Yueyao Feng · Xuan Li · Biyao Zhou · Rong Hu ·Huanchao Zhang · Jiading Yang · Yanming Fang

Abstract Ecological stoichiometry provides a framework for the balance and f low of elements between organisms and ecosystems. Elemental phenotypes have an important inf luence on the environmental adaptation and ecological evolution of plants. There have been few reports on interand intra-specif ic phenotypic variations of ecological stoichiometric traits for congeneric species in a mixed forest although such variations are well- documented at the species level at global, regional and local scales. In this study, total carbon (TC), nitrogen (TN), phosphorus (TP) and potassium(TK) were measured in leaves and the elemental phenotypes were statistically analyzed in four species of oaks— Quercus fabri, Q. serrata var. glandulifera, Q. acutissima and Q. variabilis—in a mixed-oak secondary forest in Yushan, Jiangsu,China. The average element concentrations in the four oak species were not relatively higher than previously reported for oaks from world and Chinese f lora. Ecological stoichiometry traits were correlated with tree height and diameter at breast height, indicating that phosphorous and potassium were positively correlated with tree size, while carbon was negatively correlated, especially the relationship between oak growth and total carbon or total phosphorus was obvious, and the study concluded that this was because plant growth depended on phosphorus storage and had opposite ef fects on leaf carbon accumulation. Based on tree plasticity index and the coef ficient of variation, there was medium variation in element concentrations. The plasticity index of total carbon levels was the lowest, and that of potassium the highest. Principal component analysis and cluster analysis showed that the intra-specif ic variation among the four oak species was higher than inter-specif ic variation. From the perspective of nutrient supply and ecological adaptation, this study creates a foundation for the management of secondary oak forest stands.

Keywords Ecological stoichiometry · Quercus ·Elemental phenotype · Intra-specif ic variation · Interspecif ic variation · Ecological evolution · Nutrient supply

Introduction

The balance of nutrient elements af fects ecological interactions and ecosystem processes (Leal et al. 2017a), for example, plant–herbivore interactions (Filipiak and Weiner 2017)and host-parasite interactions (Aalto et al. 2015). As a f ield of studies of the balance of nutrients, ecological stoichiometry (ES) is a framework that focuses explicitly on the balances and f lows of elements within and between organisms and ecosystems (Lemmen et al. 2019). To explore the ecological consequences of evolutionary changes in these elements, a new concept, elemental phenotype, has recently been proposed, def ined as the composition, acquisition,assimilation, allocation and excretion of the 25 elements found in an organism (Jeyasingh et al. 2014). These biological processes are not only related to evolutionary changes but also to ecosystem processes, and are frontier topics in ecological stoichiometry (Jeyasingh et al. 2014; Leal et al.2017a; Lemmen et al. 2019). Ecological stoichiometry patterns are still important topics although ES is also a classical f ield. Numerous reports on ecological stoichiometry patterns are based on specif ic variations at the global level(Güsewell 2004; Mcgroddy et al. 2004; Reich and Oleksyn 2004; Wright et al. 2005; Ordoñez et al. 2009; Yuan and Chen 2009; Tian et al. 2018), at the regional level (Han et al.2005; He et al. 2006, 2008), at the local (Hättenschwiler et al. 2008; Zhao et al. 2014) scale, and on intra-specif ic variations across dif ferent regions (Hu et al. 2017; Abdala-Roberts et al. 2018) and sites (Martin et al. 2017; Zhang et al. 2019).

