Isotopic evidence of partial mycoheterotrophy in the Gentianaceae: Bartonia virginica and Obolaria virginica as case studies^†

Abstract

• Premise of the study: An estimated 10% of plant species have evolved to steal C from their symbiotic fungal partners (mycoheterotrophy), and while physiological evidence for full and partial mycoheterotrophy is well developed in the Orchidaceae and Ericaceae, it is lacking for the majority of other mycoheterotrophic taxa. The family Gentianaceae not only contains several lineages of achlorophyllous mycoheterotrophs, but also contains species that are putative partially mycoheterotrophic. The North American genera Bartonia and Obolaria (Gentianaceae) are green but have leaves reduced to scales or foliose bracts and so have ambiguous mycoheterotrophic status.

• Methods: We investigated the natural abundance ¹³C and ¹⁵N profiles of both genera along with total N and chlorophyll content and investigated mycorrhizal infection using light microscopy.

• Key results: The shoots of B. virginica were significantly more enriched in ¹⁵N than the surrounding vegetation but not in ¹³C. In contrast, the shoots of O. virginica are not enriched in ¹⁵N compared to the surrounding vegetation but were significantly enriched in ¹³C. Total N concentrations were significantly higher than the surrounding vegetation in B. virginica, while the collaroid roots of both species were infected by arbuscular mycorrhizal fungi.

• Conclusions: This microscopic evidence coupled with the natural abundance stable isotope profiles strongly suggests that both species are partially mycoheterotrophic. However, differences in the root–shoot stable isotopic patterns relative to surrounding vegetation between B. virginica and O. virginica are suggestive of the utilization of different physiological pathways or extent of commitment to mycoheterotrophic C gain.

For over 150 years, we have recognized that some plants have evolved to steal C from their symbiotic mycorrhizal fungal partners (mycoheterotrophy). However, for most of this time, the mycoheterotrophic strategy was considered rare, limited to unusual botanical curiosities. More recently, mycoheterotrophy has been shown to be far more common than first thought, with ∼80 genera representing ∼10% of all plant species relying on fungus-derived C at some point during their life cycle (14).

Most mycoheterotrophic plants begin their life cycle with the production of seeds that are so small that they do not have sufficient reserves to establish underground unaided. Instead, they exploit soil fungi that supply the developing seedling with all of their C and much of their mineral nutrient requirements (12). Most plants that depend on mycoheterotrophy for establishment develop green leaves and are partially or fully autotrophic once mature (20). This is the case with the majority of the ∼34000 species of orchid and in an estimated 1000 species of basal ferns (15). However, more than 400 plant species never produce chlorophyll and remain mycoheterotrophic throughout their lives (12). Additionally, there is a third group, the partially mycoheterotrophic plants that require fungal C for establishment, and even though they produce chlorophyll on emergence from the soil, these adult plants still rely on fungi for a portion of their C requirements (23; 4; 19).

The physiology and ecology of the mycoheterotrophic mode of nutrition has been investigated in a number of plant families (12; 14), most notably in the Orchidaceae where mycoheterotrophy is characterized by high tissue N concentrations (6) and strong enrichment of tissues in ¹³C and ¹⁵N (6; 24; 4). Such characteristically enriched isotope profiles have also been observed for the family Ericaceae (23; 29; 9). Moreover, previous studies have harnessed this characteristic isotopic enrichment to estimate the degree of mycoheterotrophy in orchids that contain chlorophyll but are still reliant on fungi for the provision of some of their C requirements (6; 29).

To date, such physiological investigations of mycoheterotrophy have focused on plants associating with ectomycorrhizal (Basidiomycota) fungi with no detailed physiological investigation of mycoheterotrophic plants associated with the other main group of mycorrhizal plants that associate with arbuscular mycorrhizal (AM) fungi(Glomeromycota).

