Received: 26 November 2020 | Revised: 4 March 2021 | Accepted: 15 May 2021 DOI: 10.1002/ece3.7734 ORIGINAL RESEARCH High foliar K and P resorption efficiencies in old-­growth tropical forests growing on nutrient-­poor soils Ifigenia Urbina1,2 | Oriol Grau1,2,3 | Jordi Sardans1,2 | Olga Margalef1,2 | Guillermo Peguero1,2 | Dolores Asensio1,2 | Joan LLusià1,2 | Romà Ogaya1,2 | Albert Gargallo-­Garriga1,2 | Leandro Van Langenhove4 | Lore T. Verryckt4 | Elodie A. Courtois3 | Clément Stahl3 | Jennifer L. Soong5 | Jerome Chave6 | Bruno Hérault7,8 | Ivan A. Janssens4 | Emma Sayer9,10 | Josep Peñuelas1,2 1 CREAF, Centre de Recerca Ecològica i Aplicacions Forestals, Bellaterra, Spain 2 Consejo Superior de Investigaciones Científicas, Global Ecology Unit, Universidad Autònoma de Barcelona, Bellaterra, Spain 3 CIRAD, UMR EcoFoG (AgroParisTech, CNRS, INRA, Univ Antilles, Univ. Guyane), Kourou, French Guiana 4 Department of Biology, Centre of Excellence PLECO (Plants and Ecosystems), University of Antwerp, Wilrijk, Belgium 5 Climate and Ecosystem Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA 6 Laboratoire Evolution et Diversité Biologique, UMR5174, CNRS–­Université Paul Sabatier–­IRD, Toulouse cedex 9, France 7 Cirad, UR Forêts & Sociétés, Université de Montpellier, Montpellier, France 8 Institut National Polytechnique Félix Houphouët-­Boigny (INP-­HB), Yamoussoukro, Ivory Coast 9 Lancaster Environment Centre, Lancaster University, Lancaster, UK 10 Smithsonian Tropical Research Institute, Panama, Republic of Panama Correspondence Ifigenia Urbina, CREAF, Centre de Recerca Abstract Ecològica i Aplicacions Forestals, E08193 Resorption is the active withdrawal of nutrients before leaf abscission. This mecha- Bellaterra (Cerdanyola del Vallès), Catalonia, Spain. nism represents an important strategy to maintain efficient nutrient cycling; however, Email: ifigeniaurbinab@gmail.com resorption is poorly characterized in old-­growth tropical forests growing in nutrient-­ Funding information poor soils. We investigated nutrient resorption from leaves in 39 tree species in two French ANR (CEBA, ref. ANR-­10-­L ABX-­ tropical forests on the Guiana Shield, French Guiana, to investigate whether resorp- 25-­01; TULIP, ref. ANR-­10-­L ABX-­0 041; ANAEE-­France: ANR-­11-­INBS-­0 001); tion efficiencies varied with soil nutrient, seasonality, and species traits. The stocks Spanish Government, Grant/Award Number: of P in leaves, litter, and soil were low at both sites, indicating potential P limitation CGL2016-­79835-­P; Spanish Ministry of Science grant PID2019-­110521GB-­I 00, of the forests. Accordingly, mean resorption efficiencies were higher for P (35.9%) Grant/Award Number: PID2019-­ and potassium (K; 44.6%) than for nitrogen (N; 10.3%). K resorption was higher in 110521GB-­I 00; ERC Synergy project, IMBALANCE-­P, Grant/Award Number: the wet (70.2%) than in the dry (41.7%) season. P resorption increased slightly with SyG-­2013-­610028; Catalan government, decreasing total soil P; and N and P resorptions were positively related to their foliar Grant/Award Number: SGR2017-­1005 concentrations. We conclude that nutrient resorption is a key plant nutrition strategy in these old-­growth tropical forests, that trees with high foliar nutrient concentra- tion reabsorb more nutrient, and that nutrients resorption in leaves, except P, are This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2021 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. Ecology and Evolution. 2021;11:8969–8982.  www.ecolevol.org | 8969 8970 | URBINA et al. quite decoupled from nutrients in the soil. Seasonality and biochemical limitation played a role in the resorption of nutrients in leaves, but species-­specific require- ments obscured general tendencies at stand and ecosystem level. KEYWORDS nitrogen, nutrient, phosphorus, potassium, resorption, soil, stocks, tropical forest 1 | I NTRO D U C TI O N Nutrient resorption has been postulated as a response of plant species to nutrient-­limiting conditions, which could play an important Tropical forests store large amounts of aboveground biomass, role in efficient nutrient cycling in tropical forests on infertile soils even though they grow on old, highly weathered infertile soils (Vitousek, 1982). However, despite decades of research, the results (Gersmehl, 1976; Vitousek & Sanford, 1986). This issue has attracted about relationship between soil nutrient concentration and nutrient the attention of the ecological community, and several mechanisms resorption are contradictory and not clear (Aerts & Chapin, 2000; have been proposed to explain this apparent paradox (Bond, 2010; Brant & Chen, 2015; Killingbeck, 1996; Wright & Westoby, 2003; Sayer & Banin, 2016; Turner et al., 2018; Vitousek & Sanford, 1986). Yuan & Chen, 2009). On another hand, foliar nutrient concentrations One such mechanism is efficient nutrient cycling, whereby greater seem to play an important role in nutrient resorption. It has been nutrient retention and internal recycling of essential nutrients reported at global scale that resorption decreases with increasing by plants minimize nutrient losses in tropical forests with low soil foliar nutrient status (Kobe et al., 2005; Vergutz et al., 2012). Several nutrient concentrations (Brant & Chen, 2015; Grau et al., 2017; factors can though influence nutrient resorption, which can obscure Vitousek, 1984). direct relationships with soil nutrient availability. Water availability Resorption (also called “retranslocation,” “remobilization,” or is clearly a major driver of resorption, as the reduction in phloem flux “reabsorption”) is the active withdrawal of nutrients before abscis- transport and water scarcity leads to lower nutrient resorption due sion (Fries, 1952; Hill, 1980; Killingbeck, 1996) during senescence, to the premature abscission of leaves (Estiarte & Peñuelas, 2015; which leads to a series of metabolic changes associated with a Killingbeck, 2003). Phenology also substantially affects resorption decrease in auxin levels, protein breakdown, chlorophyll degra- because flower or fruit production acts as a sink from nutrients in dation, and eventually leaf abscission (Fuente & Leopold, 1968). senescent leaves (Estiarte & Peñuelas, 2015; Killingbeck, 2003). The resorption of foliar nutrients is activated by kinetin signals, Phylogenetic relatedness, understanded as the results of contrast- which promote nutrient mobilization through the phloem from ing ecological pressures that make specific niches for different nu- old senescent leaves (the source) to other plant organs such as trient acquisition strategies, is also expected to play a role in the stems, roots, or new leaves (sinks) (Hill, 1980). This process of in- efficiency of nutrient resorption among plant species, as closely re- ternal nutrient recycling plays an important role in plant nutrition lated taxa often have similar resorption efficiencies (Aerts & Van der and survival, allowing plants to be more independent of external Piejl, 1993; Killingbeck, 2003; Cantón et al., 2005). conditions (Aerts, 1996b; Brant & Chen, 2015; van Heerwaarden The resorption of nitrogen (N), phosphorus (P), and potassium (K) et al., 2002; Killingbeck, 1996; Reed et al., 2012). Nutrient re- is of particular interest, because they are three of the most import- sorption can be estimated by resorption efficiency, defined as ant among the 20 essential nutrients for plant development (Aerts & the percent reduction of a nutrient between green and senesced Chapin, 2000). Nitrogen in plant tissues is mainly present in organic leaves (Killingbeck, 2003). It can be also estimated as resorption forms as the principal constituent of proteins and nucleic acids. Plants proficiency, defined as the level to which nutrient concentration acquire N through root uptake of nitrate or ammonium from the soil is reduced in senesced leaves (Aerts, 1996a; Killingbeck, 1996). and return N to the soil through litterfall. Biological fixation of atmo- The resorption of nutrients within a plant can vary greatly from spheric molecular N (N2) can also provide large inputs of N to support year to year depending on the environmental conditions, such plant growth (Chapin, 1980) that is subsequently cycled within the as soil water availability, timing of abscission and the extent of ecosystem. As the cycling of N is predominantly biological, young shading (Killingbeck, 2003). In addition, the resorption capacity of natural ecosystems are generally poor in N (Aerts & Chapin, 2000; plants is determined by biochemical nutrient limitation, whereby Turner & Condron, 2013; Walker & Syers, 1976), whereas N is typ- the level of nutrient immobilization in leaf compounds (enzymes, ically not considered to be limiting in old-­growth lowland tropical proteins, DNA, RNA, etc.) prevents their mobilization from se- forests (Tanner et al., 1998; Vitousek & Sanford, 1986). nescing leaves. Overcoming biochemical nutrient limitation can In contrast to N, P in plant tissues is present in inorganic forms require high energetic costs and some nutrients in structural and and as a key component of co-­enzymes (ATP and NADPH), nucleic recalcitrant compounds, or those used for enzymatic machinery acids, and in larger proportions in phospholipid's membranes. P is during foliar senescence may be largely unavailable for export provided almost entirely by the weathering of P-­containing miner- from leaves (Killingbeck, 2003). als (e.g., apatite) from the parent material (Walker & Syers, 1976). URBINA et al. | 8971 Old-­growth tropical forests on ancient soils are therefore generally with high growth rates and low wood densities (acquisitive life-­ considered to be P-­limited (Bruijnzeel, 1991; Vitousek et al., 2010) history strategy), as persistence strategy. because long-­term weathering of primary minerals depletes total P 5. Resorption efficiencies will be more similar among more closely during soil development and results in increasing dominance of oc- related than phylogenetically distant species. cluded P fractions (Yang & Post, 2011). High concentrations of 1:1 clay and ferric sesquioxides bind P in soils, making it inaccessible to To address these hypotheses, we explored the stocks and the most organisms and bringing the soil system toward what is known resorption efficiencies of essential elements (N, P, and K) in abo- as “terminal steady state” (Walker & Syers, 1976). veground (leaves and leaf-­litter) and soil compartments during the Potassium represents the most abundant cation (K+) in plant cells, wet and dry season in two old-­growth tropical forests growing on and its central role in water economy, photosynthetic capacity, and old, nutrient-­poor soils. nutrient transport in plants makes K essential for plant development (Aerts & Chapin, 2000; Sardans & Peñuelas, 2015). Potassium is very soluble and therefore highly mobile in plant tissues and is almost 2 | M ATE R I A L S A N D M E TH O DS completely provided by the weathering of soil parental material. Old tropical soils, especially those of the Precambrian shield, con- 2.1 | Study area tain little available K, as described for P, due to the weathering and leaching of this cation from the soil over a very long time (Rosolem The study was conducted in French Guiana, a tropical region in the et al., 2010; Sayer & Banin, 2016). Potassium has been described as northeastern part of South America (Figure S1a). French Guiana a co-­limiting nutrient for tree growth in wet lowland tropical forests is dominated by tropical forest growing on old, nutrient poor soils in Costa Rica (Baribault et al., 2012), and the addition of both K and developed from the Precambrian Guiana Shield formation (Bongers N increased sapling growth in a long-­term fertilization experiment in et al., 2001; Courtois et al., 2018; Epron et al., 2006), including the Panama (Wright et al., 2011). two study sites: Paracou (5°18′N, 52°53′W) and Nouragues (4°05′N, Based on the differences in the relative mobilities and avail- 52°41′W; Figure S1a). Paracou is located 15 km inland from the coast, abilities of N, P, and K in lowland tropical forests, we would expect in an undisturbed forested area characterized by smooth mosaic hills the resorption efficiencies to differ among these nutrients. Given (Epron et al., 2006; Janssens et al., 1998). The Nouragues Research that soil N availability is not likely to be limiting and is less mobile Station is situated in the center of the country, which is covered by in plant tissues due to their highly presence in organic forms, N extensive primary forest with granitic hills (Bongers et al., 2001). resorption efficiency should be lower than P or K resorption effi- The climate at both sites is typical of seasonally evergreen tropical ciencies. By contrast, we would expect similar P and K resorption rainforests, with a rainy season from December to July and a dry efficiencies because their availability in the soil is low, whereas both season from August to November. Mean annual rainfall is similar in elements are highly mobile in plant tissues (Rosolem et al., 2010; both study sites (2,990 and 3,041 mm y-­1 at Nouragues and Paracou, Wright et al., 2011). Although previous work demonstrated that re- respectively) (Aguilos et al., 2018; Bongers et al., 2001). The soils sorption efficiency in trees is generally higher for P than N (Vergutz at both sites are classified as Oxisols in the USDA soil classification et al., 2012; Yuan & Chen, 2009), less studies have reported K re- (Anjos et al., 2015), with pH values between 3.7 and 4.5. sorption efficiencies (but see Ławniczak, 2011; Vergutz et al., 2012). Furthermore, there is a dearth of information on nutrient resorp- tion for tropical tree species, and thus, the importance of resorp- 2.2 | Experimental design tion for efficient nutrient cycling in lowland tropical forests remains unquantified. We calculated N, P, and K resorption efficiencies for At each site, 12 experimental plots were established to represent 39 tropical tree species to investigate the role of phylogeny in deter- the spatial variability, including the top of the hills (top plots), the mining resorption efficiencies, and to explore potential relationships slope of the hills (slope plots), and the bottom of the hills next to between nutrient resorption and plant traits. We hypothesized that: rivulets (bottom plots; Figure S1b). At each topographical posi- tion, we established four plots of 50 m × 50 m, each with an inner 1. Resorption efficiencies for P and K will be higher than for 20 m × 20 m sampling area, and a buffer of 20–­100 m between ad- N because of their lower availability in the soil and greater jacent plots. We marked, tagged, and identified all trees in each plot mobility in the plant. (50 m × 50 m) and measured diameter at breast height (DBH) for all 2. Nutrient resorption will be higher for trees with higher foliar nu- trees with larger than 10 cm. trient concentrations, as strategy to avoid nutrient loss. Two field campaigns were carried out in 2015, during the rainy 3. Nutrient resorption efficiencies will be higher in the wet than the season from May to the end of June and during the dry season from dry season, because water conductance and nutrient transport the beginning of October until late November. To collect soil sam- are greater in the wet season. ples and leaf litter on the soil surface, we established five sampling 4. Species with slow growth rates and high wood densities (conserv- points within each plot, one in the center and one in each corner ative life-­history strategy) will resorb more nutrients than species of the inner sampling area (Figure S1c). At each sampling point, 8972 | URBINA et al. we collected all mixed leaf litter from the soil surface within an Standard Reference Materials (IAEA-­N1, IAEA-­N2, IAEA-­N3, area of 20 cm × 20 cm and we took soil samples using a soil auger USGS-­4 0, and USGS-­41). (4-­cm-­diameter) at two depths (0–­15 cm and 15–­30 cm). The litter The concentrations of P and K were determined by inductively samples were oven-­dried at 70°C to constant weight. The soil sam- coupled plasma (ICP) mass spectrometry (ICP-­MS Agilent 7500 ce) ples were sieved through a 2-­mm sieve and dried at 105°C for 24 hr using oven-­dried, pulverized samples (0.1 g soil and 250 mg leaf ma- to determine dry weight, which was divided by the corer volume to terial) digested in 5 ml of concentrated HNO3 (Milestone Ultrawave obtain the soil bulk density. digestor; Sorisole, BG, Italy). The accuracies of the digestions and To quantify nutrient resorption, we collected green and senes- analytical procedures were assessed using blanks (5 ml of HNO3 and cent leaves from three emergent canopy trees and two subcanopy 2 ml of H2O2) and certified standards: tomato leaf (NIST 1573a) for trees inside the sampling area of each plot. Trees were chosen based biomass and Montana soil (NIST 2711a) and sewage-­amended Soil on DBH and to maximize the functional traits and number of species (CRM005) for soil (NIST). sampled (i.e., if two trees belonged to the same species, we chose Soil extractable-­P was determined by two methods, Bray's acid the next tree of a different species but with similar size). Green fluoride extraction (Bray-­P; Bray & Kurtz, 1945) and Olsen's bicar- leaves (5–­10 leaves per tree depending on leaf size) were collected bonate extraction (Olsen-­P; Olsen et al., 1954), using sieved, dried by tree climbers from the upper, mostly sunlit leaves and lower, soils. P concentrations in both extracts were measured by induc- mainly shaded leaves to represent two contrasting canopy condi- tively coupled plasma optical emission spectrometry (iCAP 6300 tions. Green