argentina. Some exceptions to this general trend include studies related to finding biocontrol agents (e.g. Sheppard et al. 1994) and one study of
40 mycorrhizal fungal associations (Rydlova and Vosatka 2001). The approach of Jongejans et al. (in revision) may be particularly useful; using matrix models, they examine the dynamics of C. nutans in France to explore the impacts on population growth of several factors, including insect attack and grazing by sheep. Willis et al. (2000) also studied C. nutans from its native and invaded ranges using a common garden experiment. They demonstrate that C. nutans from its invaded ranges is not significantly larger than C. nutans from native areas, although the shoot mass was slightly higher in plants grown from seed from invaded areas (Willis et al. 2000).
Previous thistle reviews (Desrochers et al. 1988, Popay and Medd 1990, Sindel
1991) mostly predate the large body of work on allelopathy by Wardle et al. (1991a,
1991b, 1992, 1993, 1994a, 1995). Recent work, such as the identification of the role of
(-)-catechin for Centaurea maculosa (Bais et al. 2003), has led to a more mechanistic understanding of allelopathy. It would be useful to identify the mechanism of Carduus allelopathy, and verify whether it is important in the field.
Below-ground aspects are extremely poorly studied for both of these species. It is known that C. acanthoides is colonized by arbuscular mycorrhizal fungi (AMF) in its native ranges (Rydlova and Vosatka 2001), but the role that AMF may play in invaded ranges has not been studied. C. nutans has been shown to impact soil microbial biomass and thus nutrient cycling (Wardle and Nicholson 1996). Additionally, below-ground competition may be very important to thistle growth and establishment (Z. Sezen, pers. obs.), and the role of belowground herbivory is also not well understood.
In conclusion, many important aspects of the life-cycles of C. nutans and C. acanthoides, as well as the efficacy of various management tools have been studied. We
41 anticipate that the understanding gained from past studies will lead to targeted future research and hence to better management of these troublesome species.
Notes
This review was written in collaboration with Zeynep Sezen.
References
Ahmed, M., and D. A. Wardle. 1991. Increasing temperatures may enhance emergence
and seedling growth of nodding thistle, summer grass and spiny emex.
Proceedings of the 44th New Zealand Weed and Pest Control Conference:288-
291.
Ahmed, M., and D. A. Wardle. 1994. Allelopathic Potential of Vegetative and Flowering
Ragwort (Senecio-Jacobaea L) Plants against Associated Pasture Species. Plant
and Soil 164:61-68.
Allen, M., and K. Shea. 2006. Spatial segregation of congeneric invaders in central
Pennsylvania, USA. Biological Invasions 8:509-521.
Alonso-Zarazaga, M. A., and M. Sanchez-Ruiz. 2002. Revision of the Trichosirocalus
horridus (Panzer) species complex, with description of two new species infesting
thistles (Coleoptera : Curculionidae, Ceutorhynchinae). Australian Journal of
Entomology 41:199-208.
42 Austin, M. P., R. H. Groves, L. M. F. Fresco, and P. E. Kaye. 1985. Relative growth of
six thistle species along a nutrient gradient with multispecies competition. Journal
of Ecology 73:667-684.
Bais, H. P., R. Vepachedu, S. Gilroy, R. M. Callaway, and J. M. Vivanco. 2003.
Allelopathy and exotic plant invasion: From molecules and genes to species
interactions. Science 301:1377-1380.
Batra, S. W. T. 1978. Carduus thistle distribution and biological control in the
northeastern states.18-22.
Baudoin, A., R. G. Abad, L. T. Kok, and W. L. Bruckart. 1993. Field-Evaluation of
Puccinia-Carduorum for Biological-Control of Musk Thistle. Biological Control
3:53-60.
Boldt, P. 1976. Habitat of Carduus nutans L. in Italy and Two Phytophagous Insects.
Pages 98-100 in T. E. Freeman, editor. Proc. International Symposium on
Biological Control of Weeds.
Bruckart, W. L., D. J. Politis, G. Defago, S. S. Rosenthal, and D. M. Supkoff. 1996.
Susceptibility of Carduus, Cirsium, and Cynara species artificially inoculated
with Puccinia carduorum from musk thistle. Biological Control 6:215-221.
Buntin, G. D., R. D. Hudson, and T. R. Murphy. 1993. Establishment of Rhinocyllus-
Conicus (Coleoptera, Curculionidae) in Georgia for Control of Musk Thistle.
Journal of Entomological Science 28:213-217.
Desrochers, A. M., J. F. Bain, and S. I. Warwick. 1987. Biosystematic of Carduus-
Nutans L in Canada. American Journal of Botany 74:732-732.
43 Desrochers, A. M., J. F. Bain, and S. I. Warwick. 1988. The biology of Canadian weeds
89. Carduus nutans L. and Carduus acanthoides L. Canadian Journal of Plant
Science 68:1053-1068.
Doing, H., E. F. Biddiscombe, and S. Knelhans. 1969. Ecology and distribution of the
Carduus nutans group (nodding thistles) in Australia. Vegetatio 17:313-351.
Dunn, P. H. 1976. Distribution of Carduus nutans, C. acanthoides, C. pycnocephalus and
C. crispus, in the United States. Weed Science 24:518-524.
Feldman, S. R., and J. P. Lewis. 1990. Output and Dispersal of Propagules of Carduus-
Acanthoides L. Weed Research 30:161-169.
Feldman, S. R., J. L. Vesprini, and J. P. Lewis. 1994. Survival and establishment of
Carduus acanthoides L. Weed Research 34:265-273.
Giurfa, M., and J. A. Nunez. 1992. Foraging by Honeybees on Carduus-Acanthoides -
Pattern and Efficiency. Ecological Entomology 17:326-330.
Giurfa, M., and J. A. Nunez. 1993. Efficient Floret Inspection by Honeybees in Capitula
of Carduus-Acanthoides. Ecological Entomology 18:116-122.
Hamrick, J. L., and J. M. Lee. 1987. Effect of Soil Surface-Topography and Litter Cover
on the Germination, Survival, and Growth of Musk Thistle (Carduus-Nutans).
American Journal of Botany 74:451-457.
Hegi, G. 1987. Carduus Distel. in Ilustrierte Flora von Mitteleuropa. Paul Parey, Berlin.
Holm, L., J. V. Pancho, J. P. Herberger, and D. L. Plucknett. 1979. A Geographical Atlas
of World Weeds. John Wiley and Sons, New York.
44 Holst, P. J., C. J. Allan, M. H. Campbell, and A. R. Gilmour. 2004. Grazing of pasture
weeds by goats and sheep. 1. Nodding thistle (Carduus nutans). Australian
Journal of Experimental Agriculture 44:547-551.
Huwer, R. K., D. T. Briese, P. M. Dowling, D. R. Kemp, W. M. Lonsdale, D. L. Michalk,
M. J. Neave, A. W. Sheppard, and T. L. Woodburn. 2005. Can an integrated
management approach provide a basis for long-term prevention of weed
dominance in Australian pasture systems? Weed Research 45:175-192.
Inderjit, and R. M. Callaway. 2003. Experimental designs for the study of allelopathy.
Plant and Soil 256:1-11.
James, T. K., A. Rahman, D. A. Wardle, and K. I. Bonner. 1998. Survival of nodding
thistle (Carduus nutans) seed buried at different depths in four soils. Pasture
Weeds and Pests:33-37.
Jessep, C. T. 1990. Aspects of the biology of nodding thistle (Carduus nutans L.) in
Canterbury, New Zealand. New Zealand Journal of Agricultural Research 33:173-
183.
Jongejans, E., A. Sheppard, and K. Shea. in revision. What controls the population
dynamics of the invasive thistle Carduus nutans in its native range? Journal of
Applied Ecology.
Julien, M. H., and M. W. Griffiths. 1999. Biological Control of Weeds: A world
catalogue of agents and their target weeds (4th ed.).
Kelly, D., H. Cameron, and J. Alex. 1988. Wind dispersal of nodding thistle seeds and
pappi. New Zealand Weed and Pest Control Conference Proceedings 41:207-209.
45 Kelly, D., and K. McCallum. 1990. Demography, seed biology and biological control of
Carduus nutans in New Zealand. Pages 72-79 in J. P. Palmer, editor. The biology
and control of invasive plants: Proceedings of a conference of the B.E.S. British
Ecological Society, Cardiff.
Kelly, D., and K. McCallum. 1992. Evaluating the impact of Rhinocyllus conicus on
Carduus nutans in New Zealand. Pages 205-211 in R. R. Scott, editor.
Proceedings of the 8th International Symposium on Biological Control of Weeds.
DSIR/CSIRO, Lincoln University, Caterbury, New Zealand.
Kelly, D., and A. I. Popay. 1985. Pasture production lost to unsprayed thistles at two
sites. Pages 115-118 in Proceedings of the 38th New Zealand Weed and Pest
Control Conference.
Kok, L. T. 1975. Host specificity studies on Ceuthorhynchidius horridus (Panzer)
(Coleoptera: Curculionidae) for the biocontrol of musk and plumeless thistle.
Weed Research 15:21-25.
Kok, L. T. 1980. Compatibility of Rhinocyllus conicus, Trichosirocalus horridus and 2,4-
D for Carduus thistle control. Pages 441-445 in Proceedings of the 5th
International Symposium on the Biological Control of Weeds, Brisbane,
Australia.
Kok, L. T. 2001. Classical biological control of nodding and plumeless thistles.
Biological Control 21:206-213.
Kok, L. T., and R. G. Abad. 1994. Transmission of Puccinia carduorum by the musk
thistle herbivores, Cassida rubiginosa (Coleoptera: Chrysomelidae),
46 Trichosirocalus horridus and Rhinocyllus conicus (Coleoptera, Curculionidae).
Journal of Entomological Science 29:186-191.
Kok, L. T., R. G. Abad, and A. B. A. M. Baudoin. 1996. Effects of Puccinia carduorum
on musk thistle herbivores. Biological Control 6:123-129.
Kok, L. T., and W. T. Mays. 1991. Successful biological control of plumeless thistle,
Carduus acanthoides L. [Campanulatae: Asteraceae (= Compositae)], by
Trichosirocalus horridus (Panzer) (Coleoptera: Curculionidae) in Virginia.
Biological Control 1:197-202.
Kok, L. T., and J. E. Roberts. 1987. Biological control and management of musk and
plumeless thistles. 444-800, Virginia Tech.
Kok, L. T., and W. W. Surles. 1975. Successful biocontrol of musk thistles by an
introduced weevil, Rhinocyllus conicus. Environmental Entomology 4:1025-1027.
Kok, L. T., and J. T. Trumble. 1979. Establishment of Ceuthorhynchidius horridus
(Coleoptera: Curculionidae), an imported thistle-feeding weevil, in Virginia.
Environmental Entomology 8:221-223.
Laterra, P. 1997. Post-burn recovery in the flooding Pampa: Impact of an invasive
legume. Journal of Range Management 50:274-277.
Laterra, P., and M. E. Bazzalo. 1999. Seed-to-seed allelopathic effects between two
invaders of burned Pampa grasslands. Weed Research 39:297-308.
Lee, J. M., and J. L. Hamrick. 1983. Demography of two natural populations of musk
thistle (Carduus nutans). Journal of Ecology 71:923-936.
Louda, S. M., D. Kendall, J. Connor, and D. Simberloff. 1997. Ecological effects of an
insect introduced for the biological control of weeds. Science 277:1088-1090.
47 Louda, S. M., R. W. Pemberton, M. T. Johnson, and P. A. Follett. 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.
McCarty, M. K. 1982. Musk Thistle (Carduus-Thoermeri) Seed Production. Weed
Science 30:441-445.
McCarty, M. K. 1985. A nursery Study of Large-Flowered Taxa of Carduus. Weed
Science 33:664-668.
McCarty, M. K., and J. L. Hatting. 1975. Effects of herbicides or mowing on musk thistle
seed production. Weed Research 15:363-367.
McCarty, M. K., and C. J. Scifres. 1969. Life cycle studies with musk thistle. University
of Nebraska, Nebraska.
McCarty, M. K., C. J. Scifres, A. L. Smith, and G. L. Horst. 1969. Germination and Early
Seedling Development of Musk and Plumeless Thistles. University of Nebraska,
Nebraska.
Medd, R. W., and J. V. Lovett. 1978. Biological studies of Carduus nutans (L.) ssp.
nutans II. Vernalization and phenological development. Weed Research 18:369-
372.
Medd, R. W., and R. C. G. Smith. 1978. Prediction of Potential Distribution of Carduus-
Nutans (Nodding Thistle) in Australia. Journal of Applied Ecology 15:603-612.
Milbrath, L. R., and J. R. Nechols. 2004a. Indirect effect of early-season infestations of
Trichosirocalus horridus on Rhinocyllus conicus (Coleoptera : Curculionidae).
Biological Control 30:95-109.
48 Milbrath, L. R., and J. R. Nechols. 2004b. Individual and combined effects of
Trichosirocalus horridus and Rhinocyllus conicus (Coleoptera : Curculionidae) on
musk thistle. Biological Control 30:418-429.
Milton, S. J., W. R. J. Dean, and S. Klotz. 1997. Effects of small-scale animal
disturbances on plant assemblages of set-aside land in central Germany. Journal
of Vegetation Science 8:45-54.
Moore, R. J., and C. Frankton. 1974. The thistles of Canada.
Moore, R. J., and G. A. Mulligan. 1956. Natural hybridization between Carduus
acanthoides and Carduus nutans in Ontario. Canadian Journal of Botany 34:71-
85.
Moore, R. J., and G. A. Mulligan. 1964. Further Studies on Natural Selection among
Hybrids of Carduus Acanthoides and Carduus Nutans. Canadian Journal of
Botany 42:1605-&.
Nicholson, K. S., A. Rahman, and D. A. Wardle. 1990. Interactions between established
nodding thistle and pasture seedlings. Proceedings of the 43rd New Zealand Weed
and Pest Control Conference:225-228.
Panetta, F. D., and D. A. Wardle. 1992. Gap Size and Regeneration in a New-Zealand
Dairy Pasture. Australian Journal of Ecology 17:169-175.
Peterson-Smith, J., and K. Shea. unpublished manuscript. Invasion response of Carduus
spp. to disturbances across three contiguous habitats.
Popay, A. I., J. H. Butler, and F. A. Meeklah. 1989. Chemical Control of Nodding Thistle
(Carduus-Nutans L) in New-Zealand Pastures. Weed Research 29:21-28.
49 Popay, A. I., and R. W. Medd. 1990. The biology of Australian weeds 21. Carduus
nutans ssp. nutans. Plant Protection Quarterly 5:3-13.
Popay, A. I., and A. Thompson. 1979. Some aspects of the biology of Carduus nutans in
New Zealand pastures. Proceedings of the 7th Asian-Pacific Weed Science
Society Conference:343-346.
Popay, A. I., and A. Thompson. 1980. Nodding thistle: its growth and control. Ruakura
Farmers Conference Proceedings:63-69.
Popay, A. I., A. Thompson, and D. D. Bell. 1987. Germination and emergence of
nodding thistle (Carduus nutans L.). Proceedings of the Eighth Australian Weeds
Conference:175-178.
Popay, I., and R. Field. 1996. Grazing animals as weed control agents. Weed Technology
10:217-231.
Popay, I., and D. Kelly. 1986. Seasonality of emergence, and survival, of nodding thistle.
Proceedings of the 39th New Zealand Weed and Pest Control Conference:187-
191.
Rand, T. A., and S. M. Louda. 2004. Exotic weed invasion increases the susceptibility of
native plants attack by a biocontrol herbivore. Ecology 85:1548-1554.
Rand, T. A., F. L. Russell, and S. M. Louda. 2004. Local- vs. landscape-scale indirect
effects of an invasive weed on native plants. Weed Technology 18:1250-1254.
Rhodes, A. F., and T. A. Block. 2000. The Plants of Pennsylvania: An Illustrated Manual.
University of Pennsylvania Press, Philadelphia.
50 Ruggiero, D. 2004. Factors Affecting the Germination of the Invasive Thistles: Carduus
nutans and C. acanthoides. Honors Thesis. The Pennsylvania State University,
University Park.
Rydlova, J., and M. Vosatka. 2001. Associations of dominant plant species with
arbuscular mycorrhizal fungi during vegetation development on coal mine spoil
banks. Folia Geobotanica 36:85-97.
Seefeldt, S. S., J. M. C. Stephens, M. L. Verkaaik, and A. Rahman. 2005. Quantifying the
impact of a weed in a perennial ryegrass-white clover pasture. Weed Science
53:113-120.
Shea, K., and D. Kelly. 1998. Estimating biocontrol agent impact with matrix models:
Carduus nutans in New Zealand. Ecological Applications 8:824-832.
Shea, K., D. Kelly, A. W. Sheppard, and T. L. Woodburn. 2005. Context-dependent
biological control of an invasive thistle. Ecology 86:3174-3181.
Shea, K., A. Sheppard, and T. L. Woodburn. 2006. Seasonal life-history models for the
integrated management of the invasive weed nodding thistle Carduus nutans in
Australia. Journal of Applied Ecology 43:517-526.
Sheppard, A. W., J. M. Cullen, and J. P. Aeschlimann. 1994. Predispersal Seed Predation
on Carduus-Nutans (Asteraceae) in Southern Europe. Acta Oecologica-
International Journal of Ecology 15:529-541.
Sieburth, P. J., and L. T. Kok. 1982. Ovipositional preference of Trichosirocalus horridus
(Coleoptera: Curculionidae). The Canadian Entomologist 114:1201-1202.
51 Silverman, E. 2006. Colonization-Competition Trade-Off and Post-Dispersal Seed
Removal of Carduus nutans and C. acanthoides. Honors Thesis. The
Pennsylvania State University, University Park.
Sindel, B. M. 1991. A Review of the Ecology and Control of Thistles in Australia. Weed
Research 31:189-201.
Skarpaas, O., R. Auhl, and K. Shea. 2006. Environmental variability and the initiation of
dispersal: turbulence strongly increases seed release. Proceedings of the Royal
Society B-Biological Sciences 273:751-756.
Skarpaas, O., and K. Shea. in revision. Seed dispersal by wind: the relative importance of
environmental conditions and species characteristics. Journal of Ecology.
Skarpaas, O., and K. Shea. unpublished manuscript. Dispersal patterns, dispersal
mechanisms and invasion wave speeds for Carduus thistles.
Smyth, C. A., and J. L. Hamrick. 1984. Variation in Estimates of Outcrossing in Musk
Thistle Populations. Journal of Heredity 75:303-307.
Smyth, C. A., and J. L. Hamrick. 1987. Realized Gene Flow Via Pollen in Artificial
Populations of Musk Thistle, Carduus-Nutans L. Evolution 41:613-619.
Stace, C. A. 1975. Hybridization and the Flora of the British Isles. in Carduus L.
Stoyer, T. L., and L. T. Kok. 1989. Oviposition by Trichosirocalus horridus (Coleoptera,
Curculionidae), a biological control agent for Carduus thistles, on plants treated
with low dosages of 2,4-dichlorophenoxyacetic acid. Environmental Entomology
18:715-718.
52 Stuckey, R. L., and J. L. Forsyth. 1971. Distribution of Naturalized Carduus nutans
(Compositae) Mapped in Relation to Geology in Northwestern Ohio. The Ohio
Journal of Science 71:1-15.
Surles, W. W., L. T. Kok, and R. L. Pienkowski. 1974. Rhinocyllus conicus
Establishment for Biocontrol of Thistles in Virginia. Weed Science 22:1-3.
Thompson, A., S. A. E., and P. Martin. 1987. The effect of nodding thistle (Carduus
nutans) on pasture production. Proceedings of the 40th New Zealand Weed and
Pest Control Conference:222-225.
Tipping, P. W. 1991. Effects of Mowing or Spraying Carduus thoermeri on Rhinocyllus
conicus. Weed Technology 5:628-631.
Tipping, P. W. 1992. Density of Carduus and Cirsium thistles in selected areas of
Maryland. Weed Technology 6:434-436.
Trumble, J. T., and L. T. Kok. 1979. Ceuthorhynchidius horridus (Coleoptera:
Curculionidae): life cycle and development on Carduus thistles in Virginia.
Annals of the Entomological Society of America 72:563-564.
Tuesca, D., E. Puricelli, and J. C. Papa. 2001. A long-term study of weed flora shifts in
different tillage systems. Weed Research 41:369-382.
Tutin, T. G., V. H. Heywood, N. A. Burges, D. M. Moore, D. H. Valentine, S. M.
Walters, and D. A. Webb. 1964. Carduus L. in Flora Europaea. Cambridge
University Press.
Wardle, D. A., M. Ahmed, and K. S. Nicholson. 1991a. Allelopathic Influence of
Nodding Thistle (Carduus-Nutans L) Seeds on Germination and Radicle Growth
of Pasture Plants. New Zealand Journal of Agricultural Research 34:185-191.
53 Wardle, D. A., and K. S. Nicholson. 1996. Synergistic effects of grassland plant species
on soil microbial biomass and activity: Implications for ecosystem-level effects of
enriched plant diversity. Functional Ecology 10:410-416.
Wardle, D. A., K. S. Nicholson, and M. Ahmed. 1991b. Residual Allelopathic Effects of
Pasture Grasses and Legumes on Nodding Thistle Seedling Emergence and
Growth. Proceedings of the 44th New Zealand Weed and Pest Control
Conference:284-287.
Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1994a. Interference Effects
of the Invasive Plant Carduus-Nutans L against the Nitrogen-Fixation Ability of
Trifolium-Repens L. Plant and Soil 163:287-297.
Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1995. Influence of Pasture
Forage Species on Seedling Emergence, Growth and Development of Carduus
Nutans. Journal of Applied Ecology 32:225-233.
Wardle, D. A., K. S. Nicholson, and A. Rahman. 1992. Influence of Pasture Grass and
Legume Swards on Seedling Emergence and Growth of Carduus nutans L and
Cirsium-Vulgare L. Weed Research 32:119-128.
Wardle, D. A., K. S. Nicholson, and A. Rahman. 1993. Influence of Plant-Age on the
Allelopathic Potential of Nodding Thistle (Carduus-Nutans L) against Pasture
Grasses and Legumes. Weed Research 33:69-78.
Wardle, D. A., K. S. Nicholson, and A. Rahman. 1994b. Influence of Herbicide
Applications on the Decomposition, Microbial Biomass, and Microbial Activity
of Pasture Shoot and Root Litter. New Zealand Journal of Agricultural Research
37:29-39.
54 Wardle, D. A., K. S. Nicholson, and A. Rahman. 1996. Use of a comparative approach to
identify allelopathic potential and relationship between allelopathy bioassays and
''competition'' experiments for ten grassland and plant species. Journal of
Chemical Ecology 22:933-948.
Wardle, D. A., M. C. Nilsson, C. Gallet, and O. Zackrisson. 1998. An ecosystem-level
perspective of allelopathy. Biological Reviews of the Cambridge Philosophical
Society 73:305-319.
Warwick, S. I. 1987. Estimates of Outcrossing Rates in Pure Populations and Hybrid
Swarms of Carduus-Nutans and Carduus-Acanthoides. American Journal of
Botany 74:763-763.
Warwick, S. I., J. F. Bain, R. Wheatcroft, and B. K. Thompson. 1989. Hybridization and
Introgression in Carduus-Nutans and Carduus- Acanthoides Reexamined.
Systematic Botany 14:476-494.
Warwick, S. I., and B. K. Thompson. 1989. The Mating System in Sympatric Populations
of Carduus-Nutans, C-Acanthoides and Their Hybrid Swarms. Heredity 63:329-
337.
Warwick, S. I., B. K. Thompson, and L. D. Black. 1990. Comparative Growth-Response
in Carduus nutans, Carduus acanthoides, and Their F1 Hybrids. Canadian
Journal of Botany 68:1675-1679.
Warwick, S. I., B. K. Thompson, and L. D. Black. 1992. Hybridization of Carduus-
Nutans and Carduus-Acanthoides (Compositae) - Morphological Variation in F1
Hybrids and Backcrosses. Canadian Journal of Botany-Revue Canadienne De
Botanique 70:2303-2309.
55 Willis, A. J., J. Memmott, and R. I. Forrester. 2000. Is there evidence for the post-
invasion evolution of increased size among invasive plant species? Ecology
Letters 3:275-283.
Woodward, M. D., and S. Glenn. 1983. Allelopathic Effects of Three Thistle Species.
Proceedings of the 37th Annual Meeting of the Northeastern Weed Science
Society:114.
Chapter 3
Models of spatial segregation in Carduus nutans and C. acanthoides
Abstract
It has long been recognized that invasive species can negatively affect native species through competitive interactions. As the numbers of biological invasions increase, interactions between different invasive species will become increasingly important. We develop models of competition between Carduus nutans and C. acanthoides, two economically important invasive weeds that are spatially segregated in central Pennsylvania, USA. A landscape simulation model examines the patterns created by these two species invading into the same region. A smaller-scale, field simulation model investigates spatially explicit interactions between these species. The results of these spatially-explicit models are generally consistent with the results of classic Lotka-
Volterra competition models, with widespread coexistence predicted if interspecific effects are weaker than intraspecific effects for both species. However, spatial segregation can be maintained between the two species under certain circumstances.
Introduction
According to the Competitive Exclusion Principle (Hardin 1960), if two species require the same limited resources, the species that is better at competing for those resources is expected to competitively exclude the other over time. Many species, such as
57 plants, do compete in similar ways for the same resources, but manage to coexist stably, implying the operation of coexistence-promoting mechanisms. Although several authors have made the general statement that for coexistence to occur, intraspecific effects must be greater than interspecific effects, there is a lack of clarity on what exactly is meant by this statement, and how this relates to classic competition models.
Much attention has been given to mechanisms promoting coexistence among similar competitors (Tilman 1982, Hulme 1996, Chesson 2000, Molofsky et al. 2001,
Levine and Rees 2002, Amarasekare 2003, Murrell and Law 2003, Abrams and Wilson
2004, Roxburgh et al. 2004, Shea et al. 2004, Snyder and Chesson 2004). Coexistence can be maintained via competition-colonization trade-offs (Levins and Culver 1971), resource partitioning in heterogeneous environments, or through environmental fluctuation dependent mechanisms such as the storage effect (Chesson and Warner 1981).
Frequency dependence can lead to coexistence through the formation of stable bands
(Molofsky et al. 2001). In particular, the role of spatial structure in promoting coexistence has received attention (Levin 1974, Neuhauser and Pacala 1999), although Wang et al.
(2005) find that, for metapopulation models, including spatial structure does not necessarily facilitate coexistence.
Many competition studies focus on well-mixed populations; however, the spatial location of individuals will determine how they actually interact (Coomes et al. 1999). If species are separated spatially, this will reduce levels of competition experienced and may stop or slow competitive exclusion (Pacala 1986, Neuhauser and Pacala 1999). In fact, considerable spatial segregation can be created through competitive interactions themselves (Weiner and Conte 1981, Seabloom et al. 2005); competition can help
58 maintain spatial structure, and spatial structure can mediate the strength of interactions between species (Coomes et al. 1999). Initial distribution of species has also been shown to be important in determining the outcome of competition (Silvertown et al. 1992).
Invasive species offer unique opportunities to examine competition and coexistence because such systems are often heavily perturbed, and may not have yet reached a steady state. In fact, many well documented cases of competition involve invasive species (Crowell 1968, MacKinnon 1978). As the ecological and economic problems from invasive species continue to grow (Vitousek et al. 1997), understanding their role in the communities that they invade will become of increasing importance
(Shea and Chesson 2002). While much research has focused on studying the interactions between native and exotic species, few studies have acknowledged that invasive species can also interact with each other.
We are interested in the spatial patterns of Carduus nutans and C. acanthoides, two very similar congeneric invasive species, in central Pennsylvania. Their spatial distributions have been described recently by Allen and Shea (2006). In an extensive roadside survey over an area of 100 x 50 km, they found considerable spatial segregation between the two species with an area of overlap with low densities of both species. This distribution was not well correlated with environmental covariates, nor is it a case of range limitation for either species (both are found across the US and in Canada). Both species are non-native to North America, and given that both have been in Pennsylvania for over 100 years, it is surprising that they are not distributed more widely in the state.
The stability of this pattern is currently unknown, although it has been relatively stable
for four years (Chapter 5).
59 It is rarely possible to discern process from pattern, and it is often difficult to demonstrate whether pattern is intrinsically or extrinsically produced (Wilson et al.
1999). Possible causes of this pattern include spread history, habitat preferences, reproductive interference, different management practices in different areas, direct competition between the species, and indirect competition through shared predators or pollinators. Historical records are not adequate to dismiss spread history, although given that the first record of C. nutans in North America was at the southern boundary of Allen and Shea’s (2006) survey area, it seems surprising that it has spread across the country but not 50 km to the north. Habitat preference is also unlikely to explain this distribution:
Allen and Shea (2006) did not find any particular environmental correlates that explained
the pattern seen, and C. nutans can grow well in C. acanthoides areas in Pennsylvania (E.
Rauschert, pers. obs.). Although the species are known to hybridize in their native ranges
(Hegi 1987) and in Canada (Moore and Mulligan 1956, Warwick et al. 1989), this
appears to be an unlikely explanation since considerable self-fertilization occurs and most
hybrids are sterile (Warwick et al. 1989). Additionally, the overlap in phenology is likely
smaller in Pennsylvania than in Ontario, thus many flowers of both species are only
pollinated by their own species. Management practices are fairly uniform across this area
and are unlikely to cause segregation (A. Gover, pers. comm.).
For these reasons, we have chosen to focus on models of direct interactions
between the two species as a possible mechanism underlying their segregated
distribution. We discuss what is meant by interspecific versus intraspecific interactions in
terms of classical competition models in general and in the special case of similar
species. We develop spatially-explicit models both at the landscape scale and at the field
60 scale to describe direct competition between these species. Simulation models allow us to focus on the relevant unique initial conditions, with both species invading from opposite ends of a landscape or a field. We examine the resulting abundances and spatial patterns that are created under different competition scenarios and compare this to predictions of classic competition models.
Methods
Study System
Carduus nutans L. and C. acanthoides L. (Asteraceae) are Eurasian-origin
monocarpic perennials with similar life histories. Both are commonly found in disturbed
areas such as pastures, marginal land and on roadsides (Batra 1978). They live as rosettes
for some time and then flower following a period of vernalization (Desrochers et al.
1988). They are quite similar in size, especially during the rosette stages. Pollination of
C. nutans is accomplished by several insect species, predominantly Bombus spp. An
individual flower may only be visited by a few pollinators (Smyth and Hamrick 1987);
this behavior can be affected by population densities of the plants, with fewer visits per
flower head at higher population densities (Heinrich 1976). C. nutans and C. acanthoides
both reproduce by means of wind-dispersed seeds; there is no evidence of asexual
reproduction, such as clonal growth, in either species (Desrochers et al. 1988). Seeds can
remain viable for 20 years (Kok 2001), and there appears to be little or no inherent
dormancy (Doing et al. 1969, Lee and Hamrick 1983, Feldman and Lewis 1990). Most
seeds that are not buried germinate quickly (Popay and Thompson 1979). Their
61 interspecific competitive interactions have been little studied (but see Warwick et al.
1990 for greenhouse study).
Classical competition models
The Lotka-Volterra model was developed as a simple way to investigate the
outcome of competition between species (Lotka 1924). The Lotka-Volterra model for two
species is generally formulated as:
dN1 r1N1 = (K1 − N1 −α12N2 ) 3.1 dt K1
dN 2 r2 N 2 = (K 2 − N 2 − α 21 N1 ) 3.2 dt K 2
where N1 is the population density of species 1, K1 is the carrying capacity, r1is the intrinsic rate of increase, and α12 is a “competition coefficient”, which is the amount by
which an individual of species 2 lowers the population growth rate of species 1 (Case
2000). The ordering of the subscripts allows α12 to be interpreted as the effect on species
1 by species 2. Note that this model does not include α11 or α22, the intraspecific
coefficients, which are implied to be equal to 1.
Typically, this coupled differential equation model is analyzed by looking at the
zero-growth isoclines in phase space (N2 versus N1), by setting the population growth rate
equal to zero (Figure 3-1), thus:
r1N1 0 = (K1 − N1 −α12N2 ) 3.3 K1
62
K1 − N1 N2 = 3.4 α12
And similarly for N2:
K 2 − N 2 N1 = 3.5 α 21
The outcome of the competitive interaction depends on the relative magnitudes of
K1 K 2 versus K2, and versus K1. α12 α 21
There are four possible outcomes, (besides the trivial equilibrium if the system only starts with one species):
K1 K 2 Coexistence, if > K 2 and > K1 α12 α 21 3.6
K1 K 2 Species 1 only, if > K 2 and < K1 3.7 α12 α 21
K1 K 2 Species 2 only, if < K 2 and > K1 3.8 α12 α 21
K1 K 2 Dependent on initial conditions, if < K 2 and < K1 α12 α 21 3.9 (one species only persists or unstable coexistence)
Note that the coexistence conditions can also be derived through the “invasion criteria”: by requiring positive growth rate at low densities of one species when the other species is already established (at its carrying capacity).
63 Chesson (2000) formulates this model in a different way. He uses the absolute competition coefficients, rather than the relative coefficients of the classic Lotka-Volterra model. He also uses α for competition coefficients; however, for clarity and comparison purposes, we use α′ to refer to Chesson’s absolute competition coefficients. The Chesson
formulation is:
dN 1 = r N (1−α′ N −α′ N ) 3.10 dt 1 1 11 1 12 2
dN 2 = r N (1−α′ N −α′ N ) 3.11 dt 2 2 22 2 21 1
The absolute competition coefficients are related to the more commonly known
relative competition coefficients and K as follows:
1 α11′ = 3.12 K1
α12 α12′ = 3.13 K1
1 α 22′ = 3.14 K 2
α 21 α 21′ = 3.15 K 2
There are several obvious benefits to the Chesson formulation. It does not
explicitly include a carrying capacity, K, which has been criticized as being difficult to
measure and as being subject to size-bias (Connolly et al. 2001). This formulation also
readily allows incorporation of multiple stages (such as juveniles and adults) and
competition between different life stages.
