Discussion
The continuous flowering and fruiting in Ruellia subsessilis, as
reported by Lima and Vieira (2006), were confirmed in this study.
According to the classification of Newstrom et al. (1994), these
phenophases have a continuous pattern, which is actually unusual
for tropical forests species (Opler et al., 1976; Newstrom et al.,
1994; Martin-Gajardo and Morellato, 2003; Marques and Oliveira,
2004). In the same study area, aside from R. subsessilis, this con-
tinuous pattern of flowering and fruiting was also reported for
herbaceous vines of the genus Oxypetalum (Vieira and Shepherd,
2002) and Dalechampia aff. triphylla (Calac ̧ a and Vieira, 2012) and
for the shrubs Ruellia brevifolia and Ruellia menthoides (Lima et al.,
2005; Lima and Vieira, 2006). These data show that in understory
species, this pattern may occur more commonly than in species of
the upper stratum,the canopy.According to Newstromet al.(1994),continuous flowering rarely occurs at the plant level, but is more
common at higher levels of analysis, as of the guild or community.
A significant relationship between the number of flowers and
precipitation showed that the first months of the rainy season
acted as a trigger that induced the flowering peak and conse-
quently, peaks in fruit and seed production. The influence of rainfall
on the reproductive phenology of R. subsessilis was confirmed by
the results of the greenhouse plants. Under the latter conditions,
the different levels of water availability affected the flowering
intensity, fruit set and seed production. According to Reich and
Borchert (1984), water availability is the environmental variable
with strongest influence on the phenology, since temperature and
photoperiod are less influential, due to their smaller variation
in tropical regions. Our results reinforce the statement of these
authors.
In the cultivated plants, seed production was only reduced in
plants under severe drought stress, demonstrating some resistance
of R. subsessilis to drought stress. This resistance may be not only
based on the experimental or natural plant response to drought
stress, which is determined by the duration and severity thereof,
but also by the genetic characteristics of the respective plant pop-
ulation (Larcher, 2004).
The water availability also directly affected the production of
the different floral morphs that can be observed in R. subsessilis. In
the natural population, the production of different morphs varied
according to the level of water availability (dry and rainy seasons),
which explains the seasonal variations of morphs. In the cultivated
plants, it was confirmed that the production of different morphs
is sensitive to the soil water potential. These results confirm our
hypothesis and corroborate studies that reported water availability
as a decisive factor in the differentiation of floral morphs in various
Ruellia species (Munguías-Rosas et al., 2012).
The NC morphs, which grow on plants in the rainy season or
in moist soil, and RC morphs, including those with early self-
pollination, produced by plants in the dry season or in soil with
insufficient water availability, are previously unreported charac-
teristics of Ruellia. Morphs with early selfing were also observed in
Mimulus nasutus and described as flowers with pre-anthesis cleis-
togamy (Diaz and Macnair, 1998), considering that the opening of
flowers which were self-pollinated in pre-anthesis occurs later, as
observed in R. subsessilis.
Therefore, drought stress seems to be an inducing condition
of the production of RC morphs in R. subsessilis. Similarly, in sev-
eral Ruellia species with typical cleistogamy (sensu Lord, 1981), the
production of cleistogamous flowers was associated with drought
stress (Long, 1971; Sigrist and Sazima, 2002; Munguías-Rosas et al.,
2012). The low frequency (2.8%) of the morph with pre-anthesis
cleistogamy indicates that this is an occasional variation of the
RC morph, differing from the latter by early dehiscence of the
anthers and stigma receptivity in the bud. The frequency of its
production, however, may depend on genetic factors, aside from
the environmental, as observed by Diaz and Macnair (1998). Long
(1977) presumed a genetically determined typical cleistogamy in
three Ruellia species (R. tuberosa, R. nudiflora and R. brittoniana).
According to this author, the production of cleistogamous flowers
must be determined by genetic factors, but the seasonal produc-
tion of these flowers might be triggered by environmental factors.