leaves were immediately frozen in liquid N. If the climb- Duo; Thermo Fisher Scientific, Germany). ers could not reach trees in the 20 m × 20 m sampling area, then we To calculate nutrient and carbon stocks per hectare, we used chose trees in the remaining area of the 50 m × 50 m plot follow- the BRIDGE database (Baraloto et al., 2010) to obtain mean wood ing the same criteria. Tree climbers also collected senescent leaves density and SLA for all species, and foliar elemental composition (% that were still attached to the tree and could be easily detached by C, N, P, and K) for trees that were not sampled but were present in shaking the branch. Across both study sites, due to the availability our plots. Mean growth rates (mm of DBH increase y-­1) were ob- of fresh litter in different seasons and logistical constrains, we were tained from the Guyafor database (Grau et al., 2017). Preliminary able collected green and senescent leaves from a total of 39 species analyses showed that foliar C and nutrient concentrations did not during the dry season (Table S1) and from a subset of 18 species in differ significantly between the upper and lower parts of the canopy the wet season (Table S2). Foliar samples were freeze-­dried (Christ (Figure S2), so we used the means of both canopy levels for the cal- Freeze Dryer ALPHA 1-­2 LDplus, Osterode am Harz, Germany) and culations of stocks and resorption efficiencies. senescent leaves were oven-­dried at 70°C to constant weight. The total leaf weight for each species was calculated using an allometric coefficient obtained by the power-­law fit, with DBH (cm) as a predictor of leaf weight (kg), using data from Chave et al. (2014). 2.3 | Chemical analysis Total leaf weight was then calculated as: Green leaves, litter, senescent leaves, and soil samples were Leaf weight (kg) = 0.02634434 × basal area (1) ground with a ball mill (Retsch, model MM400, Restch GmbH) and weighed with an AB204 Mettler Toledo (Mettler Toledo) bal- where the basal area of a tree is π × (DBH/2)2. For more information ance. Foliar C and N were determined by gas chromatography about the allometric coefficient, see the Supporting Information. and expressed per unit dry weight. The amount of sample used We then calculated foliar C and nutrient stocks per tree as for subsequent analyses was based on the C and N concentra- the product of leaf weight and foliar C or nutrient concentrations. tions (%) of each sample type: For green leaves, senescent leaves, Finally, foliar C and nutrient stocks per unit area were calculated as and leaf litter, we used 4.5 mg of pulverized dry sample, and for the sum of all the stocks contained in leaves inside the 20 x 20 m soil samples, we used 9 mg for 0–­15 cm depth (samples with high sampling area: organic-­matter content) and 11 mg for 15–­3 0 cm depth (samples n with moderate to low organic matter content). The C and N con- (2) ∑ Y= X centrations of leaves, senescent leaves, and leaf-­litter samples n=1 were analyzed using an elemental analyzer interfaced to an iso- tope ratio mass spectrometer (PDZ Europa ANCA-­G SL and PDZ where Y represents the stocks in kg/ha of a given element (C, N, P, Europa 20-­20; Sercon Ltd.), and soils samples were analyzed with or K) per plot, n represents the number of trees in the 20 m × 20 m an elemental analyzer (Elementar Vario EL Cube or Micro Cube; sampling area, and X represents the stocks of a given element in the Elementar Analysensysteme GmbH). During analysis, the samples leaves of each tree. were interspersed with several replicates of at least two labora- Leaf-­litter nutrient and C stocks were estimated as the product tory standards, which were selected to be compositionally similar of the sample dry weight per unit area and the element concentra- to the samples being analyzed, and had been previously calibrated tions in the sample. against National Institute of Standards and Technology (NIST) URBINA et al. | 8973 Soil weight per area was calculated as the product of bulk density 2.5 | Phylogenetic analysis 3 (g/cm ) and core depth (15 cm), and soil C and nutrient stocks were then calculated as the product of soil weight and the C or nutrient We assessed phylogenetic effects on nutrient resorption efficiencies concentrations in the sample. by calculating Pagel's λ and Blomberg's K, following Münkemüller We calculated nutrient-­resorption efficiencies as described by et al. (2012), that are two of the most common phylogenetic indi- Killingbeck, 1996: ces used to measure the phylogenetic signals. Both indices quan- tify the tendency of related species to resemble each other more XGl − XSl than species drawn at random from the same phylogenetic tree, × 100 (3) XGl where values close to 1 indicate a significant phylogenetic effect. We based our phylogenetic tree on only 31 of the 39 species due to where XGl and XSl represent the nutrient concentrations of green the lack of precise phylogenetic information for eight of the species and senescent leaves, respectively. As we did not measure mass (Aniba rosaeodora, Chrysophyllum poniferum, Eugenia culcullata, Inga loss during senescence, we were unable to apply the mass-­area loss jenmanii, Licania densiflora, Myrcia splendens, Paloue guianensis, and correction (van Heerwaarden et al., 2002), which could lead to un- Vochysia sabatieri). The phylogenetic tree was based on a phyloge- derestimates of resorption efficiencies by c. 10% (Han et al., 2013). netic tree that covers most tree species present in French Guiana (J. However, the mass-­loss correction would equally affect the calculated Chave et al., unpublished data). resorption efficiencies of all nutrients. We also calculated resorption Results were considered significant at p < .05, and we report proficiencies, defined as nutrient concentration in senescent leaves marginally significant trends for p < .1; means are given ±standard (Killingbeck, 1996) (Table S4). deviations throughout. 2.4 | Statistical analyses 3 | R E S U LT S All analyses were performed in R 3.5.2 (R Core Team, 2017) using Nouragues generally had greater nutrient stocks than Paracou, the “FactoMineR” package (Le et al., 2008) for multivariate analyses, but P was the scarcest nutrient stored in the leaves, leaf-­litter, the “nlme” package (Pinheiro et al., 2017; RCoreTeam, 2017) for lin- and soil in both study sites (Table 1, Figure S3). Soil P stocks were ear mixed effects models, and the “picante” (Kembel et al., 2010) and similarly low between sites, whereas soil C, N, and K stocks were “phytools” (Revell, 2012) packages for phylogenetic analyses. higher at Nouragues than Paracou at both depths. Accordingly, soil We used principal component analysis (PCA) with standardized N:P and K:P ratios (per unit mass) were also higher at Nouragues variables to visualize the relationship between the N, P, and K re- than Paracou (Figure S4). Leaf litter mass per area was higher in sorption efficiencies, foliar and soil chemical compositions (C, N, Nouragues than Paracou (Table 1), and leaf litter stocks of N, P, and P, and K), and functional characteristics (mean growth rate, wood K were also higher at Nouragues, although C and nutrient stocks density, DBH, and SLA). We then used linear models to test for dif- in green leaves did not differ significantly between sites (Table 1, ferences between sites in C, N, P, or K stocks for each compartment Figure S3). The mean foliar N:P ratio (across all species at both study (leaves, leaf litter, upper soil layer, and lower soil layer) separately, in sites) was 29.4 ± 7.0. order to determine whether there were differences regarding the Resorption efficiencies (mean across all species at both study nutrient scarcity between the two sites selected. sites in dry season, N = 39) were higher for P and K (35.9% and We used linear mixed effects models to test the influence of sea- 44.6%, respectively) than N (10.3%; p <.0001. Table 2), and the re- son and species on nutrient resorption efficiencies across sites (both sorption efficiencies for P and K were generally higher than for N for studies sites were considered in the models). First, we assessed dif- all species studied (Figure 1), in agreement with our first hypothesis. ferences in the resorption efficiencies for N, P, and K using the data The range of variation in the resorption efficiencies was smaller for for all species sampled in both study sites (N = 39), with nutrient as N (−31% to 69.5%) than for P (−91.6% to 86.7%) or K (−91.6% to the predictor and species identity as the random effect in the model; 90.6%), and there were more negative values for N (Figure 1 and the significance of differences between N, P, and K was subse- Table S1 for resorption efficiencies by species). In contrast, we found quently determined using Tukey post hoc analysis. We then used the lower variation in the P (0.069%–­0.007%) and K (0.95%–­0.043%) data for the species sampled in both seasons (N = 18) to test the ef- resorption proficiencies than N (2.7%–­0.91%) (Table S4), and pro- fect of season on resorption efficiency for each nutrient separately; ficiencies were not related with leaf nutrient concentration. The re- models included season as the predictor and species identity as the sorption efficiencies of all three nutrients tended to be higher in the random effect. Finally, we assessed the influence of tree functional wet than in the dry season (Figure 2). Across all species, K resorption traits (SLA, DBH, wood density, and growth rate) on resorption effi- was significantly higher in the wet (70.19 ± 18.7%) than the dry sea- ciency; we constructed separate models for each nutrient and trait, son (41.7 ± 25.26%; N = 18. p <.01), whereas N and P resorption using the functional trait as a predictor and species identity as the efficiencies did not differ significantly between seasons (Tables S3), random effect. and thus, our third hypothesis was only partially supported. 