Leaf physiological and structural plasticity of two Asplenium (Aspleniaceae) species coexisting in sun and shade conditions

1Department of Plant Biology, Educational and Scientific Centre “Institute of Biology and Medicine” Taras Shevchenko National University of Kyiv, 64/13, Volodymyrska Street, Kyiv, ISO 3166-2:UA-01601,Ukraine 2Department of Environmental Biology, Sapienza University of Rome, P.le A. Moro 5, Rome, ISO 3166-2:IT-00185, Italy 3Current address: Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Tartu, Estonia *Corresponding author: olena_vasheka@knu.ua REGULAR PAPER


INTRODUCTION
Leptosporangiate ferns (Polypodiidae) are the most numerous group of extant spore-bearing vascular plants (Pteridophyte Phylogeny Group 2016). Despite their ancient origin, the vast majority of modern ferns has emerged in the period of the last 100 million years (Pryer & Schuettpelz 2009) together with the evolution and spreading of flowering plants (Schneider et al. 2004). However, evolutionary competition with angiosperms resulted in habitat diversification, specialization, and extension of biological diversity as well as increasing adaptability of many fern species. According to Page (2002), the "low-light photosynthetic ability" is among the twelve important advantages of pteridophyte biology, providing a broad framework for exploitation of ecological habitats. Vasheka, Gratani & Puglielli, Physiological and structural plasticity of two Asplenium species Comparison of the main functional traits of ferns and their adaptive significance with that of seed plants has largely been made (Choy-Sin & Suan 1974;Ludlow & Wolf 1975;Bauer et al. 1991;Brach et al. 1993;Karst & Lechowicz 2007). In general, it was established that some important leaf traits and physiological characteristics of ferns have values that correspond to the lowest ones reported for seed plants (Karst & Lechowicz 2007;Tosens et al. 2016). This applies for instance to leaf mass per unit of leaf area (Karst & Lechowicz 2007), hydraulic conductivity of xylem (Brodrib & Holbrook 2004;Watkins et al. 2010), stomatal and mesophyll conductance (Nishida et al. 2015;Tosens et al. 2016), photosynthetic capacity Tosens et al. 2016) and photosynthetic nitrogen use efficiency (Durand & Goldstein 2001;Stuntz & Zolt 2001;Tosen et al. 2016).
Apart from the differences between ferns and seed plants, relatively few studies have specifically addressed fern's plastic responses to different environment factors as different light conditions. Phenotypic plasticity is defined as the capacity of a given genotype to express different phenotypes under different environmental conditions (Valladeres et al. 2007) and it is considered to play an adaptive role in survival of the species in a changing environment. The standard plasticity experiment involves two or more taxa (clones, genotypes, families, populations, species) that grow in a series of different environments in the field, or under controlled condition (Schlichting 1986).
Some works dedicated to the sun/shade response between different fern species (Choy-Sin & Suan 1974;Ludlow & Wolf 1975;Winter et al. 1986;Brach et al. 1993;Arens 1997;Nurul Hafiza et al. 2014) highlighted that fern response to sun/shade agrees with that of seed plants. In fact, ferns growing in high light condition tend to be characterized by lower total chlorophyll content and higher chlorophyll a/b ratio, stomatal density, light compensation and saturation points, net photosynthetic rates and nitrogen contents per unit leaf area. However, there is still lack of information on the morpho-functional response of different fern species to different light environments. In particular, there is no information on the range of these differences. That is, how much morphological, anatomical and physiological plasticity can be expressed by different fern species in the sun/shade transition.
Changes of photosynthetic rates across leaves with different structure and chemistry have been successfully expressed by the leaf economics spectrum (Wright et al. 2004). Relationships between leaf structure and function in ferns have received less attention, with two exceptions: Karst & Lechowicz (2007) analysed the so-called 'frond economics spectrum' by including 20 understory fern species and, by exploring only area-based correlations, found that the relationships between foliar traits were in agreement with those of the leaf economic spectrum. However, the mass-based photosynthetic capacity characterizes the biochemical capacity of single cells and is the key player in the worldwide trade-off between the physiological and structural characteristics of leaves (Westoby et al. 2013;Niinemets et al. 2004;Tosens et al. 2016). Accordingly, Tosens et al. (2016), analysing the bivariate relationships between leaf traits in 35 ferns grown under non-stressful conditions found mass-based relationships to be stronger than the area-based ones, in agreement with Wright et al. (2004) for seed plants, but in contrast with Karst & Lechowicz (2007). Despite this evidence, at the best of our knowledge there is no previous study that has addressed how the sun-shade transition differentially affect the relationship between leaf structure and function in ferns. As a result, we do not know how the expected changes in leaf structure affect changes in mass-based net photosynthesis in ferns growing in different light environments (i.e. cost in terms of carbon assimilation). This is of particular concern since bivariate relationships between leaf structure and function are useful tool for predicting species-specific adaptations to particular environments (Wright et al. 2004).
Considering these gaps of knowledge, the present study aimed at analysing the leaf morpho-physiological response of two Asplenium L. species coexisting under both sun and shade conditions in the field but characterized by different ecological preferences. Asplenium ceterach L. is a thermophilic species that tends to colonize habitats characterized by higher radiation levels, while Asplenium trichomanes L. is more linked to shaded and humid environments (see below).
In particular, we wanted to address: (i) whether leaf morphological trait variations pose a physiological cost in the sun/shade transition independently of the species and (ii) whether growing under full sunlight can be achieved through a common pattern of response across coexisting (and congeneric) species.
We hypothesized that: (i) sun leaves may be characterized by a greater cost in terms of carbon assimilation, as highlighted by a steeper slope for the relationships involving mass based net photosynthesis and leaf structural traits, due to a greater structural investment in leaves; (ii) even if a common pattern of response (in terms of mean values) could be highlighted, we still expected that the selected species could differ in plasticity of key traits (i.e. high range of variability) that could reflect their different ecological requirements.