Oak forests are one of the major forest ecosystems in China, covering 6.72 million ha 2 (10.15%) with a standing stock of 1.3 billion m 3 (8.9%) (Zeng et al. 2016). Oak forests constitute widely distributed deciduous broad-leaved species, present mostly in young secondary forests with an age class of 20–40 years (Zhang et al. 2017b).Quercusis a highly diverse genus with more than 400 species worldwide, and more than 100 species in China (Chen and Huang 1998; Kremer et al. 2012; Denk et al. 2017). Because of the importance and universality of oak forests, numerous studies have been published recently on oak ecological stoichiometry. However, oak ecological stoichiometry studies have mainly included the following: (1) plant-litter-soil system: based on the investigation of C, N and P elements s in the plant-litter-soil system, a natural oak forest was demonstrated to be more adaptable to an arid site on the Loess Plateau in China compared to introduced species (Bai et al. 2019). In another study on ecological stoichiometry of a plant-litter-soil system in a mountain oak-pine mixed forest, the researchers advocated the selection of nitrogenf ixing species and a mixed forest structure in forest restoration owing to the limited levels of N and P in the soil(Sun et al. 2019); (2) element homeostasis: on phosphorousrich sites leaf nutrient levels but not element ratios were variable, indicating that plants are capable of maintaining element homeostasis (Zhou et al. 2015); (3) phosphorous limitation: a study on foliar phosphorous resorption inQ.variabilisBlume across sites revealed that this species had a level of phosphorus adaptability characterized by low resorption at the P-rich site and a high resorption at the P-def icient site (Ji et al. 2018). The tissue N:P ratios inQ.aquifolioidesRehd. & Wils. on Balang Mountain in China increased with elevation, indicating a potential P limitation for plant growth at high elevations (Wang et al. 2018). In broad-leaved deciduous forest on the Loess Plateau, foliar P of the dominant species,Q. wutaishanicaMayr, was positively related to total soil potassium (Xing et al. 2018); and,(4) variation patterns in altitude gradients, developmental stages, sites and provenances:Q.variabilisleaf N, K, and NRE (nitrogen resorption ef ficiency) increased with altitude,C and CRE (carbon resorption ef ficiency) decreased with altitude, and P and N:P did not change signif icantly with altitude. Regulation of green leaf element stoichiometry and resorption ef fects may contribute to the responses of the oaks to environment change (Du et al. 2017). In leaf tissues ofQ. variabilis, C concentrations and N:P were not age-dependent, whereas with stand age, N and P decreased and C:N and C:P increased (Jiang et al. 2017). Three simple experiments revealed an insignif icant provenance ef fect but a marked site ef fect with 43.5%, 44.0%, and 59.4% of the total variations explained by leaf N and P concentrations and N:P ratio, respectively (Zhang et al. 2016). These investigations involved several Chinese oak species, such asQ. variabilis(Sun et al. 2012, 2015; Cao and Chen 2015),Q. acutissimaCarr. (Noh et al. 2007; Wu et al. 2014; Zhang et al. 2017a),Q. wutaishanica(Qi et al. 2009),Q. aquifolioides(Wang et al. 2018), as well as foreign oaksQ. ilexL. (Sardans et al.2008; Oliet et al. 2011),Q. suberL. (Andivia et al. 2010),Q.rubraL. andQ. prinusL. (Lovett and Goodale 2011). Interspecif ic comparative studies on ecological stoichiometry are also reported for both native and alien oaks (Wu et al. 2012a;Zhang et al. 2018a).

Previous studies have explored ecological stoichiometry patterns, processes, and homeostasis of individual oak species or oak forests. However, investigations of inter- and intra-specif ic phenotypic variations of ES traits in a mixedoak secondary forest are lacking. In this study, a subtropical oak stand with four coexisting oak species,Q. fabriHance,Q. serratavar.brevipetiolata(A.DC.) Nakai,Q. acutissima, andQ. variabiliswas selected as the sampling site.Our objectives were: (1) to investigate ES traits of carbon(C), nitrogen (N), phosphorus (P) and potassium (K) of the four species; (2) to analyze relationships between element concentrations and growth performance; and, (3) to explore the inter- and intraspecif ic variation patterns of ES phenotypes within the four oak species.

Materials and methods

Site description

The study area, Yushan forest farm of Changshu, is located in the south of Jiangsu Province (31°36′ N, 120°40′ E) on the edge of the northern subtropical monsoon climate zone.

The area is characterized by four distinct seasons, with warm springs, cool autumns, hot and humid summers, and cold dry winters. Annual average temperature is about 15 °C,while the annual average precipitation is approximately 1062.5 mm, with more precipitation in summer than in winter. The yellow–brown soil has a pH of 5–6, and is moist and well-drained.

Field sampling and measurements

Samples were collected in late July during the growing season. The sampling site is a mixed plot with four deciduous oak species,Q. fabri,Q. serratavar.glandulifera,Q. acutissimaandQ. variabilis. In this stand, 292 samples of the four species were randomly collected. Healthy, disease free leaves without insect spots—were collected from the tree canopies and height (H) and diameter at breast height (DBH)of each were recorded. The sampled trees were divided into three growth stages according to the DBH: seedlings(DBH ≤ 1), saplings (1 < DBH ≤ 10) and trees (DBH > 10).The sampling data are shown in Table 1.

Chemical analysis

The fresh leaf samples were packed into zipper bags with silica gel desiccant; the samples were dried in a 60 °C oven in the laboratory, ground with a mortar with liquid nitrogen into a 0.15 mm powder. The ground samples were stored under dry conditions in self-sealing bags.

