1 LEAF LITTER LEACHING AND NUTRIENT CYCLING IN LOWLAND TROPICAL FORESTS By LAURA A. SCHREEG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DE GREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011
2 2011 Laura A. Schreeg
3 ACKNOWLEDGMENTS I am very thankful for the support I have received from both the University of Florida (UF) and the Smithsonian Tropical Research Institute (STRI) in Panama. I would like to thank my advisor Michelle Mack who encouraged me to explore the questions that motivated me and expanded my view of plant soil interactions. I am also greatly appreciative of support from Ben Turner (STRI). The Turner lab hosted me for a year and half at STRI and cemented my interest in phosphorus biogeochemistry. Nick Comerford set the groundwork for how I view terrestrial phosphorus cycling and helped initiate the ideas I will be exploring during my post doctoral position. Ted Schu urs previous work in tropical systems was instrumental in motivating my PhD research direction. Jack Ewel has been a mentor throughout my time at UF. My peers, especially Alex Cheesman, Hollie Hall, Jordan Mayor, Alexander Shenkin, and Kelly Anderson, hav e provided a great deal of encouragement and constructive feedback. My family showed abundant support. Tania Romero and Julia Reiskind provided invaluable guidance in the lab Ed Tanner and Emma Sayer were gracious with allowing me to sample soil from their long term leaf litter manipulation plots. Mirna Fernandez and Noelle Beckman taught me the tricks of collecting from the canopy crane. Ben Bolker, Andrew Hein and Paulo Brando were always kind and helped me with statistical hurdles. Jan Jansa and Fabienne Zeugin introduced me to radioactive phosphate tracers. Joe Wright generously provided leaf litter samples and a wealth of information about tropical forest ecology. Osvaldo Calderon sorted leaf litter to the species level. Kaoru Kitajima taught me how to measure leaf toughness and donated equipment time. Jefferson Hall supported sampling from the PRORENA plots in Soberania National Park.
4 I have truly enjoyed my time at UF and this is largely due to the communities that I was able to form through School of Natural Resources and Environment, the Tropical Conservation and Development program, the Working Forests in the Tropics program and the Department of Biology. My dissertation work has been influenced by ecologists that I worked with prior to starti ng my PhD. Rich Kobe and Mike Walters were my coadvisors during my MS at Michigan State University. Corine Vriesendorp instilled in me the value of natural history. Candy Feller, at the Smithsonian Environmental Research Center, served as my mentor while I was an undergraduate intern. Her patience and support was fundamental in my career choice. I am grateful to a number of grants and funding sources, including an NSF Doctoral Dissertation Improvement Grant (DEB 0909734), a Smithsonian Institute Predoctoral Fellowship, a UF Alumni Fellowship, a STRI Short term Fellowship, a Tropical Conservation and Development Summer Field Grant, a Working Forests in the Tropics Summer Research Grant, a UF CLAS Grinter Fellowship, a UF Graduate Student Council Travel grant, a UF Department of Biology Travel Grant, the James Davidson Travel Scholarship, an IFAS Travel Grant, the Organization for Tropical Studies and the Christopher Davidson and Sharon Christoph Scholarship Fund. Permits were supported by ANAM (Autoridad Nacional del Ambiente) and the Panamanian Ministry of Health.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 3 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 ABSTRACT ..................................................................................................................... 9 CHAPTER 1 INTRODUCTION .................................................................................................... 11 2 OPTIMIZING A METHOD FOR WATER EXTRACTION OF LEAF LITTER NUTRIENTS ........................................................................................................... 16 Overview ................................................................................................................. 16 Background and Hypotheses .................................................................................. 17 Methods .................................................................................................................. 19 Experimental Approach .................................................................................... 19 Litter Collection and Preparation ...................................................................... 20 Water Extracts and Chemical Analyses ........................................................... 21 Statistical Analyses .......................................................................................... 23 Results .................................................................................................................... 24 Water Soluble Element as a Function of Time Hypothesis 1 ......................... 24 Effect of Litter to solution Ratio on Water S oluble Element Hypothesis 2 ..... 25 Effect of Litter Drying Temperature on Water Soluble Element Hypothesis 3 ................................................................................................. 25 Microbial Influe nces on Water Soluble C and Phosphate Hypothesis 4 ........ 26 Forms of Water Soluble Phosphorus ................................................................ 26 Discussion .............................................................................................................. 26 Summary of Recommendations for Future Studies ................................................ 31 3 DIFFERENTIAL LEAF LITTER NUTRIENT SOLUBILITY ACROSS 41 LOWLAND TROPICAL FOREST WOODY SPECIES ............................................ 47 Overview ................................................................................................................. 47 Background and Hypotheses .................................................................................. 48 Methods .................................................................................................................. 51 Sample Collection and Preparation .................................................................. 51 Water Extractions and Chemical Analysis ........................................................ 51 Statistical Analysis ............................................................................................ 54 Results .................................................................................................................... 55 Litter Element Solubility .................................................................................... 55 Predictors of Element Solubility ........................................................................ 56
6 Leachate Chemistry ......................................................................................... 58 Discussion .............................................................................................................. 59 4 LEAF LITTER INPUTS DECREASE PHOSPHATE SORPTION IN A STRONGLY WEATHERED TROPICAL SOIL OVER TWO TIME SCALES ........... 73 Overview ................................................................................................................. 73 Background and Hypotheses .................................................................................. 74 Methods .................................................................................................................. 78 Leaf Litter Leachate and Soil Collection ........................................................... 78 Phosphate Sorption Experiments ..................................................................... 80 Analysis ............................................................................................................ 82 Results .................................................................................................................... 84 Soil Carbon Content and Leachate Characteristics .......................................... 84 Effect of Field Litter Manipulation on Phosphate Sorption ................................ 84 Effect of Leachate Pulses on Phosphate Sorption ........................................... 85 Effect of Soil Treatment on Bray Extractable 32P .............................................. 86 Solution to soil Ratio s ...................................................................................... 86 SUVA of Leachate ............................................................................................ 87 Discussion .............................................................................................................. 87 Mechanisms Underlying Phosphate Sorption Results ...................................... 88 Leachate Concentrations .................................................................................. 91 Scaling Phosphate Sorption Results ................................................................ 92 Summary ................................................................................................................ 94 5 CONCLUSION ...................................................................................................... 109 Litter Solubility among Ecosystems and Life History Strategies ........................... 109 Other Roles of Leachate from Recently Senesced Litter in Nutrient Cycling ........ 113 APPENDIX: ALPHABETICAL LIST OF SPECIES USED IN CHAPTER 3. ................. 118 LIST OF REFERENCES ............................................................................................. 119 BIOGRAPHICAL SKETCH .......................................................................................... 132
7 LIST OF TABLES Table page 2 1 Species information and total element concentration and standard errors of freshly senesced leaf litter .................................................................................. 34 2 2 Element solubilit y over time modeled with Michaelis Menten and linear functions. ............................................................................................................ 35 2 3 The effects of litter to solution ratio and litter drying temperature on water soluble carbon (C), phosphateP and hydrogen (H) from litter ........................... 38 2 4 Eff ect of litter to solution ratio and litter drying temperature on water soluble nitrogen (N), calci um (Ca) and magnesium (Mg) ................................................ 41 2 5 The effect of drying temperature on litter element solubility using twosided t tests for five additional species not included in the larger study ......................... 42 2 6 The effect inhibiting microbial activity on water soluble carbon (C) and phosphateP for replicates of litter from the same tree ....................................... 44 3 1 Water solubility of litter elements during a 4 hr 1:50 litt er to solution ratio extract. ................................................................................................................ 66 3 2 First three components of Principle Components Analysis using soluble fraction. ............................................................................................................... 67 4 1 Char acteristics of leachate and simulated throughfall. ....................................... 96 4 2 Characteristics of the Gigante Leaf Litter Manipulation project (GLiMP) soils (0 10 cm), which is referred to as 'field litter manipul ation' treatments. .............. 97 4 3 ANOVA and Tukey multiple comparison res ults for the effect of solution on 32P phosphate sorption to soil. ........................................................................... 98 4 4 Species specific leachate effects on 32P phosphate sorption compared to a throughfall control. .............................................................................................. 99 4 5 Concentration of treatment solutions used in sorption experiments and net DOC s orption to soil. ........................................................................................ 100 4 6 Simple linear regressions evaluating leachate properties and net DOC sorbed as predictors of %32P sorbed to soil. ..................................................... 101
8 LIST OF FIGURES Figure page 2 1 Water soluble C and phosphateP shown as a function of extraction time for freshly s enesced air dried leaf litter. ................................................................. 45 2 2 Water soluble Ca, Mg, and total dissolved N expressed as a function of extraction time for freshly senesced air dried leaf litter ....................................... 46 3 1 Linear relationships between sol uble element on a litter mass basis and total initial litter element concentration.. ..................................................................... 68 3 2 Linear relationship between soluble C fraction litter lignin. ................................. 69 3 3 Stoichiometry of soluble C on a litter mass basis, phosphateP and total N. .... 70 3 4 Regression between inorganic and total N and P in water extracts. ................. 71 3 5 Stoichiometry of initial litter, leached litter and litter leachate. ............................ 72 4 1 The effect of field litter additions on phosphate-3 2P sorption to soil during a 1 h incubation.. .................................................................................................... 102 4 2 Leachate effects on reducing phosphate-32P sorption to soil during a 1 h incubation.. ....................................................................................................... 103 4 3 Phosphate-32P in supernatant for leachate relative to the throughfall control, calculated as %32P in leachate supernatant/ %32P in throughfall supernatant. 104 4 4 Relation ship between leachate effect on phosphate-32P sorp tion and net DOC sorbed to soil .......................................................................................... 105 4 5 The percent of phosphate-32P sorbed that was extractable using Bray 1. l. .... 106 4 6 The effect of solution to soil ratios on phosphate sorption for the three field litter manipulation soils.. ................................................................................... 107 4 7 SUVA wavelength scans of initi al leachates (averaged from three replicates). Standard deviations are shown for SUVA at 254 nm. ....................................... 108 5 1 Conceptual figure for how the two main metrics of litter solubility used in this study may vary for species adapted to different resource availability. ............. 1 17
9 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy LEAF LITTER LEACHING AND NUTRIENT CYCLING IN LOWLAND TROPICAL FORESTS By Laura A. Schreeg August 2011 Chair: Michelle Mack Major: Interdisciplinary Ecology Nutrient availability to plants and microbes exerts key control s on the terrestrial carbon cycle. In many systems, nutrients recycled from leaf litter, through decomposition, are considered vital for maintaining primary productivity. Decomposition is often investigated as being biologically mediated while the role of leaching, or water extraction of nutrients, has received less attention. In addition to releasing nutrients from litter, leaching may also influence nutrient interactions with the soil environment. Here I investigated the solubility of leaf litter nutrien ts and the role of leaf litter leachate in promoting plant phosphorus (P) availability in a strongly weathered soil. This work illustrated that leaf litter nutrient solubility differs am ong elements and in many cases water soluble elements on a litter mass basis were predictable by total litter concentration of the respective element. Furthermore, when a species ranked high in the soluble fraction of one element, it also ranked high in solubility of other elements. Differential nutrient solubility suggests that elements cannot be treated equally in our conceptual and empirical models of decomposition. Element cycles, especially that of P, may be more sensitive to variation in precipitation than previously appreciated.
10 Furthermore, this work demonstrates tha t litter inputs can decrease both the magnitude and strength o f phosphate sorption in a strongly weathered soil through both long term litter inputs and leachate pulses. Leachate pulses had a greater influence on decreasing soil phosphate sorption in the field litter removal treatment compared to the control and litter addition soils, showing that field litter manipulations and leachate pulses interacted in a predictable manner. In addition to decreasing the quantity of phosphate sorbed, litter inputs wer e found to decrease the strength of phosphate sorption. The effect of leachate on reducing the magnitude and strength of phosphate sorption was related to differences in leachate carbon chemistry and initial solution phosphate concentration.
11 CHAPTER 1 IN TRODUCTION L eaf production can represent more than two thirds of total aboveground net annual primary productivity i n tropical forests (Clark et al. 2001). In the absence of large disturbance, leaf litter production (i.e., senesced leaves that drop to the forest floor) is equal to foliar production, resulting in annual litter inputs that can be >10 Mg ha1 (dry mass basis; Vitousek 1984). Once on the forest floor, t ropical forest leaf litter can have residence times <1 yr (Santiago 2007). Leaf biomass in t ropical forests is therefore characterized by high rates of production and rapid decomposition making it a key flux of carbon in tropical forests. In addition, litter serves as a major flux of mineral nutrients. Although retranslocation of mineral nutrient s reclaims foliar nutrients during senescence, there are limits to the extent to which nutrients can be retranslocated (resorption proficiencies; Killingbeck 1996). Senesced leaves therefore return substantial quantities of nutrients to the soil environment. For example, in a lowland tropical forest in Costa Rica, litter returned ~ 100 kg N ha1 yr1, 5 kg P ha1 yr1 and 50 kg Ca ha1 yr1 to the forest floor in addition to substantial quantities of other nutrients (Wood et al. 2006). While there is large variation in decomposition rates among species (Wieder et al. 2009; Salinas et al. 2011) and nutrient returns through litterfall vary among and within sites (Vitousek 1984; Porder et al. 2005), leaf litter is nonetheless a large and dynamic flux of both ca rbon and mineral nutrients in tropical forests. Given the contribution that litterfall makes in returning nutrients to the soil environment controls over rates of mass loss and nutrient release have major implications for the cycling of nutrients in thes e systems. Both litter chemistry and
12 environmental conditions influence rates of decomposition. Decomposition rates often decrease with increases in lignin, lignin:N (Melillo et al. 1982) or C:N (Enriquez et al. 1993). Furthermore, decomposition rates ca n be inversely related to evapotranspiration (Meentenmeyer 1978) and can be pos it ively related to precipitation (Austin and Vitousek 2000; Powers et al. 2009) although this relationship can be the inverse in sites with anaerobic soil conditions (Schuur 2001). In addition, decomposition rates can be positively related to soil temperature (Salinas et al. 2011) The extent to which litter chemistry and environmental conditions predict rates of mass and nutrient loss is often investigated using litterbag studies, which focus on net changes in mass and nutrients over time. From these studies, high initial rates of mass and potassium loss are often attributed to leaching (Melin 1930; Attiwill 1968; Gosz et al. 1973; Berg and Staaf 1980). Yet d espite a long standing appreciation in our conceptual models of decomposition, the role of leaching in decomposition has seldom been quantified across elements and species. Evaluating the extent to which elements differ in solubility is necessary for understanding differenti al availability of nutrients, which can improve our conceptual and empirical models of decomposition. In addition to affecting nutrient rele ase from leaf litter, leaching can also indir ectly influence soil and microbial interactions with nutrient s. In turn these interactions can influence plant nutrient availability. A number of factors, such as microbial uptake and sorption and desorption reactions with soil minerals, govern the path that nutrients take as they move f rom litter to the rhizosphere. T hese f actors can all be potentially influenced by leachate. In highly weathered soils, interactions between leachate, phosphorus and soil chemistry have the potential to be especially important to
13 ecosystem function. Phosphate (the inorganic form of phosphorus t hat plants can use) is quickly sorbed, or removed from soil solution through adsorption or precipitation, in highly weathered soils Much of the sorbed phosphate is not readily plant available because of interactions with metal oxides (Mattingly 1975; Uehara and Gillman 1981; Fox and Searle 1978). Phosphorus can limit forest primary productivity (Vitousek and Farrington 1997) and, because many tropical forests occur on highly weathered soils, controls on phosphate sorption can be key to understanding the pr imary productivity of these systems. T herefore, mechanisms that either decrease phosphate sorption or inc rease the ease of desorption could have fundamental implications for carbon cycling. Litter leachate has the potential to both decrease phosphate sor ption and decrease the strength of sorption by altering soil chemistry (Dalton et al. 1952; Moshi et al. 1974; Ohno and Crannell 1996; Gerke 2010). The most likely mechanisms by which leachate may influence phosphate sorption include competition between c arbon and phosphate for soil sorption sites (Perrott 1978; de Mesquita and Torrent 1993) and the effect of increased solution phosphate on the rate of sorption (van Riemsdijk et al. 1984). T h e effect of leachate inputs on phosphate sorption has traditional ly been investigated in agricultur al systems and have been seldom addressed in forests. While the underlying mechanisms should be similar between manag ed and nonmanaged systems, results from agricultural systems are not directly applicable to forests for two main reasons. First, studies in agricultural systems work with levels of soil phosphorus that represent the high inputs of inorganic fertilizer that these systems receive (Sibanda and Young 1989; Hunt et al. 2007) and do not reflect phosphorus status of forest systems. Secondly, agricultural studies investigate leachate from crop residues and not
14 leaf litter. Given that litter leachate varies greatly among woody species, with distinctions between deciduous and evergreen trees (Hongve et al. 2008), ther e are likely differences between crop and tropical forest litter leachate chemistry. My dissertation has two main objectives. First, I aim to understand how leaf litter solubility and predictors of solubility vary among elements across species. This is i nvestigated with the goal of better understanding the potential role of leaching in litter decomposition. Secondly, I consider if litter inputs can modify soil chemistry to promote phosphorus cycling, thereby acting as not only a source of phosphorus to the soil environment but also indirectly influencing phosphorus movement through the soil. I investigate this question by studying both long term f ield inputs and leachate pulses on phosphate sorption. The work is divided into three chapters. Two chapters focus on litter solubility, while the third considers the effect of litter inputs on phosphate sorption. All field work and most lab work was completed in the Republic of Panama with the support of the Smithsonian Tropical Research Institute. Chapter 2 (th e first research chapter) address es the main factors that are likely to affect comparison studies quantifying litter solubility I address how the quantity of water soluble element on litter mass basis varies during a 24 h water extraction period. I also test if litter to solution ratio and litter drying temperature influence water soluble element. In addition, the effect of microbial activity on water soluble elements is considered. The experiments were done using recently senesced litter collected from a canopy crane in Parque Metropolitano and a reforestation experiment in Soberania National Park (PRORENA, conducted by the Smithsonian Tropical Research Institute
15 and the Yale School of Forestry and Environmental Studies). Carbon, phosphorus, nitrogen, calcium, magnesium and pH were investigated. Chapt er 3 uses l eaf litter from 41 species collected on Ba rro Colorado Island in Panama and investigates how solubility, both the soluble fract ion of total litter element and water soluble element on a li tter mas s basis, vary among elements. Furthermore, this chapter considers if water soluble element can be predicted by litter traits such as total element c oncentrations, lignin C:N and leaf toughness. In addition, this chapter invest igates stoichiometry and nut rient forms (i.e., inorganic and organic) in leachate. In Chapter 4, a radioactive carrier free phosphate tracer (32P) was used to determine how leaf litter inputs influence soil phosphate sorption in lowland tropical forest soil. The use of a phosphate t racer made it possible to investigate small yet potentially important effects of litter inputs on phosphate cycling that would be difficult to detect using traditional colorimetric methods for measuring phosphate. The effects of field litter manipulation on phosphate sorption and recovery were evaluated using soil that had received three different amounts of leaf litter: litter doubled, normal and litter removed (Sayer and Tanner 2010) The influence of leachate pulses on phosphate sorption was evaluated ac ross the three soil treatments using leachate from five species.
