Pyrrolizidine alkaloid variation in Senecio vulgaris populations from native and invasive ranges

Introduction

An alien invasive plant species is a species that expands its natural range with facilitation from intentional or non-intentional human activities, tending to hazard biodiversity, ecosystem services and human well-being in its new range (Vilà & Hulme, 2017). Many hypotheses have been proposed to explain the invasive mechanisms of alien invasive plants (Catford, Jansson & Nilsson, 2009). Some explanations focus on plant-herbivore interactions and the role of plant defense. For instance, the “Enemy Release Hypothesis” (ERH) states that in a new range, introduced plants may leave behind their specialist herbivores and gain a rapid increase in distribution and abundance (Keane & Crawley, 2002). Loss of enemies leads to lower defense levels as plants allocate fewer resources to defense and more to growth, as according to the “Evolution of Increased Competitive Ability” (EICA) hypothesis (Blossey & Notzold, 1995). The “Shift Defense Hypothesis” (SDH) argues that invasive plants decrease the level of quantitative defense but increase their qualitative defense, as invasive plants still face pressure from generalist herbivores even though they escaped attack from specialists (Doorduin & Vrieling, 2011; Joshi & Vrieling, 2005; Müller-Scharer, Schaffner & Steinger, 2004).

Chemical defense in plants can be divided into qualitative defense and quantitative defense in relation to their effect on herbivores. Plant secondary metabolites (SMs) involved in qualitative defense are toxic to many herbivores and not very costly to produce. Those involved in quantitative defense are based on digestibility-reducing chemicals and more expensive to produce and to maintain due to the typically higher complexity of the molecules (Feeny, 1976; Rhoades & Cates, 1976). Specialist and generalist herbivores react in different ways to toxic SMs: generalist herbivores are deterred by high concentrations of toxic chemicals, while specialists are often adapted to these chemicals and use them as a cue to find their host plant. Thus, plants containing high concentrations of toxic chemicals suffer more from specialist herbivores (Cates, 1980). Hence, specialist and generalist herbivores inflict different selective pressures on plants, and the concentration of SMs is balanced by the opposing selective forces of specialists and generalists (“Specialist-Generalist Dilemma”, Van der Meijden, 1996).

Moreover, different plant metabolites, even from the same groups of chemicals, may have different effects on herbivores (Kleine & Mülller, 2010; Macel et al., 2005; Van Dam et al., 1995). It is assumed that plants with a more diverse and/or with higher concentrations of SMs can better protect themselves when the specialist herbivores adapted to the qualitative defense chemicals are absent. Therefore, for introduced plants variation in both concentration and composition of defense chemicals is important to defend themselves against the guild of herbivores in a new range.

Senecio and Jacobaea, possessing pyrrolizidine alkaloids (PAs) as their characteristic defense compounds, have been chosen in several studies as model species to assess the quantitative and qualitative variation in SMs in native and introduced populations. PAs act as deterrents or toxins to non-adapted herbivores and pathogens. However, specialist herbivores that are adapted to PAs can utilize them from host plants for their own benefit, such as for a food cue and oviposition (Joosten & Van Veen, 2011; Macel, 2011; Trigo, 2011). Higher concentrations of PAs have been found in invasive rather than native populations of Jacobaea vulgaris (syn. Senecio jacobaea; Joshi & Vrieling, 2005; Lin, Klinkhamer & Vrieling, 2015), and invasive Senecio pterophorus was found to have a higher concentration of PAs than its conspecific relatives (Caño et al., 2009; Castells, Mulder & Perez-Trujillo, 2014). Beside PAs from Senecio and Jacobaea, more than 350 PAs have been identified in an estimated 6,000 plants in the Boraginaceae, Asteraceae, and Leguminosae families (Stegelmeier et al., 1999). In this study, we selected Senecio vulgaris (common groundsel, Senecioneae: Asteraceae) as a model organism for the comparison of quantitative and qualitative PA variation between native and invasive populations. S. vulgaris, a cosmopolitan weed in temperate regions, probably originated from southern Europe (Kadereit, 1984), and has spread to America, North Africa, Asia, Australia and New Zealand in the 18th century (Robinson et al., 2003). The occurrence of S. vulgaris was first recorded in China in the 19th century, and it is nowadays mainly distributed in northeastern and southwestern China (Li & Xie, 2002; Xu et al., 2012). S. vulgaris plants of some European and Canadian populations contain high amounts (>0.6 mg/g fresh weight) of PAs (Von Borstel, Witte & Hartmann, 1989). Handley et al. (2008) investigated the invasive mechanisms of this species with respect to the interaction between plants and pathogens and the outcomes did not support the EICA hypothesis. Zhu et al. (2017) found that although S. vulgari s might have been introduced into China on multiple occasions, the Chinese populations contained smaller genetic diversity compared to European populations.

