Different leaf cost–benefit strategies of ferns distributed in contrasting light habitats of sub-tropical forests

Abstract

Background and Aims Ferns are abundant in sub-tropical forests in southern China, with some species being restricted to shaded understorey of natural forests, while others are widespread in disturbed, open habitats. To explain this distribution pattern, we hypothesize that ferns that occur in disturbed forests (FDF) have a different leaf cost–benefit strategy compared with ferns that occur in natural forests (FNF), with a quicker return on carbon investment in disturbed habitats compared with old-growth forests.

Methods We chose 16 fern species from contrasting light habitats (eight FDF and eight FNF) and studied leaf functional traits, including leaf life span (LLS), specific leaf area (SLA), leaf nitrogen and phosphorus concentrations (N and P), maximum net photosynthetic rates (A), leaf construction cost (CC) and payback time (PBT), to conduct a leaf cost–benefit analysis for the two fern groups.

Key Results The two groups, FDF and FNF, did not differ significantly in SLA, leaf N and P, and CC, but FDF had significantly higher A, greater photosynthetic nitrogen- and phosphorus-use efficiencies (PNUE and PPUE), and shorter PBT and LLS compared with FNF. Further, across the 16 fern species, LLS was significantly correlated with A, PNUE, PPUE and PBT, but not with SLA and CC.

Conclusions Our results demonstrate that leaf cost–benefit analysis contributes to understanding the distribution pattern of ferns in contrasting light habitats of sub-tropical forests: FDF employing a quick-return strategy can pre-empt resources and rapidly grow in the high-resource environment of open habitats; while a slow-return strategy in FNF allows their persistence in the shaded understorey of old-growth forests.

INTRODUCTION

As important components of tropical and sub-tropical forests, ferns play crucial roles in nutrient cycling (Dearden and Wardle, 2008), microclimate maintenance (Flinn, 2007; Song et al., 2012) and forest regeneration (Coomes et al., 2005; Walker et al., 2010). It is a general point of view that some ferns are primarily distributed in the shaded understorey, with quite a few species on the brink of extinction (Aldasoro et al., 2004; Wild et al., 2006), while others are able to thrive in disturbed open habitats, forming monospecific dense stands that are problematic as invaders (Walker, 1994; Page, 2002). This contrasting distribution pattern of ferns has long been observed across forests of different regions (Kenoyer, 1928; Nasrulhaq-Boyce and Mohamed 1987; Walker et al., 1996; Saldana et al., 2005; Flinn, 2007), but the underlying and pivotal mechanisms responsible for these distribution patterns are not well understood.

Plant functional traits and their response to environmental variability have been proved valuable to illuminate different strategies of carbon fixation (i.e. photosynthetic rates and specific leaf weight) or water use (i.e. water-use efficiency and leaf hydraulic conductance) used by ferns with contrasting life forms (Watkins et al., 2007, 2010; Zhang et al., 2014), or to elucidate the different response of stomatal conductance to drought or CO₂ enhancement between ferns and other vascular plants (Brodribb and Holbrook, 2004; Flexas et al., 2014). In particular, leaf functional traits of ferns have been found to be tightly associated with their habitat preferences (Saldana et al., 2007; Parra et al., 2009). For example, Saldana et al. (2005) observed that the wide range of light environments occupied by Blechnum mochaenum might be due to its high plasticity of specific leaf area (SLA) to changes in light availability. In another study, three tree fern species distributed under either closed or open canopies differed significantly in leaf photosynthetic and morphological traits, including maximum electron transport rate, light saturation point and SLA (Riano and Briones, 2013).