8974 | URBINA et al. Nitrogen and P resorption efficiencies increased with foliar N and on foliar nutrient concentrations, but largely decoupled from soil nu- 2 2 P concentrations across species (R = 0.27, p < .0001 and R = 0.15, trient stocks (except for P), indicating a key role for plant-­internal p = .013, respectively), whereas K resorption efficiency was only mar- nutrient cycling for maintaining growth despite infertile tropical ginally related to foliar K concentration (R2 = 0.08, p = .08; Figure 3), soils. Interestingly, despite the lack of a phylogenetic signal for re- so our second hypothesis establishing that nutrient resorption will be sorption efficiency, we found some evidence that P resorption might higher for trees with higher foliar nutrient concentrations was only be greater in trees with acquisitive life history strategies, contrary to partially supported. Phosphorus resorption efficiency declined with what we expected. increasing total soil P (R2 = 0.14, p = .02) but was not related to ex- tractable P (Olsen or Bray's P; Figure S5), and there was no relationship between N or K resorption efficiencies and N or K soil concentrations, 4.1 | Nutrient stocks respectively (Figure S6). Accordingly, the PCA indicated that foliar C, N, and P concentrations were uncoupled from soil C, N and P concen- Our results indicated that Nouragues has better nutrient status trations. The first and second principal components explained 22.61% than Paracou, with higher nutrient inputs from leaf litter to the and 18.75% of the total variability, respectively, whereby soil variables soil at Nouragues. Nonetheless, the low stocks of P in the soil at generally aligned with the first axis, while resorption and leaf concen- both sites (Table 1), as well as high soil N:P and K:P ratios, indi- tration aligned with the second axis (Figure 4). cate potential P-­limitation, as expected in this tropical region (Grau The relationships between nutrient resorption efficiencies and et al., 2017; Oliveira et al., 2015; Sayer & Banin, 2016; Vitousek & functional characteristics of the species were usually weak but statis- Farrington, 1997). The N:P ratios we measured (>30) are much higher tically significant (Figure S7). Nitrogen resorption efficiency declined than the mean value for tropical forests growing on Oxisols (23.5; with increasing DBH (R 2 = 0.14, p = .037), whereas P resorption effi- Townsend et al., 2007), suggesting P limitation (Mo et al., 2019) of plant ciency increased marginally with SLA (R 2 = 0.13, p = .096; Figure S7) growth at our study sites. 2 and declined with increasing wood density (R = 0.16, p = .026) and The allocation patterns of C, N, P, and K stocks were consistent there was also a trend toward declining K resorption efficiency with with what has been described for other tropical forests in South increasing wood density (R 2 = 0.12, p = .052). There was no relation- America and Africa (Bruijnzeel, 1991; Greenland & Kowal, 1960; ship between nutrient resorption and mean growth rates, and thus, Sayer & Banin, 2016; Yavitt et al., 2009), although our study did not our fourth hypothesis that stated that species with conservative life-­ include stocks in woody tissue, which represent a large part of the history strategy will resorb more nutrients than species acquisitive C and nutrients stored in aboveground biomass in tropical forests life-­history strategy was rejected. (Heineman et al., 2016; Tanner, 1985). The higher N stocks in soil Finally, in contrast to fifth hypothesis, we detected no phylo- than leaves and leaf-­litter conform to our expectation of greater genetic signal for nutrient resorption efficiency, as Pagel's λ and N availability in old tropical soils (Sayer & Banin, 2016; Turner & Blomberg's K indices showed no relationship between resorption ef- Condron, 2013; Vitousek et al., 2010). The total concentrations of ficiencies and the phylogenetic distances among species (Table S6). K in the soil were surprisingly high, and roughly equivalent to N. However, the amount of soil K present in available (in solution or exchangeable) or unavailable (fixed or structural) forms varies by soil 4 | D I S CU S S I O N type (Rosolem et al., 2010) and our measurements of total soil K may therefore not reflect K availability to plants. Our study of 39 tropical tree species across two sites demonstrated that the resorption efficiencies of N, P, and K are partially dependent TA B L E 1 Mean values for the mass and C, N, P, and K stocks (kg/ha) at two tropical forest sites in French Guiana, showing stocks in plant compartments (leaves and leaf litter) and in the soil at 0–­15 and 15–­30 cm depth Compartment Site Mass (kg/ha) C (kg/ha) N (kg/ha) P (kg/ha) K (kg/ha) Leaf Nouragues 8,314 (2,144) 4,112 (2,144) 158 (80.3) 4.65 (2.50) 41.4 (24.2) Litter Nouragues 9,544 (1,411) 4,565 (720) 139* (19.2) 2.21 · (0.47) 16.4* (5.54) Soil (0–­15 cm) Nouragues 1,533,250 (275,876) 50,104* (4,436) 3,583* (290) 205 (137) 2,033 (83.6) Soil (15–­3 0 cm) Nouragues 1,699,068 (323,762) 28,209* (4,410) 2,222.35* (279) 206 (140) 2,551 (80.4) Leaf Paracou 12,243.61 (5,388) 5,689 (5,388) 207 (217) 6.52 (6.41) 54.6 (62.6) Litter Paracou 8,641 (1,603) 4,157 (820) 113* (21.6) 1.84 · (0.53) 9.68* (6.22) Soil (0–­15 cm) Paracou 1,675,000 (346,661.1) 36,320* (8,391) 2,566* (630) 158 (61.2) 1,009 · (182) Soil (15–­3 0 cm) Paracou 1,777,000 (436,931.8) 17,172* (4,552) 1,361* (375) 142 (49.1) 1,265 · (108) Note: Values are means and standard deviations (in parentheses) for n = 5 per site for litter and soil, stocks in leaves are the results of allometric equation. Asterisks indicate significant differences at p < .05 and dots marginally significant differences at p < .1 between sites. URBINA et al. | 8975 TA B L E 2 Significant differences at community level in the N, P, and K resorption efficiencies. (a) Output of the estimated regression parameters, standard errors, and t and p values for the linear mixed model. The estimated standard deviation associated with the random effect, σspecies, is 23.25. (b) Results of the post hoc test (pairwise comparison) using Tukey's method (a) Estimate Standard error df t-­value p-­Value Intercept 44.61 5.17 76 8.62 <.0001 N −34.31 5.07 76 −6.75 <.0001 P −8.71 5.07 76 1.71 .09 (b) Contrast Estimate Standard error df t-­ratio p-­Value K-­N 34.31 5.08 76 6.75 <.0001 K-­P 8.71 5.08 76 1.71 .2055 P-­N −25.6 5.08 76 −5.04 <.0001 4.2 | Nutrient resorption and soil nutrient less clear. Meta-­analyses revealed that phosphorus resorption effi- concentration ciency increases from high to low latitudes (Yuan & Chen, 2009) and N:P resorption efficiency increases with the latitude and decreases Plant adaptations to low P availability are common in tropical for- with mean annual temperature and precipitation (Reed et al., 2012). ests and include a range of strategies such as mycorrhizal asso- These patterns reflect variation in soil types and nutrient status in ciations, carboxylates, and phosphatase production to solubilizes different climatic zones (Reed et al., 2012), and greater P resorp- inorganic P bounded in the soil (Batterman et al., 2018; Bucking tion efficiency in the tropics is interpreted as a response to low et al., 2012; Hofmann et al., 2016; Hohenheim & Stuttgart, 1989; soil P availability (Hidaka & Kitayama, 2011; Yuan & Chen, 2009). Orwin et al., 2011; Sheldrake et al., 2017). However, nutrient resorp- However, several empirical studies have found no relationships be- tion from senescent leaves allows plants to be more independent tween nutrient resorption and soil nutrient concentrations along fer- from external resources and saves the metabolic cost of symbiotic tility gradients or in fertilization experiments (Aerts, 1996b; Brant & associations or the production of