(13C) concentration (expressed as d

(13C) concentration (expressed as dPDB values) was quantified at
harvest in shoots and roots to assess C-assimilation and allocation
belowground, and one nonlabelled plant per species was harvested
for background 13C concentration. Lipids were extracted from
freeze-dried fine roots (n = 8 per species) and analyzed as described
earlier (van Aarle&Olsson, 2003; Lekberg et al., 2012). In order to
assess the C-allocation specifically to AM fungi, we measured the
13C concentration in theAMfungal storage lipid NLFA 16:1x5 in
all labeled plant roots by compound specific isotope ratio mass
spectrometry (Olsson et al., 2005). To control for differences in
overall allocation belowground between the two plant species, we
also compared 13C concentration in NLFA 16:0 (occurs both in
plants and AM fungi) and in NLFA 18:2x6,9 (in plant roots but
not in AM fungi).
Similar to previous studies involving nonhost plants (Hildebrandt
et al., 2001), Dianthus roots were colonized by hyphae and
vesicles but rarely by arbuscules (two Dianthus plants had < 5%
arbuscular colonization), and overall AM colonization was
substantially lower (P < 0.001, t-test) than in Hypochoeris plants
(Fig. 1). This was supported by large differences (P < 0.001, t-test
on loge-transformed values) in 16:1x5 concentrations between
Dianthus (11.1  5.7 nmol g

1 root, mean  standard error (SE))
and Hypochoeris roots (536  180 nmol g

1 root). We do not
know if Dianthus can formAMin the absence of mycorrhizal plants
(sensu Atriplex confertifolia in Miller et al., 1983), but the colonization
of field plants suggests that the nearly nonhost status of this
plant is not due to cell wall structures or antifungal compounds that
deter hyphal growth and penetration. Lack of signaling may explain
the low AM colonization (discussed in Koide & Schreiner, 1992)
and this could be tested in controlled inoculation experiments with
Dianthus grown with and without good host plants.
Reflecting the lower AM colonization, only three Dianthus
plants had roots with detectableNLFA16:1x5, which clearly limits
our ability to test if Dianthus allocates C to AM fungi. Due to this,
and because of the large variation among Dianthus replicates, we
chose not to analyze the data statistically but instead to plot