16 CHAPTER 2 OPTIMIZING A METHOD FOR WATER EXTRACTION OF LEAF LITTER NUTRIENTS Overview Leaf litter leaching influences plant nutrient availability in terrestrial ecosystems by returning nut rients to the soil and influencing biogeochemical cycles at the litter soil interface. However, laboratory procedures for estimating litter solubility vary greatly, which complicates comparison among studies and ecosystems. To address this, recently senes ced litter from lowla nd tropical trees was used to investigate how element solubility on a litter mass basis responds to six litter to solution ratios ( ranging from 1:300 to 1:20), five extract ion times ( from 1 24 h) and two litter drying temperatures. The study focused on water soluble organic c a rbon (C), phosphorus (total P and phosphate) and hydrogen ions (H), and also investigated nitrogen (N), calcium (Ca), and magnesium (Mg) for a subsample of treatment levels. Water soluble element on a litter mass b asis was relatively unaffected by litter to solution ratio for the shorter extraction times However, for longer extraction times (i.e. 16 and 24 h) the more concentrated litter to solution ratios had higher phosphate solubility for one species. Water soluble phosphate was generally exhausted within 24 h; however, C, Ca, and Mg were not (results for N varied by species). Water soluble phosphateP did not differ from total soluble P for any of the species investigated (only evaluated for the 4 h extractions) Drying litter at 60C compared to air drying (22C) increased solubility for a few species/element combinations. Finally, the absence of a microbial inhibitor decreased concentrations over longer extraction times. In summary, air dried litter is preferre d over ovendried litter and longer duration extractions should use less concentrated litter to solution ratios and inhibit microbial activity.
17 Background and Hypotheses L eaf litter l eaching or water extraction of solu ble components of senesced leaves, ca n provide a key flux of nutrients to microbes, plant roots, the soil profile and the surrounding environment (McDowell and Fisher 1976; Meyer et al. 1998; Wetzel 1992; Guggenberger and Kaiser 2003; Cleveland et al. 2004). The role of leaching in potassium (K) cycling has long been appreciated (e.g., Attiwill 1968; Gosz et al. 1973), and leaching also has the potential to remove significant quantities of phosphorus (P) and nitrogen (N) from litter (Parsons et al. 1990; Qiu et al. 2005). However, studies inve stigating leaching of litter elements (Nykvist 1963; Tukey 1970; Qualls et al. 1991; Cleveland et al. 2004; Wieder et al. 2008) tend to vary in a number of factors, including extraction duration, litter to solution ratio, and litter drying temperature. A better understanding of the influence of methodological choices on litter element solubility would therefore facilitate comparisons across studies, provide recommendations for future work and advance our overall understanding of leaching as a vector in nut rient release from litter. Understanding how litter element solubility changes as a function of time will guide interpretation among studies that have similar extraction durations. If extraction rates are constant over short time periods, such as 1 versus 2 h, studies varying in short extraction duration may be easily compared. Similarly, comparison among longer extraction periods requires knowledge of when element solubility is exhausted. When measured as percent of dry weight, exhaustion of water extract able materials has been reported to occur within 24 h for broadleaf species (Nykvist 1962; Taylor and Parkinson 1988) but knowledge of the variation among individual elements is needed. If extraction times for reaching exhaustion are similar among species for a given element,
18 comparisons may be relatively straightforward by using any extraction time that is greater than the established time to exhaustion. Litter to solution ratio is an additional concern for interpreting laboratory indices of litter leachin g. Litter to solution ratios may simply act as a dilution factor, which would have no effect on element solubility expressed on a litter mass basis. On the other hand, litter to solution ratio could decrease the quantity of element extracted if solubility decreases when a common ion already exists in solution, as predicted by the common ion effect and Le Chatleliers principle. Alternatively, the quantity of element extracted could increase with more concentrated litter to solution ratios if the presence of certain ions (e.g., protons) resulted in increased chemical extraction of nutrients in litter. Any influence of litter drying temperature on element water solubility could be related to the disruption of cell integrity. A large fraction of the water in f reshly senesced leaves is removed during air drying (water contents decreases from ~50% water to ~7% water; Cornelissen 1996; Taylor and Barlcher 1996), while oven drying removes a much smaller percent (i.e., the remaining ~7% based on 60C drying). Cell membrane integrity is disrupted during desiccation of live plants (Vicre et al. 2004) and if similar changes occur in senesced material, drying may increase solubility. If the disruption caused during the process of air drying is large compared to any dis ruption caused by additional oven drying oven, element solubility may be similar between the air and oven dried samples. Finally, the reactivation of microbes on rewetted litter could immobilize nutrients and mineralize carbon (C), leading to underestim ates of water soluble elements. However, a study investigating the bioavailability of leachate C found that < 5% of the
19 total litter leachate C is bioavailable during the first 24 h of the assay (Don and Kalbitz 2005), and this study intentionally added a microbial inoculum. Therefore, microbial activity during a 24 h index of leaching potential (which lacks the addition of an inoculum) may not be significantly influenced by microbial activity. This study investigated how solubility of litter nutrients is affected by four variables: extraction time, litter to solution ratio, litter drying temperature, and microbial activity. The study focuses on water extracted elements expressed on a dry mass basis and defines water soluble as element extracted per gram of dry leaf litter. The hypotheses were: 1) water soluble elements will increase over the first hours of extraction and reach exhaustion by 24 h; 2) more concentrated litter to solution ratios will decrease solubility due to the common ion effect and the effect will be more pronounced at longer extraction times; 3) element solubility will be similar among litter drying temperatures; 4) microbial activity will have little influence on solution concentrations, and therefore little influence on the water soluble element expressed on a litter mass basis. Methods Experimental Approach The investigation of how extraction time, litter to solution ratio, litter drying temperature and microbial activity influenced water soluble element was divided into three experim ents. Experiment 1 investigated how water soluble organic C, phosphate and H varied over six litter to solution ratios (1:300, 1:100, 1:50, 1:30, 1:25 and 1:20), five extraction times (1, 4, 8, 16, and 24 h) and two litter drying temperatures (air dried at 22C or ovendried at 60C) for Anacardium excelsum Castilla elastic a and Luehea seemannii (species information is provided in Table 21). In addition, extracted Ca, Mg,
20 N and P (phosphateP and total P) were determined for all extraction times of the 1: 50 litter to solution ratio and all ratios of the 4 h extraction time. To expand the number of species investigated, the effect of drying temperature on water soluble organic C, Ca, Mg, N, phosphateP and total P were also evaluated for Cecropia longipes Dipteryx panamensis Inga punctata Ochroma pyramidale, Terminalia amazonia using a 1:50 extraction ratio and 4 h extraction time (Experiment 2). Finally, data investigating the effect of microbe activity during water extractions (Experiment 3) is present ed. Microbial activity was not controlled in Experiments 1 and 2, but data from those experiments led us to believe it was worthy of investigation. Litter Collection and P reparation Senesced leaves of A. excelsum C. elastic a and L. seemannii were collecte d from three individuals of each species and used in Experiments 1 and 3. Leaves were collected from the Metropolitan National Park ( Parque Metropolitano) via a canopy crane (Parker et al. 1992). Parque Metropolitano is a seasonal forest with 1740 mm mean annual precipitation (Kitajima et al. 1997) in Panama City, Panama. Senesced leaves were identified by sight and were collected if they fell when lightly tapped by hand, indicating that the abs cission zone was well formed and translocation of mineral nutr ients and C had ceased. For Experiment 2, which investigated drying temperature, C. longipes was also collected via canopy crane from Parque Metropolitano, while D. panamensis I. punctata, O. pyramidale, T. amazonia were collected from a reforestation exp eriment in Soberania National Park (PRORENA, conducted by the Smithsonian Tropical Research Institute and the Yale School of Forestry and Environmental Studies; BastienHenri et al. 2010) Recently senesced leaves of D. panamensis and T. amazonia were coll ected from low hanging b ranches or by usi ng a
21 stick to tap branches I. punctata and O. pyramidale and some D. panamensis sampl es were collected on tarps and in baskets that were checked weekly. All litter samples were collected during the 2009 dry season between January and the beginning of May Petioles were removed and litter was air dried under ambient laboratory temperature and humidity conditions ( 22 0.5C and 55 5%, respectively ) in thin layers for two weeks. Litter was then stored in plastic zi p bags. Litter was cut into ~ 25 cm2 squares, with the exception of T amazonia leaves, which were small enough to use whole. Litter was then thoroughly mixed, either by individual tree ( A excelsum C elastic a and L seemannii in Experiment 1 and 3) or by species (Experiment 2). Litter for the ovendried treatment was pulled from the bulked samples and dried at 60C for 48 h, cooled in a desiccator and then stored in plastic zip bags. Gloves were worn while handling collected litter in the laboratory. W ater Extracts and Chemical A nalyses All extractions used 60 ml square polyethylene widemouth bottles. Litter was weighed on an ovendried (60C) equivalent basis and gently placed in bottles before being covered with 60 ml of deionized water. In Experim ent 3, which investigated the effect of microbial activity during extractions, a treatment of 1 mM sodium azide (NaN3) was included. Bottles were placed sideways on a shaker table and set on low shaking speed (180 oscillations per minute). At the end of the extraction time, solutions were poured off, centrifuged (8000 g for 10 min) and the supernatant was immediately decanted. M olybdate reactive phosphorus (MRP) was determined at 880 nm using a Hach DR 5000 UV vis spectrophotometer Molybdate reactive phosphorus is referred to as phosphate throughout the rest of the study; although, it should be noted that in addition to phosphate, MRP can include small quantities of organic and condensed
22 phosphorus compounds hydrolyzed during the assay (Worsfold et al. 2005). Water extractable organic C was determined as dissolved organic carbon (D OC ) remaining in solution after centrifuging, which was measured using combustion and infrared detection using a S himadzu TOC VCSH Total Organic Carbon analyzer. Dissolved organic carbon is referred to as water soluble C when reported on a litter mass basis. Solution pH was determined with a Hach Sension 3 pH meter and electrode. Cation and total P concentrations were determined by inductively coupled plasma optical emission spec trometry (ICP OES) using an Optima 2100 ( Perkin Elmer Waltham, MS ). Total dissolved nitrogen (TDN) was determined by a sodium hydroxide/potassium persulfate oxidation in sealed glass tubes with colorimetric nitrate detection on a Latchet Quickchem 8500 au t oanalyzer (Latchett, Loveland, CO). Total dissolved N is referred to as water soluble N when reported on a litter mass basis. For bulk litter, total C and N were determined on ground samples by combustion and gas chromotography ( Flash EA 1112, Thermo Bre men, Germany). C ations and total P of ground litter were determined by nitric acid digest ion under pressure at 180C in sealed Teflon vessels with detection by ICP OES All pH and phosphate values were determined within 48 h of sampling with storage at 4 C. All DOC solutions were acidified, stored at 4C and analyzed < 48 h after extraction, or samples were frozen immediately after extraction. Cations were determined in solutions that were frozen after extraction. Total dissolved N was determined with di gestion on refrigerated samples with digestion within 48 h of extraction.
23 Statistical A nalyses The effect of litter to solution ratio and drying temperature on water soluble element were investigated within each extraction time using twoway ANOVA. Significant factors were further investigated using Tukey multiple comparison tests. Relationships between time and water soluble element were evaluated by fitting both linear and Michaelis Menten functions. The Michealis Menten function (y = Vm*x/(k+x), where Vm is the expected maximum water soluble element and k is the time when half of the expected maximum water soluble element is reached) was fitted using the nonlinear least squares Michaelis Menten function in R (R Development Core Team) Model fit was evaluated by Akaike's information criterion (AIC). To determine if water soluble element was exhausted within 24 h, Vm from significant Michealis Menten models was compared to the water soluble element from the 24 h extraction using onetailed t tests. For Experiment 2, t tests were used to evaluate the effect of leaf litter drying temperature on water soluble element for five additional species not evaluated in Experiment 1. Tests were conducted within each species. Contributions of phosphateP versus organic P were evaluated for water extracts from Experiment 1 and 2. For these eight species, t tests were used to investigate if phosphateP (i.e., molybdate reactive phosphorus) and total P (i.e., ICP values) differed. T tests were also used to test the effect o f microbial activity on water soluble element within species during 4 and 24 h extract durations (Experiment 3). Tests were done within species because it was known a priori that species vary in litter element solubility, and the interest here was in treat ment effects and not overall differences in solubility among species. Bonferroni corrections were not included because the goal was to conservatively evaluate the retention, not rejection, of the null hypothesis. Therefore, use Bonferroni corrections,
24 whic h control Type I error and the probability of reporting false positives, would be nonconservative. Results Water Soluble Element as a Function of Time Hypothesis 1 For the three species evaluated, 24 h was not sufficient for reaching exhaustion of water soluble C, Ca or Mg but it was sufficient for phosphate (Table 22, Figure 21). Nitrogen results varied by species, with water soluble N reaching an asymptote in <24 h for C. elastica but not for the other two species (Table 22). For L. seemannii not o nly was C not exhausted in 24 h, but for a number of the ratios investigated, C solubility was better modeled using linear compared to Michaelis Menten functions (Table 22). This suggests the rate of extraction did not decline during the 24 h investigati on. Linear models were also better fits than Michaelis Menten functions for L. seemannii and water soluble N, Ca and Mg (Table 22, Figure 22). However, N, Ca and Mg were only investigated at a 1:50 litter to solution ratio and data from C and P show that model fit was affected by litter to solution ratios. For example, at more concentrated litter to solution ratios (1:20, 1:25, 1:30), C and phosphate extracted from A. excelsum and C. elastica were found to be better fit by Michaelis Menten than linear functions ( Table 22 ). At less concentrated ratios (1:100 and 1:300), Michaelis Menten and linear fits were found to be similar in many cases (as evaluated by AIC values) and parameters were less likely to be significant ( Table 22 ). Because the same volume of deionized water was used for all samples, less concentrated litter to solution ratios had less litter, which may underlie increased variation among samples and less ability to fit models.