64 The four competitive outcomes described above are unchanged; the criteria for the four outcomes in terms of absolute coefficients are:
Coexistence, ifα 22′ > α12′ and α11′ > α 21′ 3.16
Species 1 only, ifα 22′ > α12′ and α11′ < α 21′ 3.17
Species 2 only, ifα 22′ < α12′ and α11′ > α 21′ 3.18
Dependent on initial conditions, ifα 22′ < α12′ and α11′ < α 21′ 3.19
This formulation better allows comparison of competition coefficients without
need for reference to carrying capacities. Coexistence occurs if for both species, their
absolute intraspecific effect is stronger than their absolute interspecific effect (the impact
each has on the other species). Another way to state this is that one additional individual
of species 1 depresses the species 1 growth rate more than it depresses the species 2
growth rate; thus it is impossible for species 1 to increase in population size without
harming itself more than it harms species 2 and vice versa. The mathematical
implications of this were recognized by Lotka and Volterra (Lotka 1924). The
implications of this have sometimes been stated as “intraspecific effects are stronger than
interspecific effects” (Chesson 2000); to be more precise, what this means is that for each
species the intraspecific effect on its own growth rate is larger than the interspecific effect
on the other species’ population growth rate. Exclusion occurs when adding another
individual of species 1 depresses the growth rate of species 2 more than the growth rate
of species 1.
65 Special case: similar species with similar carrying capacity
For species, which have a similar K, Equations 3.1 and 3.2 can be simplified:
dN1 r1 N1 = (K − N1 −α12 N 2 ) 3.20 dt K
dN 2 r2 N 2 = (K − N 2 −α 21 N1 ) 3.21 dt K
In this case, the coexistence condition becomes:
α12 < 1 and α 21 < 1 3.22
Recall that in the classic formulation, α11 and α 22 are inherently set to 1. Looking at the equations, it is apparent that this condition will make adding an individual of species 1 have more of an effect than adding an individual of species 2 on the growth rate of species 1; however, this is more clearly seen mathematically using the Chesson formulation:
If K1 = K2, then from Equations 3.12 and 3.14,α11′ = α 22′ . These two variables are thus interchangeable with each other in Equation 3.16 , leading to:
Coexistence, ifα11′ > α12′ and α 22′ > α 21′ 3.23
In this sense, “intraspecific effects have to be stronger than interspecific effects,” although this now means something different: Species 1 can coexist with species 2 if, for both species, their growth rates are more affected by intraspecific competition than by interspecific competition with the other species.
In the literature, there is some confusion about what leads to competitive exclusion. For example, Vila and Weiner (2004) claim that an invader is a better
66 competitor if the effect it has on natives is larger than the effect natives have on it (thus looking at relative interspecific competition effects). This is not what is predicted by interspecific competition models.
For the non-native congeners, C. nutans and C. acanthoides, which are very
similar in terms of size and life-history requirements, it is likely that the carrying
capacities are similar, and thus it is likely that the effects of intraspecific competition are
approximately equivalent for these species. Based on these simple models, we would
predict that they may not be able to coexist, if interspecific interactions are stronger than
intraspecific interactions in either sense described above.
The Lotka-Volterra models separate possible outcomes into four areas based on
the relative strengths of intraspecific versus interspecific effects. Using spatially-explicit
simulation models, we examine a range of interspecific competition coefficients that
cover the cases described for the analytical models (Figure 3-2) to assess the pattern
formation under the different conditions. We focus on cases where C. nutans interspecific
effects would be greater, as laboratory experiments suggested that this may be the case
(Warwick et al. 1990). We therefore only investigate the case where C. nutans
interspecific effects on C. acanthoides (α ′NA , using N to denote C. nutans, and A to denote C. acanthoides) are greater than or equivalent to C. acanthoides interspecific
effects on C. nutans (α ′AN ). Since all other parameters in the models are equivalent (i.e. the models are symmetric), the model results would be identical but reversed if C.
acanthoides coefficients were greater. Thus we examine three Lotka-Volterra cases:
where interspecific effects are smaller than intraspecific effects, where interspecific
67 effects are greater than intraspecific effects only for one species (C. nutans), and where interspecific effects are stronger than intraspecific effects for both species. As the
outcomes depend only on the relative intraspecific versus interspecific effects, we keep
the same intraspecific coefficients, taken to be equivalent for these species, and vary the
interspecific effects to examine the range of behavior.
Landscape model
A series of simulation models were developed (in C++) to examine the dynamics
of this field system with particular initial conditions: two species invading from opposite ends of a landscape. The landscape model is a coupled map-lattice model where each cell represents one thistle patch. We examine a landscape 60 patches long and 20 patches wide. Patches may contain individuals of both species. This model is different from the classic competition models in one key aspect: immigration and emigration are included through seed dispersal. Dispersal can occur between neighboring thistle patches only, so the size of a patch is a reflection of actual dispersal distances in the field. Thus patches are considered to be approximately 100 m x 100 m in size, and the arena size is 6 km x 2 km. Boundaries of the landscape are reflecting, as we are not interested in the particular properties of a landscape of this size. Competition within and between species takes place only within a patch using a discrete Lotka-Volterra model, thus it does not address the
spatial competition issues raised by Pacala (1997). Within a patch, the mean-field model
is used, ignoring potential heterogeneities in the actual amount of competition
experienced.
68 Demographic stochasticity is included as population sizes within a patch can only have integer values. The fractional portion of the number of individuals predicted is accounted for by comparing it to a random number between 0 and 1, and rounding up if the fractional portion is greater than the random number. We include demographic stochasticity in this way rather than by simply truncating (rounding down from fractional individuals), because this would exclude rare invasion events that can be important.
A list of model parameters used is shown in Table 3-1. Population growth rates
are based on growth rates for C. nutans in Australia (Shea et al. 2005). In all simulations,
both species are considered to be equivalent except in terms of their interspecific
competition coefficients. Competition coefficients are formulated similarly to Chesson’s
absolute competition coefficients (Chesson 2000). The range of competition coefficients
used is presented in Figure 3-2. The dynamics as two spreading waves meet may be
different than the dynamics of one species invading into the established area of another
species because patches at the front of an invasion have fewer individuals. In order to
mimic invasion of these two species from two different areas of introduction, we simulate
what happens when two invading waves meet. First the model is initialized with small
populations of C. nutans and C. acanthoides at opposite ends of the landscape (length-
wise). The model is run until the two populations first meet in the middle. The resulting
landscape is saved and used as a starting landscape for all subsequent simulations, to
minimize computational time simulating dynamics when the two species do not yet
interact. We use the same basic parameter values except for the interspecific competition
coefficients (the effect of C nutans on C. acanthoides and vice versa). All simulations
presented were run 1000 times for 100 generations, and the outcomes were examined.
69 Generally, the criterion for coexistence is whether or not species can recover from low abundance (MacArthur and Levins 1967, Chesson 2000). This is often referred to as the “invasion” criteria because it is identical to a few individuals invading into an established community. This criterion is rather restrictive, and is not appropriate for our models of this system, where we already know that at least to a limited extent, these species can co-occur. In this instance, we wish to examine two species invading into each other’s range. Three outcomes are possible: 1) The two species interact, and only one is able to invade the other’s range, driving the other species extinct, 2) The two species spread through each other’s ranges and both occupy patches across the full range (i.e. widespread coexistence), or 3) The two species cannot fully invade into each other’s ranges (i.e. segregation is maintained). In the latter case, a stable overlap band may develop. In a sense, we are still interested in the invasion criteria, but with the addition that we are also interested in how a species growth rate from low densities is affected by the density of the population of the other invader in the system. How this interaction plays out has implications for how wide the overlapping “band of coexistence” in the middle will be.
In our results, we focus on the pattern created. We focus on the patterns created after 100 years, to roughly mimic how long the Carduus thistles are known to be in
Pennsylvania. We quantify any spatial pattern that arises by examining how the number of thistles in each field changes across the landscape. We quantify the relative invasion distance by comparing the distance invaded when both species are present to the speed of the invasion when only one species is present. We first run the model for 100 years with only one species present (separately for both species) and record how far across the
70 landscape the species invaded. We then run the model for 100 years with both species present to compare how far a species spreads when both species are present relative to when in isolation. We refer to this as the relative invasion distance. We also examine how far the species spread in 50 versus 100 generations, in order to examine whether or not the species are still spreading. A ratio of around 0.5 indicates that spread is similar from 0
to 50 years compared to between 50 and 100 years. Ratios closer to 1 indicate that most
spread occurred earlier in the process. The width of overlap observed after 100
generations is also quantified. We examine a longitudinal slice of the landscape
graphically, in order to avoid potential edge effects (Pacala and Silander 1985).
A within-field model
In order to include spatially explicit competition as well as dispersal, and to examine a smaller spatial scale, we develop an individual based model. In this coupled map-lattice model, each cell represents the space necessary for one individual, and can only be occupied by one individual. Physically, we assume that this space is approximately 20 x 20 cm. We model a patch of 60 x 60 cells, which has a maximum population size of 3600 individuals. The plants in the model are assumed to have a winter annual life cycle, which is most commonly observed in our field experiments. First fecundity is calculated, and then seed dispersal occurs, followed by lottery competition for germination sites. Random mortality is assumed to occur during winter, and all individuals are assumed to be reproductive the following year. The parameters for this model are listed in Table 3-2.
71 Thistles disperse randomly locally to locations within a dispersal neighborhood, and seeds that disperse off the field are assumed to be lost. The dispersal distances are calculated using the WALD (inverse Gaussian) model. The parameters of this model are related to biologically measurable phenomena, such as release heights, terminal velocities, mean horizontal wind speeds and wind turbulence (Katul et al. 2005).
Estimates for these species (in meters) are taken from Skarpaas and Shea (unpublished manuscript) and are noted in Table 3-2. Within a field, we assume that these species have approximately equivalent dispersal distances. Using these parameters, we draw inverse
Gaussian distributed numbers for dispersal distances (using the PROB library in C++, developed by Burkardt (2005)). The direction of dispersal is assumed to be random.
Seeds are assumed to disperse to the new locations; seeds that disperse off the field are lost. In contrast to the landscape model, dispersal is not restricted to nearest neighbor cells only.
Although the Lotka-Volterra model is a population level model, not an individual level model, this type of a simulation model will converge on the Lotka-Volterra model in the limit of infinite competition or dispersal neighborhoods (Pacala 1997). We examine a range of competition parameters for the reduction in fecundity (see Figure 3-
2), but assume that the two species have equal germination probabilities (thus there is no asymmetry in competition at the germination stages). The results from the models using a more realistic 10,000 seeds produced per individual (Kok 2001) was similar to using
1,000 seeds per individual, thus for computational efficiency, we only simulate 1000 seeds per individual. A proportion of seeds attempt to germinate; those that fail are assumed to die, and the rest remain in the seedbank.
72 The model was initialized with both species invading at opposite ends of the field.
Dispersal is sufficiently extensive to allow some individuals to arrive at the opposite end of the field quickly. For the output of this model, we examine the average number of C. nutans and C. acanthoides present after 100 generations from 100 trials for each combination of competition coefficients. Whenever one species is predicted to go extinct
(no individuals or seeds are present), the run is terminated, and the time to extinction is recorded. Graphically, we examine the number of individuals of each species present at each distance across the field, averaged over 100 trials.
Results
Landscape model
The results of the landscape model generally agree with the predictions of the Lotka-
Volterra model. The overlap and invasion distances are summarized for all parameter
combinations in Figure 3-3. The largest area of overlap was seen when interspecific
effects were small for both species. C. nutans was able to invade furthest when the effect
of C. acanthoides on C. nutans was small, and when the effect of C. nutans on C.
acanthoides was larger. C. acanthoides was only able to invade far when interspecific
interactions were small for both species. The results of symmetric interspecific effects
after 50 and 100 years are shown in Figure 3-4. Totally symmetric cases are only possible
for two of the Lotka-Volterra states (it is not possible to have interspecific effects greater
than intraspecific effects for C. nutans only and yet have competitive equivalence). When
interspecific interactions are weaker than interspecific interactions for both species, both
73 species co-occur at many points in the landscape (Figure 3-4a). Figure 3-4b shows the
case of complete equivalence of all interspecific and intraspecific coefficients. In this
case, the area of overlap between the two species is still expanding, although at a slower
rate than when interspecific effects are smaller than intraspecific effects (Figure 3-4a).
When interspecific effects are greater than intraspecific effects (Figure 3-4c), invasion is
not possible, and the narrow area of overlap created is stable (i.e. is the same at 50 and
100 years).
The results of three asymmetric runs after 50 and 100 years, which correspond to
the three Lotka-Volterra situations examined, are shown in Figure 3-5. As in the
symmetric case, when interspecific effects are smaller than intraspecific effects for both
species, widespread coexistence occurs, with C. nutans present at higher densities than C.
acanthoides (Figure 3-5a). When interspecific interactions are greater than intraspecific
interactions for C. nutans only (Figure 3-5b), the Lotka-Volterra model predicts that C.
nutans will exclude C. acanthoides, which can be seen in the shift of the overlap zone
toward the right. Eventually, C. nutans would drive C. acanthoides out of the landscape.
Note that Figure 3-5b is not a counterpart to Figure 3-4b, which has symmetric
interspecific effects. When interspecific effects are stronger than intraspecific effects for
both species, but interspecific effects are not equivalent (i.e. the interspecific effect of C.
nutans is greater than the interspecific effect of C. acanthoides, Figure 3-5c), an area of
overlap occurs with a stable width but an unstable location (i.e. it moves as the species
with stronger interspecific effect expands its range at the expense of the other species).
When interspecific interactions are stronger than intraspecific interactions for
both species, the Lotka-Volterra model predicts either an unstable coexistence or one
74 species will drive the other species extinct, dependent on the initial conditions. Unstable coexistence is not likely to be maintained in a model including even a small amount of demographic stochasticity, this for our models this translates to a prediction of one species driving the other species extinct. There are still co-occurrences of both species in a narrow area (Figure 3-4c and 3-5c). This area of overlap is small relative to the overlap when intraspecific interactions are stronger than interspecific interactions. This overlap zone is stationary when interspecific effects are the same for both species (Figure 3-5 c) but moves when C. nutans has a larger interspecific effect than C. acanthoides (Figure 3-
4c), indicating that again C. nutans would drive C. acanthoides out of the landscape.
Note that we only examine cases where C. nutans interspecific effects are stronger than
or equal to C. acanthoides effects; the results would be similar but reversed when
examining cases of greater C. acanthoides interspecific effects.
Figure 3-6 summarizes the results with respect to the three Lotka-Volterra
situations; the parameter values for these are described in Figure 3-2. The largest overlap
occurred when interspecific effects were smaller than intraspecific effects, and the
smallest overlap occurred when interspecific effects were stronger than intraspecific
effects (Figure 3-6a). C. nutans spread was not slowed at all when interspecific
interactions were weaker than intraspecific interactions, nor when interspecific
interactions were only stronger for C. nutans, as indicated by 50 to 100 year ratios close
to 0.5. When interspecific effects were stronger, C. nutans spread was generally slower in
the second 50 years (Figure 3-6b). Not surprisingly, C. nutans invaded the furthest when
the effect of C. acanthoides was smaller than intraspecific effects, regardless of whether
C. nutans interspecific effects were stronger or weaker than intraspecific effects. C.
75 acanthoides only successfully invaded when interspecific effects were weaker than intraspecific effects (Figure 3-6c, d).
Parameter sets that led to a large overlap width also lead to higher total thistle
densities in the area of overlap (Figure 3-7). This is likely because smaller areas of
overlap are observed as interspecific competition becomes greater for both species;
greater interspecific effects also lead to lower total densities of thistles when both species
are present. Not surprisingly, the largest area of overlap after 100 years occurs when both
species experience less intraspecific competition than interspecific competition, in which
case widespread coexistence is predicted. The smallest area of overlap occurs when
interspecific competition is greater than intraspecific competition, in which case the
overlap (i.e. apparent coexistence) in some fields is only maintained by immigration from
the source population. Note that this is the overlap observed after 100 years, which is not
necessarily stable.
Field model:
The predictions of the field model also generally agree with the Lotka-Volterra model, although the special case of identical interspecific effects produces slightly unexpected outcomes. C. acanthoides extinctions only occurred when the effect of C. nutans on C. acanthoides was quite large (Figure 3-8a); these extinctions occurred most quickly when the effect of C. nutans on C. acanthoides was largest, and when the effect of C. acanthoides on C. nutans was smaller (Figure 3-8b). The number of C. nutans
present in a field after 100 years was highest (and the number of C. acanthoides was
lowest) when interspecific effects of C. nutans on C. acanthoides were larger than
76 intraspecific effects (Figure 3-8c and d). An exception to this general trend is that when
interspecific effects were equivalent for both species, C. acanthoides population sizes
were high even at large values of C. nutans interspecific effects.
When examining the average numbers of individuals present across the landscape
width (Figures 3-9 and 3-10), it is apparent that the initial spatial pattern of the species
starting on opposite ends of the landscape mostly disappears. This is not surprising
initially, as watching individual simulations unfold reveals that the extensive dispersal
capabilities of both species quickly lead to some individuals appearing at the opposite
end of the landscape after only a few generations. The behavior at the edge is different,
with slight increase or decrease of species abundance at the edges. When the species are
equivalent, there is often widespread coexistence (Figure 3-9 a and b). Interestingly,
when interspecific effects are equivalent and stronger than intraspecific interactions, the
initial spatial pattern persists: there is more of a species on the half of the field where it
first invaded, and this pattern appears to be stable from 50 to 100 years (Figure 3-9c).
The situations examined are fairly consistent with the predictions of the analytical
(Lotka-Volterra) models, with one exception. In the case of identical competitors,
coexistence occurs even when interspecific effects are stronger than intraspecific effects
for both species. This result is extremely unstable and likely unimportant biologically:
when there are very slight differences in interspecific effects, this unexpected coexistence
no longer occurs. Widespread coexistence occurs if interspecific interactions are smaller
than intraspecific interactions (Figure 3-10a). Otherwise, C. acanthoides is driven extinct
in some cases (Figure 3-10b and c; Figure 3-11 a). The extinctions happened most
quickly when interspecific interactions are greater than intraspecific for C. nutans only
77 (Figure 3-11b), which is also where the largest numbers of C. nutans are present
(Figure 3-11c). The largest number of C. acanthoides was present when interspecific interactions where smaller than or equal to intraspecific interactions (Figure 3-11d).
Discussion
In the landscape model, after 100 years, some degree of overlap is predicted in all cases. The overlap, in terms of extent and location, may not change very rapidly under certain circumstances. In the landscape model, in all cases, there are still co-occurrences of both species, some of which are maintained by dispersal from the invasion fronts of both species. However, in some of these cases, if we simulated more than 100 years, one species would be lost from the landscape. Nevertheless, the fact that there were no cases where one species was lost in 100 years indicates that the time for one species to be competitively excluded is relatively long, even without considering spatially explicit competition, which is generally predicted to slow exclusion.
This modeling work is motivated by an observed pattern of spatial segregation between C. nutans and C. acanthoides, with an area of overlap with lower densities of both species. If the width of overlap is stable, this pattern would be consistent with interspecific interactions being stronger than intraspecific interactions for both species
(Figure 3-4c or Figure 3-5c). If the location of the overlap is stable, it is most consistent with Figure 3-4c, and implies that interspecific effects of these species are the same.
However, Figure 3-7 predicts that the same conditions (strong interspecific competition)
78 that lead to low densities of both species also predict a narrow area of overlap (less than 1 km), which is maintained by immigration. The observed overlap is certainly wider than this (greater than 20 km), and populations are likely not only maintained by dispersal. We feel that the distribution observed may not have reached equilibrium yet, given the time since introduction and the slow natural spread rates of these species, and thus we cannot infer much about the relative strengths of competition based on the distributions of these species. Resurveys of the distribution in a decade would prove very informative.
The results of competition experiments (Chapter 4) seem to indicate that there is little difference in interspecific versus intraspecific effects. This would indicate that the behavior observed in the system should resemble Figure 3-4b, with an overlap zone that slowly decays. Eventually, the landscape should contain both species everywhere, and the total thistle density should be uniform across the landscape. We do not predict that a stable overlap area exists based on the parameters that we observed. Again, resurveys of
the overlap distribution after some years should provide evidence to support or refute this
model prediction.
The results of the field model largely concurred with the predictions of the Lotka-
Volterra model. Thus, the incorporation of spatially-explicit competitive interactions does
not appear to affect the outcome. Pacala and Silander (1990) found that a non-spatial
version fit as well as a spatial model in a weed system; they attribute the lack of
importance of spatial processes to only weak aggregation from non-random dispersal and
plasticity in plant performance. Spatial interactions have, however, been shown to be
important in other cases (Cain et al. 1995, Pacala and Deutschman 1995). For example,
Neuhauser and Pacala (1999) found that a stochastic, spatially explicit version of the
79 Lotka-Volterra model led to a reduction in the region of parameter space leading to coexistence, and spatial segregation in some regions of parameter space predicted to lead to coexistence. It may be that for the Carduus thistle system, dispersal is too extensive relative to the scale of competitive interactions, for spatial structure to play a significant role.
Implications of model assumptions
The Lotka-Volterra competition models have been criticized as being overly
simplistic for a variety of reasons. Classic competition models do not incorporate scale-
dependent changes and may predict exclusion where species can coexist (Wiens 1989).
These models assume that there is no effect of age or genetic structure. They also assume
that there are no time lags, which can have important consequences for population
dynamics (Caswell 1972), and the mechanisms of competition are not specified (Tilman
1987).
Simulation models have been criticized as being difficult to analyze fully, making
it more difficult to generalize the results (Bolker and Pacala 1997). To overcome this,
moment closure techniques (Bolker and Pacala 1999, Ellner 2001) are a promising
avenue for analytically examining spatial competition; however, we chose to use
simulation models in this case because we were interested in particular initial conditions
(two species invading into the same area).
Both the landscape model and the field model assume environmental
homogeneity. Environmental heterogeneity can be an important coexistence promoting
factor if the species have different competitive responses in different environments
80 (Amarasekare 2003). In a heterogeneous competitive environment, relative non- linearities in species growths, spatial storage effect and growth-density covariance can generate coexistence (Amarasekare 2003). The within-field patterns observed in Chapter
5 suggest that both species aggregate in favorable areas of the fields; however, since these areas appear to be the same for both species, there is no evidence that each species performs better in different habitats. Thus including environmental heterogeneity would likely not affect the outcomes of competition.
We did not incorporate the effects of hybridization in these models. C. nutans and
C. acanthoides are known to form hybrids in both their native and invaded ranges (Moore and Mulligan 1956, Hegi 1987), but their hybrids are mostly sterile (Warwick et al.
1989). According to Case et al. (2005), producing sterile hybrids can lead to strong priority effects – whichever species is present in higher abundances initially will win.
While this may be true for some species, these species only have partial overlap in flowering phenology, thus hybridization may not play such a strong role.
Future avenues of research
Many cases of competition may be apparent (i.e. mediated through other species)
rather than direct (Connell 1990). C. nutans and C. acanthoides are attacked by similar
herbivores, which may lead to indirect interactions between them. The receptacle-feeding
weevil, Rhinocyllus conicus, released for biocontrol of C. nutans, is also an agent of C. acanthoides, although its phenology may not always be as well synchronized (Surles and
Kok 1978). R. conicus has been found to mediate interactions between two native thistles in Nebraska (Russell and Louda 2005). In central Pennsylvania, the phenology appears to
81 be variable from year to year: it is possible C. acanthoides escapes attack more in some years than in others. It would be useful to explicitly include apparent competition in
models of thistle interactions.
In both models, space was homogeneous and fully occupied by thistles. In reality,
however, most of space is occupied by other species. At the landscape scale, there are
many fields and roadsides that do not contain either species. At the field scale, even in
fields of co-occurrence of these species, many areas of the field do not contain thistles
either. At a landscape level, it is possible that no propagules of either species have
reached an area. Within a field that contains thistles, some habitat that is suitable for
thistle growth may not be occupied by thistles despite the presence of propagules. This
may be due to competition with other plant species. The establishment of both of these
species is sensitive to microsite characteristics (Chapter 7). Within fields of co-
occurrence, there are differences in the plant community in areas where thistles are found
versus where there are no thistles, although it is unknown whether the resident
community resists thistle establishment or whether the establishment of thistles changes
the community (Chapter 6). It would be useful to model the interactions of these two
Carduus thistles in the context of competition with other plant species, although pairwise
competition models may not be the best approach for this, as indirect effects can have a
large impact on outcomes (Tilman 1987).
Stochastic disturbance may play an important role in the population dynamics of
these species, thus even if exclusion would be predicted based on calculating competition
coefficients, this may never be observed in the field. It has been suggested that the
incorporation of stochasticity is of vital importance in models of invasions, due to their
82 inherently stochastic nature, where typically many invasions attempts will fail (Korniss and Caraco 2005).
Both models were quite simple, particularly the landscape model, as full parameters for more complex models are not yet available. It may be best to consider the landscape model as the interactions of two generic species, rather than truly representing the Carduus system. As more and more is known about the Carduus system, it will
become possible to parameterize more complex models, to both incorporate more
biological realism and relax many of the simplifying assumptions.
Conclusion
In conclusion, a wide variety of dynamics can potentially be observed when two
competing invasive species invade into the same area. The resulting dynamics will likely
depend at least in part on the strength of their effects on each other versus intraspecific
effects, as well as interactions with the resident community, including their predators and
competitors. Competitive exclusion was originally studied in microcosms of organisms
with short generation times (Gause 1934). In a field system where the generation time is
at least one year, it may be difficult to come to a definitive conclusion about whether or
not competitive exclusion occurs and whether or not the distribution that we observe is
stable. Models, in conjunction with experimental and observational studies, offer a way
to improve our understanding of such situations considerably.
83 Acknowledgements
This research is in collaboration with my advisor, Katriona Shea. This work was supported by USDA-CSREES (Biology of Weedy and Invasive Plants) NRI grant #2002-
35320-1228 to KS and a NASA Space Grant Fellowship to ER. Thanks to Olav Skarpaas,
Zeynep Sezen and Ingmar Rauschert for helpful suggestions. Thanks to John Burkardt for making the PROB library of functions publicly available.
References
Abrams, P. A., and W. G. Wilson. 2004. Coexistence of competitors in metacommunities
due to spatial variation in resource growth rates; does R* predict the outcome of
competition? Ecology Letters 7:929-940.
Allen, M., and K. Shea. 2006. Spatial segregation of congeneric invaders in central
Pennsylvania, USA. Biological Invasions 8:509-521.
Amarasekare, P. 2003. Competitive coexistence in spatially structured environments: a
synthesis. Ecology Letters 6:1109-1122.
Batra, S. W. T. 1978. Carduus thistle distribution and biological control in the
northeastern states.18-22.
Bolker, B., and S. W. Pacala. 1997. Using moment equations to understand stochastically
driven spatial pattern formation in ecological systems. Theoretical Population
Biology 52:179-197.
84 Bolker, B. M., and S. W. Pacala. 1999. Spatial moment equations for plant competition:
Understanding spatial strategies and the advantages of short dispersal. American
Naturalist 153:575-602.
Burkardt, J. 2005. PROB Probability Density Functions.
Cain, M. L., S. W. Pacala, J. A. Silander, and M. J. Fortin. 1995. Neighborhood Models
of Clonal Growth in the White Clover Trifolium-Repens. American Naturalist
145:888-917.
Case, T. J. 2000. An illustrated guide to theoretical ecology. Oxford University Press,
New York.
Case, T. J., R. D. Holt, M. A. McPeek, and T. H. Keitt. 2005. The community context of
species' borders: ecological and evolutionary perspectives. Oikos 108:28-46.
Caswell, H. 1972. Simulation Study of a Time Lag Population Model. Journal of
Theoretical Biology 34:419-&.
Chesson, P. 2000. Mechanisms of maintenance of species diversity. Annual Review of
Ecology and Systematics 31:343-+.
Chesson, P. L., and R. R. Warner. 1981. Environmental Variability Promotes
Coexistence in Lottery Competitive-Systems. American Naturalist 117:923-943.
Connell, J. H. 1990. Apparent versus Real Plant Competition. in J. B. Grace and D.
Tilman, editors. Perspectives on Plant Competition. Academic Press, New York.
Connolly, J., P. Wayne, and F. A. Bazzaz. 2001. Interspecific competition in plants: How
well do current methods answer fundamental questions? American Naturalist
157:107-125.
85 Coomes, D. A., M. Rees, and L. Turnbull. 1999. Identifying aggregation and association
in fully mapped spatial data. Ecology 80:554-565.
Crowell, K. L. 1968. Rates of Competitive Exclusion by Argentine Ant in Bermuda.
Ecology 49:551-555.
Desrochers, A. M., J. F. Bain, and S. I. Warwick. 1988. The biology of Canadian weeds
89. Carduus nutans L. and Carduus acanthoides L. Canadian Journal of Plant
Science 68:1053-1068.
Doing, H., E. F. Biddiscombe, and S. Knelhans. 1969. Ecology and distribution of the
Carduus nutans group (nodding thistles) in Australia. Vegetatio 17:313-351.
Ellner, S. P. 2001. Pair approximation for lattice models with multiple interaction scales.
Journal of Theoretical Biology 210:435-447.
Feldman, S. R., and J. P. Lewis. 1990. Output and Dispersal of Propagules of Carduus-
Acanthoides L. Weed Research 30:161-169.
Gause, G. F. 1934. The Struggle for Existence. Hafter Publishing Company, New York.
Hardin, G. 1960. Competitive Exclusion Principle. Science 131:1292-1297.
Hegi, G. 1987. Carduus Distel. in Ilustrierte Flora von Mitteleuropa. Paul Parey, Berlin.
Heinrich, B. 1976. Foraging Specializations of Individual Bumble-Bees. Ecological
Monographs 46:105-128.
Hulme, P. E. 1996. Herbivory, plant regeneration, and species coexistence. Journal of
Ecology 84:609-615.
Katul, G. G., A. Porporato, R. Nathan, M. Siqueira, M. B. Soons, D. Poggi, H. S. Horn,
and S. A. Levin. 2005. Mechanistic analytical models for long-distance seed
dispersal by wind. American Naturalist 166:368-381.
86 Kok, L. T. 2001. Classical biological control of nodding and plumeless thistles.
Biological Control 21:206-213.
Korniss, G., and T. Caraco. 2005. Spatial dynamics of invasion: the geometry of
introduced species. Journal of Theoretical Biology 233:137-150.
Lee, J. M., and J. L. Hamrick. 1983. Demography of two natural populations of musk
thistle (Carduus nutans). Journal of Ecology 71:923-936.
Levin, S. A. 1974. Dispersion and Population Interactions. American Naturalist 108:207-
228.
Levine, J. M., and M. Rees. 2002. Coexistence and relative abundance in annual plant
assemblages: The roles of competition and colonization. American Naturalist
160:452-467.
Levins, R., and D. Culver. 1971. Regional Coexistence of Species and Competition
between Rare Species (Mathematical Model/Habitable Patches). Proceedings of
the National Academy of Sciences of the United States of America 68:1246-&.
Lotka, A. J. 1924. Elements of Physical Biology. Dover, New York.
MacArthur, R., and R. Levins. 1967. Limiting Similarity Convergence and Divergence of
Coexisting Species. American Naturalist 101:377-&.
MacKinnon, K. 1978. Competition between Red and Grey Squirrels. Mammal Review
8:185-190.
Molofsky, J., J. D. Bever, and J. Antonovics. 2001. Coexistence under positive frequency
dependence. Proceedings of the Royal Society of London Series B-Biological
Sciences 268:273-277.
87 Moore, R. J., and G. A. Mulligan. 1956. Natural hybridization between Carduus
acanthoides and Carduus nutans in Ontario. Canadian Journal of Botany 34:71-
85.
Murrell, D. J., and R. Law. 2003. Heteromyopia and the spatial coexistence of similar
competitors. Ecology Letters 6:48-59.
Neuhauser, C., and S. W. Pacala. 1999. An explicitly spatial version of the Lotka-
Volterra model with interspecific competition. Annals of Applied Probability
9:1226-1259.
Pacala, S. 1997. Dynamics of Plant Communities. in M. Crawley, editor. Plant Ecology.
Blackwell Science, Malden.
Pacala, S. W. 1986. Neighborhood Models of Plant-Population Dynamics .2.
Multispecies Models of Annuals. Theoretical Population Biology 29:262-292.
Pacala, S. W., and D. H. Deutschman. 1995. Details that matter: The spatial distribution
of individual trees maintains forest ecosystem function. Oikos 74:357-365.
Pacala, S. W., and J. A. Silander. 1985. Neighborhood Models of Plant-Population
Dynamics .1. Single-Species Models of Annuals. American Naturalist 125:385-
411.
Pacala, S. W., and J. A. Silander. 1990. Field-Tests of Neighborhood Population-
Dynamic Models of 2 Annual Weed Species. Ecological Monographs 60:113-
134.
Popay, A. I., and A. Thompson. 1979. Some aspects of the biology of Carduus nutans in
New Zealand pastures. Proceedings of the 7th Asian-Pacific Weed Science
Society Conference:343-346.
88 Roxburgh, S. H., K. Shea, and J. B. Wilson. 2004. The intermediate disturbance
hypothesis: Patch dynamics and mechanisms of species coexistence. Ecology
85:359-371.
Russell, F. L., and S. M. Louda. 2005. Indirect interaction between two native thistles
mediated by an invasive exotic floral herbivore. Oecologia 146:373-384.
Seabloom, E. W., O. N. Bjørnstad, B. M. Bolker, and O. J. Reichman. 2005. Spatial
signature of environmental heterogeneity, dispersal, and competition in
successional grasslands. Ecological Monographs 75:199-214.
Shea, K., and P. Chesson. 2002. Community ecology theory as a framework for
biological invasions. Trends in Ecology & Evolution 17:170-176.
Shea, K., D. Kelly, A. W. Sheppard, and T. L. Woodburn. 2005. Context-dependent
biological control of an invasive thistle. Ecology 86:3174-3181.
Shea, K., S. H. Roxburgh, and E. S. R. Rauschert. 2004. Moving from pattern to process:
coexistence mechanisms under intermediate disturbance regimes. Ecol Letters
7:491-508.
Silvertown, J., S. Holtier, J. Johnson, and P. Dale. 1992. Cellular Automaton Models of
Interspecific Competition for Space - the Effect of Pattern on Process. Journal of
Ecology 80:527-534.
Skarpaas, O., and K. Shea. unpublished manuscript. Dispersal patterns, dispersal
mechanisms and invasion wave speeds for Carduus thistles.