Three experiments, repeated three times, were set up for each sample. The concentration of total organic carbon (TC)was determined by the potassium dichromate volumetric method (Bai et al. 2019), and 0.02 g of sample was consumed for each experiment. To determine total phosphorus(TP), total nitrogen (TN), and total potassium (TK) concentrations, 0.2 g samples were f irst digested with H2SO4andH2O2in each experiment, and TP, TN and TK concentration were then determined by the colorimetric method (ultraviolet spectrophotometer), the Kjeldahl method and the f lame photometry method, respectively (Bai et al. 2019).

Table 1 Features of the sample collection

Data analysis

Microsoft Of fice Excel 2007, SPSS 21.0, and R studio 3.6.2 software were used for data statistics and analysis. Concentrations of TC, TN, TP and TK were expressed by mass fraction, and the stoichiometric ratios of the four elements expressed by mass ratio. Before data processing, SPSS software was used to test the normal distribution of C, N, P and K concentrations in oak leaves. The results showed that the concentrations of the four elements in 292 oak leaves basically were a normal distribution. SPSS21.0 was used to carry out an ANOVA for each variable for the four species: TC, TN, TP, TK, C:N, C:P, C:K, N:P, P:K and N:K,expressed as mean ± standard error. The correlation between the four elemental concentrations were tested and the dif ference calculated between the four species and the stoichiometric coef ficient of variation (CV%). Before the correlation analysis, the concentrations of each element were converted to logarithmic values to eliminate errors caused by variance and homogeneity. In order to study the stoichiometric plasticity and stability of oak, the phenotypic plasticity index(PPI) of the above indicators was calculated as:

where P max is the maximum phenotypic value; P min is the minimum phenotypic value (Zhang et al. 2011). The value of PPI is between 0 and 1; the higher the value, the greater the plasticity (Yang et al. 2016a).

A correlation analysis between the leaf stoichiometric traits and the growth status of trees was carried out by R studio3.6.2 before testing the logarithmic conversion of H,DBH versus the concentration of each element. Leaf stoichiometric traits were then compared at dif ferent growth stages.A principal component analysis and cluster analysis of the stoichiometric traits and stoichiometric ratios were carried out with R studio3.6.2.

Results

Stoichiometric traits of leaves of four oak species

The mean concentrations of carbon, nitrogen, phosphorous and potassium were 297.1 ± 4.2, 17.9 ± 0.3, 0.8 ± 0.02 and 5.1 ± 0.1 mg g −1 , respectively (Table 2). Total carbon varied as:Q. serratavar.glandulifera 0.05).

Mamun did come to my roof one day to talk to me but I wanted him to go away. I didn t want any one to see us talking. As you know, in Bangladesh rumors8 go around so fast. When we talked, I saw deep love in his eyes. I always smiled at him; I didn t talk to him much. Still, life was going on so wonderfully. Mamun never told me he loved me. I thought that was because, I was 5/6 years younger than him.

Stoichiometric ratio of leaves in four oak species

The mean values of C:N, C:P, C:K, N:P, N:K and P:K ratios were 18.8 ± 0.6, 414.6 ± 10.8, 73.7 ± 2.5, 24.8 ± 0.7,4.5 ± 0.2 and 0.2 ± 0.004, respectively (Table 3). There were signif icant dif ferences in C:P and C:K ratios among the four species (p≤ 0.01), while there were no inter-specif ic dif ferences in C:N, N:P and P:K (p> 0.05). In the four species, there was a correlation between carbon and nitrogen levels (R2 = 0.0262,p< 0.05, Fig. 1 a), a signif icant correlation between phosphorous and potassium concentrations (R2 = 0.3741,p< 0.01, Fig. 1 b), and no correlation between other elements.

Correlation between growth performance and element concentrations

Concentrations of the dif ference elements varied signif icantly between the three size classes (seedling, sapling and tree). Average P and K levels in each size class were sizedependent: seedling < sapling < tree (Fig. 2 c, d). Carbon and nitrogen concentrations did not show this pattern; there was no relationship between N concentrations and tree size(Fig. 2 a, b).