25 Effect of Litter to solution Ratio on Water Soluble E lement H ypothesis 2 Results suggest that increasing litter to solution ratio has no systematic effect on water soluble C, Ca, Mg or N and, although there were a few significant responses, they do not support a generalization about litter to solution ratio (see bel ow). Ratio therefore acts as a solution dilution factor and does not influence water solubility of these nutrients when evaluated on a litter mass basis. This is also true for phosphate for extraction times <16 h; however, at longer extraction times more concentrated ratios had elevated water soluble P for C. elastica (Table 23). Similarly, more concentrated ratios resulted in greater solubility of protons for most extraction times (Table 23). The significant litter to solution ratio results that do not support general trends but deserve mention include that of A. excelsum for which two of the more concentrated litter to solution ratios had significantly greater water soluble C than the 1:100 ratio at 16 h (for both air and ovendried litter), but ovendried litter showed an inconsistent result at 24 h (as evaluated through Tukey multiple comparisons; Table 23). A similar result was found for A. excelsum ovendried litter and N, with the 1:300 ratio having higher solubility than other ratios. The signif icant ratio effect during the 4 h extractions for water soluble Ca and Mg of C. elastica do not represent a trend among the continuum of six litter to solution concentrations and appear to be due to low values for the 1:50 litter to solution treatment (Table 2 4). Effect of Litter Drying Temperature on Water Soluble E lement Hypothesis 3 In a few cases, ovendried litter had significantly higher element solubility than air dried litter (Table 23, 2 4, 2 5). Significantly higher C solubility in oven dried litter was found for L. seemannii at three extraction times (Table 23). For A. excelsum ovendried litter had significantly greater DOC extracted per gram of litter during the 16 h
26 extraction. In contrast, water soluble C did not differ with drying tem perature for C. elastica regardless of extraction time (Table 23). Drying temperature of litter had some influence on water soluble phosphate. Two extraction times for L. seemannii (Table 23) and the one extraction time (4 h) investigated for D panamens is (Table 25) showed greater water soluble phosphate in oven compared to air dried litter. Ovendrying litter was found to significantly increase water soluble N for L. seemannii (Table 24). Microbial Influences on Water Soluble C and Phosphate Hypothe sis 4 Additions of 1 mM sodium azide were intended to inhibit microbial activity with minimal cell lysis beyond that induced by rewetting. In some cases, the presence of sodium azide led to increased water soluble C, especially during the 24 h extraction time (Table 26). In addition, in one case ( C. elastica 24 h) soluble phosphate was greater with the sodium azide addition. Tests using samples from within an individual tree were more likely to show significance than the tests using samples from three in dividual trees (Table 26). Forms of W ater Soluble Phosphorus Water soluble phosphateP did not differ from soluble total P for air dried litter of any of the eight species investigated (using 1:50 litter to solution ratios and 4 h extraction and t tests for each species with p = 0.05 as significance; data not shown). This demonstrates that water extractable P from fresh litter is inorganic phosphateP, with little organic P present. Discussion Litter to solution ratio did not, in general, affect water soluble C or P during shorter extraction times (with the exception of H ions). This suggests that litter to solution ratio largely acts as solution dilution factor, which is accounted for when extracted elements
27 are expressed on a litter mass basis. Therefo re, data across studies using different litter to solution ratios can be easily scaled when similar short extraction times are used. Comparing among studies that use different extraction times is more difficult because time response curves of water solubl e elements vary among species and initial extraction rates can decrease quickly (Figure 21; Table 22). In addition, exhaustion of extractable element in many cases requires >24 hours (Table 22). Oven dried litter (60C) yielded higher concentrations of water soluble C, phosphate and H ions in a few cases. This suggests the additional drying that occurs between air and 60C, and the removal of an additional 8% of mass through water loss (on average for this study) may have an influence beyond the effect that occurs during initial desiccation from freshly fallen to air dried. Alternatively, oven drying litter could induce changes in microbial communities and enzyme activities. Because the effect of oven drying is predictable and not often statistically sig nificant, the use of ovendried litter seems acceptable. The effect of inhibiting microbial activity suggests that microbial respiration can decrease DOC that was extracted during the course of the experiment. This was especially true for the 24 h extract ion compared to the 4 h extraction. A study on the bioavailability of leachate C from recently senesced litter, which intentionally added a microbial inoculum, found rates of mineralization < 5% of the total DOC during the first 24 h (Don and Kalbitz 2005) In contrast, any microbial activity in the current study would be due to autochthonous organisms because an inoculum was not included. Initially a microbial inhibiter was not added based on an assumption that the extraction times would not be long enough to cause significant oxidation of DOC. However, results
28 from the sodium azide additions suggest that autochthonous microbial communities on litter can, in some cases (Table 26), quickly mineralize organic C upon rewetting. This study could have been impr oved by using sodium azide in all samples, especially the longer extractions. Although microbial activity can decrease the concentrations of elements in solution, in turn leading to lower calculated values of water soluble element on a litter mass basis, it seems unlikely that microbial activity influences the release of elements from litter. The highest rates of extraction were during the first hours (Figure 21 and 22), suggesting the release of elements from litter during the time periods investigated are due to chemical and physical extraction by water. The possibility of residual enzymes on recently senesced litter, however, deserves mention. Most work on litter enzymes has been done with decomposing litter. Enzyme activities of rewetted litter may be due to both reactivated microbes and desorption of residual enzymes (AlarcnGutirrez et al. 2010). It is difficult to extrapolate the values reported from studies using decomposing litter to this work using leachates of recently senesced litter becaus e 1) decomposing litter is likely to have higher fungal colonization and, therefore, higher enzyme activities (Fioretto et al. 2001) and 2) the indices of enzyme activities in these studies use a finely milled litter (<0.5 mm) which can release enzymes that may not be active during the shorter extraction times used in this study. The role of enzymes in the water extraction of freshly senesced litter remains unknown and testing enzyme activities in leachates would be a worthwhile endeavor for future studies, specifically given the dominance of phosphateP in the extracts.
29 The definition of DOC, which is then used for calculating C extracted on a per mass litter basis, is another consideration when synthesizing results from multiple studies. Dissolved organic C is operationally defined as organic C remaining in supernatant after centrifuging. In the literature, protocols for determining DOC vary greatly and include centrifuging, filtering through glass wool, and filtering through a range of pore sizes (0.2 1 2000; Frberg et al. 2007; Wieder et al. 2008). In practice, there is no commonly used consensus for filter pore size (although suggestions exist; Thurman 1985), yet this can make an important differ ence in DOC concentrations. By not filtering, these data may not be directly comparable to studies that report DOC after filtering solutions through a small pore size. Dissolved organic C values may also not be directly comparable if different filter sizes were used. The use of centrifuged, nonfiltered values due to cost and time considerations is recommended. Collection methods in this study isolated leaves that had reached senescence but had not yet experienced leaching on the ground or in a trap. By working with litter that was standardized to time zero of the litter stage, this study clearly shows the first hours of extraction represent an important time period for the flux of soluble nutrients, especially C and P, from recently senesced leaves (Figure 2 1 and 2 2). Studies that use litter of an unknown age, even if presumed to be fairly fresh, may miss the initial (and likely substantial) flux of C and mineral nutrients from leaching. The distinction between leaching of fresh litter and decomposing l itter may be important for building a mechanistic understanding of litter leaching in ecosystem processes. For example, the stage of litter decomposition can influence expectations for
30 quantities released and the forms of extracted elements. This study, which used fresh litter, shows that 13% of the total C is initially soluble (calculated using the asymptote of water soluble C and total litter C for A. excelsum and C. elastica ; data in Table 21 and 2 2). The proportion of total litter C that is water extr actable would likely increase during decomposition due to microbial activity (Magill and Aber 2000; Cleveland et al. 2006). However, the stage of decomposition can influence leachate chemistry. For example, bioavailability of water extractable C varies wi th litter age (Yavitt and Fahey 1986; Hongve et al. 2000; Don and Kalbitz 2005) and the extent of decomposition may also influence chemistry of other elements. Extracts with significant concentrations of organic P have been reported (Qualls et al. 1991) w hile this study did not detect organic P. When soils are re wetted, increases in soluble organic phosphorus can be positively correlated with microbial P due to release of P through cell lysis (Turner and Haygarth 2001) and it is possible that the same rel ationship holds for P forms extracted from leaf litter. Therefore, if microbial colonization of litter increases during decomposition (Fioretto et al. 2000; Duarte et al. 2010), extractable organic P concentrations may increase. This is likely to have env ironmental significance given that many organic P compounds are more mobile in soil than inorganic phosphate (Frossard et al. 1989) This study found that 24 h did not exhaust water extractable C for the three species investigated (Table 22). For A. excelsum or C. elastica the rate of solubility decreased but in the case of L. seemannii water soluble C was often equally or better modeled by linear functions compared to to Michaelis Menten functions (Table 22), suggesting the rate of solubility did not dec rease during the 24 h investigated. The solubility of elements may be related to how quickly litter reaches maximum water
31 content (Ibrahima et al. 1995). Leaf traits such as lignin content, thicker cutin layers and general leaf toughness (as hypothesized by Gallardo and Merino 1993) may be good predictors of species that will require longer extraction times to develop full response curves of water soluble element. While lignin content from the samples analyzed here was not determined, data for A. excelsum and C. elastica from an earlier collection (same site) and data for L. seemannii from a different Panama site (Barro Colorado Island) are 12%, 19% and 31% (L. Schreeg, unpublished data using freshly senesced litter compiled from multiple individuals analyz ed through Ankom fiber digest), which suggest the high lignin content of L. seemannii could be related to C solubility. But, this is only speculation because lignin content is known to change with site conditions (Vitousek 1998; Mack and DAntonio 2003). Although great efforts were made to standardize for time since senescence in this study, foliar endophytes, insect herbivory or parasitic infections were not controlled for. The effect of such interactions on living and senescing leaves could add natural variation to water extractable compounds and extraction rates. Foliar endophytes are ubiquitous (see citations in Arnold et al. 2003) and abundances have been shown to increase in density as leaves age and approach senescence (Bernstein and Carroll 1977; R odrigues 1994), which could be related to enzyme activities of recently senesced litter. Herbivory could influence extractable compounds by mechanically altering litter. Parasitic infections can also damage cell walls and alter C and mineral nutrient content in the leaf (English et al. 1972; El Hajj et al. 2004). Summary of Recommendations for Future S tudies Based on the results of this study using freshly senesced litter the following recommendations can be made to facilitate comparisons of future work. T hese
32 recommendations should be further tested for other species that span a greater range of litter traits, in addition to litter at varying stages of decomposition. (1) Water soluble C and P did not change as function of litter to solution ratios at short er extraction times for the three species investigated in this study. Water soluble C and P from short extraction times can therefore be compared across litter to solution ratios. (2) More concentrated litter to solution ratios may yield greater water sol uble P at longer extraction times (16 and 24 h); therefore, lower concentration litter to solution ratios (1:300) should be used when longer extracts are of interest. (3) Drying litter at 60C, compared to air drying, led to increased water soluble element in a few instances. Studies should use air dried litter when possible, but the use of ovendried litter is acceptable when clearly stated because the direction of effect was consistent and often nonsignificant in this study. (4) Microbial activity can s ignificantly decrease nutrient concentrations (especially DOC) during the course of an extraction. Sodium azide should be used to inhibit microbial activity, especially at longer extraction times, and appropriate sample storage should be emphasized and cl early stated in the methods section. (5) Because r ate of extraction was generally greatest in the first hours, using 4 h extraction time is reasonable for determining an index of initial solubility I n a number of cases, however, exhaustion of extractabl e nutrients was not achieved in 24 h. For maximum potential solubility, longer extraction times are needed, in which case air dried litter and a microbial inhibitor should be used. As a recommended protocol for determining exhaustion of soluble litter elem ents I suggest the following: (1) Use freshly senesced leaves that are air dried (~22C); (2) Remove petioles and cut large leaves if necessary; (3) Shake for the desired time in a
33 1:300 litter to solution ratio with deionized water, including 1 mM sodium azide to minimize microbial activity. Use at least 2 g of litter (60C dry weight basis) and widemouth bottles. Bottles should be nearly full with the solution present; (4) At the end of the extraction, pour off solution and centrifuge at 8000 g for 10 minutes, or filter through element concentrations or stabilize solutions for later analysis.
34 Table 21. Species information and total element concentration and standard errors of freshly senesced leaf litter Total element (mg g 1 dry mass leaf litter) Species Distribution in Panama Ca Mg P N C Anacardium excelsum Common in moist forests, reported from dry and wet forests 19.3 1.3 5.5 0.8 0.5 0.0 6.5 0.3 410. 9 2.0 Castilla elastica Known in moist forests and premontane forests 23.6 1.4 3.2 0.3 0.5 0.0 10.7 0.3 381.3 2.7 Luehea seemannii Characteristic of tropical moist forests 27.2 1.2 4.2 0.4 0.6 0.1 10.0 0.3 441.3 1.2 Cecropia longipes Tropical and premontane moist forests 31.8 5.5 0.8 10.6 380.4 Dipteryx panamensis Tropical and premontane moist forests 10.3 2.2 0.6 14.0 471.8 Inga punctata Tropical moist and tropical and pre montane wet forests 16.6 1.8 0.5 16.3 440.6 Ochro ma pyramidale Often in disturbed areas, ranges across tropical dry to tropical and premontane moist and wet forests 17.3 3.6 1.2 10.9 471.4 Terminalia amazonia Characteristic of tropical moist forests, also found in tropical wet and premontane mois t and wet forest 26.0 2.7 1.7 7.9 432.7 Species information was taken from Croat (1978). Means of litter from three individual trees are shown for the three species used in Experiment 1. Only one analysis was completed from the bulked litter of the other species. Standard errors are shown in parentheses.