Smyth, C. A., and J. L. Hamrick. 1987. Realized Gene Flow Via Pollen in Artificial
Populations of Musk Thistle, Carduus-Nutans L. Evolution 41:613-619.
89 Snyder, R. E., and P. Chesson. 2004. How the spatial scales of dispersal, competition,
and environmental heterogeneity interact to affect coexistence. American
Naturalist 164:633-650.
Surles, W. W., and L. T. Kok. 1978. Carduus thistle seed destruction by Rhinocyllus
conicus. Weed Science 26:264-269.
Tilman, D. 1982. Resource Competition and Community Structure. Princeton University
Press, Princeton.
Tilman, D. 1987. The Importance of the Mechanisms of Interspecific Competition.
American Naturalist 129:769-774.
Vila, M., and J. Weiner. 2004. Are invasive plant species better competitors than native
plant species? evidence from pair-wise experiments. Oikos 105:229-238.
Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo. 1997. Human
domination of Earth's ecosystems. Science 277:494-499.
Wang, Z. L., D. Y. Zhang, and G. Wang. 2005. Does spatial structure facilitate
coexistence of identical competitors? Ecological Modelling 181:17-23.
Warwick, S. I., J. F. Bain, R. Wheatcroft, and B. K. Thompson. 1989. Hybridization and
Introgression in Carduus-Nutans and Carduus- Acanthoides Reexamined.
Systematic Botany 14:476-494.
Warwick, S. I., B. K. Thompson, and L. D. Black. 1990. Comparative Growth-Response
in Carduus nutans, Carduus acanthoides, and Their F1 Hybrids. Canadian
Journal of Botany 68:1675-1679.
Weiner, J., and P. T. Conte. 1981. Dispersal and Neighborhood Effects in an Annual
Plant Competition Model. Ecological Modelling 13:131-147.
90 Wiens, J. A. 1989. Spatial Scaling in Ecology. Functional Ecology 3:385-397.
Wilson, W. G., S. P. Harrison, A. Hastings, and K. McCann. 1999. Exploring stable
pattern formation in models of tussock moth populations. Journal of Animal
Ecology 68:94-107.
91
Table 3-1: Landscape model parameters Symbol Parameter Estimate λ Population growth rate 1.2
d Diffusion rate to neighboring 0.125 fields
Intraspecific coefficients 0.01 α ′NN , α′AA (Given λ=1.2, this sets K for the patch at 167 individuals)
α′NA , α ′AN Interspecific coefficients Examined a range: 0.0001, 0.0002, 0.0005, 0.001, 0.002, 0.005, 0.01 N refers to C. nutans, A refers to C. acanthoides
92
Table 3-2: Field model parameters Symbol Parameter Estimate f Fecundity (max. number of 1000 seeds produced)
a C. nutans germination advantage 0.5 (no advantage)
g Proportion germinating 0.5
s Survivorship 0.8
Intraspecific competition 0.01 α ′NN , α′AA coefficients (affect fecundity)
α′NA , α ′AN Interspecific competition Examined a range: coefficients (affect fecundity) 0.0025, 0.005, 0.01, 0.025, 0.05, 0.1, 0.25
µ Scale parameter of WALD 2 model
λ Mean (location) parameter of 1 WALD model
93
a b
N2 N2
K1 K2
α12
K1 K2 α12
N N K2 K1 1 K1 K2 1
α 21 α 21
c d
N2 N2
K1 K2
α12 K1 K2 α12
N N K1 K2 1 K2 K1 1
α 21 α 21
Figure 3-1: Four scenarios of the Lotka-Volterra model The outcomes of the Lotka-Volterra model depend on the relative strengths of intraspecific and interspecific competition. In the classical formulation, this is expressed in terms of the relative sizes of K1/α12 and K2, and the relative sizes of K2/α21and K1. The lines shown are the zero growth isoclines for species 1 (blue solid) and species 2 (red dashed). Above its zero growth isocline, the population density of a species decreases, and below the line, population density increases. In Panel a, species 1 only will persist, regardless of initial conditions. Similarly, in Panel b, species 2 only will persist, and species 1 will go extinct. In Panel 3, stable coexistence occurs; any non-zero initial population sizes of species 1 and 2 will lead to coexistence at the population sizes where the two lines meet. In Panel d, the outcome depends on the initial conditions; there is a point of unstable coexistence where the two lines meet, which can be achieved if the species start at this equilibrium or in a particular relative ratio; otherwise one species drives the other extinct.
94
0.01 0.005 0.002 ) 0.001 AN ´ α
( 0.0005 0.0002 Effect A on N 0.0001 0.0001 0.0002 0.0005 0.001 0.002 0.005 0.01 Effect N on A (α´NA)
Symbol Meaning Lotka-Volterra model prediction
C. nutans interspecific effects > intraspecific effects, C. nutans persists only C. acanthoides interspecific effects ≤intraspecific C. acanthoides goes extinct effects
Interspecific effects ≤intraspecific effects for both Stable coexistence species
Unstable equilibrium, or one Interspecific effects >intraspecific effects for both species wins, depends on the species initial conditions
Simulations not performed
(mathematically equivalent to the lower triangle of cells.)
Figure 3-2: Competition coefficient ranges for the landscape mode The intraspecific effects were always equal to 0.001. This setup is similar for the field model, although a different range of parameters was used.
95 Overlap 0.01
N 0.005 n
o 0.002 A f 0.001 Not surprisingly, the largest area of overlap o t was seen when interspecific effects were c 0.0005 small for both species ffe 0.0002 E 0.0001 1 2 5 1 2 5 01 00 00 00 0. 000 000 000 0. 0. 0. 0. 0. 0. Effect of N on A
C. nutans distance 0.01
N 0.005 n
o 0.002 C. nutans invaded the furthest under when the A f 0.001 effect of C. acanthoides on C. nutans was o t
c 0.0005 low, and when the effect of C. nutans on C.
ffe 0.0002 acanthoides was higher. E 0.0001 1 2 5 1 2 5 01 00 00 00 0. 000 000 000 0. 0. 0. 0. 0. 0. Effect of N on A
C. acanthoides distance 0.01
N 0.005 n C. acanthoides only invaded to far distances o 0.002 when interspecific interactions were small A f 0.001 (smaller than intraspecific) for both species. o t Not surprisingly, it went the shortest c 0.0005 distances (usually retreated) when the effect ffe 0.0002 E of C. nutans on C. acanthoides was large but 0.0001 the effect of C. acanthoides on C. nutans was 1 2 5 1 2 5 small. 01 00 00 00 0. 000 000 000 0. 0. 0. 0. 0. 0. Effect of N on A
Figure 3-3: Landscape model: overlap and distances invaded More intense colors indicate higher values of overlap (a), C. nutans distance invaded (b), or C. acanthoides distance invaded (c). N refers to C. nutans, A refers to C. acanthoides
96
50 years 100 years 0 0 2 2 s s 1 1 e e tl tl s s i i 0 0 th th f a: Interspecific f 8 8 o o r effects < r e e b b 0 0 m intraspecific m u u N effects 1e-04 N 1e-04 04 04
010 30 50 010 30 50
0 0 2 2 s s 1 1 e e tl tl s s i i 0 0 th b: Interspecific th f f 8 8 o o r effects = r e e b b 0 intraspecific 0 m m u u N effects 0.001 N 0.001 04 04
010 30 50 010 30 50
0 2 1 120 es es l l t t s s i i h h t t 0 0
c: Interspecific 8 8 of effects > of er er b b 0 0 m intraspecific m 4 u u N effects 0.005 N 0.005 0 04
010 30 50 010 30 50
location location
Figure 3-4: Landscape model: symmetric cases (i.e. interspecific equivalence) C. nutans populations are shown in black, and C. acanthoides populations are in grey. The number is the parameter value used for the interspecific effect (α´NA= α´AN). As both interspecific coefficients increase simultaneously, the area of overlap, which is quite wide when interspecific effects are small (a), becomes narrower (b and c), and the total thistle density is lower in the area of overlap. The results after 50 years are shown in the left column, and the results after 100 years are in the right column. The total thistle density is higher in the overlap zone in a, constant in b, and greater in c
97
50 years 100 years 0 0 2 2 s s 1 1 e e tl tl s s i i 0 0 th th f f 8 8 o a: Interspecific o r r e e b effects < b 0 0 m m u intraspecific 2e-04 u 2e-04 N effects 1e-04 N 1e-04 04 04
010 30 50 010 30 50
0 0 2 2 s s 1 1 e e tl tl s s i i 0 0 th th f b: Interspecific f 8 8 o o r effects > r e e b b 0 0 m intraspecific m u 0.002 u 0.002 N effects for C. 0.001 N 0.001 04 nutans only 04
010 30 50 010 30 50
0 0 2 2 s s 1 1 e e tl tl s s i i 0 0 th th f f 8 8 o c: Interspecific o r r e e b effects > b 0 0 m m u intraspecific 0.01 u 0.01 N effects 0.005 N 0.005 04 04
010 30 50 010 30 50
location location
Figure 3-5: Landscape model: outcomes of 3 different Lotka-Volterra scenarios C. nutans populations are shown in black, and C. acanthoides populations are in grey. The upper number is the effect of C. nutans on C. acanthoides (α´NA) and the lower number is the effect of C. acanthoides on C. nutans (α´AN). When interspecific effects are weaker than intraspecific effects, coexistence is predicted, with a lower density of C. acanthoides (a). When only C. nutans has stronger interspecific effects, it is predicted to eventually win (b). When interspecific effects are greater than intraspecific effects for both species, overlap width is stable but the location shifts to the right (c). The results after 50 and 100 years are shown in the left and right columns, respectively.
98 a b 0 2 . 6 io 1 at r
0 0 . ce 1 n a st h 05 8 i . d 0 dt i
r w a 04 6 . 3 ye 0 lap
4 0 100 over : 0. 50
2 02 . 1 0 n.
C. 0
0 0. Inter<=Intra Inter>Intra Inter>IntraNOnly Inter<=Intra Inter>Intra Inter>IntraNOnly
c d d 8 d . e 0 e d a v vad 0 6 . . in in 1 0
ce ce n 4 . 8 a 0 an 0. st st i i d d 2
.
6 0 . e e 0 v v i i t t a l a 0 l . 0 4 re re
0.
. h 2 . t -0 ans 2 . an 0 4 . nut ac
-0
0 . C. C. 0
Inter<=Intra Inter>Intra Inter>IntraNOnly Inter<=Intra Inter>Intra Inter>IntraNOnly Lotka-Volterra state Lotka-Volterra state
Figure 3-6: Landscape model results with respect to Lotka-Volterra states The simulations were separated into the 3 Lotka-Volterra states based on relative competition coefficients, as described in Figure 1. The error bars shown are the standard errors. When intraspecific interactions were greater, a large overlap area was created (a). The smallest area of overlap occurred when interspecific interactions were stronger. The distance invaded by C. nutans in 50 versus 100 years is close to 0.5 when intraspecific interactions are stronger and when C. nutans has a greater interspecific than intraspecific effect (b). C. nutans invaded the smallest distance when interspecific interactions are stronger than intraspecific interactions (c), and C. acanthoides retreats when C. nutans has stronger intraspecific effects (d).
99 ea 200 p ar a l over n i y t i 150 e dens l t s i h t l a t 100
o L-V State
an t Inter>Intra e Inter<=Intra M Inter>IntraNOnly
0 1020304050
Overlap width
Figure 3-7: Landscape model: total thistle density relationship to the overlap width after 100 years Parameter sets that led to a large overlap width also had higher thistle densities in the overlap areas. The largest overlap, as well as the largest total density of thistles, is seen when interspecific interactions are smaller than intraspecific interactions.
100 a b C. acanthoides extinction Time both species persisted 0.25 0.25 N 0.1 N 0.1 n n o 0.05 o 0.05 A A f 0.025 f 0.025 o o t t c 0.01 c 0.01 ffe 0.005 ffe 0.005 E E 0.0025 0.0025 1 1 5 5 5 5 5 5 01 05 25 01 05 25 0. 0. 00 02 00 02 0. 0. 0. 0. 0. 0. 002 002 0. 0. 0. 0. 0. 0. Effect of N on A Effect of N on A
c d Number C. nutans Number C. acanthoides 0.25 0.25 N 0.1 N 0.1 n n o 0.05 o 0.05 A A f 0.025 f 0.025 o o t t c 0.01 c 0.01 ffe 0.005 ffe 0.005 E E 0.0025 0.0025 1 1 5 5 5 5 5 5 01 05 25 01 05 25 0. 0. 00 02 00 02 0. 0. 0. 0. 0. 0. 002 002 0. 0. 0. 0. 0. 0. Effect of N on A Effect of N on A
Figure 3-8: Field model: outcomes under different interspecific competition coefficients More intense colors indicate higher values. Not surprisingly, C. acanthoides extinction occurred when the effect of C. nutans on C. acanthoides was larger, but not equal to the effect of C. acanthoides on C. nutans (a). The persistence time of both species was generally 100 years (the maximum time studied) and was only smaller when the effect of C. nutans on C. acanthoides was larger, but not equal to the effect of C. acanthoides on C. nutans (b). The highest numbers of C. nutans (c) present after 100 years (and lowest C. acanthoides, d) was when the interspecific effects of C. nutans on C. acanthoides were greater than C. acanthoides effects and greater than intraspecific effects. Note that when interspecific effects were identical for both species, C. acanthoides persisted in high numbers even when interspecific effects were greater than intraspecific.
101
50 years 100 years 5 3 s s 0 e e 3 tl tl s a: s 25 i i th th f 0 Interspecific f 2 o o r effects < r e e 15 0 mb intraspecific mb 0.0025 0.0025 u u N N
effects 5 01 0
010 30 50 010 30 50
35 s s 0 e e 3 tl tl s b: s 25 i i th th f Interspecific f 0 2 o o r r effects = 5 e e 1 0 mb intraspecific mb 0.025 u u 0.025 N N
effects 5 01 0
010 30 50 010 30 50
0 0 5 5 es es l l t t s c: s i i h h t t
0 Interspecific 0 3 3 of effects > of ber ber m intraspecific m u u 0 0 N effects 0.25 N 0.25
01 01
010 30 50 010 30 50
location location
Figure 3-9: Field model: symmetric cases (i.e. interspecific equivalence) C. nutans populations are shown in black, and C. acanthoides populations are in grey. The numbers in the graphs are the interspecific effects (α´NA= α´AN). Both species co-occur at all locations along the width of the landscape. The results after 50 and 100 years are shown in the left and right columns respectively. As both interspecific coefficients increase simultaneously, the priority effect of starting on one end of the landscape becomes more pronounced. There is an interesting edge effect on all cases, due to reduced competition at edges, where the maximum number of neighbors is smaller
102 50 years 100 years s s 0 0 e e l l t t s s i i h a: Interspecific h t t 03 03 of effects < of er er b intraspecific b 02 02 m m 1 1 u effects u N 0.005 N 0.005 0.0025 0.0025 0 0 010 30 50 010 30 50
0 0 s s e e tl tl s b: Interspecific s i i 04 th th f f
effects > 05 o o r r intraspecific 03 e e b b m effects for C. m 02 u u 03
1 0.05 N N nutans only 0.05 1 0.025 0.025 0 0
010 30 50 010 30 50
0 5 es l t s i
c: Interspecific h t
0 3
effects > of er
intraspecific b Extinction m
effects u 0.1
0 N 0.05 01 010 30 50
location
Figure 3-10: Field model: Outcomes of 3 different Lotka-Volterra scenarios C. nutans and C. acanthoides populations are shown in black and red, respectively. The upper number is the effect of C. nutans on C. acanthoides (α´NA), and the lower number is the effect of C. acanthoides on C. nutans (α´AN). When interspecific effects are weaker than intraspecific effects and widespread coexistence occurs (a). When only C. nutans has stronger interspecific effects, in the 100 year plot C. acanthoides appears to be close to extinction (b). When interspecific effects are greater than intraspecific effects for both species, C. acanthoides is nearly extinct in 50 years, and has gone extinct by 100 years (c).
103 a b 0 14 on 8 ent i t 120 0. es tinc pr 100 6 . 0 ies 0 8 th ex i ec p w 4 0 6 0. ials r 0 t 4
both s 2 . e 0 0 op. r m 2 i P T 0 0 0. Inter<=Intra Inter>Intra Inter>IntraNOnly Inter<=Intra Inter>Intra Inter>IntraNOnly
c d 00 40 s 00 e 15 ns 00 oid a h 30 t t n nu
1000 00 aca C.
20
r e C. b
0 r 0 m 5 e b 1000 Nu
m Nu
0 0
Inter<=Intra Inter>Intra Inter>IntraNOnly Inter<=Intra Inter>Intra Inter>IntraNOnly
Figure 3-11: Field model results with respect to Lotka-Volterra states The error bars shown are the standard errors. As predicted by the Lotka-Volterra models, coexistence (no extinctions) always occurred when interspecific interactions were smaller than intraspecific interactions. In all other cases, there were extinctions (a). The average persistence time of both species was lowest when only C. nutans had a stronger interspecific effect (b). Not surprisingly, the most C. nutans were present when interspecific interactions were stronger than intraspecific interactions for C. nutans only, and the most C. acanthoides were present when interspecific interactions were small for both species (c).
.
Chapter 4
Competitive interactions between two invasive thistle species, Carduus nutans and Carduus acanthoides
Abstract
As the number of biological invasions increases, interactions between invasive
species and the impacts of these interactions on their distributions is becoming an
important area of research. Carduus nutans and C. acanthoides, two invasive Eurasian weeds, have a segregated distribution in central Pennsylvania; competition between these species is a possible mechanism for this distribution. We used a response surface experimental design to examine competitive interactions between two cohorts of C.
nutans and C. acanthoides throughout their life cycles. Contrary to expectation, we found limited evidence for density dependence and little evidence for the importance of direct competition between these species. Minimal response to high densities may confer an important advantage to these invasive plants.
Introduction
Invasive species can have profound impacts on communities and ecosystems in their invaded ranges (Mooney and Hobbs 2000). Competition among plants is thought to be particularly important as they often have very similar niche requirements; this may be one important way in which the effects of invasive species are expressed. Although many
105 studies have examined interactions between native and exotic species (e.g. Meekins and
McCarthy 1999, Byers 2000, Seabloom et al. 2003), few studies have focused on interactions between invasive species at the same trophic level (but see Laterra 1999).
The potential for invader-invader interactions was already recognized by Elton (1958), who noted that in Hawaii, one species of invasive ant (Solenopsis rufa) was subsequently
replaced by a second species of invasive ant (Pheidole megacephala), and predicted that
P. megacephala would eventually be replaced by the Argentine ant if it reached Hawaii.
Thus, if invaders have similar niches, they may compete and thus influence each other’s dynamics. Such interactions between invaders, which will only increase in number as the number of invaders increase, may have a strong influence on invasion outcomes.
Carduus nutans L. and C. acanthoides L., two very similar invasive species, have
a strikingly segregated distribution in central Pennsylvania (Allen and Shea 2006). They
occur in similar habitats, primarily in pastures and roadsides (Batra 1978). The causes of
this distribution are unknown, but it does not appear to be driven by different abiotic
conditions or management practices in areas occupied by C. nutans versus areas occupied by C. acanthoides. We were interested in exploring competition between the two species as a possible factor affecting this distribution, as well as elucidating the potential for
interactions between them that might alter predictions about invasion success made from
single-species studies.
Milbrath and Nechols (2004) suggest that the role of intraspecific thistle
competition has not been well studied; interspecific competition between these species is
also not well understood. A greenhouse study with a target-neighbor design had
previously shown that for C. nutans, germination and seedling biomass were negatively
106 affected by intraspecific and interspecific competition with C. acanthoides, with interspecific competition having a larger (negative) impact (E. Rauschert, unpublished data). Warwick et al. (1990), using a target-neighbor design for a greenhouse study of C.
nutans, C. acanthoides and their hybrids, found that C. nutans was a more aggressive
competitor than C. acanthoides. If such results were observed in a field setting, it would imply that C. nutans might eventually replace C. acanthoides.
An important step in examining the role of competition on the invasion outcome is to compare the strength of interspecific competition as compared to intraspecific competition, to see whether or not competition is strong enough to cause competitive exclusion in the field. It is important to examine competition experimentally, as strong competitive interactions may be hard to detect observationally because they lead to spatial segregation of competing species (Freckleton and Watkinson 2000b). We study the full life cycle, as competition at early life history stages can lead to segregation and lack of competition at later stages, which in turn can lead to misinterpretations about the role of competition (Kubota and Hara 1996).
How to measure competition experimentally has been the focus of considerable debate. Much of this has focused on the relative merits of additive designs, which keep the target species density constant and increase the density of competitors, versus substitutive/replacement series designs (de Wit 1960), which keep the overall density constant and vary the relative proportion of species. Substitutive experiments have been criticized because intra- and interspecific competition are confounded (Snaydon 1991,
Gibson et al. 1999, Jolliffe 2000) and this may affect the estimates of the strength of competition (Firbank and Watkinson 1985, Snaydon 1991). Additive designs have been
107 criticized because they confound relative composition and overall density (Harper 1977), do not address frequency dependence and are difficult to analyze (Freckleton and
Watkinson 2000a). Antonovics and Fowler (1985) used a hexagonal fan design, which focuses on nearest neighbor effects only and is more complicated to interpret. Response surface designs, which combine substitutive and additive designs, can avoid the limitations of just using additive or substitutive designs (Inouye 2001), although they may be time-consuming (Cousens 1991). These designs also better allow fitting of competition models (Inouye 2001). As these designs also include monocultures, it may be possible, by measuring resource levels, to link the outcome of competition to predictions of Tilman’s more mechanistic R* models (Tilman 1982). Connolly et al. (2001) recommend that a response surface design with sequential measurements be used, in order to avoid size bias problems.
We use three response surface design field experiments to investigate interactions between C. nutans and C. acanthoides at different densities and different life history
stages. The first experiment involved planting seeds in known locations, which we refer
to as the seed placement experiment. Because germination rates can be variable, a related
experiment involved transplanting seedlings to known locations, which we term the
seedling placement experiment. There was an upper limit to the densities that we could
investigate with these approaches due to limitations in marking individual seed/seedling
locations. We thus also performed a third experiment, which consisted of scattering
seeds, in order to examine higher densities, which we call the seed scattering experiment.
In all three experiments, we hypothesized that C. nutans would be a better competitor, at
least at the earliest stages, due to its larger seed size. We present the results of two years
108 of the seed and seedling placement experiments, and one year of the seed scattering experiment, although a second year of this is currently in progress.
Methods
Species description
C. nutans and C. acanthoides are herbaceous monocarpic perennials; they are sometimes annual, winter annual or perennial (Desrochers et al. 1988). Individuals of both species can produce thousands of seeds (McCarty 1982, Feldman and Lewis 1990)
Both species colonize by seed dispersal, and there is no record of vegetative reproduction
(Desrochers et al. 1988). C. nutans is generally taller than C. acanthoides when flowering (Desrochers et al. 1988). In Pennsylvania, C. nutans generally has fewer stems
and fewer flowerheads than C. acanthoides, and C. acanthoides flowerheads are smaller than C. nutans flowerheads (1.2-1.6 cm, versus 1.5-4.5 cm , Rhodes and Block 2000).
Germination has been found to be microsite dependent (Chapter 7, Panetta and
Wardle 1992, Feldman et al. 1994, Ruggiero 2004), with larger gaps generally having
higher germination, although desiccation can occur if gaps are too large (Wardle et al.
1995). It is difficult to tell the species apart at the seedling and rosette stages (McCarty
and Scifres 1969, Desrochers et al. 1988); however, after overwintering, species
differences are more pronounced. Both species have been reported as being allelopathic
(Woodward and Glenn 1983, Wardle et al. 1991a), although they do not appear to have
allelopathic effects on each other (E. Rauschert, unpublished data). We do not explicitly
attempt to separate allelopathy and resource competition, as this would require creating
109 very artificial situations (Inderjit and delMoral 1997). They are known to co-occur in some locations in their native and invaded ranges and occasionally form hybrids (Moore and Mulligan 1956, Hegi 1987).
Site description
All field experiments were performed at the Russell E. Larson Agricultural Center
at Rock Springs, which is located about 10 miles southwest of University Park,
Pennsylvania. This location in the Ridge and Valley Province is typical of a central
Pennsylvania landscape. The site is a former pasture, which has been left ungrazed for
more than a decade, with mostly weedy grasses and dicots present. The site is typical of a
central Pennsylvanian hayfield; the dominant species are mostly the grasses Elytrigia
repens, Arrhenatherum elatius, Dactylis glomerata and Phleum pretense, and the dicots
Plantago lanceolata, Taraxacum officinale, Trifolium repens, Trifolium pratense and
Gallium species. The soils in the field are located within the Hagerstown silt-loam series.
Since C. nutans is not present in the area where the experiments were conducted, care was taken to prevent seed dispersal of flowering plants by using pollen bags to contain seeds. Greenhouse studies were conducted in Buckhout Greenhouse in University Park,
Pennsylvania.
All seeds were collected from naturally occurring populations from within the areas known to have only one species present (near Carlisle, PA for C. nutans, and in
State College, PA for C. acanthoides). After dissecting flowerheads to remove seeds,
seeds were sifted using mesh screens to remove small flat seeds, which are known to be
mostly non-viable (Kelly et al. 1988).
110 Experimental design
All three experiments have a response surface design, where both relative composition and density are varied (Marshall and Jain 1969). Of six response surface designs reviewed by Inouye (2001), a group of substitution series at different densities performed best in terms of estimating competition coefficients and population growth parameters efficiently, and thus this is what we use for our experiments. All experiments had five relative composition treatments (C. nutans: C. acanthoides ratios of 100:0,
75:25, 50:50, 25:75, 0:100) at three different densities (high, medium and low) (see
Figures 4-1 and 4-2). Although some have criticized equating species on the basis of
numbers (Connolly 1988) these species are roughly equivalent in size, therefore replacing
one individual with an individual of another species affects only the composition, not the
overall density. A randomized block design was used with all experiments.
Seed placement experiment
The seed placement experiment involved planting seeds in known locations, in order to follow individuals and enable species identification throughout the life cycle. In
September 2003 and 2004, six blocks were planted with densities of 100 seeds/m2 (low
density), 400 seeds/m2 (medium density) and 2500 seeds/m2 (high density). Prior to the
start of the experiment, the plots were tilled as some level of disturbance is necessary for
germination (K. Shea, unpublished data). Although we did not continue to remove non-
thistle individuals after planting, strong differences persisted between plots and the areas
around plots (i.e. only a limited amount of other plants re-invaded the plots). Seeds were
planted at a depth of 5 mm, which enhances germination (McCarty and Scifres 1969).
111 Each plot consisted of 16 individuals, regularly spaced on a 4 x 4 grid, and the size of the plot was varied to create different densities. The spacing of the thistle seeds was 0.1, 0.05 and 0.02 m between individuals for the low, medium and high density plots respectively, with a boarder area of the same width as the spacing (i.e. plots were 0.5 x 0.5 m, 0.25 x
0.25 m and 0.1 x 0.1 m). In 2003, each plot planted was in the center of a 0.5 x 0.5 m square, which is the size required for the largest plot, and there was no extra space between squares due to field constraints. In 2004, this experiment was planted in a different portion of the field, to avoid using previously contaminated soil, and plots were created in the middle of 2 x 2 m squares, in a checkerboard fashion. Within each plot, given the number of individuals each species necessary for the desired relative plot composition, each location was randomly assigned to either C. nutans or C. acanthoides.
Seedling placement experiment
Because germination can be quite variable (E. Rauschert, unpublished data), we
initiated an experiment with seedlings at known locations, in order to study specific
proportions and densities at later life stages. In early September 2003 and 2004, seeds
were planted individually in pots in a greenhouse. They were allowed to develop under
favorable conditions until their true leaves had extended beyond cotyledons for most
individuals (approximately three weeks), at which point they were transplanted to the
field site. The field was mowed once prior to initiation of the experiment. Six blocks
were planned for both years, but due to poor germination of C. nutans in the greenhouse,
only five blocks were possible in 2003. As in the seed placement experiment, each plot
consisted of 16 individuals. The densities studied were 4, 25 and 100 thistles/m2, which
112 were created by varying plot sizes. The spacing of the thistle seedlings was 0.5, 0.2 and
0.1 m between individuals for the low, medium and high density plots respectively, with a boarder area of the same width as the spacing (i.e. plots were 2.5 x 2.5 m, 1 x 1 m and
0.5 x 0.5 m). Each plot planted was in the center of a 2.5 x 2.5 m square, which is the size required for the largest plot. Holes of the same size as the germination pots
(approximately 7 cm diameter) were drilled in the existing vegetation to allow transplantation of the thistle seedlings without undue disturbance of the root structure. To better quantify the strong block effect observed in 2003-2004, soil cores were taken to a depth of 30 cm at all high density plots in each block and analyzed for basic nutrient content in 2005.
Seed scattering experiment
In order to investigate higher densities, a related experiment was planted at the field site in the fall of 2004 and 2005. The experiment had a similar response surface design, but seeds were sown at densities too high to allow for individual marking. All plots were the same size (20 x 20 cm), and density was varied by changing the numbers of seeds sown. Seeds were sown at densities of 1000 (low density), 2000 (medium density) and 4000 (high density) seeds/m2, which, for the plot sizes used, required sowing
40, 80 and 160 seeds in a 20 x 20 cm area. The lowest density is similar to that which would be found near a parent plant (O. Skarpaas, pers. comm.), and we chose to also investigate higher densities, as this has been shown to improve parameter estimation
(Inouye 2001). A control treatment, with no sown seeds, allowed us to monitor background germination. This experimental setup was replicated six times in each year of
113 the experiment. Each plot was tilled prior to planting to remove all other vegetation.
Again, although we did not continue to remove non-thistle individuals after planting, only a limited number of individuals of other plant species re-invaded the plots. The seeds mixtures for each plot were prepared in the laboratory, and distributed randomly in the plots and covered lightly with soil. Results are only reported for the cohort planted in
2004, as the 2005 cohort has not yet flowered.
Censuses
The seed placement experiment was monitored for germination three times a week during the one-month germination period in September-October. All experiments were censused for presence/absence and size (longest leaf length and rosette diameter) in the fall and spring, and censused destructively in July when plants were flowering.
Measures of plant size are particularly important, as they have been linked to fecundity in
C. nutans (Lee and Hamrick 1983). Factors measured at the destructive census include: the number of stems, the number of flowerheads, height and root crown diameter for bolting plants. Longest leaf length and diameter were measured for those plants that did not flower. For the seed scattering experiment, we only present the post-vernalization results, as after this time we can be certain of species identity.
The seed and seedling placement experiments were terminated after the first summer, when all flowering individuals had been removed and the remaining plot compositions were no longer representative of the treatments. The seed scattering experiment had higher numbers of individuals planted; lower numbers flowered in the
first year, so plots were left intact and monitored for a second year.
114 Germination trial of seedling experiment
Maternal effects can influence the characteristics of seeds produced (Roach and
Wulff 1987). We performed a germination study to examine whether the quality of seeds produced was influenced by the maternal competitive environment. The first year of the seedling experiment was the only experiment where sufficient individuals flowered in each treatment for maternal effects to be analyzed. The different maternal environments we considered were the different competitive environments in the seedling placement experiment: high, medium and low overall thistle density, crossed with different levels of intraspecific vs. interspecific competition (monoculture of a species through to 75% occupied by the other species). We focused on plots in the first three blocks, as most flowering occurred there, thus we had up to three replicates for each plot type. Five plots
(out of 45) had insufficient flowering plants. If possible, replacement plots were selected from blocks 4 and 5 to represent the treatment combinations missing.
Two plants of each species were used from every plot containing that species, and fifteen seeds were removed from the most developed head. If the selected plant did not
yield seeds, a new plant was chosen from the same plot. Several treatments, especially
treatments with only four C. acanthoides planted, did not have enough flowering C.
acanthoides, which overall flowered less than C. nutans.
In February 2005, these seeds were planted in pots in randomized locations in the greenhouse and their subsequent germination, and size two weeks after germination, were recorded. This experiment was monitored for germinations until June 2005, when all germination had ceased (no new germinations had been observed in the previous two weeks).
115 Analysis
The average plot response for each species was the basic unit for analysis, to avoid pseudoreplication (Crawley 2002). Logistic regressions using a vector of the
number of successes to failures were examined for germination, flowering success and
survivorship (Crawley 2002). The average number of heads produced was analyzed using
quasipoisson regression, as the average of count data has a variance function proportional
to a Poisson distribution. Size data (longest leaf length) were first log transformed to
obtain normally distributed data and then analyzed using linear regression. For the seed
placement analysis, average days to germination were analyzed using quasipoisson
regression. All analyses were performed using R (R Development Core Team 2005).
The basic explanatory variables used were the combined plot density of both
species and the proportion of the plot that was planted with the other species. At a given
density, the total number of plants remains constant, as the proportion of one species
increases, the proportion of the other decreases. Densities were normalized to be between
0 and 1, to be in the same range as the “proportion other species” variable. Note that the
planned and actual densities and proportions often differed when plants failed to develop
to the next life history stage. Planting densities and proportions were used for analyses of
germination and early survival, and actual densities and proportions were used for later
life-history stages in experiments where the planting density and actual densities were
considerably different (i.e. experiments started from seeds). If a response variable was
found to be positively related to the proportion of the other species present, we took this
to mean that individuals had higher performance in plots with more heterospecifics vs.
conspecifics, which implies that the strength of intraspecific competition was greater than
116 interspecific competition. The only exception is for analyses of days to germination. A negative relationship with explanatory variables implies that the average days to germination was shorter, which implies faster germination.
Interactions between density and the proportion of the other species were only included if the coefficients were significant at the p<0.05 level. The significance of the terms in the models was determined through an analysis of variance of the fitted models.
Akaike’s Information Criterion (AIC) values were used to decide whether to include parental head diameter as a covariate in the germination trial results. For the seedling placement experiment, planned densities were always used, as survivorship was very high. The effects of blocking and year were incorporated by using mixed models with the block/year effect included as a random effect (glmmPQL in R).
Results
There was considerable variation in the response to plot composition and density in the numbers of seeds, seedlings and flowering plants in each experiment (Table 4-1).