Plant material
Two largely distributed Asplenium L. species were selected. Both species are perennials with overwintering leaves that persist 12-18 months in the study area. Previous studies have demonstrated that longer-term leaf traits versus light rela-tionships are stable during most of the growing season Hallik et al. 2012).
Asplenium trichomanes L. grows in a range of habitats from open to shaded rocky ones, but prefers calcareous rocks (such as limestone and dolomite) or mortared walls. It sometimes occurs in terrestrial habitats. It is widespread in temperate and subarctic areas of northern hemisphere and also occurs in mountainous regions in the tropics extending to Australia and western regions of South America. In Italy, it occurs up to 2910 m a.s.l. (Pignatti 1982;Marchetti 2001;Soster 2001;Preston et al. 2002;Tutin et al. 2010).
Despite the fact that both species often prefer rocky outcrops and can grow together, A. trichomanes has much wider ecological amplitude and geographical range. According to Didukh's (2011) approach for the estimation of ecological scales, both species have similar requirements to damping (ω = 0.05-0.25), soil aeration (Ae = 50-95), and salt regime (30-200 mg/l). Asplenium trichomanes grows in a wide range of soil humidity (Wnp = 55-250 mm), light condition (from scyophytes to heliophytes), and thermal climate (25-85 kcal cm -2 year -1 ). On the other hand, ecological requirements of A. ceterach are shifted to warmer (Wnp = 25-75 mm) and higher insolation condition (it belongs to the group of sub-heliophytes or heliophytes).

Morphological leaf traits
Morphological measurements were carried out on fully expanded fertile leaves from each selected plant (n = 9). Measurements included leaf area (LA, cm 2 ), leaf length (cm) and leaf width (cm) measured by an Image Analysis System (Delta-T Devices, UK) and leaf dry mass (DM, g), measured after desiccation at 80°C to constant mass. LA and DM of the whole lamina without rachis (for A. trichomanes) and massive midrib (for A. ceterach) were considered. Leaf width was measured at the midpoint of the leaf.
Leaf mass per unit of leaf area (LMA, g m −2 ) was calculated by the ratio between DM and LA and leaf tissue density (LTD, mg cm −3 ) by the ratio between LMA and total leaf thickness (LT, µm) (Wright et al. 2004).

Anatomical leaf traits
Anatomical measurements were carried out on pinnae (AT) or lateral lobes (AC) from the central part of the fully expanded fresh leaves (n = 9) and analysed by light microscopy (Zeiss Axiocam MRc 5 digital camera (Carl Zeiss) using an Image Analysis System (Axiovision AC software). The following parameters were determined in transverse sections: total leaf thickness (µm), abaxial and adaxial cuticle with cell wall thickness (µm), abaxial and adaxial epidermis thickness (µm), mesophyll tissue thickness (µm).