35 Table 22. Element solubility over time modeled with Michaelis Menten and linear functions Water soluble element (mg g-1 dry mass) Parameter 1 (V m or Slope) Vm reached at 24 h? Parameter 2 (k or Intercept) Species Ratio (linear MM) Best function Estimate SE t stat P value Estimate SE t stat P value C A. excelsum 1:20 18.4 MM 56.6 2.0 28.2 < 0.001 1.4 0.3 5.4 < 0.001 1:25 17.8 MM 51.5 1.6 33.2 < 0.001 0.9 0.2 5.4 < 0.001 1:30 19.3 MM 50.8 1.4 35.2 < 0.001 0.7 0.1 5.4 < 0.001 1:50 6.1 MM 48.6 2.6 18.7 < 0.001 0.8 0.3 2.9 0.013 1:100 4.0 MM 41.3 3.3 12.4 < 0.001 0.6 0.3 1.6 0.125 1:300 5.4 MM 46.7 4.0 11.6 < 0.001 0.7 0.4 1.7 0.115 C. elastica 1:20 11.5 MM 45.1 1.9 24.4 < 0.001 1.9 0.4 5.3 < 0.001 1:25 19.2 MM 46.0 2.1 21.9 < 0.001 1.7 0.4 4.6 0.001 1:30 14.5 MM 41.8 2.4 17.2 < 0.001 1.6 0.5 3.5 0.004 1:50 2.7 MM 38.7 3.5 11.0 < 0.001 1.8 0.8 2.3 0.036 1:100 0.8 MM 43.3 6.1 7.0 < 0.00 1 2.3 1.4 1.7 0.123 1:300 4.6 MM 35.8 4.1 8.7 < 0.001 1.0 0.7 1.5 0.154 L. seemannii 1:20 12.3 linear 1.1 0.1 18.7 < 0.001 6.5 0.8 8.4 < 0.001 1:25 13.9 MM 30.5 1.7 17.8 < 0.001 6.7 1.0 6.4 < 0.001 1:30 2.8 MM 37.5 4.6 8.2 < 0.001 10. 3 2.9 3.5 0.004 1:50 11.4 linear 0.9 0.0 18.7 < 0.001 4.0 0.7 6.0 < 0.001 1:100 6.0 linear 0.8 0.1 8.8 < 0.001 6.6 1.2 5.6 < 0.001 1:300 3.4 MM 30.5 6.4 4.74 < 0.001 7.03 4.05 1.74 0.106 Phosphate P A. excelsum 1:20 10.6 MM 0.21 0.01 18.9 < 0.001 yes 1.14 0.34 3.4 0.005 1:25 5.4 MM 0.21 0.01 20.7 < 0.001 yes 1.00 0.28 3.6 0.004 1:30 5.9 MM 0.18 0.01 22.3 < 0.001 yes 0.73 0.22 3.4 0.005 1:50 2.4 MM 0.16 0.01 13.4 < 0.001 yes 0.47 0.29 1.6 0.126 1:100 0.1 MM 0.14 0.02 7.7 < 0.001 0.11 0.33 0.3 0.747 1:300 1.3 MM 0.16 0.03 6.1 < 0.001 0.51 0.65 0.8 0.452
36 Table 22. Continued. Water soluble element (mg g-1 dry mass) Parameter 1 (V m or Slope) Vm reached at 24 h? Parameter 2 (k or Intercept) Species Ratio (linear MM) Best function Estimate SE t stat P value Estimate SE t stat P value Phosphate P C. elastica 1:20 6.3 MM 0.18 0.01 15.9 < 0.001 1.29 0.43 3.0 0.010 1:25 10.5 MM 0.19 0.01 13.6 < 0.001 yes 1.08 0.45 2.4 0.033 1:30 11.2 MM 0.17 0.01 17.7 < 0.001 yes 0.88 0.31 2.9 0.013 1:50 4.2 MM 0.17 0.02 10.2 < 0.001 yes 0.90 0.53 1.7 0.116 1:100 1.2 MM 0.13 0.02 5.8 < 0.001 yes 0.49 0.69 0.7 0.485 1:300 0.2 MM 0.11 0.02 5.4 < 0.001 yes 0.29 0.60 0.5 0.630 L. seemannii 1:20 2.7 MM 0.23 0.04 6.5 < 0.001 yes 5.07 2.43 2.1 0.058 1:25 2.6 MM 0.19 0.03 5.8 < 0.001 yes 3.22 2.08 1.5 0.146 1:30 2.6 MM 0.22 0.04 5.3 < 0.001 yes 4.69 2.87 1.6 0.127 1:50 2.1 MM 0.28 0.12 2.4 0.031 yes 12.17 10.9 6 1.1 0.287 1:100 0.7 MM 0.20 0.0 9 2.4 0.034 5.63 7.10 0.8 0.442 1:300 4.9 MM 0.13 0.03 5.0 < 0.001 yes 2.12 1.88 1.1 0.279 N A. excelsum 1:50 4.1 MM 0.64 0.04 17.5 < 0.001 0.32 0.19 1.7 0.113 C. elastica 1:50 7.5 MM 0.74 0.05 16.4 < 0.001 yes 0.59 0.26 2.2 0.043 L. seemannii 1:50 6.4 linear 0.02 0.00 5.3 < 0.001 0.25 0.05 4.7 < 0.001 Ca A. excelsum 1:50 0.1 MM 0.34 0.08 4.3 < 0.001 0.48 0.91 0.5 0.610 C. elastica 1:50 1.1 MM 5.99 1.02 5.9 < 0.001 4.62 2.54 1.8 0.092 L. seemannii 1:50 9.5 linear 0.03 0.00 9.0 < 0.0 01 0.15 0.04 3.8 0.002
37 Table 22. Continued Water soluble element (mg g-1 dry mass) Parameter 1 (V m or Slope) Vm reached at 24 h? Parameter 2 (k or Intercept) Species Ratio (linear MM) Best function Estimate SE t stat P value Esti mate SE t stat P value Mg A. excelsum 1:50 0.6 MM 3.07 0.37 8.2 < 0.001 0.43 0.45 1.0 0.358 C. elastica 1:50 3.1 MM 2.03 0.32 6.4 < 0.001 5.33 2.56 2.1 0.058 L. seemannii 1:50 0.8 linear 0.06 0.01 7.5 < 0.001 0.13 0.12 1.1 0.292 Results are for air dried litter. Model selection first considered AIC values and then parameter significance. Mic haelis Menten functions were chosen if the asymptote (Vm) parameter was significant and the linear model did not improve fit by more than two AIC units. For Michaelis Menten (MM) models, parameter 1 is the asymptote (Vm) and parameter 2 is the time when half of the asymptote value is reached (k). For linear models, parameter 1 is the slope and parameter 2 is the intercept. Overall model significance was evaluated by considering the pvalue of parameter 1. For MM models, a onetailed t test was used to evaluate if Vm was significantly greater than the water soluble element measured at 24 h (i.e., was the asymptote reached at 24 h?)
38 Table 23. The effects of litter to solution ratio (1:300, 1:100, 1:50, 1:30, 1:25, 1:20 in g:g) and litter drying temperature (22 and 60C dried) on water soluble carbon (C), phosphateP and hydrogen (H) from litter Water Soluble Element (mg g-1 dry mass) C PhosphateP H Species h Factor F P value Tukey results F P value Tukey results F P value Tukey results A. excelsum 1 Ratio 0.13 0.985 0.23 0.946 7.88 <0.001 1:20, 1:25, 1:30, 1:50 > 1:300 Temp 0.67 0.420 0.40 0.534 0.01 0.937 R:T 0.34 0.886 0.35 0.875 0.16 0.974 4 Ratio 1.15 0.361 0.89 0.503 32.79 <0.001 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300 Temp 0.47 0.499 1.52 0.230 1.71 0.204 R:T 0.62 0.686 0.28 0.922 1.95 0.123 8 Ratio 0.60 0.703 1.77 0.157 9.71 <0.001 1: 20, 1:25, 1:30, 1:50, 1:100 > 1:300 Temp 0.00 0.987 2.70 0.114 0.37 0.551 R:T 1.07 0.404 1.24 0.322 0.05 0.999 16 Ratio 3.46 0.017 1:25, 1:50 > 1:100 1.81 0.150 6.13 0.001 1:20, 1:25, 1:30 > 1:300 Temp 13.59 0.001 oven > air 0.31 0 .584 0.08 0.778 R:T 1.85 0.141 0.21 0.957 0.79 0.566 24 Ratio 2.20 0.088 2.50 0.059 5.30 0.002 1:20, 1:25, 1:30 > 1:300 Temp 16.24 <0.001 0.04 0.844 0.56 0.461 R:T 4.38 0.006 1:300 > 1:30, 1:50 for oven oven > air for 1:100, 1:3 00 0.09 0.994 0.34 0.882 C. elastica 1 Ratio 0.17 0.972 0.63 0.677 12.78 <0.001 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300 Temp 0.07 0.801 0.01 0.904 0.00 0.949 R:T 0.32 0.897 0.35 0.878 0.35 0.878
39 Table 23. Continued Water so luble element (mg g-1 dry mass) C PhosphateP H Species h Factor F P value Tukey results F P value Tukey results F P value Tukey results C. elastica 4 Ratio 0.16 0.975 0.18 0.968 12.21 <0.001 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300 Temp 0.27 0.610 0.36 0.552 1.20 0.284 R:T 0.21 0.953 0.97 0.459 0.44 0.814 8 Ratio 0.46 0.800 0.23 0.946 3.87 0.011 1:20, 1:25, 1:30, 1:50 > 1:300 Temp 0.01 0.933 0.34 0.567 0.00 0.949 R:T 1.19 0.346 0.38 0.859 0.04 0.999 1 6 Ratio 0.33 0.888 6.95 <0.001 1:20, 1:25, 1:30, 1:50 > 1:300; 1:50 > 1:100 1.38 0.267 Temp 2.70 0.114 0.34 0.565 0.54 0.470 R:T 0.71 0.625 0.62 0.688 0.18 0.968 24 Ratio 0.60 0.697 16.04 <0.001 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300; 1:20, 1:25, 1:30 > 1:100 1.95 0.122 Temp 1.03 0.320 0.97 0.334 0.44 0.513 R:T 0.71 0.619 0.20 0.959 0.43 0.823 L. seemannii 1 Ratio 1.68 0.177 1.17 0.352 21.78 <0.001 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300 Temp 0.01 0.935 0.00 0.961 1.20 0.284 R:T 0.56 0.728 0.45 0.808 0.30 0.908 4 Ratio 2.16 0.092 0.27 0.925 36.69 <0.001 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300; 1:20 > 1:100
40 Table 23 Continued Water soluble element (mg g-1 dry mass) C PhosphateP H Species h Factor F P value Tukey results F P value Tukey results F P value Tukey results L. seemannii 4 Temp 16.71 <0.001 oven > air 7.76 0.010 oven > air 6.04 0.022 oven > air R:T 1.64 0.188 0.36 0.873 0.48 0.789 8 Ratio 1.28 0.303 0.38 0.859 54.84 <0.001 Temp 10.52 0.003 oven > air 19.63 <0.001 oven > air 5.34 0.030 R:T 0.90 0.496 1.04 0.419 3.46 0.017 1:20, 1:25, 1:30, 1:50 > 1:300 for oven; 1:20, 1:25, 1:30 > 1:100 for oven; 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300 f or air oven > air for 1:100 16 Ratio 1.32 0.288 0.21 0.953 10.44 <0.001 1:20, 1:25, 1:30, 1:50, 1:100 > 1:300 Temp 17.82 <0.001 oven > air 2.08 0.162 0.39 0.541 R:T 0.80 0.559 0.10 0.992 0.05 0.998 24 Ratio 1.88 0.135 0.66 0.661 3.16 0.025 Temp 4.07 0.055 3.07 0.093 9.09 0.006 R:T 0.78 0.573 0.57 0.724 4.04 0.008 1:25, 1:30 > 1:100 for air; air > oven for 1:100 Analyses of variance were conducted within each extract time (1, 4, 8, 16 and 24 h) for each of three response v ariables. Ratio and ratio and temperature interaction (R:T) have five degrees of freedom while temp erature (Temp) has one. Error degrees of freedom are 24 in all cases. Litter mass refers to oven dried weight equivalent. Main effects were only expl ored when the interaction was not significant.
41 Table 24 . Effect of litter to solution ratio (1:300, 1:100, 1:50, 1:30, 1:25, 1:20) and litter drying temperature (22 and 60C) on water soluble nitrogen (N), calcium (Ca) and magnesium (Mg), expressed as mg element g1 dry mass litter, after a 4 h extraction N Ca Mg Species F P value Tukey results F P value Tukey results F P value Tukey results A. excelsum Ratio 5.58 0.002 1.44 0.247 0.83 0.542 Temp 0.66 0.425 0.00 0.964 0.00 0.956 R:T 2.79 0.041 1:300 > 1:20, 1:25, 1:30, 1:50, 1:100 for oven 0.51 0.769 0.11 0.989 C. elastica Ratio 0.34 0.886 4.88 0.003 1:25, 1:30, 1:100> 1:50 7.46 0.000 1:20, 1:25, 1:30, 1:100> 1:50 Temp 0.13 0.725 1.48 0.235 1.98 0.173 R:T 2.00 0.116 1.06 0.408 1.54 0.215 L. seemannii Tatio 1.32 0.288 2.74 0.043 1:20 > 1:300 1.79 0.154 Temp 8.12 0.009 oven > air 4.07 0.055 3.88 0.061 R:T 1.22 0.328 2.01 0.114 0.99 0.446 Ratio and Ratio:Temp (R:T) have 5, 24 degrees of freedom and t emp has 1, 24 degree of freedom
42 Table 25. The effect of drying temperature on litter element solubility using twosided t tests for five additional species not included in the larger study Litter drying temperature 60 C 22C Wate r soluble element or pH t stat P value mean SE mean SE C (mg g-1 dry mass) Cecropia longipes 0.38 0.73 22.01 2.72 23.22 1.72 Dipteryx panamensis 1.31 0.26 21.76 2.44 18.37 0.88 Inga punctata 1.03 0.36 6.24 0.72 4.96 1.01 Ochroma pyramidal e 0.83 0.45 40.73 5.17 33.41 7.10 Terminalia amazonia 1.74 0.16 19.05 3.26 13.39 0.18 Anacardium excelsum 44.68 2.94 39.98 6.39 Castilla elastica 28.11 2.33 29.15 5.68 Luehea seemannii sig 10.64 1.12 8.30 1.26 Ca (mg g-1 dry mass) C. longipes 1.39 0.24 1.47 0.28 1.87 0.07 D. panamensis 0.25 0.82 0.29 0.08 0.27 0.02 I. punctata 1.51 0.21 0.38 0.08 0.26 0.01 O. pyramidale 0.34 0.75 3.29 0.34 3.01 0.73 T. amazonia 1.21 0.29 0.86 0.07 0.75 0.05 A. excelsum 0.24 0.06 0.39 0.1 6 C. elastica 2.03 0.34 2.07 0.02 L. seemannii 0.35 0.03 0.26 0.00 Mg (mg g-1 dry mass) C. longipes 1.18 0.30 1.12 0.34 1.59 0.21 D. panamensis 1.16 0.31 0.44 0.09 0.34 0.02 I. punctata 1.65 0.17 0.34 0.07 0.22 0.02 O. pyramidale 0.15 0.89 1.71 0.09 1.78 0.43 T. amazonia 1.27 0.27 0.82 0.04 0.75 0.04 A. excelsum 2.32 0.70 2.58 0.81 C. elastica 0.68 0.05 0.58 0.05 L. seemannii 0.43 0.04 0.28 0.04 N (mg g -1 dry mass) C. longipes 1.58 0.19 1.12 0.07 0.90 0.12 D. panamensis 1.14 0.32 0.54 0.05 0.48 0.02 I. punctata 1.22 0.29 0.62 0.07 0.51 0.07 O. pyramidale 0.16 0.88 1.14 0.16 1.10 0.21 T. amazonia 1.03 0.36 0.71 0.23 0.47 0.03 A. excelsum 0.55 0.12 0.68 0.08 C. elastica 0.72 0.03 0.76 0.05 L. seemannii sig 0.42 0.05 0.32 0.06 Phosphate P (mg g -1 dry mass) C. longipes 1.06 0.35 0.32 0.02 0.26 0.04 D. panamensis 3.96 0.02 0.32 0.03 0.19 0.02 I. punctata 1.86 0.14 0.11 0.01 0.08 0.01 O. pyramidale 0.06 0.95 0.37 0.07 0.36 0.06 T. amazonia 1.73 0.16 1.48 0.24 0.92 0.22 A. excelsum 0.15 0.02 0.14 0.02 C. elastica 0.14 0.00 0.15 0.03 L. seemannii sig 0.13 0.04 0.09 0.04
43 Table 25 Continued Litter drying temperature 60 C 22 C Water soluble element or pH t st at P value mean SE mean SE pH C. longipes 0.71 0.52 7.65 0.23 7.48 0.06 D. panamensis 0.03 0.98 5.76 0.06 5.77 0.12 I. punctata 2.02 0.11 5.78 0.06 6.02 0.10 O. pyramidale 3.03 0.04 6.42 0.12 6.02 0.05 T. amazonia 0.62 0.57 4.66 0.03 4 .64 0.03 A. excelsum 5.19 0.02 5.15 0.09 C. elastica 6.28 0.18 6.24 0.10 L. seemannii sig 5.45 0.11 5.72 0.14 Significant P values are highlighted in bold. All tests have four degrees of freedom. All values are extracted elements per gram oven dried leaf litter. Means and standard errors from three species of the main study are included for comparision but were not reanalyzed here. They were i ncluded in a twoway ANOVA ( Table 2 3 and 24 ) and significance is indicated by 'sig' here. All data sho wn here are from extractions using a 1:50 litter to solution ratio and 4 h extract time.