The results of all experiments with respect to evidence for density and competition are shown in Table 4-2. Overall, there were not strong responses to competition in the response variables studied (i.e. plant size, survivorship, proportion flowering and number of flowers produced). Even when some effects were statistically significant, the magnitudes of the effects estimated were small. The evidence for density dependence with respect to both seeds and seedlings was limited and variable, as density was not always a significant predictor in the models, and the magnitude of the estimated effect
117 was occasionally quite small. The evidence for a different effect of interspecific competition versus intraspecific competition was also not strong. Generally, there were no significant interactions between density and the proportion of the other species; all exceptions are noted in Table 4-2.
In the seed placement experiment, there was conflicting evidence that C.
acanthoides was impacted by density and weak evidence than C. nutans was more strongly impacted by interspecific versus intraspecific competition. We saw weak evidence for a negative effect of density on C. acanthoides and for greater intraspecific competition in C. nutans in the seedling placement experiment. The germination trial
results suggested that interspecific effects may be greater than intraspecific for C.
acanthoides. While C. nutans appears to be negatively impacted by density, there was conflicting evidence for the effects of density on C. acanthoides. There appeared to be weak density-dependent effects in the seed scattering experiment, as well as potentially greater effects of intraspecific than interspecific competition on C. nutans. The results of each experiment are discussed in more detail below.
Seed placement experiment
In 2003, 52% of C. nutans and 41% of C. acanthoides germinated (720 seeds planted for both species), whereas in 2004 only 41% of C. nutans and 28% of C.
acanthoides germinated. There were strong differences in plant performance between
blocks. Analyses of proportion germinating (Table 4-3, Figure 4-3), average days to
germination (Figure 4-4), size in spring and number of heads produced did not
demonstrate strong impacts of the other species’ presence. The proportion of the other
118 species planted was only significant for the proportion of C. nutans germinating, where it
had a negative effect; however, this effect appears to be largely driven by the 75% C.
acanthoides category, which had much lower C. nutans germination. There was a
significant interaction of density by proportion of the other species in the model of C.
acanthoides proportion germinating; density had a negative effect in monoculture but a positive effect with more C. nutans present. Density had a significantly negative impact
on the days to C. acanthoides germination, which implies faster germination at higher densities. There were no significant relationships with density for C. nutans.
Seedling placement experiment
This experiment was the only experiment where a considerable number of
individuals flowered, but the two years of this experiment had very different flowering
rates. In 2003, 49% of C. nutans and 38% of C. acanthoides flowered. In 2004 13% of C.
nutans and 6% of C. acanthoides flowered.
In most of the models of the probability of flowering (Figure 4-5), size of plants,
and number of flowers produced (Figure 4-6), density and the proportion of the other
species present were not significant, and the magnitude of any significant effects
predicted are small (Table 4-4). The block effect was quite strong in this experiment; we
attempted to quantify the effect of block with soil nutrient analyses; however, these were
not well related to the thistle response. It appeared that nutrient availability was generally
higher in blocks where thistles did poorly (data not shown). The number of flowers
produced by C. nutans was positively affected by increased presence of C. acanthoides;
119 however, this effect is only marginally significant. The number of heads produced was negatively related to overall density for C. acanthoides but not for C. nutans.
In the germination trial of seeds produced by the 2003-2004 seedling placement
plants, over 75% of the seeds germinated. Analyses of the proportion of seeds
germinating (Table 4-5 , Figure 4-7), average days to germination and size at 14 days
were not significantly related to the composition of the parental plots. However, three
results were marginally significant with respect to the proportion other species experience
by maternal plants: a negative relationship with the proportion of C. acanthoides
germinating, a negative relationship of the size at two weeks of C. nutans, and a positive relationship of the average days to germination for C. nutans, which indicates slower
germination. All three of these suggest stronger interspecific effects than intraspecific
effects. The average diameter of the head from which the seeds were taken was found to
be positively correlated with the germination rates, and was always included in the
models. Overall the combined density of both species was significantly positively related
to the proportion germinating for C. nutans and significantly negatively related for C.
acanthoides; however, the magnitude of the effect estimated was quite small.
Seed scattering experiment
4,200 seeds of each species were planted each year. In fall 2004 there were 527 C.
nutans and 480 C. acanthoides seedlings present (Table 4-1). No germinations were observed in the control plots. Again, there appeared to be no strong effects of density- dependence or competition. For C. nutans, overall plot density of both species had a small but significant negative impact on germination and survival to July, whereas the
120 proportion of the other species planted had a marginally significant positive impact
(Table 4-6; Figure 4-8). For C. acanthoides, neither density nor the proportion of the
other species planted was found to be significantly related to germination and survival to
July.
The average rosette longest leaf length was not significantly related to the
proportion of other species planted, nor to the proportion of the other species present
(Table 4-6 ; Figure 4-9 and 4-10). Plots with higher actual density tended to have slightly
larger average longest leaf lengths.
In the analysis of whether or not the plants flowered, for C. nutans, the actual proportion of the other species had a marginally significant positive effect, whereas the actual density did not significantly impact flowering. For C. acanthoides, not enough
individuals flowered to perform this analysis. The block effect was not generally
significant, with the exception that block 6 had lower proportions of C. acanthoides
germination and survival, and blocks 2 and 4 had higher proportion of C. nutans
germination and survival. No interactions between density and the proportion of the other
species were significant.
Discussion
Although the importance of competition is often debated, plants are thought to be
subject to competition because most plants require the same resources, such as nutrients
(i.e. water and macronutrients such as nitrogen and phosphorus), light and germination space (Tilman 1997). Interestingly, however, many convincing examples of competitive
121 displacement are actually animals rather than plants (Argentine ant (Crowell 1968), Grey squirrel (MacKinnon 1978), mud snails (Byers 2000)). Hubbell (2005) also points out
that there are few documented examples of character displacement in plants, and that few
plant species have been proven to have disappeared due to competitive exclusion
Overall, we only found weak evidence for density-dependent effects, whether
interspecific or intraspecific. This is somewhat in contrast to the observations of Lee and
Hamrick (1983), who found that generally, rosette size was smaller and seed production
was lower in more dense C. nutans populations. The seedling placement experiment was
planted into established vegetation; thus presumably competition with the established
vegetation was also important and may explain why thistle density was not significant in
that experiment. Perhaps the overall lack of evidence for density dependence is not very
surprising, as weedy invasive species are commonly used as examples of species that
grow almost exponentially. It is possible that these species are more like fugitive species,
which are adapted to a certain amount of disturbance, and take advantage of good
conditions by growing as rapidly as possible. However, no population can continue to
grow indefinitely; at some point, some resources should become limited, and the negative
effects of density should become evident. It is noteworthy that even at very high
densities, as seen in the seed scattering experiment, we did not consistently find evidence
for density dependence.
In some instances, plant performance was even better in high density plots (i.e.
positive density-dependence). There may be some benefits associated with proximity to
other thistles. For example, both of these species are reported to be allelopathic
(Woodward and Glenn 1983, Wardle et al. 1991a), although the particular chemicals
122 involved have not been identified. It may be that both species secrete chemicals that enhance thistle growth but suppress the growth of other species. In a more natural setting, thistles of various ages and sizes are present, and there may be nurse plant effects, where larger individuals enhance the establishment of smaller seedlings, such as by preventing desiccation through shading. Additionally, both species may respond to underlying heterogeneities in a field, leading to higher densities and larger plants in favorable sites.
The amount of variation in plant performance was greater than we expected, and may have important implications for species distributions. Even if there is a strong mean effect of competition, given enough variation in response to competition, inferior competitors will still be able to persist for extended periods because they sometimes are able to succeed. Whether this is slow competitive exclusion or whether some mechanism such as the storage effect or relative non-linearities (Chesson 2000) operate in this system remains to be elucidated. This high variation also argues that competition between these species would be unlikely to lead to a stable segregated pattern, as the degree of inter versus intraspecific competition experienced by individuals can be very different and hence lead to different outcomes. Our experiments did not point to any particular explanation for this variation in success. Although the seedling placement experiment was strongly influenced by the effect of blocks, the seed scattering experiment blocks were more uniform, and there was still considerable variation in plant performance not attributable to our treatments. This may have been caused by a variety of small-scale heterogeneities: for example, Hamrick and Lee (1987), in a study of germination and early growth, found that the effects of intraspecific competition were masked by the effects of microtopography, litter cover, moisture and their interactions.
123 We found weak evidence that maternal effects can affect early performance of seeds in the germination trials of plants from the seedling placement experiment. C. acanthoides germination appeared to be slightly negatively impacted by the proportion of
C. nutans and overall density experienced by mother plants. It is unclear why C. nutans seeds from higher density environments appeared to germinate better than those from lower density environments. These maternal effects did not appear to persist for long, as we saw no significant effect of density on size at 14 days in a greenhouse setting. The maternal effects we observed may be more important under field conditions, as maternal effects in seedlings have been shown to persist longer under competition or otherwise stressful conditions (Stratton 1989).
When designing competition experiments, some have emphasized a need to vary both relative proportion of species, and the overall density due to the possibility of different responses at different densities (Law and Watkinson 1987, Gibson et al. 1999).
Interestingly, in the statistical analysis of our results, interactions between the proportion of the other species and density were not usually significant. The only exception was in the average number of days to germination in the seed placement experiment. There may be interactions between frequency and density when considering indirect ways in which the species may interact. For example, both species are attacked by the same specialist predator, Rhinocyllus conicus, which has a preference for C. nutans (Surles and Kok
1978). R. conicus individuals are influenced by the density of thistle patches (Z. Sezen, unpublished ms), and it is possible that a single C. acanthoides might be less attacked when surrounded by the preferred species C. nutans. Although in our field, we did not see
124 strong effects of R. conicus because the overall attack rate was very low, in areas with
higher R. conicus levels, such interactions could potentially be important.
Both of these species have been present in Pennsylvania since the mid-19th century (Rhodes and Klein 1993) and are presumably well-adapted to the conditions here.
Blossey and Notzold (1995) suggested that invasive species are more competitive in their native ranges (the Evolution of Increased Competitive Ability (EICA) hypothesis) but this was not well supported in a study comparing invasive weeds (including C. nutans) of
European and Australian origin in a common environment (Willis et al. 2000). A study of
intraspecific competition in garlic mustard, Alliaria petiolata, in Pennsylvania suggested that garlic mustard was less competitive in its invaded ranges (Bossdorf et al. 2004), in contrast to the EICA hypothesis. The authors propose instead the Evolutionary Reduced
Competitive Ability (ERCA) hypothesis, which states that, if there is reduced competition in invaded ranges and there is a cost to competitiveness, there is selection for species to become less competitive (Bossdorf et al. 2004). Thus it is possible that C.
nutans and C. acanthoides are less competitive than they might be in their native ranges.
C. nutans and C. acanthoides establishment has been shown to be microsite limited (Chapter 7, Panetta and Wardle 1992, Feldman et al. 1994, Ruggiero 2004). We did not address the role of other species in mediating interactions between the two thistle species. The seedling placement experiment was conducted in the presence of background vegetation; competition with the other species present may have played a large role in determining the subsequent growth of the thistles. The two experiments started with seed were initiated on bare ground, which should provide a “worst-case- scenario,” in the sense that competition between the two species is likely most strong
125 when no other species are present, and the thistles are present at high densities). Given the lack of evidence that we found for competition, it is unlikely that the two thistles compete strongly with each other in a pasture setting, where the immediate neighbors are
most often not thistles.
Sfenthourakis et al. (2006), in a recent meta-analysis of species co-occurrences,
found no evidence for widespread competition among congeneric species. They also
found no evidence for strong associations between species pairs. They suggest that
patterns observed are more a result of history or habitat preferences. Since there are no
clear differing habitat preferences for these two thistles (Allen and Shea 2006), this may suggest that historical invasion patterns are the cause of the segregated pattern observable today. We did not find evidence that interactions between Carduus nutans and C.
acanthoides would lead to competitive exclusion. Due to the lack of competition between
Carduus nutans and C. acanthoides documented in this study, we would expect the
segregation of these two species to degrade over time, unless other factors are involved in
maintaining this pattern.
In conclusion, we found no strong evidence for competition between C. nutans
and C. acanthoides and little evidence for density dependence. We feel that it is extremely important to report such results, as omitting them can bias estimates of the importance of competition from meta-analyses (e.g. Connell 1983, Schoener 1983,
Gurevitch et al. 1992). We were very intrigued that there was such poor evidence for competition between C. nutans and C. acanthoides. Even when the results are statistically
significant, often the magnitude of the estimated effect is small and thus potentially
biologically unimportant. Overall, it seems that for these species, intraspecific
126 competition may be slightly greater than interspecific competition, and that the response to density may be minimal at commonly observed densities. In fact, in the seed scattering experiment, size appeared to be larger in plots that had higher actual densities of plants.
These results seem to imply that density dependence is not particularly strong for these thistles. It is likely that in the field other mechanisms such as space, soil quality and resource availability and interactions with other species swamp any density-dependent effects, perhaps contributing to their success as invaders.
Acknowledgements
This work is in collaboration with my advisor, Katriona Shea. This research was partially supported by USDA grant 2002-35320-1228 to KS and a NASA Space Grant
Fellowship to Emily Rauschert. Tony Omeis and Roxanne Lease generously provided greenhouse space; Scott Smiles provided field space and Scott Harckom provided field equipment. Ottar Bjørnstad and Olav Skarpaas made helpful statistical suggestions, and
Dave Mortensen improved the design of the greenhouse study. Ingmar Rauschert assisted
with several figures. The fieldwork was carried out with the critical assistance of
Elizabeth Dlugosz, John Mellon, Jeffrey Butterbaugh, Adam Reese and Stephen Selego,
with significant contributions also from Zeynep Sezen, Ingmar Rauschert, Eelke
Jongejans, Olav Skarpaas, Pyush Agrawal, Emily Phillips, Melanie Northrup, Jessica
Peterson, Brian Jones, Andrew Jálics, Michael Claus, Diana Sitt, Tatiana Sitt and many
other people. Lidewij Keser, Simone Adeshina and Kerry Lynott assisted with the
greenhouse study.
127 References
Allen, M., and K. Shea. 2006. Spatial segregation of congeneric invaders in central
Pennsylvania, USA. Biological Invasions 8:509-521.
Antonovics, J., and N. L. Fowler. 1985. Analysis of Frequency and Density Effects on
Growth in Mixtures of Salvia Splendens and Linum grandiflorum Using
Hexagonal Fan Designs. Journal of Ecology 73:219-234.
Batra, S. W. T. 1978. Carduus thistle distribution and biological control in the
northeastern states.18-22.
Blossey, B., and R. Notzold. 1995. Evolution of Increased Competitive Ability in
Invasive Nonindigenous Plants - a Hypothesis. Journal of Ecology 83:887-889.
Bossdorf, O., D. Prati, H. Auge, and B. Schmid. 2004. Reduced Competitive Ability in an
Invasive Plant. Ecology Letters 7:346-353.
Byers, J. E. 2000. Competition between two estuarine snails: Implications for invasions
of exotic species. Ecology 81:1225-1239.
Chesson, P. 2000. Mechanisms of maintenance of species diversity. Annual Review of
Ecology and Systematics 31:343-+.
Connell, J. H. 1983. On the Prevalence and Relative Importance of Interspecific
Competition - Evidence from Field Experiments. American Naturalist 122:661-
696.
Connolly, J. 1988. Experimental Methods in Plant Competition Research in Crop Weed
Systems. Weed Research 28:431-436.
128 Connolly, J., P. Wayne, and F. A. Bazzaz. 2001. Interspecific competition in plants: How
well do current methods answer fundamental questions? American Naturalist
157:107-125.
Cousens, R. 1991. Aspects of the Design and Interpretation of Competition (Interference)
Experiments. Weed Technology 5:664-673.
Crawley, M. 2002. Statistical Computing: an Introduction to Data Analysis using S+.
John Wiley and Sons, New York.
Crowell, K. L. 1968. Rates of Competitive Exclusion by Argentine Ant in Bermuda.
Ecology 49:551-555.
de Wit, C. T. 1960. On competition. Verslagen van landbouwkundige onderzoekingen.
Desrochers, A. M., J. F. Bain, and S. I. Warwick. 1988. The biology of Canadian weeds
89. Carduus nutans L. and Carduus acanthoides L. Canadian Journal of Plant
Science 68:1053-1068.
Elton, C. S. 1958. The Ecology of Invasions by Animals and Plants. The University of
Chicago Press, Chicago.
Feldman, S. R., and J. P. Lewis. 1990. Output and Dispersal of Propagules of Carduus-
Acanthoides L. Weed Research 30:161-169.
Feldman, S. R., J. L. Vesprini, and J. P. Lewis. 1994. Survival and establishment of
Carduus acanthoides L. Weed Research 34:265-273.
Firbank, L. G., and A. R. Watkinson. 1985. On the Analysis of Competition within Two-
species Mixtures of Plants. Journal of Ecology 22:503-517.
129 Freckleton, R. P., and A. R. Watkinson. 2000a. Designs for greenhouse studies of
interactions between plants: an analytical perspective. Journal of Ecology 88:386-
391.
Freckleton, R. P., and A. R. Watkinson. 2000b. On detecting and measuring competition
in spatially structured plant communities. Ecology Letters 3:423-432.
Gibson, D. J., J. Connolly, D. C. Hartnett, and J. D. Weidenhamer. 1999. Designs for
greenhouse studies of interactions between plants. Journal of Ecology 87:1-16.
Gurevitch, J., L. L. Morrow, A. Wallace, and J. S. Walsh. 1992. A Metaanalysis of
Competition in Field Experiments. American Naturalist 140:539-572.
Hamrick, J. L., and J. M. Lee. 1987. Effect of Soil Surface-Topography and Litter Cover
on the Germination, Survival, and Growth of Musk Thistle (Carduus-Nutans).
American Journal of Botany 74:451-457.
Harper, J. L. 1977. Population Biology of Plants. Academic Press, New York.
Hegi, G. 1987. Carduus Distel. in Ilustrierte Flora von Mitteleuropa. Paul Parey, Berlin.
Hubbell, S. P. 2005. Neutral theory in community ecology and the hypothesis of
functional equivalence. Functional Ecology 19:166-172.
Inderjit, and R. delMoral. 1997. Is separating resource competition from allelopathy
realistic? Botanical Review 63:221-230.
Inouye, B. D. 2001. Response surface experimental designs for investigating interspecific
competition. Ecology 82:2696-2706.
Jolliffe, P. A. 2000. The replacement series. Journal of Ecology 88:371-385.
Kelly, D., H. Cameron, and J. Alex. 1988. Wind dispersal of nodding thistle seeds and
pappi. New Zealand Weed and Pest Control Conference Proceedings 41:207-209.
130 Kubota, Y., and T. Hara. 1996. Allometry and competition between saplings of Picea
jezoensis and Abies sachalinensis in a sub-boreal coniferous forest, northern
Japan. Annals of Botany 77:529-537.
Law, R., and A. R. Watkinson. 1987. Response-Surface Analysis of Two-species
Competition: An Experiment on Phleum Arenarium and Vulpia Fasciculata.
Journal of Ecology 75:871-886.
Lee, J. M., and J. L. Hamrick. 1983. Demography of two natural populations of musk
thistle (Carduus nutans). Journal of Ecology 71:923-936.
MacKinnon, K. 1978. Competition between Red and Grey Squirrels. Mammal Review
8:185-190.
Marshall, D. R., and S. K. Jain. 1969. Inference in Pure and Mixed Populations of Avena
Fatua and A. Barbata. Journal of Ecology 57:251-270.
McCarty, M. K. 1982. Musk Thistle (Carduus-Thoermeri) Seed Production. Weed
Science 30:441-445.
McCarty, M. K., and C. J. Scifres. 1969. Life cycle studies with musk thistle. University
of Nebraska, Nebraska.
Meekins, J. F., and B. C. McCarthy. 1999. Competitive ability of Alliaria petiolata
(garlic mustard, Brassicaceae), an invasive, nonindigenous forest herb.
International Journal of Plant Sciences 160:743-752.
Milbrath, L. R., and J. R. Nechols. 2004. Individual and combined effects of
Trichosirocalus horridus and Rhinocyllus conicus (Coleoptera : Curculionidae) on
musk thistle. Biological Control 30:418-429.
131 Mooney, H. A., and R. J. Hobbs. 2000. Invasive Species in a Changing World. Island
Press, Washington, D.C.
Moore, R. J., and G. A. Mulligan. 1956. Natural hybridization between Carduus
acanthoides and Carduus nutans in Ontario. Canadian Journal of Botany 34:71-
85.
Panetta, F. D., and D. A. Wardle. 1992. Gap Size and Regeneration in a New-Zealand
Dairy Pasture. Australian Journal of Ecology 17:169-175.
R Development Core Team. 2005. R: A language and environment for statistical
computing. in. Austria: R Foundation for Statistical Computing, Vienna.
Rhodes, A. F., and T. A. Block. 2000. The Plants of Pennsylvania: An Illustrated Manual.
University of Pennsylvania Press, Philadelphia.
Rhodes, A. F., and W. M. Klein. 1993. The vascular flora of Pennsylvania: Annotated
checklist and atlas. American Philosophical Society, Philadelphia.
Roach, D. A., and R. D. Wulff. 1987. Maternal Effects in Plants. Annual Review of
Ecology and Systematics 18:209-235.
Ruggiero, D. 2004. Factors Affecting the Germination of the Invasive Thistles: Carduus
nutans and C. acanthoides. Honors Thesis. The Pennsylvania State University,
University Park.
Schoener, T. W. 1983. Field Experiments on Interspecific Competition. American
Naturalist 122:240-285.
Seabloom, E. W., E. T. Borer, V. L. Boucher, R. S. Burton, K. L. Cottingham, L.
Goldwasser, W. K. Gram, B. E. Kendall, and F. Micheli. 2003. Competition, seed
132 limitation, disturbance, and reestablishment of California native annual forbs.
Ecological Applications 13:575-592.
Sfenthourakis, S., E. Tzanatos, and S. Giokas. 2006. Species co-occurrence: the case of
congeneric species and a causal approach to patterns of species association.
Global Ecology and Biogeography 15:39-49.
Snaydon, R. W. 1991. Replacement or Additive Designs for Competition Studies. Journal
of Applied Ecology 28:930-946.
Stratton, D. A. 1989. Competition Prolongs Expression of Maternal Effects in Seedlings
of Erigeron-Annuus (Asteraceae). American Journal of Botany 76:1646-1653.
Surles, W. W., and L. T. Kok. 1978. Carduus thistle seed destruction by Rhinocyllus
conicus. Weed Science 26:264-269.
Tilman, D. 1982. Resource Competition and Community Structure. Princeton University
Press, Princeton.
Tilman, D. 1997. Mechanisms of Plant Competition. in M. J. Crawley, editor. Plant
Ecology. Blackwell, Malden.
Wardle, D. A., M. Ahmed, and K. S. Nicholson. 1991. Allelopathic Influence of Nodding
Thistle (Carduus-Nutans L) Seeds on Germination and Radicle Growth of Pasture
Plants. New Zealand Journal of Agricultural Research 34:185-191.
Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1995. Influence of Pasture
Forage Species on Seedling Emergence, Growth and Development of Carduus
Nutans. Journal of Applied Ecology 32:225-233.
133 Warwick, S. I., B. K. Thompson, and L. D. Black. 1990. Comparative Growth-Response
in Carduus nutans, Carduus acanthoides, and Their F1 Hybrids. Canadian
Journal of Botany 68:1675-1679.
Willis, A. J., J. Memmott, and R. I. Forrester. 2000. Is there evidence for the post-
invasion evolution of increased size among invasive plant species? Ecology
Letters 3:275-283.
Woodward, M. D., and S. Glenn. 1983. Allelopathic Effects of Three Thistle Species.
Proceedings of the 37th Annual Meeting of the Northeastern Weed Science
Society:114.
134
Table 4-1: Summary of total response 2003-2004 2004-2005
Experiment Response C. nutans C. acanthoides C. nutans C. acanthoides variable Seed Seeds - - 4200 4200 scattering planted Seedlings - - 527 480 in fall Thistles in - - 281 233 June Thistles in - - 358 245 July Percentage - - 9.2% 1.63% flowering
Seedling Seedlings 581 619 720 720 placement planted Seedlings 542 608 692 688 surviving Percentage 48.9% 39.3% 12.9% 5.7% flowering
Seed Seeds 720 720 720 720 placement planted Seeds 376 297 293 205 germinating Percentage 78.5% 69.0% 56.7% 49.3% surviving Percentage 20.0% 13.7% 6.6% 4.0% flowering
In the seed scattering experiment, higher numbers of thistles present in July than in June indicates that some spring germinations did occur. For the seedling placement, in 2003-2004 there were only 5 blocks, and in one area, some incorrect species were planted; for this case we used the actual proportions that were planted in our analyses.
135
Table 4-2: Summary of results Experiment Variable C. nutans significantly C. acanthoides significantly impacted by: impacted by:
Density Proportion Density Proportion other sp. other sp. Seed Proportion No Yes, - Yes, - 1 No placement germinating Days to No No Yes, - No germination Summary: Conflicting evidence that C. acanthoides is impacted by density and weak evidence than C. nutans was more strongly impacted by interspecific competition Seedling Rosette Size No No No No placement in May Proportion No No No No Flowering Number of No Marginal, + Yes, - No Flowerheads Summary: Weak evidence for a negative effect of density on C. acanthoides and for greater intraspecific competition in C. nutans Germination Proportion Yes, + No Yes, - Marginal - trial germinating (seedling Days to Marginal, + No Yes + No placement) germination Seedling size No No Marginal, - Marginal, - Summary C. nutans appears to have responded positively to density, whereas C. acanthoides had a negative response to density, and interspecific effects may be greater than intraspecific for C. acanthoides. Seed Germination Yes, - Marginal, + No No scattering & survival Rosette Size No No Yes, - No Proportion No No - - Flowering Number of No No - - Flowerheads Summary: Weak evidence for negative effect of density, weak evidence for greater intraspecific competition than interspecific for C. nutans
Significance implies p≤0.05, and marginal significance implies p≤0.10. For the seed scattering experiment, all results are from July unless otherwise noted. Note that for average days to germination, a negative relationship implies faster germination. 1. There was a significant interaction in this model. Density had a negative effect in monoculture but a positive effect with more C. nutans present (see Table 3).
136
Table 4-3: Results from the seed placement experiment C. nutans C. acanthoides Response Explanatory Coefficient p-value Coefficient p-value variable variable Proportion Intercept 0.018 -0.40 germinating Proportion -0.53 0.05* -0.22 0.31 other species Combined -0.072 0.64 -0.80 0.01** density Interaction 1.34 0.05*
Average Intercept 2.64 2.69 Days to germination Proportion 0.167 0.18 -0.081 0.71 other species Combined -0.105 0.18 -0.18 0.01** density
• indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01
137
Table 4-4: Results from the seedling placement experiment C. nutans C. acanthoides Response Explanatory Coefficient p-value Coefficient p-value variable variable
Longest Intercept 2.08 2.39 leaf length in May Proportion 0.088 0.47 -0.089 0.35 other species Combined -0.045 0.59 -0.086 0.17 density
Proportion Intercept -0.94 -1.4 flowering Proportion -0.0075 0.99 -0.30 0.55 other species Combined -0.062 0.80 -0. 43 0.14 density
Number Intercept 1.11 2.9 of heads Proportion 0.38 0.06• -0.22 0.36 produced other species Combined -0.20 0.18 -0.37 0.02* density
Most results are not significant, and the magnitude of the coefficients estimated is somewhat small. • indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01
138
Table 4-5: Results from the germination trial C. nutans C. acanthoides Response Explanatory Coefficient p-value Coefficient p-value variable variable
Proportion Intercept 2.35 -0.20 Germinating Proportion -0.31 0.71 -1.4 0.10• other species Combined 1.04 0.03* -1.10 0.03* density Average Head -0.03 0.55 0.14 0.23 Diameter
Average Intercept 0.69 1.70 Days to Proportion 0.40 0.11 0.48 0.63 Germination other species Combined 0.24 0.09• 0.38 0.03* density Average Head 0.037 0.04* 0.0042 0.91 Diameter
Size at 14 Intercept 1.74 1.30 days Proportion -0.083 0.12 0.11 0.31 other species Combined 0.0077 0.77 -0.012 0.08• density Average Head -0.0033 0.43 -0.0032 0.84 Diameter
• indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01
139
Table 4-6: Results from the seed scattering experiment C. nutans C. acanthoides Response Explanatory Coefficient p-value Coefficient p-value variable variable Germination Intercept -2.09 -2.48 and survival Proportion 0.58 0.06• 0.052 0.88 to July other species planted Combined -0.61 0.02* -0.46 0.12 density planted
Longest leaf Intercept 1.66 1.81 length of Proportion 0.144 0.66 -0.26 0.32 rosettes (log other species transformed) planted Combined 0.55 0.11 0.702 0.05* density planted
Proportion Intercept -2.64 - Flowering Proportion 1.68 0.21 - other species planted Combined 0.053 0.84 - density planted
Number of Intercept 0.74 - heads Proportion -0.0060 0.99 - produced other species planted Combined -0.93 0.20 - density planted
Note that not enough C. acanthoides flowered to analyze proportion flowering or the number of heads produced. • indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01
140
Figure 4-1: Experimental design The experimental setup of the three competition experiments is shown on the top. The seed placement experiment consisted of placing 16 seeds (a), the seedling placement involved planting 16 seedlings (b), and the seed scattering experiment involved scattering seeds (c, the densities used were actually higher). Density was varied in the seed and seedling placement experiments by changing plot size (d), and in the seed scattering experiment by increasing the number of seeds (e). Plot composition was varied in the seed and seedling placement experiments by randomly allocating species to particular sites (f), constrained by the proportions of each plot, and in the seed scattering experiment, by scattering pre-counted seeds (g) of the desired proportion randomly
141
0 0 1 0 8 y t i s n de 0
s e d i o h t 06 an ac
C.
04 2 0
020406080100
C. nutans density
Figure 4-2: Response-surface design: seedling placement densities The response surface design chosen is basically a set of three replacement series, shown by the diagonal lines of points. Moving along a line, the overall density is constant, but the relative species composition changes. This pattern is the same for all competition experiments, but the densities are higher in the seed 2 placement and seed scattering experiments. Densities shown are seedlings/m .
142 g n i at 6 ng n i i 5 t 0. . m r na i ge 5 m
. r
0 40 . ge on 0 i
t 4 on i 0. t opor 3 . pr 3
. opor s 0 pr e
20 d . i 0 2 ns . o a 0 h t t 1 nu . an 1
. 0 ac C.
0 00 C. .
0. 0 100 400 2500 100 400 2500
Total density planted Total density planted
6 g . 7 n 0 . i 0 at ng n i i t 5 6 . m r na 0 0. i ge m
r
5 . 4 ge on 0 i
0. t or on i 4 t op 0. 3 or . pr 0
op 3 s . pr e 0
d 2 s i n 0. o 2 a h t 0. t 1 nu . an
1 . 0 0 ac C.
0 0 C.
0. 0. 00.250.50.75 00.250.50.75
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-3: Seed placement experiment: germination response to density and the proportion of the other species present C. nutans had no response to the density of seeds planted. C. acanthoides germination had a significant interaction between density and the proportion of the other species planted leading to higher germination at higher densities when more C. nutans is present. The proportion of C. nutans germinating significantly declined with the proportion of C. acanthoides planted; however, this appears to be mostly based on the 75% C. acanthoides planted group. The units of density are seeds/m2.
143
n on o i i t 0 a 2 at n n i i 0 m 2 m r ger ge
to 15 to
s s 15 y y da da
e g 10 a age 10 er er av av
. h 5 t ns 5 a t an nu ac
C. C. 0 0
100 400 2500 100 400 2500
Total density planted Total density planted
n on o i i t a at n 20 n i i m m r 20 ger ge
to to 15
s s y y 15 da da
e g 10 a age 10 er er av av
. h t ns 5 5 a t an nu ac
C. C. 0 0
00.250.50.75 00.250.50.75
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-4: Seed placement experiment: average days to germination response to density and the proportion of the other species planted For C. nutans, the average days to germination was not significantly related to density; however, for C. acanthoides, the average days to germinate was significantly negatively related to density. There were no significant relationships between the average days to germination and proportion of the other species planted. The units of density are seeds/m2
144 g n i g er 0 n i 4 3 r ow l 0. 0. f we
n o fl o i
t 3 or on . i t 0 0 op 2 or pr 0.
op s pr 2 e
d 0. i ns o a h t 10 t . 0 1 nu . an
0 ac C.
C.
0 00 . 0. 0 4 25 100 4 25 100
Total density planted Total density planted
g n i g er 25 . n 12 i 0 r ow 0. l f
we o fl on 20 i
t 0. on i t 08 opor 0. 15 pr
opor 0. pr
des s i 10 n . o 4 0 a h t 0 t 0. nu an
05 0. ac C.
C. 0
0 00 . 0. 0 00.250.50.75 00.250.50.75
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-5: Seedling placement experiment: flowering response to planting density and the proportion of the other species present There is no clear trend in the proportion of C. nutans flowering in relationship to density. Slightly fewer C. acanthoides flowered at high densities; however, this relationship is not significant. Few C. acanthoides flowered in plots that contained mostly C. nutans, but otherwise the proportion flowering was not well correlated with the proportion of other species. The lower two graphs are for year 2 only, because planting errors in year 1 lead to intermediate proportions of species planted. The units of density are seedlings/m2.
145 s s 6 ad 30 e ead h h er er ow ow l 25 l f f
f f o o
r r 45 e 20 e b b num num
15 e g age a er er 23 10 av av
. s h t n 1 5 a t an nu ac
0 0 C. C.
4 25 100 4 25 100
Total density planted Total density planted
s s ead ead 4 h h er er w 20 o ow l l f f
f f o o 3
r r e e 15 b b m nu num
e e g 10 ag a er er v a av
. s h 5 t n a t an nu ac
012 0 C. C.
00.250.50.75 00.250.50.75
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-6: Seedling placement experiment: response in the number of flowers produced to total planting density and the proportion of other species present For both species, the average number of flowers produced appears to be slightly lower at higher densities, but this relationship is only significant for C. acanthoides. For C. nutans, there is a significant increase in the number of flowers produced with increasing proportion of C. acanthoides. For C. acanthoides, there is no significant trend in the response to the proportion other species present. The lower two barplots are with year 2 data only, as planting error from year 1 led to intermediate proportions. For C. acanthoides there is no error bar for the 0.75 proportion of the other species category, as only one plot produced a flowering plant, thus the variance is undefined. The units of density are seedlings/m2.