In R. subsessilis, if the production of flowers with pre-anthesis cleis-
togamy were determined by the genetic constitution, apart from
environmental factors, its expression level in the study population
would be considered low.
Production of RC morphs must be correlated with less energy
spent by the plant, as there is a lower production of pollen grains
and ovules and the corolla has a smaller size than in NC morphs.
In two species of Impatiens (Balsaminaceae), Schemske (1978)
found that floral morphs resulting from low energy investment
do increase the adaptive capacity of a plant under limiting condi-
tions (e.g., drought stress). Possibly, the RC morphs of R. subsessilis
similarly increase the adaptive capacity of the plant to the given
habitat situation, since their production in the dry season – along
with the production of NC morphs in the rainy season – will maxi-
mize the reproductive success, resulting in fruiting throughout the
year.
Another characteristic that increases the adaptive capacity of
the plant, which favors the reproductive success of R. subsessilis, is
the long time of stigma receptivity that begins in the pre-anthesis
flowers and lasts until corolla drop. In the bud stage, the receptivity
favors early self-pollination and therefore allows the occurrence of
pre-anthesis cleistogamy. This possibility was confirmed finding a
higher fruiting in pre-anthesis flowers after manual self-pollination
than after cross-pollination. In addition, during the corolla drop,the
stigma receptivity favors late self-pollination, delayed in this case,
in both morphs.
In fact, it was found that the morphs in this population are pre-
dominantly autogamous, due to the absence of pollinators (Braz
et al., 2000; personal obs.). The low fruiting in the test of open pol-
lination with emasculation reinforces this statement. However, NC
morphs seem to experience higher amounts of cross-pollination,
while RC morphs rely more intensely on self-pollination. A mixed
mating system and the presence of different morphs with the same
adaptations mentioned above are also observed in species with dis-
tinctly separated chasmogamous and cleistogamous morphs. The
continued flower opening of morphs in cleistogamy in pre-anthesis
is the main difference between R. subsessilis and other species of
the genus with typical cleistogamy. Lima and Vieira (2006) investi-
gated, in the same study area, the breeding system of three species
of Ruellia, R. brevifolia and R. menthoides – which are cleistogamous,
and of R. subsessilis. These authors recorded high fruiting in the
open-pollination test (control) for the three species (76.5%, 66.7%
and 62.1%, respectively), showing that their reproductive strate-
gies are equally efficient. It is noteworthy that, in these species,
delayed self-pollination is a common characteristic, increasing the
percentage of fruiting in the control.
Autogamy is therefore themain guarantee ofthemaintenance of
R. subsessilis plants atthe study site by sexual reproduction. Despite
this statement, no inbreeding depression was recorded, consider-
ing the high fruiting and high viability of the seeds derived from
selfing. This mating system can result in a low genetic variability
(Schemske and Lande, 1985), but the cross-pollinations that will
still occur may be sufficiently compensatory for the population. A
factor thatfavors xenogamous crosses is the existence of functional
nectaries in both morphs. Future studies on the genetic population
structure may shed light on these issues.
The higher seed weight of autogamous compared to xenog-
amous seeds may result in greater survival of the R. subsessilis
seedlings near the mother plant, as similarly observed in the
cleistogamous Viola stagnina/Violaceae (Eckstein and Otte, 2005).
Heavier seeds may also result in more vigorous seedlings. This
latter characteristic, according to Waller (1984), is highly impor-
tant, as it increases the individual plant competitiveness when the
population density is high. Additionally, our results for the ger-
mination speed of seeds contradict the theoretical predictions of
Schoen and Lloyd (1984). These authors stated that autogamous
seeds would germinate more slowly than xenogamous seeds. How-
ever, a faster germination of autogamous seeds can ensure the
quick establishment of new plants in the population, forming a
vigorous bank of seedlings adapted to the conditions in the vicinity of the mother plant. These results reinforce the importance of
self-fertilization to maintain R. subsessilis in the study area and,
on the other hand, show the possibility of cross-pollination as a
complementary system, which even includes a different strategy of
germination.