44 Table 26. The effect i nhibiting microbial act iv i ty on water soluble carbon (C) and phosphateP for replicates of litter from the same tree (a) and replicates using l itter from different trees (b) a) Carbon (mg C g-1 dry mass) Phosphate (mg P g-1 dry mass) Species h w/ DI w/ NaN 3 t stat P value w/ DI w/ NaN 3 t stat P value A. excelsum 4 38.6 0.9 45.1 1.0 4.7 0.009 0.14 0.01 0.14 0.02 0.1 0.891 24 49.7 .6 71.6 4.8 3.3 0.030 0.23 0.01 0.22 0.01 0.7 0.520 C. elastica 4 30.0 1.2 30.0 2.1 0.0 1.000 0.13 0.01 0.14 0.02 0.3 0.751 24 33.8 0.3 43.5 1.1 8.8 0.001 0.15 0.01 0.20 0.01 3.6 0.023 L. seemannii 4 6.1 0.6 6.3 0.7 0.3 0.805 0.09 0.02 0.11 0.01 1.0 0.387 24 25.3 1.6 26.6 1.5 0.6 0.605 0.29 0.03 0.38 0.05 1.5 0.208 b) Carbon (mg C g-1 dry mass) Species h w/ DI w/ NaN 3 t stat P value A. excelsum 4 40.0 2.6 45.5 3.2 1.3 0.259 24 46.8 7.5 58.6 8.0 1.1 0.342 C. elastica 4 34.7 3.2 31.1 3.3 0.8 0.471 24 34.7 1.0 45.5 4.0 2.6 0.058 L. seemannii 4 8.8 1.3 7.7 1.1 0.6 0.557 24 22.0 0.8 25.0 0.3 3.5 0.025 Results are for ai r dried leaf litter (22C) using a 1:50 litter to solution extraction ratio and two extraction times (4 and 24 h). Means standard errors are reported. All tests have four degrees of freedom. Deionized water (DI) is compared to sodium azide (NaN3) treat ments, which inhibit microbial activity. Significant P values (<0.05) are in bold.
45 Figure 21. Water soluble C and phosphateP shown as a function of extraction time for freshly senesced air dried leaf litter. Points are jittered along the x axis. Thr ee litter to solution extraction ratios are shown with means 1 standard error (n=3 per extraction ratio). Michaelis Menten or linear fits are shown based on comparison of Akaike's information criterion (AIC) values. All models that are displayed were si gnificant, as evaluated through the P value for the main parameter (either Vm or slope; Table 22). All water soluble element values are expressed an oven dried weight basis.
46 Fig ure 2 2 Water soluble Ca, Mg, and total dissolved N expressed as a functi on of extraction time for freshly senesced air dried leaf litter using 1:50 litter to solution extraction ratio. Means 1 standard error are shown. Akaike's information criterion (AIC) scores were used to choose Michaelis Menten versus linear fits and the function is shown when the main parameter (either Vm or slope) was significant (Table 22).
47 CHAPTER 3 DIFFERENTIAL L EAF LITTER NUTRIENT SOLUBILITY ACROSS 41 LOWLAND TROPICAL FOREST WOODY SPECIES Overview Leaching is a mechanism for the release of nutri ents from litter, or senesced leaves, and can dr ive plant soilmicrobe interactions in lowland tropical forests. Although litter leaching is well established in our conceptual models of decomposition, nutrient solubility and predictors of solubility have seldom been quantified across species for multiple elements This study used a standardized extraction procedure (1:50 litter to solution ratio and 4 h extraction) and leaf litter from 41 tropical tree and liana species to investigate how solubility varies among elements, and whether soluble nutrients could be predicted by litter traits (e.g., lignin, total element concentrations). In addition, nutrient forms (i.e., inorganic and organic) and ratios in leachate were investigated. Across the 41 species, ele ments were found to have differential solubility. On average 100% of total potassium (K), 35 % of total P 28% of total sodium (Na), 5% of total N and 3% of total C, were soluble during the 4 h extract ion The high solubility of P suggests this nutrient bec omes rapidly available to litter microbes without a metabolic cost. Water soluble elements on a litter mass basis were strongly predicted by t otal initial litter element concentration for P (r2=0.66), K (r2=0.79), and Na (r2=0.51), while a significant but weaker relationship was found for N (r2=0.36) and there was no significant relationship for C or calcium (Ca). Common predictors of decomposition rates had few significant correlations with element solubility. Soluble C on a litter mass basis was negativ el y related to lignin content (r2=0.19; p< 0.004) Solubility of nutrients was linked within a species: when a species ranked high in the soluble fraction of one element, it also ranked high in solubility of other elements, as evaluated through
48 Principle Co mponents Analysis. Differential solubility of litter elements emphasizes that elements cannot be treated equally in our conceptual and empirical models of decomposition. Furthermore, terrestrial element cycles, especially that of P and Na, may be more sens itive to precipitation patterns than previously appreciated. Background and Hypotheses Litter leaching has long been a key component of conceptual models of decomposition (Melin 1930; Attiwill 1968; Gosz et al. 1973; Lousier and Parkinson 1978; Berg and St aaf 1980). The role of leaching is also prominent in our thinking of how substances move from layers of decomposing litter through the soil profile (Schoenau and Bettany 1987; Qualls et al. 1991). In addition, litter leachate plays an important role in pl ant soilmicrobe interactions (Wurzburger and Hendrick 2009; Bowman et al. 2004). However, a basic understanding of how litter element solubility varies among species, factors regulating litter element solubility and how leachate stoichiometry compares to that of litter have been relatively little quantified. A better understanding of litter element solubility has the potential advance our knowledge of key questions related to decomposition and plant soilmicrobe interactions. For example, biological acqui sition of nutrients released through water extraction do not require a metabolic investment, in contrast to microbial mineralization of nutrients from organic matter. Information on the solubility of elements from litter could therefore advance our underst anding of nutrient bioavailability, which is paramount for understanding controls on nutrient cycles (Attwill 1968) and for refining the role of ecological stoichiometry in predicting ecosystem processes (Httenschwiler and Jrgensen 2010). Litter leaching may also offer insight into additive and nonadditive effects of mixed species litter combinations on decomposition. Transfer of
49 nutrients among species has been proposed as a mechanism underlying nonadditive effects of mixed species litter on decomposit ion rates (Gartner and Cardon 2004; Ball et al. 2009). Nutrient transfer is often considered to be due to biotic activity (Briones et al. 1996), but species variation in litter element solubility may be an additional mechanism. Moreover, differences in lit ter nutrient solubility might underpin positive relationships between precipitation and decomposition rates (Austin and Vi tousek 2000; Powers et al. 2009; but see Schuur 2001). This study focused on two metrics of litter element solubility: 1) water soluble element, defined as the quantity of water extracted element expressed on a litter mass basis (dry weight equivalent) and 2) the soluble fraction, or the percent of total litter element that is water extracted. Many studies interpret the initial phase of decomposition to be a leaching phase ( Swift et al. 1981; Prescott et al. 1993), although the role of leaching as the main mechanism underlying net element losses during this phase is seldom tested. Nevertheless, such studies provide a baseline for the pr ediction that litter elements should differ in their soluble fraction. For example, litter potassium (K) has been reported as having high solubility, while litter calcium (Ca) is noted as having low solubility (e.g., Gosz et al. 1973). Furthermore, element s are known to differ in their mobility in both live leaves and during senescence (Fisher 2007) which may lead to differences in water solubility among elements for a given species (Attiwill 1968). If litter elements show differential solubility, this may result in soluble fractions that are consistent in rank across species, or soluble fractions may converge on an average that is similar across species (e.g., K > Ca and the soluble fraction for each
50 element is similar across species). Predictability of s oluble fractions would in turn translate into patterns in leachate stoichiometry. Forming hypotheses about the role of litter solubility in ecosystem processes also requires an investigation of litter element solubility as a general species trait. That is if litter solubility is high for one element, does the species also show relatively higher soluble fractions for other elements? Moreover, the ability of well studied litter traits, such as lignin content and total element content, to predict litter elem ent solubility would place litter solubility within context of already rich data sets. It is noted that studies on solubility of live leaves, which are of interest in throughfall studies, are of limited use for understanding litter leaching because solubil ity of elements of live leaves and senesced leaves can differ (Tukey 1970) due to biochemical changes during senescence. This study investigated litter solubility and leachate chemistry for 41 woody species from a lowland tropical moist forest. A 4 h water extract ion with 1:50 litter to solution ratio was used to generate two indices of initial solubility. There were four hypotheses : 1) Litter elements differ in their solubility and the soluble fraction is consistent among species for a given element. If so the total litter element concentration will predict water soluble element, and intercepts from these regressions will not be significantly different from zero. 2) Inclusion of common metrics of litter quality (i.e., litter lignin content, leaf toughness C:N, lignin:N) improve the ability of linear models to predict water soluble elements when added as additional explanatory variables to models already including total litter element concentration. 3) Element solubility is a general litter trait. That is if a species has high solubility of one element, it also shows high relative solubility of other elements. 4) Because element soluble
51 fractions are consistent, leachate chemistry will show predictable relationships in nutrient stoichiometry. Furthermore, relationships in leachate inorganic/organic nutrient forms, leachate pH and C quality were investigated. Methods Sample Collection and P reparation Leaf litter was collected weekly between January 2006 and January 2007 from Poachers Peninsula, the southern point of Barro Colorado Island (BCI) in Pana ma. BCI supports a lowland tropical moist forest and mean annual precipitation is 2600 mm. Approximately 95% of the rainfall occurs during the 8 month wet season and rate of litter fall is highest during the distinct dry season between January and April (Wieder and Wright 1995). Litter was collected in 59 0.25 m2 traps elevated above the forest floor ( Wright and Cornejo 1990) Litter was removed from the traps and dried at 60C, sorted to the species level st ored in plastic zip bags and well mixed within a species. Petioles were removed before use and litter was cut into ~25 cm2 pieces. Sub samples were taken to determine conversion factors of stored litter to fresh ovendried (60C) weights. Litter represented 41 species, spanning 25 families and including canopy, midstory, and understory trees and lianas. All collected litter with more than 30 g per species was used (the average collection per species was 244 g litter). Species are listed in the Appendix. Wat er Extractions and Chemical A nalysis A 1:50 litter to solution ratio and a 4hour extract ion time were used for all water extracts and there were three replicates per species with the exception of a mmonium and nitrate measurements, which onl y had two repl icates per species Water extract replicates were separate leaching events, on different days, using bulked species litter.
52 The 4 h extract ion time serves as an index of initial nutrient solubility and has been shown to encompass the highest rate of loss, but does not represent exhaustion of extractable elements (Chapter 2). Two grams of litter (oven dried weight equivalent (ODW)) were placed carefully in 125 ml wide mouth polyethylene bottles and 100 ml deionized water was added. Bottles w ere placed on a shaker table on the low setting (180 oscillations per minute) for 4 h. Solution was poured off into beakers and then into tubes for centrifuging ( 8000 g for 10 min) After centrifuging, supernatant was immediately poured off. P hosphateP and pH were either analyzed immediately, or samples were stored at 4C overnight and analyzed within 24 hours. Dissolved organic carbon (DOC) samples were acidified to pH ~ 3 and stored at 4C until analysis the next day. These same samples were then used for specific ultraviolet absorption wavelength scans. Subsamples for future digests of total nitrogen, analysis of ammonium and nitrate and analysis for cations and total phosphorus were frozen after centrifuging. Molybdate reactive P (MRP) was read at 880 nm with a Ha ch DR 5000 UV vis spectrophotometer. Molybdate reactive P is commonly considered to be phosphateP, but may include some organic and condense d P compounds that are hydrolyzed during the assay. Dissolved organic carbon concentrations, operationally defined in this study as being the organic carbon remaining in supernatant after centrifuging, were determined by a Shimadzu TOC VCSH Total Organic C arbon analyzer. Solution pH was determined using Hach Sension 3 pH meter. A sodium hydroxide/potassium persulfate oxidation was used to determine total solution N concentrations. All N concentrations (total N, ammonium and nitrate) were determined colorimetrically on a Latchet Quickchem 8500 autoanalyzer. C ation and total P concentrations were determined by
53 Inductiv ely coupled plasma optical emission spectrometry (ICP OES) using an Optima 2100 (Perkin Elmer, Waltham, MS). The specific ultraviolet absorption at 254 nm (SUVA254), which is the measured absorbance divided by the C concentration, was used as an index of aromaticity (Hur and Schlautman 2003). Samples were diluted to 5 mg C L1 for the scans and are expressed here as L g1 C cm1. All samples had similar pH (~3). Standardizing for pH aids in comparison among samples, as large pH differences can influence S UVA (Hautala et al. 2000). Elemental composition of initial litter was determined by grinding and homogenizing 10 g of litter ( petioles removed) per species. From this bulked sample, percent C and N of non extracted litter were determined on ground sampl es using a total element combustion analyzer ( Thermoelectron Flash EA 1112 series ). A nitric acid digest of ground litter followed by ICP OES analysis was used to determine total P and cations. Leaf toughness was determined by a fracture toughness test us ing a portable universal tester with an attached pair of scissors (Lucas et al. 2000 ) after re drying stored litter at 40C. Following the methods of Kitajima and Poorter (2010), a small rectangle (~15 mm long) perpendicular to the central vein was cut out of the center of the leaf and sheared with the scissors. Secondary veins were avoided and this was verified by checking the force displacement curve. The work to shear the lamina was divided by the rectangular cross sectional area, which is equal to the fracture length times the lamina thickness. The lamina thickness was measured with a micrometer, avoiding visible veins. This method was developed for fresh leaves but was adapted to leaf litter. It was note d that dried litter of some species shattered during the test.