146
g n i 2 . at 0 1 . ng n i i 1 t m r na i 0 . ge m
8 r
0. ge on i
81 t . 0 on i 6 t . opor 0 6 pr .
opor s pr e 4
d 0. i 40 . ns o 0 a h t t 2 . nu 2 an .
0 ac C.
00 0 C. .
0 0. 4 25 100 4 25 100
Total density planted Total density planted
2 ng . i t 2 1 . 1 na ng i i t m r .0 na 0 i e . 1 g 1 m
r
on ge 8 i . t
8 0 0. or on i t op pr .6 6 . 0 opor
0 s pr e
d 4 4 . i ns 0 o 0. a t h t nu 2 .2 . an
0 0 C. ac
0 0 . C.
0. 0 00.250.50.75 0.25 0.5 0.75 1
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-7: Germination trial: Germination response to the density and the proportion of the other species experienced by the mother plant The proportion of C. nutans germinating appears to be higher in plants whose mothers experienced higher densities, but the opposite trend occurs for C. acanthoides. Both of these results are significant. For C. nutans there was no response in seed germination to the proportion of the other species experienced by the mother plant. For C. acanthoides there is a marginally significant trend for decreased germination of seeds whose mothers experienced a higher proportion of C. nutans, which may be largely driven by high germination in the 25% C. nutans group. The units of density are seedlings/m2.
147 g g in in iv iv v v r r 5 u u s 1 s 10
.
0. d 0 d an an
8 0 ng ng i i t 0. t na 10 na i i 0. m m 06 r r e 0. e g g
. . p 04 o op . 0 pr 05 pr
. 0. s h t n 02 a t 0. an nu ac
00 00 C. C.
0. 0.
1000 2000 4000 1000 2000 4000
Total density planted Total density planted
g g in in iv iv v 0 v r r 2 u u s 0. s 10
d 0. d an an
g g 15 08 n . n i i 0. 0 at at n n i i 6 m m 0 r r 0 0. 1 ge ge
0.
04 op. op. r 0. pr p
. 05 h t ns 0. 02 a t 0. an nu ac
00 00 . . C. C.
0 0
00.250.50.75 00.250.50.75
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-8: Seed scattering experiment: the response of germination and survival to July to density and the proportion of the other species planted For both species, there is a decrease in the proportion germinating with increasing density; however, this relationship is only significant for C. nutans. The units of density are seeds/m2. The monoculture plots are on the left of the graph (with zero proportion other species planted). The error bars represent standard errors. The proportion germinating and surviving until July appears to be increase slightly with the proportion other species planted. This relationship is marginally significant for C. nutans. Note that the scales of the y-axes are different: the axis for C. nutans is twice as large
148
h th t 2 2 g g 1 1 n n le le
f f a 0 a 0 1 le 1 le
t t s e g 8 ges 8 n n o lo l
e 6 age ag r e er v av a
4
. h t ns a t an 246 nu ac
C. C.
0 02
1000 2000 4000 1000 2000 4000
Total density planted Total density planted
h h t t g g 2 n 2 n 1 1 le le
f a 0 eaf le 0 l 1
1
t t s s e e g g n 8 n lo lo
e g a age er er av av
. 46 s h t n a t an 2468 nu ac
C. C.
02 0
00.250.50.75 00.250.50.75
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-9: Seed scattering experiment: plot average longest leaf length relationship to density and the proportion of the other species planted For both species, there is a slight negative response to planting density; however, this relationship is not significant for either species (analyses were performed with log-transformed data). Neither species had a significant relationship to the proportion of the other species planted. The units of density are seeds/m2.
149 h th t g g n n le le 5
.
5 f f . a a le le
02 st . st 2 e e g g n n 02 lo . lo 5 .
2 e e g a ag 01 er . er p=0.12 1 av av
5
. . h t 5 ns . a t an nu ac 00
. 0 01 . C. C. p=0.03
1 g g Lo Lo
0 100 200 300 400 500 0 100 200 300 400 500
Total density present Total density present
h h t t g g n n le le 5
.
5 f . a eaf le l
02 st . st 2 e e g g n n 02 lo . lo 5 .
2 e g a age 01 er er . p=0.66 1 av av
5
. . s h t 5 n . a t an nu ac 00
. 0 01 . C. C. p=0.32
1 g g Lo Lo
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Proportion C. acanthoides planted Proportion C. nutans planted
Figure 4-10: Seed scattering experiment: average longest leaf length response to actual densities and actual proportions of the other species present Both species have slightly higher average longest leaf lengths with higher plot density. This relationship is significant for C. acanthoides but not for C. nutans. The units of density are seeds/m2. The relationship between average longest leaf length and the proportion of the other species planted is weak for both species, and not significant. The lines shown are the fitted models, and the p-values are for the slope.
Chapter 5
Spatial coexistence patterns of two invasive thistle species, Carduus nutans and C. acanthoides
Abstract
We examine coexistence patterns of two invasive species, Carduus nutans and C.
acanthoides at two spatial scales. A roadside survey of 5 x 5 km blocks in a previously identified overlap zone of these two species provided information at the regional scale. At a smaller scale, we surveyed four fields of natural co-occurrence of the two species, quantifying the spatial patterns by randomly placed 1 x 1 m quadrats, and by detailing spatial position within the quadrats. The patterns observed are strikingly different at the
different scales. At the regional scale, there is positive autocorrelation in both species but
negative cross-correlation between them, consistent with previous surveys. However, at
the field scale, although there is positive autocorrelation in both species, we generally see
positive cross-correlation between the two species. These results suggest that at smaller
scales, the two species are aggregated in good habitat and that there is no evidence for
strong competition between them.
151 Introduction
Two important related themes in ecology are development and maintenance of species patterns in space and time and the consequences of spatiotemporal pattern for population and ecosystem dynamics (Levin 1992). Studies of negative associations in the distribution of ecologically similar species dates back at least to Diamond’s (1975) study of birds on islands off New Guinea, where he found a “checkerboard” pattern in species
associations, in which ecologically similar species did not coexist on islands. Diamond
argued that competition led certain pairs of species to not co-occur. In a study of many
assemblages of stream fishes, Winston (1995) found that morphologically similar species
were less likely to co-occur, and that competition was the most likely explanation for this
pattern. Wilson (1988) studied plant communities in New Zealand, and found an excess
of both negative associations and positive associations compared to those expected under
null models. The excess of negative associations would suggest competition, although in
their study it is not possible to exclude different habitat preferences as a partial
explanation.
Spatial heterogeneities in population abundances can arise from environmental and community factors and are not simply random effects (Legendre 1993). Although larger spatial patterns, such as range limitations, are generally thought to be driven by climatic variables, biotic interactions can also be important. In animals, it is well known that competition through behavioral patterns, such as aggression in ants, can lead to the exclusion of species from certain areas (Cole 1983). Competition can also play a large role in plant interactions; for example, competition has been shown to limit species
152 ranges in two Ulex species (Bullock et al. 2000). Because they are sessile, plants cannot
leave environments where they experience high levels of competition. It is their
interactions with their immediate neighbors that are of most importance, thus the spatial
distribution of neighbors can influence the degree of competition actually experienced
(Pacala 1986).
Patterns of spatial covariances can be important in joining data with ecological
processes (Bjørnstad and Falck 2001). Competition, dispersal and environmental
heterogeneities can have significant effects on the spatial patterns observed in plants.
Direct (resource) competition between species should lead to negative cross-correlations
at small distances but possibly positive cross-correlations at greater distances, indicative of spatial segregation (Dale 1999). For example, in a study of the development of spatial structure in an experiment in California grasslands, competition was found to cause spatial segregation between grass species (Seabloom et al. 2005). Within species, limited dispersal should cause species to have an aggregated (autocorrelated) pattern.
Environmental heterogeneities can cause aggregation within a species and either aggregation or segregation between species, depending on whether the species have the same or different habitat preferences. Indirect interactions between species, such as apparent competition, can operate at a different spatial scale than direct interactions, and
can influence patterns in different directions. The resulting spatial structure can be
studied by examining autocorrelations or cross correlations at different spatial distances,
which can be described by structure functions such as correlograms or variograms
(Legendre 1993). The net effect of all of these processes on the spatial structure can be
153 predicted by using plant population dynamic models and compared to spatial structures found in empirical studies (Seabloom et al. 2005).
Spatial scaling is fundamental to ecology, and different patterns may be observed
at different scales (Wiens 1989). It is important to study correlation at several scales, as a
lack of autocorrelation at one scale does not imply that there is no autocorrelation at other
scales (Legendre and Fortin 1989). For example, in a study of two Ulex spp, Bullock et al. (2000) found that apparent co-occurrences at larger scales disappeared at finer scales.
The patterns that can be observed are limited by the scale at which they are studied;
inappropriate studies of scale can lead to misleading conclusions about the system (Wiens
1989, Freckleton and Watkinson 2002). A common approach to scale issues is to use
several scales; however, these are often chosen arbitrarily (and anthropocentrically)
(Wiens 1989). Kunin (1998) has proposed scale-area curves, where presence-absence
data of the area occupied by a species and the scale of resolution are plotted on a log-log
scale, as a robust way to transition between scales.
Allen and Shea (2006) documented a striking distributional pattern consisting of
strong spatial segregation on a large-scale study (50 x 100 km) in central Pennsylvania
(see Figure 5-1). Carduus nutans L. and C. acanthoides L. (Asteraceae) are two thistle species from Europe and Asia, which have spread to become invasive species in North and South America, South Africa, Australia and New Zealand (Julien and Griffiths
1999). Both species are present in all directions away from Allen and Shea’s (2006) study area; thus this is not an obvious case of range limits, nor do any environmental correlates
particularly explain this pattern. Spatial patterns can be caused by many different
mechanisms (Pielou 1961). Allen and Shea (2006) propose several hypotheses to explain
154 this distributional pattern, including competition between the two species; however, many of the mechanisms which could create such a pattern are not observable at the larger scale of their study.
In order to further investigate the distributional patterns of Carduus nutans and C.
acanthoides, we monitored the region of overlap identified in 2002 by Allen and Shea
(2006) in 2004 and 2005. We also documented smaller scale spatial patterns within four fields of natural co-occurrence from 2003-2005. We analyze in detail the co-occurrence
patterns of these species at these two levels of spatial resolution, and specifically ask the
question whether or not the patterns of co-occurrence are consistent with the effects of
direct competition between these two species.
Methods
Species description
Carduus nutans and C. acanthoides are both monocarpic perennials of Eurasian
origin (Desrochers et al. 1988). They are difficult to distinguish morphologically at the
rosette stages (Desrochers et al. 1988). C. acanthoides is often confused with C. nutans because they are so similar; however, pubescence on the underside of the leaf and leaf shape may be used to separate the species (McCarty et al. 1969). Both species have similar sized rosettes with leaves that are up to 30 cm long (Desrochers et al. 1988).
Vernalization is required for both species to bolt, after which their basal leaves decay. C.
nutans grows to be approximately 2 m tall; C. acanthoides is generally slightly shorter, approximately 1.5 m tall (Rhodes and Block 2000) or taller (E. Rauschert, pers. obs.).
155 Both species produce conspicuous purple flowerheads that are easily visible. When flowering, the species differences are most pronounced: C. nutans produces fewer, larger
flowerheads and fewer stems, whereas C. acanthoides produces many small flowerheads and more stems.
In North America, C. nutans was first recorded near Harrisburg, PA in 1853, and
C. acanthoides was first recorded in New Jersey in 1879 (Desrochers et al. 1988). Both
occupy overgrazed pastures and disturbed roadsides, sometimes occurring in mixed
stands (Batra 1978). These species colonize new areas by seed dispersal; there is no
record of vegetative reproduction (Desrochers et al. 1998). Individuals of both species
can produce many thousands of seeds (McCarty 1982, Feldman and Lewis 1990). They
are wind-dispersed, although Kelly et al. (1988) conclude that wind dispersal is not
always effective as seeds easily detach from pappi. Studies of dispersal through seed
tracking suggest that the two species disperse to similar distances, with estimates of mean
dispersal ranging from 0.99-3.43 m for C. nutans and 1.32-4.22 m for C. acanthoides
(Skarpaas and Shea, in revision). According to Medd and Smith (1978), C. nutans is also dispersed by water, birds and farm animals, and vehicles can disperse seeds; contaminated agricultural seed is an important mechanism of dispersal; C. acanthoides
can presumably also be dispersed by similar agents. Contaminated hay bales are a
potentially important mechanism of localized, human-mediated dispersal for both species
(D. Mortensen, pers. comm.).
C. nutans and C. acanthoides are known to hybridize in their native ranges (Hegi
1987) and in Canada (Moore and Mulligan 1956, Warwick et al. 1989). Because C.
nutans and C. acanthoides have different numbers of chromosomes, most hybrids are
156 sterile. It is currently unknown whether or not hybrids form in Pennsylvania. Certainly most of the individuals we observed were unlikely to be hybrids due to incomplete phenological overlap of these two species. Since the native ranges of these species are overlapping, it is possible that these phenological differences have evolved to minimize pollen competition. Both species can self-fertilize: Warwick and Thompson (1989) found significant departures from random outcrossing which they attribute to some level of self- fertilization.
Skinner et al. (2000) rank C. nutans and C. acanthoides as the second and
fifteenth most commonly listed noxious weeds in the US. Chemical, mechanical and
biological control have been attempted with some success (Kok and Surles 1975,
Desrochers et al. 1988, Kok and Mays 1991a, Sindel 1991, Kok 2001). The receptacle
feeding weevil, Rhinocyllus conicus was introduced for the control of C. nutans (Kok and
Surles 1975) and is also considered to be a biocontrol agent for C. acanthoides (Kok and
Mays 1991a). R. conicus adults lay eggs on the bracts of thistle flower heads; the larvae feed on the receptacle tissue, which leads to a loss in seed production (Kok 2001). C.
nutans is more heavily infested by R. conicus and has a higher seed reduction than C.
acanthoides (Surles and Kok 1978). R. conicus attack can be identified in the field by the presence of egg cases on the bracts of thistles.
Survey methods
Surveys were conducted at two spatial scales, as a multi-scale approach better allows an understanding of both the patterns and processes involved in invasions
(Pauchard et al. 2003, Pauchard and Shea 2006). The spatial scale of the regional and
157 field surveys are summarized in Figure 5-2. Extent and grain are specified, as both are important aspects of scale (Wiens 1989).
Regional survey methods
In order to monitor the stability of the regional distribution, the area of overlap was monitored in 2004 and 2005 using similar methods to Allen and Shea’s 2003 survey
(Allen and Shea 2006); we summarize the approach here. The survey area is a 100 x 50 km area divided into blocks of 5 by 5 km. Surveillance was done by driving along pre- chosen roads at a constant speed for 20 km per block in 2002, and 10 km per block in
2003. In 2004 and 2005, we surveyed blocks in the core 2002 areas of overlap. We surveyed 43 blocks in 2004 and 46 blocks in 2005. When a species was found, the location (using a Garmin ETrex Legend GPS unit), elevation, population size, abundance, spatial extent, road type, slope location (i.e. before, on or beyond the slope at the road edge), environment, slope and aspect were recorded. In contrast to Allen and Shea
(2006), there was no “stopping rule”- all 10km of each block were surveyed regardless of whether a species had previously been recorded in a given block. Moreover, unlike Allen and Shea (2006), we also recorded populations that were sighted within 2 km of the last population. These differences allow us to have better indicators of the densities of C.
nutans and C. acanthoides populations in a block. We also recorded the surrounding vegetation in 2005. Surveys were conducted when both species were flowering, as inflorescences can be seen from at least 100 m away (Allen and Shea 2006). In 2005, we also revisited all co-occurrence sites (14) that had been found in 2002-2004, to examine the short-term persistence of these populations.
158 Description of the fields
Three of the sites identified as co-occurrences by Allen and Shea in 2002 were suitable for within-field surveying because they were accessible and had more than ten individuals of both species still present in 2003. We located a fourth site by driving in the area of overlap in 2003 because we wanted to include an additional pasture. Both pastures chosen were permanent pastures, as rotations to tilling and cropping may break
the cycle of biennials and perennials and obscure the co-occurrence patterns. All of these
sites of overlap were located in Perry and Cumberland Counties in Pennsylvania. One site
was an abandoned industrial site (Site I), two sites were permanent pastures (Sites P1 and
P2) and one site was a long roadside over a forested ridge (Site R); all are common habitat types for these species (Batra 1978).
The P1 site (coordinates 40.379 N, 77.306W) consisted mostly of C. acanthoides,
with a few C. nutans individuals within the field. The area surveyed was about 80 m long
and 30 m wide. This field was used for occasional cattle grazing, despite the high density
of thistles.
Site P2 (coordinates 40.225 N, 77.431 W) was the only site not identified in the
2002 Allen and Shea survey. In this pasture, two large patches of thistles were chosen: an
80 x 25 m (main) section near a temporary stream and a 40 x 45 m (middle) section in the center of the pasture.
Site I (40.183 N, 77.238W) was an abandoned industrial site. This field contained the highest densities of C. nutans we saw in Pennsylvania – parts of the field were like
impenetrable forests of thistles. We surveyed a 40 x 45 m portion of the field, which
contained both species.
159 Site R (coordinates 40.301 N, 77.400 W) was located along a major road through
Colonel Denning State Park. This site was a linear habitat: there was dense forest and thistles were only found immediately adjacent to the road. The ridge ran east-west, and the northern slope had virtually no thistles on it: no C. nutans or C. acanthoides were
observed on this slope. The top and southern sides of the ridge were surveyed. The lower
boundary was about one kilometer down the ridge; no thistles were seen for more than a
mile after this.
2004 and 2005 Field Surveys
For the actual surveys, a nested sampling design was chosen, as sampling at several scales can give more information about spatial pattern, and is particularly recommended when nothing is known in advance about the spatial structure (Fortin et al.
1989). Fields were sampled using 1 x 1 m quadrats covering at least 10% of the area of
the field. In order to avoid problems of possible periodicity in the pattern (Krebs 1989),
quadrat locations were chosen randomly each year. While re-randomizing between years
does not allow us to track the fate of individuals, it does allow more independent
estimates of the spatial pattern each year. Within a quadrat, the spatial location and
species identity of each thistle was noted at a 5 cm resolution, giving spatial information
at a yet finer scale. For flowering plants, the height of each individual and numbers of
stems and flower heads were quantified. The presence of R. conicus attack on flowering plants was determined by looking for egg cases at up to 10 of the most developed flower heads. For rosettes, the longest leaf length was recorded. We gathered information about the other vegetation in the quadrats, which is discussed in Chapter 6. In 2004 and 2005,
160 there were so few C. nutans individuals within the survey plots in P1 that it was decided to collect an extra dataset by centering quadrats on each C. nutans adult and surveying as usual; this extra dataset is referred to as P1X.
Analyses
All analyses were performed in R (R Core Development Team 2005), and ArcGIS
(ESRI 1999-2001) was used to visualize the larger scale data. Correlograms were used to
analyze the spatial patterns at the regional and coarse-field scales. Correlograms are plots
of correlation at various distance classes, which can be used for both autocorrelation and
cross-correlations. Where the correlogram line first crosses the y=0 line (i.e. the x-
intercept) indicates to what spatial extent spatial extent correlations remain positive (or
negative). Often at distances greater than 2/3 the maximum distance, the uncertainty in
the correlations increases and the estimate degenerates into random noise. Thus we are
primarily interested in whether the correlations at the smallest distances are positive or
negative (i.e. whether the y-intercept is positive or negative). We term “positive
autocorrelation” when autocorrelation at the smallest distance classes is positive, and
“negative autocorrelation” when the autocorrelation at the smallest distances is negative,
and similarly for cross-correlations.
The significance of the correlation coefficients is calculated by permutation, and
in our case we are interested in a two-sided test – significantly positive or significantly
negative coefficients. A correlogram is considered globally significant if at least one
correlation coefficient is significant at the level α´= α/υ (Bonferroni corrected level),
where υ is equal to the number of distance classes, and we consider the α = 0.05 level
161 (Legendre and Fortin 1989). All correlograms were calculated using the “correlog” function in the “ncf” package in R (Bjørnstad 2005), which uses Moran’s I for the correlation coefficients.
At the regional scale, we examined autocorrelation in block densities of populations (the number of populations of a species seen along a 10 km stretch of road in a 5 x 5 km block) for both species, and cross-correlograms between the two species, using a binning increment of 10 km. We only included blocks within the area of co- occurrence (32 blocks) in our analyses. We used a square root transformation of the density indicators to stabilize the variance in the data (Sokal and Rohlf 1995). We also examined autocorrelation and cross-correlation in block presence or absence of a species, in order to compare our results to those of Allen and Shea (2006).
At the field scale, we examined correlograms and cross-correlograms of the square-root transformed densities. We examined autocorrelations in rosettes, flowering plants and all individuals for both species. For both auto- and cross-correlations, analyses of the rosettes, flowering plants and all individuals were similar, so we only present the results for all individuals. We also examined cross-correlations with a species between rosettes and flowering plants, which was generally similar to the autocorrelation results.
P1 and P1X datasets were merged for this analysis. For each field, the distance classes were created using a binning increment of 2 m. For Site R, we also present additional correlograms with a binning increment of 50 m, in order to look at the pattern at the scale of the site (approximately 1200 m long). The two large patches (Main and Middle) in P2 were analyzed separately.
162 Results
At the regional scale, we continue to see evidence of spatial segregation (negative
cross-correlation) between the two species; however, at a smaller scale, we see
aggregation (positive cross-correlation) of the two species within areas of coexistence.
Generally the magnitude of the C. acanthoides autocorrelation was larger than the magnitude of the C. nutans autocorrelation, which may be due to the fact that in every field, there were much higher abundances of C. acanthoides.
Regional Survey
The regional scale results are summarized in Table 5-1 and are shown in Figure 5-
3. The regional scale results were consistent with the findings of Allen and Shea (2006).
The presence-absence auto-correlograms (Figure 5-5) were only globally significant for
C. nutans in 2005, where positive autocorrelation was observed. The other three auto- correlograms suggest positive autocorrelation but are not globally significant. The density auto-correlograms (Figure 5-4) show the same pattern. Only the 2005 C. acanthoides
density correlogram is not globally significant.
Both the presence-absence and density cross-correlograms show negative cross-
correlation at shorter distances and positive cross-correlations at longer distances. The
spatial extent of negative correlations is smaller than that found by Allen and Shea (2006)
for the full dataset, although sub-setting their data to only include the blocks that we
resurveyed leads to correlations similar to those we see (negative cross-correlation up to
15-20 km), and so may just be a function of the smaller extent of the later surveys.
163 Of the 14 previously identified co-occurrence sites revisited in 2005, 3 had no flowering thistles, possibly due to management changes. Two sites had only C.
acanthoides present, and one site had only C. nutans present. All other sites still had at least 10 individuals of both species.
Within-Field Surveys
Table 5-2 presents a summary of the numbers of flowering plants and rosettes of
each species found in the four fields of co-occurrence in 2004 and 2005. There was
considerable variation in the plot mean thistle densities, and the standard deviation is
generally larger than the mean. There were differences among fields and years: P2
generally had more thistles of both species (both flowering plants and rosettes) in 2005.
Sites P1, I and R generally had more rosettes of both species in 2004 than in 2005; these fields had more bolting plants of both species in 2005 than in 2004, except Site R, which had more C. nutans flowering in 2005 than in 2004.
The general pattern for Sites P1, P2 and I (Figure 5-6, 5-7, 5-8, and 5-9) was
significant positive cross-correlations between species and also positive autocorrelation
within species. This pattern was generally consistent regardless of whether just rosettes,
just flowering plants, or all individuals of a species were considered. An exception to
these trends is the P2 Main Patch in 2004, where C. nutans had non-significant negative autocorrelation, probably due to the small sample size (only 16 individuals). All correlograms presented are globally significant, again except for the autocorrelation in C.
nutans for Site P2 Main Patch 2004. The spatial scale of autocorrelation was generally
similar, ranging between 10-20 m, with considerable variation seen. Cross-correlations
164 between flowering plants and rosettes within a species were generally positive and were very similar to the autocorrelation results.
Site R had a very different pattern. It is important to point out that there were almost no plots of actual co-occurrence in Site R. The top of the ridge tended to have C. nutans plants only, and the lower portion of the ridge site tended to have C. acanthoides
plants only. The correlograms of the smaller distances (Figure 5-10) showed negative
cross-correlation, but the coefficients were not significant. The autocorrelations were still
positive. The larger distance correlograms of Site R (using a binning increment of 50 m,
Figure 5-11) showed positive autocorrelation for both species, but the cross-correlation
was negative in both years.
Discussion
At the regional scale, we found positive autocorrelation in both species (although
the C. acanthoides pattern was not as strong), and we found negative cross-correlation between the two species. The regional scale pattern appears to be consistent over the four years it was studied (2002-2005). These findings are consistent with those reported in
Allen and Shea (2006), and may suggest negative interactions between the two species, although abiotic factors are generally considered to drive distributions (particularly range limits) at larger scales (Levin 1989, Wiens 1989). However, at finer scales, where the biological mechanisms underlying negative biotic interactions (such as competition) operate, we find positive cross-correlations between the two species, which does not support the idea that there is strong competition between these species, leading to
165 exclusion. In fact, these results are the opposite of what others have found (e.g. Bullock et al. 2000, Purves and Law 2002): instead of apparent co-occurrences disappearing at finer scales, we see apparent segregation disappear at finer scales. These results highlight
the importance of studying distributions at multiple scales and at scales relevant to the
mechanisms under study.
Preliminary results analyzing co-occurrences at the within-quadrat scale also
indicate positive associations between these species (results not shown). These positive
auto- and cross-correlations observed at the within quadrat scale are consistent with the
field scale results. This may indicate that there are domains of scales, where patterns are
relatively constant over a particular range of scales and change abruptly after this (Wiens
1989). It would be helpful to understand where the edges of domains occur and why.
The fact that we see both positive autocorrelation and positive cross-correlation
(at similar scales) within a field is most consistent with the hypothesis that both species
aggregate in good habitats. It may be that positive correlation in microsite suitability for
germination induces the positive cross-correlation between these species; this effect is
strong enough to outweigh any potential negative effects of competition. The beneficial
effects of being in good sites can be greater than the effects of competition with
neighbors, particularly when patch quality varies in time (Fowler 1988). Although this
study was observational, making it generally more difficult to assign causality, the results
of the pattern are consistent with the experiments presented in Chapter 4, which indicate
that competition between these species does not appear to be strong. Although limited
dispersal can cause aggregation within a species, it cannot be responsible for positive
cross-correlation between species. In Chapter 6, where we discuss the vegetation in the
166 fields of co-occurrence, we find that there is a difference in the vegetation community between thistle plots and non-thistle plots, but that there are not strong differences between the plots occupied by C. nutans and C. acanthoides. We also find significant
spatial structure in the vegetation community, which further suggests that the fields are
not homogeneous.
A potentially valuable extension to this study would be to try to quantify the
heterogeneities (such as soil texture or nutrients) in these fields. Once the heterogeneities
to which the thistles are responding are known, it would be possible to account for them
in the analysis, and see how the remaining pattern is distributed. In a study of aggregated
congeneric fungi, Komonen (2005) found that once spatial clumping of suitable habitat
was accounted for, the species distribution appeared to be random. A better
understanding of the smaller scale environmental variation could be used try to make
predictions about where we might find thistles at larger scales. At larger scales,
Collingham et al. (2000) found that the same environmental variables that were important
at a smaller scale (2 km) were also important at a larger scale (10 km). These authors
claim that it is possible to scale up within the same spatial extent, and it is possible to
focus down from larger extents to smaller extents at the same resolution. Allen and Shea
(2006) found no differences in habitat preferences of these two species at the regional
scale; it appears that they also prefer the same habitats at the field scale.
An important exception to the general pattern of positive cross-correlation at
smaller scales was found in Site R. It represents more of a transition between the regional
scale and the field scale, since it is quite long (approximately 1 km). The two species
were not well mixed at this site, because the upper portion surveyed only had C. nutans,
167 and the lower section just C. acanthoides. It is likely that these two species were introduced into this site when contaminated soil was brought in for the road edge; neither species is present for over a mile in either direction of the road, nor are they able to grow off the road in the forested areas. It is possible that soil from different areas, contaminated with different species was used in sections of the road, leading to areas which only contain one species or the other, and that it is difficult in such poor habitat for the species to spread beyond the area of original introduction.
Seabloom et al. (2005), in a grassland experiment started from spatial randomness, found that different processes (dispersal, competition, environmental heterogeneities) took different amounts of time to cause spatial structure, and this spatial structure had different extents. Environmental heterogeneity caused aggregation in just one year at smaller spatial scales; dispersal and competition led to increasingly negative cross-correlation between species, which was not very apparent after one year, but became more apparent in subsequent years at greater than 4 m distances. Spatial patterns may sometimes develop quite rapidly: Stoll and Bergius (2005) found that competition led to regular spacing of plants within a five week experiment. In the areas that we studied, there should have been significant time for competition between the two species to leave a spatial signal: the two pasture fields that we studied were permanent pastures,
and Site I appeared to have been managed similarly for at least the last five years.
At least 30 sampling units should be used to detect spatial autocorrelation, and
more than 100 sampling locations may be necessary to reliably estimate spatial structure
(Fortin and Dale 2005). While we had close to 200 units for each of the field surveys, the
number of regional survey units (blocks) is rather low, which may account for the lack of
168 global significance in some correlograms, as well as their rough appearance. Perhaps future efforts should divide this area of overlap into units slightly smaller than 5 x 5 km, or again increase the spatial extent of the survey, in an effort to clarify this relationship.
Random sampling, such as we used in the field surveys, may not be the most efficient way to survey these species, given their aggregated distribution. Rew et al.
(2006), using models of seven different survey methods, suggest that a targeted transect design provides the information most efficiently.
As both of these species are of management concern, their movement and range expansion are of particular interest. Collingham et al. (2000) found that scale influences conclusions reached about whether or not a species had expanded and filled its potential range; larger (coarser) spatial scales were more likely to indicate that range expansion had ceased. Although these species occur in across Pennsylvania, they are not found in all potentially suitable habitats (E. Rauschert, pers obs). Their projected invasion wave speeds, calculated using data collected in Pennsylvania, are very low: for C. acanthoides
it is approximately 3 m per year, and for C. nutans is approximately 10 m per year
(Skarpaas and Shea , unpublished manuscript). It is obvious that they have not moved
across the state in less than 200 years via wind dispersal: presumably humans have
moved them to areas far from the original introductions, but they may still be slowly
filling in areas in between. Thus, it is possible that their current segregated distribution is
a historical artifact, which will decay over time.
C. nutans and C. acanthoides had markedly different spatial relationships at the two scales at which we studied them: segregation at a larger scale but aggregation at a
smaller scale. This is counter to the pattern that has been seen in other species. It is
169 possible that such relationships are more likely to be seen for invasive species as compared to native species. At larger scales, invader presence may be primarily due to human activities, and may be due to their human-mediated invasion history. At finer scales, their interactions with each other, with resident species and local habitat heterogeneity, as well as local disturbance regimes, may lead to quite different patterns of association. Given the slow natural spread of many species, it may take a very long time for natural small scale processes to influence large-scale distributions.
Acknowledgements
This work is in collaboration with my advisor Katriona Shea. This research was partially supported by USDA-CSREES (Biology of Weedy and Invasive Plants) NRI grant #2002-35320-1228 to KS and a NASA Space Grant Fellowship to ER. Thanks to
Ottar Bjørnstad and Matt Ferrari for statistical assistance. Thanks to Zeynep Sezen and
Olav Skarpaas for support throughout the project. Eelke Jongejans, Zeynep Sezen and
Olav Skarpaas provided useful comments on the manuscript. Thanks to John Mellon, Jeff
Butterbaugh, Elizabeth Dlugosz, Adam Reese, Stephen Selego, Gabby Hryschyn and Ane
Schjolden for field assistance. Thanks to the Frey Family, the Shirk family, Eric Revene for use of their fields.
170 References
Allen, M., and K. Shea. 2006. Spatial segregation of congeneric invaders in central
Pennsylvania, USA. Biological Invasions 8:509-521.
Batra, S. W. T. 1978. Carduus thistle distribution and biological control in the
northeastern states.18-22.
Bjørnstad, O. N. 2005. NCF package. http://onb.ent.psu.edu/.
Bjørnstad, O. N., and W. Falck. 2001. Nonparametric spatial covariance functions:
Estimation and testing. Environmental and Ecological Statistics 8:53-70.
Bullock, J. M., R. J. Edwards, P. D. Carey, and R. J. Rose. 2000. Geographical separation
of two Ulex species at three spatial scales: does competition limit species' ranges?
Ecography 23:257-271.
Cole, B. J. 1983. Assembly of Mangrove Ant Communities - Patterns of Geographical-
Distribution. Journal of Animal Ecology 52:339-347.
Collingham, Y. C., R. A. Wadsworth, B. Huntley, and P. E. Hulme. 2000. Predicting the
spatial distribution of non-indigenous riparian weeds: issues of spatial scale and
extent. Journal of Applied Ecology 37:13-27.
Dale, M. R. T. 1999. Spatial pattern analysis in plant ecology. Cambridge University
Press, Cambridge.
Desrochers, A. M., J. F. Bain, and S. I. Warwick. 1988. The biology of Canadian weeds
89. Carduus nutans L. and Carduus acanthoides L. Canadian Journal of Plant
Science 68:1053-1068.
171 Diamond, J. M. 1975. Assembly of Species Communities. Pages 342-444 in M. L. Cody
and J. M. Diamond, editors. Ecology and Evolution of Communities. Belknap
Press of Harvard University Press, Cambridge, MA.
ESRI. 1999-2001. ArcGIS Desktop, Version 8.3.
Feldman, S. R., and J. P. Lewis. 1990. Output and Dispersal of Propagules of Carduus-
Acanthoides L. Weed Research 30:161-169.
Fortin, M. J., and M. R. T. Dale. 2005. Spatial analysis: a guide for ecologists.
Cambridge Univeristy Press, New York.
Fortin, M. J., P. Drapeau, and P. Legendre. 1989. Spatial Auto-Correlation and Sampling
Design in Plant Ecology. Vegetatio 83:209-222.
