Global leaf nitrogen and phosphorus stoichiometry and their scaling exponent

Global leaf nitrogen and phosphorus stoichiometry and their scaling exponent Abstract Leaf nitrogen (N) and phosphorus (P) concentrations constrain photosynthetic and metabolic processes, growth and the productivity of plants. Their stoichiometry and scaling relationships regulate the allocation of N and P from subcellular to organism, and even ecosystem levels, and are crucial to the modelling of plant growth and nutrient cycles in terrestrial ecosystems. Prior work has revealed a general biogeographic pattern of leaf N and P stoichiometric relationships and shown that leaf N scales roughly as two-thirds the power of P. However, determining whether and how leaf N and P stoichiometries, especially their scaling exponents, change with functional groups and environmental conditions requires further verification. In this study, we compiled a global data set and documented the global leaf N and P concentrations and the N:P ratios by functional group, climate zone and continent. The global overall mean leaf N and P concentrations were 18.9 mg g−1 and 1.2 mg g−1, respectively, with significantly higher concentrations in herbaceous than woody plants (21.72 mg g−1 vs. 18.22 mg g−1 for N; and 1.64 mg g−1 vs. 1.10 mg g−1 for P). Both leaf N and P showed higher concentrations at high latitudes than low latitudes. Among six continents, Europe had the highest N and P concentrations (20.79 and 1.54 mg g−1) and Oceania had the smallest values (10.01 and 0.46 mg g−1). These numerical values may be used as a basis for the comparison of other individual studies. Further, we found that the scaling exponent varied significantly across different functional groups, latitudinal zones, ecoregions and sites. The exponents of herbaceous and woody plants were 0.659 and 0.705, respectively, with significant latitudinal patterns decreasing from tropical to temperate to boreal zones. At sites with a sample size ≥10, the values fluctuated from 0.366 to 1.928, with an average of 0.841. Several factors including the intrinsic attributes of different life forms, P-related growth rates and relative nutrient availability of soils likely account for the inconstant exponents of leaf N vs. P scaling relationships. leaf, nitrogen, phosphorus, stoichiometry, scaling exponent, functional group, biogeography INTRODUCTION Plant growth and metabolism depend on ribosome numbers and protein concentrations in living cells [1,2]. Nitrogen (N) and phosphorus (P), especially the N in Rubisco that drives photosynthesis and the P in ribosomal RNA that drives the generation and maintenance of proteins, are essential nutrients [3–6] that are consequently tightly linked and important parameters in stoichiometric growth models [7–9]. Therefore, the leaf N and P stoichiometric patterns including N and P concentrations and N:P ratios are largely explored both at regional and global levels [10–13], which are important bridges linking elemental compositions or allocations with organismal metabolic processes, and even energy flow in the whole ecosystem [9]. In particular, the strong correlation between leaf N and P concentrations can be quantified via a stoichiometric scaling relationship described by a power function as N = βPα, where α and β indicate the slope (i.e. the scaling exponent) and the ‘elevation’ or Y-intercept (i.e. normalization constant), respectively, of the log–log linear leaf N concentration vs. P concentration regression line [14,15]. As is the case for other allometric ‘rules’ in ecology, leaf N vs. P scaling provides a simple but useful empirical model [8,9,15] in which the exponent is considered to be critical for predictions of plant and ecosystem functioning [16]. For example, a previous stoichiometric growth model (the growth rate hypothesis) posited that organisms with higher growth rates required disproportionately higher P than N concentrations, resulting in a scaling exponent of N to P concentrations below unity (i.e. α < 1.0) [1]. The large-scale patterns of the leaf N and P stoichiometry in relation to environmental conditions (especially temperature), biogeochemical gradients, intrinsic genetic factors and species composition have been generalized in previous studies [10–12,17–19]. Moreover, a constant exponent of leaf N vs. P scaling is appealing to, and chased by, ecologists for its simplicity in model operations. For example, empirical studies have reported that N concentrations scale, on average, across species as a three-quarter-power function of the P concentrations [20]. If the invariant exponent is true, the plant growth rate could be predicted by leaf N and P stoichiometry [8,21]. Nevertheless, a core issue remains: whether the N vs. P scaling exponent is invariably ‘constant’, or whether its numerical value depends on species functional groupings or other ecosystem properties. Numerous studies have also reported N vs. P scaling relationships with statistically significant variation in the scaling exponent [11,14,15,22,23]. For example, Niklas and Cobb [24] suggested that the scaling exponents of woody and herbaceous species are, on average two-thirds and three-quarters, respectively. Based on an extensive worldwide collection of leaf N and P concentrations, Wright et al. [14] claimed that the N vs. P scaling exponent was approximately two-thirds. In contrast, using a larger leaf N and P data set consisting of 7445 entries compiled by Reich and Oleksyn [10], Niklas [21] reports that the scaling exponent is 0.73, approaching three-quarters rather than two-thirds. In addition, several other studies have indicated that the exponent was approximately 0.67 [11], 0.72 [22], 0.78 [25] or 1.0 [26]. Based on what is to our knowledge the most comprehensive data to date, with 9300 pairwise observations of leaf N and P concentrations, Reich et al. [15] found similar two-thirds exponents across biomes, taxonomic divisions and angiosperm life forms, and therefore proposed the general two-thirds-power law (N ∝ P2/3) across major plant groups and biomes. If true, a simple stoichiometric scaling relationship exists governing leaf stoichiometry and metabolism, despite differences among specific case studies. However, similar to the inconstant scaling of leaf respiration with N and P [27] and the biological quarter-power scaling [28], any N vs. P scaling relationship derived from a large global data set may hide important biological variations rooted in species-dependent, or region- or site-related, differences. According to some stoichiometric scaling models, plant growth rates are purported to influence the N vs. P scaling relationship [1,21,29]. If true, plants from different functional groups or different geographical locations should have different growth rates due to specific evolutionary traits and environmental conditions, resulting in potential differences in their N vs. P scaling exponents. Additionally, according to some growth rate hypotheses [5,8,30], growth rates are purported to be more closely correlated to P concentrations than N concentrations, such that changes in P (and not N) demand or status are predicted to be a strong driver for variation in the numerical values of N vs. P scaling exponents. To test the generality of the leaf N and P stoichiometry and the constancy of the leaf N vs. P scaling exponent, we firstly explore the global leaf N and P stoichiometry that may be used as a basis for the comparison of other individual studies. We then determine the N vs. P scaling exponent for different functional groups (i.e. woody species and herbs) and spatial scales (i.e. latitudinal zones, ecoregions/continents and local sites), to evaluate whether the numerical value of the governing scaling exponent varies as a function of species composition, spatial distribution, or other variables of interest (e.g. climate conditions, and soil-relative N and P availabilities) as sometimes reported by others [10,18,31,32]. Furthermore, we explore the patterns of scaling exponent numerical variation across different sites to test whether the global N vs. P scaling relationship obscures site-related significant differences. We hypothesize that the scaling exponent at each site should reflect site-specific N vs. P stoichiometric relationships, because plants growing at the same site represent the characteristics shaped by the combination of local climatic conditions, geological processes, soil nutrient availabilities and other environment factors. GLOBAL DATA SET AND STATISTICAL ANALYSES In order to carry out the aforementioned studies, we compiled a large and geographically comprehensive global data set of pairwise leaf N and P concentration distributions, including global, regional and site-level records, for as many of the variables of interest as possible. We adopted only those records reporting paired N and P concentrations of green leaves with detailed location information, and excluded all records without site information or with unpaired N–P records. Using a detailed review of the literature, our own field sampling and the open TRY data set (Table S1; https://www.try-db.org) [33], a total of 12 055 records were acquired spanning 486 sites worldwide, in which 142 sites had more than 10 records for each site and 94 sites had more than 20 records for each site. Duplicated records were removed. The data set included 3441 terrestrial plant species in 1342 genera and 222 families. All of the plant samples in the data set were collected during the growing season. We used the Flora of China (http://frps.eflora.cn/), Useful Tropical Plants (http://tropical.theferns.info/), Australian Native Plants (https://www.anbg.gov.au/index.html) and Wikipedia (https://en.wikipedia.org/wiki) websites to identify plant functional groups and verify taxonomic classifications. We conducted similar statistical analysis to Reich et al. [15]. To perform data analyses, we log10-transformed the N and P concentration data for all plants in the data set, and then used reduced major axis (RMA) regression [34] to determine the N vs. P scaling relationship at four levels (i.e. species functional groupings, latitudinal zones, ecoregions and individual sites). Therefore, we first classified plants into herbaceous and woody species, and further divided woody plants into coniferous, deciduous broad-leaved and evergreen broad-leaved woody species. We then grouped the data from sites into tropical (0–25°), temperate (25–50°) and boreal (>50°) zones, to compare the N vs. P scaling exponents among these zones. We also analysed the latitudinal patterns of the scaling exponent for each of the functional groups, but excluded those of coniferous plants because most of the coniferous plants in our data set were distributed in the temperate zone. Our data analysis was then focused on patterns among ecoregions (continents), including North America (the United States of America, Canada and Mexico), Europe (Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden), Asia (China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia), Oceania (Australia and New Zealand), Africa (South Africa, Uganda and Cameroon) and South America (Brazil, Bolivia and Venezuela). Lastly, we analysed the N vs. P scaling relationships at each site (with n ≥ 10 and n ≥ 20 records) and quantified the numerical variation of the scaling exponent across different sites. We used a likelihood ratio test to evaluate the heterogeneity of RMA regression exponents within the aforementioned levels of analyses [34]. Additionally, we performed general linear regressions to explore the changes in the scaling exponents with the geometric mean N and P concentrations and N:P ratios, using the statistical package R 2.15.2 [35]. GLOBAL LEAF N AND P STOICHIOMETRY ACROSS DIFFERENT SCALES We used the above-described global data set of 12 055 pairwise leaf N and P concentration records to characterize large-scale leaf N and P stoichiometry by functional group, latitudinal zone, ecoregion and site. The geometric mean values of leaf N and P concentrations, and N:P mass ratios, of the pooled data were 18.9 mg g−1 and 1.2 mg g−1, and 15.8, respectively, but these numerical values differed significantly among the contrasting functional groups (Table 1). Compared to woody plants, herbaceous plants showed significantly higher N and P concentrations (21.72 mg N g−1 vs. 18.22 mg N g−1; 1.64 mg P g−1 vs. 1.10 mg P g−1), and lower N:P ratios (13.3 vs. 16.6). The mean leaf N (and P) concentrations of coniferous, deciduous broad-leaved and evergreen broad-leaved woody species were 12.13 (and 0.98), 21.13 (and 1.37), and 15.45 (and 0.79) mg g−1, respectively. The corresponding N:P ratios for the three woody groups were 12.4, 15.4 and 19.5, respectively. Table 1. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants for different functional groups. Note: woody plants were divided into three groups: coniferous gymnosperms and two angiosperm groups, deciduous broad-leaved and evergreen broad-leaved woody. Mean indicates the geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b and c) denote significant difference (p < 0.05) among functional groups based on a likelihood ratio test. Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 View Large Table 1. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants for different functional groups. Note: woody plants were divided into three groups: coniferous gymnosperms and two angiosperm groups, deciduous broad-leaved and evergreen broad-leaved woody. Mean indicates the geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b and c) denote significant difference (p < 0.05) among functional groups based on a likelihood ratio test. Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 View Large The leaf N and P stoichiometry also changed significantly with latitude. For the pooled data, the leaf N and P concentrations significantly increased from the tropical to boreal regions, but N:P ratios decreased (Table 2). Specifically, the geometric mean value of leaf N concentration was 17.41 mg g−1 in the tropical region, 19.24 mg g−1 in the temperate region and 19.83 mg g−1 in the boreal region. The geometric mean value of leaf P concentration was 0.83 mg g−1 in the tropical region, 1.28 mg g−1 in the temperate region and 1.49 mg g−1 in the boreal region. The geometric mean value of leaf N:P mass ratio was 21.0, 15.1 and 13.2 in the tropical, temperate, and boreal regions, respectively. For each of the functional groups, the latitudinal patterns of N–P stoichiometry were generally consistent with the pooled data set, i.e. leaf N and P concentrations increased, but leaf N:P ratios decreased from tropical, to temperate, to boreal regions (Table 2). Table 2. Summary of RMA regression results between leaf N and leaf P concentrations (e.g. log10 leaf N = α log10 leaf P + log10β), the statistics of leaf N and P concentrations, and the N:P mass ratio in terrestrial plants along the latitudinal zones. The exponent of herbs in the tropical zone was excluded in the comparison due to the paucity of samples in our data set. We also did not consider the latitudinal pattern of conifers because most of the coniferous samples in our data set were distributed in the temperate zone. All the plants, including conifers and tropical herbs, were pooled together during the evaluation of the latitudinal pattern for all plants. Mean indicates geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b, c) denote significant difference (p < 0.05) among latitude zones based on a likelihood ratio test. Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 View Large Table 2. Summary of RMA regression results between leaf N and leaf P concentrations (e.g. log10 leaf N = α log10 leaf P + log10β), the statistics of leaf N and P concentrations, and the N:P mass ratio in terrestrial plants along the latitudinal zones. The exponent of herbs in the tropical zone was excluded in the comparison due to the paucity of samples in our data set. We also did not consider the latitudinal pattern of conifers because most of the coniferous samples in our data set were distributed in the temperate zone. All the plants, including conifers and tropical herbs, were pooled together during the evaluation of the latitudinal pattern for all plants. Mean indicates geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b, c) denote significant difference (p < 0.05) among latitude zones based on a likelihood ratio test. Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 View Large Furthermore, there are remarkable differences in leaf N–P stoichiometry across different ecoregions (six continents) (Table 3). The geometric mean values of leaf N concentrations were 18.48 mg g−1, 20.79 mg g−1, 19.33 mg g−1, 10.01 mg g−1, 10.32 mg g−1 and 18.51 mg g−1 for North America, Europe, Asia, Oceania, Africa and South America, respectively. The respective geometric mean values of leaf P concentrations were 1.46 mg g−1, 1.54 mg g−1, 1.25 mg g−1, 0.46 mg g−1, 0.51 mg g−1 and 0.69 mg g−1, and the geometric mean values of leaf N:P mass ratio were 12.7, 13.5, 15.5 21.5, 20.4 and 26.7, respectively. Table 3. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants across different continents. Note: each continent only contained those countries that had sampling sites in our data set (for details, see text). Mean indicates geometric mean, and n is the number of observations. Each regression relationships were statistically significant with p < 0.05. Different letters (i.e. a, b, c, d, e) denote significant difference (p < 0.05) among continents based on a likelihood ratio test. Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 The countries located in the six continents were as follows: North America: The United States of America, Canada and Mexico; Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; Oceania: Australia and New Zealand; Africa: South Africa, Uganda and Cameroon; and South America: Brazil, Bolivia and Venezuela. View Large Table 3. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants across different continents. Note: each continent only contained those countries that had sampling sites in our data set (for details, see text). Mean indicates geometric mean, and n is the number of observations. Each regression relationships were statistically significant with p < 0.05. Different letters (i.e. a, b, c, d, e) denote significant difference (p < 0.05) among continents based on a likelihood ratio test. Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 The countries located in the six continents were as follows: North America: The United States of America, Canada and Mexico; Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; Oceania: Australia and New Zealand; Africa: South Africa, Uganda and Cameroon; and South America: Brazil, Bolivia and Venezuela. View Large At different study sites, the leaf N–P stoichiometry also showed statistically significant differences. The site-level geometric mean values of leaf N and P concentrations and N:P mass ratios exhibited large variations among 142 sites with records of n ≥ 10, ranging between 4.6 and 30.5 mg g−1 for N, 0.16 and 2.83 mg g−1 for P, and 6.3 and 35.4 for N:P, with respective geometric means of 17.8 mg g−1, 1.1 mg g−1 and 15.8 (Fig. 1). Figure 1. View largeDownload slide Frequency distribution of leaf N and P contents and N:P ratio of samples from the 142 sites with >10 records. (a) Leaf N content; (b) leaf P content; and (c) leaf N:P ratio. Figure 1. View largeDownload slide Frequency distribution of leaf N and P contents and N:P ratio of samples from the 142 sites with >10 records. (a) Leaf N content; (b) leaf P content; and (c) leaf N:P ratio. In short, in this section, we have documented the numerical values of the global leaf N and P stoichiometry by functional group, latitudinal zone, ecoregion and local site, which reveals a large variation in the leaf N and P concentrations and N:P ratios, both biologically and ecologically. Our results supported the general patterns of leaf N and P stoichiometry reported by Reich and Oleksyn [10], and the numerical values could be used as a basis for comparing other individual studies with the global averages documented here. VARIATION IN THE LEAF N–P SCALING EXPONENT Leaf N–P scaling exponents across different scales In this section, we examine whether the leaf N–P scaling exponent varies among plant functional groups, latitudes, ecoregions and local sites. As shown in Fig. 2 and Table 1, the scaling exponent for the pooled data was 0.678 (95% confidence interval (CI) = 0.669–0.688), and thus statistically indistinguishable from two-thirds. However, the numerical values of the scaling exponent differed significantly among the contrasting functional groups (Fig. 2). Compared to woody plants, herbaceous plants showed numerically lower N vs. P scaling exponents (0.659 vs. 0.705). The scaling exponents for coniferous, deciduous broad-leaved and evergreen broad-leaved woody species were 0.610, 0.712 and 0.731, respectively (Fig. 2 and Table 1). Figure 2. View largeDownload slide Relationships between leaf N and leaf P concentrations in terrestrial plants among functional groups. (a) All species pooled together, (b) herbaceous plants, (c) woody plants pooled together and (d) woody plants classified by life forms (EB, evergreen broad-leaved; DB, deciduous broad-leaved; and C, conifers). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentrations, P concentrations and N:P ratio, respectively. Numbers in square brackets denote the lower and upper 95% confidence limits of the scaling exponents. Figure 2. View largeDownload slide Relationships between leaf N and leaf P concentrations in terrestrial plants among functional groups. (a) All species pooled together, (b) herbaceous plants, (c) woody plants pooled together and (d) woody plants classified by life forms (EB, evergreen broad-leaved; DB, deciduous broad-leaved; and C, conifers). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentrations, P concentrations and N:P ratio, respectively. Numbers in square brackets denote the lower and upper 95% confidence limits of the scaling exponents. The leaf N–P scaling exponents also showed significant latitudinal differences. For the pooled data, the scaling exponents decreased from 0.747 in the tropical region to 0.715 in the temperate region, and to 0.603 in the boreal region (Table 2 and Fig. S1a). For each of the functional groups, the latitudinal patterns of the scaling exponents were generally consistent with the pooled data set, i.e. the N vs. P scaling exponents decreased from tropical to temperate, to boreal regions (Table 2 and Fig. S1b-d). For example, the exponent of evergreen broad-leaved woody species decreased from 0.783 in the tropical region to 0.689 in the temperate region, to 0.643 in the boreal region. The numerical values of the deciduous broad-leaved woody species in the tropical (0.704) and temperate (0.766) regions were much higher than those in the boreal region (0.424). Similarly, the scaling exponent of the herbaceous species in the temperate region was statistically significantly higher than that in the boreal region (0.681 vs. 0.609). Further investigation indicated a large difference in the leaf N–P scaling exponents across different ecoregions (six continents). The respective values of the scaling exponents in North America, Europe, Asia, Oceania, Africa and South America were 0.603, 0.672, 0.712, 0.786, 0.835 and 1.071 (Fig. 3 and Table 3). Figure 3. View largeDownload slide Relationship between leaf N and leaf P concentrations in terrestrial plants among six ecoregions (continents). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentration, P concentration and the N:P ratio, respectively. Note: each continent only contains those countries with sampling sites that were documented in our data set. The countries located on the six continents were as follows: (a) North America: the USA, Canada and Mexico; (b) Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; (c) Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; (d) Oceania: Australia and New Zealand; (e) Africa: South Africa, Uganda and Cameroon; and (f) South America: Brazil, Bolivia and Venezuela. Numbers in square brackets indicate the lower and upper 95% confidence limits of the scaling exponents. Figure 3. View largeDownload slide Relationship between leaf N and leaf P concentrations in terrestrial plants among six ecoregions (continents). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentration, P concentration and the N:P ratio, respectively. Note: each continent only contains those countries with sampling sites that were documented in our data set. The countries located on the six continents were as follows: (a) North America: the USA, Canada and Mexico; (b) Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; (c) Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; (d) Oceania: Australia and New Zealand; (e) Africa: South Africa, Uganda and Cameroon; and (f) South America: Brazil, Bolivia and Venezuela. Numbers in square brackets indicate the lower and upper 95% confidence limits of the scaling exponents. The data of species growing in different sites manifested statistically significant differences in the numerical values of the leaf N–P scaling exponents. When the data from all 142 sites were pooled, the overall N vs. P scaling exponent was 0.664, or approximately two-thirds (p > 0.05). However, when scaling analyses were performed for each site, the numerical values of the 142 individual exponents showed a log-normal distribution with a mean value of 0.841 (a geometric mean of 0.804) and ranged from 0.366 to 1.928 (Fig. 4). Similarly, when only sites with larger sample sizes were considered (sites for which sampling records were n ≥ 20, a total of 94 individual sites), statistically significant differences in the scaling exponents were detected, i.e. numerical values ranged between 0.441 and 1.492 with an average of 0.817 (Fig. S2b). Figure 4. View largeDownload slide The leaf N to P scaling at 142 sites with more than 10 records. (a) Relationships between leaf N and P concentrations with overall RMA regression line (red) and separate RMA regression lines for sites (black); (b) Frequency distribution of the scaling exponents from the 142 sites (slopes of the black lines in a). All regression slopes were significantly larger than zero (p < 0.05). Figure 4. View largeDownload slide The leaf N to P scaling at 142 sites with more than 10 records. (a) Relationships between leaf N and P concentrations with overall RMA regression line (red) and separate RMA regression lines for sites (black); (b) Frequency distribution of the scaling exponents from the 142 sites (slopes of the black lines in a). All regression slopes were significantly larger than zero (p < 0.05). In summary, our data reveal large differences in the numerical values of leaf N vs. P scaling exponents that are dependent on species functional groups, ecoregions and sample sites, although the numerical value when all records are pooled is close to two-thirds. Possible mechanisms of the different N–P scaling exponents As shown in the section described above, extensive analysis of a large worldwide data set consisting of 12 055 records shows that the numerical value of the exponent governing the leaf N vs. P scaling relationship is approximately two-thirds when all species are pooled. This is consistent with the finding of Reich et al. [15]. However, more detailed analyses of the same data set show that statistically significant differences exist in the exponents governing the leaf N vs. P scaling relationships depending on the functional group, latitudinal zone, ecoregion or local site conditions being considered. These results show that the overall N vs. P scaling relationship derived from any global data set likely hides variation that is biologically (and thus ecologically) important, particularly with regard to modelling N and P dynamics on the whole ecosystem or at site-specific levels. The data presented here also shed some light as to why variations in N and P scaling relationships exist, although a detailed study of causalities is outside of the scope of this paper. For example, the scaling exponents governing the N vs. P relationships are negatively correlated with leaf P concentrations and positively correlated with N:P ratios, but not significantly related to leaf N concentrations (Fig. 5). We speculate that leaf P concentration plays a pivotal role in ‘shaping’ the numerical values of the N vs. P scaling exponent, i.e. metabolic requirements for P mainly account for scaling differences across different functional groups, latitudinal zones and the six continental ecosystems. This speculation resonates with previous studies reporting that changes in N:P ratios are mainly driven by variations in P concentration [4,5,36]. Figure 5. View largeDownload slide Relationships between the N to P scaling exponents and leaf N, P and N:P mass ratios at different scales. (a) Relationships between the N to P scaling exponents and leaf N content; (b) relationships between the N to P scaling exponents and leaf P content; and (c) relationships between the N to P scaling exponents and N:P mass ratios. The error bars indicate the 95% CI of the scaling exponents. Leaf N and P contents and N:P ratios were derived from the geometric means of the analysed samples during each scaling relationship analysis. Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. The black, red, blue and green dots in the legend, respectively, represent mean values from five functional group levels (cf. Table 1), six continents, eleven latitudinal zones (cf. Table 2) and one global level. Figure 5. View largeDownload slide Relationships between the N to P scaling exponents and leaf N, P and N:P mass ratios at different scales. (a) Relationships between the N to P scaling exponents and leaf N content; (b) relationships between the N to P scaling exponents and leaf P content; and (c) relationships between the N to P scaling exponents and N:P mass ratios. The error bars indicate the 95% CI of the scaling exponents. Leaf N and P contents and N:P ratios were derived from the geometric means of the analysed samples during each scaling relationship analysis. Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. The black, red, blue and green dots in the legend, respectively, represent mean values from five functional group levels (cf. Table 1), six continents, eleven latitudinal zones (cf. Table 2) and one global level. In this regard, the influence of the requirement for P on growth rates might have the most important influence on variations in the scaling exponents across functional groups. For example, compared to woody species, herbaceous species tend to be small-sized and have higher growth rates [37,38]. According to the growth rate hypothesis of Sterner and Elser [1], plants with higher growth rates are predicted to have higher P demands and lower N:P mass ratios [39]. Our data reveal that herbaceous species have higher leaf P concentrations and lower N:P ratios (Table 1) [22], which may account for N vs. P scaling exponents with lower numerical values (see Fig. 5). Similarly, compared to evergreen broad-leaved plants, deciduous broad-leaved plants tend to have low leaf longevity and rapid leaf growth rates [40–42], resulting in high P demands and low N:P ratios. This difference may be responsible for the numerically smaller scaling exponents in deciduous relative to evergreen broad-leaved plants. In contrast, the numerically smaller N–P scaling exponents observed for coniferous species relative to broad-leaved woody species might reflect leaf morphological and anatomical characteristics in addition to P-related growth rate. The specific leaf area of conifers is generally smaller than that of broad-leaved species [43,44], which may result in relatively lower nutrient demands, especially for N (Table 1) [6]. In addition, coniferous species are mostly distributed in cold habitats that may require the storage of lipid P in thick and narrow leaves for cold resistance [45]. Thus, compared to broad-leaved woody species, coniferous species tend to have low N:P ratios and thus numerically lower N vs. P scaling exponents (Table 1). Species composition, P-related growth rates and soil relative nutrient availabilities collectively appear to be the most important factors governing the numerical values of N vs. P scaling exponents across latitudinal zones. Plant functional group traits correlate significantly with the latitudinal pattern of leaf N and P concentrations and N:P ratios [18], and thus likely control leaf N vs. P scaling relationships across latitudinal zones. In our global data set, evergreen broad-leaved trees dominate tropical regions, deciduous broad-leaved trees and herbaceous plants co-dominate temperate regions, while coniferous and herbaceous species are more frequent than other plant species groups in boreal regions (Fig. 6a). In accordance with the aforementioned variations in the N–P scaling exponent across functional groups (Table 1), these distinct species groupings explain some, but clearly not all, of the latitudinal patterns of the N vs. P scaling exponent for each of the specific functional groups (i.e. evergreen broad-leaved woody species, deciduous broad-leaved woody species and herbs). Other factors are likely involved. For example, the length of the growing season, and thus leaf longevity tend to decrease from the tropics to boreal regions, resulting in higher leaf growth rates and higher P demands, particularly in the growing season, with increasing latitude [41,46]. Furthermore, soil P availability relative to N availability tends to increase with increasing latitude or from the humid to arid regions, resulting in decreasing leaf N:P ratios [10,31,32,47]. These trends help to explain why the numerical values of the N vs. P scaling exponent decline from the tropics to higher latitudes for each of the major plant functional groups. Figure 6. View largeDownload slide Changes in species composition along (a) the latitudinal zones and (b) across six continents. The four functional groups (herb, evergreen broad-leaved woody species, deciduous broad-leaved woody species and conifers) occupied most of the sample cases, whereas some species (e.g. fern) or unknown plants were not included in the analysis. Figure 6. View largeDownload slide Changes in species composition along (a) the latitudinal zones and (b) across six continents. The four functional groups (herb, evergreen broad-leaved woody species, deciduous broad-leaved woody species and conifers) occupied most of the sample cases, whereas some species (e.g. fern) or unknown plants were not included in the analysis. Species composition and soil relative nutrient availability are additional factors that are related to variation of the N vs. P scaling exponent across different ecoregions (continents) because they influence the N and P stoichiometry [48]. For example, herbs and conifers are especially well represented in North America and Europe compared to the other four continents, whereas broad-leaved deciduous species dominate the species composition in Asia, and evergreen broad-leaved species dominate Oceania, Africa and South America (Fig. 6b). Considering that N vs. P scaling exponents differ as a function of functional groupings, we suggest that the biases in the species composition observed for the different continents are important contributors to the differences in the numerical values of N vs. P scaling exponents. Furthermore, previous studies have shown that soil relative nutrient availability differs among the six continents. According to world maps of nutrient limitations [32,49] and soil P distribution [50], European and North American sample sites are mainly N-limited, the Asian sample sites are located in both N-limited and P-limited regions, whereas sample sites in South America, Africa and Oceania are mostly located in P-limited regions. Moreover, P limitation tends to increase with elevated N:P ratios, whereas N limitation increases with declining N:P ratios [4,5,51]. This feature, even in isolation, could explain the lower N vs. P scaling exponents observed for sample sites in North America and Europe compared to those in Africa, Oceania and South America. Based on the Global Gridded Soil Phosphorus Distribution Maps [50], we extracted the total P density in the top 50 cm soil and analysed the relationships between the leaf N vs. P scaling exponent and soil total P density at the different scales of globe, latitude range and continent. We found that the scaling exponent tended to decrease with increasing soil P density (Fig. S3, r2= 0.28, p = 0.024). This result further demonstrates the crucial role of soil P in influencing the leaf N vs. P scaling exponent [52]. In addition, plants tend to uptake excess N (i.e. luxury consumption of N) when soil P availability is deficient [7,29,53] and vice versa [54,55]. Excess uptake of elements could mask the stoichiometric requirements to varying degrees and dramatically modify the N–P scaling relationship. Presumably, excess uptake of N for P-limited plants causes a higher scaling exponent of leaf N vs. P, whereas excess uptake of P for N-limited plants results in a lower scaling exponent [29]. Moreover, leaf N concentration could occasionally be negatively correlated with leaf P concentration in the case of luxury N uptake [5]. However, given the pervasiveness of N and P limitation in terrestrial ecosystems [56,57] and the difficulty in detecting the magnitude of excess nutrient uptake, the exact extent of N vs. P scaling flexibility resulting from the excess nutrient uptake at various nutrient-limited sites may be hard to evaluate. Finally, as most of the sample sites in Europe and North America are located at high latitudes and regarded as N-limited, whereas the South American sample sites are mainly located at low latitudes generally limited by P, the variations in the scaling exponent across these continents are coincident with the aforementioned latitudinal patterns. Our results have revealed large variations in leaf N and P stoichiometry and the N vs. P scaling exponent, although the fundamental mechanisms underpinning these patterns remain poorly interpreted. Initially, the three-quarters or two-thirds power law empirically proposed by studies with limited data, or pooled data combining all the functional groups and primarily ignored the phylogeny of plants, lacked convincing theoretical foundation [8,15,17,24]. In actual fact, 13 years ago, Savage et al. [28] recognized the similar variability in the allometric scaling exponent in biology by synthesizing studies of basal metabolic rates. Due to the lack of supporting data regarding soil nutrient availability for each record in our data set, we failed to build strong correlations between leaf N vs. P scaling exponents and soil parameters at site levels. Hence, there is a need for extensive data regarding paired N–P concentration of leaves and soils with supporting biogeographic and physiological information to provide more evidence for the explanation of the variation in the scaling exponent. CONCLUSIONS Our results document detailed information on leaf N and P concentrations and N:P ratios of different functional groups, latitudinal zones and ecoregions. Leaf N and P concentrations of herbaceous plants were significantly higher than those for woody plants, which showed decreasing trends from boreal to tropical regions. Among six continents, Europe had the highest N and P concentrations and Oceania showed the smallest values. In particular, our data indicate that large differences in the numerical values of leaf N vs. P scaling exponents depend on species functional groups, ecoregions and sample sites, although the numerical value for the exponent when all records were pooled was close to two-thirds. Compared to woody species, herbaceous species have lower N vs. P scaling exponents (0.659 vs. 0.705). Among woody species, conifers have the lowest scaling exponent (0.610), whereas deciduous and evergreen broad-leaved woody species have the highest values (0.712 and 0.731, respectively). The numerical values of exponents manifest a significant latitudinal pattern, with decreasing values from tropical to temperate to boreal regions. Notable differences in the scaling exponent also occur across North America, Europe, Asia, Oceania, Africa and South America (i.e. α = 0.603, 0.672, 0.712, 0.786, 0.835 and 1.071, respectively). In addition, the numerical values of the exponent differed as a function of sample site, with a geometric mean value of 0.804, ranging from 0.366 to 1.928. These results show that there is no canonical numerical value for the N vs. P scaling exponent, and that the analysis of pooled data for this scaling relationship may hide biologically and ecologically significant variation. Our findings have important implications for predicting plant growth rate and ultimately vegetation productivity, helping parameterize vegetation climate models [13,17,21], and increasing our understanding of plant adaptation and evolution [22]. Our results also suggest that we need to incorporate specific exponents into scaling leaf N to P in plant growth and ecosystem functioning models according to specific functional groups, biogeographic regions and ecosystem nutrient availabilities. Acknowledgements We thank Chao Li, Wengjing Fang, Weinan Chen, Suhui Ma, Ming Ouyang, Shaopeng Wang and Zhiheng Wang for their helpful suggestions regarding data collection and analysis. We appreciate the researchers who contributed their available data in the global TRY database, listed in the supporting references. FUNDING This work was supported by National Natural Science Foundation of China (31621091 and 31330012), the National Key Research and Development Program of China (2017YFC0503900), and the TRY initiative on plant traits (http://www.try-db.org). The TRY database is hosted at the Max Planck Institute for Biogeochemistry (Jena, Germany) and supported by DIVERSITAS/Future Earth, the German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig and the European Union BACI project (640176). REFERENCES 1. Sterner RW , Elser JJ . Ecological stoichiometry: the biology of elements from molecules to the biosphere . 