The correlation between C, P and K concentrations and size class were further analyzed (Fig. 3). Consistent with the results in Fig. 2, except for nitrogen (Fig. 3 b, f), levels of the other elements were correlated with height and diameter(p< 0.01). However, there was no correlation between C, P,K levels and growth stage. Carbon levelsshowed a signif icantly negative correlation with height ofQ. fabri,Q. serratavar.glanduliferaandQ. variabilis(Q. fabri:R2 = 0.0680,p= 0.01;Q. serratavar.glandulifera:R2 = 0.2064,p< 0.01;Q. variabilis:R2 = 0.0869,p= 0.01, Fig. 3 a). It was also negatively correlated with the DBH of these three species(Q. fabri:R2 = 0.0710,p= 0.01;Q. serratavar.glandulifera:R2 = 0.2146,p< 0.01;Q. variabilis:R2 = 0.0720,p= 0.01,Fig. 3 e). Phosphorous concentrations were positively correlated with height ofQ. fabriandQ. variabilis(Q. fabri:R2 = 0.0494,p= 0.03;Q. variabilis:R2 = 0.3665,p< 0.01,Fig. 3 c), and signif icantly positive correlation with diameter ofQ. fabriandQ. variabilisandQ. acutissima(Q. fabri:R2 = 0.0612,p= 0.02;Q. variabilis:R2 = 0.2892,p< 0.01;Q. acutissima:R2= 0.1679,p< 0.01, Fig. 3 g). Pottasium levels were positively correlated with height (R2 = 0.1281,p< 0.01, Fig. 3 d) and DBH (R2 = 0.1191,p< 0.01, Fig. 3 h)inQ. variabilis.

Table 2 Concentrations of total carbon, total nitrogen, total phosphorous, total potassium in leaves of four oak species

Table 3 Leaf stoichiometric ratios of all individuals of the four species of oaks

Intraspecif ic variations of carbon, nitrogen,phosphorous and potassium in the four oak species

According to the violin diagrams of the stoichiometric traits of the four species, N concentrations were relatively similar. Distribution of C, P and K betweenQ. fabriandQ. variabilisand C and P inQ. serratavar.glanduliferawere scattered with a wide range of variations. Distribution of C and N inQ. serratavar.glanduliferawas relatively uniform, while P and K were highly variable. There was an obvious discrete value of K concentration inQ.acutissima(Fig. 4).

The ecological stoichiometry of the four species varied considerably because of internal physiological status and natural external environmental factors at the Yushan sample site. The phenotype plasticity index (PPI) and coef ficients of variation (CV) of the elemental phenotypes (Eps) of the four species were calculated (Fig. 5). The PPI were slightly dif ferent in the four oaks, notably the small plasticity index of carbon levels ofQ. acutissima. The PPI of K and N were the largest inQ. variabilis,while C and P were the largest inQ. fabri(Fig. 5 a). The greater the phenotypic plasticity,the larger the inf luence of external environmental factors on oaks. Comparing dif ferent elements, the PPI of carbon was small and its concentration relatively stable; the PPIs of K and P were relatively large and were signif icantly af fected by environmental factors. Coef ficient of variation of K had the highest variation except forQ. acutissima,and for C it was the lowest variation followed by N. ForQ. acutissima,carbon levels varied little while potassium concentrations were highly variable. For the other oak species, the concentrations of the four elements varied moderately (20%–50%,Fig. 5 b) (Hu et al. 2019).

PCA analysis of stoichiometric traits in the four oaks species

Figure 6 shows the concentration of elements of the four oak species and the stoichiometric ratios of these elements,a total of 10 variables. According to the screen test, the four principal components explain best the phenotypes of the four elements and the stoichiometric ratios among them (Fig. S1).The f irst four principal components accounted for 94% of the variance (PC1 accounted for 42%, PC2 for 22%, PC3 for 18%, and PC4 for 12%). The N:K ratio contributed the most to the ranking model of PC1, the C:N ratio the greatest contribution to PC2, the P:K ratio the most to PC3, and carbon levels contributed the most to PC4. On these four principal component axes, the four oak species did not show an obvious grouping (Fig. 6). Intra-specif ic variation of the four oak species was greater than interspecif ic variation (Table S1).

Fig. 1 Correlations between a total carbon and total nitorgen concentrations and b total phosphorous and total potassium concentrations for the four oak species ( Q. fabri, Q. serrata var. glandulifera, Q.acutissima and Q. variabilis are represented by green, blue, red and purple, respectively)

Cluster analysis of stoichiometric traits in four oak species

Previous studies have conf irmed that inter-specif ic introgression is a common phenomenon inQuercus, and similar variations may occur when four species of oaks are present in the same habitat. Cluster analysis was carried out on the values and ratios of four elements for the four oak species.Before the cluster analysis, two optimal cluster numbers were tested,one contained 289 individuals, the other only three individuals – two individuals ofQ. variabilisand one individual ofQ. serratavar.glandulifera.