Fowler, N. L. 1988. What Is a Safe Site - Neighbor, Litter, Germination Date, and Patch
Effects. Ecology 69:947-961.
Freckleton, R. P., and A. R. Watkinson. 2002. Large-scale spatial dynamics of plants:
metapopulations, regional ensembles and patchy populations. Journal of Ecology
90:419-434.
Hegi, G. 1987. Carduus Distel. in Ilustrierte Flora von Mitteleuropa. Paul Parey, Berlin.
Julien, M. H., and M. W. Griffiths. 1999. Biological Control of Weeds: A world
catalogue of agents and their target weeds (4th ed.).
Kok, L. T. 2001. Classical biological control of nodding and plumeless thistles.
Biological Control 21:206-213.
Kok, L. T., and W. T. Mays. 1991. Successful Biological Control of Plumeless Thistle,
Carduus acanthoides L. [Campanulatae: Asteraceae (= Compositae) by
172 Trichosirocalus horridus (Panzer) (Coleoptera:Curculionidae) in Virginia.
Biological Control 1:197-202.
Kok, L. T., and W. W. Surles. 1975. Successful biocontrol of musk thistles by an
introduced weevil, Rhinocyllus conicus. Environmental Entomology 4:1025-1027.
Komonen, A. 2005. Local spatial pattern in the occurrence of two congeneric wood-
decaying fungi in an old-growth boreal forest. Scandinavian Journal of Forest
Research 20:393-399.
Krebs, C. J. 1989. Ecological methodology. Harper & Row, New York.
Kunin, W. E. 1998. Extrapolating species abundance across spatial scales. Science
281:1513-1515.
Legendre, P. 1993. Spatial Autocorrelation - Trouble or New Paradigm. Ecology
74:1659-1673.
Legendre, P., and M. J. Fortin. 1989. Spatial Pattern and Ecological Analysis. Vegetatio
80:107-138.
Levin, S. A. 1989. Challenges in the development of a theory of ecosystem structure and
function. in J. Roughgarden, R. M. May, and S. A. Levin, editors. Perspectives in
Ecological Theory. Princeton Univeristy Press, Princeton, N.J.
Levin, S. A. 1992. The Problem of Pattern and Scale in Ecology. Ecology 73:1943-1967.
McCarty, M. K. 1982. Musk Thistle (Carduus-Thoermeri) Seed Production. Weed
Science 30:441-445.
McCarty, M. K., C. J. Scifres, A. L. Smith, and G. L. Horst. 1969. Germination and Early
Seedling Development of Musk and Plumeless Thistles. University of Nebraska,
Nebraska.
173 Medd, R. W., and R. C. G. Smith. 1978. Prediction of Potential Distribution of Carduus-
Nutans (Nodding Thistle) in Australia. Journal of Applied Ecology 15:603-612.
Moore, R. J., and G. A. Mulligan. 1956. Natural hybridization between Carduus
acanthoides and Carduus nutans in Ontario. Canadian Journal of Botany 34:71-
85.
Pacala, S. W. 1986. Neighborhood Models of Plant-Population Dynamics .2.
Multispecies Models of Annuals. Theoretical Population Biology 29:262-292.
Pauchard, A., P. B. Alaback, and E. G. Edlund. 2003. Plant invasions in protected areas
at multiple scales: Linaria vulgaris (Scrophulariaceae) in the west Yellowstone
area. Western North American Naturalist 63:416-428.
Pauchard, A., and K. Shea. 2006. Integrating the study of non-native plant invasions
across spatial scales. Biological Invasions 8:399-413.
Pielou, E. C. 1961. Segregation and Symmetry in 2-Species Populations as Studied by
Nearest-Neighbour Relationships. Journal of Ecology 49:255-269.
Purves, D. W., and R. Law. 2002. Fine-scale spatial structure in a grassland community:
quantifying the plant's-eye view. Journal of Ecology 90:121-129.
R Development Core Team. 2005. R: A language and environment for statistical
computing. in. Austria: R Foundation for Statistical Computing, Vienna.
Rew, L. J., B. D. Maxwell, F. L. Dougher, and R. Aspinall. 2006. Searching for a needle
in a haystack: evaluating survey methods for non-indigenous plant species.
Biological Invasions 8:523-539.
Rhodes, A. F., and T. A. Block. 2000. The Plants of Pennsylvania: An Illustrated Manual.
University of Pennsylvania Press, Philadelphia.
174 Seabloom, E. W., O. N. Bjørnstad, B. M. Bolker, and O. J. Reichman. 2005. Spatial
signature of environmental heterogeneity, dispersal, and competition in
successional grasslands. Ecological Monographs 75:199-214.
Sindel, B. M. 1991. A Review of the Ecology and Control of Thistles in Australia. Weed
Research 31:189-201.
Skarpaas, O., and K. Shea. in revision. Seed dispersal by wind: the relative importance of
environmental conditions and species characteristics. Journal of Ecology.
Skarpaas, O., and K. Shea. unpublished manuscript. Dispersal patterns, dispersal
mechanisms and invasion wave speeds for Carduus thistles.
Skinner, K., L. Smith, and P. Rice. 2000. Using noxious weed lists to prioritize targets for
developing weed management strategies. Weed Science 48:640-644.
Sokal, R. R., and F. J. Rohlf. 1995. Biometry: The Principles and Practice of Statistics in
Biological Research. W.H. Freeman and Company, New York.
Stoll, P., and E. Bergius. 2005. Pattern and process: competition causes regular spacing
of individuals within plant populations. Journal of Ecology 93:395-403.
Surles, W. W., and L. T. Kok. 1978. Carduus thistle seed destruction by Rhinocyllus
conicus. Weed Science 26:264-269.
Warwick, S. I., J. F. Bain, R. Wheatcroft, and B. K. Thompson. 1989. Hybridization and
Introgression in Carduus-Nutans and Carduus- Acanthoides Reexamined.
Systematic Botany 14:476-494.
Warwick, S. I., and B. K. Thompson. 1989. The Mating System in Sympatric Populations
of Carduus-Nutans, C-Acanthoides and Their Hybrid Swarms. Heredity 63:329-
337.
175 Wiens, J. A. 1989. Spatial Scaling in Ecology. Functional Ecology 3:385-397.
Wilson, J. B. 1988. Community Structure in the Flora of Islands in Lake Manapouri,
New-Zealand. Journal of Ecology 76:1030-1042.
Winston, M. R. 1995. Cooccurrence of Morphologically Similar Species of Stream
Fishes. American Naturalist 145:527-545.
176
Table 5-1: Regional survey summary Population summary
Year Number of Sites with of Sites with C. Sites with both thistle sites C. nutans acanthoides species 2004 77 51 33 7 2005 82 61 34 13
Block Summary
Number of C. nutans C. acanthoides Blocks with Year blocks surveyed blocks blocks both species 2004 44 21 16 6 2005 47 20 20 9
177
Table 5-2: Summary of the numbers of thistles in four fields of co-occurrence
Field P1 P1X P2 I R
Year 2004 2005 2004 2005 2004 2005 2004 2005 2004 2005
Flowering C. 1 11 12 62 10 39 20 21 52 78 nutans
Flowering C. 152 426 14 159 71 344 437 612 33 10 acanthoides
C. nutans 24 4 10 2 6 35 46 17 64 9 rosettes
C. acanthoides 902 354 5 41 33 148 1465 612 14 3 rosettes
C. nutans 0.11 0.07 2.00 1.36 0.05 0.23 0.37 0.21 0.53 0.36 plot density ± ± ± ± ± ± ± ± ± ± (plants/m2) 0.43 0.43 2.10 0.76 0.28 1.11 1.25 0.73 1.71 1.44 C. 4.49 3.71 1.73 4.26 0.32 1.52 10.75 6.80 0.21 0.05 acanthoides ± ± ± ± ± ± ± ± ± ± plot density 6.66 13.98 1.56 5.19 1.32 4.23 18.49 11.17 1.09 0.43 (plants/m2) C. nutans % attacked by - 64% 83% 55% 50% 46% 50% 52% 81% 58% R. conicus C. acanthoides 36% 53% 43% 30% 11% 33% 30% 48% 60% 60% % attacked by R. conicus Plot densities are reported as the mean ± standard deviation. For Site P2, the main and middle portions of the field are combined.
178
Figure 5-1: 2002 distributions of C. nutans and C. acanthoides Figure 1 shows the distribution of C. nutans and C. acanthoides in central Pennsylvania. redrawn from Allen and Shea (2006). The species are highly segregated, and there are fewer populations of both species in the area of overlap.
Thistle status 179 Empty C. acanthoides C. nutans Regional scale Both species Grain: 5 x 5 km
Field scale
Grain: 1 x 1 m
) ) ") ) ) ) ) ) ) ) ") ") ) ) ") ) ) ) ) ) ) ) ) ) )) ) ) ") ") ) ) ) ") ") ) ") ")") ") ") ") ) ) ") ") ) ") ) ") ") ") ) ")") ) ") ")) ) ") ) ) ") ") ") ) ") ") ") ")") ") ") ")") ") ") ") ") ") ") ") ") ") ") ") )) ") ") ") ")") ) ") ")") ) ") ") ") ) ") ") ") ") ") ") ") ") ") ") ") ") ") ") ") ") ") ") ")") ") ") ") ") ") ") ") ") ") ) ) ") ) ) ") ") ") ) ) ")") ) ) ) ") ) ) )" ") ) ") ) )) ) ) ") ") ) ) )") ) ") ") ") ) ") ) ") ") ") ) )
Plot scale
Grain: 5 x 5 cm )))))))))))))))))))) )))))))))"))))))))))) ))"))))))")))))))))))) )))))))))))))))))))) ))))))")))))))))))))) )))))))))))))))))))) ))))")))))))))))))))) )))))))))))))))))))) ))))))))))))))))")))) )))))))))))))))))))) ))"))))))))))))")"))))) )))))))))"))))"))))))) ))")))))")))")))))))))) )))))))))))))))))))) ))))))))"))))))")))))) ))))))))")))))))))))) )))))))))))))))))))) ))))))))"))))))))")))) )")))))))"))))")))))))) )")))))))))))))"))))))
Figure 5-2: Scales of study Each square indicates an area studied. Each empty square represents an absence, red squares indicate C. acanthoides presence, green squares indicate C. nutans presence, and blue squares indicate presence of both species. At the field scale, we studied 4 fields of co-occurrence, and at the plot scale, we had between 180 and 320 plots in each site. The data shown (from 2004, Site I, Plot 63) are presence-absence data; we also have density information for each square 180
2004 Populations !( C. acanthoides ± )" C. nutans Kilometers 0 2.5 5 10 15 20
") !(!( !"() ") !( ") !( ") ") !( !( ")") ") !(") ") ") ")") ") !( ")") ") ") ") ")") ") ")") ") ")") ") ")")") ")")") ") ")") !"() !(!( ") !( !( "()!") !( !( !(!"() ") !( !"()!( !(!( ") ") ") !( !(!"()!"() ")
2005 Populations (! C. acanthoides ± )" C. nutans Kilometers 0 2.5 5 10 15 20
)" )" )" )" )" !( !( )"!( )")" !( !( )")" !( )" )" )")" )" !( )" )" )"!( )" )" )" )" !( !( )" !( !( )" )" !( )"!()" )")" )")" )" )")" )")" )" )"!()" )" !(!( !()" )" !( )"!( )" )")" !( !( )")" )" )" )" !( !()"!( )")" )")"!( !(!( !( )" )"! )" !( !( !( !( )")"(
Figure 5-3: Regional survey maps from 2004 and 2005 ± The smaller map shows the relationship of our survey area to the larger area surveyed by Allen and Shea (2006).
181 2004 2005 .3 4 0 . 0 2 .2 . 0 0 0 . 0 .1 n 0 o .2 ti a -0 l 0 . rre 0 C. nutans o C
autocorrelation .6 .1 -0 -0 .2 0 - .0 -1
5 101520253035 5 101520253035 0 1 4 0. . 0 05 0. .2 0 n o ti a l 00 . 0 rre 0 o C. acanthoides . 0 C autocorrelation 05 0. - .2 -0 0 1 . 0 - 5 101520253035 5 101520253035 .3 0 0 .2 0 2 . 0 0 n .1 o 0 ti .1 a 0 l rre
Cross- o 0 C 0 . .0 0
correlation 0 .1 -0 0 .1 -0
5 101520253035 5 101520253035
Distance (km) Distance (km)
Figure 5-4: Presence–absence correlograms from the regional survey Correlograms show the correlations (within or among species) at various distances. The filled dots indicate that a particular coefficient was significantly negative or positive. An increment of 10 km was used for these analyses. The 2004 autocorrelation correlograms are not globally significant, nor is the 2005 C. acanthoides autocorrelation correlogram. There is positive autocorrelation for each species but negative cross-correlation between species.
182
2004 2005 4 . .5 0 0 .2 0 0 . n 0 o i at 0 . el 0 r r o C. nutans .5 0 C -
autocorrelation .2 0 - .0 -1 .4 -0
5 101520253035 5 101520253035 4 . 0 5 .1 3 0 . 0 0 .1 2 . 0 0 n o i 5 at 0 . .1 el r 0 0 r
C. acanthoides o C 0 0 . autocorrelation .0 0 0 5 .1 0 . -0 0 -
5 101520253035 5 101520253035 3 . 0 5 1 0. 2 0. n o 5 i 0 at 1 . 0. el 0 r r
Cross- o C
correlation 0 05 0. 0. - 1 . 0 - 5 1 . 0 - 5 101520253035 5 101520253035
Distance (km) Distance (km)
Figure 5-5: Density index correlograms from the regional survey The filled dots indicate that a particular coefficient was significantly negative or positive. An increment of 10 km was used for these analyses. All correlograms except the C. acanthoides autocorrelation correlogram in 2005 are globally significant. There is positive autocorrelation for each species but negative cross- correlation.
183
2004 2005 5 6 .0 .0 0 0 4 .0 3 0 0 n . o 0 ti a l 2 rre .0 o C. nutans 0 C 1
autocorrelation .0 0 0 .0 0 1 2 .0 .0 -0 -0 020406080 0 20406080 5 .2 0 5 .2 0 5 .1 0 n o 5 ti .1 a l 0 rre 5 o
C. acanthoides .0 C 0
autocorrelation 5 .0 0 5 .0 -0 5 .0 -0 020406080 0 20406080 4 02 0 . 0. 0 3 .0 0 00 0. 2 n .0 o ti 0 a l 02 1 rre 0 0. . - o
Cross- 0 C
correlation 0 4 .0 0 0 . 0 - 1 .0 -0
020406080 0 20406080
Distance (m) Distance (m)
Figure 5-6: Site P1 correlogram results The filled dots indicate that a particular coefficient was significantly negative or positive. All correlograms were globally significant, and they indicate both positive autocorrelation for both species, as well as positive cross-correlation. An increment of 2m was used for all correlograms.
184 2004 2005 20 . 0 6 0 0 . 15 . 0 0 on i t a 10 . 2 0 rrel 00 o C. nutans . C autocorrelation 0 0.05 0 0 0. -0.002
0 20406080 0 20406080
5 .2 0 5 1 0. 5 .1 0 n o 10 ti 0. a l rre 5 C. acanthoides o .0 C 0 05
autocorrelation 0. .05 00 0 - 0.
0 20406080 0 20406080 6 0 .0 1 0 0 . 0 4 .0 0 005 n 2 o 0. ti .0 a 0 l rre o Cross- 0 C .0 000 0
correlation 0. 2 .0 -0 005 . 0 -
0 20406080 0 20406080
Distance (m) Distance (m)
Figure 5-7: Site P2 main patch correlogram results For 2004 the correlograms are for flowering plants only for C. nutans, since no C. nutans rosettes were found, and the sample size is small (only 16 individuals). The filled dots indicate that a particular coefficient was significantly negative or positive. All correlograms were globally significant, except the 2004 autocorrelation in C. nutans. There is positive autocorrelation for both species, as well as positive cross-correlation. An increment of 2m was used for all correlograms.
185
2004 2005 3 0 0. 06 . 0 02 0. n 04 o 0. ti a 1 l 0 rre 0. C. nutans o 02 C autocorrelation 0. 00 0. 00 0.
0 1020304050 0 10203040 0 .2 3 . 0 0 5 .1 0 2 0. 0 n .1 o i 0 at l 1 e . r 0 r 5
C. acanthoides .0 Co autocorrelation 0 0 0 0. .0 0 5 1 .0 . 0 -0 - 0 1020304050 010203040
4 0 . 4 30 .0 0 0 . n 20 o 2 0 i . 0 . at l 0 e r r 10
Cross- 0 . Co
correlation 0 .0 0 00 0 . 0 2 1 .0 0 . -0 0 - 0 1020304050 010203040
Distance (m) Distance (m)
Figure 5-8: Site P2 middle patch correlogram results The filled dots indicate that a particular coefficient was significantly negative or positive. All correlograms were globally significant, and they indicate both positive autocorrelation for both species, as well as positive cross-correlation. An increment of 2m was used for all correlograms.
186 2004 2005 4 0 . 0 08 0. 3 .0 0 06 0. n 2 o .0 ti 0 a l 04 . 0 rre 1 C. nutans o .0 C 0 autocorrelation 02 0. 0 .0 0 00 0. 1 .0 -0 0 1020304050 0 102030405060 4 .5 0 0. 4 3 0. . 0 3 . n 0 o 2 ti a 0. l 2 rre 0. C. acanthoides o C 1 . 1 0
autocorrelation . 0 0 0 0. 0.
0 1020304050 0 102030405060 8 0 0 . .1 0 0 6 .0 0 6 4 n .0 o 0 .0 ti 0 a l 2 rre o Cross- .0 C 2 0 .0
correlation 0 0 .0 0 2 2 .0 .0 -0 -0
0 1020304050 0 102030405060
Distance (m) Distance (m)
Figure 5-9: Site I correlogram results The filled dots indicate that a particular coefficient was significantly negative or positive. All correlograms were globally significant, and they indicate both positive autocorrelation for both species, as well as positive cross-correlation. An increment of 2m was used for all correlograms.
187 2004 2005 5 .1 0 10 0. 0 .1 0 05 n 5 0. o ti .0 a l 0 rre 00 o 0
C. nutans 0. C 0 . autocorrelation 0 05 5 0. .0 - -0 0 1 -0. 0 1020304050 0 1020304050 20 0. .10 0 15 0. n o 6 ti 10 a .0 l 0. 0 rre C. acanthoides o C 05 0. autocorrelation 2 .0 0 00 0. 2 .0 -0 0 1020304050 0 1020304050 15 0. 10 0. 10 0. n o ti a l 05 05 rre 0. 0. Cross- o C correlation 00 00 0. 0.
0 1020304050 0 1020304050
Distance (m) Distance (m)
Figure 5-10: Site R correlogram results (shorter distances) The filled dots indicate that a particular coefficient was significantly negative or positive. All correlograms were globally significant, and an increment of 2m was used for all correlograms. There is still positive autocorrelation for each species, but the cross-correlation is generally negative, although the smallest distance coefficient is not significant either year.
188 2004 2005 6 .06 0 .0 0 4 4 .0 .0 0 0 n 2 o ti .0 .02 a l 0 0 rre C. nutans o 0 C 0 .0 .0 0 autocorrelation 0 2 2 .0 .0 -0 -0
0 200 400 600 800 1000 0 200 400 600 800 1000
5 .0 0 8 0 0 0. 3 .0 0 n o 004 ti a l 0. rre 1 o
C. acanthoides .0 C autocorrelation 0 000 0. 1 .0 -0 004 . 0 - 0 200 400 600 800 1000 0 200 400 600 800 1000
3 .0 0 010 0. 2 .0 0 n 005 o . ti 0 1 a l .0 0 rre 0 o
Cross- 0 C 0 0 0. correlation .0 0 1 005 .0 . 0 -0 -
0 200 400 600 800 1000 0 200 400 600 800 1000
Distance (m) Distance (m)
Figure 5-11: Site R correlogram results (longer distances) These correlogram use the same data as Figure 10, but with an increment of 50m. The filled dots indicate that a particular coefficient was significantly negative or positive. All correlograms were globally significant, and they indicate both positive autocorrelation for both species, but negative cross-correlation between the two species.
Chapter 6
Plant community associations of the invasive thistles Carduus nutans and C. acanthoides
Abstract
Much attention has been focused on identifying which communities are most
vulnerable to invasion by exotic species. Once established, invasive species can
significantly change the composition of the communities that they invade. Carduus
nutans and C. acanthoides are two similar invasive species, which have caused considerable economic damage worldwide. As a first step in understanding their
interactions with other plants, we recorded their plant associations in four fields in central
Pennsylvania, which contained both species. We found significant differences in
community composition in plots with and without Carduus thistles. However, we did not find significant differences between plots with C. nutans versus C. acanthoides, suggesting that they have similar interactions with the vegetation community.
Introduction
Invasive species are having enormous impacts globally, and are detrimentally impacting communities throughout the world (Mack et al. 2000). It seems obvious that invasives that become the dominant species will significantly change the communities they invade. However, even when invasives are not completely dominant, they may have
190 still have significant, perhaps more subtle effects. For example, Werner (1977) studied
the impact of Dipsacus sylvestrus (teasel) rosettes, which create openings in the
vegetation that favor establishment of not only teasel but also of winter annuals.
The opposite is also true: communities can impact whether or not invasive species
become established. Some communities seem more invasible than others, which may be
due to ecosystem and species characteristics, and propagule pressure (Lonsdale 1999).
Much of the literature has focused on whether or not more diverse communities are more
invasible (Elton 1958, Lonsdale 1999, Stohlgren et al. 1999, Naeem et al. 2000). Various
studies have reported both positive and negative relationships between native and exotic
species richness; these results appear to be a function of the scale at which this is studied
(Shea and Chesson 2002), and may be due to larger-scale studies encompassing more
spatial heterogeneity (Davies et al. 2005).
It may not always be useful to group together the impacts of invasive species, as it
has also been recognized that species identity of invaders may be critical. For example,
the novel weapons hypothesis proposes that certain invasive species have biochemical
exudates that allow them to succeed against previously unexposed species in their
invaded ranges (Callaway and Ridenour 2004). Recent work has identified that
Centaurea maculosa secretes (-)-catechin, which inhibits the growth of other plant species (Bais et al. 2003). However, in terms of management, it may be more useful to focus on communities than on particular weed species, in part because the removal of one problem species may facilitate the invasion of another species (Booth and Swanton
2002).
191 The ability to understand the interactions of invasive species with the resident plant community may be influenced by the local spatial structure. It has been increasingly recognized that autocorrelation is very common in ecological data, particularly with sessile species such as plants. Including spatial structure in statistical models may lead to different conclusions than assuming spatial independence (Keitt et al. 2002). Spatial autocorrelation in abundances may arise from population dynamics or in response to environmental factors (Keitt et al. 2002).
Carduus nutans and C. acanthoides are two invasive weeds, which are common
in roadsides and pastures in Pennsylvania. C. nutans and C. acanthoides are the second and fifteenth most commonly listed noxious weeds in the US (Skinner et al. 2000). These
Carduus species have a highly segregated distribution in central Pennsylvania with a
narrow area of overlap (Allen and Shea 2006), and one possible hypothesis for this
striking distribution is that they interact differently with the resident communities.
However, their interactions with other plants are not yet well understood. In order to
investigate this question, we surveyed the vegetation communities using randomly placed
quadrats in four sites of natural co-occurrence. We investigated Carduus thistle interactions with other plant community members and whether or not they were associated with certain communities, while controlling for spatial autocorrelation. We were particularly interested in whether C. nutans was associated with different vegetation than C. acanthoides.
192 Methods
Species description:
Carduus nutans L. and C. acanthoides L. (Asteraceae) are monocarpic perennials of Eurasian origin (Desrochers et al. 1988). They are quite similar, particularly during the rosette stage. Rosettes can occupy a considerable amount of space, with leaves up to 30 cm long (Desrochers et al. 1988). Vernalization is required for both species to bolt and flower. When bolting, the basal leaves of both species decay. Flowering individuals of both species produce thousands of seeds (McCarty 1982, Feldman and Lewis 1990).
Both species are commonly found in pastures and along roadsides, and seem to do well in disturbed areas (Kok 2001). Establishment of both species depends on the characteristics of potential germination sites, with generally better germination in larger gaps (Chapter 7, Panetta and Wardle 1992, Feldman et al. 1994, Ruggiero 2004).
Members of Onopordion-communites, to which both C. nutans and C. acanthoides belong, are more likely to occur in their native ranges in areas where existing vegetation and soil have been damaged but not destroyed (Doing et al. 1969).
Site description:
Three of the sites identified as co-occurrences by Allen and Shea (2006) in 2002
were suitable for within-field surveying because they were accessible and had more than
ten individuals of both species still present in 2003 (Sites P1, I and R). Another pasture
site (Site P2) was also included in the survey because the thistles are of particular
economic concern in pastures. Both pastures chosen are permanent pastures, as rotations
193 to tilling and cropping may break the cycle of biennials and perennials and obscure the co-occurrence patterns. All sites are located in Perry and Cumberland counties in
Pennsylvania. One site is an abandoned industrial site (Site I), two sites are pastures
(Sites P1 and P2) and one site is a long roadside over a forested ridge (Site R).
Site P1 (coordinates 40.379 N, 77.306W) has mostly C. acanthoides present as well as a few C. nutans individuals within the field. We surveyed an 80 m by 30 m area, which is used for occasional cattle grazing, despite the high density of thistles. Site P2 is also a pasture (coordinates 40.225 N, 77.431 W), in which we surveyed two large patches of thistles: an 80 x 25 m section near a temporary stream and a 40 x 45 m section in the center of the pasture. Site I (40.183 N, 77.238W) is an abandoned industrial site, which contains the highest densities of C. nutans we saw in Pennsylvania. We surveyed a 40 x
45 m portion of the field, which contained both species, although the C. nutans densities
were lower in that portion of the field. Site R (coordinates 40.301 N, 77.400 W) is located
along a major road through Colonel Denning State Park. This site is highly one-
dimensional: there is dense forest and thistles are only found immediately adjacent to the
road. C. nutans is found more near the top of the slope, whereas C. acanthoides is generally found further down the slope, although it does grow well on an unsurveyed portion of the top of the ridge, indicating that the distribution of the thistles in the survey is not just due to elevational differences.
Field methods
Each field was sampled by placing 1 x 1 m quadrats at pre-chosen random locations throughout the field. Sampling locations were randomly chosen by preselecting
194 random coordinates, in order to avoid problems of periodicity (Krebs 1989). Random sampling methods work better than transect methods if there is heterogeneity, although larger numbers of samples may be required to detect rare species (Goslee in press). At least 10% of each field was sampled. Each quadrat was subdivided into nine sectors to allow quantification of the species identity and evenness of the surrounding vegetation.
The number of times each species present occurred in each sector was recorded. In addition, in 2005 an abundance estimate was recorded for each species using Daubenmire cover classes (0-5%, 5-25%, 25-50%, 50-75%, 75-95%, 95-100%) (Bonham et al. 2004).
Statistical analyses
We used the “Ecodist” package (Goslee 2006) in R (R Core Development Team,
2005) for our analyses. For each field in each year we constructed three community
matrices, one of presence-absence, one of frequencies, and one of percent cover (2005
data only) of each plot. For percent cover, we used the midpoint of the cover classes to
indicate percent cover estimates. Distance matrices were calculated using Jaccard
distances for presence-absence data, which is suitable for species presence-absence
analyses because it does not consider joint absences, whose meaning is confounded in
ecological data (Legendre and Legendre 1998). Bray-Curtis distances, which are more
suitable for species abundance data (Legendre and Legendre 1998), were used for
frequency and percent cover data. Carduus thistles were excluded from the community
matrix, because we wanted to test whether or not the other community members were
different in plots with and without thistles. As differences in ecological data are often
better represented by non-Euclidean distance metrics, we use Nonmetric Multi-
195 dimensional Scaling (NMDS) to graph similar plots closer together and dissimilar plots further apart. NMDS is an ordination technique based on ranking distances (Legendre and Legendre 1998), which is more robust than other ordination techniques for analyzing community ecological data (Minchin 1987).
Because we were concerned with possible autocorrelation in the vegetation, as we knew there was significant autocorrelation in the Carduus thistle distribution in these fields (Chapter 5), Mantel correlograms of the community matrix were constructed
(Legendre and Fortin 1989). A correlogram is a plot of correlation at different distance classes. The global significance of a correlogram is determined by assessing whether at least one correlation coefficient is significant at the α´= α/υ (Bonferroni corrected level), where υ is equal to the number of distance classes, and we consider the α=0.05 level
(Legendre and Fortin 1989). Because significant spatial autocorrelation was detected, we used partial Mantel tests (Legendre and Fortin 1989) to test for community differences in the thistle and non-thistle plots, after spatial structure was accounted for.
To test whether or not there were differences between the communities in plots which contained C. nutans versus C. acanthoides, we subset the dataset to include only
plots that had either C. nutans or C. acanthoides present. Plots that contained both species were excluded from this analysis to avoid counting them twice. We calculated distance matrices for the subset data in the same manner as for the full dataset. NMDS was used to visualize the differences between C. nutans and C. acanthoides plots. Partial
Mantel tests were used to determine the significance of these differences, due to the presence of strong spatial structure.
196 Results
The proportion of plots that contained Carduus thistles varied considerably between sites and somewhat between years: between 14 and 69 % of the plots that we surveyed contained Carduus thistles (Table 6-1). All species found in all fields are listed
in Table 6-2; mostly grasses, such as Elytrigia repens, Arrhenatherum elatius and
Dactylis glomerata and dicots, such as Plantago species, Trifolium species, and
Taraxacum officinale, were present, with tree seedlings, vines and bushes more common
in Site R. All sites except Site R had more plots with C. acanthoides than C. nutans in both years. The Mantel correlograms of the presence-absence data (Figure 6-1) reveal the presence of significant spatial structure in the vegetation in these communities. Mantel correlograms of frequency and percent cover data also reveal similar spatial structure (not shown). Plots were generally positively correlated up to distances of around 20-50 meters in most sites. Site R had significant positive autocorrelation up to approximately 300 m in
2005. All correlograms were globally significant.
The results of the NMDS of the distances between all plots for presence-absence data are shown in Figure 6-2. There appear to be differences in the communities in the plots in areas of thistle presence compared to areas of thistle absence. Some degree of clustering in thistle plots is apparent in all fields in all years, although the pattern seems least strong in Site R. The results of the partial Mantel tests (Table 6-3) reveal significant differences in the presence-absence community data in thistle versus non-thistle plots in every field. The Mantel R values are particularly large (more than 0.32) for Site P2 and
Site I in both years, which indicate stronger differences between areas where thistles
197 occur and areas where they do not. The Mantel R values for site R are smaller, and indicate that the differences between the community composition in thistle and non- thistle plots are not so large at this site, which is on a forested roadside.
The patterns in plot differences in terms of the frequency data (NMDS shown in
Figure 6-3) are not as clear. Site P2 still has significant differences in the frequencies of the community members, as does Site P1 in 2004 and Site I in 2005. The Mantel R values are rather small for all significant results (less than 0.10). Site R does not have significant differences in frequencies in either year. It appears that the differences in the plant communities are not as strong in terms of plant frequencies, as they are for plant presence-absence.
The percent cover results are shown visually in Figure 6-4. Surprisingly, Sites P1,
P2 and I are all significantly different in terms of percent cover community composition in thistle and non-thistle plots; as with the frequency data, the Mantel R values are small.
Thus even though there are statistically significant differences, the magnitude of the difference may be small and is biologically not as important. Again, Site R did not have
significant community differences in percent cover of thistle versus non-thistle plots.
In order to examine whether or not there were differences in plots with C. nutans,
compared to plots with C. acanthoides, we examined NMDS of the distances between plots with thistles in them only, which are shown in Figure 6-5 for presence-absence data,
Figure 6-6 for frequency data and Figure 6-7 for percent cover data. Note that in Sites P1,
P2 and I there were very few plots, which had only C. nutans. Site I was excluded from
this analysis, because there were no plots in 2004 with just C. nutans, and there was only
1 plot in 2005. It appears that there are not strong differences in areas of C. nutans
198 presence, compared to areas of C. acanthoides presence. Generally the results indicate
that there are not significant differences between plots that had C. nutans present and
plots that had C. acanthoides only present (Table 6-4), although the small number of C. nutans available for this analysis limits the strength of this conclusion.
Discussion
We found strong evidence that there are different vegetation communities in plots
with and without Carduus thistles. The magnitude of the differences between communities was relatively consistent for the two years that we surveyed. The presence- absence data showed the strongest differences, with very large (often greater than 0.3)
Mantel R estimates. Differences in the frequency and percent cover analyses were often significant, but the magnitude of the differences was usually smaller. Thus there are more pronounced differences in presence-absence than in frequency or percent cover. The vegetation community in Site R is more that of a forest edge than a pasture, which may contribute to the lack of differences between thistle and non-thistle plots when using frequency and percent cover data.
We found considerable spatial structure in all sites and all years. We were particularly concerned with this as we knew there was spatial structure in the thistle distribution, with both auto- and cross-correlation being generally positive to approximately 20 m (Chapter 5). The spatial extent of autocorrelation was similar to that which we observed in the thistle distribution (about 20-30m) except for Site R, which had positive autocorrelation on a much larger scale (300 m). The larger spatial extent in the
199 community correlation of Site R is somewhat consistent with the spatial scale of negative cross-correlation observed between the two species but longer than the autocorrelation observed in each of these species. Spatial structure however is not solely responsible for the differences we saw, which were still significant after accounting for this spatial structure.
We were intrigued to see differences in the vegetation present in thistle and non- thistle plots. Unfortunately, this is a chicken and egg problem – is the community influencing the thistles or are thistles influencing the community or are both of these mechanisms occurring? It would be useful to conduct experiments to determine whether thistles cannot invade some areas of these pastures or whether, once present, they change the community in some way or if both of these are true.
The vegetation present in an area can have a large impact on thistle establishment.
Both of these species are more likely to invade disturbed areas, in part because there is less competition with established vegetation. Their germination and establishment are known to be microsite dependent (Chapter 7, Panetta and Wardle 1992, Feldman et al.
