Enemy release does not increase performance of Cirsium arvense in New Zealand
- First Online:
- Cite this article as:
- Cripps, M.G., Edwards, G.R., Bourdôt, G.W. et al. Plant Ecol (2010) 209: 123. doi:10.1007/s11258-010-9728-7
- 133 Views
Cirsium arvense (L.) Scop. (Californian, Canada, or creeping thistle) is an exotic perennial herb indigenous to Eurasia that successfully established in New Zealand (NZ) approximately 130 years ago. Presently, C. arvense is considered one of the worst invasive weeds in NZ arable and pastoral productions systems. A mechanism commonly invoked to explain the apparent increased vigour of introduced weeds is release from natural enemies. The enemy-release hypothesis (ERH) predicts that plants in an introduced range should experience reduced herbivory, particularly from specialists, and that release from this natural enemy pressure facilitates increased plant performance in the introduced range. In 2007, surveys were carried out in 13 populations in NZ (7 in the North Island and 6 in the South Island) and in 12 populations in central Europe to quantify and compare growth characteristics of C. arvense in its native versus introduced range. Altitude and mean annual precipitation for each population were used as covariates in an attempt to explain differences or similarities in plant traits among ranges. All plant traits varied significantly among populations within a range. Shoot dry weight was greater in the South Island compared to Europe, which is in line with the prediction of increased plant performance in the introduced range; however, this was explained by environmental conditions. Contrary to expectations, the North Island was not different from Europe for all plant traits measured, and after adjustment for covariates showed decreased shoot density and dry weight compared to the native range. Therefore, environmental factors appear to be more favourable for growth of C. arvense in both the North and South Islands. In accordance with the ERH, there was significantly greater endophagous herbivory in the capitula and stems of shoots in Europe compared to both NZ ranges. In NZ, capitulum attack from Rhinocyllus conicus was found only in the North Island, and no stem-mining attack was found anywhere in NZ. Thus, although C. arvense experiences significantly reduced natural enemy pressure in both the North and South Islands of NZ there is no evidence that it benefits from this enemy release.
KeywordsEnemy release hypothesisHerbivoryPlant invasionBiogeography
Invasive plants can have severe detrimental ecological (Vitousek et al. 1997; Mooney and Hobbs 2000) and economic (Pimentel et al. 2005) impacts. A common assumption is that these detrimental effects are incurred primarily in the introduced ranges, since invasive plants are often considered more vigorous in their exotic, compared to their native range. There has been a long history of anecdotal reports of introduced plants being more vigorous in their exotic range, but relatively little quantification of these observed differences (Thébaud and Simberloff 2001). For instance, in Canterbury New Zealand early reports of invasive watercress were stated to ‘attain a size and strength quite unknown in its native country’, but no measure of size or vigour was provided (Armstrong 1879). Crawley (1987) also noted observational reports of alien plants growing larger in their introduced range and presented a comparison of plant heights from floral records that supported this claim. However, mere observational notes and data from floral records that typically report only the size range of some plant parts are insufficient to demonstrate that an introduced plant performs better in its exotic range. Conducting comparative biogeographical studies of invasive plants in their native and introduced ranges has been recognized as an important step towards understanding plant invasions (Hierro et al. 2005). A review of quantitative studies comparing plants in their native versus introduced range showed that the majority of invasive plants are more vigorous in their exotic range, although this was not always consistent (Hinz and Schwarzlaender 2004).
The mechanism most commonly invoked to explain the increased performance of invasive weeds is release from natural enemies (Maron and Vilà 2001; Keane and Crawley 2002; Mitchell and Power 2003). The enemy-release hypothesis (ERH) asserts that upon introduction to an exotic range, plants experience a decrease in natural enemy pressure that facilitates their dispersal and increased abundance. Introduced plants can experience direct fitness benefits from decreased natural enemy pressure and may also experience selection for more vigorous genotypes that invest more in growth and reproduction, and less in defence against herbivores and pathogens (Blossey and Nötzold 1995). Studies comparing herbivory and the natural enemy complexes in native and introduced ranges have found reduced herbivory, lower overall diversity of natural enemies and a shift from specialists to generalists in the introduced range of invasive plants (e.g. Goeden 1974; Memmott et al. 2000; Wolfe 2002; Hinz and Schwarzlaender 2004; Colautti et al. 2004; Genton et al. 2005; Cripps et al. 2006; Liu and Stiling 2006).