enzymes (Brant & Chen, 2015). In Chen, 2015; Chapin, 1980; Killingbeck, 2003), whereas other studies support of our first hypothesis, the resorption efficiencies of all the reported a decrease in N and P resorption efficiencies with fertiliza- species were generally higher for P and K than for N (Figure 1 and tion (Mayor et al., 2014; Yuan & Chen, 2015). Total soil P at our study Table S1), suggesting that greater resorption of P and K might be an sites was lower than in other tropical forests on similar weathered important mechanism to conserve nutrients that would otherwise soils (Yang & Post, 2011; Yavitt, 2000), and the narrow range of soil P be lost through leaching (K) or rendered inaccessible by sorption to concentrations (20–­470 ppm) may account for the lack of correlation soil minerals (P). Indeed, assuming that our mass-­based measure- between P resorption and soil P concentration. However, our study ments underestimated resorption by c. 10% (Han et al., 2013; Van demonstrates clearly that resorption efficiency is higher for P than Heerwaarden et al., 2003), around half of our tree species had P N, which we interpret as an adaptive strategy of plants in response resorption efficiencies above the global average for evergreen an- to low P availability, and the lack of clear relationships between soil giosperms (50%–­60%; Yuan & Chen, 2009; Vergutz et al., 2012) and fertility and nutrient resorption efficiencies might be explained by several species had resorption efficiencies >80% for P and K, which biochemical limitation. indicates a scarce condition of these nutrients at our sites (Tsujii et al., 2017). By contrast, only one species had an N resorption effi- ciency >60%, which conforms to our expectation of lower N resorp- 4.3 | Nutrient resorption and biochemical limitation tion compared to P or K because of the relatively high availability of N in these old tropical forests (Lambers et al., 2008). Biochemical limitation influences resorption efficiency because Nutrient resorption efficiency is assumed to be higher when the level of nutrient immobilization in foliar tissue largely deter- soil nutrient availability is limited (Hidaka & Kitayama, 2011; Reed mines the capacity of nutrient recovery from senescent leaves et al., 2012; Tsujii et al., 2017). Although we found a weak rela- (Killingbeck, 2003; Kobe et al., 2005). Hence, the functional roles and tionship between P resorption efficiencies and total soil P con- storage forms of nutrients likely also explain the variation in resorp- centrations, there was no relationship between P resorption and tion efficiencies. In our study, it is striking that the order of nutrient extractable P or between N or K resorption efficiencies and soil N or resorption efficiencies for all species (K ≥ P > N) was the opposite K concentrations. Links between soil fertility and nutrient resorption of the level of nutrient immobilization in plant tissues (N > P > K). have been established on a global scale, but the evidence for links N is highly immobilized in organic forms in plants (e.g., in proteins, between soil fertility and nutrient resorption at regional scales is enzymes, and nucleic acids; Chapin, 1980; Cantón et al., 2005), 8976 | URBINA et al. F I G U R E 1 Resorption efficiencies for (a) nitrogen (N), (b) phosphorus (P), and (c) potassium (K) in 39 tropical tree species sampled during the dry season at two lowland tropical forest sites in French Guiana; note that the y axis and species rankings differ among panels whereas a high proportion of P is present in inorganic forms (Condron we observed in our study species. In our study, P and K resorption et al., 2013; Mo et al., 2019; Walker & Syers, 1976) and K is a free varied more among species than N (Figure 2, Table S1) and greater cation in plant cells (Sardans & Peñuelas, 2015). Nutrients more in- variation in P than N resorption efficiencies has been observed for volved in the enzymatic metabolism of foliar senescence, such as N, different plant growth forms (Aerts & Chapin, 2000) as well as in or energetic forms necessary to export nutrients from leaves, such a global dataset of woody plant species (Han et al., 2013). Greater as P, will be partially unavailable for resorption (Killingbeck, 2003). variability in P versus N resorption efficiency could reflect the com- Potassium can thus be reabsorbed more easily than P, and a greater parably greater importance of P resorption for plant nutrient conser- proportion of P can be reabsorbed than N, which matches the pattern vation (Han et al., 2013), or the greater extent of N immobilization URBINA et al. | 8977 F I G U R E 2 Resorption efficiencies for (a) nitrogen (N), (b) phosphorus (P) and (c) potassium (K) for 18 tropical tree species sampled in the wet (left-­hand panels) and dry (right-­hand panels) seasons at two tropical forest sites in French Guiana; note that the y axis and species rankings differ among panels in plant cell walls (Tsujii et al., 2017). However, foliar concentrations In general, trees with higher foliar nutrient concentrations were of P are generally more variable than N due to luxury P consumption able to reabsorb more nutrients than trees with low foliar nutrient and greater capacity of plants to store P (Ostertag, 2010); it is there- concentration at our sites, in accordance with our second hypoth- fore possible that variation in resorption efficiencies for N and P is at esis. Although global meta-­analyses demonstrated that nutrient least partly related to their concentrations in leaves. resorption efficiency decreases with increasing leaf nutrient status 8978 | URBINA et al. F I G U R E 3 Nutrient resorption efficiencies and their correlations with foliar nutrient concentration (dw/dw) based on the 39 tropical tree species sampled at both study sites in the dry season. (a) N resorption efficiency versus N in leaves, (b) P resorption versus P in leaves, and (c) K resorption versus K in leaves. Coefficients for the significant regressions and R 2 are displayed in the lower-­right corner of each panel (Kobe et al., 2005; Vergutz et al., 2012), this relationship may not in the wet than the dry season. Given the high mobility of K, greater hold true within highly infertile sites. Several studies in tropical for- resorption efficiencies during the wet season could suggest an im- ests on P-­poor soils have demonstrated similarly high P resorption portant effect of water availability on internal K cycling. However, efficiencies or low P concentrations in leaf litter to those measured as K is readily leached from the canopy, greater leaching of K from at our site (e.g., Cai & Bongers, 2007; Richardson et al., 2008; Wood senesced compared to green leaves during the wet season would et al., 2011). We thus propose that resorption may be a particularly present as higher resorption efficiencies, and we cannot currently important plant strategy to avoid nutrient losses in this very infertile discount this possibility. Furthermore, the interpretation of seasonal tropical soil and high interspecific variation in leaf and litter chemis- differences in resorption efficiencies may be confounded by differ- try (Hättenschwiler et al., 2008) may also entail distinct resorption ences in the nutrient requirements and acquisition strategies among efficiencies. tree species. The differences in the species rankings (from highest to lowest resorption efficiency) between the wet and dry season (Figure 2) indicates that seasonal differences in resorption may be 4.4 | Nutrient resorption and the effect of highly species-­specific. seasonality In contrast to our third hypothesis, we found no seasonal differences 4.5 | Effects of phylogeny and species functional in N and P resorption efficiencies, possibly because higher immobi- characteristics on nutrient resorption lization of these nutrients in recalcitrant foliar compounds may pre- vent retranslocation even when phloem water fluxes are high. By Species with high wood densities are associated with slow growth contrast, mean K resorption efficiency across all species was higher rates and more conservative strategies (Chave et al., 2009; URBINA et al. | 8979 species-­specific process in these tropical forests to conserve nutri- ent that are less available in soil. 4.6 | Conclusions Our results show that nutrient resorption is an important plant nutrition strategy in these old-­growth tropical forests growing on weathered, nutrient-­poor soils. Trees with high foliar nutrient con- centration reabsorb more nutrients, and nutrients that are scarcer in soil presented higher resorption efficiencies. The species-­specific requirements and the functional strategy seem to strongly deter- mine the resorption of nutrients. Furthermore, seasonality and bi- ochemical limitation play also a role in the resorption of nutrients, which make it difficult