1994, Ruggiero 2004). C. nutans is particularly sensitive to competition during the rosette
stages (Austin et al. 1985), and is vulnerable to allelopathic effects of other species
(Wardle et al. 1995). Competition with other species has been suggested as a possible
management option (Kok et al. 1986, Wardle et al. 1995). Jongejans et al. (in revision)
found that mowing the surrounding vegetation influenced whether or not C. acanthoides
could invade; frequent mowing lead to dense, lawn-like vegetation which was not
conducive to C. acanthoides establishment.
200 C. nutans and C. acanthoides may also be having an influence on the surrounding vegetation. Both species have been reported as having allelopathic effects on other species (Woodward and Glenn 1983, Wardle et al. 1991a). In particular C. nutans may interfere with the nitrogen fixing abilities of T. repens through allelopathic effects of decaying rosette leaves; this may lead to lower nitrogen availability (Wardle et al.
1994a). It has also been suggested that C. nutans’ allelopathic effects may alter the
outcome of competitive interactions between grasses and legumes to favor grasses
(Wardle et al. 1994).
The species identity of immediate neighbors may play a large role in the community dynamics. We saw the strongest differences in presence-absence data: it may be that certain species have a particular effect on these Carduus thistles or are particularly affected by Carduus thistles. At this stage, we are not able to identify which species are
responsible for the differences that we see in the community. A useful next step is to use
logistic regressions to identify which of the many species are responsible for the
differences we observed in thistle versus non-thistle plots. This will allow a better
biological understanding of what causes the differences observed. It may be that there are
differences in the functional groups that are present in areas with and without Carduus
thistles. In particular, the abovementioned effects of Carduus thistles on legumes such as
T. repens may be of importance.
Importantly, we found no strong differences between communities associated with C. nutans and communities associated with C. acanthoides. These two species are
generally quite similar, and it appears that they have similar niches and play similar roles
in pastures. It would be interesting to see whether there are differences at larger spatial
201 scales between the vegetation with which they are associated. We gathered some preliminary presence-absence vegetation data from a larger scale survey of these two species, which seems to support such a hypothesis. This pattern is consistent with what we observed in Chapter 5: that the patterns observed are different at different spatial scales. However, different associations at larger spatial scales are more likely an artifact of their invasion history rather than an indication that they can only coexist with different plant communities, as their small scale patterns imply that they are associated with the same vegetation in areas of co-occurrence.
Invasive species are often viewed as having a different effect than residents
because they have no shared evolutionary history with the species in their invaded ranges.
Interestingly, many of the species that we identified in these surveys are also non-native
species. It appears that several of the species that we found are also associated with these
Carduus thistles in their native ranges. Doing et al. (1969) list other common members of the C. nutans and C. acanthoides group (Onopordion communities) in their native ranges;
many of the species listed we also found in our sites (Meliolotus albus, M. officinalis,
Verbascum thapsus, Datura stramonium, Cirsium arvense, Achillea millefolium), although the dominant species appear to be different. Thus presumably a large number of these species are not co-occurring for the first time in central Pennsylvania. It would be interesting to know whether their interactions are similar in their native and invaded ranges.
Our results highlight the need to examine the invaded plant community when studying invasive species. Many studies focus in the interactions of only a few species and consider the rest of the vegetation to be uniform. The differences that we were able to
202 detect were not immediately obvious when examining these fields and potentially play a major role in the invasion dynamics. We feel that more attention should be paid to interactions between all species present at an invaded site.
Acknowledgements
This research is in collaboration with my advisor, Katriona Shea. This work was supported by USDA-CSREES (Biology of Weedy and Invasive Plants) NRI grant #2002-
35320-1228 to KS and a NASA Space Grant Fellowship to ER. Sarah Goslee was extremely helpful with statistical analyses. Zeynep Sezen, Jessica Peterson, Joel McNeal and John Mellon helped with vegetation identification. Elizabeth Dlugosz, Adam Reese,
Stephen Selego, John Mellon, Jeff Butterbaugh and Erin Daewood helped tremendously with fieldwork, as well as Heather Alt, Ashley D’Antonio, Matt Clark, Paul Chen and
Adrianne Whitehair. Thanks to the Fry Family for generous use of their farm, the Shirk family and Eric Revene for use of their land.
References
Allen, M., and K. Shea. 2006. Spatial segregation of congeneric invaders in central
Pennsylvania, USA. Biological Invasions 8:509-521.
Austin, M. P., R. H. Groves, L. M. F. Fresco, and P. E. Kaye. 1985. Relative growth of
six thistle species along a nutrient gradient with multispecies competition. Journal
of Ecology 73:667-684.
203 Bais, H. P., R. Vepachedu, S. Gilroy, R. M. Callaway, and J. M. Vivanco. 2003.
Allelopathy and exotic plant invasion: From molecules and genes to species
interactions. Science 301:1377-1380.
Bonham, C., D. E. Mergen, and S. Montoya. 2004. Plant Cover Estimation: A
Continuous Daubenmire Frame. Rangelands 26:17-22.
Booth, B. D., and C. J. Swanton. 2002. Assembly theory applied to weed communities.
Weed Science 50:2-13.
Callaway, R. M., and W. M. Ridenour. 2004. Novel weapons: invasive success and the
evolution of increased competitive ability. Frontiers in Ecology and the
Environment 2:436-443.
Davies, K. F., P. Chesson, S. Harrison, B. D. Inouye, B. A. Melbourne, and K. J. Rice.
2005. Spatial heterogeneity explains the scale dependence of the native-exotic
diversity relationship. Ecology 86:1602-1610.
Desrochers, A. M., J. F. Bain, and S. I. Warwick. 1988. The biology of Canadian weeds
89. Carduus nutans L. and Carduus acanthoides L. Canadian Journal of Plant
Science 68:1053-1068.
Doing, H., E. F. Biddiscombe, and S. Knelhans. 1969. Ecology and distribution of the
Carduus nutans group (nodding thistles) in Australia. Vegetatio 17:313-351.
Elton, C. S. 1958. The Ecology of Invasions by Animals and Plante. The University of
Chicago Press, Chicago.
Feldman, S. R., and J. P. Lewis. 1990. Output and Dispersal of Propagules of Carduus-
Acanthoides L. Weed Research 30:161-169.
204 Feldman, S. R., J. L. Vesprini, and J. P. Lewis. 1994. Survival and establishment of
Carduus acanthoides L. Weed Research 34:265-273.
Goslee, S. C. in press. Behavior of vegetation sampling methods in the presence of spatial
autocorrelation. Plant Ecology.
Jongejans, E., O. Skarpaas, P. W. Tipping, and K. Shea. in revision. Establishment and
spread of founding populations of an invasive thistle: the role of competition and
seed limitation.
Keitt, T. H., O. N. Bjornstad, P. M. Dixon, and S. Citron-Pousty. 2002. Accounting for
spatial pattern when modeling organism-environment interactions. Ecography
25:616-625.
Kok, L. T. 2001. Classical biological control of nodding and plumeless thistles.
Biological Control 21:206-213.
Kok, L. T., T. J. McAvoy, and W. T. Mays. 1986. Impact of Tall Fescue Grass and
Carduus Thistle Weevils on the Growth and Development of Musk Thistle
(Carduus-Nutans). Weed Science 34:966-971.
Krebs, C. J. 1989. Ecological methodology. Harper & Row, New York.
Legendre, P., and M. J. Fortin. 1989. Spatial Pattern and Ecological Analysis. Vegetatio
80:107-138.
Legendre, P., and L. Legendre. 1998. Numerical Ecology. Elsevier, New York.
Lonsdale, W. M. 1999. Global patterns of plant invasions and the concept of invasibility.
Ecology 80:1522-1536.
205 Mack, R. N., D. Simberloff, W. M. Lonsdale, H. Evans, M. Clout, and F. A. Bazzaz.
2000. Biotic invasions: Causes, epidemiology, global consequences, and control.
Ecological Applications 10:689-710.
McCarty, M. K. 1982. Musk Thistle (Carduus-Thoermeri) Seed Production. Weed
Science 30:441-445.
Minchin, P. R. 1987. An Evaluation of the Relative Robustness of Techniques for
Ecological Ordination. Vegetatio 69:89-107.
Naeem, S., J. M. H. Knops, D. Tilman, K. M. Howe, T. Kennedy, and S. Gale. 2000.
Plant diversity increases resistance to invasion in the absence of covarying
extrinsic factors. Oikos 91:97-108.
Panetta, F. D., and D. A. Wardle. 1992. Gap Size and Regeneration in a New-Zealand
Dairy Pasture. Australian Journal of Ecology 17:169-175.
Ruggiero, D. 2004. Factors Affecting the Germination of the Invasive Thistles: Carduus
nutans and C. acanthoides. Honors Thesis. The Pennsylvania State University,
University Park.
Shea, K., and P. Chesson. 2002. Community ecology theory as a framework for
biological invasions. Trends in Ecology & Evolution 17:170-176.
Skinner, K., L. Smith, and P. Rice. 2000. Using noxious weed lists to prioritize targets for
developing weed management strategies. Weed Science 48:640-644.
Stohlgren, T. J., D. Binkley, G. W. Chong, M. A. Kalkhan, L. D. Schell, K. A. Bull, Y.
Otsuki, G. Newman, M. Bashkin, and Y. Son. 1999. Exotic plant species invade
hot spots of native plant diversity. Ecological Monographs 69:25-46.
206 Wardle, D. A., M. Ahmed, and K. S. Nicholson. 1991. Allelopathic Influence of Nodding
Thistle (Carduus-Nutans L) Seeds on Germination and Radicle Growth of Pasture
Plants. New Zealand Journal of Agricultural Research 34:185-191.
Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1994. Interference Effects
of the Invasive Plant Carduus-Nutans L against the Nitrogen-Fixation Ability of
Trifolium-Repens L. Plant and Soil 163:287-297.
Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1995. Influence of Pasture
Forage Species on Seedling Emergence, Growth and Development of Carduus
Nutans. Journal of Applied Ecology 32:225-233.
Werner, P. A. 1977. Colonization Success of a Biennial Plant Species - Experimental
Field Studies of Species Cohabitation and Replacement. Ecology 58:840-849.
Woodward, M. D., and S. Glenn. 1983. Allelopathic Effects of Three Thistle Species.
Proceedings of the 37th Annual Meeting of the Northeastern Weed Science
Society:114.
207
Table 6-1: Percentages of plots with thistles in the four fields of co-occurrence P1 P2 I R Year 2004 2005 2004 2005 2004 2005 2004 2005
Plots 235 210 324 324 177 180 220 240 sampled
Plots with Carduus 69% 64% 14% 30% 60% 56% 25% 15% thistles
Plots with C. nutans 8% 4% 4% 8% 14% 13% 19% 13%
Plots with C. nutans 0.4% 1% 2% 4% 0% 1% 19% 12% only
Plots with C. 68% 63% 11% 26% 60% 55% 6% 3% acanthoides
Plots with C. acanthoides 60% 60% 10% 22% 46% 43% 6% 3% only
Plots with Both 8% 3% 1.2% 5% 14% 12% 0.5% 0.4% Species
208
Table 6-2: Plant species list Plants Present in Present in Scientific name P1 P2 I R Scientific name P1 P2 I R Acer rubrum x x Meliolotus officinalis x x x x Ailanthus altissima x Nepeta cataria x x x x Alliaria petiolata x Oxalis strica x x x Parthenocissus Allium candense x x quinquefolia x x x Allium vineale x x Phleum pratense x x x Arcticum minus x x x Phytolacca americana x Arrhenatherum elatius x x x Plantago lanceolata x x x x Asclepias syriaca x x Plantago major x x x x Barbarea vulgaris x Polygonum cuspidatum x Capsella bursa-pastoris x Polygonum persica x x x Carduus acanthoides x x x x Portunlaca oeracea x x x Carduus nutans x x x x Potentilla recta x x Carex lurida x Prunus species x Centauria maculosa x x x Quercus species x Chenopodium album x x Rhus species x Chrysanthemum leucanthemem x Robinia pseudoacacia x Cirsium arvense x x x x Rorippa sylvestrus x Cirsium vulgare x x x x Rosa multiflora x Commelina communis x Rosa species x Coronilla varia x x x Rubus species x x Dactylis glomerata x x Rumex crispus x x Datura stromonoium x Solidago species x x x Daucus carota x x x x Stellaria media x x Dipsacus fullonum x x x x Taraxacum officinale x x x x Elytrigia repens x x x x Thalaspi arvense x x Erigeron philadelphicus x x x x Toxicodendron radicans x Festuca arundinacea x Tragopogon pratensis x Gallium species x x Trifolium arvense x Glechoma species x x Trifolium pratense x x x Hesperis matronalis x x Trifolium repens x x x Impatiens pallida x Verbascum thapsus x Juncus effusus var. solutus x Veronica officinalis x Lepidium campestre x Vinca minor x Linaria vulgaris x x x Vitus species x x Liriodendron tulipifera x Zea mays x Lonicera species x Unknown moss x Malva neglecta x x Unknown dicot x x x x Medicago lupulina x x Unknown grass x x x x Meliolotus alba x x x
Abiotic categories Bare ground x x x x Water x x Rock x x
209
Table 6-3: Differences in community in thistle versus non-thistle plots: Results of partial Mantel tests 2004 2005 Type of Site Mantel R p-value Mantel R p- value Data Presence- P1 0.33 0.001*** 0.32 0.001*** absence P2 0.19 0.001*** 0.25 0.001*** I 0.38 0.001*** 0.36 0.001*** R 0.13 0.001*** 0.08 0.001***
Frequency P1 0.05 0.036* 0.02 0.112 P2 0.08 0.024* 0.09 0.003** I 0.01 0.219 0.10 0.001* R -0.02 0.720 0.01 0.303
Percent P1 - - 0.05 0.002** cover P2 - - 0.10 0.001*** I - - 0.10 0.001*** R - - 0.01 0.285 • indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01 *** indicates significance of p≤0.001
210
Table 6-4: Differences in community in C. nutans vs. C. acanthoides plots: Results of partial Mantel tests 2004 2005 Type of Site Mantel R p-value Mantel R p- value Data Presence- P1 0.02 0.378 -0.03 0.693 absence P2 -0.09 0.886 0.08 0.135 R 0.07 0.096• -0.03 0.632 Frequency P1 0.02 0.354 -0.03 0.703 P2 -0.09 0.898 0.08 0.091 R 0.07 0.093• -0.03 0.594 Percent P1 - - -0.03 0.703 cover P2 - - 0.08 0.122 R - - -0.03 0.623 Note that site I was not included in this analysis, as there were too few plots with C. nutans only. • indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01 *** indicates significance of p≤0.001
211
2004 2005 5 .1 0 0 .1 0 n 0 o .1 i 0 5 elat .0 5 0 rr .0 o 0 C 0 el 0
P1 .0 .0 0 0 ant M 5 5 0 . .0 -0 -0
0 20406080 0 20406080
15 . 0 10 n 10 . . 0 o 0 i elat 05 rr 5 0. 0 o 0. 0 C 0 el P2 0. ant 00 05 0. M 0. - 0 1 . 0 - 0 20 40 60 80 100 120 020406080100120
5 15 .1 0. 0 n o 0 i 10 . .1 0 0 elat 5 rr 05 .0 o 0. 0 C 0 00 el .0 0.
I 0 ant 05 5 M 0. .0 - 0 - 0 1 . 0 - 0 1020304050 0 1020304050
0 5 .1 .1 0 0 n o i 0 .1 5 0 elat .0 0 rr o 5 .0 C 0 el 0
R .0 0 0 ant .0 0 M 5 5 .0 .0 -0 -0
0 200 400 600 800 1000 0 200 400 600 800 1000 Distance (m) Distance (m) Figure 6-1: Mantel correlograms using presence-absence data The correlograms shown are plots of the correlation in the vegetation community at different distance classes. There is significant positive autocorrelation in all cases, meaning that plots that are closer (in geographical distance) are more likely to be similar. Correlation coefficients that are significantly different from zero are shown with filled dots.
212
2004 2005 .6 6 . 0 0 4 4 . . 0 0 2 .2 0 0. 0 . 0 X2 2 . .2 0 -0 P1 - 6 . 6 . -0 0 -
-0.5 0.0 0.5 -0.5 0.0 0.5
.5 0 5 . 0 0 . 0 X2 0 P2 0. .5 0 - 5 -0.
-0.5 0.0 0.5 -0.5 0.0 0.5
.5 .6 0 0 4 . 0 .2 0 0.0 X2 0.0 .2
I 0 - .5 0 - .6 0 -
-0.6 -0.2 0.0 0.2 0.4 0.6 0.8 -0.6 -0.2 0.0 0.2 0.4 0.6
5 5 . . 0 0 0 0 0. 0. X2 R 5 . 5 . -0 -0
-0.5 0.0 0.5 -0.5 0.0 0.5 X1 X1 Figure 6-2: Nonmetric multi-dimensional scaling of all plots: presence-absence data X1 and X2 are the two dimensions of the NMDS (not Euclidean coordinates). NMDS graphs more similar plots closer to each other, as a way to visualize the differences in the community in the plots studied. C. nutans plots are shown as red circles, and C. acanthoides plots are shown as blue squares. In many fields, it appears that plots with thistles in them are closer together, implying that they are more similar to each other than to non-thistle plots.
213
2004 2005 6 . 0 4 .5 . 0 0 2 . 0 0 . 0 . 0 0 X2
P1 4 . -0 .5 0 - 8 . -0
-0.8 -0.4 0.0 0.2 0.4 0.6 -0.6 -0.2 0.0 0.2 0.4 0.6
8 . 0 .6 0 5 . 0 0.4 .2 0 0 X2 0. 0 . P2 0 2 . 0 - 5 -0. .6 0 - -0.5 0.0 0.5 -0.5 0.0 0.5
0.4 6 . 0 .2 0 4 0. 0 . 0 2 . 0 .2 X2 0 0 - 0. 2 I .4 0 - -0. 6 . 0 - 6 . .8 -0 0 - -0.5 0.0 0.5 -0.4 -0.2 0.0 0.2 0.4 0.6
5 . .5 0 0 0 0 . . X2 0 0 R .5 0 5 - . -0
-0.5 0.0 0.5 -0.5 0.0 0.5 X1 X1 Figure 6-3: NMDS of all plots: frequency data X1 and X2 are the two dimensions of the NMDS. C. nutans plots are shown as red circles, and C. acanthoides plots are shown as blue squares.
214
P1 P2 .5 0 .5 0 0 . X2 0 0 . 0 .5 -0 .5 -0
-0.5 0.0 0.5 -0.5 0.0 0.5
I R 8 0. .6 0 .5 0 4 0. .2 0 0 X2 . 0 0 0. 2 -0. .5 0 - 6 -0.
-0.6 -0.4 -0.2 0.0 0.2 0.4 -0.5 0.0 0.5 X1 X1 Figure 6-4: NMDS of all plots: percent cover data (2005 only) X1 and X2 are the two dimensions of the NMDS. C. nutans plots are shown as red circles, and C. acanthoides plots are shown as blue squares.
215 2004 2005 5 . .5 0 0 0 . 0 . 0 X2 0 P1 .5 .5 -0 0 -
-0.5 0.0 0.5 -0.6 -0.2 0.0 0.2 0.4 0.6
.6 .6 0 0 4 . 4 . 0 0 .2 .2 0 0 0 . 0 0 . 0 X2 2 . -0
P2 4 . 4 -0 . -0 .6 0 - .8 0 -
-0.6 -0.2 0.0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
6 .6 . 0 0 4 4 . . 0 0 .2 .2 0 0 0 0 . . 0 0 X2 R .4 .4 0 0 - - 8 . .8 -0 -0 -0.6 -0.2 0.0 0.2 0.4 0.6 -0.5 0.0 0.5 X1 X1 Figure 6-5: NMDS of Carduus plots only: presence-absence data X1 and X2 are the two dimensions of the NMDS. C. nutans plots are shown as red circles, and C. acanthoides plots are shown as blue squares.
216 2004 2005 .5 5 . 0 0 0 . 0 . 0 X2 0 P1 .5 .5 -0 -0
-0.6 -0.2 0.0 0.2 0.4 0.6 -0.6 -0.2 0.0 0.2 0.4 0.6
.8 6 0 . 0 6 . 0 4 . 0 4 . 0 .2 0 2 . 0 0 . X2 0 .2 0 P2 .2 - 0 - .4 0 - .6 .6 0 0 - -
-0.6 -0.4 -0.2 0.0 0.2 0.4 -0.6 -0.2 0.0 0.2 0.4 0.6
6 . 6 . 0 0 4 . 4 . 0 0 .2 0 .2 0 0 . 0 0 . X2 0 2 .
R -0 .4 0 - .6 0 - 8 . -0 -0.5 0.0 0.5 -0.5 0.0 0.5 X1 X1 Figure 6-6: NMDS of Carduus plots only: frequency data X1 and X2 are the two dimensions of the NMDS. C. nutans plots are shown as red circles, and C. acanthoides plots are shown as blue squares.
217
P1 P2 .8 0 6 . 0 .5 0 .4 0 2 . 0 0 . X2 0 .2 0 - .5 0 - .6 0 -
-0.6 -0.2 0.0 0.2 0.4 0.6 -0.6 -0.2 0.0 0.2 0.4 0.6
R
.5 0 0 . 0 5 . -0
-0.6 -0.2 0.0 0.2 0.4 0.6 0.8 X1 Figure 6-7: NMDS of Carduus plots only: percent cover data (2005 only) X1 and X2 are the two dimensions of the NMDS. C. nutans plots are shown as red circles, and C. acanthoides plots are shown as blue squares.
Chapter 7
Effect of microsite characteristics on Carduus nutans and C. acanthoides germination and survival
Abstract
The successful establishment of invasive species has been shown to depend on
aspects of the invaded community such as gap characteristics. Biotic resistance may be
particularly critical for stopping invaders at early life history stages. We examine the
response of two invasive thistle species, Carduus nutans and C. acanthoides to three different microsite characteristics: size, water availability and disturbance type. Increased germination and survival were observed for both species in larger gaps. Watering generally had a negative impact on germination but a positive impact on size. The two species responded differently to the different types of disturbance: C. acanthoides had higher germination in tilled plots (with both above- and belowground disturbance), whereas C. nutans had better performance in clipped plots (with above-ground
disturbance only). These results suggest that microsite characteristics have a large impact
on establishment of these two Carduus species.
Introduction
A suitable microsite for germination and seedling establishment is a necessary condition for the success of any plant species. Understanding the effect of microsite
219 characteristics has particular importance in the case of invasive species, as this is the first barrier offered by a community against invasion. Gaps in the existing vegetation create opportunities for plants to establish due to the exposure of soil, removal of competing vegetation and removal of litter (Goldberg and Gross 1988). In mature plant communities, gaps in the vegetation occur due to death of individuals or through animal activity. Even among species with generally similar life histories, species may have very different responses to gap opportunities (Gross and Werner 1982).
Gap characteristics can be important in understanding the regeneration niche,
which includes dispersal, germination, establishment and seedling development (Grubb
1977). Gap characteristics, such as size, can have an important impact on plant
establishment, growth and reproduction (McConnaughay and Bazzaz 1990). Although
larger gaps have fewer potential competitors, they are also more vulnerable to
desiccation, leading to trade-offs in optimal gap sizes, which may be different among
species. The probability of finding a suitable germination site can also be affected by soil
heterogeneities (Harper et al. 1965), which can vary between gaps.
The speed of emergence, which can be affected by gap characteristics, can lead to
competitive advantage (Bergelson 1997), as individuals have an opportunity to increase
in size when there is less competition. In fact, the effects of early emergence can be even
larger than just the effect of having extra time to grow (Ross and Harper 1972). The
availability of gaps may also differ throughout the year (Grubb 1977), thus the timing of
germination relative to gap availability may be critical. Short-term changes in water
availability have been shown to impact the establishment of novel species even one year
after the changes were applied (Davis and Pelsor 2001).
220 Better plant growth in gaps may be because of a reduction in the intensity of belowground competition and not just due to light availability, as root biomass has been shown to be lower in gaps in an abandoned hayfield (Cahill and Casper 2003). The effects of above-ground and belowground competition are not necessarily additive; Cahill
(2002) found that for two rosette forming species, the combined competitive response to above-ground and below-ground competition were subadditive, indicating that these responses are not independent.
The ability of invasive species to establish must be understood in order to predict their dynamics (King and Grace 2000). For invading plants, competition with established vegetation can make it difficult for seedlings to emerge. Thus, in the case of invasive species, understanding the factors that affect establishment is of critical applied importance.
Carduus nutans and C. acanthoides are invasive species of Eurasian origin which
have spread to become invasive throughout the world (Holm et al. 1979). They cause
considerable economic damage in pastures, which may have many gaps of various sizes;
thus their establishment response to gap characteristics is important to understand.
Despite substantial propagule pressure, C. nutans seldom establishes in well managed
pastures, presumably due to a lack of suitable gaps for colonization (Hamrick and Lee
1987).
C. nutans and C. acanthoides germination and survival were previously shown to be microsite size dependent in a one year study, although, surprisingly, a watering treatment appeared to have no effect on thistle germination and survival (Ruggiero 2004).
It has been suggested that for C. nutans there may be decreased emergence in extremely
221 large plots and that intermediate sized gaps may best promote germination (Panetta and
Wardle 1992). It is also important to consider whether gaps are primarily above-ground or both above and below-ground, as below-ground interactions may also be important
(Wardle et al. 1992).
It is very important to conduct such studies in realistic field settings, as the conclusions reached may be different in different environments. For example, Panetta and Wardle (1992) found that C. nutans had the highest emergence with 10 cm gaps in a field setting, but with no gaps in a greenhouse setting.
Our aim was to better understand how gap characteristics affect germination and establishment of these species in a field setting, and to examine whether there were any differential responses in the C. nutans and C. acanthoides. Although we mainly focus on the germination and seedling stages, we also follow individuals through to flowering. We expand the approach of Ruggiero (2004) to include disturbance type: above ground disturbance only (similar to extreme overgrazing) versus above and below-ground disturbance (similar to tilling). We hypothesized that germination would be best for both species in larger microsites, that watering would have a beneficial impact, and that thistle performance would be best in plots that had received both above-ground and below- ground disturbance.
222
Methods
Study species
Carduus nutans and C. acanthoides are herbaceous monocarpic perennials; they
are sometimes annual, winter annual or perennial (Desrochers et al. 1988). Individual
plants produce thousands of seeds (up to approximately 20,000 seeds, Kok 2001), which
are wind dispersed. C. nutans seeds are larger (2-4 mm) and heavier ( 4 mg) than C.
acanthoides (1-3 mm, 2 mg, McCarty et al. 1969). There is no record of vegetative propagation of either species (Desrochers et al. 1988).
C. nutans and C. acanthoides germination was well studied in a laboratory context (McCarty et al. 1969). Both species have maximal emergence at depths of 0.5-1.0 cm in the soil. C. nutans is able to germinate from greater depths than C. acanthoides,
which does not emerge from depths greater than 4 cm. Seeds require maximal contact
with soil in order to absorb enough moisture for germination. C. acanthoides appeared to
be more sensitive to moisture stress, although C. nutans is also affected by moisture conditions. Germination is not strongly light dependent (McCarty et al. 1969). There is no period of dormancy required for C. nutans’ germination (Doing et al. 1969, Lee and
Hamrick 1983).
These species are difficult to distinguish at the rosette stages; however, after overwintering, species differences in leaf form and pubescence are more pronounced.
The seedlings of both species are subject to similar, moderate levels of herbivory in the
223 field, mainly by grasshoppers, crickets and other insect herbivores (E. Rauschert, pers obs).
Experimental design
The experiments were conducted in a field at the Russell E. Larson Agricultural
Center at Rock Springs, which is located about 10 miles southwest of University Park,
Pennsylvania. This location in the Ridge and Valley province is typical of a central
Pennsylvania landscape. The site is a former pasture that has been left ungrazed for more
than a decade, with mostly weedy grasses and dicots present. The site is typical of a
central Pennsylvanian hayfield; the dominant species are mostly the grasses Elytrigia
repens, Arrhenatherum elatius, Dactylis glomerata and Phleum pretense, and the dicots
Plantago lanceolata, Taraxacum officinale, Trifolium repens, Trifolium pratense and
Gallium species. The soils are located within the Hagerstown silt-loam series.
The experiment had a randomized, full block design. There were three treatments:
watering, microsite size and type of microsite disturbance. Four microsite sizes were
investigated: squares with lengths of 5, 15, 30 and 50 cm along an edge. In 2004, due to
size constraints on the available portion of the field, the outer size of the plots was 50 x
50 cm, which is the same size as the largest disturbance. In 2005, we were able to use a
different portion of the field that was larger; all plots were 1 x 1 m in size, in order to
have a larger buffer area between plots. Disturbances were created in two different ways:
a clipping treatment disturbed only the above-ground biomass but left the soil intact, and
a tilling treatment disturbed both above and below-ground areas. The clipping treatment
involved removing all above-ground biomass with electric clippers or scissors. The tilling
224 treatment consisted of either roto-tilling the plot (for 30 and 50 cm edge length disturbances) or drilling holes with a bulb planting attachment (for the 5 and 15 cm
disturbances). Half of the plots received a watering treatment, which consisted of adding
1 L of water over a 50 x 50 cm area twice weekly. There were a total of 32 plots types (2
species x 2 watering types x 4 microsite sizes x 2 microsite disturbance types), which
were replicated ten times in each of two years.
The seeds used in this experiment were collected from naturalized Carduus
populations from areas that only have species present (near Carlisle, PA for C. nutans, and in State College for C. acanthoides). Flowerheads collected were dissected to remove
seeds, after which seeds were sifted with mesh screens to remove small flat seeds, which
are not typically viable (Kelly et al. 1988).
Four seeds of a species were planted in the center of each disturbance in
September of 2004 or 2005. This controls for seed limitation and allows us to focus on
the effect of microsite. After planting, plots were lightly maintained to retain differences
between the treatments during the germination period (i.e. for the first month). Plots were
censused three times weekly for new germinations for the first month. During censuses,
herbivory and death of previously emerged seedlings were recorded. At the end of the fall growing season, in early November, the longest leaf length and diameter of each seedling was recorded, in addition to whether or not the plant was herbivorized. The watering treatment was continued until the end of the fall growing season each year.
In May and June of 2005, the longest leaf length and diameter of the surviving rosettes was measured. All mature flowerheads were bagged with pollen bags in the field after pollination had occurred, to avoid contaminating the soil further with thistle seeds.
225 In July 2005, a destructive census of flowering plants was carried out. The height, number of stems, root crown diameter and number of heads was recorded for flowering plants, and the longest leaf length and diameter was measured for rosettes. The second cohort is still in progress and will be censused in the same manner.
Statistical methods
Analyses were performed separately for the two species. In order to avoid pseudoreplication problems, the average plot response was used as the response variable in all analyses (Crawley 2002). We examined logistic regressions of the proportion of seeds planted that germinated, the proportion of seeds planted that germinated and
survived in the fall, the proportion of germinating individuals that survived in the fall and
the proportion of germinating individuals that survived until the following summer. In
each case we used a vector of the number of successes to failures in order to account for
sample size in estimating proportions (Crawley 2002). We used quasipoisson regression
to analyze the average number of days to germination. We examined linear regressions of
log transformed size data (longest leaf length at the end of fall). We also examined the
presence or absence of herbivory at the fall full census using logistic regressions. All
analyses were done using R (R Development Core Team 2005).
The explanatory variables were whether or not a plot was watered, the area of the
microsite disturbance and the type of the microsite disturbance. The area of the microsite
was calculated by squaring the length of an edge, as the disturbances were square. The
area of microsite disturbance was then normalized by dividing all sizes by the largest size
(2500 cm2) so that the maximum was 1, to allow easier comparison of their relative
226 effects with the watering and disturbance type variables, which were coded as 0 and 1.
This rescaling has no effect on the statistical significance of parameters. In most analyses, if a response variable is positively related to one of the treatment variables, this implies that individuals had better performance in plots with that treatment; two exceptions to this are analyses of the average days to germination and the presence or absence of herbivory. A positive relationship with the average days to germination implies slower germination, and a positive relationship with herbivory implies greater herbivory in plots with that treatment.
Akaike’s Information Criterion (AIC) was used for model selection to decide whether or not to include interactions between treatment effects. The random effect of blocking and year were incorporated by using mixed models (glmmPQL in R). The random year effect was also not included for models of survival until destructive census, as only one year of data was available (the second cohort is currently in progress). The
significance of the terms in the models was determined through an analysis of variance of
the fitted models.
Results
Microsite characteristics strongly influenced the establishment of both species,
which had rather different germination rates. C. nutans had higher germination rates than
C. acanthoides (Table 7-1). There was more germination in the second year for both
species; however, the average time to germination was also longer in 2005 (Figure 7-1 ).
Nearly all germinations occurred in the fall; in 2005 only four new germinations were
227 observed in the spring for the cohort planted in 2004 (the cohort planted in 2005 is still in progress). C. nutans grew to slightly larger sizes than C. acanthoides by early November.
The results of the germination and survival models are summarized in Table 7-2; the
analyses of the average days to germination, fall size and herbivory are described in
Table 7-3.
Larger microsite size was generally beneficial to germination, survival and fall
size (Figure 7-2). The average days to germination were slightly longer in larger
microsite environments, indicating slower germination; however, this relationship was
only significant for C. nutans. The herbivory models suggest less herbivory in larger
microsites, which are where plants were larger; however, these results are not significant.
The watering treatment had a negative impact on germination and survival
(Figure 7-3). C. acanthoides survival appeared to be more impacted than C. nutans.
Interestingly, seeds in watered plots appeared to germinate faster. There is no size response (average longest leaf length) to watering for C. acanthoides, whereas C. nutans
was larger in watered plots. The results of the herbivory model suggest that there was
more herbivory in watered plots, but these results are not significant, especially for C.
acanthoides.
Interestingly, the two species appear to have had a different response to the type
of disturbance (Figure 7-4). C. nutans always performed better in clipped versus tilled
plots. C. acanthoides had better germination in tilled plots. The survival of individuals
that germinated appeared to be higher in clipped plots, but this relationship was not
significant for C. acanthoides. The average rosette size in fall was higher and the average
days to germination was lower for C. acanthoides in tilled plots. Herbivory patterns
228 appear to act in the opposite direction: C. acanthoides experienced greater levels of herbivory in tilled plots, and C. nutans experienced greater levels of herbivory in clipped plots, although this is not significant.