In this study, S. vulgaris plants from seeds collected from 6 native (Europe) and 6 invasive (China) populations were grown under identical conditions in a greenhouse. PAs were extracted from the roots and shoots of harvested S. vulgaris plants and measured using Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS). According to the SDH, invasive plants tend to evolve higher levels of qualitative defense chemicals. Hence, we hypothesized that plants from invasive S. vulgaris populations would produce higher concentrations of PAs than those from native ranges. We also compared PA profiles in the native and invasive populations.

Materials and Methods

Results

PA diversity

Of the 21 PAs reported from S. vulgaris in the literature, 16 PAs were included in the mass spectrometric method and detected in our samples (Fig. 1). An additional eight putative PA N-oxides, with unknown identity were detected, of which it could be ascertained, based on their protonated molecular mass, fragmentation spectra and retention times, that they were different from the 21 PAs reported previously (Table S1). In many cases, both forms of PAs (tertiary amine and N-oxide) were detected in at least part of the samples. Exceptions were senecivernine and usaramine N-oxide that were detected in a number of samples, but their counterparts senecivernine N-oxide and usaramine were below the limit of detection in all samples. Similarly, no tertiary amine counterparts of the 8 unknown PA N-oxides could be identified. Thus, in total 22 PAs were detected (Table 3).

Senecionine, integerrimine, seneciphylline, and their respective N-oxides were present in the roots and shoots of all plants and all populations. Spartioidine, retrorsine and their respective N-oxides were found in all populations and in more than 90% of the individual root and shoot samples. Riddelliine N-oxide was detected in ten populations (83%), while senecivernine was detected in five populations (42%). Two PAs, riddelliine and usaramine N-oxide, were rarely found; usaramine N-oxide was only detected in the root and shoot extracts of one plant from a native population located in Potsdam, Germany, while riddelliine was found in two plants originating from Potsdam and in one plant from an invasive population in Shennongjia, China. Six of the eight unidentified PA N-oxides were found in all populations, and three of them (unk 3-5) were found in more than 90% of the shoot and root samples (Table 3, Tables S2–S3).

Variation in PA composition

Overall PA diversity (H′) as well as evenness (J′) was higher in shoots than in roots, and lower in the invasive populations than in native ones (Fig. 2). Differences in PA composition were significant between organs (shoots and roots) and among populations (two factor Adonis test; organ: df = 1, r² = 0.41, p = 0.005; populations: df = 11, r² = 0.19, p = 0.005; Fig. 3). Senecionine N-oxide was the dominant component of the PA profile in the roots, followed by integerrimine N-oxide, senecionine and seneciphylline N-oxide (Fig. 4A). In the shoots, the four above mentioned PAs were also prevalent, in combination with retrorsine N-oxide and two unidentified PA N-oxides (unk 4 and 5, Fig. 4B). The ratio between the concentration of senecionine and that of seneciphylline (Sn/Sp ratio, including the free base and N-oxide forms of these PAs) for individual plants ranged from 0.56 to 6.87; the ratio at population level was greater than 1 and differed significantly between populations (ANOVA test: df = 11 and 43, F = 9.7, P < 0.001, Fig. 5).

Generally, the relative abundance of individual PAs was significantly different among populations (Table 4). However, the clustering of the populations did not show any geographically related pattern (Fig. 6).

Table 4:

Results of the nested ANOVA tests of difference among Senecio vulgaris populations and ranges (native or invasive) for 13 selected pyrrolizidine alkaloids (PAs).

PA code	Root		Shoot
	Population (range)	Range	Population (range)	Range
Concentration of PAsa,b
Sn	1.24ns	0.17ns	1.12ns	0.37ns
Sn.ox	2.23*	<0.00ns	1.82ns	0.33ns
Ir	0.12ns	1.48ns	1.34ns	0.29ns
Ir.ox	2.23*	0.13ns	2.59*	0.69ns
Rt	2.98**	13.05**	3.26**	12.98***
Rt.ox	3.22**	14.15***	2.30**	14.2***
Sp	1.18ns	0.59ns	0.81ns	1.49ns
Sp.ox	2.23ns	1.26*	1.62ns	1.37ns
St	2.21*	2.49ns	2.30*	3.03ns
St.ox	3.14**	2.87ns	3.13**	3.32ns
Unk3	1.82ns	0.42ns	2.01ns	2.40ns
Unk4	3.12**	2.39ns	4.02**	0.49ns
Unk5	3.02**	1.20ns	2.00ns	0.10ns
Total PA concentration	2.05*	0.11ns	1.81ns	1.48ns
Relative abundance of PAsa,c
Sn	0.35ns	1.97ns	2.12*	0.004ns
Sn.ox	2.68*	6.67*	2.10*	1.64ns
Ir	2.56*	0.16ns	1.46ns	0.51ns
Ir.ox	6.25**	0.09ns	2.95**	0.004ns
Rt	2.21**	7.96**	3.66**	2.39ns
Rt.ox	3.33ns	9.99**	5.44***	3.27ns
Sp	1.01*	0.76ns	1.60ns	0.02ns
Sp.ox	6.67ns	2.83**	2.27*	1.06ns
St	1.74***	1.25ns	2.83**	0.57ns
St.ox	6.34ns	15.51**	10.93***	5.09ns
Unk3	0.73***	0.49ns	0.56ns	0.68ns
Unk4	4.13***	1.78ns	2.45*	2.67ns
Unk5	3.22**	1.39ns	2.05*	5.48ns