Gas exchange measurements
Measurements of gas exchange were carried out using the infrared gas analysers ADC LCPro+ (UK) equipped with a leaf chamber (PLC, Parkinson Leaf Chamber). Measurements were made on three leaves per each sun and shade plants (n = 9). Net photosynthetic rate (Aa, µmol CO 2 m ̶ 2 s ̶ 1 ), photosynthetic photon flux density (PPFD, µmol photon m ̶ 2 s ̶ 1 ), stomatal conductance (gs, mol H 2 O m ̶ 2 s ̶ 1 ) and transpiration rate (E, mmol H 2 O m ̶ 2 s ̶ 1 ), were measured. Measurements were carried out under natural conditions, on cloudfree days at saturating PPFD (> 1,500 µmol photon m ̶ 2 s ̶ 1 in sun conditions), in the morning from 8:00 am to 12:00 am. CO 2 concentration in the leaf chamber was set at 400 µmol CO 2 mol ̶ 1 air, and relative air humidity of the incoming air ranged between 40% and 60%. The intrinsic water-use efficiency (iWUE, µmol CO 2 mol H 2 O ̶ 1 ) was calculated as the ratio between Aa and gs. Water-use efficiency (WUE, µmol CO 2 mmol H 2 O ̶ 1 ) was calculated as the ratio between Aa and E. Mass based net photosynthesis (Am, nmol CO 2 g -1 s -1 ) was calculated as the ratio between Aa and LMA (Wright et al. 2004).

Data analysis
Standardized major axis regression (SMA, Warton et al. 2006) was used to analyse the relationships LMA-LT, LMA-LTD, LTD-LT and between Am and LMA, LT and LTD with Light as the main factor. The differences in terms of slopes and intercepts were tested by the Likelihood Ratio and Wald statistic, respectively. All analyses were run with the R library SMATR (Warton et al. 2012) on log-transformed data.
To test if species shared a common pattern of response in the sun-shade transition, a Principal Component Analysis (PCA) by including all the considered traits grouped by Species×Light environment was carried out. Then, a Twoway ANOVA was performed in order to test the effect of factors Species, Light and their interaction (Species×Light) on the extracted principal components (i.e. PC1 and PC2).
Multiple comparisons were analysed by a Tukey test. Such approach was used in order to reduce multiple testing, considering that the use of emerging collective properties (expressed by PCs) as primary variables allows for an equally robust approach (Giuliani 2017).
The phenotypic plasticity index (PI, Valladares et al. 2000) was calculated for each species (hereafter PI AC and PI AT ) in order to quantify the degree of phenotypic plasticity for each of the considered traits in response to different light environments. PI was calculated as the difference between the minimum and the maximum mean value divided by the maximum mean value per each trait. Then, PI AT was regressed against PI AC via linear regression analysis (n = 25).

Traits variability and bivariate relationships
The results highlighted an overall difference in the considered leaf traits of sun and shade leaves for the two considered species ( fig. 1, table 1).

intercept, R 2 and P of the Standardized major Axis Regression for the log--log relationships.
The log-log relationships between leaf dry mass per unit leaf area (LMA) and net photosynthetic rate per unit of leaf dry mass (Am), leaf thickness (LT) and Am, leaf tissue density (LTD) and Am, LTD and LMA, LT and LMA and LTD and LT for sun and shade leaves. P in bold indicates significance at < 0.05.
Changes in LMA were due to changes in both LTD (R 2 = 0.53; P = 0.0006) and LT (R 2 = 0.59; P = 0.0002) for shade leaves while in sun leaves only the relationship LMA-LT (R 2 = 0.87; P = 1.18E-08) was significant ( fig. 2A-C, table 2). LT and LTD were in turn significantly and positively related only in shade leaves (R 2 = 0.42; P = 0.003). Moreover, LMA and LT scaled negatively (P < 0.05) with Am in both sun and shade leaves while the relationship Am-LTD was only significant in shade leaves (R 2 = 0.39; P = 0.005) ( fig.  2D-F, table 2). Significant differences between slopes for the considered bivariate relationships were not found. However, intercepts were always found to significantly differ between sun and shade leaves ( fig. 2A-F) except for the relationship LMA-LTD.

Patterns of phenotypic plasticity: convergences and divergences between species
The PCA returned two axes of variation with percentage of explained variance of 38 % and 21 % for PC1 and PC2, respectively ( fig. 3; supplementary file 1). PC1 was significantly and positively related to gs, E, LMA, LTD, LT, mesophyll thickness and stomatal density while negatively with WUE, leaf length and abaxial epidermal cell length. PC2 was significantly and positively related to Aa, Am and adaxial cuticle thickness while negatively with adaxial epidermal cell width, iWUE, LP, water content and stomatal cell length showed the same magnitude of correlation with both of the PCs. The Two-way ANOVA showed that the factors Species and Light were significant for both the PCs while the interaction terms were not significant in both of the analyses (table 3). Overall, as expected, Light accounted for most of the variance for PC1 while the factor Species for PC2.
The phenotypic plasticity patterns obtained for the two considered species in response to light were significantly related (R 2 = 0.68; P = 4.667e-07; fig. 4). However, on average, A. trichomanes was characterized by a slightly higher anatomical plasticity as compared to A. ceterach (0.19 and 0.14, respectively), while the latter showed a greater physiological plasticity (0.60 and 0.46 in A. ceterach and A. trichomanes, respectively). The two species roughly converged in terms of the magnitude of morphological plasticity (0.35 and 0.38 in A. ceterach and A. trichomanes, respectively).