2002 ; Press Princeton University Press , Princeton . 2. Lambers H , Raven JA , Shaver GR et al. Plant nutrient-acquisition strategies change with soil age . Trends Ecol Evolut 2008 ; 23 : 95 – 103 . https://doi.org/10.1016/j.tree.2007.10.008 Google Scholar Crossref Search ADS 3. Elser JJ , Dobberfuhl DR , MacKay NA et al. Organism Size, Life History, and N:P Stoichiometry . BioScience 1996 ; 46 : 674 – 84 . https://doi.org/10.2307/1312897 Google Scholar Crossref Search ADS 4. Xie L-Q , Xie P , Li S-X et al. The low TN:TP ratio, a cause or a result of Microcystis blooms? Water Res 2003 ; 37 : 2073 – 80 . https://doi.org/10.1016/S0043-1354(02)00532-8 Google Scholar Crossref Search ADS PubMed 5. Güsewell S . N : P ratios in terrestrial plants: variation and functional significance . New Phytol 2004 ; 164 : 243 – 66 . https://doi.org/10.1111/j.1469-8137.2004.01192.x Google Scholar Crossref Search ADS 6. Ghimire B , Riley WJ , Koven CD et al. A global trait-based approach to estimate leaf nitrogen functional allocation from observations . Ecol Appl 2017 ; 27 : 1421 – 34 . https://doi.org/10.1002/eap.1542 Google Scholar Crossref Search ADS PubMed 7. Ågren GI . The C:N:P stoichiometry of autotrophs - theory and observations . Ecol Lett 2004 ; 7 : 185 – 91 . https://doi.org/10.1111/j.1461-0248.2004.00567.x Google Scholar Crossref Search ADS 8. Niklas KJ , Owens T , Reich PB et al. Nitrogen/phosphorus leaf stoichiometry and the scaling of plant growth . Ecol Lett 2005 ; 8 : 636 – 42 . https://doi.org/10.1111/j.1461-0248.2005.00759.x Google Scholar Crossref Search ADS 9. Elser JJ , Fagan WF , Kerkhoff AJ et al. Biological stoichiometry of plant production: metabolism, scaling and ecological response to global change . New Phytol 2010 ; 186 : 593 – 608 . https://doi.org/10.1111/j.1469-8137.2010.03214.x Google Scholar Crossref Search ADS PubMed 10. Reich PB , Oleksyn J . Global patterns of plant leaf N and P in relation to temperature and latitude . Proc Natl Acad Sci USA 2004 ; 101 : 11001 – 6 . https://doi.org/10.1073/pnas.0403588101 Google Scholar Crossref Search ADS PubMed 11. Han W-X , Fang J-Y , Guo D-L et al. Leaf nitrogen and phosphorus stoichiometry across 753 terrestrial plant species in China . New Phytol 2005 ; 168 : 377 – 85 . https://doi.org/10.1111/j.1469-8137.2005.01530.x Google Scholar Crossref Search ADS PubMed 12. Sardans J , Janssens IA , Alonso R et al. Foliar elemental composition of European forest tree species associated with evolutionary traits and present environmental and competitive conditions . Glob Ecol Biogeogr 2015 ; 24 : 240 – 55 . https://doi.org/10.1111/geb.12253 Google Scholar Crossref Search ADS 13. Tang Z-Y , Xu W-T , Zhou G-Y et al. Patterns of plant carbon, nitrogen and phosphorus concentrations in relation to productivity in the terrestrial ecosystems of China . Proc Natl Acad Sci USA 2018 ; 115 : 4033 – 8 . Google Scholar Crossref Search ADS PubMed 14. Wright IJ , Reich PB , Westoby M et al. The worldwide leaf economics spectrum . Nature 2004 ; 428 : 821 – 7 . https://doi.org/10.1038/nature02403 Google Scholar Crossref Search ADS PubMed 15. Reich PB , Oleksyn J , Wright IJ et al. Evidence of a general 2/3-power law of scaling leaf nitrogen to phosphorus among major plant groups and biomes . Proc Biol Sci 2010 ; 277 : 877 – 83 . https://doi.org/10.1098/rspb.2009.1818 Google Scholar Crossref Search ADS PubMed 16. Díaz S , Hodgson JG , Thompson K et al. The plant traits that drive ecosystems: evidence from three continents . J Veg Sci 2004 ; 15 : 295 – 304 . https://doi.org/10.1111/j.1654-1103.2004.tb02266.x Google Scholar Crossref Search ADS 17. Wright IJ , Reich PB , Cornelissen JHC et al. Assessing the generality of global leaf trait relationships . New Phytol 2005 ; 166 : 485 – 96 . https://doi.org/10.1111/j.1469-8137.2005.01349.x Google Scholar Crossref Search ADS PubMed 18. Han W-X , Fang J-Y , Reich PB et al. Biogeography and variability of eleven mineral elements in plant leaves across gradients of climate, soil and plant functional type in China . Ecol Lett 2011 ; 14 : 788 – 96 . https://doi.org/10.1111/j.1461-0248.2011.01641.x Google Scholar Crossref Search ADS PubMed 19. Ågren GI , Weih M . Plant stoichiometry at different scales: element concentration patterns reflect environment more than genotype . New Phytol 2012 ; 194 : 944 – 52 . https://doi.org/10.1111/j.1469-8137.2012.04114.x Google Scholar Crossref Search ADS PubMed 20. Niklas KJ , Cobb ED . N, P, and C stoichiometry of Eranthis hyemalis (Ranunculaceae) and the allometry of plant growth . Am J Bot 2005 ; 92 : 1256 – 63 . https://doi.org/10.3732/ajb.92.8.1256 Google Scholar Crossref Search ADS PubMed 21. Niklas KJ. Plant allometry, leaf nitrogen and phosphorus stoichiometry, and interspecific trends in annual growth rates . Ann Bot 2006 ; 97 : 155 – 63 . https://doi.org/10.1093/aob/mcj021 Google Scholar Crossref Search ADS PubMed 22. Kerkhoff AJ , Fagan FM , Elser JJ et al. Phylogenetic and growth form variation in the scaling of nitrogen and phosphorus in the seed plants . Am Nat 2006 ; 168 : E103 – 22 . https://doi.org/10.1086/507879 Google Scholar Crossref Search ADS PubMed 23. Yan Z-B , Han W-X , Peñuelas J et al. Phosphorus accumulates faster than nitrogen globally in freshwater ecosystems under anthropogenic impacts . Ecol Lett 2016 ; 19 : 1237 – 46 . https://doi.org/10.1111/ele.12658 Google Scholar Crossref Search ADS PubMed 24. Niklas KJ , Cobb ED . Biomass partitioning and leaf N,P-stoichiometry: comparisons between tree and herbaceous current-year shoots . Plant Cell Environ 2006 ; 29 : 2030 – 42 . https://doi.org/10.1111/j.1365-3040.2006.01578.x Google Scholar Crossref Search ADS PubMed 25. Zhao N , Yu G-R , He N-P et al. Invariant allometric scaling of nitrogen and phosphorus in leaves, stems, and fine roots of woody plants along an altitudinal gradient . J Plant Res 2016 ; 129 : 647 – 57 . https://doi.org/10.1007/s10265-016-0805-4 Google Scholar Crossref Search ADS PubMed 26. McGroddy ME , Daufresne T , Hedin LO . Scaling of C:N:P stoichiometry in forests worldwide: implications of terrestrial Redfield-type ratios . Ecology 2004 ; 85 : 2390 – 401 . https://doi.org/10.1890/03-0351 Google Scholar Crossref Search ADS 27. Rowland L , Zaragoza-Castells J , Bloomfield KJ et al. Scaling leaf respiration with nitrogen and phosphorus in tropical forests across two continents . New Phytol 2017 ; 214 : 1064 – 77 . https://doi.org/10.1111/nph.13992 Google Scholar Crossref Search ADS PubMed 28. Savage VM , Gillooly JF , Woodruff WH et al. The predominance of quarter-power scaling in biology . Funct Ecol 2004 ; 18 : 257 – 82 . https://doi.org/10.1111/j.0269-8463.2004.00856.x Google Scholar Crossref Search ADS 29. Ågren GI . Stoichiometry and nutrition of plant growth in natural communities . Annu Rev Ecol Evol Syst 2008 ; 39 : 153 – 70 . https://doi.org/10.1146/annurev.ecolsys.39.110707.173515 Google Scholar Crossref Search ADS 30. Elser JJ , Acharya K , Kyle M et al. Growth rate-stoichiometry couplings in diverse biota . Ecol Lett 2003 ; 6 : 936 – 43 . https://doi.org/10.1046/j.1461-0248.2003.00518.x Google Scholar Crossref Search ADS 31. Vitousek PM , Howarth RW . Nitrogen limitation on land and in the sea: how can it occur? Biogeochemistry 1991 ; 13 : 87 – 115 . https://doi.org/10.1007/BF00002772 Google Scholar Crossref Search ADS 32. Norby RJ , De Kauwe MG , Domingues TF et al. Model-data synthesis for the next generation of forest free-air CO2 enrichment (FACE) experiments . New Phytol 2016 ; 209 : 17 – 28 . https://doi.org/10.1111/nph.13593 Google Scholar Crossref Search ADS PubMed 33. Kattge J , Diaz S , Lavorel S et al. TRY - a global database of plant traits . Global Change Biol 2011 ; 17 : 2905 – 35 . https://doi.org/10.1111/j.1365-2486.2011.02451.x Google Scholar Crossref Search ADS 34. Warton DI , Wright IJ , Falster DS et al. Bivariate line-fitting methods for allometry . Biol Rev 2006 ; 81 : 259 – 91 . https://doi.org/10.1017/S1464793106007007 Google Scholar Crossref Search ADS PubMed 35. R Development Core Team . R: A Language and Environment for Statistical Computing . 2015 ; R Foundation for Statistic Computing , Vienna . 36. Niinemets Ü , Kull K . Leaf structure vs. nutrient relationships vary with soil conditions in temperate shrubs and trees . Acta Oecologica 2003 ; 24 : 209 – 19 . https://doi.org/10.1016/S1146-609X(03)00094-8 Google Scholar Crossref Search ADS 37. Niklas KJ. Plant Allometry: the scaling of form and process . 1994 ; University of Chicago Press , Chicago . 38. Güsewell S , Koerselman W . Variation in nitrogen and phosphorus concentrations of wetland plants . Perspect Plant Ecol Evol Syst 2002 ; 5 : 37 – 61 . https://doi.org/10.1078/1433-8319-0000022 Google Scholar Crossref Search ADS 39. Delgado-Baquerizo M , Reich PB , García-Palacios P et al. Biogeographic bases for a shift in crop C:N:P stoichiometries during domestication . Ecol Lett 2016 ; 19 : 564 – 75 . https://doi.org/10.1111/ele.12593 Google Scholar Crossref Search ADS PubMed 40. Chapin FS III , Shaver GR . Differences in growth and nutrient use among arctic plant growth forms . Funct Ecol 1989 ; 3 : 73 – 80 . https://doi.org/10.2307/2389677 Google Scholar Crossref Search ADS 41. van Ommen Kloeke AEE , Douma JC , Ordonez JC et al. Global quantification of contrasting leaf life span strategies for deciduous and evergreen species in response to environmental conditions . Global Ecol Biogeogr 2012 ; 21 : 224 – 35 . https://doi.org/10.1111/j.1466-8238.2011.00667.x Google Scholar Crossref Search ADS 42. Díaz S , Kattge J , Cornelissen JHC et al. The global spectrum of plant form and function . Nature 2016 ; 529 : 167 – 71 . https://doi.org/10.1038/nature16489 Google Scholar Crossref Search ADS PubMed 43. Ackerly DD , Reich PB . Convergence and correlations among leaf size and function in seed plants: A comparative test using independent contrasts . Am J Bot 1999 ; 86 : 1272 – 81 . https://doi.org/10.2307/2656775 Google Scholar Crossref Search ADS PubMed 44. Reich PB , Ellsworth DS , Walters MB et al. Generality of leaf trait relationships: a test across six biomes . Ecology 1999 ; 80 : 1955 – 69 . https://doi.org/10.1890/0012-9658(1999)080[5b1955:GOLTRA]5d2.0.CO;2 Google Scholar Crossref Search ADS 45. Chapin FS , Shaver GR , Kedrowski RA . Environmental controls over carbon, nitrogen and phosphorus fractions in Eriophorum vaginatum in Alaskan tussock tundra J Ecol 1986 ; 74 : 167 – 95 . https://doi.org/10.2307/2260357 Google Scholar Crossref Search ADS 46. Yan Z-B , Li P , Chen Y-H et al. Nutrient allocation strategies of woody plants: an approach from the scaling of nitrogen and phosphorus between twig stems and leaves . Sci Rep 2016 ; 6 : 20099 . https://doi.org/10.1038/srep20099 Google Scholar Crossref Search ADS PubMed 47. Mariotte P , Canarini A , Dijkstra FA . Stoichiometric N:P flexibility and mycorrhizal symbiosis favour plant resistance against drought . J Ecol 2017 ; 105 : 958 – 67 . https://doi.org/10.1111/1365-2745.12731 Google Scholar Crossref Search ADS 48. Yang X-J , Huang Z-Y , Zhang K-L et al. C:N:P stoichiometry of Artemisia species and close relatives across northern China: unravelling effects of climate, soil and taxonomy . J Ecol 2015 ; 103 : 1020 – 31 . https://doi.org/10.1111/1365-2745.12409 Google Scholar Crossref Search ADS 49. Wang Y-P , Law RM , Pak B . A global model of carbon, nitrogen and phosphorus cycles for the terrestrial biosphere . Biogeosciences 2010 ; 7 : 2261 – 82 . https://doi.org/10.5194/bg-7-2261-2010 Google Scholar Crossref Search ADS 50. Yang X , Post WM , Thornton PE et al. Global Gridded Soil Phosphorus Distribution Maps at 0.5-degree Resolution . 2014 ; ORNL DAAC , Oak Ridge, Tennessee, USA . http://dx.doi.org/10.3334/ORNLDAAC/1223 . 51. Yan Z-B , Tian D , Han W-X et al. An assessment on the uncertainty of the nitrogen to phosphorus ratio as a threshold for nutrient limitation in plants . Ann Bot 2017 ; 120 : 937 – 42 . https://doi.org/10.1093/aob/mcx106 Google Scholar Crossref Search ADS PubMed 52. Zhou X , Sun X , Du B-M et al. Multielement stoichiometry in Quercus variabilis under natural phosphorus variation in subtropical China . Sci Rep 2015 ; 5 : 937 – 42 . https://doi.org/10.1038/srep07839 53. Garten CT . Correlations between concentrations of elements in plants . Nature 1976 ; 261 : 686 – 8 . https://doi.org/10.1038/261686a0 Google Scholar Crossref Search ADS 54. Chapin FS III . The mineral nutrition of wild plants . Annu Rev Ecol Syst 1980 ; 11 : 233 – 60 . https://doi.org/10.1146/annurev.es.11.110180.001313 Google Scholar Crossref Search ADS 55. Sistla SA , Appling AP , Lewandowska AM et al. Stoichiometric flexibility in response to fertilization along gradients of environmental and organismal nutrient richness . Oikos 2015 ; 124 : 949 – 59 . https://doi.org/10.1111/oik.02385 Google Scholar Crossref Search ADS 56. Elser JJ , Brachen MES , Cleland EE et al. Global analysis of nitrogen and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems . Ecol Lett 2007 ; 10 : 1135 – 42 . https://doi.org/10.1111/j.1461-0248.2007.01113.x Google Scholar Crossref Search ADS PubMed 57. Sullivan BW , Alvarez-Clare S , Castle SC et al. Assessing nutrient limitation in complex forested ecosystems: alternatives to large-scale fertilization experiments . Ecology 2014 ; 95 : 668 – 81 . https://doi.org/10.1890/13-0825.1 Google Scholar Crossref Search ADS PubMed © The Author(s) 2017. Published by Oxford University Press on behalf of China Science Publishing & Media Ltd. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png National Science Review Oxford University Press

Loading next page...
 
/lp/ou_press/global-leaf-nitrogen-and-phosphorus-stoichiometry-and-their-scaling-FDX14mIbAB
Publisher
Oxford University Press
Copyright
© The Author(s) 2017. Published by Oxford University Press on behalf of China Science Publishing & Media Ltd.
ISSN
2095-5138
eISSN
2053-714X
D.O.I.