to obtain a general conclusion about this mechanism in this ecosystem. More studies that address the role of functional strategy and the individual's competition on the nutrient F I G U R E 4 Links between nutrient resorption efficiencies, resorption will help to clarify the control of this important nutrient foliar nutrient concentrations, soil nutrients, and tree species traits visualized in an ordination plot based on principal component recycling process in tropical forests. analysis (PCA), showing foliar carbon (C), nitrogen (N), phosphorus (P), and potassium (K) concentrations (blue arrows), soil C, N, P, and AC K N OW L E D G M E N T S K concentrations (black arrows), N, P, and K resorption efficiencies This research was financially supported by the Spanish Government (dotted line arrows), as well as tree growth rate, diameter at (project CGL2016-­79835-­P), the Catalan government (project breast height (DBH), wood density, and specific leaf area (SLA) as functional traits (red arrows) SGR2017-­1005), and the ERC Synergy project, SyG-­2013-­610028 IMBALANCE-­P. We acknowledge support from the French ANR (CEBA, ref. ANR-­10-­L ABX-­25-­01; TULIP, ref. ANR-­10-­L ABX-­0 041; Muller-­L andau, 2004; Nascimento et al., 2005). As nutrient resorp- ANAEE-­France: ANR-­11-­INBS-­0 001). tion represents an important plant strategy to conserve nutrients, we hypothesized that more conservative species, with slow growth C O N FL I C T O F I N T E R E S T rates and high wood density, would have higher resorption efficien- There are not conflicts of interest. cies than fast-­growing species. By contrast, we found that species with high wood densities reabsorbed less P and K, and that growth AU T H O R C O N T R I B U T I O N rate was not correlated with N, P or K resorption efficiency. We thus Ifigenia Urbina Barreto: Conceptualization (lead); Data cura- rejected our fourth hypothesis. Furthermore, closely related species tion (lead); Formal analysis (lead); Funding acquisition (sup- did not have more similar resorption efficiencies than distantly re- porting); Investigation (lead); Methodology (lead); Project lated species, and we thus rejected our fifth hypothesis. Previous administration (lead); Resources (lead); Software (lead); Supervision work in tropical forest established that N and P resorption efficiency (lead); Validation (lead); Visualization (lead); Writing-­original draft declined with increasing leaf toughness, which was attributed to (lead); Writing-­review & editing (lead). Oriol Grau: Conceptualization greater allocation of nutrients to structural compounds, which are (lead); Data curation (lead); Formal analysis (lead); Funding acquisi- not easily remobilized (Wood et al., 2011). It is thus conceivable that tion (supporting); Investigation (lead); Methodology (lead); Project the allocation of nutrients in species with conservative growth strat- administration (supporting); Resources (supporting); Software (sup- egies actually reduces their capacity to reabsorb nutrients. Overall, porting); Supervision (lead); Validation (lead); Visualization (lead); our data did not provide strong evidence for a relationship between Writing-­original draft (equal); Writing-­review & editing (equal). resorption efficiency and functional characteristics, although the Jordi Sardans: Conceptualization (equal); Data curation (sup- negative relationship between DBH and N resorption efficiency porting); Formal analysis (supporting); Funding acquisition (lead); could reflect the capacity of large trees with extensive roots systems Investigation (equal); Methodology (supporting); Project admin- to acquire more N from the soil than smaller trees. Additional work istration (lead); Supervision (lead); Validation (equal); Writing-­ on inter-­ versus intraspecific differences in nutrient resorption and original draft (equal); Writing-­review & editing (lead). Olga Margal: the relationship between functional traits and nutrient resorption Writing-­review & editing (supporting). Guille Peguero: Writing-­ could clarify the relative importance of abiotic and biotic controls review & editing (supporting). Dolores Asensio: Writing-­review & of resorption efficiencies. The high P and K resorption efficien- editing (supporting). Juan Llusia: Writing-­review & editing (support- cies for all species, despite the differences between them and the ing). Romà Ogaya: Writing-­review & editing (supporting). Albert lack of phylogenetic relationship, suggest that resorption is a highly Gargallo Gariga: Writing-­review & editing (supporting). Leandro 8980 | URBINA et al. Van Langenhove: Data curation (equal); Formal analysis (support- and creating legends for soil maps. In P. Schad, C. Van Huyssteen, & E. Micheli (Eds.). World Soil Resources Reports No. 106. FAO. https:// ing); Methodology (equal); Writing-­review & editing (supporting). doi.org/10.1017/S0014​47970​6394902 Lore Verryckt: Writing-­review & editing (equal). Elodie A. Courtois: Baraloto, C., Timothy Paine, C. E., Patiño, S., Bonal, D., Hérault, B., & Writing-­review & editing (supporting). Clément Stahl: Writing-­ Chave, J. (2010). Functional trait variation and sampling strategies in review & editing (supporting). Jennifer L. Soong: Data curation (sup- species-­rich plant communities. Functional Ecology, 24(1), 208–­216. https://doi.org/10.1111/j.1365-­2435.2009.01600.x porting); Formal analysis (supporting); Investigation (supporting); Baribault, T., Kobe, R., & Finley, A. (2012). Tropical tree growth is cor- Methodology (supporting); Supervision (supporting); Validation related with soil phosphorus, potassium, and calcium, though not (supporting); Writing-­review & editing (supporting). Jerome Chave: for legumes. Dryad Digital Repository. Ecological Monographs, 82(2), Data curation (lead); Funding acquisition (supporting); Investigation 189–­203. https://doi.org/10.1890/11-­1013.1 (supporting); Methodology (supporting); Resources (equal); Writing-­ Batterman, S. A., Hall, J. S., Turner, B. L., Hedin, L. O., LaHaela Walter, J. K., Sheldon, P., & van Breugel, M. (2018). Phosphatase activity and review & editing (supporting). Bruno Hérault: Data curation (sup- nitrogen fixation reflect species differences, not nutrient trading porting); Formal analysis (supporting); Methodology (supporting); or nutrient balance, across tropical rainforest trees. Ecology Letters, Supervision (supporting); Writing-­review & editing (supporting). 21(10), 1486–­1495. https://doi.org/10.1111/ele.13129 Ivan A. Janssens: Conceptualization (supporting); Funding acqui- Bond, W. J. (2010). Do nutrient-­poor soils inhibit development of forests? A nutrient stock analysis. Plant and Soil, 334(1), 47–­60. https://doi. sition (lead); Investigation (supporting); Resources (supporting); org/10.1007/s1110​4-­010-­0 440-­0 Supervision (supporting); Validation (supporting); Writing-­review & Bongers, F., Charles-­Dominique, P., Forget, P. M., & Théry, M. (Eds.). (2001). editing (supporting). Emma Sayer: Investigation (lead); Supervision Nouragues : Dynamics and plant-­animal interactions in a neotropical rain- (lead); Validation (lead); Visualization (lead); Writing-­review & editing forest. https://www.sprin​ger.com/gp/book/97814​02001239 Brant, A. N., & Chen, H. Y. H. (2015). Patterns and mechanisms of nutri- (lead). Josep Peñuelas: Conceptualization (lead); Data curation (sup- ent resorption in plants. Critical Reviews in Plant Sciences, 34(5), 471–­ porting); Formal analysis (supporting); Funding acquisition (lead); 486. https://doi.org/10.1080/07352​689.2015.1078611 Investigation (lead); Methodology (equal); Project administration Bray, R. H., & Kurtz, L. T. (1945). Determination of total, organic and avail- (lead); Resources (supporting); Software (supporting); Supervision able forms of phosphorus in soils. Soil Science, 59, 39–­46. https://doi. org/10.1097/00010​694-­19450​1000-­0 0006 (lead); Validation (lead); Visualization (lead); Writing-­original draft Bruijnzeel, L. A. (1991). Nutrient input—­output budgets of tropical forest (lead); Writing-­review & editing (lead). ecosystems: A review. Journal of Tropical Ecology, 7(1), 1–­24. https:// doi.org/10.1017/S0266​46740​0 005010 DATA AVA I L A B I L I T Y S TAT E M E N T Bucking, H., Liepold, E., & Ambilwade, P. (2012). The role of the mycorrhi- zal symbiosis in nutrient uptake of plants and the regulatory mech- The data used in this study will be made publicly available at time of anisms underlying these transport processes. In N. K. Dhal & S. C. publication in the public repository http://glonu​teco.creaf.cat/ Sahu (Eds.), Plant science, Vol. 1, 1st edn. (pp. 107–­138). IntechOpen. https://doi.org/10.5772/52570 ORCID Cai, Z. Q., & Bongers, F. (2007). Contrasting nitrogen and phosphorus Ifigenia Urbina https://orcid.org/0000-0002-8541-950X resorption efficiencies in trees and lianas from a tropical montane rain forest in Xishuangbanna, south-­west China. Journal of Tropical Guillermo Peguero https://orcid.org/0000-0002-6464-1486 Ecology, 23(1), 115–­118. https://doi.org/10.1017/S0266​46740​ Emma Sayer https://orcid.org/0000-0002-3322-4487 6003750 Josep Peñuelas https://orcid.org/0000-0002-7215-0150 Cantón, F. R., Suárez, M. F., & Cánovas, F. M. (2005). Molecular aspects of nitrogen mobilization and recycling in trees. Photosynthesis Research, 83, 265–­278. REFERENCES Chapin, F. S. (1980). The mineral nutrition of wild plants. Annual Aerts, R. (1996a). Nutrient resorption from senescing leaves of peren- Review of Ecology and Systematics, 11, 233–­260. https://doi. nials: Are there general patterns? British Ecological Society, 84(4), org/10.2307/2096908 597–­608. Chave, J., Coomes, D., Jansen, S., Lewis, S. L., Swenson, N. G., & Zanne, Aerts, R. (1996b). Nutrient resorption from senescing leaves of peren- A. E. (2009). Towards a worldwide wood economics spectrum. nials: Are there general patterns. Journal of Ecology, 84(4), 597–­608. Ecology Letters, 12(4), 351–­366. https://doi.org/10.1111/j.1461-­0248.