54 Despite this, the test should provide an index of lamina toughness of litter. Percent lignin was estimated using an Ankom fiber digester. The method operationally defines lignin as the fraction of mass remaining after sequential extraction of a neutral detergent, acid detergent and acid extract, minus the ash weight. Analytical triplicates were analyzed from the 10 g of ground litter from the bulked sample of each species. Statistical A nalysis Water soluble element on a litter mass basis w as calculated as element extracted per gram dry mass litter during the 4 h index. The soluble fraction, or the percent of total nutrient released during the water extract, was calculated as the water extractable element on a mass basis divided by the total litter concentration of the respective nutrient on a dry mass basis. Soluble fractions across specie s were summarized as means with standard deviations and coefficients of variation (CV) using the species averages A Principle Components Analysis (PCA) was used centered and scaled data in order to evaluate if solubility was a general property (i.e., if a species ranks high in the solubility of one element, does a species also tend to rank high in solubility of other elements?). Linear regressions were used to investigate if water soluble element on a litter mass basis could be predicted by the initial litter element concentration, percent lignin, lignin:N, C:N, or litter toughness Linear regressions were also used to evaluate if the soluble fraction could be predicted by percent lignin, lignin:N or C:N. For water soluble element on a litter mass basis, t he dependent variables were investigated with simple linear regressions and then select combinations were investigated using multiple regressions. For each element, m odels of litter toughness + total litter element lignin + total litter element lignin:N + total litter element and C:N + total litter element were
55 compared to models which included only total litter element. Model selection was based on Akaike' s information criterion (AIC) values. A difference of two AIC units was the criteria for considering models to be different. Because litter toughness values were only available for 35 species, this set of regressions omitted six species. For N and P, relationships between total extracted inorganic nutrient and total nutrient (inorganic + organic forms) were investigated using simple linear regressions. Correlations among solution nutrients (C:N, C:P, N:P on a molar basis ), pH and SUVA254 were analyzed using Spearmans rho. All analyses were conducted in R (R Core Development Team). Results Litter E lement Solubility Litter elements showed differential solubility. Averaged across species, nutrient solubility was found to range greatly among nutrients. Potas sium had the highest average soluble fraction (101%) while C had low est soluble fraction ( 3%) (Table 3 1). Measurements of water soluble element on a litter mass basis and total element concentration of litter were measured on separate samples drawn from t he bulked bag, which may explain the greater than 100% soluble fraction in the case of potassium (Table 3 1). In comparison to K, Na had a much lower soluble fraction, despite being another monovalent cation. On average, 28% of total Na was extracted. The soluble fraction of P was 35%, much greater than that of C, and also higher than that of N (5%; Table 3 1). There was high variation in the soluble fraction of Mg among species (Table 3 1). On a litter mass basis, more C was extracted than any of the other elements investigated (14.7 mg g1 litter; Table 3 1) during the 4 h extraction. Potassium ranked
56 second (4.5 mg g1 litter). All other elements had less than 1 mg extracted element per gram of leaf litter (Table 3 1). It is noted that litter for this s tudy was oven dried (60C), which may lead to small increases in solubility for some elements (i.e., C, P) compared to air drying for some species (Chapter 2). Predictors of E lement Solubility Total litter nutrient concentration was a significant predictor of soluble element on a litter mass basis for K, Mg, N, Na and P (Figure 3 1) but not for Ca ( P = 0.13) or C ( P = 0.12) as evaluated through linear regression. The linear relationships were strongest for elements that showed the highest average soluble fraction. For K, total litter K was found to explain 79% of the variance in soluble K on a litter mass basis Total litter P explained 66% of the variation in soluble P on a litter mass basis and litter Na accounted for 51% of the variance in soluble Na on a litter mass basis. For Mg, N and P, the intercepts were not significantly different from zero ( P =0.07, 0.40, 0.09 for Mg, N and P), demonstrating the soluble fraction of these elements did not vary with litter element concentration. The intercept terms were sig nificantly different from zero in the case of K (intercept=0.95, P =0.006) and Na (intercept=0.07, P =0.04), indicating that the soluble fraction of these elements decreases with increased litter concentration of the respective element. Overall, litter toughness and percent lignin had little to no ability to explain variation in water soluble elements among species. In simple linear regressions p ercent lignin was only a significant predictor for s oluble C on a litter mass basis and this relationshi p was weak ( r2= 0.19; P = 0.004 ). L itter toughness was not a significant predictor of s oluble element on a litter mass basis for any of the elements. In multiple regression analyses, including lignin as a predictor variable in models of soluble
57 element on a litter mass basis versus total litter concentration slightly imp roved model fit ) and Na predictor variable in regressions of s oluble element on a litter mass basis versus total l itter element concentration did not improve model fit for any of the relationships. C:N showed a negative relationship with water soluble P, N and K (r2= 0.18 and P = 0.007 for P; r2= 0.37 and P < 0.001 for N; r2= 0.28 and P < 0.001 for K). Lignin:N was also negatively related to soluble N on a litter mass basis ( r2= 0.25; P < 0.001) K ( r2= 0.19; P = 0.005) in addition to C ( r2= 0.18; P = 0.006) These significant regressions appear to be driven by relationships between water soluble N, P, K and total litter N concentrations ( r2= 0.18 and P =0.007 for water soluble K and total litter N; r2= 0.22 and P =0.002 for water soluble P and total litter N; Fig 31). When included as an additional variable in multiple regressions of water soluble element as a function of t otal litter element, neither the addition of C:N nor the addition lignin:N improved the model fit. Soluble fractions were investigated as a function of %lignin, lignin:N and C:N. The soluble fraction of C was negatively related to %lignin (r2= 0.30 and P < 0.001; Figure 32) and lignin:N (r2= 0.10 and P = 0.002). The soluble fraction of N was negatively related to lignin:N (r2= 0.10 and P = 0.04) and C:N (also r2= 0.10 and P = 0.04). Other elements did not have significant relationships between their soluble fr actions and %lignin, lignin:N or C:N. Solubility of elements was a general property of the species evaluated. A PCA with the soluble fraction showed that 57% of the variation was explained by the first axis and loadings for C, Ca, K, Mg, Na, P and N were all in the same direction (Table 32). This suggests that when a species ranked high in the soluble fraction of one element, it also
58 had high solubility of other elements, relative to the other species. Results were similar for element solubility on a litter mass basis, although the variance explained by the first axis was slightly less (49%, data not shown). Tw o outliers were removed. For P, Faramea occidentalis was removed because it had a notably low soluble P fraction. It also had the highest concentrat ion of extracted aluminum (Al) out of the 41 species investigated (1.3 mg g1 dry mass), which was 160 times great er than any other species. The high Al concentration likely resulted in significant aluminum phosphate precipitation, decreasing th e reading f or solution P concentration (calculations not shown) In addition, Uncaria tomentosa was identified as an outlier with low total litter K and removed for K analyses. Leachate C hemistry Leachate stoichiometry among species was similar for N and P (r ho=0.63 P < 0.0001; 6.6 2.6 (average on molar basis SD ); Figure 3). The relationships between leachate C and N (r ho=0.53, P < 0.0004 31.03 16.5) and C and P (r ho=0.42, P = 0.007, 192.1 129.7) were also significant but these ratios were not as similar among species as N:P. Phosphate was strongly related to total P in the extract s (r2 = 0.99; Figure 3 4b) with > 94 % of all extracted P in the inorganic form when averaged across species Inorganic N was also positively related to total N in extracts across species (Figure 3 4a). However, in contrast to P a smaller percent of extractable N wa s in in organic form (25% when averaged across species (range of 6 52%). Solution pH, (range 4.3 6.4), was weakly negatively correlated with SUVA254, an index of carbon a romaticity (rho= 0.30; P = 0.056). Leachate stoichiometry was compared to stoichiometry of fresh litter and leached litter (Figure 3 5; stoichiometry of leached litter was calculated and not directly
59 measured). Due to the higher solubility of P compared to N or C, leaching had a greater influence on N:P and C:P ratios compared to C:N (Figure 35 a,b,c). Discussion Litter elements were found to vary greatly in their solubility when evaluated as soluble fraction and as soluble element on a litter mass basis (Table 31). Potassium was highly soluble, with 100% of the total litter K released during the 4 h water extract ion Sodium and P were less soluble than K (28% and 35%, respectively) but much more soluble than N and C, when averaged across species (solubl e fraction of N was 5% and C was 3%; Table 31) Litter K has often been noted as having high solubility (e.g., Attiwill 1968; Gosz et al. 1973; Brinson 1977; Lousier and Parkinson 1978) but explicit tests of K solubility across species are rare. By quant ifying initial solubility of 41 species, this study confirms that K is highly soluble in lowland tropical forest litter Potassium receives much less attention than N and P in studies of terrestrial nutrient cycling, even though K has been found to limit f orest primary productivity in temperate forests (Tripler et al. 2006) and acts as a control on fine root investment in a lowland tropical forest (Wright et al. in press ) The high solubility of litter K for all 41 species suggests that ecosystem cycling of K may be sensitive to variation and pulses in precipitation. High solubility of leaf litter phosphorus has often been noted ( Chapin et al. 1978, Lousier and Parkinson 1978; Attiwill 1993; Parsons et al. 1990; Polglase et al. 1994) but seldom quantified ac ross species. In this study, P was much more soluble than C or N: on average 35% of the total phosphorus was extracted by water during the 4 h index (range 16 53 % ) Large fluxes of soluble P contrast with numerous litterbag studies in which P remained c onstant or becomes immobilized from the surrounding environment
60 during decomposition (Attiwill 1968; Staaf and Berg 1982; Hobbie and Vitousek 2000; McGroddy et al. 2004). However, an element can be solubilized from litter without leaving the litter enviro nment, perhaps due microbial immobilization by litter microbes. Alternatively, water extracted P may be removed from litter while P from the surrounding soil environment is immobilized. In addition, the soluble flux may have been removed prior to initiating the litterbag study. Understanding which of these factors is at work is important for understanding the fate of the soluble flux and its role in decomposition. In contrast to P, N solubility was fou nd to be much lower, only 5% when averaged across speci es (range 1.4 11% ). Moreover, almost all water extracted P was in inorganic form while N was largely organic (Figure 3 4). Because P solubility was high, and because a large percentage of P was leached as phosphate, it is likely to be readily available to litter microbes or involved in physiochemical interactions with soil minerals (Mattingly 1975) Microbial use of N in contrast, may incur greater metabolic cost. The dominance of phosphate in leachate, rather than organic P, is likely related to chemist ry in live leaves and chemical changes during senescence. In live leaves, phosphate is critical part of ATP and ADP cycling and is also stored in vacuoles as a reserve to maintain cytoplasm phosphate at a constant concentration (Lauer et al. 1989; Sinclair and Vadez 2002). Foliar P concentration is dependent on soil P conditions and species, but inorganic P has been shown to range from 10 50% of the total foliar P (Bieleski 1973; Ostertag 2010). Up to 75% of leaf N can be present in leaf chloroplasts, larg ely as proteins (Hortensteiner and Feller 2002). During senescence, catabolism converts compounds from the growth phase of the leaf to compounds that are exportable before leaf abcission (Lim et al. 2007). The break down of lipid membranes
61 and RNA, which are rich in P, results in increased phosphate concentrations (Taylor 1993; Thompson et al. 2000), which can then be remobilized (Prez Amador et al. 2000). In contrast, N can be retranslocated in organic forms, such as amino acids (Hortensteiner and Feller 2002). Due to both live leaf concentrations and requirements for remobilization, the inorganic form of P is likely a larger percentage of total element in a senesced leaf compared to N, which may underlie differences in N and P solubility and nutrient form in leachate. The differences in N and P solubility result in leachate stoichiometry with a much lower N:P ratio than that of litter; leachate N:P was 6.6 (range 2.3 14.8), while initial litter N:P was 51.4 (range 30.1 85.1) and the calculated N:P of leach ed litter was 76.9 (range 47.7 126.4; Figure 35; molar ratios). The large difference in these ratios sets up contrasting expectations for N versus P limitation to biological processes (Redfield 1958). Although t hresholds for N and P limitation of microbi al activity are not clear a microbial ratio of 7:1, determined by a review of global studies (Cleveland and Liptzin 2007) can be used as a guide. The contrast between leachate and litter stoichiometry suggests leachate production may make phosphorus relatively less limiting to decomposition than litter stoichiometry predicts. If so, nutrient solubility could play into proposed models of vertical nitrogen limitation in tropical forests (Hedin et al. 2009). It is argued that the litter layer is rich in C but relatively low in N, creating a vertical layer in the forest profile where active N fixation could be beneficial despite tropical systems being generalized by overall N richness ( Jenny 1950; Vitousek 1984) This decoupling between the N rich soil and the N poor litter is proposed as a hypothesis to explain the paradox that tropical forests can sustain N richness over time, without down regulating
62 N inputs through fixation (Hedin et al. 2009) Experimentally manipulating litter so it is relatively more N p oor compared to other nutrients provides some support for this argument. Litter from P fertilized plots in Hawaii was shown to have higher N fixation rates, measured as acetylene reduction (Vitousek and Hobbie 2000). Litter from P fertilized plots had low er lignin concentrations and, because external P addition to the soil had little influence on N fixation rates, it was hypothesized that carbon quality rather than P may constrain heterotrophic N fixation (Vitousek and Hobbie 2000). However, litter in P fe rtilized plots also had higher P concentrations and, if the soil and litter are decoupled, we suggest that an alternative mechanism underlying increased N fixation in the P plots may be related to increased soluble P (speculation based on Figure 31). In summary, it is proposed that the low solubility of N and higher solubility of P add support to a hypothesis that N may be likely to be limiting to decomposition. Sodium was found to have a much lower soluble fraction than K, which may be due to location in leaf cells. Sodium is a nonessential element but can sometimes serve as a K replacement (Subbarao et al. 2003), as both are small monovalent cations. Potassium is an essential element used for a number of functions and is found in the cytoplasm and vacuol e. In contrast, Na can be concentrated in the vacuole but very little is found in cytoplasm (Leigh and Wyn Jones 1986). It has been reported that the cytoplasm disappears during leaf senescence, but vacuole membrane permeability does not change (Thomas and Stoddart 1980), which could potentially influence solubility of nutrients from senesced litter. Sodium is not essential for plant growth but it is essential in animal physiology. Animal consumers of plants have much higher Na concentrations than plants (Subbarao
63 et al. 2003), creating a disconnect between the Na requirements of animals and the ability of plant consumption to meet the requirement Sodium availability was demonstrated to limit litter decomposition in a lowland tropical forest (Kaspari et al. 2009), showing that Na cycling can exert direct controls over carbon cycling in some sites This study demonstrates that Na is relatively soluble and total litter Na is a predictor of extracted Na in fresh litter. Although Na has a lower soluble fracti on than K, in instances where Na is limiting to decomposition (Kaspari et al. 2009), any movement away from litter during decomposition has the potential to significantly impact decomposition rates. Differences in Na solubility among systems and species may be an important research question for better understanding connections among plants, decomposers, climate and carbon cycling. Water soluble C showed significant but weak negative relationships with lignin ( r2= 0.19 ; P = 0.004) and lignin:N ( r2= 0.18; P = 0 .006 ) which are common predictors of litter decomposition rates. This suggests that the soluble flux of carbon may partially underlie commonly observed relationships between decomposition rates and lignin predictors. This is supported by a study in a lowl and tropical forest with high rainfall, where C solubility correlated positively with initial decomposition (050 days), but was not a predictor of overall decomposition rates (Wieder et al. 2009). In this study the direct contribution of extracted C to mass loss is small. Soluble C on a litter mass basis ranged from 3 37 mg g1, yielding a maximum of 3.7% of mass loss due to C solubility during a 4 h solubility index. Soluble C on a litter mass basis and lignin may be related simply because soluble compone nts increase when nonsoluble components of a leaf decrease. If the relationships between soluble C and lignin, and lignin:N, underlie
64 relationships between decomposition rates and lignin, and lignin:N, it is reasoned that the mechanism would be due to an indirect effect of C solubility on decomposition rates and not direct effects of initial solubility on mass loss. Although the soluble fraction was low, the quantity of carbon in solution was high (average of 293 mg L1 across species, using 1:50 litter t o solution ratio and 4 h extraction) and this initially soluble C could fuel microbial activity and enhance rates of decomposition (Fontaine et al. 2003). In addition to increasing decomposition rates, leachate C can influence nutrient cycling through a number of other pathways. Nutrients recycled from litter can only be used for plant growth if th ey can reach plant roots, and l eaching of carbon from leaf litter may influence this process by providing a carbon source to microbes (Kalbitz et al. 2003). In turn, microbial biomass can then serve as a reservoir of potentially available nutrients, thus preventing leaching out of the system (e.g., Brooks et al. 1998) and/or reducing movement to soil pools of low availability (e.g., Oberson and Joner 2005). As another potential pathway C leached from litter may influence sorpt ion and desorption reactions of nutrients on soil surfaces (Hunt et al. 2007; Chapter 4), potentially leading to increased plant nutrient availability. Our results illustrate that litter el ements have differential solubility. These results emphasize that elements cannot be treated equally in our conceptual and empirical models. Terrestrial element cycles, especially the P cycle, may be more sensitive to variation in precipitation than previ ously appreciated. Differential element solubility, the strong ability of water soluble elements to be predicted by total litter concentration of a given element, and the emergence of solubility as a general litter trait may be similar in other systems. Th e generality of these results to other biomes should be a focus of
65 future work. If supported in other systems, differential litter nutrient solubility may be another avenue by which precipitation pulses and timing (Kieft et al. 1987; Fierer and Schimel 20 02) drive terrestrial nutrient cycling.
66 Table 3 1. Water solubility of litter elements during a 4 hr 1:50 litter to solution ratio extract. Soluble fraction Water soluble element (% of total element extracted) (mg element g 1 dry mass) element me an SD CV mean SD CV C 3.3 1.8 0.6 14.69 7.49 0.51 Ca 4.2 4.5 1.1 0.69 0.80 1.17 K 101.2 25.4 0.3 4.50 1.98 0.44 Mg 20.5 12.6 0.6 0.82 0.68 0.83 N 4.6 2.1 0.5 0.63 0.37 0.27 Na 28.0 10.3 0.4 0.25 0.11 0.45 P 34.6 10.4 0.3 0.21 0.10 0.49 Si 0.09 0.06 0.75 Data represent 41 species with the exception of P and K which only had 40 species included in the analysis (see text). Each species value was created by averaging three replicates
67 Table 3 2. First three components of Principle Comp onents Analysis using soluble fraction. Axis Variable or statistic 1 2 3 Soluble fraction: Component Loading C 0.401 Ca 0.379 0.23 0.531 K 0.363 0.287 0.597 Mg 0.48 0.179 N 0.387 0.446 0.253 Na 0.351 0.459 0.272 P 0.245 0.668 0.43 1 Component value Proportion of Variance 0.57 0.17 0.10 Cumulative Proportion 0.57 0.73 0.83 Standard deviation 1.99 1.08 0.83
68 Figure 3 1. Linear r elationships between soluble element on a litter mass basis and total initial litter element concent ration. Extracted elements are the average of three extracts SE using a 4 h extraction time. Total litter concentration is based on one analysis drawn from 10 g ground litter. D ashed lines indicate slopes of 1:1, 1:2 and 1:4, which represent soluble fractions of 100%, 50% and 25%, respectively (refer to labels in the Ca plot).
