Schima superba outperforms other tree species by changing foliar chemical composition and shortening construction payback time when facilitated by shrubs

A 3.5-year field experiment was conducted in a subtropical degraded shrubland to assess how a nurse plant, the native shrub Rhodomyrtus tomentosa, affects the growth of the target trees Pinus elliottii, Schima superba, Castanopsis fissa, and Michelia macclurei, and to probe the intrinsic mechanisms from leaf chemical composition, construction cost (CC), and payback time aspects. We compared tree seedlings grown nearby shrub canopy (canopy subplots, CS) and in open space (open subplots, OS). S. superba in CS showed greater growth, while P. elliottii and M. macclurei were lower when compared to the plants grown in the OS. The reduced levels of high-cost compounds (proteins) and increased levels of low-cost compounds (organic acids) caused reduced CC values for P. elliottii growing in CS. While, the levels of both low-cost minerals and high-cost proteins increased in CS such that CC values of S. superba were similar in OS and CS. Based on maximum photosynthetic rates, P. elliottii required a longer payback time to construct required carbon in canopy than in OS, but the opposite was true for S. superba. The information from this study can be used to evaluate the potential of different tree species in the reforestation of subtropical degraded shrublands.


Results
Plant growth, photosynthetic and structural parameters. Results showed that relative growth rates of seedling height (RGR-H, P < 0.001), and basal diameter (RGR-B, P = 0.001), maximum photosynthetic rates (A max , P = 0.023) and specific leaf areas (SLA, P < 0.001) were significantly affected by species, and the interaction between species and treatment (P ≤ 0.023 for all) (Table 1). Meanwhile, RGR-B was also significantly affected by treatment factor (P = 0.012). Pinus elliottii showed significantly lowered RGR-H, RGR-B, and A max (P ≤ 0.033 for all) in canopy subplots (CS) than in open subplots (OS) (Fig. 1a-d). Similarly, the RGR-B and A max of Michelia macclurei were significantly decreased (P < 0.001, and P = 0.012), but SLA was increased (P = 0.012) when grown in CS than in OS. In contrast, Schima superba had much higher RGR-B, RGR-H, and A max (P ≤ 0.015 for all) grown in CS than in OS. These parameters for Castanopsis fissa, however, were not significantly different between shrub nurse treatments and controls.
Leaf main element concentration. Leaf total carbon (C), nitrogen (N), and phosphorus (P) concentrations were significantly affected by species (P < 0.001 for all), but not affected by treatment (P ≥ 0.223 for all) (Table 1). Meanwhile, leaf N concentration was significantly affected by the interactions between species and treatment (P = 0.003). For P. elliottii, leaf C and N concentrations were significantly lower (P = 0.008, and  Table 1. P values from two-way ANOVA statistics for the effects of species (4 levels) and treatment (2 levels) on plant growth parameters, maximum photosynthetic rate, specific leaf area, and leaf total carbon, nitrogen and phosphorus concentrations.
Scientific RepoRts | 6:19855 | DOI: 10.1038/srep19855 P = 0.006) when grown in CS than in OS ( Fig. 1e-g). In contrast, for S. superba and C. fissa, leaf N concentrations were significantly increased when grown in CS than in OS (P = 0.001, and P = 0.009). Leaf P concentrations were not significantly changed between treatment for any of the four species.
Leaf chemical composition. Leaf minerals, organic acids, protein, lipid, soluble phenolics, lignin, TSC and TNC concentrations were significantly affected by species (P ≤ 0.041 for all), but only leaf organic acids and soluble phenolics concentrations were significantly affected by treatment (P = 0.002, and P = 0.023, respectively) ( Table 2). Leaf minerals, organic acids, protein, lipid concentrations were also significantly affected by the interactions between species and treatment (P ≤ 0.019 for all). For P. elliottii seedlings, leaf organic acids was significantly increased (P < 0.001) but leaf protein (P = 0.018) and soluble phenolics (P = 0.039) were significantly decreased when grown in CS than in OS (Fig. 2). The minerals concentration of S. superba was significantly lower (P = 0.046) but protein concentration of this species was significantly higher in CS than in OS (P = 0.025). For C. fissa, minerals, organic acids and protein concentrations were all significantly higher when grown in CS than in OS (P ≤ 0.037 for all). For M. macclurei, only soluble phenolics were significantly lower when grown in CS than in OS (P = 0.016). Lignin, TSC or TNC concentrations were not significantly affected by shrub nurse effect for any of the tested target species.

