LETTERS PUBLISHED: 6 JULY 2015 | ARTICLE NUMBER: 15080 | DOI: 10.1038/NPLANTS.2015.80 Grassland productivity limited by multiple nutrients Philip A. Fay et al.* Terrestrial ecosystem productivity is widely accepted to be agro-ecosystems, where the importance of P, K and micronutrients nutrient limited1. Although nitrogen (N) is deemed a key deter- is better understood20. minant of aboveground net primary production (ANPP)2,3, the Here we report the frequency, magnitude and global extent of prevalence of co-limitation by N and phosphorus (P) is increas- nutrient limitation of grassland ANPP by N, P and K combined 4–8 ingly recognized . However, the extent to which terrestrial with micronutrients (K+µ) at 42 grassland sites in the Nutrient productivity is co-limited by nutrients other than N and P has Network (NutNet; Supplementary Information and Borer et al.21). remained unclear. Here, we report results from a standardized All sites conducted the same standardized N-P-K+µ addition exper- factorial nutrient addition experiment, in which we added N, P iment, the definitive test for nutrient limitation. This approach over- and potassium (K) combined with a selection of micronutrients comes major constraints of previous meta-analyses4, that is, fi (K+μ), alone or in concert, to 42 grassland sites spanning ve differing nutrient addition rates and methodologies and the rarity continents, and monitored ANPP. Nutrient availability limited of factorial nutrient treatments in the underlying studies5. productivity at 31 of the 42 grassland sites. And pairwise com- Sites were located on five continents (Supplementary Fig. 1) and binations of N, P, and K+μ co-limited ANPP at 29 of the sites. spanned a 23-fold range in ANPP (Supplementary Fig. 2a), over 25 ° Nitrogen limitation peaked in cool, high latitude sites. Our find- in absolute latitude (actual latitudes 54 °N to 37 °S), nearly 3,500 m ings highlight the importance of less studied nutrients, such as in elevation, and wide ranges in mean annual precipitation (260– K and micronutrients, for grassland productivity, and point to 1,900 mm), mean annual temperature (0.3–22 °C; Supplementary significant variations in the type and degree of nutrient limit- Fig. 1), soil texture, and pretreatment soil N, P and K pools ation. We suggest that multiple-nutrient constraints must be (Supplementary Table 1). The sites included both native and previously considered when assessing the ecosystem-scale consequences cultivated grassland, and some sites were managed with burning, of nutrient enrichment. grazing or other practices. Thus, we were able to evaluate single and Terrestrial ecosystem productivity is widely accepted to be nutri- multiple-nutrient limitation of ANPP at global, continental and site ent limited1, and many studies have focused on limitation by a single spatial extents, as influenced by management, climate and soils. 2,3 2 nutrient, nitrogen . In grasslands, and in other systems, the role of N, P and K+µ were factorially applied annually to replicated 5 m additional nutrients is increasingly recognized. A recent meta- plots at the beginning of each site’s growing season at rates com- 4 22,23 analysis of 1400 N and P fertilization studies showed the limitation monly used in grassland fertilization experiments . The K+µ of terrestrial productivity by both N and P. Furthermore, these treatment included a micronutrient mix in the first treatment year nutrients were often synergistically co-limiting, where together only, to avoid micronutrient toxicity. Nutrient limitation of ANPP they limited productivity more than the sum of their individual was quantified by harvesting current year standing crop biomass limitations. This meta-analysis is the most comprehensive assess- for 3 to 5 years, and computing the log response ratio (LRR), the ment of ecosystem nutrient limitation to date. However, the natural logarithm of the ratio of treatment plot to control plot global extent and magnitude of multiple limitation by nutrients ANPP, a metric commonly used in meta-analyses24. other than N and P remains poorly understood in natural Multiple-nutrient limitation of ANPP often occurs in the form of systems4,5,9, including grasslands3,10, a critically endangered co-limitation5. Co-limitation is synergistic when the response to biome that accounts for approximately one-third of Earth’sterres- multiple nutrients is greater than the sum of the responses to each trial net primary production11. More importantly, multiple-nutri- nutrient added individually, additive if the multiple-nutrient ent limitation has not been experimentally tested in grasslands on response equals the sum of the individual nutrient responses, and a global scale using a standardized experimental approach. The sub-additive if less than the sum of the single-nutrient responses5. potential for