Leaf cost–benefit strategies that are quantified using a suite of leaf functional traits including SLA, maximum net photosynthetic rates (A), leaf nitrogen content (N), photosynthetic nitrogen-use efficiency (PNUE), leaf construction costs (CCs) and leaf life span (LLS) (Wright et al., 2004; Poorter et al., 2006; Shipley et al., 2006; Coste et al., 2011), are widely used to investigate potential mechanisms underlying the distribution patterns of seed plants in contrasting light habitats (Villar and Merino, 2001; Navas et al., 2003; Poorter and Bongers, 2006; Zhu and Cao, 2010). For example, plants from high-light habitats employ quick return energy-use strategies, having leaves with high SLA and A, low CC and short LLS. In contrast, plants from low-light environments adopt slow-return energy-use strategies, bearing low SLA and A, high CC and long LLS. A leaf cost–benefit analysis has been employed to explain successfully the rapid spread and success of an invasive tree fern Sphaeropteris cooperi that could achieve high photosynthetic rates at a low leaf carbon cost (Durand and Goldstein, 2001). As a contrasting pattern, an endangered fern species Adiantum reniforme var. sinensis is confined to shaded and moist habitats, which is due to its high energy requirements for leaf construction (Liao et al., 2007). Yet, whether and how the leaf cost–benefit strategy of ferns can be correlated with their contrasting distribution patterns in forests with different light conditions (e.g. shaded understorey vs. open habitats) remains to be understood.

The Dinghushan Forest Ecosystem Research Station (DFERS), located in the sub-tropical region of southern China, consists of both heavily disturbed forests and natural forests that have been protected for >400 years (Kong et al., 1997). Previous studies have identified different distribution patterns of ferns in different types of forests (i.e. disturbed forests vs. natural forests) in the DFERS (Zhang, 2011). For example, Dicranopteris dichotoma (Thunb.) Bernh. and Blechnum orientale L. represent ferns that are widespread in open sites of disturbed forests, while Arachniodes exilis (Hance) Ching, an endangered fern species, is restricted to shaded understorey of old-growth forests.

In this study, we selected 16 fern species from the two forest types that represent contrasting light habitats (with eight species from each forest), and conducted a leaf cost–benefit analysis for these fern species in terms of leaf functional traits including LLS, leaf N and P content, SLA, A, CC, photosynthetic nitrogen- and phosphorus-use efficiency (PNUE and PPUE) and payback time (PBT). Our hypothesis was that ferns from disturbed forests (FDF) possess higher A, greater PNUE and PPUE, and shorter PBT and LLS as compared with ferns from natural forests (FNF). This quick-return energy-use strategy of FDF helps them to capture and utilize resources in open habitats of disturbed forests where resource (e.g. light) availability is generally greater. In contrast, the slow-return energy-use strategy of FNF is inclined to enhance their survival in shaded understorey where resource availability appears to be lower.

MATERIALS AND METHODS

Study site and plant material

The present study was carried out in the DFERS (21°09′21″−21°11′30″N, 112°30′39″−112°33′41″E), Chinese Academy of Sciences, Guangdong Province, southern China. The mean annual precipitation at this site is about 1900 mm, of which 80 % occurs in the wet season from April to September. Mean annual temperature is 21·4 °C, with monthly temperature ranging from 12·6 °C (January) to 28·0 °C (July). The studied fern plants were collected at two specific sampling sites: (1) a disturbed forest at 200–250 m altitude located in the outer areas of the DFERS; this forest has experienced anthropogenic disturbances (e.g. logging), and has canopy height and coverage of 10 m and 50 %, respectively; and (2) a natural forest at 220–300 m altitude located in the core areas of the DFERS. This well-protected forest is >400 years old, with canopy height and coverage of 15 m and 95 %, respectively.

All of the 16 fern species are terrestrial and perennial. According to the Flora of China (http://www.floraofchina.org), they are found across a wide range of elevations, but in this study all samples were obtained within the elevation range 200–300 m. Only healthy, matured vegetative leaves were used for trait measurements.

To identify light levels experienced by fern species at the two sites, photosynthetic photon flux density (PPFD) in the two forests was measured simultaneously between sunrise and sunset (0700−1900 h, local time) over three consecutive sunny/cloudy days using quantum sensors (Li-1400; LiCor, Lincoln, NE, USA). The sensors were positioned at similar heights above fern plants that were sampled for trait measurements, and the PPFD data were recorded at 10 min intervals (Supplementary Data Fig. S1).

Leaf life span

At the beginning of the growing season (April) in 2013, a total of 30–50 newly emerged leaves from 10–20 individuals per species were labelled. Leaf growth and mortality were recorded twice a month until the end of 2014. Average LLS was calculated as the average time elapsed between leaf emergence and fall.