Like the majority of angiosperms, the family Gentianaceae (∼1700 species) are primarily autotrophic plants forming AM associations (12; 2). Gentians are an intriguing group in which to study the evolution of mycoheterotrophy because they contain taxa with apparently different degrees of commitment to the mycoheterotrophic lifestyle. Less than 2% of the Gentianaceae are achlorophyllous and thus obligate mycoheterotrophic plants (∼25 species); the genera Cotylanthera,Voyria, and Voyriella are composed entirely of obligate mycoheterotrophs, and the genus Sebaea has one achlorophyllous representative, Sebaea oligantha (21). Less well understood and circumscribed within the family are the partially mycoheterotrophic and green Gentianaceae.

The small North American genera Bartonia and Obolaria (Gentianaceae) are green but have leaves reduced to scales or foliose bracts (7) and reportedly have “low chlorophyll content” (17) although the provenance of the data are unclear. Bartonia is comprised of three species (16), and Obolaria is monotypic (7). In these genera, overall leaf reduction, the presence of coralloid roots, and the absence of root hairs, termed structural simplification (15), suggest their partial dependence on mycorrhizal associations for fungus-derived C. Based on these superficial criteria, the latest flora of the mid-Atlantic USA considers Bartonia, “presumably partially myco(hetero)trophic” and that Obolaria has “well-developed mycorrhizae and may be substantially myco(hetero)trophic” (27, p. 684), but physiological investigation of these hypotheses has not been undertaken. Here, using field-collected individuals of O. virginica and B. virginica to investigate the physiological characteristics of these putative, partial mycoheterotrophs, we analyzed (1) the natural abundance isotope signatures (δ¹⁵N and δ¹³C) of O. virginica and B. virginica in relation to those of their surrounding nongentian vegetation because mycoheterotrophic plants have been shown to be enriched in these heavy elements, (2) the chlorophyll content (because chlorophyll has been shown to be deficient in some partial mycoheterotrophs; 4), and (3) the morphological characteristics of the mycorrhizal symbiosis.

MATERIALS AND METHODS

Sampling regime

All plant material was collected from southeastern Virginia, USA. For all analyses, B. virginica was collected (13 October 2007) in forested wetlands (36°49′23.18″N, 76°50′53.78″W) from the Old Dominion University Blackwater Ecologic Preserve, Isle of Wight County. Plants of O. virginica were collected (12 May 2008) from a rich mesic hardwood forest (36°49′59.12″N, 76°51′33.63″W) adjacent to the preserve, Southampton County.

For stable isotopic analyses of each study species, we collected five flowering plants (root and shoot) that were at least 5 m distant from the next plant collected. To estimate the average stable isotope ratios of the surrounding vegetation for comparison to the putative mycoheterotrophic plants, we established center points at each B. virginica and O. virginica sampling location. From these center points, we sampled the fully expanded new leaves of the year from the first five plant species (other than B. virginica and O. virginica) within a 2-m radius.

For chlorophyll content estimation, whole green stems (including reduced leaves) of B. virginica (N = 9) were sampled, excluding the inflorescence. Two fully expanded green foliose bracts from the terminal flower of each O. virginica (N = 9) were collected. Plant material for chlorophyll content estimation was stored in the dark on ice and assayed ∼4 h after collection.

Microscopy

The whole, freshly excised root system of five B. virginica and O. virginica plants was divided into 0.5 cm long sections, and the roots were cleared in 2 M potassium hydroxide at 80°C for 2 h. All root pieces were then washed in distilled water and incubated in lactophenol cotton blue at room temperature for a further 4 h to stain for mycorrhizal infection. The stained root pieces were washed in 50% (v/v) glycerol for 30 min to remove excess stain and were mounted as squashes in glycerol on glass microscope slides under coverslips sealed with Hystomount (TAAB Laboratories Equipment, Berkshire, UK). Digital images were captured using an Olympus BX51 microscope coupled to an Olympus DP71 digital camera (Olympus UK, Essex, UK).