Figure 1.
Plant relative seedling height growth rate (a), plant relative basal diameter growth rate (b), maximum photosynthetic rate (c), specific leaf area (d), and leaf total carbon (e), nitrogen (f) and phosphorus (g) concentrations. Values are means + SD (n = 5). Single, double and triple asterisks indicate a significant difference between CS and OS plots at p < 0.05, p < 0.005, and p < 0.001, respectively. Construction costs and payback time. CC value and payback time were significantly affected by species (P < 0.001, and P = 0.009), but not significantly affected by treatment and the interaction between species and treatment (Table 1). For P. elliottii, CC value was significantly decreased (P = 0.003) but payback time was significantly increased (P = 0.047) when grown in CS than in OS (Fig. 3). Also, CC value of C. fissa and payback time of S. superba were found to be decreased in CS than in OS (P = 0.044, and P = 0.001). CC values and payback time of M. macclurei were not significantly different between treatments.  As revealed by other studies, nurse plant can simultaneously exert both facilitative and competitive effects on target plants seedlings, and the relative dominance of either positive or negative effects largely relies on the traits of tested species 6,7,10,14 . Our previous studies also showed that the differences in target plant growing performances in canopy subplots were mainly caused by their adaptations to light environment 8,17 . S. superba is a typical late-successional species that adapted to a wide range of light intensity, i.e. it is shade tolerant in juvenile stage but grow in higher light conditions as they mature 13 . In contrast, P. elliottii, is a fast growing species, and M. macclurei is a light demanding species, thus the canopy shade composed negative growing conditions for their growth during the 3.5-year experiment. Moreover, changes in plant resource use efficiencies may also help explain such differences, i.e. some species outperform others by acquiring limited resources or by using resources more efficiently 12,27 . For M. macclurei, the lowered photosynthetic rates together with the unchanged leaf C, N, P concentrations under R. tomentosa may have decreased photosynthetic energy, nitrogen or phosphorus -use efficiency when grown in canopy subplots. In contrast, S. superba can highly increase its resource use efficiencies when nursed by shrub R. tomentosa. Thus, we conclude that the shrub nurse plant does not benefit the growth and resource-use efficiency of P. elliottii and M. macclurei during the 3.5 years. It follows that the former two species may not be suitable for long-term shrubland restoration when the shrub R. tomentosa is used as a nurse plant.

Source of Variation
Although attempts to understand the mechanisms underlying differences in plant utilization and allocation of assimilated carbon among species have often focused on the role of photosynthesis, process of downstream carboxylation may also provide important information on how carbon assimilation is related to the chemical composition of plant organs 18 . The metabolism of organic acids is fundamentally important at the cellular level for several biochemical pathways, or at the individual level in modulating plant adaptation to the environment 28 . Organic acids are also involved in transporting micronutrients in the transpiration stream in the xylem 29,30 . In our study, the highly increased organic acids in foliar tissues of P. elliottii and C. fissa grown in canopy subplots showed that the nurse shrub coverage may largely improve their cellular metabolism such as participate in the balance of charges formed during the extensive metabolism of anions 28,31 .
As a component of functional proteins, structural proteins, and the photosynthetic machinery, nitrogen is an important plant nutrient 32 . In this study, higher leaf nitrogen concentrations were associated with elevated leaf protein contents in S. superba and C. fissa growing in canopy subplots. Reduced leaf nitrogen contents in P. elliottii, in contrast, were associated with decreased protein contents in canopy subplots. There may be a trade-off between investing nitrogen in the photosynthetic apparatus such as chlorophyll rather than in structural compounds such as cell wall proteins 33,34 . The proportions of leaf nitrogen partitioned to different nitrogen pools are affected by irradiance 35 , nutrition 36 , and other environmental factors 34 . In this study, as affected by the microenvironment changed by shrub nurse plant, more nitrogen may have been transformed to protein in leaves of S. superba and C. fissa, while less nitrogen was available for constructing different forms of protein in needles of P. elliottii growing in canopy subplots vs. open subplots.