Photosynthetic light response curve

We measured photosynthetic light response curves between 0800 and 1130 h on consecutive sunny days, using a Li-6400 (LiCor) portable photosynthesis system with an LED red/blue light source. The PPFD was set at 1500, 1000, 500, 200, 100, 50, 20, 10 and 0 µmol m^–2 s^–1, respectively. Ambient conditions were a CO₂ concentration of 390 µmol mol^–1, air humidity in the range 55–65 %, leaf temperature at 28 °C and vapour pressure deficit of 1·5 kPa. Under each PPFD step, plants were exposed to the above conditions for 5–15 min to allow the stabilization of photosynthetic parameters. Five fully expanded and healthy leaves were measured from five individuals for each species. Since a few fern species (i.e. Schizoloma ensifolium, Schizoloma heterophyllum and Pteris vittata) have relatively small leaves that could not completely cover the area within the chamber size (2 × 3 cm²), gas exchange values given by Li-6400 were corrected using the chamber area/actual leaf area ratio as a correction factor. Light response curves were obtained by fitting the data to a non-rectangular hyperbola:

$$A_{\text{n}}=\frac{\text{PPFD}\times \varphi +A_{\text{a}}-\sqrt{{(\varphi \times \text{PPFD}+A_{\text{a}})}^2-4\times \varphi \times \theta \times \text{PPFD}\times A_{\text{a}}}}{2\times \theta }-R_{\text{d}}$$

where A_n is the net photosynthetic rate (μmol m^–2 s^–1), A_a is the light-saturated net photosynthetic rate (μmol m^–2 s^–1), ϕ is the apparent quantum yield (mol C mol^–1 photons), θ is the curvature of the non-rectangular hyperbola and R_d is area-based respiration (μmol m^–2 s^–1). In this study, a common apparent quantum yield (ϕ = 0·05) and curvature (θ = 0·80) were used for all species. The light saturation point (LSP) was calculated as the value of PPFD for which the photosynthetic rate reached 90 % of A_a. The light compensation point (LCP) denoted the x-intercept where the net photosynthetic rate is equal to zero.

Specific leaf area, leaf nutrient and construction cost

For each species, the leaves of individuals used for photosynthetic measurements were collected. Leaf area was measured with a leaf area meter (Li-3000A; LiCor), and then the leaves were oven-dried for 48 h at 70 °C to determine dry mass. The SLA (cm² g^–1) was calculated as leaf area per dry mass. Dry leaves were ground and homogenized for subsequent chemical analyses.

Mass-based total nitrogen concentration (N_m; mg g^–1) was determined by a Kjeldhal analysis. Mass-based total phosphorus concentration (P_m; mg g^–1) was determined using atomic absorption spectrum photometry. Ash concentration was determined gravimetrically after the combustion of samples for 4 h at 500 °C. Heat of combustion was determined with an oxygen bomb calorimeter (Model 6400, Parr, IL, USA), and the ash-free heat of combustion was calculated by converting the heat of combustion on a total dry mass basis to the corresponding ash-free mass. Mass-based leaf construction cost (CC_m; g glu g^–1) was defined as the amount of glucose required to produce 1 g of biomass from glucose and minerals, and was calculated following Williams et al. (1987):

$${\text{CC}}_{\text{m}}=\frac{(\mathrm{0·06968}\times Hc-\text{0}\mathrm{·65)}\times \text{(1}-\text{ash})+\mathrm{7·5}\times (k\times \text{N}/\mathrm{14·0067})}{\mathrm{0·89}}$$

where Hc is the ash-free heat of combustion (kJ g^–1), ash is the ash concentration (g g^–1), N is the nitrogen concentration (g g^–1) and k is the oxidation state of the nitrogen source (+5 for nitrate or −3 for ammonium). We used k = –3 because it has been shown that ammonium is the main source of soil nitrogen at the study site (Zhang, 2011).

To calculate area-based leaf construction cost (CC_a) and nutrient concentrations (N_a and P_a), the above measured mass-based values were divided by the SLA. PNUE and PPUE were calculated from the ratio of mass-based maximum photosynthetic rate (A_m) to N_m and P_m, respectively.