Successful classical biological control also lends pertinent evidence in support of the ERH, but does not dictate that plants are more vigorous in their introduced range due to lack of natural enemies. Although the implicit principle of the ERH has been the premise of most classical biological control programs, in most cases little is known about how an invasive plant grows in its native range (Hierro et al. 2005). In New Zealand (NZ), there is renewed interest in biological control of Cirsium arvense (L.) Scop. As part of the ongoing biological control effort against this weed in NZ, field surveys were carried out in Europe and the North and South Islands of NZ to quantitatively compare the performance of the plant and the levels of herbivory among ranges. Our purpose for conducting these surveys was twofold: first, to provide baseline data on the growth of C. arvense in NZ that could be used for subsequent assessment of the impacts of biological control agents and secondly to test assumptions of the ERH, which has implications for classical biological control. In line with the assumptions of the ERH, we hypothesized that (1) plant performance would be greater in the introduced ranges of the North and South Islands of New Zealand and (2) herbivory would be greater in the native range.
Cirsium arvense (L.) Scop. (Asteraceae) (Californian, Canada, or creeping thistle) is a perennial herb and a member of the tribe Cardueae, which comprises the plants commonly known as thistles (Bremer 1994). It is indigenous to Eurasia but has been accidently spread throughout temperate regions of the world, where it is considered one of the worst invasive weeds (Holm et al. 1977). It was first reported as an accidental introduction to NZ in 1878 (Kirk 1878) and is presently considered one of the worst weeds in NZ arable and pastoral production systems (Bourdôt and Kelly 1986; Bourdôt et al. 2007). C. arvense spreads clonally by means of creeping lateral roots and also reproduces by seeds. C. arvense is almost completely dioecious with individual plants having capitula containing either male (staminate) or female (pistillate) florets. Seed production is dependent on insect pollination and proximity to male plants. Population sex ratios range from equal male:female ratios (Lloyd and Myall 1976) to extremely female-biased sex ratios (Lalonde and Roitberg 1994). In general, few seeds are produced where male and female plants are separated by more than 50 m (Lalonde and Roitberg 1994). In New Zealand, C. arvense flowers from December to February, and seeds are produced from December to April (Webb et al. 1988). In the native range of Europe, C. arvense flowers from July to September (Clapham et al. 1987), and seeds are also produced from July to September (M. Cripps, personal observation).
There is a long history of control efforts against C. arvense, including cultural, chemical and biological methods (Donald 1990). Classical biological control has been attempted in Canada, USA and NZ. In North America, biological control of thistles in general was hampered by evidence that Rhinocyllus conicus (Frölich) was negatively impacting populations of related native thistles (Louda et al. 1997; Louda, 1999). However, in NZ, there are no native plants in the tribe Cardueae (Webb et al. 1988), enabling the recent release of two oligophagous thistle herbivores from Europe: Cassida rubiginosa Müller and Ceratapion onopordi Kirby. Previously, from 1979 to 1996, four insect herbivores [Altica carduorum Guér., Hadroplontus litura (F.), Lema cyanella (L.) and Urophora cardui (L.)] had been released in NZ for biological control of C. arvense (Julien and Griffiths 1998), but all have failed to establish (Harman et al. 1996), except for L. cyanella, which has reportedly established at one site in the North Island (Landcare Research, unpublished data). Additionally, R. conicus was released for control of Carduus nutans L. in 1973 (Julien and Griffiths 1998), and is known to also attack C. arvense (Zwölfer and Harris 1984). Other than insect herbivores, the highly specialized rust fungus, Puccinia punctiformis (Str.) Röhl., is also known to occur on C. arvense in NZ, and was present as early as 1881 (Cunningham 1927).
Populations of Cirsium arvense surveyed in Europe, New Zealand North and South Islands with corresponding 10-year mean climatic data
Mean annual temperature (°C)
Mean max temperature of warmest month (°C)
Mean min temperature of coldest month (°C)
Mean annual precipitation (mm)
NZ North Island
Bay of Plenty
Bay of Plenty
Bay of Plenty
NZ South Island
Climatic data was gathered from both ranges to assess weather conditions that could account for differences in plant growth. For the European range, weather data were collected from the National Climate Center database (http://mi3.ncdc.noaa.gov/), and for NZ, weather data were collected from the National Institute of Water and Atmospheric Research (http://cliflo.niwa.co.nz/). For each surveyed population of C. arvense, the data from the closest available weather station were used. Distance between the surveyed population and the nearest weather station ranged from 3 to 54 km. For each population, daily means were collected for each climatic metric, and 10-year means were calculated for annual mean temperature, mean temperature of the warmest month, mean temperature of the coldest month and total annual precipitation.