10.1093/nsr/nwx142
Publisher site
See Article on Publisher Site

Abstract

Abstract Leaf nitrogen (N) and phosphorus (P) concentrations constrain photosynthetic and metabolic processes, growth and the productivity of plants. Their stoichiometry and scaling relationships regulate the allocation of N and P from subcellular to organism, and even ecosystem levels, and are crucial to the modelling of plant growth and nutrient cycles in terrestrial ecosystems. Prior work has revealed a general biogeographic pattern of leaf N and P stoichiometric relationships and shown that leaf N scales roughly as two-thirds the power of P. However, determining whether and how leaf N and P stoichiometries, especially their scaling exponents, change with functional groups and environmental conditions requires further verification. In this study, we compiled a global data set and documented the global leaf N and P concentrations and the N:P ratios by functional group, climate zone and continent. The global overall mean leaf N and P concentrations were 18.9 mg g−1 and 1.2 mg g−1, respectively, with significantly higher concentrations in herbaceous than woody plants (21.72 mg g−1 vs. 18.22 mg g−1 for N; and 1.64 mg g−1 vs. 1.10 mg g−1 for P). Both leaf N and P showed higher concentrations at high latitudes than low latitudes. Among six continents, Europe had the highest N and P concentrations (20.79 and 1.54 mg g−1) and Oceania had the smallest values (10.01 and 0.46 mg g−1). These numerical values may be used as a basis for the comparison of other individual studies. Further, we found that the scaling exponent varied significantly across different functional groups, latitudinal zones, ecoregions and sites. The exponents of herbaceous and woody plants were 0.659 and 0.705, respectively, with significant latitudinal patterns decreasing from tropical to temperate to boreal zones. At sites with a sample size ≥10, the values fluctuated from 0.366 to 1.928, with an average of 0.841. Several factors including the intrinsic attributes of different life forms, P-related growth rates and relative nutrient availability of soils likely account for the inconstant exponents of leaf N vs. P scaling relationships. leaf, nitrogen, phosphorus, stoichiometry, scaling exponent, functional group, biogeography INTRODUCTION Plant growth and metabolism depend on ribosome numbers and protein concentrations in living cells [1,2]. Nitrogen (N) and phosphorus (P), especially the N in Rubisco that drives photosynthesis and the P in ribosomal RNA that drives the generation and maintenance of proteins, are essential nutrients [3–6] that are consequently tightly linked and important parameters in stoichiometric growth models [7–9]. Therefore, the leaf N and P stoichiometric patterns including N and P concentrations and N:P ratios are largely explored both at regional and global levels [10–13], which are important bridges linking elemental compositions or allocations with organismal metabolic processes, and even energy flow in the whole ecosystem [9]. In particular, the strong correlation between leaf N and P concentrations can be quantified via a stoichiometric scaling relationship described by a power function as N = βPα, where α and β indicate the slope (i.e. the scaling exponent) and the ‘elevation’ or Y-intercept (i.e. normalization constant), respectively, of the log–log linear leaf N concentration vs. P concentration regression line [14,15]. As is the case for other allometric ‘rules’ in ecology, leaf N vs. P scaling provides a simple but useful empirical model [8,9,15] in which the exponent is considered to be critical for predictions of plant and ecosystem functioning [16]. For example, a previous stoichiometric growth model (the growth rate hypothesis) posited that organisms with higher growth rates required disproportionately higher P than N concentrations, resulting in a scaling exponent of N to P concentrations below unity (i.e. α < 1.0) [1]. The large-scale patterns of the leaf N and P stoichiometry in relation to environmental conditions (especially temperature), biogeochemical gradients, intrinsic genetic factors and species composition have been generalized in previous studies [10–12,17–19]. Moreover, a constant exponent of leaf N vs. P scaling is appealing to, and chased by, ecologists for its simplicity in model operations. For example, empirical studies have reported that N concentrations scale, on average, across species as a three-quarter-power function of the P concentrations [20]. If the invariant exponent is true, the plant growth rate could be predicted by leaf N and P stoichiometry [8,21]. Nevertheless, a core issue remains: whether the N vs. P scaling exponent is invariably ‘constant’, or whether its numerical value depends on species functional groupings or other ecosystem properties. Numerous studies have also reported N vs. P scaling relationships with statistically significant variation in the scaling exponent [11,14,15,22,23]. For example, Niklas and Cobb [24] suggested that the scaling exponents of woody and herbaceous species are, on average two-thirds and three-quarters, respectively. Based on an extensive worldwide collection of leaf N and P concentrations, Wright et al. [14] claimed that the N vs. P scaling exponent was approximately two-thirds. In contrast, using a larger leaf N and P data set consisting of 7445 entries compiled by Reich and Oleksyn [10], Niklas [21] reports that the scaling exponent is 0.73, approaching three-quarters rather than two-thirds. In addition, several other studies have indicated that the exponent was approximately 0.67 [11], 0.72 [22], 0.78 [25] or 1.0 [26]. Based on what is to our knowledge the most comprehensive data to date, with 9300 pairwise observations of leaf N and P concentrations, Reich et al. [15] found similar two-thirds exponents across biomes, taxonomic divisions and angiosperm life forms, and therefore proposed the general two-thirds-power law (N ∝ P2/3) across major plant groups and biomes. If true, a simple stoichiometric scaling relationship exists governing leaf stoichiometry and metabolism, despite differences among specific case studies. However, similar to the inconstant scaling of leaf respiration with N and P [27] and the biological quarter-power scaling [28], any N vs. P scaling relationship derived from a large global data set may hide important biological variations rooted in species-dependent, or region- or site-related, differences. According to some stoichiometric scaling models, plant growth rates are purported to influence the N vs. P scaling relationship [1,21,29]. If true, plants from different functional groups or different geographical locations should have different growth rates due to specific evolutionary traits and environmental conditions, resulting in potential differences in their N vs. P scaling exponents. Additionally, according to some growth rate hypotheses [5,8,30], growth rates are purported to be more closely correlated to P concentrations than N concentrations, such that changes in P (and not N) demand or status are predicted to be a strong driver for variation in the numerical values of N vs. P scaling exponents. To test the generality of the leaf N and P stoichiometry and the constancy of the leaf N vs. P scaling exponent, we firstly explore the global leaf N and P stoichiometry that may be used as a basis for the comparison of other individual studies. We then determine the N vs. P scaling exponent for different functional groups (i.e. woody species and herbs) and spatial scales (i.e. latitudinal zones, ecoregions/continents and local sites), to evaluate whether the numerical value of the governing scaling exponent varies as a function of species composition, spatial distribution, or other variables of interest (e.g. climate conditions, and soil-relative N and P availabilities) as sometimes reported by others [10,18,31,32]. Furthermore, we explore the patterns of scaling exponent numerical variation across different sites to test whether the global N vs. P scaling relationship obscures site-related significant differences. We hypothesize that the scaling exponent at each site should reflect site-specific N vs. P stoichiometric relationships, because plants growing at the same site represent the characteristics shaped by the combination of local climatic conditions, geological processes, soil nutrient availabilities and other environment factors. GLOBAL DATA SET AND STATISTICAL ANALYSES In order to carry out the aforementioned studies, we compiled a large and geographically comprehensive global data set of pairwise leaf N and P concentration distributions, including global, regional and site-level records, for as many of the variables of interest as possible. We adopted only those records reporting paired N and P concentrations of green leaves with detailed location information, and excluded all records without site information or with unpaired N–P records. Using a detailed review of the literature, our own field sampling and the open TRY data set (Table S1; https://www.try-db.org) [33], a total of 12 055 records were acquired spanning 486 sites worldwide, in which 142 sites had more than 10 records for each site and 94 sites had more than 20 records for each site. Duplicated records were removed. The data set included 3441 terrestrial plant species in 1342 genera and 222 families. All of the plant samples in the data set were collected during the growing season. We used the Flora of China (http://frps.eflora.cn/), Useful Tropical Plants (http://tropical.theferns.info/), Australian Native Plants (https://www.anbg.gov.au/index.html) and Wikipedia (https://en.wikipedia.org/wiki) websites to identify plant functional groups and verify taxonomic classifications. We conducted similar statistical analysis to Reich et al. [15]. To perform data analyses, we log10-transformed the N and P concentration data for all plants in the data set, and then used reduced major axis (RMA) regression [34] to determine the N vs. P scaling relationship at four levels (i.e. species functional groupings, latitudinal zones, ecoregions and individual sites). Therefore, we first classified plants into herbaceous and woody species, and further divided woody plants into coniferous, deciduous broad-leaved and evergreen broad-leaved woody species. We then grouped the data from sites into tropical (0–25°), temperate (25–50°) and boreal (>50°) zones, to compare the N vs. P scaling exponents among these zones. We also analysed the latitudinal patterns of the scaling exponent for each of the functional groups, but excluded those of coniferous plants because most of the coniferous plants in our data set were distributed in the temperate zone. Our data analysis was then focused on patterns among ecoregions (continents), including North America (the United States of America, Canada and Mexico), Europe (Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden), Asia (China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia), Oceania (Australia and New Zealand), Africa (South Africa, Uganda and Cameroon) and South America (Brazil, Bolivia and Venezuela). Lastly, we analysed the N vs. P scaling relationships at each site (with n ≥ 10 and n ≥ 20 records) and quantified the numerical variation of the scaling exponent across different sites. We used a likelihood ratio test to evaluate the heterogeneity of RMA regression exponents within the aforementioned levels of analyses [34]. Additionally, we performed general linear regressions to explore the changes in the scaling exponents with the geometric mean N and P concentrations and N:P ratios, using the statistical package R 2.15.2 [35]. GLOBAL LEAF N AND P STOICHIOMETRY ACROSS DIFFERENT SCALES We used the above-described global data set of 12 055 pairwise leaf N and P concentration records to characterize large-scale leaf N and P stoichiometry by functional group, latitudinal zone, ecoregion and site. The geometric mean values of leaf N and P concentrations, and N:P mass ratios, of the pooled data were 18.9 mg g−1 and 1.2 mg g−1, and 15.8, respectively, but these numerical values differed significantly among the contrasting functional groups (Table 1). Compared to woody plants, herbaceous plants showed significantly higher N and P concentrations (21.72 mg N g−1 vs. 18.22 mg N g−1; 1.64 mg P g−1 vs. 1.10 mg P g−1), and lower N:P ratios (13.3 vs. 16.6). The mean leaf N (and P) concentrations of coniferous, deciduous broad-leaved and evergreen broad-leaved woody species were 12.13 (and 0.98), 21.13 (and 1.37), and 15.45 (and 0.79) mg g−1, respectively. The corresponding N:P ratios for the three woody groups were 12.4, 15.4 and 19.5, respectively. Table 1. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants for different functional groups. Note: woody plants were divided into three groups: coniferous gymnosperms and two angiosperm groups, deciduous broad-leaved and evergreen broad-leaved woody. Mean indicates the geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b and c) denote significant difference (p < 0.05) among functional groups based on a likelihood ratio test. Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 View Large Table 1. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants for different functional groups. Note: woody plants were divided into three groups: coniferous gymnosperms and two angiosperm groups, deciduous broad-leaved and evergreen broad-leaved woody. Mean indicates the geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b and c) denote significant difference (p < 0.05) among functional groups based on a likelihood ratio test. Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 Functional group N αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean All 12 055 0.678b (0.669–0.688) 0.33 18.93 1.20 15.8 Functional group  Herb 2776 0.659bc (0.637–0.681) 0.20 21.72 1.64 13.3  Woody 8888 0.705a (0.693–0.717) 0.34 18.22 1.10 16.6   Conifer woody 526 0.610c (0.574–0.648) 0.50 12.13 0.98 12.4   Deciduous broad-leaf 5035 0.712a (0.695–0.730) 0.22 21.13 1.37 15.4   Evergreen broad-leaf 3267 0.731a (0.710–0.753) 0.29 15.45 0.79 19.5 View Large The leaf N and P stoichiometry also changed significantly with latitude. For the pooled data, the leaf N and P concentrations significantly increased from the tropical to boreal regions, but N:P ratios decreased (Table 2). Specifically, the geometric mean value of leaf N concentration was 17.41 mg g−1 in the tropical region, 19.24 mg g−1 in the temperate region and 19.83 mg g−1 in the boreal region. The geometric mean value of leaf P concentration was 0.83 mg g−1 in the tropical region, 1.28 mg g−1 in the temperate region and 1.49 mg g−1 in the boreal region. The geometric mean value of leaf N:P mass ratio was 21.0, 15.1 and 13.2 in the tropical, temperate, and boreal regions, respectively. For each of the functional groups, the latitudinal patterns of N–P stoichiometry were generally consistent with the pooled data set, i.e. leaf N and P concentrations increased, but leaf N:P ratios decreased from tropical, to temperate, to boreal regions (Table 2). Table 2. Summary of RMA regression results between leaf N and leaf P concentrations (e.g. log10 leaf N = α log10 leaf P + log10β), the statistics of leaf N and P concentrations, and the N:P mass ratio in terrestrial plants along the latitudinal zones. The exponent of herbs in the tropical zone was excluded in the comparison due to the paucity of samples in our data set. We also did not consider the latitudinal pattern of conifers because most of the coniferous samples in our data set were distributed in the temperate zone. All the plants, including conifers and tropical herbs, were pooled together during the evaluation of the latitudinal pattern for all plants. Mean indicates geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b, c) denote significant difference (p < 0.05) among latitude zones based on a likelihood ratio test. Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 View Large Table 2. Summary of RMA regression results between leaf N and leaf P concentrations (e.g. log10 leaf N = α log10 leaf P + log10β), the statistics of leaf N and P concentrations, and the N:P mass ratio in terrestrial plants along the latitudinal zones. The exponent of herbs in the tropical zone was excluded in the comparison due to the paucity of samples in our data set. We also did not consider the latitudinal pattern of conifers because most of the coniferous samples in our data set were distributed in the temperate zone. All the plants, including conifers and tropical herbs, were pooled together during the evaluation of the latitudinal pattern for all plants. Mean indicates geometric mean, and n is the number of observations. Each regression relationship was statistically significant with p < 0.05. Different letters (i.e. a, b, c) denote significant difference (p < 0.05) among latitude zones based on a likelihood ratio test. Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 Latitudinal zone n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Latitude zone for all plants  0–25° (tropical) 2278 0.747a (0.721–0.775) 0.22 17.41 0.83 21.0  25–50° (temperate) 8225 0.715b (0.703–0.728) 0.38 19.24 1.28 15.1  >50° (boreal) 1470 0.603c (0.576–0.631) 0.21 19.83 1.49 13.2 Latitude zone for evergreen broad-leaved plants  0–25° (tropical) 1, 679 0.783a (0.750–0.818) 0.17 17.01 0.78 21.9  25–50° (temperate) 1, 350 0.689b (0.663–0.716) 0.48 13.78 0.76 18.2  >50° (boreal) 219 0.643b (0.584–0.707) 0.49 15.23 1.27 12.0 Latitude zone for deciduous broad-leaved plants  0–25° (tropical) 313 0.704a (0.642–0.772) 0.32 19.77 1.10 18.0  25–50° (temperate) 4218 0.766a (0.746–0.787) 0.23 21.06 1.35 15.6  >50° (boreal) 469 0.424b (0.388–0.464) 0.03 22.89 1.81 12.6 Latitude zone for herbaceous plants  25–50° (temperate) 2, 039 0.681a (0.655–0.708) 0.19 22.38 1.74 12.9  >50° (boreal) 673 0.609b (0.570–0.651) 0.21 20.08 1.39 14.3 View Large Furthermore, there are remarkable differences in leaf N–P stoichiometry across different ecoregions (six continents) (Table 3). The geometric mean values of leaf N concentrations were 18.48 mg g−1, 20.79 mg g−1, 19.33 mg g−1, 10.01 mg g−1, 10.32 mg g−1 and 18.51 mg g−1 for North America, Europe, Asia, Oceania, Africa and South America, respectively. The respective geometric mean values of leaf P concentrations were 1.46 mg g−1, 1.54 mg g−1, 1.25 mg g−1, 0.46 mg g−1, 0.51 mg g−1 and 0.69 mg g−1, and the geometric mean values of leaf N:P mass ratio were 12.7, 13.5, 15.5 21.5, 20.4 and 26.7, respectively. Table 3. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants across different continents. Note: each continent only contained those countries that had sampling sites in our data set (for details, see text). Mean indicates geometric mean, and n is the number of observations. Each regression relationships were statistically significant with p < 0.05. Different letters (i.e. a, b, c, d, e) denote significant difference (p < 0.05) among continents based on a likelihood ratio test. Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 The countries located in the six continents were as follows: North America: The United States of America, Canada and Mexico; Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; Oceania: Australia and New Zealand; Africa: South Africa, Uganda and Cameroon; and South America: Brazil, Bolivia and Venezuela. View Large Table 3. Summary of RMA regression results between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β, the statistics of leaf N and P concentrations, and the N:P ratios in terrestrial plants across different continents. Note: each continent only contained those countries that had sampling sites in our data set (for details, see text). Mean indicates geometric mean, and n is the number of observations. Each regression relationships were statistically significant with p < 0.05. Different letters (i.e. a, b, c, d, e) denote significant difference (p < 0.05) among continents based on a likelihood ratio test. Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 Region n αRMA (95% CI) r2 N mean (mg g−1) P mean (mg g−1) N:P mean Global 12 055 0.678d (0.669–0.688) 0.33 18.94 1.20 15.8 North America 706 0.603e (0.563–0.646) 0.13 18.48 1.46 12.7 Europe 1852 0.672d (0.646–0.699) 0.26 20.79 1.54 13.5 Asia 7951 0.712c (0.699–0.726) 0.30 19.33 1.25 15.5 Oceania 380 0.786b (0.741–0.834) 0.66 10.01 0.46 21.5 Africa 100 0.835b (0.750–0.930) 0.71 10.32 0.51 20.4 South America 942 1.071a (1.011–1.134) 0.19 18.51 0.69 26.7 The countries located in the six continents were as follows: North America: The United States of America, Canada and Mexico; Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; Oceania: Australia and New Zealand; Africa: South Africa, Uganda and Cameroon; and South America: Brazil, Bolivia and Venezuela. View Large At different study sites, the leaf N–P stoichiometry also showed statistically significant differences. The site-level geometric mean values of leaf N and P concentrations and N:P mass ratios exhibited large variations among 142 sites with records of n ≥ 10, ranging between 4.6 and 30.5 mg g−1 for N, 0.16 and 2.83 mg g−1 for P, and 6.3 and 35.4 for N:P, with respective geometric means of 17.8 mg g−1, 1.1 mg g−1 and 15.8 (Fig. 1). Figure 1. View largeDownload slide Frequency distribution of leaf N and P contents and N:P ratio of samples from the 142 sites with >10 records. (a) Leaf N content; (b) leaf P content; and (c) leaf N:P ratio. Figure 1. View largeDownload slide Frequency distribution of leaf N and P contents and N:P ratio of samples from the 142 sites with >10 records. (a) Leaf N content; (b) leaf P content; and (c) leaf N:P ratio. In short, in this section, we have documented the numerical values of the global leaf N and P stoichiometry by functional group, latitudinal zone, ecoregion and local site, which reveals a large variation in the leaf N and P concentrations and N:P ratios, both biologically and ecologically. Our results supported the general patterns of leaf N and P stoichiometry reported by Reich and Oleksyn [10], and the numerical values could be used as a basis for comparing other individual studies with the global averages documented here. VARIATION IN THE LEAF N–P SCALING EXPONENT Leaf N–P scaling exponents across different scales In this section, we examine whether the leaf N–P scaling exponent varies among plant functional groups, latitudes, ecoregions and local sites. As shown in Fig. 2 and Table 1, the scaling exponent for the pooled data was 0.678 (95% confidence interval (CI) = 0.669–0.688), and thus statistically indistinguishable from two-thirds. However, the numerical values of the scaling exponent differed significantly among the contrasting functional groups (Fig. 2). Compared to woody plants, herbaceous plants showed numerically lower N vs. P scaling exponents (0.659 vs. 0.705). The scaling exponents for coniferous, deciduous broad-leaved and evergreen broad-leaved woody species were 0.610, 0.712 and 0.731, respectively (Fig. 2 and Table 1). Figure 2. View largeDownload slide Relationships between leaf N and leaf P concentrations in terrestrial plants among functional groups. (a) All species pooled together, (b) herbaceous plants, (c) woody plants pooled together and (d) woody plants classified by life forms (EB, evergreen broad-leaved; DB, deciduous broad-leaved; and C, conifers). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentrations, P concentrations and N:P ratio, respectively. Numbers in square brackets denote the lower and upper 95% confidence limits of the scaling exponents. Figure 2. View largeDownload slide Relationships between leaf N and leaf P concentrations in terrestrial plants among functional groups. (a) All species pooled together, (b) herbaceous plants, (c) woody plants pooled together and (d) woody plants classified by life forms (EB, evergreen broad-leaved; DB, deciduous broad-leaved; and C, conifers). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentrations, P concentrations and N:P ratio, respectively. Numbers in square brackets denote the lower and upper 95% confidence limits of the scaling exponents. The leaf N–P scaling exponents also showed significant latitudinal differences. For the pooled data, the scaling exponents decreased from 0.747 in the tropical region to 0.715 in the temperate region, and to 0.603 in the boreal region (Table 2 and Fig. S1a). For each of the functional groups, the latitudinal patterns of the scaling exponents were generally consistent with the pooled data set, i.e. the N vs. P scaling exponents decreased from tropical to temperate, to boreal regions (Table 2 and Fig. S1b-d). For example, the exponent of evergreen broad-leaved woody species decreased from 0.783 in the tropical region to 0.689 in the temperate region, to 0.643 in the boreal region. The numerical values of the deciduous broad-leaved woody species in the tropical (0.704) and temperate (0.766) regions were much higher than those in the boreal region (0.424). Similarly, the scaling exponent of the herbaceous species in the temperate region was statistically significantly higher than that in the boreal region (0.681 vs. 0.609). Further investigation indicated a large difference in the leaf N–P scaling exponents across different ecoregions (six continents). The respective values of the scaling exponents in North America, Europe, Asia, Oceania, Africa and South America were 0.603, 0.672, 0.712, 0.786, 0.835 and 1.071 (Fig. 3 and Table 3). Figure 3. View largeDownload slide Relationship between leaf N and leaf P concentrations in terrestrial plants among six ecoregions (continents). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentration, P concentration and the N:P ratio, respectively. Note: each continent only contains those countries with sampling sites that were documented in our data set. The countries located on the six continents were as follows: (a) North America: the USA, Canada and Mexico; (b) Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; (c) Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; (d) Oceania: Australia and New Zealand; (e) Africa: South Africa, Uganda and Cameroon; and (f) South America: Brazil, Bolivia and Venezuela. Numbers in square brackets indicate the lower and upper 95% confidence limits of the scaling exponents. Figure 3. View largeDownload slide Relationship between leaf N and leaf P concentrations in terrestrial plants among six ecoregions (continents). Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. N, P and N:P indicate the geometric mean values of leaf N concentration, P concentration and the N:P ratio, respectively. Note: each continent only contains those countries with sampling sites that were documented in our data set. The countries located on the six continents were as follows: (a) North America: the USA, Canada and Mexico; (b) Europe: Spain, Greece, France, Ukraine, Czech, Germany, Poland, Belgium, the Netherlands, United Kingdom, Norway and Sweden; (c) Asia: China, Malaysia, India, Sri Lanka, Kazakhstan and Indonesia; (d) Oceania: Australia and New Zealand; (e) Africa: South Africa, Uganda and Cameroon; and (f) South America: Brazil, Bolivia and Venezuela. Numbers in square brackets indicate the lower and upper 95% confidence limits of the scaling exponents. The data of species growing in different sites manifested statistically significant differences in the numerical values of the leaf N–P scaling exponents. When the data from all 142 sites were pooled, the overall N vs. P scaling exponent was 0.664, or approximately two-thirds (p > 0.05). However, when scaling analyses were performed for each site, the numerical values of the 142 individual exponents showed a log-normal distribution with a mean value of 0.841 (a geometric mean of 0.804) and ranged from 0.366 to 1.928 (Fig. 4). Similarly, when only sites with larger sample sizes were considered (sites for which sampling records were n ≥ 20, a total of 94 individual sites), statistically significant differences in the scaling exponents were detected, i.e. numerical values ranged between 0.441 and 1.492 with an average of 0.817 (Fig. S2b). Figure 4. View largeDownload slide The leaf N to P scaling at 142 sites with more than 10 records. (a) Relationships between leaf N and P concentrations with overall RMA regression line (red) and separate RMA regression lines for sites (black); (b) Frequency distribution of the scaling exponents from the 142 sites (slopes of the black lines in a). All regression slopes were significantly larger than zero (p < 0.05). Figure 4. View largeDownload slide The leaf N to P scaling at 142 sites with more than 10 records. (a) Relationships between leaf N and P concentrations with overall RMA regression line (red) and separate RMA regression lines for sites (black); (b) Frequency distribution of the scaling exponents from the 142 sites (slopes of the black lines in a). All regression slopes were significantly larger than zero (p < 0.05). In summary, our data reveal large differences in the numerical values of leaf N vs. P scaling exponents that are dependent on species functional groups, ecoregions