Payback time

Payback time (PBT; expressed in days) denotes the time that a leaf requires to amortize its construction costs through photosynthesis (Poorter et al., 2006; Karagatzides and Ellison, 2009; Feng et al., 2011). PBT was calculated as the ratio of CC_m to specific daily carbon gain (Williams et al., 1989), assuming that sugars fixed throughout the lifetime of a leaf have an equal value to the plant and no leaf respiration is involved in growth processes (Poorter et al., 2006; Cavatte et al., 2012):

$$\text{PBT}=\frac{{\text{CC}}_{\text{m}}}{({\displaystyle \int A_{\text{m}}-{\displaystyle \int R_{\text{m}})\times 12\times 180/72}}}$$

where $\int A_{\text{m}}$ and $\int R_{\text{m}}$ are the mass-based instantaneous rates of net CO₂ fixation integrated over the daytime period and leaf respiration integrated at night, and 180 is the molar mass of glucose, 72 the molar mass of carbon in glucose and 12 the molar mass of carbon. To determine daily net carbon gain ($\int A_{\text{m}}$ – $\int R_{\text{m}}$), we followed the method described in Coste et al. (2011) and Cavatte et al. (2012). For each fern species, light levels experienced by leaves over three typical days (sunny and cloudy) were measured between 07:00 and 19:00 h using a quantum sensor (Li-1400; LiCor) at 10 min intervals. Then, photosynthetic light response curves (Supplementary Data Fig. S2) were used to calculate A_n over 10 min steps, and corrected for SLA to express A_n on a mass basis. The daily carbon gain for fully expanded leaves was estimated as the sum of all A_n values minus plant respiration at night, which is assumed to be 0·07 times the daily net carbon assimilation (Givnish, 1988; Poorter et al., 2006; Coste et al., 2011).

Statistical analysis

Differences in leaf functional traits between the two fern groups were tested using the independent-samples t-test. The relationships between pairs of traits were analysed by a linear regression analysis. A principal component analysis (PCA) was performed for the 16 traits across the studied fern species, using log₁₀-transformed values. All the analyses were performed using SPSS version 13·0 software packages (SPSS, Chicago, IL, USA).

RESULTS

Ferns from disturbed forests (FDF) had significantly higher photosynthetic rates, dark respiration rates and photosynthetic light saturation and compensation points compared with ferns in natural forests (FNF) (Fig. 1). The two groups, FDF and FNF, had similar values (on average) of leaf N_m and P_m (Fig. 2A, B), but FDF showed higher PNUE and PPUE (Fig. 2C, D) than FNF. There were no significant differences in SLA and CC between the two fern groups, but FDF showed shorter LLS as compared with FNF (Fig. 3).

Table 1.

The 16 study fern species investigated in this study

Species	Family	Code	Texture	Elevation	Light preference	Sample site
Blechnum orientale L.	Blechnaceae	Bo	Leathery	300–800 m	High light	Disturbed forest
Cyclosorus parasiticus (L.) Farewell.	Thelypteridaceae	Cp	Papery	100–700 m	High light	Disturbed forest
Dicranopteris dichotoma (Thunb.) Bernh.	Gleicheniaceae	Dd	Papery	100–1100 m	High light	Disturbed forest
Lygodium japonicum (Thunb.) SW.	Lygodiaceae	Lj	Papery	50–800 m	High light	Disturbed forest
Nephrolepis auriculata (L.) Trimen	Nephrolepidaceae	Na	Leathery	30–750 m	High light	Disturbed forest
Pteris vittata L.	Pteridaceae	Pv	Papery	25–800 m	High light	Disturbed forest
Schizoloma ensifolium (SW.) J. Sm.	Lindsaeaceae	Se	Papery	120–600 m	High light	Disturbed forest
Schizoloma heterophyllum (Dryand.) J. Sm.	Lindsaeaceae	Sh	Papery	120–600 m	High light	Disturbed forest
Alsophila podophylla Hook.	Cyatheaceae	Ap	Papery	100–550 m	Low light	Natural forest
Angiopteris fokiensis Hieron.	Angiopteridaceae	Af	Papery	100–600 m	Low light	Natural forest
Arachniodes exilis (Hance) Ching	Dryopteridaceae	Ae	Leathery	400–1100 m	Low light	Natural forest
Cibotium barometz (L.) J. Sm.	Dicksoniaceae	Cb	Leathery	100–900 m	Low light	Natural forest
Hemigramma decurrens (Hook.) Cop.	Tectariaceae	Hd	Papery	100–700 m	Low light	Natural forest
Microsorum fortunei (T. Moore) Ching	Polypodiaceae	Mf	Leathery	200–1800 m	Low light	Natural forest
Pronephrium triphyllum (Sw.) Holtt.	Thelypteridaceae	Pt	Papery	120–600 m	Low light	Natural forest
Pteris fauriei Hieron.	Pteridaceae	Pf	Leathery	50–900 m	Low light	Natural forest