The cluster analysis showed that the 292 individuals of the four species of oaks in the Yushan secondary forest could not be grouped according to species. However,Q. acutissimatended to be grouped together withQ. variabilis, whereasQ.fabriandQ. acutissimawere rarely grouped together in the smaller cluster unit (Fig. 7). Figure 4 shows the clustering of the average element concentrations of the four species at dif ferent growth stages; the clustering distance betweenQ.acutissimaandQ. fabriwas high and their relationship was also more distant. This is consistent with the most recent results ofQuercusgrouping.

Discussion

Nutrient status of the forest stand

Fig. 2 Concentrations of a total carbon, b total nitrogen, c total phosphorous, d total pottasium in the three growth stages of seedling, sapling and tree

Several studies have reported on element concentrations inQ. acutissimaandQ. variabilis(Table 4), however few onQ.fabriandQ. serratavar.brevipetiolatawere found. For the latter species, only one study on a small sample (n < 4) was reported (Wu et al. 2012a). Moreover, most of these studies only concerned nitrogen and phosphorous levels. Therefore,this study was the f irst analysis of ecological stoichiometric traits of four elements in a multi-oak mixed stand. Nitrogen and phosphorus concentrations were generally lower than or close to previously reported values for oaks, e.g.,Q.acutissima(Wu et al. 2014; Zhang et al. 2016),Q. variabilis(Sun et al. 2015), andQ. wutaishanica(Xing et al. 2018).The values were also lower than the average values from world f lora (Güsewell 2004; Tian et al. 2018) and Chinese f lora (Han et al. 2005; Ren et al. 2007; Yang et al. 2016b),except for evergreen woody plants (Güsewell 2004). The N:P ratio obtained in this study was similar to that of oak species studied by Zhang et al. ( 2016), Wu et al. ( 2014) and Sun et al. ( 2015), and higher than that compared to world f lora (Güsewell 2004; Tian et al. 2018) and Chinese f lora(Han et al. 2005; Ren et al. 2007). Foliar nitrogen and phosphorus may be af fected by many factors such as organ type,geographical location and site conditions. Leaf nitrogen is usually higher in non-photosynthetic organs such as stems and fruits (Bai et al. 2019), while it increased with altitude;phosphorous and N:P ratios did not change signif icantly with altitude (Du et al. 2017). Leaf nitrogen and phosphorus declined with age, while N:P ratios did not vary signif icantly among age classes (Jiang et al. 2017). Oaks on phosphorousrich sites had higher nitrogen and phosphorus levels than those on phosophorous-def icient sites (Zhou et al. 2015; Ji et al. 2018; Yuan et al. 2019). According to previous studies on plant nutrition (Koerselman and Meuleman 1996; Aerts and Chapin 2000; Wu et al. 2012b; Jiang et al. 2017), when leaf nitrogen levels are < 20 mg g –1 , phosphorous concentrations < 1 mg g −1 , and N:P ratios > 16, the nutritional status of the plant can be interpreted as phosphorus limited. However,when leaf nitrogen concentrations are < 20 mg g−1, nitrogen is limited; when the N:P ratio > 20, phosphorus is limited(Güsewell 2004). In this study, the oak nutrition showed signif icant phosphorus limitations.

Fig. 3 Correlations between a height (H) and total carbon, b total nitrogen, c total phosphorous, d total potassium concentrations, DBH and e total carbon, f total nitrogen, g total phosphorous, h total potassium concentrations in four oaks ( Q. fabri, Q. serrata var. glandulifera, Q. acutissima and Q. variabilis are represented by green, blue,red and purple, respectively)

Fig. 4 Violin diagrams of a total carbon, b total nitrogen, c total phosphorous, d total potassium of the four species ( Q. fabri, Q. serrata var.glandulifera, Q. acutissima and Q. variabilis are represented by green, blue, red and purple, respectively)