​ https://doi.org/10.1016/S0020​-­1693(97)05934​-­3 2009.01285.x Aerts, R., & Chapin, F. S. (2000). The mineral nutrition of wild plants re- Chave, J., Réjou-­Méchain, M., Búrquez, A., Chidumayo, E., Colgan, M. S., visited: A re-­evaluation. Advances in Ecological Research, 30, 55. Delitti, W. B. C., Duque, A., Eid, T., Fearnside, P. M., Goodman, R. C., Aerts, R., & Van der Piejl, M. J. (1993). A simple model to explain the dom- Henry, M., Martínez-­Yrízar, A., Mugasha, W. A., Muller-­L andau, H. C., inance of low-­productive perennials in nutrient-­poor habitats. Oikos, Mencuccini, M., Nelson, B. W., Ngomanda, A., Nogueira, E. M., Ortiz-­ 66(1), 144–­147. https://doi.org/10.2307/3545208 Malavassi, E., … Vieilledent, G. (2014). Improved allometric models to Aguilos, M., Stahl, C., Burban, B., Hérault, B., Courtois, E., Coste, S., estimate the aboveground biomass of tropical trees. Global Change Wagner, F., Ziegler, C., Takagi, K., & Bonal, D. (2018). Interannual Biology, 20(10), 3177–­3190. https://doi.org/10.1111/gcb.12629 and seasonal variations in ecosystem transpiration and water use Condron, L. M., Spears, B. M., Haygarth, P. M., Turner, B. L., & efficiency in a tropical rainforest. Forests, 10(1), 14. https://doi. Richardson, A. E. (2013). Role of legacy phosphorus in improving org/10.3390/f1001​0 014 global phosphorus-­use efficiency. Environmental Development, 8(1), Anjos, L., Gaistardo, C., Deckers, J., Dondeyne, S., Eberhardt, E., 147–­148. https://doi.org/10.1016/j.envdev.2013.09.003 Gerasimova, M., Harms, B., Jones, A., Krasilnikov, P., Reinsch, T., Courtois, E. A., Stahl, C., Van den Berge, J., Bréchet, L., Van Langenhove, Vargas, R., & Zhang, G. (2015). World reference base for soil re- L., Richter, A., Urbina, I., Soong, J. L., Peñuelas, J., & Janssens, I. A. sources 2014. International soil classification system for naming soils (2018). Spatial variation of Soil CO2, CH4 and N2O fluxes across URBINA et al. | 8981 topographical positions in tropical forests of the Guiana shield. Killingbeck, K. T. (1996). Nutrients in senesced leaves: Keys to the search Ecosystems, 21(7), 1445–­1458. https://doi.org/10.1007/s1002​ for potential resorption and resorption proficiency. Ecology, 77(6), 1-­018-­0232-­6 1716–­1727. https://doi.org/10.2307/2265777 Epron, D., Bosc, A., Bonal, D., & Freycon, V. (2006). Spatial variation of Killingbeck, K. T. (2003). Nutrient resorption. Plant Cell Death Processes, soil respiration across a topographic gradient in a tropical rain forest 1, 215–­226. https://doi.org/10.1016/B978-­01252​0915-­1/50017​-­5 in French Guiana. Journal of Tropical Ecology, 22(5), 565–­574. https:// Kobe, R. K., Lepczyk, C. A., & Iyer, M. (2005). Resorption efficiency doi.org/10.1017/S0266​46740​6003415 decreases with increasing green leaf nutrients in a global data set. Estiarte, M., & Peñuelas, J. (2015). Alteration of the phenology of leaf Ecology, 86(10), 2780–­2792. https://doi.org/10.1890/04-­1830 senescence and fall in winter deciduous species by climate change: Lambers, H., Raven, J. A., Shaver, G. R., & Smith, S. E. (2008). Plant Effects on nutrient proficiency. Global Change Biology, 21(3), 1005–­ nutrient-­acquisition strategies change with soil age. Trends in 1017. https://doi.org/10.1111/gcb.12804 Ecology and Evolution, 23(2), 95–­103. https://doi.org/10.1016/j. Fries, N. (1952). Variation in the content of phosphorus, nucleic acids and tree.2007.10.008 adenine in the leaves of some deciduous trees during the autumn. Ławniczak, A. E. (2011). Nitrogen, phosphorus, and potassium resorp- Plant and Soil, 4(1), 29–­42. tion efficiency and proficiency of four emergent macrophytes from Fuente, R. K., & Leopold, A. C. (1968). Senescence processes. Plant nutrient-­rich wetlands. Polish Journal of Environmental Studies, 20(5), Physiology, 43(1), 1496–­1502. 1227–­1234. Gersmehl, P. (1976). An alternative biogeography. Annals of the Association Le, S., Josse, J., & Husson, F. (2008). FactoMineR: An R package for of American Geographers, 66(2), 223–­241. https://doi.org/10.1111/ multivariate analysis. Journal of Statistical Software, 25. https://doi. j.1467-­8306.1976.tb010​87.x org/10.18637/​jss.v025.i01 Grau, O., Peñuelas, J., Ferry, B., Freycon, V., Blanc, L., Desprez, M., Mayor, J. R., Wright, S. J., & Turner, B. L. (2014). Species-­specific re- Baraloto, C., Chave, J., Descroix, L., Dourdain, A., Guitet, S., Janssens, sponses of foliar nutrients to long-­term nitrogen and phosphorus ad- I. A., Sardans, J., & Hérault, B. (2017). Nutrient-­c ycling mechanisms ditions in a lowland tropical forest. Journal of Ecology, 102(1), 36–­4 4. other than the direct absorption from soil may control forest struc- https://doi.org/10.1111/1365-­2745.12190 ture and dynamics in poor Amazonian soils. Scientific Reports, 7, Mo, Q., Li, Z., Sayer, E. J., Lambers, H., Li, Y., Zou, B. I., Tang, J., Heskel, 45017. https://doi.org/10.1038/srep4​5017 M., Ding, Y., & Wang, F. (2019). Foliar phosphorus fractions reveal Greenland, D. J., & Kowal, J. M. (1960). Nutrient content of the moist how tropical plants maintain photosynthetic rates despite low soil tropical forest of Ghana. Plant and Soil, 2, 154–­174. https://doi. phosphorus availability. Functional Ecology, 33(3), 503–­513. https:// org/10.1007/BF013​77368 doi.org/10.1111/1365-­2435.13252 Han, W., Tang, L., Chen, Y., & Fang, J. (2013). Relationship between the Muller-­L andau, H. C. (2004). Interspecific and inter-­site variation in relative limitation and resorption efficiency of nitrogen vs phospho- wood specific gravity of tropical Trees1. Biotropica, 36(1), 20. https:// rus in woody plants. PLoS One, 8(12), 1–­11. https://doi.org/10.1371/ doi.org/10.1646/02119 journ​al.pone.0083366 Münkemüller, T., Lavergne, S., Bzeznik, B., Dray, S., Jombart, T., Schiffers, Hättenschwiler, S., Aeschlimann, B., Coûteaux, M.-­M ., Roy, J., & K., & Thuiller, W. (2012). How to measure and test phylogenetic Bonal, D. (2008). High variation in foliage and leaf litter chem- signal. Methods in Ecology and Evolution, 3(4), 743–­756. https://doi. istry among 45 tree species of a neotropical rainforest com- org/10.1111/j.2041-­210X.2012.00196.x munity. The New Phytologist, 179(1), 165–­175. https://doi. Nascimento, H. E. M., Laurance, W. F., Condit, R., Laurance, S. G., org/10.1111/j.1469-­8137.2008.02438.x D’Angelo, S., & Andrade, A. C. (2005). Demographic and life-­history Heineman, K. D., Turner, B. L., & Dalling, J. W. (2016). Variation in wood correlates for Amazonian trees. Journal of Vegetation Science, 16(6), nutrients along a tropical soil fertility gradient. New Phytologist, 625. 10.1658/1100-­9233(2005)016[0625:dalcfa]2.0.co;2 211(2), 440–­454. https://doi.org/10.1111/nph.13904 Oliveira, R. S., Galvão, H. C., de Campos, M. C. R., Eller, C. B., Pearse, Hidaka, A., & Kitayama, K. (2011). Allocation of foliar phospho- S. J., & Lambers, H. (2015). Mineral nutrition of campos rupestres rus fractions and leaf traits of tropical tree species in re- plant species on contrasting nutrient-­impoverished soil types. New sponse to decreased soil phosphorus availability on Mount Phytologist, 205(3), 1183–­1194. https://doi.org/10.1111/nph.13175 Kinabalu, Borneo. Journal of Ecology, 99(3), 849–­857. https://doi. Olsen, S. R., Cole, C. V., Watanabe, F. S., & Dean, L. A. (1954). Estimation org/10.1111/j.1365-­2745.2011.01805.x of available phosphorus in soils by extraction with sodium bicarbonate. Hill, J. (1980). The remobilization of nutrients from leaves. Journal of United States Department of Agriculture. Plant Nutrition, 2(4), 407–­4 44. https://doi.org/10.1080/01904​16800​ Orwin, K. H., Kirschbaum, M. U. F., St John, M. G., & Dickie, I. A. (2011). 