Chlorophyll concentration

The chlorophyll extraction protocols of 3) were used to estimate chlorophyll a and b concentrations from fresh shoots. Once excised in the field, shoots were stored on ice in the dark for no more than 4 h when chlorophyll was extracted from homogenized material with ice-cold acetone. Optical density was measured at a wavelength (λ) of 645 nm and 663 nm using a FLOUstar (BMG Labtech, Offenburg, Germany) spectrophotometer. The chlorophyll concentration (mg·L⁻¹ of extract) was calculated according to 1 using Eqs. 1 and 2 below and re-expressed as milligrams of chlorophyll per gram of tissue (fresh mass, mg·g⁻¹).

Isotope ratio mass spectrometry (IRMS)

Tissues of B. virginica and O. virginica were separated into shoot and root portions. Root tissues were triple rinsed in deionized water to remove soil. Then all tissues of B. virginica and O. virginica and those of vegetation associated with each sample (leaf only) were heated in a drying oven at 70°C until no further change in mass was detected. Dried tissues were homogenized separately to a fine power with a mortar and pestle, and a 5-mg subset was weighed into ultraclean tin cups, then analyzed for ¹⁵N and ¹³C by continuous-flow mass spectrometry (PDZ Europa 2020 Isotope Ratio Mass Spectrometer (IRMS) coupled to a PDZ ANCA GSL preparation unit). Data were collected as atom % ¹³C and re-expressed as delta (δ) relative to the Pee Belemnite international standard using Eq. 1.

Where R_Sample = ¹³C : ¹²C ratio in the sample and R_Standard = ¹³C : ¹²C ratio in the Pee Belemnite international standard. Tissue N concentrations (%) were simultaneously collected from the same samples also using mass spectrometry.

RESULTS

Whole plant morphology and chlorophyll content of B. virginica and O. virginica

The morphology of B. virginica and O. virginica exhibited features often associated with the transition from autotrophic to mycoheterotrophic plants; leaves are reduced to foliose bracts or scales (Fig. 1). The entire length of all root samples examined (N = 5) was colonized by mycorrhizal fungi with the exception of the main root axis. Mycorrhizal fungi formed both arbuscule-like structures and hyphal coils (Fig. 2). There was no significant difference in the total tissue chlorophyll content of O. virginica compared to B. virginica (t = 0.79; df = 15; P = 0.44 and Fig. 3A) although the chlorophyll A : B ratio was significantly higher in O. virginica compared to B. virginica (t = 2.60; df = 15; P = 0.02 and Fig. 3B).

Figure 1

Photographs of (A) shoots, (B) roots, and (C) whole plant of Bartonia virginica and of (D) shoots and (E) roots of Obolaria virginica. Number codes: 1, flowers; 2, reduced leaves; 3, green stem; 4, reduced “coralloid” root network.

Figure 2

Cleared root squashes of (A and B) Bartonia virginica and (C) Obolaria virginica stained for mycorrhizal infection with lactophenol cotton blue after clearing in hot potassium hydroxide. Number codes: 1, arbuscule-like structures; 2, root vascular stele; 3, fungal hyphae/hyphal coils. All root squashes (N = 7 for each species) showed fungal infection in the form of arbuscules in B. virginica and O. virginica. NS, nonsignificant.

Figure 3

Average (A) total chlorophyll content and (B) chlorophyll a:b ratio for Bartonia virginica (lightly shaded bars) and Obolaria virginica (darkly shaded bars). Error bars represent +1 standard error, pairs of bars marked with an asterisk (*) differed significantly (Student's t test; P < 0.05).

Tissue ¹⁵N and ¹³C signature of B. virginica and O. virginica and surrounding vegetation

The roots and shoots of B. virginica are significantly more enriched in ¹⁵N than the surrounding nongentian vegetation (Fig. 4A, Table 1) with the exception of Liriodendron tulipifera (AM) and Pinus taeda (ectomycorrhizal) (F_10,58 = 12.28; P < 0.001). The roots of B. virginica are significantly more enriched in ¹³C than all surrounding nongentian vegetation (F_10,57 = 12.79; P < 0.001) (Table 1); but the shoots exhibit the same ¹³C signature as the majority of the surrounding reference plants (Fig. 4A, Table 1). The roots and shoots of O. virginica are not enriched in ¹⁵N compared to the surrounding reference vegetation (Fig. 4B, Table 2). However, both the roots and shoots of O. virginica are significantly more enriched in ¹³C than all surrounding plants (Fig. 4B, Table 2; F_8,70 = 5.66; P < 0.001).