Soluble phenolics are thought to have antioxidant effects in stressed plants 37 . Researchers have proposed that a high concentration of phenolics may protect leaves against high light conditions and may help avoid photosynthetic down-regulation and photoinhibition 38,39 . In our study, plants growing in the open rather than canopy subplots may have suffered from excessive radiation. In two of our tested target species, P. elliottii and M. macclurei, phenolic concentrations were substantially higher in open subplots than in canopy subplots. The increased phenolic concentrations may have protected their photosynthetic apparatus against the excessive energy. These results also indicate that P. elliottii and M. macclurei can cope with high light conditions and that their establishments in subtropical degraded shrublands may not require shade condition composed by shrub coverage.
The increase in phenolics is mostly associated with lignification because lignin compounds often create barriers between injured and healthy tissues 40 . Unlike previous studies, however, our study did not detect significant differences in lignin concentration in open vs. canopy subplots for any of the four target species. Total structural carbohydrate and structural carbon in plants are mostly involved in morphogenesis of the plant cytoskeleton, and total structural carbohydrate content is relatively stable within a species 41 . If the growth of a plant is limited by photosynthesis, total non-structural carbohydrate may be lower in a declining stand than in a healthy stand 42 . In our study, however, total structural and non-structural carbohydrate concentrations did not differ in open subplots vs. canopy subplots for any of the four species, although there were clear differences in growth and photosynthetic performances in the two subplot types.
Researchers have concluded that there is a compromise in chemical composition between growth and defence within a plant because of limitations in available energy 43 . According to this theory, a plant must choose to invest energy in growth-related processes or in the accumulation of defence-related compounds 44 . For example, species with high growth potential invest more energy in primary compounds (proteins) and less in secondary compounds with a possible defence role (such as phenols or lignin); the opposite is true for species with a low growth potential 18,45 . This theory was partially supported by the relationships between chemical compounds and the growth of target species in canopy subplots in the current study. Because the shrub canopy provides a relatively benign micro-environment for other plants, S. superba and C. fissa seedlings growing in canopy subplots invested more energy in the synthesis of proteins than in the synthesis of defence compounds (e.g. phenolics or lignin). In contrast, P. elliottii had decreased levels of proteins and soluble phenolics in canopy subplots, indicating that this target species did not benefit from R. tomentosa.
The CC value is an important parameter relative to the carbon budget of plants. It is a "black box", however, because the underlying mechanisms that produce different values across treatments are unclear 18 . Increasing our understanding of CC values requires additional analyses of the chemical compositions of plants. For example, positive correlations between compounds with high energy costs (e.g. proteins, soluble phenolics, and lignin) and low energy costs (e.g. minerals and organic acids) could buffer variations in CC values and help explain the difference in CC values between treatments 22,46,47 . In this study, low levels of high-cost compounds (e.g. proteins) accompanied by high levels of low-cost compounds (e.g. organic acids) contributed to significantly lower CC values for P. elliottii growing in canopy subplots than in open subplots. For S. superba, because the increased levels of low-cost minerals could not balance the increased levels of high-cost proteins, CC values did not significantly differ between open and canopy subplots. In contrast, the elevated levels of low-cost minerals together with low-cost organic acids resulted in significantly decreased CC values for C. fissa seedlings growing in canopy subplots. For M. macclurei, except for soluble phenolics, most of the chemical compounds did not significantly differ in canopy vs. open subplots, which resulted in little change in CC values.
