Abstract

INTRODUCTION

Around 1 % of angiosperm species have evolved to abstract resources from other plants in the form of root parasitism (Press and Graves, 1995). Compared with stem parasites that grow upon other plants above ground, identifying the hosts of root parasites is not straightforward in the field (Piehl, 1963; Musselman and Mann, 1977; Gibson and Watkinson, 1989). The Santalales is primarily composed of hemi- and holoparasitic plants that have a variety of life forms. However, apart from stem parasites in the Viscaceae, Loranthaceae and Misodendraceae, host associations in the remaining root parasites are little understood.

Thesium (Santalaceae) is a genus of herbaceous and woody root hemiparasites that are widely distributed in temperate and tropical regions of the Old World (Pilger, 1935). Some species are well-known agricultural weeds. The physiology of Thesium has thus been studied in an agronomic context; for example, mannitol metabolism of Thesium has been particularly well studied for the purpose of controlling its weedy habit (Fer et al., 1993; Simier et al., 1993, 1994, 1998; Williamson et al., 2002). However, very little is known about its host range and selectivity in wild populations. For example, Thesium chinense is thought to parasitize species of Poaceae (Numata and Yoshizawa, 1975), but information is based on a limited number of observations and circumstantial evidence that these plants usually occur near plants from the family Poaceae. The situation is more or less the same in other members of Thesium, but the genus as a whole seems to be capable of using a wide range of angiosperms, including Themeda and Poa (Poaceae; Scarlett et al., 2003), Galium (Rubiaceae; Renaudin et al., 1981), barley and onion (Abu-Irmaileh, 1980) and grape (Dasgupta, 1988). Therefore, it is possible that each Thesium species parasitizes several host species within a population while having some preference for particular groups of hosts (e.g. Poaceae). Hence, it is necessary to examine root associations between Thesium and potential host plants in a given population and to quantify the strength of parasitism in order to understand the host range and post-attachment selectivity of Thesium. Knowledge of the parasitic life history not only provides basic biological information, but also contributes to a better understanding of how plants of the Santalales have evolved the current diversity of parasitic life forms.

The host range and selectivity of the root hemiparasite T. chinense were examined in a wild population situated on a riverside in central Japan. First, a vegetation survey was conducted to determine whether Thesium plants are significantly more likely to occur in association with particular plant species. Plants of Thesium as well as other root parasites take up host resources through a nodule-like structure called a haustorium. The abundance of haustoria in roots of all potential hosts was therefore directly examined and quantified and the host range and selectivity of T. chinense determined. This is the first quantitative analysis of host range and selectivity in root hemiparasitic plants of Santalales.

MATERIALS AND METHODS

Study plant

Thesium chinense (Santalaceae) is a perennial herb (Fig. 1) that is usually erect, but occasionally prostrate, and has branched stems up to 60 cm in length. The plant commonly occurs in disturbed habitats such as grasslands or riversides and quickly regenerates after fire. It has narrow, linear leaves, and the flowers are cylindrical, greenish-white, and approx. 2 mm long. Thesium chinense, in common with other root and stem parasites, takes up water and nutrients from its host by means of a specialized structure known as a haustorium (Fig. 1), which provides a physical as well as physiological bridge between parasite and host (Kuijt, 1969). The haustorium constitutes a hyaline body (the structure rich in nuclei involved in resource translocation and processing; Riopel and Timko, 1995), a penetration peg (the projection that enters host tissue; Tennakoon and Cameron, 2006) and a xylem element (the channel of nutrients and water to be absorbed from the host). Field studies were conducted by the riverside of the Kizu River, Kyoto Prefecture, Japan, during November and December 2007. During the study, T. chinense was mostly fruiting, but some plants were still in flower.

Open in a separate window

Fig. 1.

(A) Riverside vegetation with Thesium chinense. Tc, Thesium chinense; Ac, Artemisia capillaris; Ec, Eragrostis curvula; Gv, Galium verum. (B) Thesium chinense and its host (Artemisia princeps). (C) Root of T. chinense (white) entangled in the root of its host (Artemisia capillaris; dark brown). (D) Haustorial connection of T. chinense and host (Eragrostis curvula) roots. Scale bar = 1 mm.