DOI: 10.7717/peerj.3686/table-4

Notes:

aNested ANVOA tests were conducted separately for each individual PA (or total PA concentration) from root and shoot samples. Concentration or relative abundance of PAs were used as independent variable, population nested in ranges (df = 10) and range (df = 1) as fixed factors. In total 59 individual plants were used, and they were from 6 native and 6 invasive populations. The relative abundance of the 13 selected PAs was at least 1%, averaged among all samples.

bConcentration of PAs was calculated as µg/g dry weight and log transformed for the tests.

cRelative abundance of PAs was calculated as individual PA percentage of total PA concentration and root square transformed for the tests.

Level of significance:

nsP > 0.05

*p < 0.05

**p < 0.01

***p < 0.001.

Discussion

The great asset of LC-MS/MS is that it can analyse PA tertiary amine and N-oxide forms simultaneously in a single run with high sensitivity and specificity in combination with minimal sample clean-up. However, like other mass spectrometric techniques that have evolved in recent years, such as LC-QToF-MS (Skoneczny et al., 2015) and LC-Orbitrap-MS (These et al., 2013), it requires a comprehensive set of authentic analytical standards for a full quantitative result. Furthermore, although most LC-MS techniques are capable of annotating tentative PAs–based on their fragmentation spectra and or elementary composition—to establish the chemical structure of the unknowns, additional techniques, such as NMR are required.

It has been reported that PA profiles of the aboveground parts of S. vulgaris plants comprise seneciphylline, senecionine, retrorsine and the corresponding E-geometrical isomers, spartioidine, integerrimine and usaramine (Hartmann & Zimmer, 1986; Pieters & Vlietinck, 1988). In S. vulgaris PAs are primarily produced as N-oxides in the roots, which is also the dominant form of PAs in the other parts of the plant (Hartmann & Dierich, 1998). Apart from the PAs mentioned above, riddelliine, senecivernine, platyphylline and neoplatyphylline have been reported in the aerial parts of S. vulgaris plants (Von Borstel, Witte & Hartmann, 1989; Yang et al., 2011), as well as neosenkirkine (Von Borstel, Witte & Hartmann, 1989) and othonnine (Xiong et al., 2012). The 21 PAs with identified structures detected from S. vulgaris plants in previous studies have been summarized in Table S1 and structures of most of them were shown in Fig. 1.

Of these 21 PAs reported previously, 16 PAs were included in the mass spectrometric method, most of which were detected in this study. However, due to a lack of suitable reference standards, we were unable to search for platyphylline, neoplatyphylline, neosenkirkine, or othonnine in the root and shoot extracts of S. vulgaris. However, three unidentified PA N-oxides (Unk 3-5), with the same molecular mass as retrorsine N-oxide and that could be structural isomers of the latter were present in more than 90% of the samples. In particular, one PA N-oxide (Unk 5) comprised about 10% of the total PA concentration in shoot samples, although a reliable quantification of this compound due to lack of a standard could not be made. It would be worthwhile to elucidate the structure of these three PAs and explore whether they are dominant in the PA profiles of certain S. vulgaris plants.

Some studies have found either senecionine (Hartmann & Zimmer, 1986) or seneciphylline to be dominant (Lüthy, Heim & Schlatter, 1983), while others have found both PAs to be dominant in S. vulgaris (Von Borstel, Witte & Hartmann, 1989; Brown & Molyneux, 1996). In this study, senecionine was generally present in higher concentrations than seneciphylline.

The shoots and roots of S. vulgaris plants differed in that shoots showed more divergent PA profiles and that the shoots had a lower total PA content than the roots. Although there were significant differences in PA variation between the shoots and roots, these parts were positively correlated regarding the concentrations of total PAs and most of individual PAs (Table 3). This pattern could be explained by the processes of PA synthesis and accumulation in S. vulgaris plants, as PAs are primarily produced as senecionine N-oxide in the roots, while structural transformation mainly occurs in the shoots. Usually there is little turnover of PAs once being produced and they translocate to plant tissues mainly via the phloem (Hartmann & Dierich, 1998). Similar patterns regarding differences and correlations of PAs between the roots and shoots have been found in J. vulgaris (Cheng et al., 2011; Joosten et al., 2011).