Data analyses occurred in three stages. First, data from individual quadrats were used to test for differences among populations within each range for shoot density, shoot dry weight, shoot height, and percent cover estimates of C. arvense, other herbaceous plants, grasses and bare ground, using one-way analysis of variance. Second, sample means for each population were used to test for differences among ranges (Europe, NZ North Island and NZ South Island) in each of these variables, again using one-way analysis of variance. The proportion of shoots attacked among ranges was analysed by using a generalized linear model (GLM) with a logit-link function, allowing for over-dispersion and assuming a binomial distribution (with the binomial total being the sum of the total number of dissected shoots). Analyses for shoot attack were done separately for capitula and stem attack, respectively. Third, analysis of covariance incorporating the 10-year mean annual precipitation and altitude for each population was used to adjust for abiotic differences which could potentially explain differences in growth patterns among ranges. Mean temperatures for each population are reported in Table 1 but not included in the analysis since lack of overlap among the ranges would confound the analysis. All analyses were conducted using GenStat (Version 11).
Latitude and all climatic variables included were significantly different among all three ranges (P < 0.001; Table 1). Precipitation was greatest in the NZ North Island followed by the NZ South Island and lowest in Europe. Altitude was significantly greater in Europe compared to the North Island, but not different from the South Island; the North and South Island were not significantly different in altitude.
Contrary to our first hypothesis, we failed to find any convincing evidence that C. arvense is more vigorous in its introduced range. Dry weight was greater in the South Island, which is in line with our prediction of increased performance in the introduced range; however, this was explained by more favourable conditions (i.e. increased precipitation in NZ) and therefore does not support the hypothesis that increased vigour is due to release from natural enemies. In the introduced range of the NZ, North Island plant performance was not different than Europe for any plant trait compared and after adjusting for differences in altitude and precipitation shoot density and dry weight were in fact lower in the North Island, which is contrary to the ERH and our first prediction. This would suggest that the similar performance of C. arvense in the North Island and Europe is due to more favourable environmental conditions for C. arvense growth in the NZ North Island.
The general lack of increased performance of C. arvense in both introduced NZ ranges is in contrast to most other similar studies, which indicate that invasive weeds are more vigorous in their introduced range (Hinz and Schwarzlaender 2004). However, only a limited number of such comparative surveys have been conducted, and it has been noted that many of these conducted to date have been rather low in quality (e.g. low number of populations surveyed, few plant traits measured, herbivory not reported, no weather data) (Hinz and Schwarzlaender 2004). Recently, more rigorous comparative surveys have produced evidence that the invasive weeds Solidago gigantea and Buddleja davidii grow more vigorously in their introduced range, and that this increased growth performance cannot be attributed to more favourable climatic conditions, which lends support for the ERH (Jakobs et al. 2004; Ebeling et al. 2008). In contrast, our data indicate that the increased dry weight of C. arvense in the South Island can be explained by more favourable abiotic conditions and even the similar performance of C. arvense in the North Island compared to Europe can be attributed to more favourable conditions without which the plant’s growth would be reduced to levels significantly less than in its native range. The general decreased performance of C. arvense in the North Island might be due to other environmental factors associated with lower latitudes. This is supported by the fact that C. arvense is not a problematic plant at latitudes lower than 37° North or South in North America (Moore 1975) and Australia (Amor and Harris 1974), respectively. Thus, our data also indicate that the plant is a more vigorous weed in higher latitude temperate conditions, and additionally show that this is regardless of whether it is in the introduced or native range.
The lack of evidence for increased vigour of an invasive plant in its introduced range is not completely unique. Some studies have found invasive plants to have equal or reduced growth in their introduced range depending on the plant trait measured (Lonsdale and Segura 1987; Sheppard et al. 1996; Erfmeier and Bruelheide 2004), or reduced plant performance in the introduced range was considered to be related to nutrient conditions of the particular habitat (Edwards et al. 1998). In the case of broom, Cytisus scoparius, Paynter et al. (2003) found no difference in plant size and growth rate between its native and introduced ranges, but a significantly greater population density in the introduced range. It was suggested that the similar size and growth rate between ranges might be a result of increased intraspecific competition due to the higher population densities in the introduced range, and that this may have concealed effects of release from natural enemies that might have been evident at lower densities. In our study, density of C. arvense was not different among ranges, and therefore there was no reason to believe that intraspecific competition was significantly different. Additionally, there were no differences in the percentage cover of other herbaceous plants and grasses, which might suggest that interspecific competition was also not different. Therefore, it is unlikely that competition may have concealed effects of release from natural enemies on individual shoot growth. However, there was significantly more bare ground in the European range, which would suggest greater disturbance that could allow for increased growth and spread of the plant. Interspecific competition from grasses is known to have a strong adverse effect on the growth of C. arvense, and microsite availability, such as bare ground, has been shown to be critical for new shoot recruitment (Edwards et al. 2000). Nevertheless, when percent cover of other herbs, grasses and bare ground were used as covariates there was no change in the density, height or dry weight of C. arvense among ranges.