RESULTS

Examination of haustorial connections

The direct examination of haustorial connections revealed 22 species belonging to 11 families as hosts of T. chinense (Table 1; including four species that were found associated with T. chinense in preliminary observations: Artemisia princeps, Diodia teres, Sporobolus fertilis and Sedum bulbiferum). The observed and expected numbers of haustoria were significantly different (χ² test, P < 0·0001; Table 2), suggesting that there is post-attachment selectivity. Significant differences were also found when data were analysed according to host plant families (P < 0·0001; Table 3). Based on χ² values, Andropogon virginicus was a highly preferred host, whereas Dianthus superbus and Potentilla chinensis were less preferred (Table 2). In the same manner, Poaceae was highly preferred, whereas Caryophyllaceae and Rosaceae were hardly parasitized (Table 3). However, it should be noted that the observed/expected numbers of haustoria were too small in some species/families to confidently infer the significance of preference.

Table 2.

Observed and expected numbers of Thesium haustoria on the roots of various host plants

Open in a separate window

Table 3.

Observed and expected numbers of Thesium haustoria in various host-plant families

	Number of haustoria
Family	Observed	Expected	χ2
Fabaceae	1155	1238·8	5·7
Poaceae	2184	1021·8	1321·9
Asteraceae	367	505·7	38·0
Caryophyllaceae	86	465·3	309·2
Rosaceae	14	419·0	391·5
Cyperaceae	42	195·7	120·7
Polygonaceae	167	149·3	2·1
Rubiaceae	25	49·3	12·0
Oxialidaceae	9	22·9	8·5
Violaceae	30	9·8	41·8
Total	4079	4079	2251·3

Open in a separate window

The expected numbers for Valerianaceae and Crassulaceae were less than five and were thus excluded from the calculation of χ². Significant chi-squares (χ² > 16·92, P < 0·05) are in bold.

Haustorium size

The majority of haustoria were <1 mm in most of the species examined. However, the haustorium size differed significantly among species (Kruskal–Wallis test, P < 0·0001; Table 4). The haustoria formed on Lespedeza juncea were significantly larger than those formed on Andropogon virginicus, Cymbopogon tortilis, Eragrostis curvula, Agrostis sp., Rumex acetosella and Dianthus superbus (Scheffe's test, P < 0·05). In addition, Pueraria lobata had significantly larger haustoria than did Rumex acetosella. Similarly, the size of haustoria in Fabaceae was significantly larger than that in other families, expect Rosaceae and Oxalidaceae. Asteraceae had significantly larger haustoria than did Polygonaceae.

Table 4.

Size distributions of Thesium haustoria found on the roots of each host species

		Size range (mm)
Species no.	Species	<1·0	1·0–1·5	1·5–2·0	2·0–2·5	2·5–3·0	>3·0	Significance
1	Lespedeza juncea	548	259	106	40	3	0	2, 3, 4, 5, 8, 9
2	Andropogon virginicus	659	136	41	7	1	0	1
3	Cymbopogon tortilis	410	111	18	8	2	1	1
4	Eragrostis curvula	368	75	4	1	0	0	1
5	Agrostis sp.	259	54	5	0	0	0	1
6	Artemisia capillaris	214	69	23	1	0	0
7	Pueraria lobata	124	37	17	12	3	0	8
8	Rumex acetosella	149	16	2	0	0	0	1,7
9	Dianthus superbus	74	11	1	0	0	0	1
10	Erigeron annuus	48	12	0	0	0	0
11	Praecoces sp.	38	4	0	0	0	0
12	Viola mandshurica	28	2	0	0	0	0
13	Galium verum	24	1	0	0	0	0
14	Briza maxima	18	0	0	0	0	0
15	Potentilla chinensis	12	2	0	0	0	0
16	Oxalis corniculata	9	0	0	0	0	0
17	Poaceae sp.	5	0	0	0	0	0
18	Fabaceae sp.	4	0	0	0	0	0
19	Vicia sepium	1	1	0	0	0	0
20	Agropyron sp.	1	0	0	0	0	0

Open in a separate window

The size distributions were significantly different among species (Kruskal–Wallis test, P < 0·0001). Pairs of species that had significant differences in the sizes of Thesium haustoria (Scheffe's test, P < 0·05) are indicated by species number.

