Genetic and chemical diversity of the toxic herb Jacobaea vulgaris Gaertn. (syn. Senecio jacobaea L.) in Northern Germany Author Jung, Stefanie ∗ & Systematic Botany, Justus Liebig University Giessen, Germany Author Lauter, Jan Author Hartung, Nicole M. Author These, Anja Author Hamscher, Gerd Author Wissemann, Volker text Phytochemistry 2020 112235 2020-04-30 172 1 9 http://dx.doi.org/10.1016/j.phytochem.2019.112235 journal article 10.1016/j.phytochem.2019.112235 1873-3700 8294291 2.1. PA profile of J. vulgaris We coupled chromatography with high resolution mass spectrometry (LC-HR-MS) in order to detect low PAs contents and different structures of PAs. A total of 98 different PA structures were determined in 367 J. vulgaris plant samples of which 347 were located in Schleswig-Holstein and 20, respectively 2 populations, in Hesse ( Fig. 1 , Table.1 ). At the time of the study, 13 of these 98 PAs were available as reference standards and could therefore be unambiguously determined. The most abundant PA in J. vulgaris individuals was erucifoline- N -oxide (16.33%) followed by senecionine- N -oxide_senecivernine- N -oxide (13.75%) and jacobine- N -oxide (10.84%) ( Table 2 ). Fig. 2 shows the structural formula of the most abundant PAs and a comprehensive list of all identified PAs is shown in Table 5 . The average total PA content of an individual was 1032 ± 365 mg / kg corresponding to 0.1% of dry weight. The mean amount of PA between populations differed from the lowest 777 ± 237 mg /kg of dry weight (Heiligenhafen) to the highest 1666 ± 476 mg /kg of dry weight (Breklum) ( Fig. 3A ). Both locations in Hesse show similar sums of PAs when compared to the locations in Schleswig-Holstein . Population 17 (Breklum) has a higher PA content, than the remaining populations (Kruskal-Wallis, p <0.0001). However, given our hypothesis, this content might not be relevant for selection by big herbivores, as all found concentrations do not reach the level of harmfulness for big grazing mammals like cattle. For insect herbivores it is known that the effect of the concentration of PAs depends on both PA and insect species ( Macel et al., 2005 ). Nonetheless compared to other investigations in Germany , we found lower overall PA contents in dry weight ( These et al., 2013 ). Explanations for different contents of PAs are rare. A possible influence might be soil composition and nutrient availability. This was shown by Kirk et al. (2010) who found higher PA concentrations in plants growing in soils with limited nutrients. Kirk et al. (2010) also suggested that there is no selective pressure on PA content by different herbivores. However, other studies did not point towards relevant effects of locational factors including soil composition or herbivory insects on PA content ( Joosten et al., 2009 ; Van der Meijden et al., 1989 ; Vrieling and Wijk, 1994 ), indicating the need for further research in this field. PA diversity was investigated by counting the different PAs that are produced by individuals and subsequently averaged within populations ( Fig. 3B ). Individuals from population 5 (70 ± 2), 6 (71 ± 3), 15 (73 ± 3), 16 (74 ± 3), and 19 (71 ± 3) produced a greater range of different PAs than the average of all individuals hence are chemically more diverse. Population 2 (49 ± 2), 3 (51 ± 2), 12 (57 ± 3), 23 (49 ± 3), 24 (50 ± 2), 25 (51 ± 5) and 27 (57 ± 2) in contrast produced a lower range of different PAs (ANOVA, F (27, 706) = 18.86, P <0.0001). Population 15 was genetically investigated, too and shows one of the highest genetic diversity values ( Table 4 ). Table 1 Location of collection sites of J. vulgaris populations with number of individuals used for chemical analysis (sample size PA analysis), number of individuals used for genetic analysis (sample size AFLP analysis), and coordinates. Chemotype of locations including significant differences between relative abundance of erucifoline and jacobine. - = not used, ns = not significant, *p <0.05, **p <0.01, ***p <0.001.