Discussion
Both C. nutans and C. acanthoides produce large numbers of seeds per individual, yet very few of these seeds actually lead to established rosettes. Our results confirm that
C. nutans and C. acanthoides germination, survival and growth are strongly dependent on characteristics of their germination sites. In related studies, we have observed very poor germination of these species in undisturbed vegetation (K. Shea, unpublished data). In particular, the size of the microsite in which seeds are placed appears to be particularly important. Goldberg and Gross (1988) studied gap characteristics in a mid-successional old field, and found that most gaps were small (less than 10 cm in size), and that animals
were creating most of the new gaps. In New Zealand pastures with C. nutans infestations, gap sizes were also found to be small (mostly less than 10 cm), and there were fewer gaps in spring and autumn, when most germination occurs (Panetta and Wardle 1992). Panetta and Wardle (1992) suggest that hoof sized gaps may best promote germination. This would be an unfortunate result for management, as grazing in wet pastures often creates gaps of such sizes. We observed increased survival and germination with microsite size; thus our results suggest that larger sized gaps may actually best promote germination.
Wardle et al. (1995) claim that some pasture cover may be needed for a suitable microclimate for C. nutans germination, as they often had low germination in large bare
229 ground plots. However, even in their experiments, individuals in the bare plots were more
likely to flower in the first summer and grew larger than in plots with other potential
competitors. Attainment of large size when exposed to little competition agrees with our
observation of large, naturally occurring individuals of both species on highly disturbed
areas with virtually no other vegetation, such as construction sites (E. Rauschert, pers
obs).
Interestingly, watering generally had a negative impact on germination and
survival in our study. This was not what might have been expected given previous work
on these species. For example, C. acanthoides has been shown to be sensitive to desiccation during germination and early growth (Feldman et al. 1994). It is possible that watering may have washed some seeds deeper into the soil; both species have decreased emergence at deeper in the soil, with C. acanthoides is more vulnerable to this effect
(McCarty et al. 1969). It is also possible that the addition of water initially benefited the existing vegetation more than the thistle seeds. That germination was faster in watered plots is consistent with what has been previously found; Hamrick and Lee (1987) found that variation in the timing of germination may be due to moisture conditions.
The differential response we observed to type of disturbance is very intriguing. C.
acanthoides had higher and faster germination in tilled plots than in clipped plots, which may be responsible for the larger fall size of rosettes. C. nutans had better performance
generally in clipped plots; apparently it was not as inhibited by the established below-
ground environment as C. acanthoides is. Seed predation may play a role in this interaction; predators such as ants and crickets appear to remove more C. acanthoides
230 seeds than C. nutans seeds (Silverman 2006). It is possible that predators were more abundant in more intact, clipped plots, leading to lower germination there.
Seed size may be partially responsible for the differential response to disturbance, although in some cases, seed mass is not strongly correlated with emergence (Goldberg
1987). C. acanthoides seeds are considerably smaller than C. nutans seeds. It has been suggested that species with smaller seeds need larger gaps for establishment
(McConnaughay and Bazzaz 1987), to overcome the fact that they produce initially smaller seedlings; perhaps species with smaller seeds also benefit from a greater amount of disturbance. The relatively high number of C. acanthoides present at the end of fall in tilled plots may be largely driven by the germination response rather than better survival, as fall survival and survival to July, given germination, were actually lower (but not significantly so) in tilled plots than in clipped plots. There may have been higher levels of herbivory on C. acanthoides in tilled plots, and on C. nutans in clipped plots (result not
significant) because their average sizes were larger in each of these environments, which
may have made them more attractive to herbivores. We feel that it might be interesting to
further examine this differential response to disturbance type.
According to Lee and Hamrick (1983), for C. nutans, germination occurs at approximately the same time in the greenhouse; in our experience, the same is true for C.
acanthoides. That the period of germination is more spread out in the field is likely due to the environments in which germination occurs (Lee and Hamrick 1983). For C.
acanthoides, the longest time to germination was experienced in unwatered, tilled plots, and for C. nutans, this occurred in large microsites that were tilled. These environments potentially represent some of the more stressful environments.
231 Future avenues:
A potentially useful avenue of research would be to combine the information about the influences of microsite quality on thistle germination and establishment gained from this experiment with information about the dispersal capabilities of these species, as well as the dynamics of gaps in pastures, to predict the population dynamics of these thistles. Simulation models would allow combining these three factors in order to both understand the dynamics in a field, and how management might be able to manipulate these factors to reduce thistle infestation levels. For example, it would be possible to model the effect of changing the frequency or size of gaps in a field on the thistle population sizes in subsequent years.
Feldman et al. (1994) point out that different conditions are optimal for different parts of C. acanthoides establishment: closed canopy or litter cover gave the best
protection from predation, but gaps in litter cover improve germination. Although
herbivory was not a specific focus of this study, we did find significant relationships
between herbivory and microsite disturbance type for C. acanthoides, although this could also be mediated by thistle quality rather than the characteristics of the microsite. As stated above, given the large amount of seedling herbivory we observed, it would be useful to investigate the effects of herbivory on early seedling growth more explicitly.
Few studies of microsite establishment follow individuals through to reproduction; the conditions necessary for reproduction may be more restrictive than the conditions for germination (Turnbull et al. 2000). We did follow individuals through to
flowering; however, very few individuals flowered in this experiment. It would be
possible to repeat this experiment with higher replication; however, a more reasonable
232 way to answer questions about effects of microsite beyond the seedling stage might involve either transplanting individuals to ensure sufficient replication or planting more seeds and culling extra germinants. It might also be beneficial to incorporate grazing, as this may also strongly influence thistle plant performance.
It may be useful to explicitly study the response to soil topography, which is influenced by gap characteristics. Soil topography may have a significant effect on germination and subsequent growth of seedlings. Soil topography can impact the depth to which seeds are covered with soil, and can also be very important for moisture availability. Different species, even of the same genus, may prefer different types of soil topography (Harper et al. 1965). For example, germination of C. acanthoides was
previously shown to be higher in rough soil than smooth (Feldman et al. 1994).
It is also important to consider the identity of the potential competitors.
McConnaughay and Bazzaz (1990) found that the identity of neighbors influenced
survivorship. In this study, we simply used an existing former pasture, and assumed that
the influence of the resident community was fairly uniform. Some studies of these species
have revealed specific species as having a particular influence on these Carduus species,
such as Daucus carota for C. nutans (Moore and Mulligan 1964), Lolium perenne for C.
nutans (Wardle et al. 1992) and Lotus tenuis for C. acanthoides (Laterra 1997). In fact, the use of other species’ allelopathic effects on C. nutans has been suggested as a
management strategy (Wardle et al. 1992, 1996). It might be interesting to investigate
whether there is a difference in response to microsite characteristics where the
background vegetation is of species known to specifically depress these thistles. The
effects may be different at different points in the life cycle: Wardle et al. (1992) suggest
233 that the species that most effectively suppress thistle seedling growth may not be the same as those that most inhibit germination.
Both the autocorrelation and cross-correlation in the distribution of these species within a field is positive (Chapter 5). Thus it appears that these species aggregate in favorable sites. To some extent, they may even create favorable sites; individuals may benefit from being in sites already occupied by thistles. This may be due to allelopathic influences, whereby chemicals are secreted which favor Carduus thistles and suppress
other species. There may also be inter-generational benefits, with older rosettes providing
some degree of protection to seedlings (a nurse plant effect). In a pilot study, we
observed greater germination of seeds in pots with rosettes present than in pots without rosettes present. It would be interesting to investigate the response to microsites where thistles are or were previously established.
Summer drought followed by overgrazing can lead to pastures with many bare patches, which increases C. nutans germination in the fall (Popay and Thompson 1979).
Alternation of dry and wet summers is favorable for C. nutans, as drought increases gaps
necessary for germination, but wet periods are necessary for life cycle completion (Doing
et al. 1969). Our results demonstrate that gap characteristics play an important role in
whether or not these two species can germinate and establish; these are both critical
stages in the invasion process. In order to better understand how to manage these species,
it would be useful to study the characteristics of gaps typically found in pastures, and
determine whether or not it is possible to manipulate them so that fewer thistles can
establish.
234 Acknowledgements
This research is in collaboration with my advisor, Katriona Shea. This work was supported by USDA-CSREES (Biology of Weedy and Invasive Plants) NRI grant #2002-
35320-1228 to KS and a NASA Space Grant Fellowship to ER. Zeynep Sezen and Eelke
Jongejans provided useful comments on this manuscript. Elizabeth Dlugosz, Elizabeth
Larcom, and Ingmar Rauschert were particularly helpful with the fieldwork. Justin
Robinson, John Mellon, Amber Hoover, Christina Saylor, Simone Adeshina, Laura
Warg, Laura Wentzel, Michael Claus and Andy Jálics also helped with fieldwork.
References
Bergelson, J. 1997. Competition between Two Weeds. Pages 65-70 in P. Kareiva, editor.
Exploring Ecology and Its Applications. Sinauer Associates, Sunderland.
Cahill, J. F. 2002. Interactions between root and shoot competition vary among species.
Oikos 99:101-112.
Cahill, J. F., and B. B. Casper. 2003. Canopy gaps are sites of reduced belowground plant
competition in a productive old field. Plant Ecology 164:29-36.
Crawley, M. 2002. Statistical Computing: an Introduction to Data Analysis using S+.
John Wiley and Sons, New York.
Davis, M. A., and M. Pelsor. 2001. Experimental support for a resource-based
mechanistic model of invasibility. Ecology Letters 4:421-428.
235 Desrochers, A. M., J. F. Bain, and S. I. Warwick. 1988. The biology of Canadian weeds
89. Carduus nutans L. and Carduus acanthoides L. Canadian Journal of Plant
Science 68:1053-1068.
Doing, H., E. F. Biddiscombe, and S. Knelhans. 1969. Ecology and distribution of the
Carduus nutans group (nodding thistles) in Australia. Vegetatio 17:313-351.
Feldman, S. R., J. L. Vesprini, and J. P. Lewis. 1994. Survival and establishment of
Carduus acanthoides L. Weed Research 34:265-273.
Goldberg, D. E. 1987. Seedling Colonization of Experimental Gaps in 2 Old-Field
Communities. Bulletin of the Torrey Botanical Club 114:139-148.
Goldberg, D. E., and K. L. Gross. 1988. Disturbance Regimes of Midsuccessional Old
Fields. Ecology 69:1677-1688.
Gross, K. L., and P. A. Werner. 1982. Colonizing Abilities of Biennial Plant-Species in
Relation to Ground Cover - Implications for Their Distributions in a Successional
Sere. Ecology 63:921-931.
Grubb, P. J. 1977. Maintenance of Species-Richness in Plant Communities - Importance
of Regeneration Niche. Biological Reviews of the Cambridge Philosophical
Society 52:107-145.
Hamrick, J. L., and J. M. Lee. 1987. Effect of Soil Surface-Topography and Litter Cover
on the Germination, Survival, and Growth of Musk Thistle (Carduus-Nutans).
American Journal of Botany 74:451-457.
Harper, J. L., J. T. Williams, and G. R. Sagar. 1965. The Behavior of Seeds in Soil .1.
The Heterogeneity of Soil Surfaces and Its Role in Determining the Establishment
of Plants from Seed. Journal of Ecology 53:273-&.
236 Holm, L., J. V. Pancho, J. P. Herberger, and D. L. Plucknett. 1979. A Geographical Atlas
of World Weeds. John Wiley and Sons, New York.
Kelly, D., H. Cameron, and J. Alex. 1988. Wind dispersal of nodding thistle seeds and
pappi. New Zealand Weed and Pest Control Conference Proceedings 41:207-209.
King, S. E., and J. B. Grace. 2000. The effects of gap size and disturbance type on
invasion of wet pine savanna by cogongrass, Imperata cylindrica (Poaceae).
American Journal of Botany 87:1279-1286.
Kok, L. T. 2001. Classical biological control of nodding and plumeless thistles.
Biological Control 21:206-213.
Laterra, P. 1997. Post-burn recovery in the flooding Pampa: Impact of an invasive
legume. Journal of Range Management 50:274-277.
Lee, J. M., and J. L. Hamrick. 1983. Demography of two natural populations of musk
thistle (Carduus nutans). Journal of Ecology 71:923-936.
McCarty, M. K., C. J. Scifres, A. L. Smith, and G. L. Horst. 1969. Germination and Early
Seedling Development of Musk and Plumeless Thistles. University of Nebraska,
Nebraska.
McConnaughay, K. D. M., and F. A. Bazzaz. 1987. The Relationship between Gap Size
and Performance of Several Colonizing Annuals. Ecology 68:411-416.
McConnaughay, K. D. M., and F. A. Bazzaz. 1990. Interactions among Colonizing
Annuals - Is There an Effect of Gap Size. Ecology 71:1941-1951.
Moore, R. J., and G. A. Mulligan. 1964. Further Studies on Natural Selection among
Hybrids of Carduus Acanthoides and Carduus Nutans. Canadian Journal of
Botany 42:1605-&.
237 Panetta, F. D., and D. A. Wardle. 1992. Gap Size and Regeneration in a New-Zealand
Dairy Pasture. Australian Journal of Ecology 17:169-175.
Popay, A. I., and A. Thompson. 1979. Some aspects of the biology of Carduus nutans in
New Zealand pastures. Proceedings of the 7th Asian-Pacific Weed Science
Society Conference:343-346.
R Development Core Team. 2005. R: A language and environment for statistical
computing. in. Austria: R Foundation for Statistical Computing, Vienna.
Ross, M. A., and J. L. Harper. 1972. Occupation of Biological Space During Seedling
Establishment. Journal of Ecology 60:77-&.
Ruggiero, D. 2004. Factors Affecting the Germination of the Invasive Thistles: Carduus
nutans and C. acanthoides. Honors Thesis. The Pennsylvania State University,
University Park.
Silverman, E. 2006. Colonization-Competition Trade-Off and Post-Dispersal Seed
Removal of Carduus nutans and C. acanthoides. Honors Thesis. The
Pennsylvania State University, University Park.
Turnbull, L. A., M. J. Crawley, and M. Rees. 2000. Are plant populations seed-limited?
A review of seed sowing experiments. Oikos 88:225-238.
Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1995. Influence of Pasture
Forage Species on Seedling Emergence, Growth and Development of Carduus
Nutans. Journal of Applied Ecology 32:225-233.
Wardle, D. A., K. S. Nicholson, and A. Rahman. 1992. Influence of Pasture Grass and
Legume Swards on Seedling Emergence and Growth of Carduus nutans L and
Cirsium-Vulgare L. Weed Research 32:119-128.
238 Wardle, D. A., K. S. Nicholson, and A. Rahman. 1996. Use of a comparative approach to
identify allelopathic potential and relationship between allelopathy bioassays and
''competition'' experiments for ten grassland and plant species. Journal of
Chemical Ecology 22:933-948.
239
Table 7-1: Summary of germination and survival in fall 2004 and 2005 Planted 2004 Planted 2005 C. nutans C. acanthoides C. nutans C. acanthoides
Number of plots 160 160 160 160 Percentage of plots with 52% 36% 75% 61% germinations Number of seeds 640 640 640 640 planted Percentage of 22% 13% 39% 24% germinations Percentage 18% 8% 33% 18% surviving fall Percentage 13% 6% in progress in progress surviving to July
240
Table 7-2: Results of logistic regressions of germination and survival C. nutans C. acanthoides Explanatory Coefficient p- value Coefficient p- value variable Germination Intercept -0.96 -2.11 Watering -0.25 0.13 -.11 0.53 Area 0.90 <0.0001*** 0.21 0.33 Disturbance -0.21 0.20 0.99 <0.0001*** Type
Germination Intercept -1.04 -2.42 and fall Watering -0.38 0.04* -0.46 0.02* survival Area 0.95 <0.0001*** 0.49 0.04* Disturbance -0.43 0.01** 0.87 <0.0001*** Type
Fall Intercept 2.59 1.34 survival, Watering -0.57 0.07• -1.03 0.002** given Area 0.42 0.14 0.87 0.04* germination Disturbance -1.24 0.004** -0.31 0.38 Type
Survival to Intercept 0.25 -0.61 July (given Watering -0.59 0.28 -1.23 0.03* germination) Area 1.48 0.004** 2.14 0.002** Disturbance -1.15 0.009** -0.31 0.62 Type
• indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01 *** indicates significance of p≤0.001
241
Table 7-3: Analyses of average days to germination, fall size and herbivory C. nutans C. acanthoides Explanatory Coefficient p- value Coefficient p- value variable Days to Intercept 2.94 3.27 germination Watering -0.19 <0.0001*** -0.23 0.0005*** Area 0.14 0.04* 0.034 0.69 Disturbance 0.17 0.003** -0.22 0.0011*** Type
Fall size Intercept -0.08 -0.63 Watering 0.12 0.05* 0.027 0.55 Area 0.43 <0.0001*** 0.44 <0.0001*** Disturbance -0.22 0.007** 0.50 <0.0001*** Type
Herbivory Intercept -0.38 -1.82 Watering 0.46 0.11 0.17 0.65 Area -0.34 0.47 -0.40 0.53 Disturbance -0.44 0.18 1.06 0.04** Type
The average days to germination were analyzed using quasipoisson regression. Note that a negative coefficient value indicates faster germination. Fall size was log transformed and then analyzed using linear regression. Herbivory was analyzed with logistic regression. • indicates marginal significance (p≤0.10) * indicates significance of p≤0.05 ** indicates significance of p≤0.01 *** indicates significance of p≤0.001
242
C. nutans C. acanthoides 0 2. 2 1. t o l 0 . 5 . 1 p r 1 8 0. ns pe 0 o i 6 1. . at 0 n i 4 m . r 0 5 e . 0 G 2 . 0 0 0 0. 0. 2004 2005 2004 2005 35 n o i t 30 a 30 n i 25 5 m r 2 ge 20 20 o t
s 15 15 0 0 1 1 age day 5 5 Aver 0 0
2004 2005 2004 2005 Year Year
Figure 7-1: Overall germination The upper two figures show the germinations per plot: both species had higher germination in 2005. Note that the axes are on a different scale: C. nutans germination is considerably higher than C. acanthoides. The lower two figures show the average days to germination in a plot, which was very consistent between the two species. Both species took longer to germinate in 2005. The error bars shown here and in all figures are the standard errors of the means.
243
C. nutans C. acanthoides
l l 5 . a 0 f 20 g 0. n i v 4 i p<0.0001 v 0. p=0.04 r 15 . 0 d su 3 . 0 an g 10 . n 0 i t 2 a 0. n i 5 m 0 r 0. 1 e . g 0
. 0 op 0 r 0 0. P 0. 25 225 900 2500 25 225 900 2500 h 0 t 5 . . 2 1 ng e l p<0.0001 af 5 .
e p<0.0001 1 l 0 1. est 0 ng 1. o l e 5 g . 0 a 5 r . e 0 v A 0 0 0. 0. 25 225 900 2500 25 225 900 2500 Area (cm2) Area (cm2)
Figure 7-2: Effect of microsite area The upper two figures show the response in the proportion germinating and surviving in fall to microsite area; for both species, this proportion increases with increasing microsite size. The lower two figures show the fall rosette size response to microsite area, which is similar to the germination and survival response. P- values listed are from the fitted models, which include the other microsite characteristics.
244
C. nutans C. acanthoides 4 l l 0. a f g 20 n i 0. v i 3 v .
r p=0.04 0 15
. p=0.02 0 d su an 2 0. 0 ng 1 i t 0. na i m 1 . 0 05 0. ger . op 0 00 Pr 0. 0. No Yes No Yes 4 0 . 1 2. h t 2 . p=0.55 1 ng e 5 . l 0
p=0.05 . 1 1 af e l 8 0. est 0 1. 6 ng . 0 o l 4 5 0. age . 0 er 2 v . 0 A 0 0 0. 0. No Yes No Yes Watering treatment Watering treatment
Figure 7-3: Effect of the watering treatment The upper two figures show the effect of the watering treatment on the proportion germinating and surviving in the fall. For both species, significantly fewer individuals germinated and survived when exposed to a watering treatment. The lower two figures show the effect of watering treatment on the average rosette size: while both species had larger sizes when watered, this effect was not significant for either species. P-values listed are from the fitted models, which include the other microsite characteristics.
245 C. nutans C. acanthoides l 5 l 2 a 0. f g n i v i 3 20 . v r p=0.01 0.
d su p<0.0001 15 0. 20 an . 0 ng i 0 t 1 0. na i 1 m . 05 ger 0. . op 00 . 00 Pr 0 0. Clipping Tilling Clipping Tilling 0 5 . 2. 1 h p=0.007 engt
5 p<0.0001 l . 1 0 eaf . l
1 t 0 1. nges o 5 . 0 age l 5 . 0 er v A 0 0 0. 0. Clipping Tilling Clipping Tilling Disturbance type Disturbance type
Figure 7-4: Effect of disturbance type C. nutans and C. acanthoides appear to have the opposite response to the type of microsite disturbance: the upper two plots show that while C. nutans had higher germination and survival in clipped plots, C. acanthoides had higher germination and survival in tilled plots. The lower two plots show size response to disturbance type. Generally, C. nutans was larger in clipped plots, whereas C. acanthoides was larger in tilled plots. P-values listed are from the fitted models, which include the other microsite characteristics.
Chapter 8
Conclusions
The results of this thesis do not support the hypothesis that competition between
C. nutans and C. acanthoides causes interspecific spatial segregation. The models of competition examined in Chapter 3 predict persistent segregation only when both species have stronger interspecific effects than intraspecific effects. The competition experiments presented in Chapter 4 show that this does not seem to be the case for these species.
There was no evidence that interspecific effects were consistently stronger than intraspecific effects, and in fact, considerable variation was observed regardless of the density of species present. If interspecific effects are roughly equivalent to intraspecific effects, the models predict slow decay of the currently observed segregated pattern.
Regionally, C. nutans and C. acanthoides do not commonly co-occur (Allen and
Shea 2006, Chapter 5). However, it appears that at local scales, within certain areas, these species readily coexist and, in contrast to the predictions of competition theory, are more likely to be found near each other. Thus the results of the finer scale patterns of C. nutans
and C. acanthoides also do not support the idea of strong competition between these two
species.
The observed co-occurrence at local scales may arise from independent
aggregation on favorable sites. The models presented in Chapter 3 focused on
endogenous patterns, internally created by competition itself, and did not include
exogenous variation. It would also be possible to examine models which combine
247 endogenous and exogenous variation either by simulation or analytically by using spatial moment equations (Bolker 2003). An obvious extension would be to incorporate actual spatial heterogeneity in the landscape using GIS modeling methods.
In hindsight, it is apparent that direct competition between these species could not be solely responsible for the observed patterns. Both the region of overlap and the fields of co-occurrence have plentiful apparently suitable but unoccupied sites. Regionally, this may in part be due to a lack of propagules, but locally this is at least partially due to strong competition with other plant species. Both species are very sensitive to microsite characteristics (Chapter 7) and have difficulty establishing without suitable gaps in the vegetation.
The stratified dispersal of these species is likely occurring at three critical scales, all of which may be affected by human activity. Humans have clearly spread these species across the continental US; these extreme long-distance dispersal events may involve contaminated agricultural seed or transport on vehicles. On a more localized scale, contaminated hay bales are an important source of thistle seed, which likely spread a species throughout farms in a community, once an infestation arrives. At an even finer scale, the wind dispersal of seeds can spread the thistles to neighboring fields; management within a field, through reducing the number and height of thistles, may influence this pattern as well. Separating the effects of these different processes, which operate at different scales, can be quite complex. It may be possible to estimate human mediated dispersal by marking seeds or inflorescences; such methods have been used to trace long-distance seed dispersal by migratory sheep in Spain (Manzano and Malo
2006).
248 Interestingly, we did find differences in the vegetation communities between plots where thistles were present versus where thistles were absent (Chapter 6). Further research might focus in more detail on what leads to these differences: whether the thistles are unable to invade into certain areas, or whether, once present, the thistles change the plant community, or whether both of these processes occur to some extent.
However, there were no differences in vegetation at sites occupied by one Carduus
species or the other, suggesting that they either respond similarly to the existing
vegetation or impact it in the same way. Understanding how these thistles interact with
the dominant members of the communities they typically invade (e.g. roadsides and
pastures) may provide substantial insight into their current distribution.
It is possible that, once established, Carduus thistles persist in certain areas of a pasture because senescing adults create conditions favorable for the next generation. This may result from high numbers of propagules present near parent plants, or from gaps of the correct size are created at the optimal time for germination, or allelopathic effects of the decaying adults may be involved. Allelopathy is again receiving interest as an important factor in some plant-plant interactions due to the identification of the particular chemical involved ( (-)-catechin) for Centauria maculosa (Bais et al. 2003). As both
Carduus nutans and C. acanthoides have been reported to be allelopathic (Woodward
and Glenn 1983, Wardle et al. 1991), further research in this area may be useful. The role
of intra-generational interactions is also not well understood; however, the results from a
pilot study indicate that germination and survival may be higher under established
rosettes than in bare ground (E. Rauschert, unpublished data). Since there is considerable
249 variation in whether individual plants exhibit winter annual vs. biennial vs. triennial life cycles for these species, such inter-generational interactions may occur frequently.
Competition has classically been studied by examining equilibrial outcomes.
Because conditions are not always stable, however, many communities are not at equilibrium. C. nutans and C. acanthoides are pests of frequently disturbed communities.
For example, roadsides are mowed several times a year, as well as being sprayed with
herbicides at least once a year. Pastures are subject to grazing by animals as well as
disturbances created by animal movement, such as the hoof-sized gaps Panetta and
Wardle (1992) suggest are optimal for thistle germination. Additionally, farmers may
specifically target heavily infested areas of pastures when applying herbicides or extra
mowing. Competition between these two species may not be of great importance if
disturbances do not allow both species to occur at extremely high densities. Both species’
response to disturbance is an important area of research that could lead to management
practices to reduce the size of thistle infestations. For example, grazing by a combination
of sheep and goats has been suggested as a possible management approach (Holst et al.
2004).
Since the dispersal capabilities of these species are well-studied (Skarpaas et al.
2006, Skarpaas and Shea, in revision), it would be possible to combine information about
establishment probabilities in microsites (Chapter 7) with dispersal kernels, to examine the realized establishment probabilities. This would likely involve observational work focused on the distributions of gaps in pastures and how these change with time. Such a model could be used to contrast the effectiveness of management practices that influence gap availability.
250 Direct interactions are only one way in which species may affect each other; indirect interactions mediated via shared natural enemies or pollinators may also be important. Apparent competition through shared predators can have a large impact on population dynamics (Holt and Lawton 1994). C. nutans presence has been shown to
negatively impact native thistles in Nebraska through the biocontrol agent Rhinocyllus conicus, as attack of C. undulatum by R. conicus increased with increasing proximity to
C. nutans populations (Rand et al. 2004). Although we did not find evidence for strong direct competitive interactions between C. nutans and C. acanthoides, we did find
evidence suggestive of potential indirect interactions via R. conicus. In the four fields of
co-occurrence we studied, we found lower attack rates on C. acanthoides when the
density of C. nutans was higher (data not shown), possibly implying that the presence of
C. nutans confers a benefit to C. acanthoides by allowing it to escape attack. The attack rates on C. nutans did not appear to be impacted by C. acanthoides presence. We believe that this may be an interesting area for further research, particularly as R. conicus is generally better synchronized with C. nutans phenology.
It is well known that above-ground herbivory can influence the outcome of competition. Most of the research on interactions with other species has focused on aboveground interactions; very little is known about belowground associations of these species (Chapter 2). Soil organisms can help maintain coexistence of plant species, particularly for equivalent competitors (Bever 2003), if there is negative feedback from the soil community. This may be an interesting avenue for future research.
Competition between C. nutans and C. acanthoides does not seem to be particularly strong and is thus unlikely to be the cause of their segregated distribution.
251 Spread history is a likely candidate for the currently observed patterns. Although the dispersal of these species has been called “extensive,” using models combining thistle demography and dispersal abilities, Skarpaas and Shea (unpublished manuscript) predict very low spread rates of these species (3 m/year for C. acanthoides and 10 m/year for C.
nutans). Although these models likely underestimate spread in some circumstances, they
highlight the fact that spread through natural dispersal of these species would be
extremely slow. Stuckey and Forsyth (1971) emphasize how extremely slow invasion of
C. nutans was into adjacent suitable habitat. The fact that both species are currently widely distributed throughout the United States is most likely due to human mediated dispersal; they may then be only spreading very slowly from the areas to which humans accidentally introduce them. In conclusion, competition between C. nutans and C.
acanthoides does not appear to be affecting their spatial patterns, and human dispersal, as well as indirect interactions and interactions with other plants species, are likely the cause of their current distribution.
This research combines experimental, observational and modeling approaches.
The process of using these multiple methods allows us to understand what occurs when multiple invaders meet. For example, for C. nutans and C. acanthoides, we feel
particularly confident that their direct competitive interactions are not causing their
currently observed regionally segregated pattern. The models clarify the requirements for
competition to create such a pattern, the experiments demonstrate that these conditions
are not met for these species, and the observational studies at a finer scale confirm that
competition is unlikely to be strongly influencing patterns in the field. In future work, we
plan to combine the known dispersal capabilities of these species with our experimental
252 results on the effects of microsite qualities on establishment, as well as with observational information about gap distributions in pastures; this will be accomplished by using models that contrast the effect of various gap management strategies. We feel that this unified approach can be usefully applied to other questions regarding interactions between invasive species.
References
Allen, M., and K. Shea. 2006. Spatial segregation of congeneric invaders in central
Pennsylvania, USA. Biological Invasions 8:509-521.
Bais, H. P., R. Vepachedu, S. Gilroy, R. M. Callaway, and J. M. Vivanco. 2003.
Allelopathy and exotic plant invasion: From molecules and genes to species
interactions. Science 301:1377-1380.
Bever, J. D. 2003. Soil community feedback and the coexistence of competitors:
conceptual frameworks and empirical tests. New Phytologist 157:465-473.
Bolker, B. M. 2003. Combining endogenous and exogenous spatial variability in
analytical population models. Theoretical Population Biology 64:255-270.
Holst, P. J., C. J. Allan, M. H. Campbell, and A. R. Gilmour. 2004. Grazing of pasture
weeds by goats and sheep. 1. Nodding thistle (Carduus nutans). Australian
Journal of Experimental Agriculture 44:547-551.
Holt, R. D., and J. H. Lawton. 1994. The Ecological Consequences of Shared Natural
Enemies. Annual Review of Ecology and Systematics 25:495-520.
253 Manzano, P., and J. E. Malo. 2006. Extreme long-distance seed dispersal via sheep.
Frontiers in Ecology and the Environment 5:244-248.
Panetta, F. D., and D. A. Wardle. 1992. Gap Size and Regeneration in a New-Zealand
Dairy Pasture. Australian Journal of Ecology 17:169-175.
Rand, T. A., F. L. Russell, and S. M. Louda. 2004. Local- vs. landscape-scale indirect
effects of an invasive weed on native plants. Weed Technology 18:1250-1254.
Skarpaas, O., R. Auhl, and K. Shea. 2006. Environmental variability and the initiation of
dispersal: turbulence strongly increases seed release. Proceedings of the Royal
Society B-Biological Sciences 273:751-756.
Skarpaas, O., and K. Shea. in revision. Seed dispersal by wind: the relative importance of
environmental conditions and species characteristics. Journal of Ecology.
Skarpaas, O., and K. Shea. unpublished manuscript. Dispersal patterns, dispersal
mechanisms and invasion wave speeds for Carduus thistles.
Stuckey, R. L., and J. L. Forsyth. 1971. Distribution of Naturalized Carduus nutans
(Compositae) Mapped in Relation to Geology in Northwestern Ohio. The Ohio
Journal of Science 71:1-15.
Wardle, D. A., M. Ahmed, and K. S. Nicholson. 1991. Allelopathic Influence of Nodding
Thistle (Carduus-Nutans L) Seeds on Germination and Radicle Growth of Pasture
Plants. New Zealand Journal of Agricultural Research 34:185-191.
Woodward, M. D., and S. Glenn. 1983. Allelopathic Effects of Three Thistle Species.
Proceedings of the 37th Annual Meeting of the Northeastern Weed Science
Society:114.
VITA
Emily Sofia Jálics Rauschert
EDUCATION Ph.D. in Ecology (expected graduation August 2006) The Pennsylvania State University B.A. in Physics and German. (1999) Case Western Reserve University
RESEARCH EXPERIENCE Graduate Researcher, Biology Department, The Pennsylvania State University Advisor:Katriona Shea (2001-2006) Student Researcher, Zoological Insitute, Ludwig-Maximilians-Universität München Supervisor: Beate Nürnberger (1999-2001) Student Researcher, Physics Dept., Case Western Reserve University Supervisor:Walter Lambrecht (1997)
TEACHING EXPERIENCE Instructor for Biol 110: Biodiversity and Basic Concepts June-August 2006 Lectured, wrote exams, co-supervised teaching assistants, co-organized laboratory exercises Student Lecturer in Biol 11: Introductory Biology. Gave series of lectures in ecology (4/2005) Guest Lecturer in Biol 419/597C Ecological and Environmental Problem Solving. Lectured on stochastic models, optimal foraging theory, competition models and matrix models (3/14/2005, 1/19/2005, 4/7/2004 & 2/26/2003) Teaching Assistant for Biol 110 (8/2004-12/2004) and Bio 220: Populations and Communities(1/2004-5/2004 & 1/2003-5/2003) Taught laboratory exercises, including teaching scientific concepts and writing skills, designing quizzes and grading lab reports based on scientific content, data presentation and writing style
AWARDS J. Brian Horton Memorial Award in Ecology, 2006 Graduate Assistant Outstanding Teaching Award, Penn State 2006 Ecology Research Assistantship 2005, 2003 NASA Space Grant Fellowship 2004-2006 and 2002-2004 J. Ben and Helen Hill Memorial Award 2005, 2004, 2003 and 2002 Ecology Research Merit Award 2003 Braddock Fellowship, Pennsylvania State University 2001-2003 University Fellowship, Pennsylvania State University 2001-2002
PUBLICATIONS Shea, K., S. H. Roxburgh and E. S. J. Rauschert (2004). "Moving from pattern to process: coexistence mechanisms under intermediate disturbance regimes." Ecology Letters 7(6): 491-508.