Fig. 1.

Maximum CO₂ assimilation rate (A_a), dark respiration rate (R_d), light saturation point (LSP) and light compensation point (LCP) of the 16 fern species. Species codes are given in Table 1. Data are means ± s.e. (n = 5) for each species. Traits are compared between FDF and FNF as two groups using t-tests (right side). **P < 0·01; ***P < 0·001.

Fig. 2.

Mass-based leaf nitrogen and phosphorus contents (N_m and P_m), and photosynthetic nitrogen- and phosphorus-use efficiencies (PNUE and PPUE) of the 16 fern species. Species codes are given in Table 1. Data are means ± s.e. (n = 5) for each species. These traits are compared between FDF and FNF as two groups using t-tests (right side). ns, non-significant; *P < 0·05; **P < 0·01.

Fig. 3.

Specific leaf area (SLA), leaf construction cost per unit dry mass (CC_m) and per unit area (CC_a), and leaf life span (LLS) of the 16 fern species. Species codes are given in Table 1. Data are means ± s.e. (n = 5) for each species. These traits are compared between FDF and FNF as two groups using t-tests (right side). ns, non-significant; ***P < 0·001.

Across the 16 fern species, LLS was significantly and negatively correlated with A_a, PNUE and PPUE (Fig. 4A–C), but not with SLA and CC_a (Fig. 4D, E). There was a significant and positive correlation between PBT and LLS (Fig. 4F).

Fig. 4.

Correlations between leaf life span (LLS) and other leaf functional traits across the 16 fern species. A_a, maximum CO₂ assimilation rate; PNUE, photosynthetic nitrogen-use efficiency; PPUE, photosynthetic phosphorus-use efficiency; SLA, specific leaf area; CC_a, leaf construction cost per unit area; PBT, payback time. Open and filled circles denote FDF and FNF, respectively. ns, non-significant; *P < 0·05; ***P < 0·001.

The association of leaf functional traits across the 16 fern species was evaluated with a PCA. The first axis of the PCA explained 44 % of the variation in leaf traits, and reflected a slow to quick energy-use continuum (Fig. 5A). The positive side of the first axis loaded fern species from disturbed forest with higher A_a, PNUE and PPUE, and shorter PBT and LLS, whereas the negative side of the first axis had fern species from natural forest with opposite trait patterns (Fig. 5B). The second axis of the PCA explained 24 % of the variation, and, although fern species with higher CC and lower leaf nutrient concentrations (N_m and P_m) were clustered at the top, FDF and FNF were not well separated along the second PC axis (Fig. 5B).

Fig. 5.

A principal component analysis for 16 leaf traits (A) across the 16 fern species (B). Trait abbreviations: SLA, specific leaf area; A_a (A_m), maximum CO₂ assimilation rate per unit area (per unit dry mass); LSP, light saturation point; LCP, light compensation point; N_a (N_m), leaf nitrogen content per unit area (per unit dry mass); P_a (P_m), leaf phosphorus content per unit area (per unit dry mass); CC_a (CC_m), leaf construction cost per unit area (per unit dry mass); PNUE, photosynthetic nitrogen-use efficiency; PPUE, photosynthetic phosphorus-use efficiency; PBT, payback time; LLS, leaf life span. Species codes in (B) are given in Table 1. Open and filled circles in (B) denote FDF and FNF, respectively.