Nr.	Location	Sample size	Sample size AFLP analysis	Lat.	Long.	Chemotype
PA analysis
1	Rauischholzhausen	11	9	50.762199	8.879012	Mixed, ns
2	Cleeberg	9	–	50.454095	8.556469	Erucifoline, **
3	Stodthagen	9	9	54.281263	10.115665	Mixed, ns
4	Goosefeld	15	–	54.428561	9.808093	Jacobine, *
5	Dazendorf	15	8	54.362362	10.944441	Erucifoline, ***
6	Neustadt	15	7	54.088625	10.798694	Mixed, ns
7	Eutin	15	–	54.133031	10.6095	Erucifoline, ***
8	Mühlenbarbek	15	53.954671	9.654996	Jacobine, *
9	Drögen Eider	15	–	54.164423	10.123386	Jacobine, *
10	Bordesholm	15	–	54.193261	10.047096	Jacobine, *
11	Preetz	15	–	54.235568	10.296812	Jacobine, *
12	Langwedel	11	–	54.214269	9.921167	Jacobine, *
13	Rastorf	8	–	54.275717	10.300127	Mixed, ns
14	Dithmarscher Geest	15	8	54.130057	9.254603	Jacobine, *
15	Heiligenhafen	15	9	54.377679	10.931665	Erucifoline, **
16	Neumünster	15	–	54.039343	9.980943	Jacobine, *
17	Breklum	15	–	54.607393	8.987367	Erucifoline, ***
18	Flensburg	15	7	54.728386	9.367222	Mixed, ns
19	Füsing	15	10	54.524689	9.637699	Jacobine, *
20	Hamburg	15	8	53.643896	9.928421	Jacobine, *
21	Rehburg	16	–	54.284511	10.288206	Mixed, ns
22	Schwarzenbek	15	–	53.553038	10.414365	Mixed, ns
23	Niebüll	15	–	54.767187	8.866052	Mixed, ns
24	Lottorf	6	–	54.453106	9.571611	Mixed, ns
25	Barkauer See	14	–	54.086987	10.641261	Jacobine, ***
26	Forstkrug	15	–	53.486318	10.733531	Mixed, ns
27	Bültsee	13	–	54.497838	9.753609	Mixed, ns

Table 2 Description of the ten most abundant PA structures detected in J. vulgaris plants in Northern Germany. Senecionine and senecivernine as well as their N -oxides were respectively quantified as sum due to coelution. SEM: standard error of the mean.

average	SEM	Minimum	Maximum
[mg/kg]	[mg/kg]	[mg/kg]
Erucifoline- N -Oxide	168.43	6.93	1.10	487.79
Senecionine- N -	141.87	5.77	8.23	671.95
Oxide_Senecivernine- N -
Oxide
Jacobine- N -Oxide	111.82	5.10	1.45	473.07
Jacobine	90.50	4.10	0.00	393.19
Erucifoline	83.03	3.90	0.94	431.43
Seneciphylline- N -Oxide	82.33	4.44	2.48	475.52
Seneciphylline	64.05	3.36	1.72	457.44
Retrorsine- N -Oxide	54.34	1.84	6.58	204.07
Senecionine_Senecivernine	48.91	2.45	2.53	344.73
Retrorsine	19.43	0,71	1.80	97.72

Fig. 2. Structural formula of the 6 most abundant PAs in this study. Previous studies have specified three ( Macel et al., 2004 ) respectively four chemotypes of J. vulgaris ( Witte et al., 1992 ) . According to the dominant PAs, a jacobine type , an erucifoline type , a senecionine type and a mixed type were assigned. We found mainly individuals with both PA jacobine as well as erucifoline. However, ANOVA revealed that averaged trough all populations, 11 populations contained a significantly higher relative share of jacobine. In contrast, 5 populations produced more erucifoline ( Fig. 3C ), confirming Marcel at al. (2004) who postulated that erucifoline chemotypes are not only restricted to South-East Europe. Moreover, one population (10) contained no erucifoline and 5 populations contained no jacobine at all (2, 5, 7, 15, and 17) or less than 2% ( Fig. 4 ). Remarkably individuals from Population 3 (Stodhagen) produced neither much of jacobine nor erucifoline but plenty seneciphylline and its N -oxide. However, our data regarding PA composition widely spread so that individuals from the same population produce considerably different amounts of different PAs. As we cannot figure out any consistency or dominant PA profile we suggest that there might not be a direct evolutionary constraint as a selective pressure on PA composition in J. vulgaris at all. At least for specialist herbivores other studies confirm that they do not put any selective pressure on PA composition ( Macel et al., 2002 ; Macel and Vrieling, 2003 ; Vrieling and Boer, 1999 ). In contrast, the same authors showed that different PAs had different effects on generalist and specialist herbivores and thus hypothesized that herbivores could play a role in the evolution of PA diversity ( Macel et al., 2005 ; Macel and Klinkhamer, 2010 ). Furthermore, soil constitution and its content of microorganisms can affect PA composition ( Joosten et al., 2009 ). Although diversity within populations does exist, averaged values of our data do not support any directed evolution, therefore we interpret the overall diversity of PAs in J. vulgaris , lacking specific chemotypes to be the result of a panmictic metapopulation with no directing selection on the geographic range of our study.