Research on leaf CC values and associated traits (e.g. payback time for the carbon investment) has provided insights into carbon acquisition strategies of plants and has therefore helped explain plant growth patterns and population dynamics 48 . Our study also showed that the measurement of a single chemical component or nutrient element might not be able to reveal much about the mechanisms of the plant nurse effects on target species. CC values together with estimated payback time, however, integrate the changes of plant chemical composition and photosynthesis, which may help explain plant response to nurse plants and may enable us to predict long-term growing performances of target plants. The calculated payback time based on CC values for P. elliottii was significantly elevated, but that for S. superba was decreased when growing in canopy subplots. From the perspective of long-term regional restoration practices (> 3.5 years), we believe that P. elliottii will require a longer time to construct enough carbon for growth and metabolism when growing in the canopy than when growing in open subplots, while the opposite would be true for S. superba. Thus, we suggest that P. elliottii should be used solely as a pioneer tree to construct plantations, and that S. superba can be more widely used to accelerate the establishment of native plantations using R. tomentosa as a nurse plant in subtropical degraded shrublands. We also determined that shrub nurse plant did not shorten the payback time for carbon construction of C. fissa or M. macclurei seedlings, which means that the two species did not benefit from shrub nurse effect during their establishment on subtropical degraded shrublands.

Conclusion
Changes in plant chemical composition, construction cost and payback time contribute to the different growing performances of target plants when nursed by shrubs. Among the four tested target species, S. superba is the only one that can effectively utilize the ameliorated microclimate constructed by shrubs, thus can potentially form multi-species communities and accelerate the reforestation of degraded subtropical shrublands. The findings of this study showed that the outcomes of shrub nurse effects are highly species-and time-specific because not all species will benefitted from shrub nurse plants at the same time. This study only demonstrated the possible mechanisms on plant nurse effect at a certain time point (3.5 year after the initiation of experiment). Therefore, it is possible that the nurse benefit may there during the early establishment state and gradually dissipate thereafter. Reforestation on degraded ecosystems is a long-term practice. Thus, it is crucial that careful experimental selection of both nursing and target species is assessed using long-term studies before massive practical restoration and plantation efforts.

Study site. This study was initiated in 2007 in a subtropical shrubland located at the Heshan National Field
Research Station of Forest Ecosystems (112 o 50'E, 22 o 34'N, Heshan County, Guangdong Province, China). The regional subtropical evergreen broadleaved forest has degraded into a community dominated by shrubs and grasses. The soil type in this region is a typical laterite soil that has been seriously eroded because of a lack of forest coverage. The subtropical monsoon climate in this region is characterized by cool and dry winters and humid summers. The annual precipitation ranges from 1460 to 1820 mm, which mainly occurs as rain between March and August. At the station, the mean annual air temperature is 21.7 °C, and the mean annual solar radiation is 435.75 KJ cm −2 .
Scientific RepoRts | 6:19855 | DOI: 10.1038/srep19855 Plant species. Rhodomyrtus tomentosa (Ait.) Hassk. (Myrtaceae) is an evergreen shrub that grows naturally in infertile and acidic soils in tropical and subtropical Asia 49 . R. tomentosa, which is an indicator of acidic soil and a pioneer on bare land, has a cushion-shaped canopy and can grow as tall as 2 m. Our previous work showed that mature shrubs significantly reduced sunlight radiation and soil temperature, but significantly elevated soil volumetric water content, soil bulk density, as well as soil capillary moisture in subtropical degraded shrublands. R. tomentosa was therefore selected as a nurse plant for different tree seedlings in this and our previous studies 8,9,17 .
Four tree species were selected as target species. Schima superba Gardn. et Champ (Theaceae), Castanopsis fissa (Champ. ex Benth.) Rehd. et. Wils, and Michelia macclurei Dandy. are broadleaved evergreen native tree species commonly used for reforestation in tropical and subtropical regions of China. Among them, S. superba is a shade tolerant, C. fissa is a moderate mesophytic and M. macclurei is a light demanding species. A fourth target tree species, Pinus elliottii Engelm. is an introduced conifer that can grow as tall as 30 m. It is fast growing and resistant to drought and nutrient-poor soils on degraded hills. This pine species is widely used in reforestation in southern China, because it has ecological values such as water and soil conservation, and economic values such as a source of timber, pulp, nuts and rein 50 . Several researches have documented that pines can be facilitated by shrubs 6,7 , however, few previous researches tested the facilitation of native shrub Rhodomyrtus tomentosa on the establishment and growth of pine seedlings on subtropical shrubland 51 . In this study, four reforestation species, including three broadleaved species and one conifer species, were selected to test whether they can be well facilitated by shrubs in their initial life history, to identify the underlying mechanisms from carbon allocation aspects, and to instruct regional reforestation practices.