DISCUSSION

For many root parasites, host range and selectivity in wild populations are poorly known and little studied because this requires a careful and extensive excavation of the root. Consequently, studies have concentrated on examination of parasite performance in pots with different host species as an alternative to excavation study (Malcom, 1966). However, pot-based studies are not suitable for a full understanding of host range and may yield misleading predictions regarding the pattern of host use in the wild (Marvier and Smith, 1997). Therefore, the present study was undertaken in order to explore the pattern of host association in T. chinense in a wild population. This is the first quantitative analysis of host associations in root hemiparasitic plants of Santalales.

The association analysis indicated that the majority of plants in the study population had neither positive nor negative associations with T. chinense. However, it should be noted that the sample sizes were too small for most species to make meaningful comparisons (Table 1), and it is possible that there is a hidden preference that it was not possible to detect. Among the eight species tested for significance, Lespedeza juncea and Eragrostis curvula occurred significantly more often in plots with T. chinense. There was also a significant negative association between Diodia teres and T. chinense, but direct examination of the roots indicated that Diodia teres is in fact parasitized, suggesting that our indirect method does not correctly reflect the actual patterns of host–parasite association in T. chinense. One of the positively associated species, Eragrostis curvula, belongs to the Poaceae; thus the present results confirm previous observations that T. chinense often occurs in proximity to grasses (Numata and Yoshizawa, 1975). The present study was conducted in natural grassland on a flood-prone riverside of the Kizu River, where many indigenous endangered plant species are found. However, the habitat has recently been invaded by various alien plant species, some of which are also parasitized by T. chinense, e.g. Eragrostis curvula, Diodia teres and Andropogon virginicus, further suggesting that there is limited specialization.

Direct observation of the roots revealed a previously unsuspected diversity of host plants for T. chinense. Overall, 57·9 % of the plant species that occurred in the population were parasitized by T. chinense (Table 1) and they belonged to 11 different families encompassing a broad range of angiosperms, suggesting that specialization to particular host taxonomic groups has not occurred in T. chinense. However, an analysis of hosts indicated that species such as Andropogon virginicus are strongly preferred and that the Poaceae had considerably more haustoria than expected from their root biomass. One factor that may influence such results is haustorium size because some species may have more but smaller haustoria than other species. The analysis of haustorium size in fact suggested that those that formed on Lespedeza juncea and Pueraria lobata were larger than those in some other species. However, haustorium size is generally very similar among host species, indicating that the strong preference for Andropogon virginicus and Poaceae plants is not associated with haustorium size

The two species that had larger haustoria, i.e. Lespedeza juncea and Pueraria lobata, both belong to the Fabaceae, the family that had the largest overall haustorium size (Table 5). The reason for the large haustoria sizes is unknown, but because these plants fix atmospheric nitrogen they may be of higher nutritional value, allowing haustoria to reach larger sizes. Thus, although these plants were not the preferred hosts as judged by the number of haustoria, they may be suitable hosts for T. chinense. However, Jiang et al. (2008) showed that in the root hemiparasitic Rhinanthus minor nitrogen fixation by the host legume is of no benefit to the parasite. It should therefore be tested explicitly whether T. chinense grown on legumes and on non-legumes differ in any significant ways that affect its growth performance, and whether such a difference, if any, is attributable to nitrogen fixation (Gibson and Watkinson, 1991; Marvier, 1996; Matthies, 1996).

Table 5.