Carbon levels of the four oaks in this study were markedly lower than in previously studied oak species(Sardans et al. 2008; Cao and Chen 2015; Sun et al. 2015,2019; Zhou et al. 2015; Du et al. 2017; Jiang et al. 2017;Ji et al. 2018; Bai et al. 2019). This may be because the method used in this study is different from that reported in the literature. C:N ratios forQ. acutissima,Q. variabilis,Q. fabri, andQ. serratavar.brevipetiolatawere 19.8, 20.6, 17.2, and 17.9, respectively, while the C:P ratios were 482.3, 421.5, 373.1, and 414.1, respectively.Lower C:N ratios and median C:P ratios in oaks were found compared with the literature, e.g.Q. wutaishanica(C:N = 31.8; C:P = 222.0) (Bai et al. 2019),Q. acutissima(C:N = 36.3; C:P = 301.7) (Sun et al. 2019), andQ. ilex(C:N = 38.0; C:P = 444.0) (Sardans et al. 2008). Pottasium concentrations of the four species were lower than those from aQ. wutaishanicaforest on the Loess Plateau(6.7 mg g –1 ) (Xing et al. 2018), and forQ. variabilison phosphorous-rich sites (5.7 mg g –1 ) (Zhou et al. 2015) or from across China (6.5 mg g –1 ) (Sun et al. 2015). Concentrations of pottasium inQ. acutissimaandQ. serratavar.brevipetiolatawree no higher than inQ. variabilison phosphorous-deficient sites (5.0 mg g −1 ) (Ji et al. 2018),(4.8 mg g −1 ) (Zhou et al. 2015). In this study, N:K ratios were higher than other values in the literature, e.g.,Q. ilexand deciduous oaks (Ferretti et al. 2014).

Relationship between nutrient concentrations and tree size

Do nutrient concentrations among plant individuals vary with different sizes or ages? Ecological stoichiometric patterns may be explained by the growth rate hypothesis(Elser et al. 2000; Sterner and Elser 2002). This is based on the concept that rapidly growing organisms can uplift phosphorous allocation, up-regulate levels of phosphorousrich ribosomal RNA, promote protein synthesis, and support rapid growth. As a result, fast-growing organisms have low biomass C:P and N:P ratios. Another explanation for the relationship between carbon storage and tree size is that productivity declines on the scale of leaves with size and age increasing. The decline in growth is sometimes inferred from life-history theory to be an inevitable result of increasing resource allocation to reproduction (Stephenson et al.2014). One set of theories is that the decrease of growth after the start of reproduction is caused by cellular senescence;another set of theories concerns the larger tree size increases the burden of respiration (Mencuccini et al. 2005). In this study, logarithm values of foliage phosphorous or potassium levels were positively correlated with the logarithm values of height and diameter, which means that high concentrations of P or K can promote oak growth. In other words,the growth of oaks possibly depends on phosphorous and potassium leaves in leaves. The relationship between carbon concentrations and size was opposite to that between P and K, while the relationship between nitrogen concentration and size was insignif icant.

Fig. 5 a Plasticity indices and b coef ficient of variations of the elemental phenotypes of the four species (potassium, nitrogen, carbon and phosphorous are represented by green, blue, red and purple, respectively)

Fig. 6 All 292 individuals of the four oaks distribution def ined by a PC1 and PC2,b PC3 and PC4 ( Q. fabri, Q.serrata var. glandulifera, Q.acutissima and Q. variabilis are represented by green, blue, red and purple, respectively)

Fig. 7 Cluster analysis of stoichiometric traits in all 292 individuals of four oaks ( Q. fabri, Q. serrata var. glandulifera, Q. acutissima and Q. variabilis are represented by red, purple, blue and green, respectively)