DISCUSSION
Plant responses to reduced light environment have been investigated in several ecological and eco-physiological studies (see Valladares & Niinemets 2008 for a review) but ferns have often been disregarded. In the present study, the response of two congeneric fern species coexisting under sun and shade conditions in the field was analysed.

Traits variability and bivariate relationships
Overall, the results highlight significant divergence between leaves developed under full sunlight and shade conditions (mean PPFD = 1895±175 and 15±1, respectively) in terms of mean values of the considered traits. In particular, as expected, we found an increase in Aa and gs for sun leaves, and the values are in the range of that reported for fern species (Aa generally < 10 and gs between 0.017 and 0.177, Tosens et al. 2016). In our study, LMA and LT ranged between 40.67-123.04 g m -2 and 164-388 μm, respectively. Interestingly, such values were slightly outside the ranges reported for these traits in the literature (Tosens et al. 2016;Vasheka et al. 2016) suggesting that sun-shade transition represents a strong driver of LMA and LT variations in ferns. Accordingly, we found that both LMA and LT scaled negatively with photosynthetic rate on a mass basis (Am) and that LT was the main determinant of LMA changes in both sun and shade leaves. Moreover, the scaling relationships differ in terms of intercepts, reflecting species acclimation to different light conditions. It is in fact well known that different environmental conditions may determine different proportional changes of the parameters on the x-and y-axis, with the result that intercepts significantly differ (Wright et al. 2001;Atkinson et al. 2010), as observed here. However, an eventual cost in terms of Am due to changes in LMA and LT would be out- Figure 4 -Relationship between the phenotypic plasticity index measured in response to light for Asplenium trichomanes (PI AT ) and Asplenium ceterach (PI AT ) for the 25 morphological, anatomical and physiological leaf traits considered (see table 3). Values of PI calculated per each parameter and species were used as experimental units (n = 25). R 2 = 0.68; P = 4.667e-07. The 1:1 line, mean values for each trait pool (black dots) and the parameters that mainly drive mean differences are also shown: stomatal density (DSt), stomatal conductance (gs), water use efficiency (WUE), intrinsic water use efficiency (iWUE). lined by differences in slopes between sun and shade leaves, but we did not observe any difference in slopes. Moreover, we want to stress that the slope we found for the relationship Am-LMA (-1.48) was identical to that found by Tosens et al. (2016) by including data for 35 fern species (-1.47).
We argue that the common slope shared by sun and shade leaves for the relationship Am-LMA and Am-LT, together with differences in intercepts, reflects the ability of the considered species to adapt morphological and anatomical leaf traits in response to light with no further costs in terms of carbon assimilation. Additionally, there was a general lack of relationship between LTD and the rest of the considered parameters only for sun leaves, possibly highlighting a tendency of the considered species to better modulate LT under full sunlight conditions, supported by the lack of relationship LTD-LT. A similar result was obtained by Puglielli et al. (2017) for a shrub species growing in different light environments. Moreover, LTD is a costly trait (de la Riva et al. 2016;Puglielli & Varone 2018), as also attested by the greatest slope of the relationship LTD-Am (-4.41) as compared to Am-LMA and Am-LT (mean slope = -1.48). Therefore, such response can allow sun leaves to avoid excessive costs in terms of Am by preferably modulating LT in order to avoid a longer pay-back time for leaf construction costs (i.e. longer leaf life-span). On the contrary, this may happen in shade conditions even if this aspect needs further investigation.
Concerning the anatomical traits, few comparisons with other studies could be made even if the measured values are in agreement with Vasheka et al. (2016).