DISCUSSION

Our results showed that ferns in disturbed forests exhibited significantly greater values of photosynthetic traits such as A_a, A_m, PNUE and PPUE than ferns in natural forests (Figs 1 and 2). The results are consistent with findings for seed plants showing that species grown in high-light conditions had higher photosynthetic capacity and greater nutrient-use efficiency than those that occurred in shaded habitats (Poorter et al., 2006; Poorter, 2008). For example, a representative FDF species, Dicranopteris dichotoma, had A_a and A_m of 8·32 µmol m^–2 s^–1 and 179·30 nmol g^–1 s^–1, respectively (Fig. 1), which are comparable with those of seed plants at the same study site. We found in a previous study that values (on average) of A_a and A_m for 34 woody species were 8·90 µmol m^–2 s^–1 and 120·86 nmol g^–1 s^–1, respectively (Zhu et al., 2013). In contrast, A_a and A_m of Arachniodes exilis, a typical FNF species, were 3·93 µmol m^–2 s^–1 and 77·30 nmol g^–1 s^–1, respectively. In addition, PNUE of D. dichotoma was 2-fold higher than that of A. exilis (Fig. 2). The higher photosynthetic capacity and greater nutrient-use efficiencies could give ferns like D. dichotoma a growth advantage over species like A. exilis in the competition for light and nutrient resources.

In the present study, we found that the values of CC_m of ferns ranged from 1·10 to 1·44 g glu g^–1 (Fig. 3), which are fairly close to those of other fern species (Oikawa et al., 2004; Liao et al., 2007), as well as some seed plants (Eamus et al., 1999; Zhu and Cao, 2010; Feng et al., 2011). As shown in Fig. 3B, the CC_m of ferns from the two types of forests had non-significant differences, which is consistent with a number of comparative studies investigating different plant functional groups, such as between evergreen and deciduous woody species (Villar et al., 2006), invasive and native Eupatorium species (Feng, 2008), and tree and liana species (Zhu and Cao, 2010). Through analysing the chemical composition of plants, Villar et al. (2006) documented that the lack of significant differences in CC_m might be due to the positive relationships between highly expensive compounds (e.g. fats) and cheap ones (e.g. minerals) in leaves, leading to the relative constancy of CC_m across species (Poorter and Bergkotte, 1992; Poorter and de Jong, 1999; Villar and Merino, 2001). Taking both the energy cost of leaf biomass and leaf morphological properties into consideration, CC_a could be worked out from the ratio of CC_m to SLA. A number of studies have shown that CC_a represents a useful tool in evaluating plant energy-use strategies (Song et al., 2007; Zhu and Cao, 2010). For example, Nagel and Griffin (2001) found that lower CC_a of the invasive species Lythrum salicaria resulted in its lower energy requirement and higher abundance as compared with co-occurring native species. In the present study, we found that CC_a values of the 16 fern species were highly variable, with values ranging from 22·55 to 89·54 g glu m^–2 (Fig. 3). On average, there was no significant difference in CC_a between the two fern groups, which might be due to the interspecific variability of SLA across the 16 fern species (Fig. 3).

Leaf life span is considered as a balance between lifetime carbon gain and its cost (Suarez, 2005). In this study, we found that LLS was tightly correlated with leaf photosynthetic capacity such as A_a, PNUE and PPUE (Fig. 4A–C), which is consistent with a number of previous studies (Reich et al., 2003; Matsuki and Koike, 2006; Poorter and Bongers 2006). Shorter LLS combined with greater photosynthetic rates and photosynthetic nutrient-use efficiency may contribute to the faster growth of FDF as compared with FNF. Similar results were reported by Feng et al. (2011) who showed that an invasive species, Ageratina adenophora, showed higher photosynthetic energy-use efficiency in plants from invasive populations relative to plants from native populations, which may facilitate the growth of A. adenophora in invasive ranges compared with those in native ranges. It has been shown that low SLA species usually have a relatively high value of dry mass content and high concentrations of structural components such as lignin and cell wall (high CC), resulting in long periods of time (long LLS) required to compensate the expensive investments to leaves (Poorter and de Jong, 1999; Nagel and Griffin, 2001). Although a negative correlation between LLS and SLA (or CC) has been observed at the global scale (Wright et al., 2004; Shipley et al., 2006), neither SLA nor CC correlated significantly with LLS across the 16 fern species in the present study (Fig. 4D E). Several studies have shown that the global pattern in the LLS–SLA relationship may differ at a local scale (Williams et al., 1989; Santiago and Wright, 2007; Coste et al., 2011), because differences in resource availability may lead to the variations in leaf traits causing the LLS–SLA relationship to diverge from that documented in global-scale studies (Vincent, 2006; Gotsch et al., 2010). For example, under low-light conditions, the long-lived leaves of shade-tolerant species are likely to increase mechanical resistance of leaf tissues to avoid physical damage and herbivory, associated with a reduction of SLA (Coley, 1988; Cavatte et al., 2012). Nevertheless, these species commonly exhibit anatomical adjustment involving decreases in leaf thickness and density to enhance light capture, leading to an increase in SLA (Smith et al., 1997; Niinemets and Sack, 2006; Poorter et al., 2006). These observations may explain the lack of a significant LLS–SLA relationship across ferns with different light requirements in the present study.