Concentrations of nitrogen or phosphorus reported in the literature were mostly negative with regards to size. In 20 naturally regenerated secondary oak forests, leaf N and P levels decreased with age, and the concentrations between young and mature stands were signif icantly dif ferent (Jiang et al. 2017). Noh et al. ( 2007) found that, among three Korean oak stands (with mean ages of 10.8, 38.2, and 44.0),there was no signif icant dif ference in leaf nitrogen levels,while leaf phosphorous concentrations (0.11%, 0.07% and 0.08%) were markedly dif ferent. However, as they pointed out, the ages were not equally distributed in the study. InEucalyptusplantations with an age sequence of two, four and six years in subtropical China, there was no statistically signif icant dif ference in leaf nitrogen and phosphorous between stands of dif ferent ages, though the researchers indicated that phosphorous limitation increased with stand age (Fan et al. 2015). Some reports also found a concentration of elements was positively related to size. From young to mid-age,near-mature, and mature forests, leaf N, P, and K concentrations inPicea schrenkianaFisch. & C.A. Mey increased signif icantly (Sun et al. 2018). In a 13-year-old plantation ofMachilus pauhoiKaneh., concentrations of leaf carbon,nitrogen, and phosphorus were signif icantly higher than in a 9-year-old stand (Zhang et al. 2015). It is dif ficult to explain this discrepancy but it may be that slow-growing species have a phosphorus storage mechanism. Trees may have evolved long-term adaptive mechanisms to store phosphorous in biomass and can accumulate more phosphorous than nitrogen with size, especially for slow-growing, long-living species of late successional stages (Sardans and Peñuelas 2015). In Borneo camphor,Dryobalanops aromaticaGaertn. F., stored phosphorus accounted for 67.7% of the total phosphorus requirements for reproduction (Ichie and Nakagawa 2013). Poplar hybrids had higher uptake and storage potential for phosphorus, while willow hybrids displayed higher phosphorus resorption ef ficiencies (Ros et al. 2018).In this study, we found that nitrogen levels of four oak species had no correlation with tree size. Growth depended on nitrogen and phosphorous concentrations, similar to results of previous studies (Noh et al. 2007; Zhang et al. 2015; Sun et al. 2018). In a study ofEucalyptusplantations with different age sequences in subtropical China (Fan et al. 2015)and another of Schrenk’s spruce (P. schrenkianaFisch. &C.A. Mey.) from young to mid-age, near-mature and mature in the temperate zone in China (Sun et al. 2018), carbon reserves f irst increased and then decreased. During growth,carbon assimilation capacity was enhanced, the C concentration was the highest in a near-mature forest, and gradually decreased with individual size increasing. A global analysis of 403 tropical and temperate tree species showed that size had an opposite ef fect on leaf carbon accumulation (Stephenson et al. 2014). This study indicated that C concentration was negatively correlated with tree size, which was similar to the results of several studies on biomass at the leaf scale (Mencuccini et al. 2005; Noh et al. 2007; Zhang et al.2009; Stephenson et al. 2014). The decline of carbon may be caused by several factors. One explanation is that nutrient uptake in older stands decreases because more nutrients are sequestered in woody biomass (Rosenvald et al. 2013),and large trees have a lower carbon return on investment in leaf biomass than smaller trees (Zhang et al. 2009). Is the ef fect of size on ecological stoichiometry due to age or size itself? Most studies believe that, due to the metabolic scale theory, size itself has an independent and strong inf luence(Mencuccini et al. 2005; Noh et al. 2007). In a study of threeQ. acutissimastands with dif ferent age sequences and tree sizes, size had a more signif icant ef fect than age, and evidence indicated that size accounted for the inf luence of relative growth rates and net assimilation rates per unit leaf area (Noh et al. 2007). However, it is dif ficult to determine the age limit of mature trees, especially the mixed-oak secondary forest in Yushan as it contains a mixture of plantation and natural forests. In the middle of the last century, oak forests undergone a renewal due to severe disturbance. Some trees may have died or been removed before maturity (Ren et al. 2010) and it is dif ficult to unify the age of the forest in short-term research. Therefore, the correlation between tree size and ecological stoichiometry traits is complex and there is the function of multicollinearity which is why linear f itting models explained only part of the variance (the values of R 2 were low), although there were statistical correlations between tree size and ES in our study (p< 0.05). However,this question is still unresolved and requires further in-depth study.

Table 4 Comparision of leaf nitrogen, phosphorous concentrations (mg g −1 ) and N:P ratios in oaks and in other plants

Element phenotypic variation in oaks

Elemental phenotype (EP) is a new concept proposed by hydrobiologists which has been applied to aquatic ecosystems (Tuckett et al. 2016; Leal et al. 2017b). It is also called stoichiometric phenotype (Leal et al. 2017a, b) or chemical phenotype which refers to the compound types (Berenbaum and Zangerl 1998; Mandolino et al. 2003; de Meijer et al.2003). According to the def inition, EP involves 25 elements and def ines f ive aspects for each element: composition,acquisition, assimilation, allocation, and excretion (Jeyasingh et al. 2014; Leal et al. 2017a). Although this concept is still unfamiliar to many researchers on terrestrial ecosystems, it is an interesting and important f ield. With the development of high-throughput phenomics and ionomics, the majority of the 25 elements can be determined rapidly and quantitatively (Jeyasingh et al. 2014). The study of elemental or stoichiometric phenotypes is of considerable theoretical signif icance. Changing nutrition supply may impact f itnessrelated traits of organisms and alter allocation trade-of fs, and thus can af fect life-history evolution (Boersma and Elser 2006; Snell-Rood et al. 2015). Intraspecif ic genetic variation in stoichiometry may change under dif ferent selection regimes (stabilizing, directional, or disruptive selection),resulting in the dif ferentiation of EPs with a genetic basis and the formation of phenotypic diversity (Jeyasingh et al.2014). On the other hand, phosphorus stress stimulated the production of adventitious roots in phosphorus-ef ficient genotypes of common bean (Phaseolus vulgarisL.) rather than phosphorus-inef ficient genotypes, so as to obtain phosphorus from topsoil horizons ef fectively and extensively (Miller et al. 2003).