and sample sites, although the numerical value when all records are pooled is close to two-thirds. Possible mechanisms of the different N–P scaling exponents As shown in the section described above, extensive analysis of a large worldwide data set consisting of 12 055 records shows that the numerical value of the exponent governing the leaf N vs. P scaling relationship is approximately two-thirds when all species are pooled. This is consistent with the finding of Reich et al. [15]. However, more detailed analyses of the same data set show that statistically significant differences exist in the exponents governing the leaf N vs. P scaling relationships depending on the functional group, latitudinal zone, ecoregion or local site conditions being considered. These results show that the overall N vs. P scaling relationship derived from any global data set likely hides variation that is biologically (and thus ecologically) important, particularly with regard to modelling N and P dynamics on the whole ecosystem or at site-specific levels. The data presented here also shed some light as to why variations in N and P scaling relationships exist, although a detailed study of causalities is outside of the scope of this paper. For example, the scaling exponents governing the N vs. P relationships are negatively correlated with leaf P concentrations and positively correlated with N:P ratios, but not significantly related to leaf N concentrations (Fig. 5). We speculate that leaf P concentration plays a pivotal role in ‘shaping’ the numerical values of the N vs. P scaling exponent, i.e. metabolic requirements for P mainly account for scaling differences across different functional groups, latitudinal zones and the six continental ecosystems. This speculation resonates with previous studies reporting that changes in N:P ratios are mainly driven by variations in P concentration [4,5,36]. Figure 5. View largeDownload slide Relationships between the N to P scaling exponents and leaf N, P and N:P mass ratios at different scales. (a) Relationships between the N to P scaling exponents and leaf N content; (b) relationships between the N to P scaling exponents and leaf P content; and (c) relationships between the N to P scaling exponents and N:P mass ratios. The error bars indicate the 95% CI of the scaling exponents. Leaf N and P contents and N:P ratios were derived from the geometric means of the analysed samples during each scaling relationship analysis. Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. The black, red, blue and green dots in the legend, respectively, represent mean values from five functional group levels (cf. Table 1), six continents, eleven latitudinal zones (cf. Table 2) and one global level. Figure 5. View largeDownload slide Relationships between the N to P scaling exponents and leaf N, P and N:P mass ratios at different scales. (a) Relationships between the N to P scaling exponents and leaf N content; (b) relationships between the N to P scaling exponents and leaf P content; and (c) relationships between the N to P scaling exponents and N:P mass ratios. The error bars indicate the 95% CI of the scaling exponents. Leaf N and P contents and N:P ratios were derived from the geometric means of the analysed samples during each scaling relationship analysis. Scaling exponents (α) were calculated from the RMA regression between leaf N and leaf P concentrations, e.g. log10 leaf N = α log10 leaf P + log10β. The black, red, blue and green dots in the legend, respectively, represent mean values from five functional group levels (cf. Table 1), six continents, eleven latitudinal zones (cf. Table 2) and one global level. In this regard, the influence of the requirement for P on growth rates might have the most important influence on variations in the scaling exponents across functional groups. For example, compared to woody species, herbaceous species tend to be small-sized and have higher growth rates [37,38]. According to the growth rate hypothesis of Sterner and Elser [1], plants with higher growth rates are predicted to have higher P demands and lower N:P mass ratios [39]. Our data reveal that herbaceous species have higher leaf P concentrations and lower N:P ratios (Table 1) [22], which may account for N vs. P scaling exponents with lower numerical values (see Fig. 5). Similarly, compared to evergreen broad-leaved plants, deciduous broad-leaved plants tend to have low leaf longevity and rapid leaf growth rates [40–42], resulting in high P demands and low N:P ratios. This difference may be responsible for the numerically smaller scaling exponents in deciduous relative to evergreen broad-leaved plants. In contrast, the numerically smaller N–P scaling exponents observed for coniferous species relative to broad-leaved woody species might reflect leaf morphological and anatomical characteristics in addition to P-related growth rate. The specific leaf area of conifers is generally smaller than that of broad-leaved species [43,44], which may result in relatively lower nutrient demands, especially for N (Table 1) [6]. In addition, coniferous species are mostly distributed in cold habitats that may require the storage of lipid P in thick and narrow leaves for cold resistance [45]. Thus, compared to broad-leaved woody species, coniferous species tend to have low N:P ratios and thus numerically lower N vs. P scaling exponents (Table 1). Species composition, P-related growth rates and soil relative nutrient availabilities collectively appear to be the most important factors governing the numerical values of N vs. P scaling exponents across latitudinal zones. Plant functional group traits correlate significantly with the latitudinal pattern of leaf N and P concentrations and N:P ratios [18], and thus likely control leaf N vs. P scaling relationships across latitudinal zones. In our global data set, evergreen broad-leaved trees dominate tropical regions, deciduous broad-leaved trees and herbaceous plants co-dominate temperate regions, while coniferous and herbaceous species are more frequent than other plant species groups in boreal regions (Fig. 6a). In accordance with the aforementioned variations in the N–P scaling exponent across functional groups (Table 1), these distinct species groupings explain some, but clearly not all, of the latitudinal patterns of the N vs. P scaling exponent for each of the specific functional groups (i.e. evergreen broad-leaved woody species, deciduous broad-leaved woody species and herbs). Other factors are likely involved. For example, the length of the growing season, and thus leaf longevity tend to decrease from the tropics to boreal regions, resulting in higher leaf growth rates and higher P demands, particularly in the growing season, with increasing latitude [41,46]. Furthermore, soil P availability relative to N availability tends to increase with increasing latitude or from the humid to arid regions, resulting in decreasing leaf N:P ratios [10,31,32,47]. These trends help to explain why the numerical values of the N vs. P scaling exponent decline from the tropics to higher latitudes for each of the major plant functional groups. Figure 6. View largeDownload slide Changes in species composition along (a) the latitudinal zones and (b) across six continents. The four functional groups (herb, evergreen broad-leaved woody species, deciduous broad-leaved woody species and conifers) occupied most of the sample cases, whereas some species (e.g. fern) or unknown plants were not included in the analysis. Figure 6. View largeDownload slide Changes in species composition along (a) the latitudinal zones and (b) across six continents. The four functional groups (herb, evergreen broad-leaved woody species, deciduous broad-leaved woody species and conifers) occupied most of the sample cases, whereas some species (e.g. fern) or unknown plants were not included in the analysis. Species composition and soil relative nutrient availability are additional factors that are related to variation of the N vs. P scaling exponent across different ecoregions (continents) because they influence the N and P stoichiometry [48]. For example, herbs and conifers are especially well represented in North America and Europe compared to the other four continents, whereas broad-leaved deciduous species dominate the species composition in Asia, and evergreen broad-leaved species dominate Oceania, Africa and South America (Fig. 6b). Considering that N vs. P scaling exponents differ as a function of functional groupings, we suggest that the biases in the species composition observed for the different continents are important contributors to the differences in the numerical values of N vs. P scaling exponents. Furthermore, previous studies have shown that soil relative nutrient availability differs among the six continents. According to world maps of nutrient limitations [32,49] and soil P distribution [50], European and North American sample sites are mainly N-limited, the Asian sample sites are located in both N-limited and P-limited regions, whereas sample sites in South America, Africa and Oceania are mostly located in P-limited regions. Moreover, P limitation tends to increase with elevated N:P ratios, whereas N limitation increases with declining N:P ratios [4,5,51]. This feature, even in isolation, could explain the lower N vs. P scaling exponents observed for sample sites in North America and Europe compared to those in Africa, Oceania and South America. Based on the Global Gridded Soil Phosphorus Distribution Maps [50], we extracted the total P density in the top 50 cm soil and analysed the relationships between the leaf N vs. P scaling exponent and soil total P density at the different scales of globe, latitude range and continent. We found that the scaling exponent tended to decrease with increasing soil P density (Fig. S3, r2= 0.28, p = 0.024). This result further demonstrates the crucial role of soil P in influencing the leaf N vs. P scaling exponent [52]. In addition, plants tend to uptake excess N (i.e. luxury consumption of N) when soil P availability is deficient [7,29,53] and vice versa [54,55]. Excess uptake of elements could mask the stoichiometric requirements to varying degrees and dramatically modify the N–P scaling relationship. Presumably, excess uptake of N for P-limited plants causes a higher scaling exponent of leaf N vs. P, whereas excess uptake of P for N-limited plants results in a lower scaling exponent [29]. Moreover, leaf N concentration could occasionally be negatively correlated with leaf P concentration in the case of luxury N uptake [5]. However, given the pervasiveness of N and P limitation in terrestrial ecosystems [56,57] and the difficulty in detecting the magnitude of excess nutrient uptake, the exact extent of N vs. P scaling flexibility resulting from the excess nutrient uptake at various nutrient-limited sites may be hard to evaluate. Finally, as most of the sample sites in Europe and North America are located at high latitudes and regarded as N-limited, whereas the South American sample sites are mainly located at low latitudes generally limited by P, the variations in the scaling exponent across these continents are coincident with the aforementioned latitudinal patterns. Our results have revealed large variations in leaf N and P stoichiometry and the N vs. P scaling exponent, although the fundamental mechanisms underpinning these patterns remain poorly interpreted. Initially, the three-quarters or two-thirds power law empirically proposed by studies with limited data, or pooled data combining all the functional groups and primarily ignored the phylogeny of plants, lacked convincing theoretical foundation [8,15,17,24]. In actual fact, 13 years ago, Savage et al. [28] recognized the similar variability in the allometric scaling exponent in biology by synthesizing studies of basal metabolic rates. Due to the lack of supporting data regarding soil nutrient availability for each record in our data set, we failed to build strong correlations between leaf N vs. P scaling exponents and soil parameters at site levels. Hence, there is a need for extensive data regarding paired N–P concentration of leaves and soils with supporting biogeographic and physiological information to provide more evidence for the explanation of the variation in the scaling exponent. CONCLUSIONS Our results document detailed information on leaf N and P concentrations and N:P ratios of different functional groups, latitudinal zones and ecoregions. Leaf N and P concentrations of herbaceous plants were significantly higher than those for woody plants, which showed decreasing trends from boreal to tropical regions. Among six continents, Europe had the highest N and P concentrations and Oceania showed the smallest values. In particular, our data indicate that large differences in the numerical values of leaf N vs. P scaling exponents depend on species functional groups, ecoregions and sample sites, although the numerical value for the exponent when all records were pooled was close to two-thirds. Compared to woody species, herbaceous species have lower N vs. P scaling exponents (0.659 vs. 0.705). Among woody species, conifers have the lowest scaling exponent (0.610), whereas deciduous and evergreen broad-leaved woody species have the highest values (0.712 and 0.731, respectively). The numerical values of exponents manifest a significant latitudinal pattern, with decreasing values from tropical to temperate to boreal regions. Notable differences in the scaling exponent also occur across North America, Europe, Asia, Oceania, Africa and South America (i.e. α = 0.603, 0.672, 0.712, 0.786, 0.835 and 1.071, respectively). In addition, the numerical values of the exponent differed as a function of sample site, with a geometric mean value of 0.804, ranging from 0.366 to 1.928. These results show that there is no canonical numerical value for the N vs. P scaling exponent, and that the analysis of pooled data for this scaling relationship may hide biologically and ecologically significant variation. Our findings have important implications for predicting plant growth rate and ultimately vegetation productivity, helping parameterize vegetation climate models [13,17,21], and increasing our understanding of plant adaptation and evolution [22]. Our results also suggest that we need to incorporate specific exponents into scaling leaf N to P in plant growth and ecosystem functioning models according to specific functional groups, biogeographic regions and ecosystem nutrient availabilities. Acknowledgements We thank Chao Li, Wengjing Fang, Weinan Chen, Suhui Ma, Ming Ouyang, Shaopeng Wang and Zhiheng Wang for their helpful suggestions regarding data collection and analysis. We appreciate the researchers who contributed their available data in the global TRY database, listed in the supporting references. FUNDING This work was supported by National Natural Science Foundation of China (31621091 and 31330012), the National Key Research and Development Program of China (2017YFC0503900), and the TRY initiative on plant traits (http://www.try-db.org). The TRY database is hosted at the Max Planck Institute for Biogeochemistry (Jena, Germany) and supported by DIVERSITAS/Future Earth, the German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig and the European Union BACI project (640176). REFERENCES 1. Sterner RW , Elser JJ . Ecological stoichiometry: the biology of elements from molecules to the biosphere . 