9362788 Organic nutrient uptake by mycorrhizal fungi enhances ecosystem Hofmann, K., Heuck, C., & Spohn, M. (2016). Phosphorus resorption carbon storage: A model-­based assessment. Ecology Letters, 14(5), by young beech trees and soil phosphatase activity as dependent 493–­502. https://doi.org/10.1111/j.1461-­0248.2011.01611.x on phosphorus availability. Oecologia, 181(2), 369–­379. https://doi. Ostertag, R. (2010). Foliar nitrogen and phosphorus accumulation re- org/10.1007/s0044​2-­016-­3581-­x sponses after fertilization: An example from nutrient-­limited Hawaiian Hohenheim, U., & Stuttgart, D. (1989). Citric acid excretion and precipita- forests. Plant and Soil, 334(1), 85–­98. https://doi.org/10.1007/s1110​ tion of calcium citrate in the rhizosphere of white lupin (Lupinus albus 4-­010-­0281-­x L.). Plant, Cell & Environment, 12, 285–­292. Pinheiro, J., Bates, D., DebRoy, S., & Sarkar, D. & R Core Team (2017). Janssens, I. A., Têtè, S., Ceulemans, R., & Guyane, G. D. (1998). Soil CO2 nlme: Linear and nonlinear mixed effects models. R package version efflux rates in different tropical vegetation types in French Guiana. 3.1-­131. Annals of Forest Science, 55, 671–­680. R Core Team (2017). R: A language and environment for statistical comput- Kembel, S. W., Cowan, P. D., Helmus, M. R., Cornwell, W. K., Morlon, ing. R Foundation for Statistical Computing. H., Ackerly, D. D., Blomberg, S. P., & Webb, C. O., & Integrating, P. Reed, S. C., Townsend, A. R., Davidson, E. A., & Cleveland, C. C. R., tools for. (2010). Picante: R tools for integrating phylogenies and (2012). Stoichiometric patterns in foliar nutrient resorption across ecology. Bioinformatics, 26, 1463–­1464. https://doi.org/10.1093/ multiple scales. New Phytologist, 196(1), 173–­180. https://doi. bioin​forma​tics/btq166 org/10.1111/j.1469-­8137.2012.04249.x 8982 | URBINA et al. Revell, L. J. (2012). phytools: An R package for phylogenetic comparative Vitousek, P. M. (1984). Litterfall, nutrient cycling, and nutrient lim- biology (and other things). Methods in Ecology and Evolution, 3, 217–­ itation in tropical forests. Ecology, 65(1), 285–­298. https://doi. 223. https://doi.org/10.1111/j.2041-­210X.2011.00169.x org/10.2307/1939481 Richardson, S. J., Allen, R. B., & Doherty, J. E. (2008). Shifts in Vitousek, P. M., & Farrington, H. (1997). Nutrient limitation and soil leaf N: P ratio during resorption reflect soil P in temper- development: Experimental test of a biogeochemical theory. ate rainforest. Functional Ecology, 22(4), 738–­745. https://doi. Biogeochemistry, 37(1), 63–­75. https://doi.org/10.1023/A:10057​ org/10.1111/j.1365-­2435.2008.01426.x 57218475 Rosolem, C. A., Sgariboldi, T., Garcia, R. A., & Calonego, J. C. (2010). Vitousek, P. M., Porder, S., Houlton, B. Z., Chadwick, O. A., & Houlton, Z. Potassium leaching as affected by soil texture and residual fertil- (2010). Terrestrial phosphorus limitation: Mechanisms, implications, ization in tropical soils. Communications in Soil Science and Plant and nitrogen —­ Phosphorus interactions. Ecological Applications, Analysis, 41(16), 1934–­1943. https://doi.org/10.1080/00103​ 20(1), 5–­15. https://doi.org/10.1890/08-­0127.1 624.2010.495804 Vitousek, P. M., & Sanford, R. L. (1986). Nutrient cycling in moist tropi- Sardans, J., & Peñuelas, J. (2015). Potassium: A neglected nutrient in cal forest. Annual Review of Ecology and Systematics, 17(1), 137–­167. global change. Global Ecology and Biogeography, 24(3), 261–­265. https://doi.org/10.1146/annur​ev.es.17.110186.001033 https://doi.org/10.1111/geb.12259 Walker, T. W., & Syers, J. K. (1976). The fate of phospho- Sayer, E. J., & Banin, L. F. (2016). Tree nutrient status and nutrient cy- rus during pedogenesis. Geoderma, 15(1), 1–­19. https://doi. cling in tropical forest—­Lessons from fertilization experiments. In org/10.1016/0016-­7061(76)90066​-­5 G. Goldstein & L. Santiag (Eds.), Tropical tree physiology. Tree phys- Wood, T. E., Lawrence, D., & Wells, J. A. (2011). Inter-­specific variation in iology, Vol. 6 (pp. 275–­297). Springer. https://doi.org/10.1007/978-­ foliar nutrients and resorption of nine canopy tree species in a neo- 3-­319-­27422​-­5 tropical rainforest is related to physical leaf traits. Functional Ecology, Sheldrake, M., Rosenstock, N. P., Revillini, D., Olsson, P. A., Mangan, 43(5), 544–­551. S., Sayer, E. J., Wallander, H., Turner, B. L., & Tanner, E. V. J. (2017). Wright, I. J., & Westoby, M. (2003). Nutrient concentration, re- Arbuscular mycorrhizal fungal community composition is altered by sorption and lifespan: Leaf traits of Australian sclero- long-­term litter removal but not litter addition in a lowland tropical phyll species. Functional Ecology, 17(1), 10–­19. https://doi. forest. New Phytologist, 214(1), 455–­467. https://doi.org/10.1111/ org/10.1046/j.1365-­2435.2003.00694.x nph.14384 Wright, S. J., Yavitt, J. B., Wurzburger, N., Turner, B. L., Tanner, E. V. Tanner, E. V. J. (1985). Jamaican Montane forests: Nutrient capital and J., Sayer, E. J., Santiago, L. S., Kaspari, M., Hedin, L. O., Harms, K. cost of growth. Journal of Ecology, 73(2), 553–­568. E., Garcia, M. N., & Corre, M. D. (2011). Potassium, phosphorus, or Tanner, E. V. J., Vitousek, P. M., & Cuevas, E. (1998). Experimental inves- nitrogen limit root allocation, tree growth, or litter production in tigation of nutrient limitation of forest growth on wet tropical moun- a lowland tropical forest. Ecology, 92(8), 1616–­1625. https://doi. tains. Ecology, 79(1), 10–­22. 10.1890/0012-­9658(1998)079[0010:EI org/10.1890/10-­1558.1 ONLO]2.0.CO;2 Yang, X., & Post, W. M. (2011). Phosphorus transformations as a function Townsend, A. R., Cleveland, C. C., Asner, G. P., & Bustamante, M. M. of pedogenesis: A synthesis of soil phosphorus data using Hedley C. (2007). Controls over foliar N: P ratios in tropical rain forests. fractionation method. Biogeosciences, 8(10), 2907–­2916. https://doi. Ecology, 88(1), 107–­118. 10.1890/0012-­9658(2007)88[107:COFNRI org/10.5194/bg-­8-­2907-­2011 ]2.0.CO;2 Yavitt, J. B. (2000). Nutrient dynamics of soil derived from different Tsujii, Y., Onoda, Y., & Kitayama, K. (2017). Phosphorus and nitrogen parent material on Barro Colorado Island. Panama. Biotropica, 32(2), resorption from different chemical fractions in senescing leaves of 198–­207. https://doi.org/10.1111/j.1744-­7429.2000.tb004​62.x tropical tree species on Mount Kinabalu, Borneo. Oecologia, 185(2), Yavitt, J. B., Harms, K. E., Garcia, M. N., Wright, S. J., He, F., & Mirabello, 171–­180. https://doi.org/10.1007/s0044​2-­017-­3938-­9 M. J. (2009). Spatial heterogeneity of soil chemical properties in a Turner, B. L., Brenes-­Arguedas, T., & Condit, R. (2018). Pervasive phos- lowland tropical moist forest, Panama. Australian Journal of Soil phorus limitation of tree species but not communities in tropical Research, 47(7), 674–­687. https://doi.org/10.1071/SR08258 forests. Nature, 555(7696), 367–­370. https://doi.org/10.1038/natur​ Yuan, Z. Y., & Chen, H. Y. H. (2009). Global-­scale patterns of nutrient e25789 resorption associated with latitude, temperature and precipita- Turner, B. L., & Condron, L. M. (2013). Pedogenesis, nutrient dynamics, tion. Global Ecology and Biogeography, 18(1), 11–­18. https://doi. and ecosystem development: The legacy of T.W. Walker and J.K. org/10.1111/j.1466-­8238.2008.00425.x Syers. Plant and Soil, 367(1-­2), 1–­10. https://doi.org/10.1007/s1110​ Yuan, Z. Y., & Chen, H. Y. H. (2015). Decoupling of nitrogen and phospho- 4-­013-­1750-­9 rus in terrestrial plants associated with global changes. Nature Climate van Heerwaarden, L. M., Toet, S., & Aerts, R. (2002). Current measures of Change, 5(5), 465–­469. https://doi.org/10.1038/nclim​ate2549 nutrient resorption efficiency lead to a substantial underestimation of real resorption efficiency: Facts and solutions. Oikos, 2001, 77–­78. Van Heerwaarden, L. M., Toet, S., & Aerts, R. (2003). Current measures S U P P O R T I N G I N FO R M AT I O N of nutrient resorption efficiency lead to a substantial underestima- Additional supporting information may be found online in the tion of real resorption efficiency: Facts and solutions. Oikos, 101(3), Supporting Information section. 664–­669. https://doi.org/10.1034/j.1600-­0706.2003.12351.x Vergutz, L., Manzoni, S., Porporato, A., Novais, R. F., & Jackson, R. B. (2012). Global resorption efficiencies and concentrations of carbon and nutrients in leaves of terrestrial plants. Ecological Monographs, How to cite this article: Urbina I, Grau O, Sardans J, et al. 82(2), 205–­220. https://doi.org/10.1890/11-­0 416.1 High foliar K and P resorption efficiencies in old-­growth Vitousek, P. M. (1982). Nutrient cycling and nutrient use effi- tropical forests growing on nutrient-­poor soils. Ecol Evol. ciency. The American Naturalist, 119(4), 553–­572. https://doi. 2021;11:8969–­8982. https://doi.org/10.1002/ece3.7734 org/10.1086/283931