Phenotypic variations of elements with a genetic basis have important ecological and evolutionary consequences.Observed phenotypic variation in living organisms is shaped by genomes, environment, and their interactions (Li et al.2018). Three components, phenotypic plasticity, genetic variation, and genetic variation in plasticity, may contribute to intra-population variation in EP (Leal et al. 2017a). It is of great ecological and evolutionary signif icance to study elemental phenotypic variations of woody plants in terrestrial ecosystems. In woody plants, element concentrations are also potentially af fected by gene, site, and their interactions.InQ. acutissima, 44.0%, 7.1 and 14.0% of all variations in leaf phosphorous were explained by site, provenance, and site × provenance, respectively (Zhang et al. 2016). Genotype variability of P and K concentrations inQ. roburwere more pronounced than nitrogen variability, and coef ficients of variation of P and K were 5.9% and10.7%, respectively(Nikolic et al. 2006). We obtained a complete data set of elemental phenotypic variation ofQuercus. Although our study concerned only one forest stand, it contained 292 individuals of four oak species. The results revealed that phenotypic variation had two basic characteristics: f irst, the degree of plasticity of the four elements was dif ferent, with the lowest variation in carbon, the highest in potassium, and the median in nitrogen and phosphorus; second, intra-specif ic variation was higher than inter-specif ic variation. It was expected that intra-population variations should be higher than interpopulation variations for a given oak species. The variation patterns of intra-vs.inter-specif ic, intra-vs.inter-population are consistent with the trend of nuclear genetic variations(Zhang et al. 2018b; Li et al. 2019).

The abundant stoichiometric variations of four oak species in our study not only lay a foundation to explore the genetic basis and evolutionary implications of ecological stoichiometry (Jeyasingh and Weider 2007; Leal et al.2017b), but also provide clues for genetic improvement(Ramaekers et al. 2010; Veneklaas et al. 2012). This will provide important theoretical and technological support for the management of oak forests.

Conclusions

The examination of the stoichiometric traits of four elements in Yushan oak species produced three conclusions. First, leaf stoichiometric traits of the four oak species were relatively low. Their nutritional status was interpreted as phosphorus limiting. Second, the growth of oak trees was determined by and depended upon phosphorous and potassium concentrations in leaves to some extent. Carbon may have the opposite ef fect on tree size. But the ef fect of element concentration on growth regulation is complicated due to the size of the tree itself or to the age factor. Phosphorous storage plays an important role, especially in nutrient absorption and biomass accumulation of Yushan oak trees. Third, at the element level, the phenotypic plasticity of carbon was the lowest and potassium the highest. At the species level, intra-specif ic phenotypic variation was greater than that of inter-specif ic phenotypic variation.

These results provide information on ecological conditions in the research area such as the availability of some elements for oaks or the health of plants grown locally. The results also provide theoretical support for the management of oak forests and a foundation for the genetic and evolutionary functions of ecological stoichiometry. In order to further explore the ecological processes of secondary mixed-oak forests in Yushan, soil nutrient status and nutrient absorption ef ficiency should be studied. Previous studies could not provide a satisfactory explanation for the correlation between element concentrations and growth. This remains to be further explored as element levels are vital in protein synthesis and accumulation. Elemental phenotypes play an important role in environmental adaptation and ecological evolution of plants. Their dif ferentiation involves interaction between genetics and ecological niche. It is still necessary to integrate high-throughput molecular methods to explore the genetic diversity of Yushan oak to further study their intra- and inter-specif ic variation, and to explore the process of ecological succession of the Yushan secondary mixedoak forest.

AcknowledgmentsWe gratefully acknowledge the Yushan forest farm of Changshu for granting access to the farm to carry out the study. We appreciated valuable comments and insights from Lei Xie,Xiaoxu Guo, Yao Li, and Lu Wang.

Author’s contributionsMs. YF and Dr. XL contributed equally to this work. They conceived and designed the experiment. Ms. YF and Ms. BZ conducted the laboratory sample analysis; Ms. YF and Dr. XL analyzed the data; Dr. XL, Dr. RH and Prof. HZ contributed analysis methods/tools; Ms. YF and Dr. XL drafted the manuscript; Prof. YF contributed to manuscript revision. Prof. JY and Prof. YF contributed project administration and funding acquisition.

FundingThe work was supported by National Natural Science Foundation of China (No 3177030224) and Jiangsu Forestry Science and Technology Innovation and Promotion Project (LYKJ[2017]025)