Figure 4

Average natural abundance δ¹⁵N (relative to air) and δ¹³C (relative to the Pee Dee Belemnite international standard) (‰) for (A) Bartonia virginica and (B) Obolaria virginica roots and shoots (◆), together with adjacent ectomycorrhizal plants (▽), arbuscular/nonmycorrhizal plants (○) and plants associated with ericoid mycorrhizas (◇). Bidirectional error bars represent ±1 SE. N = 3–12.

Table 1. Mean δ¹³C and δ¹⁵ N (±1 SE) isotope profiles of the roots and shoots of Bartonia virginicaand co-occurring (reference) vegetation. Values within each column sharing the same letter are not significantly different (1-way ANOVA; P> 0.05).

Species	Mycorrhiza	Mean δ^13C ± SE	Mean δ^15 N ± SE
Bartonia virginica (root)	Gentian (AM)	−29.26 ± 0.3 g	0.60 ± 0.30 ab
Bartonia virginica (stem)	Gentian (AM)	−33.26 ± 0.63 cdef	1.58 ± 0.35 a
Acer rubrum	AM	−32.65 ± 0.63 def	−2.79 ± 0.55 d
Clethra alnifolia	AM	−34.15 ± 0.26 abcd	−1.85 ± 0.42 cd
Ilex opaca	AM	−31.58 ± 0.22 ef	−3.27 ± 0.30 d
Liriodendron tulipifera	AM	−35.20 ± 0.02 abc	−1.24 ± 0.02 bcd
Mitchella repens	AM	−35.80 ± 1.00 ab	−2.82 ± 0.82 d
Woodwardia areolata	AM	−31.30 ± 0.05 f	−1.97 ± 0.53 cd
Rhododendron viscosum	Ericoid	−33.40 ± 0.26 cde	−2.53 ± 0.25 d
Vaccinium formosum	Ericoid	−33.63 ± 0.48 bcd	−1.64 ± 1.21 cd
Pinus taeda	Ecto	−37.00 ± 0.10 a	0.18 ± 0.03 abc

• Notes: 1-way ANOVA_[Box-cox] δ¹³C, F_10,57 = 12.79, P < 0.001; 1-way ANOVA for δ¹⁵N, F_10,58 = 12.28, P < 0.001

Table 2. Mean δ¹³C and δ¹⁵ N (±1 SE) isotope profiles of the roots and shoots of Obolaria virginicaand co-occurring (reference) vegetation. Values within each column sharing the same letter are not significantly different (1-way ANOVA; P> 0.05).

Species	Mycorrhiza	Mean δ^13Cδ^15 ± SE	Mean δ^15 N ± SE
Obolaria virginica (root)	Gentian (AM)	−27.61 ± 0.20 a	10.08 ± 0.77 a
Obolaria virginica (stem)	Gentian (AM)	−27.58 ± 0.12 a	8.89 ± 0.42 ab
Acer rubrum	AM	−30.79 ± 0.28 cd	7.31 ± 0.24 b
Bignonia capreolata	AM	−31.07 ± 0.76 d	3.93 ± 1.19 b
Lindera benzoin	AM	−30.00 ± 0.30 bc	9.15 ± 1.15 ab
Liriodendron tulipifera	AM	−29.95 ± 0.27 bc	7.51 ± 0.59 b
Polystichum acrostichoides	AM	−29.50 ± 0.39 b	5.78 ± 0.91 b
Uvularia perfoliata	AM	−31.45 ± 0.14 d	9.10 ± 0.40 ab
Fagus grandifolia	Ecto	−30.07 ± 0.09 bc	9.54 ± 0.48 ab