69 Figure 32. Linear relationship between soluble C fraction (expressed as a percent of total litter C) and an index of percent leaf litter lignin as evaluated through a fiber dig est.
70 Figure 33. S toichiometry of soluble C on a litter mass basis phosphateP and total N Each point represents the average of three extracts for bulked litter of one species. Spearmans rho statistics are shown.
71 Figure 34. Regression between i norganic and total N (a) and P (b) in water extracts. Total N was determined through a persulfate digest, total P was read using an ICP, while phosphateP and inorganic N were obtained colorimetrically.
72 Figure 35 Stoichiometry of initial litter, lea ched litter and litter leachate (from the 4 h extract; n=40 or 41 species). Stoichiometry of leached litter was not measured directly. It was calculated by subtracting soluble element on a litter mass basis from the initial element litter concentration f or each species. All values are on a molar basis.
73 CHAPTER 4 LEAF LITTER INPUTS DECREASE PHOSPHATE SORPTION IN A STRONGLY WEATHERED TROPICAL SOIL OVER TWO TIME SCALES Overview In strongly weathered soils, leaf litter not only returns phosphorus to the soi l environment, it may also modify soil properties and soil solution chemistry, with the potential to decrease phosphate sorption and increase plant available phosphorus. Using a radioactive phosphate tracer (32P) and 1 h lab incubations this study investig ated the effect of two time scales of litter inputs on phosphate sorption: 1) long term field litter manipulations (litter addition, control and litter removal) and 2) pulses of litter leachate (i.e. water extracts of leaf litter) from five species. Leachate pulse effects were compared to a simulated throughfall, which served as a control solution. Soil receiving long term doubling of leaf litter maintained 5fold more phosphate in solution than the litter removal soil. In addition to the quantity of phosphate sorbed, the field litter addition treatment decreased the strength of phosphate sorption, as evaluated through Bray 1 extracts of sorbed 32P. In litter removal soil, leachate pulses significantly reduced phosphate sorption in comparison to the throughf all control for all five species evaluated. However, the ability of leachate pulses to reduce phosphate sorption decreased when soil had received field litter inputs. Across soils the effect of leachate pulses on phosphate sorption increased with net DOC s orbed, with the exception of one species that had a higher index of aromatic carbon concentration. This work demonstrates that litter inputs, as both long term inputs and leachate pulses, can decrease the quantity and strength of phosphate sorption, which may lead to increased plant availability of this key nutrient.
74 Background and Hypotheses Much of the phosphate that is sorbed, or removed from solution by either adsorption or precipitation, by strongly weathered soils is not readily available for plant u ptake due to soil mineralogy and high concentrations iron and aluminum oxides (Mattingly 1975; Uehara and Gillman 1981; Fox and Searle 1978; de Mesquita and Torrent 1993; Fontes and Weed 1996). Phosphate sorption by iron and aluminum oxides is one reason p hosphorus is thought likely limit net primary productivity in strongly weathered soils (Walker and Syers 1976; Crews et al. 1995), and mechanisms that alter phosphate sorption may therefore act as controls on net primary productivity. Investigations in agr icultural systems (Dalton et al. 1952; Moshi et al. 1974; Ohno and Crannell 1996; Negassa et al. 2008) and a pine plantation (Bhatti et al. 1998) show that organic matter additions have the potential to decrease soil phosphate sorption. Several lines of reasoning suggest that litter inputs could decrease phosphate sorption in lowland tropical forests on strongly weathered soils. Lowland tropical forests tend to have high litter inputs (Vitousek 1984) and rapid litter turnover (Cusack et al. 2009); therefor e, heavy rains, either seasonally or throughout the year, can create high concentrations of dissolved organic carbon (DOC) and solute movement from litter to the soil (Cleveland et al. 2006). In addition, long evolutionary histories between tropical fores t plants and their environments argue for selection of plant traits that provide advantages by increasing nutrient availability (van Breeman 1993; Binkley and Giardina 1998). Mechanisms underlying the potential influence of litter on phosphate sorption can be divided into two categories effects on microbial cycling and effects on soil chemistry. This study investigated litter effects on soil chemistry over two time scales: 1)
75 the cumulative effect of long term litter inputs on soil chemical properties, w hich is referred to as field litter manipulation, and 2) the short term effect of litter leachate on phosphate sorption when both leachate and phosphate are simultaneously in solution, which is referred to as leachate pulses. Potential mechanisms under lying how field litter manipulation and leachate pulse treatments can influence phosphate sorption are similar and include mechanisms related to carbon competition for, and occupation of, phosphate sorption sites, in addition to changes in pH and ionic str ength. Below each variable is explored in turn. Carbon inputs may reduce phosphate sorption by decreasing available sorption sites (Perrott 1978; de Mesquita and Torrent 1993; Ohno and Crannell 1996; Easterwood and Sartain 2000). The ability of carbon to decrease sorption sites is dependent on carbon chemistry. For example, the presence of some low molecular weight organic compounds, such as oxalate and malate, can greatly decrease phosphate sorption (Swenson et al. 1949, Nagarajah et al. 1970; Lopez Hern andez et al. 1986; Kafkafi et al. 1988; Bhatti et al. 1998) while other compounds such as simple carbohydrates have no effect (Negassa et al. 2008). The effect of higher molecular weight carbon compounds is less clear. While humic and fulvic acids may som etimes decrease phosphate sorption to pure iron and aluminum oxides (Hunt et al. 2007; but see Borggaard et al. 2005) and soils (Sibanda and Young 1989), in other instances high molecular weight organic carbon compounds can increase phosphate sorption to s oil by forming metal bridges (Levesque and Schnitzer 1967; Borie and Zuninio 1983). However, phosphate sorbed through organic metal bonds is thought to be more readily available than phosphate sorbed to metal oxides (Gerke 2010).
76 Litter inputs may also inf luence soil phosphate sorption through pH. In soils with high iron and aluminum oxide content, theory predicts that phosphate sorption through ligand exchange reactions increases with decreases in soil pH (McBride 1994). This prediction has been supported for pure iron oxides (Hingston et al. 1967; Barrow et al. 1980). However, nonlinear responses, which often have sorption plateaus in midpH ranges somewhere between 4 and 6, have been reported for aluminum oxides and kaolinite clay (Chen et al. 1973; Edzw ald et al. 1976; Bar Yosef et al. 1988; He et al. 1997). Similar nonlinear responses have also been shown for a range of soils (Murrmann and Peech 1969; White and Taylor 1977). This range of soil responses may be expected because soils are complex mixtures of the clays and metal oxides that are often studied independently. Therefore, a number of predictions are possible for the effect of litter inputs on phosphate sorption. Nevertheless, given that long term litter inputs are likely to modify soil pH within the midrange that is often found to be a plateau of phosphate sorption, long term inputs effects on pH may have little effect on phosphate sorption. For leachate pulses, soil may buffer against leachate pH values. Ionic strength of soil solution may be altered by either long term litter inputs or leachate pulses. Ionic strength has been hypothesized to have little influence on specificsorption of phosphate because these reactions are controlled by ligand exchange and not by electrostatic attraction (G oldberg and Sposito 1984; Sparks 1995). However, increased ionic strength has sometimes been reported to increase phosphate sorption (Edzwald et al. 1976 and He et al. 1997 using kaolinite clay; McBride 1997 for goethite; Ryden et al. 1977 with soil) and t he magnitude of the effect can be dependent on the ion identity. For example, calcium (Ca) has been found to have a greater
77 influence on phosphate sorption than potassium (K) or sodium (Na) (Ryden et al. 1977; Barrow et al. 1980; Pardo et al. 1992). As a further consideration, phosphate inputs through litter may increase solution phosphate concentrations, which in turn can decrease the percent of solution phosphate that is sorbed (van Riemsdijk et al. 1984). In addition to being a potential mechanism underlying the effect of organic matter additions on phosphate sorption, the simultaneous addition of phosphate and carbon is a key covariate to consider when focusing on mechanisms related to carbon competition for sorption sites (Iyamuremye et al. 1996; Guppy et al. 2005). This study used a radioactive carrier free phosphate tracer (32P) to determine how leaf litter inputs influence soil phosphate sorption in lowland tropical forest soil. The study investigated the long term effects of litter manipulations on subsequent addition of 32P, the simultaneous addition of leaf litter leachate and 32P, and interactions between field litter manipulation and leachate pulses. The effect of field litter manipulation on post sorption recovery of 32P was also investigated. The effects of field litter manipulation on phosphate sorption via changes in soil properties were evaluated using soil that had received three treatments: litter addition, control and litter removal. The influence of leachate pulses on phosphate sorption was evaluated across the three soil treatments using leachate from five species. We had two main hypotheses: 1) soil with higher carbon content (but the same mineralogy and texture) would sorb less 32P because previously sorbed carbon would occupy potent ial sorption sites and 2) leachate pulses would reduce the percent of 32P sorbed to soil, relative to a throughfall control, due to the sorption of leachate DOC and
78 phosphate. Furthermore, it was hypothesized that when initial solution phosphate was controlled for, leachate pulses would continue to show decreased 32P sorption due to DOC competition for sorption sites. Considering the interaction between field litter manipulations and leachate pulses, a third hypothesis follows: effects of leachate pulses on phosphate sorption would be greater in the leaf litter removal soil. The pH was evaluated, but it was not expected to show clear trends with phosphate sorption and litter treatments. It was predicted that any correlation between increased phosphate sorpti on and increased ionic strength of Ca, magnesium (Mg), K and Na would be weak compared to the influence of carbon decreasing sorption. When considering the effect of field litter manipulation treatment on strength of phosphate sorption, it was hypothesized that more of the sorbed 32P would be readily extractable (Bray 1 extract; Bray and Kurtz 1945) for the litter addition soil. Finally, a small experiment considering the effect of solutionto soil ratios was included so that results from this study can be more readily compared to work using different ratios. Methods Leaf L itter Leachate and Soil Collection The study was conducted in the Republic of Panama at the Smithsonian Tropical Research Institute (STRI). Freshly senesced leaves of five species ( Albizia guachapele, Anacardium excelsum, Cecropia peltata, Castilla elastica, Ficus insipida) were collected from the forest canopy using the crane in Parque Metropolitano, a seasonal forest w ith 1740 mm mean annual precipitation. A. excelsum C. peltata and C. e lastica are distributed across a range of precipitation levels, while F. insipida and A. guachapele are mostly restricted to moist forests (Croat 1978). All five species have large geographical distributions within the Neotropics (Croat 1978). Leaves were collected
79 when they could be removed by a light touch meaning the abscission zone had formed, leaves had senesced and were ready to fall. Litter was transported in paper bags and then spread out in thin layers to air dry under ambient laboratory temperatu re and humidity (22 0.5C and 55 5%, respectively ) for approximately three weeks For each species, senesced leaves were collected from at least three individual trees, litter was cut when necessary so the maximum size was approximately 5 cm x 5 cm and samples were bulked at the end of the collecting period (August October 2008). Soil was collected from the Gigante Leaf Litter Manipulation (GLi M P) project (Sayer and Tanner 2010; Sayer et al. 2006) where three treatments were initiated in 2000: leaf l itter doubled, leaf litter r emoved and a control. Soils from these plots represent the levels of the field litter manipulation treatment. The GLiMP soil is classified as an Oxisol and was formed from Miocene basalt. Soil was collected (010 cm) from three of the five plots of each treatment in early October 2008, air dried under ambient laboratory temperature and humidity (22 0.5C and 55 5%, respectively ) sieved (2 mm), bulked by treatment and stored in plastic bags. Subsamples were dried at 105C for 24 h to determine air to oven dry weight conversions. Total soil ca rbon was determined by removing all visible organic fragments remaining after the 2 mm sieve, grinding soil in a ball mill and anal yzing on an Elemental Analyzer ( Flash EA 1112 Thermo, Bremen, Germany ). Leaf litter leachates, which represent the leachate pulse treatment, were created using a 1:30 litter to deionized water ratio based on grams oven dried weight equivalent (ODE). Litter was shaken at low speed (approximately 180 rpm) f or 4 h with 250 mL deionized water in wide mouth bottles. Solutions were immediately poured off and
80 centrifuged at 8180 x g for 10 minutes. Supernatants were stored at 4C for 1 5 days. All solutions were brought to room temperature before use in the sorption experiments. Solutions were analyzed for molybdate reactive phosphate (MRP; Hach DR 5000 UV vis spectrophotometer ) DOC ( Shimadzu TOC VCSH Total Organic C arbon analyzer ), and pH (Hach Sension 3 pH meter) Samples for DOC analysis were prepared (i.e., diluted and acidified) at the time of the phosphate sorption experiment. Cation concentrations were determined by inductively coupled plasma optical emission spectrometry (ICP OES; Optima 2100 Perkin Elmer Shelton, CT). Ionic strength due to Ca, K, Mg an i*Zi 2, where Zi is the valence of the ion and ci is the ion concentration in mol L1. Scans of specific ultraviolet absorption (SUVA), which is the measured absorbance divided by the carbon concentration and expressed here as L g1 carbon cm1, were used as an index of carbon chemistry (Hur and Schlautman 2003; Jaffrain et al. 2007) and measured on the diluted DOC samples. Phosphate Sorption E xperiments The effects of field litter manipulation and species sp ecific leachate pulses on phosphate sorption were evaluated with carrier free 32P sorption experiments in November and December 2008. The control was a simulated throughfall, which was created using nutrient concentrations and ratios from a review on thr oughfall in lowland tropical forests (Zimmermann et al. 2007; concentrations in Table 41). Within each soil, each species specific leachate pulse replicate was an independently created leachate using the bulked litter. Three replicates were included per s pecies. Experiments included five levels of initial phosphate spikes (+0, 3, 6, 9, 12 g mL1 for controls and
81 +0, 2, 4, 6, 8 g mL1 for leachates) in order to analyze the effect of leachate pulses on 32P sorption using initial phosphate solution as a cov ariant. In batches of 10 samples, 3 mL of DI water was added to 5 g of air dried soil (but all reported values are expressed on an oven dried weight equivalent basis) to mimic approximately 60% gravimetric water content. Sodium azide (1 mM) was included t o inhibit microbial activity. Thirty minutes later leachate, or throughfall solution, and phosphate (31P) were added. Phosphate stock ranged fro the target concentration. The 32P radioactive isotope tracer (0.25 mL of approximately 4 Mdpm per sample) was then added and samples were shaken on low (180 rpm) for an hour. Additions of the small quantities of radioactive 32P tracer used in this study did not have any appreciable effect on solution phosphate concentration. Total solutionto soil ratio was 10 mL to 4.66 g ODW. After centrifuging for 10 minutes at 8180 x g supernatant was immediately poured off. Activity of 32P remaining in the supernatant was read using a liquid scintillation counter (Beckman Coulter LS 6500 Multipurpose Scintillation Counter Fullerton, CA) and Ultima Gold AB cocktail. Activities of samples for determining batchspecific 32P tracer additions and blanks were also determined. Subsamples of supernatant were diluted, acidified and frozen for later analysis of DOC (see above). In addition, for the throughfall samples, Bray 1 extractable (Bray and Kurtz 1945) 32P was measured to provide insight int o how 32P was partitioned among soil sorption pools. Bray 1 extract is a weakly acidic ammonium fluoride solution (0.025 N HCl and 0.03 N NH4F) that is thought to extract soil phosphorus through the formation of fluoride complexes with aluminum and iron a nd is considered an index of readily extractable
82 phosphorus (Sharpley et al. 1987; Olander and Vitousek 2004) For this analysis, immediately after centrifuging and pouring off supernatant, a clean spatula was used to break up compressed soil in each centr ifuge tube. Bray 1 extract was added (for a solutionto soil ratio of 35 mL to 4.66 g ODW soil), samples were placed on the shaker table for five minutes before centrifuging at 8180 x g for 15 min. The Bray 1 supernatant was then immediately analyzed for 3 2P activity (see above). The effect of solution to soil ratio on 32P sorption was investigated using a small number of samples with deionized water. This experiment was carried out in June 2009 using the same bulked soil as used in the above experiments, but a slightly different protocol was followed. This experiment is included here to consider the dependency of percent 32P soil sorption on the solutionto soil ratios and to determine if the slope of the relationship varies among soil, which is helpful for comparing the results presented here to other work. Total solution was 7.5 mL, as opposed to 10 mL, and the step involving incubation with DI was eliminated. The total solution consisted of 7 mL of the throughfall, 0.25 mL DI and 0.25 ml 32P, all added at the same time. Also, samples were shaken on the high setting (280 oscillations per minute), instead of low (180 oscillations per minute). Analysis The effect of field litter manipulation on soil %32P sorption was statistically analyzed using a oneway ANOVA (only the samples that did not receive a 31P phosphate spike were used in this analysis). To evaluate the effect of leachate pulses on %32P sorption, two approaches were used. This study first considered if leachate pulses affected %32P sorption wit hout controlling for initial phosphate in solution and then this study evaluated the effect of leachate pulses while controlling for initial