Patterns of phenotypic plasticity: convergences and divergences between species
The ANOVA carried out on the two PCs extracted was decisive to determine if species shared a common pattern of response in the shade-sun transition. The analysis highlighted a greater effect of Light than Species on PC1 as compared to PC2. Such result can be easily interpreted. In fact, for mathematical construction the PC1 captures the highest variance within the data, so it is reasonable to expect greater differences in terms of Light than Species along this axis. On the contrary, PC2 should also include species-specific differences for the considered traits, thus resulting in a greater effect of Species than Light, as observed. Moreover, the lack of significance for the interaction term (Species×Light) for both PC1 and PC2 demonstrates that such response patterns are independent of species-specific differences in trait values. If so, a convergence in terms of phenotypic plasticity patterns for the considered species could be expected. Accordingly, the plastic responses of the two species scaled positively (R 2 = 0.68, P = 4.667 e -07) with most of the points falling tightly around the 1:1 line in fig. 4. Mean plasticity values plotted in the same graph showed that A. trichomanes tends to be characterized by a slightly higher anatomical plasticity (mean value falling above the 1:1 line) while A. ceterach is characterized by a higher physiological plasticity (mean value falling below the 1:1 line). Such a discrepancy mainly comes from different plastic behaviour in few parameters. In fact, A. ceterach is characterized by a greater plasticity in gs, iWUE and WUE which drives the mean below the 1:1 line, while the higher anatomical plasticity in A. trichomanes is driven by a greater plasticity of stomatal density. Concerning A. ceterach, its response is due to less stomatal closure under full sunlight conditions resulting in a lower iWUE and WUE as compared to shade leaves. On the other hand, A. trichomanes tends to have a more conservative water use strategy. For this species increasing stomatal density under full sunlight condition expands opportunity to better control stomatal conductance in relation to photosynthesis (i.e. iWUE) (Heterington & Woodward 2003;Puglielli et al. 2017). However, this is not enough to increase iWUE to a great extent in sun as compared to shade leaves, possibly reflecting hydraulic limitations to photosynthesis. Strong hydraulic limitation of photosynthesis in ferns is generally reported (Zhang et al. 2014;Tosens et al. 2015), and it seems to be due to the fact that water and CO 2 share a significant fraction of the outside-xylem pathways in photosynthetic organs . While this hypothesis is valid for A. trichomanes, A. ceterach seems to display no hydraulic limitation of photosynthesis or, at least, to a lesser extent. These differences can be also reflected by the different structure of mesophyll tissues between the species ( fig. 1). The mesophyll of A. ceterach is well differentiated on palisade and spongy tissues whereas A. trichomanes has undifferentiated mesophyll which can further justify different water use and carbon fixation strategy in A. ceterach as compared to A. trichomanes. Additionally, A. ceterach belongs to the group of desiccation-tolerant ferns (Proctor & Tuba 2002;Hietz 2010) and has some additional adaptations for control of water loss such as leaf curling and densely covered abaxial leaf surface by peltate scales, the effect of which on water uptake and loss of the desiccation-tolerant epiphytic fern was recently established (John & Hasenstein 2017).
Most studies have investigated phenotypic plasticity of leaf traits in angiosperms and values between 0.20 and 0.60 are generally reported (as reviewed by Gratani 2014 andChelli et al. 2019). On the other hand, gymnosperms display much lower phenotypic plasticity, with PI ranging between 0.09 and 0.12 for anatomical and 0.14 and 0.32 for physiological traits (Wyka et al. 2007). Our results show that the investigated Asplenium species display PI values comparable to those of seed plants and that further comparisons are needed in order to place phenotypic plasticity of ferns in an evolutionary context.

CONCLUSIONS
On the whole, the present work highlights a significant degree of phenotypic plasticity for the considered species in response to different light conditions reflecting a remarkable acclimation capacity. This is supported by the lack of costs in terms of carbon assimilation possibly due to the species ability to modulate leaf thickness under full sunlight conditions. This allows reducing the leaf construction costs thus resulting in the same slopes for the considered bivariate relationships between sun and shade leaves.
Moreover, the study provides the first evidence of a common pattern of response to different light conditions in two fern species. Despite the similarities in the response to sun-shade transition we found that A. ceterach and A. trichomanes diverge in terms of water use strategies which reflect a different tolerance to full sunlight conditions. In particular, A ceterach seems to be more suited to cope with full sunlight conditions than A. trichomanes, according to the ecological requirements of the species. In fact, while both species prefer substrates with little water storage capacity such as rocks, cliffs, cracks in stone and scree slopes, A. trichomanes occurs, however, in more shady places or locations with continuous water supply.
This work contributes to the so far overlooked response of ferns to different light conditions which can further elucidate the ecological success of this group of vascular plants. Moreover, we identified a set of traits and provide data analysis procedure that can be used to test integrated sun-shade patterns of response in other species of ferns.

SUPPLEMENTARY FILE
One supplementary file is associated to this paper: Factor loadings and Species×Light coordinates for the principal component analysis (PCA) carried out on the considered traits: https://doi.org/10.5091/plecevo.2019.1525.1891