• Notes: 1-way ANOVA for δ¹³C, F_8,70 = 26.28, P < 0.001; 1-way ANOVA for ¹⁵N, F_8,70 = 5.66, P < 0.001

Tissue N content of B. virginica and O. virginica and surrounding vegetation

Both the roots and shoots of B. virginica were both significantly more concentrated in N than the shoots of other plants sampled (F_11,59 = 14.82, P < 0.001) (Fig. 5A). Moreover, B. virginica roots were significantly more concentrated in N than associated shoots (F_11,59 = 14.82, P < 0.001) (Fig. 5A). In contrast, tissue N concentrations were much more variable at the Obolaria site with the shoots of O. virginica being significantly less concentrated in N than four co-occurring species and not significantly different from the remaining three species (Fig. 5B). Again O. virginica roots were significantly more enriched in N than associated shoots (Fig. 5B).

Figure 5

Tissue N concentration (% N) of (A) Bartonia virginica and surrounding vegetation and (B) Obolaria virginica and surrounding vegetation. Bars sharing the same letter are not significantly different (P > 0.05: ANOVA followed by Tukey's multiple comparison test), error bars represent +1 standard error. Bars are shaded according to their mycorrhizal association. (B.v = Bartonia virginica; A.r = Acer rubrum; B.r = Bignonia capreolata; C.a = Clethra alnifolia; F.g = Fagus grandifolia; I.o = Ilex opaca; L.b = Lindera benzoin; L.t = Liriodendron tulipifera; M.r = Mitchella repens; O.v = Obolaria virginica; P.a = Polystichum acrostichoides; P.t = Pinus taeda R.v = Rhododendron viscosum; U.p = Uvularia perfoliata)

DISCUSSION

The overwhelming majority of the ∼1700 species within the family Gentianaceae are autotrophic with the exception of three achlorophyllous and fully mycoheterotrophic genera, Cotylanthera,Voyria, and Voyriella (21), that rely on fungi for the provision of all of their C and the majority of their nutrient requirements. Many green plant species, termed partial mycoheterotrophs, can however obtain significant amounts of C from their fungal partners during their adult life stages as well as through photosynthesis (4; 14). Such dependence of the plant on the fungal partner in mycoheterotrophic associations leads to a reduction in photosynthetic surfaces with leaves reduced in size, often to scales or foliose bracts, and reduced root production (15). Consequently, the hairless coralloid roots heavily colonized by mycorrhizal fungi and reduced photosynthetic surfaces of two members of the Gentian family, O. virginica and B. virginica, have long led botanists to suspect partial mycoheterotrophy in these genera (8; 27) despite the lack of physiological evidence supporting this hypothesis in species of either genus.

The structure of gentianoid AMs has been shown to be highly variable, with both arbuscules and intracellular hyphal coils reported, the latter being the more predominant feature (22). Furthermore, 10) suggested that the presence of hyphal coils in Voyria aphylla (Gentianaceae), a feature shared with many mycoheterotrophs, may be indicative of significant control of fungal development by the plant partner in mycoheterotrophic plant–fungal symbioses. In line with previous observations, the roots of O. virginica and B. virginica were colonized by mycorrhizal fungi forming both highly branched arbuscule-like structures and fungal coils within the cortical cells of both species (Fig. 2).

Partially mycoheterotrophic plants, such as the orchid C. trifida, have reduced chlorophyll concentrations as well as reduced photosynthetic surfaces (4). Although the photosynthetic areas of O. virginica and B. virginica are clearly reduced (Fig. 1), potentially limiting autotrophic C gain, their total chlorophyll concentration anda:b ratio are comparable to those recorded for other species from both dicotyledonous and monocotyledonous autotrophic plants obtained using the same methodology (3). Thus, further evidence beyond total chlorophyll concentration to support the notion of mycoheterotrophy in these taxa is required.