synergistic co-limitation of grasslands by multiple Across all sites and years, the combined addition of N and P nutrients, or conversely, failing to account for the absence of increased ANPP by an average of 40% over controls (LRR = 0.34), single- or multiple-nutrient limitation, means we may misesti- compared to increases of only 18% (LRR = 0.16) for N individually mate the magnitude and extent of nutrient limitation of terrestrial and 9% for P individually (LRR = 0.09; P = 0.03, Fig. 1a and net primary productivity. Supplementary Fig. 3). This provides clear evidence for globally- Humans now produce more reactive N than is produced from all averaged synergistic co-limitation of ANPP by N and P in these natural terrestrial sources, primarily as fertilizers, industrial pro- grasslands. Synergistic co-limitation by N and P across sites and ducts, and through fossil fuel combustion12,13. Anthropogenic years emerged from frequent occurrence of NP co-limitation at increases in atmospheric N will result in further terrestrial N depo- the site level. Some form of NP co-limitation averaging 67% sition, altering ecosystem function13–17 and potentially increasing (LRR = 0.51) occurred at 60% of the sites, including sites on all limitation by other nutrients, such as P, K, or trace elements16.K five continents, with synergistic NP co-limitation at 13 sites and occurs in high concentrations in plant tissues18 and its uptake is sub-additive co-limitation at 12 sites (Supplementary Table 3). correlated with that of other nutrients19,20. Our understanding of Globally averaged synergistic co-limitation of grassland ANPP by limitation by nutrients other than N in grasslands lags that of N and P contradicts the long-held perception that N is the *A full list of authors and their affiliations appears at the end of the paper. NATURE PLANTS | VOL 1 | JULY 2015 | www.nature.com/natureplants 1 LETTERS NATURE PLANTS DOI: 10.1038/NPLANTS.2015.80 a All sites Significant effects b (N = 42) (no. of sites) 0 N (10) 5 P (8) 10 Nutrients 15 K+µ (3) added 20 N NP P (25) Rank 25 K+µ NK+µ (18) 30 NP Nutrients added NK+µ 35 PK PK+µ (9) +µ 40 NPK+µ NPK+µ (25) 45 −0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 −0.8 −0.6 −0.4 −0.2 0.0 0.2 0.4 0.6 0.8 1.0 Nutrient limitation of ANPP, ln(treatment/control) Nutrient limitation of ANPP, ln(treatment/control) Figure 1 | Nutrient limitation of ANPP (LRRs, ln(treatment/control)) by N, P and K plus year 1 micronutrients (K+μ). a, Mean ± s.e.m. over all years available at each site. Means for all sites are closed symbols. Means of sites where individual nutrient treatments had significant positive effects (defined in b) are open symbols. b, Site nutrient limitation of ANPP ranked by the magnitude of limitation, averaged over all available years. Vertical dashed lines mark the effect sizes approximating a 0.05 significance level for negative and positive nutrient effects on ANPP. Note that the sites rank in a different order in each nutrient treatment. predominant nutrient limiting grassland productivity3,10 and high- single- and multiple-nutrient K limitation, potentially augmented lights a critical role for P. by micronutrient addition during year 1, occurs more frequently There was no globally averaged single- or multiple-nutrient than previously recognized20,25. limitation involving K+µ (P > 0.32, Supplementary Table 2). Although single- and multiple-nutrient limitation of ANPP was However, this did not preclude frequent site-level single- and mul- widespread, there were nonetheless 13 sites that showed no evidence tiple-nutrient K+µ limitation of grassland ANPP. ANPP was limited of multiple-nutrient limitation, and 12 of the 13 showed no evidence at three sites by K+µ alone and co-limited by NK+µ at 18 sites, both of single-nutrient limitation (Supplementary Table 3). The presence averaging 48% (LRR = 0.39; Fig. 1a and Supplementary Table 3). and magnitude of nutrient limitation may depend on site climate, 26,27 Additionally, ANPP was co-limited by PK+µ at nine sites by an soil development or fertility . For these reasons, greater site- average of 52% (LRR = 0.42), approaching the magnitude of level limitation of ANPP by one nutrient may correlate with ANPP limitation at sites where N (57%) or P (54%) individually greater limitation by one or more additional nutrients. Indeed, limited ANPP (Fig. 1a). Site-level NK+µ and PK+µ co-limitation site-level individual nutrient limitation of ANPP increased with occurred in synergistic, additive and sub-additive forms, and in limitation by other individual nutrients (R2 0.08–0.15; Fig. 2a,b,d), total, single- or multiple-nutrient limitation involving K+µ occurred and ANPP limitation by nutrient pairs increased with that of a at 21 sites again occurring on all five continents (Supplementary third individual nutrient (R2 0.07–0.21; Fig.