Higher PA concentrations in the belowground parts compared with the aboveground parts of S. vulgaris plants have been found in the vegetative stage. It has been reported that when the plants have produced buds, the highest PA concentrations are found in the capitula, while the stems and leaves generally contain lower total PA concentrations compared to the roots (Hartmann & Zimmer, 1986). This consistent with our finding that the total PA concentration was lower in the shoots than in the roots when the S. vulgaris plants were not yet flowering.

The indexes of PA diversity and evenness were somewhat lower for plants from invasive ranges than those from the native range (Fig. 2). This indicated that invasive S. vulgaris populations tended to produce less diverse PA profiles than the native ones. However, this trend is much weaker than observed in some other invasive species. For instance, native J. vulgaris populations expressed four chemotypes (Macel, Vrieling & Klinkhamer, 2004), while in invasive J. vulgaris populations one chemotype dominated (Joshi & Vrieling, 2005). PA diversity in S. pterophorus (native to South Africa) was reduced after introduction in Europe and Australia (Castells, Mulder & Perez-Trujillo, 2014). Furthermore, invasive Tanacetum vulgare plants contained a smaller number of qualitative defense compounds than the native ones (Wolf et al., 2011).

We also found that invasive S. vulgaris populations did not produce higher concentrations of individual and of total PAs than native populations. These results did not agree with our prediction deduced from the SDH. Some studies have found that PA levels of related species significantly increased in the invaded range. For example, invasive populations of S. inaequidens, S. pterophorus and J. vulgaris all showed a significantly higher total PA concentration than their native conspecifics (Joshi & Vrieling, 2005; Caño et al., 2009; Castells, Mulder & Perez-Trujillo, 2014; Lin, Klinkhamer & Vrieling, 2015). Some other invasive species appear to have evolved towards decreased chemical defense levels but they may have developed other compensatory mechanisms that contribute to their invasion success. For instance, invasive genotypes of Sapium sebiferum evolved a reduced defense and resistance ability, but were more tolerant and outperformed the native genotypes under higher levels of herbivore attack (Zou et al., 2008).

The prerequisite of the EICA hypothesis and SDH is that invasive plants face a lower specialist herbivore pressure in invasive ranges. We could confirm that S. vulgaris populations in China were attacked by insect herbivores but we did not determine whether the insects were specialists or not. Although it is likely that most insects will be generalists, it is not impossible that there may be one or more specialists among them that have adapted to S. vulgaris, since S. vulgaris has a long invasive history and more than 60 congeneric species have been identified in China (Chen, 1999). Since there are significant variations between populations, a good revisiting study on the EICA hypothesis and SDH needs enough populations for a robust statistical analysis, and it is also important to describe and cluster invasive populations by analysis of their genetic structure in the different ranges; otherwise it remains difficult to determine whether the differences between native and invasive populations are the result of evolution or of pre-adaption (Pan et al., 2013; Turner, Hufbauer & Rieseberg, 2014; Siemann et al., 2016; Schrieber et al., 2016).

Taking into account the high PA levels present in S. vulgaris and the toxic effect that PAs exert on most herbivores, it reasonable to assume that PAs play an important role in the chemical defense of S. vulgaris. However, there are also other metabolites that can function as chemical defense in S. vulgaris. For instance, an oplopane sesquiterpene and jacaranone were identified from S. vulgaris (Liu, Zhang & Wang, 2010). Both compounds (or similar compounds) have a negative effect on insect feeding (Lajide, Escoubas & Mizutani, 1996; Reina et al., 2001; Xu, Zhang & Casida, 2003). It will be interesting to investigate whether the levels of other qualitative defense compounds such as oplopane sesquiterpenes and jacaranone are higher in invasive S. vulgaris populations than in native ones, as the SDH would predict. It may be advantageous to use a non-targeted analysis approach to explore for metabolites of potential significance, as was recently shown in the study of Skoneczny et al. (2017).

Conclusions

As the Senecio vulgaris plants from native and invasive ranges were grown under identical conditions, the differences in PA concentration and PA composition between ranges and between populations might thus be explained by their genetic variation. In our study the invasive S. vulgaris populations had slightly less diverse PA profiles and tended to have lower concentrations of individual PAs compared to the native populations. This finding is in contrast to the predictions of the SDH. However, the current findings should also be treated with caution given the limited number of populations sampled, the lack of background information about herbivore guilds feeding on S. vulgaris and the limited knowledge on the genetic structure of S. vulgaris populations in the different ranges. Future studies should focus on sampling a larger number of populations and screening for a wider array of plant metabolites in order to address these questions.

Supplemental Information