Natural abundance profiling of stable isotopes (¹⁵N and ¹³C) have previously been employed as a tool to investigate an organism's position within a food web (18; 6; 25; 11) with the compounded effects of isotopic discrimination increasing along trophic hierarchies leading to an enrichment in tissue natural abundance δ¹⁵N and δ¹³C. Such natural abundance stable isotope approaches have been used to help resolve the trophic strategies of many species of the ∼10% of plants (12, 13) that rely on fungus-derived C for at least part of their lives (6; 24; 11; 4). Stable-isotope-based physiological evidence for mycoheterotrophy or indeed partial mycoheterotrophy is well developed in orchids (6; 11; 4) and Ericaceae (23; 29; 9) but is lacking for most other plant families with mycoheterotrophic species.

The extent of ¹⁵N and/or ¹³C enrichment seen in O. virginica and B. virginica is generally greater than the surrounding nongentian, autotrophic plants with an average δ¹⁵N of 8.9 ‰ (±0.4) and δ ¹³C of −27.6 ‰ (±0.1) for O. virginica and an average δ¹⁵N of 1.6 ‰ (±0.4) and δ ¹³C of −33.3 ‰ (±0.6) for B. virginica, the trend is more complex when compared to the natural abundance isotope profiles of other fully and partially mycoheterotrophic plants. The δ¹⁵N profile for O. virginica is similar to that recorded for the partially mycoheterotrophic (ectomycorrhizal) orchid Cephalanthera damasonium (δ¹⁵N of 8.6 ‰ [± 0.5]; 11), but B. virginica appears less enriched in ¹⁵N than C. damasonium. Moreover, both gentian species are less enriched in ¹⁵N than the multispecies average for fully mycoheterotrophic orchids (δ¹⁵N of 10.6 ‰ [± 2.5]; 24). Likewise, the δ¹³C signatures of the two gentians are less enriched than those recorded for the partially mycoheterotrophic orchid C. damasonium (average δ ¹³C of −26.9 ‰ [± 0.3]; 11) and than the multispecies average for fully mycoheterotrophic orchids (δ ¹³C of −22.2 ‰ [± 0.2]; 24). While these isotopic fractionation patterns and N concentrations are generally supportive of the notion of partial mycoheterotrophy in both gentian species, clear interspecific differences between B. virginica and O. virginica are suggestive of the utilization of different physiological pathways for mycoheterotrophic C gain when compared to other partial mycoheterotrophs that, in contrast, associate with ectomycorrhizas rather than AMs. These natural abundance ¹³C and ¹⁵N profiles, coupled with reduced photosynthetic surfaces of O. virginica and B. virginica, a convergent feature shared with many mycoheterotrophic and partially mycoheterotrophic orchids, and natural abundance profiles of stable isotopes strongly suggest that both species are partially mycoheterotrophic.

Mycoheterotrophy thus appears to have evolved on multiple occasions in the Gentianaceae, and the most recent tribal and subtribal molecular phylogeny (21) suggests at least two independent origins of obligate mycoheterotrophy in tribes Saccifolieae (Voyriella) and Exaceae (Cotylanthera and Sebaea) with the remaining mycoheterotrophic gentian genus, Voyria, unplaced. Bartonia and Obolaria, represent separate lineages from the obligate mycoheterotrophic Gentianaceae and are nested in the more derived tribe Gentianeae (21) although both species have been regarded as closely related based on morphology (see 28; 26). While the precise phylogenetic relationship between these genera remains unresolved (5; 26; 16), the phylogenetic analysis by 16) unequivocally shows that Bartonia and Obolaria are not sister taxa and thus suggests the independent evolution of mycoheterotrophy in both genera. However, that notion is based on the assumption that other green genera of the Gentianaceae are fully autotrophic (with the exception of Bartonia and Obolaria). Still, the possibility remains that partial mycoheterotrophy, especially during seedling establishment, is underreported in the Gentianaceae; thus, there is a need to resolve the trophic status of green and presumably autotrophic genera in the family (i.e., Gentianella,Latouchea,Megacodon,Swertia) at all life stages.