Experimental design. The detailed information regarding the establishment and design of the experiment were previously reported 12,17 and are also briefly described below. In 2007, we divided the 2-ha subtropical shrubland field site into three blocks. Within each block, four plots (5 m × 5 m) were randomly selected, and each plot was assigned to one of the four target tree species, giving a total of three plots for each target tree species. Within each plot, one 2 m × 2 m subplot was assigned the "canopy" treatment (CS), and one was assigned the "open" treatment (OS). CS subplots were located under the circular edges of the R. tomentosa canopies, which were 1.0-1. Leaf chemical compositions. After photosynthetic rates were measured, leaves on the same branches of each tree were collected and oven-dried at 60 °C for 72 h. The dried leaves were then ground, passed through a 0.08 mm sieve, and dried again in the oven. Total carbon (C) and nitrogen (N) concentrations in the leaf samples were determined spectrophotometrically with the potassium dichromate oxidation spectrophotometric method and the Kjeldahl method. Phosphorus concentration (P) was determined colorimetrically after HClO 4 -H 2 SO 4 digestion 52 . Nitrate-N contents of leaf samples were measured colorimetrically with salicylic acid 53 . Ash content was determined by combusting 1 g of plant material in a muffle furnace at 550 °C for 6 h and then weighing the residue. Ash alkalinity was determined acidimetrically by titration 54 .
Another set of oven-dried leaf samples (1g) was extracted with a solution of water, methanol, and chloroform in a volumetric ratio of 1:2:2 55 . The extracts in the chloroform phase were dried with a rotary evaporator, and the residue (total lipids) was weighed. Soluble carbohydrates were measured in the methanol-water phase using anthrone reagent 56 . The soluble phenol contents were also determined colorimetrically in the methanol-water phase with Folin-Ciocalteus reagent 57 . After extraction with the water, methanol, and chloroform mixture, the residues of the leaf samples were boiled in 3% HCl (v/v) for 3 h. Insoluble sugars were subsequently analysed in the supernatants 56 . The final residues after boiling with HCl were used for another round of determination of carbon and nitrogen concentration using the potassium dichromate oxidation spectrophotometric method and the Kjeldahl method, respectively 52 .
Chemical calculations. Total mineral, protein, and organic acid concentrations in target plant leaf samples were estimated using the following equations 58  The final residue after extraction was considered to be a mixture of lignin and TSC. Lignin concentration was calculated with the carbon and nitrogen concentration and assuming that the carbon concentration in the (hemi) cellulose complex was 444 mg g −1 and that the carbon concentration in lignin was 640 mg g −1 21 . CC values of target plant leaf samples were calculated with the following formula 59  where CC is the construction cost (g glucose g −1 ), C om is the organic carbon concentration (g g −1 ), M is the total mineral content (g g −1 ), and N org is the organic nitrogen concentration (g g −1 ). Because the quantity of ammonium and nitrate taken up by the different tree species was unknown, these CC values should be considered as maximum values. Payback time was estimated as CC/A mass by transforming the unit of CC from g glucose g −1 to nmol C g −1 and by transforming the unit of A mass from nmol CO 2 g −1 s −1 to nmol C g −1 h −1 . Payback time was calculated per hour rather than per day, because the diurnal radiation period changes during the growing season. As a consequence, the estimated payback time is considered to be the theoretically minimum amortization period 52,60 . Statistical analyses. We used IBM SPSS Statistics 19.0 for statistical analyses. Results are presented as means + standard deviation (SD). For each target species, differences between OS and CS subplots in plant growth and photosynthetic parameters, CC value, chemical composition, and payback time were analysed with one-way ANOVAs. Two-way ANOVA was applied to determine the effect of species (4 levels) and treatments (2 levels) on plant growth and photosynthetic parameters, CC value, chemical composition, and payback time.