In accordance with our second hypothesis, herbivory was significantly greater in the native range. The phytophagous insect community on C. arvense and high proportions of shoot attack are well known from its native range in Europe (Zwölfer 1965; Schröder 1980; Freese 1994). Assessing the degree of specialized attack in NZ was considered important since biological control agents have been released for this weed in NZ in the past (Julien and Griffiths 1998), and the possibility of host shifts by indigenous herbivores onto the introduced plant cannot be discounted (e.g. Olckers and Hulley 1991; Creed and Sheldon 1995; Cripps et al. 2006). The surveys conducted here confirm the lack of agent establishment on C. arvense noted by other authors (Harman et al. 1996). The absence of host shifts by indigenous insect herbivores is not surprising since no native plants occur in the Cardueae tribe in NZ. Previously, the degree of attack by R. conicus on C. arvense in NZ was unknown. Here we show that R. conicus, originally released for control of C. nutans, is also commonly encountered on C. arvense, at least in the North Island of NZ. Interestingly, R. conicus was not encountered in any of the surveyed populations in the South Island of NZ, although adult weevils have been observed in that region (M. Cripps, personal observation). Similarly, Fenner and Lee (2001) found no endophagous herbivore attack in the capitula of C. arvense from a survey carried out in 1998 in the South Island of NZ. The rust pathogen, P. punctiformis, is known to cause severe detrimental effects, often killing shoots before flowering (Watson and Keogh 1980; Thomas et al. 1994), and the proportion of shoots attacked by this pathogen has been shown to be similar in both NZ and Europe (Cripps et al. 2009).
The fact that endophagous herbivory was much greater in the native range, but that plant performance was generally not different, or explained by environmental differences among ranges could indicate that C. arvense is not influenced by endophagous herbivory. Another possibility is that the relatively low natural enemy pressure we have found in NZ is sufficient to maintain plant growth at similar levels to the native range. Successful biological control is often attributed to a single best agent (Denoth et al. 2002). Thus, it is possible that the specialized natural enemies present in NZ (i.e. R. conicus and P. punctiformis) are sufficient to decrease the plant performance to levels similar to the native range. However, this is unlikely, since capitulum feeders have been shown to have negligible influence on the population dynamics of C. arvense (Edwards et al. 2000), which is further supported by data showing that seedlings do not contribute to the growth of established C. arvense populations (Bourdôt et al. 2006). And, although P. punctiformis can kill individual shoots, population effects are minimal, even with artificial manipulation (Frantzen 1994; Kluth et al. 2003). In terms of generalist insect herbivores, only five have been reported in association with C. arvense in NZ (Spiller and Wise 1982), and the only insects commonly encountered on C. arvense in NZ were pollinators (M. Cripps, personal observation). Therefore, it is also unlikely that generalist insect herbivores have an impact on C. arvense in NZ.
In conclusion, addressing the hypothesis of increased plant performance in the introduced range compared to the native range is important to gain a more objective perspective on invasive weeds, and to better inform management decisions, particularly biological control. In accordance with the ERH, it is evident that C. arvense is released from natural enemy pressure in NZ. However, this release from natural enemies has not resulted in increased performance of C. arvense and does not explain the case of increased dry weight in the South Island. As a result, the data presented here do not offer promising support for the classical biological control of C. arvense; however, no strong conclusion can be made, since any introduced agent would in turn be released from its own specialized predators and parasitoids, which could allow it to have an unexpected impact on the plant. This study was the first step in our biogeographic comparison of C. arvense in its native and introduced ranges. Further experimental study was also carried out to assess population demographics in the native and introduced range when herbivores and pathogens were excluded, or not. This will be reported on subsequently and will offer more insights into the influence of natural enemies on the population dynamics of C. arvense.
This research was funded by the Bio-Protection Research Centre, Lincoln University, New Zealand.