Size distributions of Thesium haustoria for each host-plant family

		Size range (mm)
Family no.	Family	<1·0	1·0–1·5	1·5–2·0	2·0–2·5	2·5–3·0	>3·0	Significance
1	Fabaceae	1720	376	68	16	3	1	2
2	Poaceae	677	297	123	52	6	0	1, 3, 4, 5, 6, 7, 8
3	Asteraceae	262	81	23	1	0	0	2,4
4	Polygonaceae	149	16	2	0	0	0	2,3
5	Caryophyllaceae	74	11	1	0	0	0	2
6	Cyperaceae	38	4	0	0	0	0	2
7	Violaceae	28	2	0	0	0	0	2
8	Rubiaceae	24	1	0	0	0	0	2
9	Rosaceae	12	2	0	0	0	0
10	Oxalidaceae	9	0	0	0	0	0

Open in a separate window

The size distributions were significantly different among families (Kruskal–Wallis test, P < 0·0001). Pairs of families that had significant differences in the sizes of Thesium haustoria (Scheffe's test, P < 0·05) are indicated by family number.

Overall, the present results demonstrate that T. chinense uses various angiosperm hosts, some of which are more heavily infected than others. Several proximate factors are probably responsible for determining host range and selectivity in T. chinense, including the availability of chemical stimulants, strength of host defence, and level of osmotic pressure. Because species of Thesium are known to germinate in the absence of host chemical stimulants (Fer et al., 1993, 1994), such a signal is probably not required for seed germination in T. chinense. However, chemical stimulants are involved in haustorial formation in root parasites of Orobanchaceae (Musselman, 1980; Westwood, 2000; Yoder, 2001; Bouwmeester et al., 2003). Thus, it is possible that the availability of such signals may determine the success of root infection in Thesium. The strength of host defences such as induced lignification may also affect host quality. For example, some parasitic plants such as Rhinanthus minor (Cameron et al., 2006; Cameron and Seel, 2007; Rumer et al., 2007) and Orobanche crenata (Perez-de-Luque et al., 2005) are more likely to establish successful haustorial formation with less-defended plants. In addition, parasitic plants usually have higher root osmotic pressure than do their hosts, thereby facilitating water movement from host to parasite (Harris and Lawrence, 1916; Gworgwor and Weber, 1991; Simier et al., 1993; Williamson et al., 2002). Thus, it is possible that host plants vary greatly in these physiological attributes, which may be important for determining the pattern of host use in T. chinense.

Overall, the finding that T. chinense parasitizes an array of angiosperm hosts has broad general implication for patterns of host use in root parasites of Santalales. Although the present data suggest that T. chinense has strong host selectivity, it has obviously not become specialized to a limited number of preferred hosts. This is in marked contrast with some holoparasites such as Rafflesia, which has an extremely narrow host range (Ismail, 1988). Other members of Santalales are also known to use a broad range of angiosperms; for example, Osyris alba uses 23 host species belonging to 14 families (Jamal, 2006). Nevertheless, the host ranges of many Santalales species are still little understood, and many reported cases of host associations are probably fragmentary. For example, the holoparasitic Balanophora tobiracola (Balanophoraceae) is known to parasitize species of Pittosporum (Pittosporaceae), Ligustrum (Oleaceae), Eurya (Theaceae), Celastrus (Celastraceae) and Rhaphiolepis (Rosaceae; Akuzawa, 1982; Kawakita and Kato, 2002), but considering that these hosts belong to divergent families and are often dominant trees in the vegetation, it is possible that other plants are also used by B. tobiracola. Thus, the host ranges of Santalales species (especially the hemiparasitic species) are probably much broader than currently known, as evidenced by the present results for T. chinense, which was thought to parasitize only grasses. The evolutionary significance of a wide host range is yet unknown, but may be related to the generally large seed size in Santalales (Moles et al., 2005), which limits the number of seeds produced per plant and hence the chance of arriving at preferred hosts. Further studies of the life history and host associations in other members of the Santalales should broaden the perspectives on patterns of parasitic evolution in this intriguing plant lineage. Moreover, pot-based comparisons of performance on different hosts, or histological investigations of haustorial anatomy would provide further insights into the ecology of parasitic life style in plants of the Santalales.

ACKNOWLEDGEMENTS

LITERATURE CITED