2002 ; Press Princeton University Press , Princeton . 2. Lambers H , Raven JA , Shaver GR et al. Plant nutrient-acquisition strategies change with soil age . Trends Ecol Evolut 2008 ; 23 : 95 – 103 . https://doi.org/10.1016/j.tree.2007.10.008 Google Scholar Crossref Search ADS 3. Elser JJ , Dobberfuhl DR , MacKay NA et al. Organism Size, Life History, and N:P Stoichiometry . BioScience 1996 ; 46 : 674 – 84 . https://doi.org/10.2307/1312897 Google Scholar Crossref Search ADS 4. Xie L-Q , Xie P , Li S-X et al. The low TN:TP ratio, a cause or a result of Microcystis blooms? Water Res 2003 ; 37 : 2073 – 80 . https://doi.org/10.1016/S0043-1354(02)00532-8 Google Scholar Crossref Search ADS PubMed 5. Güsewell S . N : P ratios in terrestrial plants: variation and functional significance . New Phytol 2004 ; 164 : 243 – 66 . https://doi.org/10.1111/j.1469-8137.2004.01192.x Google Scholar Crossref Search ADS 6. Ghimire B , Riley WJ , Koven CD et al. A global trait-based approach to estimate leaf nitrogen functional allocation from observations . Ecol Appl 2017 ; 27 : 1421 – 34 . https://doi.org/10.1002/eap.1542 Google Scholar Crossref Search ADS PubMed 7. Ågren GI . The C:N:P stoichiometry of autotrophs - theory and observations . Ecol Lett 2004 ; 7 : 185 – 91 . https://doi.org/10.1111/j.1461-0248.2004.00567.x Google Scholar Crossref Search ADS 8. Niklas KJ , Owens T , Reich PB et al. Nitrogen/phosphorus leaf stoichiometry and the scaling of plant growth . Ecol Lett 2005 ; 8 : 636 – 42 . https://doi.org/10.1111/j.1461-0248.2005.00759.x Google Scholar Crossref Search ADS 9. Elser JJ , Fagan WF , Kerkhoff AJ et al. Biological stoichiometry of plant production: metabolism, scaling and ecological response to global change . New Phytol 2010 ; 186 : 593 – 608 . https://doi.org/10.1111/j.1469-8137.2010.03214.x Google Scholar Crossref Search ADS PubMed 10. Reich PB , Oleksyn J . Global patterns of plant leaf N and P in relation to temperature and latitude . Proc Natl Acad Sci USA 2004 ; 101 : 11001 – 6 . https://doi.org/10.1073/pnas.0403588101 Google Scholar Crossref Search ADS PubMed 11. Han W-X , Fang J-Y , Guo D-L et al. Leaf nitrogen and phosphorus stoichiometry across 753 terrestrial plant species in China . New Phytol 2005 ; 168 : 377 – 85 . https://doi.org/10.1111/j.1469-8137.2005.01530.x Google Scholar Crossref Search ADS PubMed 12. Sardans J , Janssens IA , Alonso R et al. Foliar elemental composition of European forest tree species associated with evolutionary traits and present environmental and competitive conditions . Glob Ecol Biogeogr 2015 ; 24 : 240 – 55 . https://doi.org/10.1111/geb.12253 Google Scholar Crossref Search ADS 13. Tang Z-Y , Xu W-T , Zhou G-Y et al. Patterns of plant carbon, nitrogen and phosphorus concentrations in relation to productivity in the terrestrial ecosystems of China . Proc Natl Acad Sci USA 2018 ; 115 : 4033 – 8 . Google Scholar Crossref Search ADS PubMed 14. Wright IJ , Reich PB , Westoby M et al. The worldwide leaf economics spectrum . Nature 2004 ; 428 : 821 – 7 . https://doi.org/10.1038/nature02403 Google Scholar Crossref Search ADS PubMed 15. Reich PB , Oleksyn J , Wright IJ et al. Evidence of a general 2/3-power law of scaling leaf nitrogen to phosphorus among major plant groups and biomes . Proc Biol Sci 2010 ; 277 : 877 – 83 . https://doi.org/10.1098/rspb.2009.1818 Google Scholar Crossref Search ADS PubMed 16. Díaz S , Hodgson JG , Thompson K et al. The plant traits that drive ecosystems: evidence from three continents . J Veg Sci 2004 ; 15 : 295 – 304 . https://doi.org/10.1111/j.1654-1103.2004.tb02266.x Google Scholar Crossref Search ADS 17. Wright IJ , Reich PB , Cornelissen JHC et al. Assessing the generality of global leaf trait relationships . New Phytol 2005 ; 166 : 485 – 96 . https://doi.org/10.1111/j.1469-8137.2005.01349.x Google Scholar Crossref Search ADS PubMed 18. Han W-X , Fang J-Y , Reich PB et al. Biogeography and variability of eleven mineral elements in plant leaves across gradients of climate, soil and plant functional type in China . Ecol Lett 2011 ; 14 : 788 – 96 . https://doi.org/10.1111/j.1461-0248.2011.01641.x Google Scholar Crossref Search ADS PubMed 19. Ågren GI , Weih M . Plant stoichiometry at different scales: element concentration patterns reflect environment more than genotype . New Phytol 2012 ; 194 : 944 – 52 . https://doi.org/10.1111/j.1469-8137.2012.04114.x Google Scholar Crossref Search ADS PubMed 20. Niklas KJ , Cobb ED . N, P, and C stoichiometry of Eranthis hyemalis (Ranunculaceae) and the allometry of plant growth . Am J Bot 2005 ; 92 : 1256 – 63 . https://doi.org/10.3732/ajb.92.8.1256 Google Scholar Crossref Search ADS PubMed 21. Niklas KJ. Plant allometry, leaf nitrogen and phosphorus stoichiometry, and interspecific trends in annual growth rates . Ann Bot 2006 ; 97 : 155 – 63 . https://doi.org/10.1093/aob/mcj021 Google Scholar Crossref Search ADS PubMed 22. Kerkhoff AJ , Fagan FM , Elser JJ et al. Phylogenetic and growth form variation in the scaling of nitrogen and phosphorus in the seed plants . Am Nat 2006 ; 168 : E103 – 22 . https://doi.org/10.1086/507879 Google Scholar Crossref Search ADS PubMed 23. Yan Z-B , Han W-X , Peñuelas J et al. Phosphorus accumulates faster than nitrogen globally in freshwater ecosystems under anthropogenic impacts . Ecol Lett 2016 ; 19 : 1237 – 46 . https://doi.org/10.1111/ele.12658 Google Scholar Crossref Search ADS PubMed 24. Niklas KJ , Cobb ED . Biomass partitioning and leaf N,P-stoichiometry: comparisons between tree and herbaceous current-year shoots . Plant Cell Environ 2006 ; 29 : 2030 – 42 . https://doi.org/10.1111/j.1365-3040.2006.01578.x Google Scholar Crossref Search ADS PubMed 25. Zhao N , Yu G-R , He N-P et al. Invariant allometric scaling of nitrogen and phosphorus in leaves, stems, and fine roots of woody plants along an altitudinal gradient . J Plant Res 2016 ; 129 : 647 – 57 . https://doi.org/10.1007/s10265-016-0805-4 Google Scholar Crossref Search ADS PubMed 26. McGroddy ME , Daufresne T , Hedin LO . Scaling of C:N:P stoichiometry in forests worldwide: implications of terrestrial Redfield-type ratios . Ecology 2004 ; 85 : 2390 – 401 . https://doi.org/10.1890/03-0351 Google Scholar Crossref Search ADS 27. Rowland L , Zaragoza-Castells J , Bloomfield KJ et al. Scaling leaf respiration with nitrogen and phosphorus in tropical forests across two continents . New Phytol 2017 ; 214 : 1064 – 77 . https://doi.org/10.1111/nph.13992 Google Scholar Crossref Search ADS PubMed 28. Savage VM , Gillooly JF , Woodruff WH et al. The predominance of quarter-power scaling in biology . Funct Ecol 2004 ; 18 : 257 – 82 . https://doi.org/10.1111/j.0269-8463.2004.00856.x Google Scholar Crossref Search ADS 29. Ågren GI . Stoichiometry and nutrition of plant growth in natural communities . Annu Rev Ecol Evol Syst 2008 ; 39 : 153 – 70 . https://doi.org/10.1146/annurev.ecolsys.39.110707.173515 Google Scholar Crossref Search ADS 30. Elser JJ , Acharya K , Kyle M et al. Growth rate-stoichiometry couplings in diverse biota . Ecol Lett 2003 ; 6 : 936 – 43 . https://doi.org/10.1046/j.1461-0248.2003.00518.x Google Scholar Crossref Search ADS 31. Vitousek PM , Howarth RW . Nitrogen limitation on land and in the sea: how can it occur? Biogeochemistry 1991 ; 13 : 87 – 115 . https://doi.org/10.1007/BF00002772 Google Scholar Crossref Search ADS 32. Norby RJ , De Kauwe MG , Domingues TF et al. Model-data synthesis for the next generation of forest free-air CO2 enrichment (FACE) experiments . New Phytol 2016 ; 209 : 17 – 28 . https://doi.org/10.1111/nph.13593 Google Scholar Crossref Search ADS PubMed 33. Kattge J , Diaz S , Lavorel S et al. TRY - a global database of plant traits . Global Change Biol 2011 ; 17 : 2905 – 35 . https://doi.org/10.1111/j.1365-2486.2011.02451.x Google Scholar Crossref Search ADS 34. Warton DI , Wright IJ , Falster DS et al. Bivariate line-fitting methods for allometry . Biol Rev 2006 ; 81 : 259 – 91 . https://doi.org/10.1017/S1464793106007007 Google Scholar Crossref Search ADS PubMed 35. R Development Core Team . R: A Language and Environment for Statistical Computing . 2015 ; R Foundation for Statistic Computing , Vienna . 36. Niinemets Ü , Kull K . Leaf structure vs. nutrient relationships vary with soil conditions in temperate shrubs and trees . Acta Oecologica 2003 ; 24 : 209 – 19 . https://doi.org/10.1016/S1146-609X(03)00094-8 Google Scholar Crossref Search ADS 37. Niklas KJ. Plant Allometry: the scaling of form and process . 1994 ; University of Chicago Press , Chicago . 38. Güsewell S , Koerselman W . Variation in nitrogen and phosphorus concentrations of wetland plants . Perspect Plant Ecol Evol Syst 2002 ; 5 : 37 – 61 . https://doi.org/10.1078/1433-8319-0000022 Google Scholar Crossref Search ADS 39. Delgado-Baquerizo M , Reich PB , García-Palacios P et al. Biogeographic bases for a shift in crop C:N:P stoichiometries during domestication . Ecol Lett 2016 ; 19 : 564 – 75 . https://doi.org/10.1111/ele.12593 Google Scholar Crossref Search ADS PubMed 40. Chapin FS III , Shaver GR . Differences in growth and nutrient use among arctic plant growth forms . Funct Ecol 1989 ; 3 : 73 – 80 . https://doi.org/10.2307/2389677 Google Scholar Crossref Search ADS 41. van Ommen Kloeke AEE , Douma JC , Ordonez JC et al. Global quantification of contrasting leaf life span strategies for deciduous and evergreen species in response to environmental conditions . Global Ecol Biogeogr 2012 ; 21 : 224 – 35 . https://doi.org/10.1111/j.1466-8238.2011.00667.x Google Scholar Crossref Search ADS 42. Díaz S , Kattge J , Cornelissen JHC et al. The global spectrum of plant form and function . Nature 2016 ; 529 : 167 – 71 . https://doi.org/10.1038/nature16489 Google Scholar Crossref Search ADS PubMed 43. Ackerly DD , Reich PB . Convergence and correlations among leaf size and function in seed plants: A comparative test using independent contrasts . Am J Bot 1999 ; 86 : 1272 – 81 . https://doi.org/10.2307/2656775 Google Scholar Crossref Search ADS PubMed 44. Reich PB , Ellsworth DS , Walters MB et al. Generality of leaf trait relationships: a test across six biomes . Ecology 1999 ; 80 : 1955 – 69 . https://doi.org/10.1890/0012-9658(1999)080[5b1955:GOLTRA]5d2.0.CO;2 Google Scholar Crossref Search ADS 45. Chapin FS , Shaver GR , Kedrowski RA . Environmental controls over carbon, nitrogen and phosphorus fractions in Eriophorum vaginatum in Alaskan tussock tundra J Ecol 1986 ; 74 : 167 – 95 . https://doi.org/10.2307/2260357 Google Scholar Crossref Search ADS 46. Yan Z-B , Li P , Chen Y-H et al. Nutrient allocation strategies of woody plants: an approach from the scaling of nitrogen and phosphorus between twig stems and leaves . Sci Rep 2016 ; 6 : 20099 . https://doi.org/10.1038/srep20099 Google Scholar Crossref Search ADS PubMed 47. Mariotte P , Canarini A , Dijkstra FA . Stoichiometric N:P flexibility and mycorrhizal symbiosis favour plant resistance against drought . J Ecol 2017 ; 105 : 958 – 67 . https://doi.org/10.1111/1365-2745.12731 Google Scholar Crossref Search ADS 48. Yang X-J , Huang Z-Y , Zhang K-L et al. C:N:P stoichiometry of Artemisia species and close relatives across northern China: unravelling effects of climate, soil and taxonomy . J Ecol 2015 ; 103 : 1020 – 31 . https://doi.org/10.1111/1365-2745.12409 Google Scholar Crossref Search ADS 49. Wang Y-P , Law RM , Pak B . A global model of carbon, nitrogen and phosphorus cycles for the terrestrial biosphere . Biogeosciences 2010 ; 7 : 2261 – 82 . https://doi.org/10.5194/bg-7-2261-2010 Google Scholar Crossref Search ADS 50. Yang X , Post WM , Thornton PE et al. Global Gridded Soil Phosphorus Distribution Maps at 0.5-degree Resolution . 2014 ; ORNL DAAC , Oak Ridge, Tennessee, USA . http://dx.doi.org/10.3334/ORNLDAAC/1223 . 51. Yan Z-B , Tian D , Han W-X et al. An assessment on the uncertainty of the nitrogen to phosphorus ratio as a threshold for nutrient limitation in plants . Ann Bot 2017 ; 120 : 937 – 42 . https://doi.org/10.1093/aob/mcx106 Google Scholar Crossref Search ADS PubMed 52. Zhou X , Sun X , Du B-M et al. Multielement stoichiometry in Quercus variabilis under natural phosphorus variation in subtropical China . Sci Rep 2015 ; 5 : 937 – 42 . https://doi.org/10.1038/srep07839 53. Garten CT . Correlations between concentrations of elements in plants . Nature 1976 ; 261 : 686 – 8 . https://doi.org/10.1038/261686a0 Google Scholar Crossref Search ADS 54. Chapin FS III . The mineral nutrition of wild plants . Annu Rev Ecol Syst 1980 ; 11 : 233 – 60 . https://doi.org/10.1146/annurev.es.11.110180.001313 Google Scholar Crossref Search ADS 55. Sistla SA , Appling AP , Lewandowska AM et al. Stoichiometric flexibility in response to fertilization along gradients of environmental and organismal nutrient richness . Oikos 2015 ; 124 : 949 – 59 . https://doi.org/10.1111/oik.02385 Google Scholar Crossref Search ADS 56. Elser JJ , Brachen MES , Cleland EE et al. Global analysis of nitrogen and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems . Ecol Lett 2007 ; 10 : 1135 – 42 . https://doi.org/10.1111/j.1461-0248.2007.01113.x Google Scholar Crossref Search ADS PubMed 57. Sullivan BW , Alvarez-Clare S , Castle SC et al. Assessing nutrient limitation in complex forested ecosystems: alternatives to large-scale fertilization experiments . Ecology 2014 ; 95 : 668 – 81 . https://doi.org/10.1890/13-0825.1 Google Scholar Crossref Search ADS PubMed © The Author(s) 2017. Published by Oxford University Press on behalf of China Science Publishing & Media Ltd. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

National Science ReviewOxford University Press

Published: Sep 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off