Payback time, estimated from leaf CC and daily carbon gain, can be regarded as a reflection of energy-use efficiency, namely the plant energy benefit during the LLS (Poorter et al., 2006). In the present study, we noted that FDF might experience light density higher than their LSP around midday (Supplementary Data Fig. S1). This might lead to midday photosynthetic suppression, because a high excess of light energy may cause the deactivation of photosynthetic enzymes and reduction of photochemistry (Ort, 2001; Demmig-Adams and Adams, 2006; Demmig-Adams et al., 2013). In that case, the calculations of carbon gain using light response curves and light intensity are likely to be overestimated. We tested this by measuring daily changes in net photosynthetic rates of the eight FDF species on sunny days (0800–1700 h) with the aid of a Li-6400 portable photosynthesis system equipped with a standard transparent leaf chamber. We found that the dynamics of net photosynthetic rates of these species were synchronous with the changes in light levels, and no obvious occurrence of photosynthetic suppression at midday was observed (Supplementary Data Fig. S3). The results showed that FDF may adapt to high-light habitats (Barker et al., 1997; Oguchi et al., 2003; Takahashi and Badger, 2011), and that the daily carbon gain of FDF we calculated from the light response curve and light intensity was not overestimated.

We also found that the calculated PBT of FDF was in agreement with the observed LLS, but PBT was significant longer (on average) than LLS in the case of FNF (Supplementary Data Table S1), indicating that FNF may have a higher carbon gain than that calculated from light response curves and light levels experienced by these ferns during a day. This could be due to the fact that FNF grown in shaded habitats may have a slow loss of photosynthesis after a sunfleck (Zhang et al., 2009), or FNF may have an unconventional photoreceptor that could increase their light-use efficiency under low-light conditions (Watkins and Cardelus, 2012), or both. Overall, although the two fern groups showed similar leaf CC, FDF could fix the amount of energy required to construct leaves through higher photosynthetic rates within a shorter period of time, thus leading to shorter PBT and LLS as compared with FNF (Fig. 5).

Conclusions

Possessing higher photosynthetic rates, greater nutrient-use efficiencies, and shorter LLS and PBT, ferns in disturbed forests exhibit a distinct quicker return energy-use strategy as compared with ferns in natural forest, which might enable them to allocate more resources to growth and reproduction, and consequently to establish and occupy the open sites of disturbed forests at the study site. In contrast, ferns in natural forest have lower photosynthetic rates, indicating that they require a longer time to amortize the construction cost of leaves, leading to longer LLS and PBT that are features of a slow-return energy-use strategy. Our results demonstrate that the distributions of fern species in different light habitats is consistent with their contrasting life history strategies, and that leaf cost–benefit analysis represents a useful approach to better understand mechanisms underlying the distribution patterns of ferns in this sub-tropical forest.

SUPPLEMENTARY DATA

Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Table S1: mean values of leaf functional traits of the 16 fern species. Table S2: factor loading, eigenvalues and the percentage of variance explained for the first two principal components. Figure S1: diurnal changes in photosynthetic photon flux density (PPFD) during three consecutive sunny/cloudy days in habitats of representative fern species from the disturbed and natural forests. Figure S2: light response curves of photosynthesis in fully mature leaves for the 16 fern species. Figure S3: diurnal changes in net photosynthetic rates for the eight fern species from the disturbed forest.