83 phosphate in solution. For both approaches, only effects within a soil treatment were analyzed and comparisons were made between each species and the throughfall control. The effect of leachate pulse on %32P sorbed was first investigated using a oneway ANOVA without controlling for initial phosphate in solution. The analysis used samples that did not receive spikes o f 31P phosphate. Differences between the control and the leachates were evaluated using Tukey multiple comparison tests. This study then evaluated if leachate additions influenced 32P sorption, as compared to a throughfall control, while controlling for i nitial phosphate in solution. This was accomplished by including samples that received spikes of 31P phosphate and using an ANCOVA. For ANCOVA, a mixed effects model was used with the solution replicates set as a random effect, which accounted for the fact that initial phosphate concentrations were created by spiking subsamples of three replicates for each leachate. Effect sizes of leachate on phosphate sorption compared to throughfall were calculated using Cohens d statistic. The initial phosphate in solution was controlled for by using the intercepts from the ANCOVA where initial 31P phosphate in solution was the covariate: Cohens d = (intercept of %32P sorbed with throughfall intercept of %32P sorbed with leachate)/ pooled standard deviation. The in tercepts of %32P sorbed were also used in simple linear regressions within each soil to investigate the trends between phosphate sorption and initial pH, net DOC sorbed and ionic strength due to Ca, K, Mg and Na while controlling for solution
84 phosphate concentration. Net DOC sorbed was calculated as the DOC in the initial solution minus that of the final solution. For Bray 1 extracted 32P, the overall effect of soil type was investigated using ANCOVA with initial phosphate in solution as a covariate, which was appropriate because slopes among soils were not found to differ significantly. For the solutionto soil ratio experiment, soil effects on slopes of the %32P sorbed as a function of solutionto soil ratio were evaluated using ANCOVA and a linear model was fit for each soil. Species differences in leachate SUVA254 were evaluated using an ANOVA and multiple comparisons were investigated with a Tukey test. All statistical analyses were conducted in R ( R Development Core Team). Results Soil Carbon Content and Leachate Characteristics The leaf litter addi tion soil had 3.8 % total so il carbon, control soil was 3.2% and litter removal was 2.8% (Table 42 ) It is noted that resin extractable phosphate and pH also showed trends related to the litter treatment (BL Turner, unpublished data; Table 42), with the litter addition treatment showing potentially higher resin phosphate and pH. Leachates varied in initial concentrations of DOC (240 to 1053 mg C L1), phosphate (0.81 to 6.4 mg L1) and cations (Table 2). Effect of Field Litter Manipulation on Phosphate Sorption Litter addition soil sorbed less phosphate during the 1 h incubation than the litter removal and control soil, and all soil comparisons were significantly different from one another (Figure 41). Although differences were very small when considered on a percent sorbed basis, with all soils sorbing more than 99% of the added tracer, the relative differences of phosphate remaining in solution are large. In the litter addition
85 soil, five fold more phosphate was retained in solution compared to the litter removal soil. Effect of L eachate Pulses on Phosphate S orption Leachates of all species were found to decrease phosphate sorption in the litter removal soil when compared to the throughfall control (T able 43). This was also true when initial phosphate in solution was controlled for using an ANCOVA (Table 44; Figure 42), demonstrating that some leachate property, or properties, besides initial phosphate in solution underlie the decreased phosphate sorption. In the control soil, fewer species had a significant effect on decreasing phosphate sorption. In the litter addition soil one species, A. excelsum increased phosphate sorption (Table 44; Figure 42). Although the differences between throughfall and leachate were small, the increase in phosphate remaining in solution for the leachate treatments relative to the throughfall was found to be greater by more than a factor of three for some species/soil combinations (Figure 43). In the ANCOVA, there w ere two cases where slopes were not homogenous ( Castilla elastica in control soil and Anacardium excelsum in addition soil); however, both slopes converged with that of the throughfall solution as the functions approached the intercept. Therefore, the diff erence between slopes would result in failing to detect differences between treatments when differences may exist at higher levels of initial phosphate concentration. The analysis is therefore conservative in evaluating significance in these two cases. Lin ear regressions showed leachate pH and net DOC sorbed (Table 45) were not significant predictors of phosphate sorption for any of the soils (Table 46). Contrary
86 to expectations, the combined ionic strength of Ca, K, Mg and Na was strongly negatively corr elated with phosphate sorption in the litter removal soil (Table 46). When considering all soils together, the effect of leachate on reducing %32P sorption increased as net DOC sorbed increased (Figure 44) for four of the species investigated. However, A. excelsum did not fit this trend. A. excelsum shows the largest net DOC sorbed and one of the lowest effects of leachate on decreasing phosphate sorption in the litter removal soil, in addition to showing increased phosphate sorption with leachate presence in the litter addition soil (Figure 44). Effect of Soil Treatment on Bray E xtractable 32P Less than 10% of the 32P sorbed during a 1 h incubation was desorbed by the Bray 1 solution (Figure 45). However, the percent extracted from the litter addit ion soil was 40% greater than that extracted from the control and litter removal soil. The recovery of the 32P in the Bray 1 extract across soil treatments serves as an index of litter influences on desorption of phosphate and suggests that plant availabil ity of sorbed phosphate increases with litter addition. S olutionto soil R atios The solutionto soil ratio used for the 1 h incubations was found to have a strong effect on the percent of the initial 32P sorbed. For example, in the litter addition soil, th e percent of 32P remaining in solution varied from 99.75 to 98.25%, with a greater percent of the phosphate tracer sorbed when there was less solution per gram of soil. Furthermore, the effect of the solutionto soil ratio on phosphate sorption differs am ong litter manipulation treatments. The effect is greater for the litter addition soil than for the control and litter removal soil, as shown by slopes for the response of 32P sorbed as a function of solutionto soil ratio (Figure 46).
87 SUVA of L eachate A. excelsum had significantly higher SUVA254, which is an index of carbon aromaticity (Hur and Schlautman 2003; Jaffrain et al. 2007) than leachates of the other species, while the other species were not significantly different from one another (Figure 4 7; P <0.001 for ANOVA, df =4,10; P <0.002 for all A. excelsum Tukey comparisons with other species). Discussion This study demonstrates that leaf litter inputs can reduce phosphate sorption to soil through both field litter manipulation and leachate pulses The field litter addition treatment resulted in soil that sorbed significantly less phosphate (Figure 41). The effect was marked with the litter addition soil retaining fivefold more phosphate in solution compared to the litter removal soil at the end of 1 h incubations (Figure 41). It is noted that this relative difference can change with the solutionto soil ratio and would be even greater if a less concentrated ratio had been used (Figure 46). Results from the leachate pulse experiments show another avenue by which organic matter inputs can influence phosphate cycling, because leachate pulses from all five species decreased phosphate sorption to the litter removal soil. Interestingly, the effect of leachate pulses on decreasing phosphate sorption diminished as field litter inputs increased (Figure 43). This interaction demonstrates that the effects of leachate pulses are sensitive to previous inputs of organic matter and, because field litter inputs attenuated the effect of the leachate pulse, it is likely that mechanisms by which field litter manipulations and leachate pulses are controlling phosphate sorption are similar.
88 Mechanisms Underlying Phosphate Sorption R esults We reason that the mechanisms underlying the effect of the field litter m anipulations on phosphate sorption are due to more than a change in solution phosphate concentration. Although resin extractable phosphate may be greater for the 1 soil; Table 4 2) is small compared to the experimental phosphate spikes. Based on the experimental phosphate spikes of the litter removal soil (Figure 41), an increase in solution phosphate 1 1 solution) is expected to influence %32P sorpti on by less than 0.005 percentage points (here the term percentage points is used to clarify the reference to an absolute differences between the %32P results rather than a relative change). This is much less than the observed 0.43 percentage points between the litter addition and removal soil (0.52% 0.09%). In the leachate pulse experiments, the analysis of covariance determined that the leachate effect was due to more than an increase in solution phosphate concentration (Figure 42). Some variable, in ad dition to phosphate concentration, must therefore underlie the litter treatment effect. Linear regressions for the leachate pulse experiment suggest that pH is not a significant predictor of leachate pulse effect on phosphate sorption (Table 46). In the l itter removal soil, ionic strength (as evaluated by considering Ca, Mg, K and Na) was negatively correlated with phosphate sorption to soil, yet previous research suggests increased ionic strength of these ions should increase or have little influence on p hosphate sorption (Ryden et al. 1977; Sparks 1995). Given that three of the leachates had pH values >6 (Table 42), and that Ca makes a large contribution to ionic strength due to its valence charge, the possibility of calcium phosphate precipitation deserves mention (Barrow et al. 1980). However, this soil buffers against
89 initial solution pH and final supernatant pH values were similar, reflecting that of the soil after the hour incubation (determined using the control soil and solutions with initial pH r anging from 4.5 to 7.9; data not shown). Furthermore, calcium phosphate precipitation would decrease phosphate in solution, which would be interpreted in this study as increased sorption with increasing Ca concentration. The observed trend is the opposite of this expectation. Mechanisms underlying the correlation of decreased phosphate sorption with increased ionic strength of Ca, K, Mg and Na in the litter removal soil remain therefore unknown. It is noted that decreased activity coefficients (which would be expected with increased ionic strength) have been suggested, but were not supported, as a potential mechanism decreasing phosphate sorption (Bar Yosef et al. 1988). Given the results of this study, investigations of the relationship between activity co efficients and phosphate sorption may be an avenue for future work. Across soils, net DOC sorbed was correlated with the difference in 32P sorption between leachate and throughfall control, with the exception of A. excelsum (Figure 4 4). However, depending on chemistry, DOC can sorb to different types of sites and not all DOC that is sorbed can compete with phosphate for sorption sites (Yuan 1980; Kafkafi et al. 1988). Leachates can differ in DOC chemistry among species (Hongve et al. 2000) and it is theref ore likely that net DOC sorbed will provide only coarse scale information on the role of DOC on phosphate sorption. Yet, in the absence of detailed chemistry of all leachates, and knowledge of how each of the carbon compounds influence phosphate sorption, results suggest that interpreting net DOC sorbed is a useful metric. The use of net sorption of DOC was strengthened when data on specific
90 ultraviolet absorption, which are indices for carbon chemistry, were included in the interpretation. Leachate of A. e xcelsum was the only leachate found to increase phosphate sorption (which occurred in the litter addition soil). Anacardium excelsum leachate also had greater SUVA254, which is an index of aromaticity (Hur and Schlautman 2003), compared to the other four s pecies (Figure 47). In some cases, high molecular weight aromatic organic decomposition products, such as humic and fulvic acids, have been suggested to increase phosphate sorption by forming new sorption sites for phosphate (Levesque and Schnitzer 1967; Appelt et al. 1975; Gerke 2010). Although the A. excelsum leachate carbon would not be a decomposition product, the aromaticity could be related to why A. excelsum increased phosphate sorption in the litter addition soil. Continuing with this speculation, A. excelsum leachate could have similarly sorbed phosphate through organic metal bridges in the litter removal soil; however, sorption of phosphate through this pathway may have had a small effect relative to the competition of other forms of carbon with phosphate for sorption sites, resulting in a net decrease in phosphate sorption compared to the throughfall control. It is interesting to note that in contrast to specific sorption to soil, phosphate sorbed to organic matter is likely more plant available (Harter 1969; Gerke 2010). Our results for Bray 1 extractable 32P in the field litter manipulation soils demonstrate that litter additions can decrease the strength of phosphate sorption. Phosphate-32P sorbed in the litter addition soil was more readily ex tracted than 32P sorbed in the control and litter removal soil (Figure 45). Other studies report that organic matter additions can increase the fraction of phosphate going to nonspecific
91 sorption pools (Sharpley and Smith 1989; Easterwood and Sartain 200 0). As mentioned above, one mechanism for this is phosphate sorbed through organic metal bonds (Gerke 2010). Furthermore, a positive correlation between soil carbon and readily extractable phosphate has been shown for a number of systems ( Harter 1969; John son et al. 2003). Results show that the strength of phosphate sorption may be manipulated by organic matter inputs and supports the proposed link between readily extractable phosphorus and soil carbon concentrations. Leachate C oncentrations The DOC concent rations used here (162 to 710 mg C L1) appear to be within the range of field data when biases of collection methods and differences in litterfall quantity between systems are considered for field data. For example, carbon concentrations of field leachates of fresh litter in a spruce stand in Norway were 170 mg C L1 (Froberg et noted). Concentrations were 145 mg C L1 in a beech forest and 106 mg C L1 in a pine lters, 7 day sampling frequency, no microbial inhibitor mentioned). However, lower concentrations of 70 mg L1 were reported for a hardwood stand in Germany (Park and Matzner 2003, 0.45 ioned) and much lower values have been reported from an oak dominated forest in Appalachia (30 mg L1, Qualls et al. 1991, GF/F filter, usually sampled after storms, mercuric chloride added if not scheduled to be sampled immediately after storms). In temperate forests litter inputs are, in general, much lower than those in seasonal tropical forests (Vitousek 1984; Scott and Binkley 1997), which can directly affect carbon concentration of leaf litter leachate, because litter thickness can be related to leac hate concentration (Park
92 and Matzner 2003). In addition, when comparing among studies, field solutions collected on an infrequent basis may provide low estimates of DOC due to oxidation by microbes if microbial inhibitors were not included. Also, the proc ess of filtering, and the chosen pore size, can significantly reduce total DOC concentrations compared to the nonfiltered supernatant values reported in this study. Furthermore, when considering appropriate leachate concentrations for sorption reactions i n laboratory experiments, the soil to solution ratio is key. It is noted that this study used a ratio of 4.66 g soil to 10 ml for the sorption experiments, resulting in less leachate carbon per gram of soil compared to the commonly used ratio of 1 gram of soil to 10 ml solution. Scaling Phosphate Sorption R esults S patial and temporal scaling of results suggest t he small but statistically significant effects of litter (both field litter manipulation and leachate pulses) on phosphate sorption have potential biological relevance for soils with high phosphorus sorption capacity A simple scaling calculation for the effect of field litter manipulation treatments on phosphate sorption first requires an estimate of soil solution phosphate fluxes, which due to a num ber of methodological challenges are scarce for strongly weathered soils (Oberson and Joner 2005). As an estimate of phosphate inputs into soil solution, phosphate inputs from microbial mineralization and the contribution of water extractable soil phosphat e are considered. I use a soil organic P mineralization rate of 0.07 mg kg1 h1 from a temperate system (Oehl et al. 2001) and the water extractable phosphate in the litter addition soil, which was 0.06 mg kg1 (using a 1 hr extract with microbes inhibited). Next, the %32P sorbed during the 1 h incubations varies as a function of soil to solution ratio (Figure 46) so this calculation requires accounting for how solutionto soil ratios used in the lab compare to field ratios. Sorption experiments
93 in this s tudy w ere carried out using a 2.15 solutionto soil ratio (10 ml of solution to 4.66 g ODW soil) and the gravimetric moisture content of the soil at time of collection was ~60%, yielding a solutionto soil ratio of 0.6 Using the linear relationship between %32P sorbed versus solution to soil ratio (Figure 4 6), an estimate for %32P at the field ratio can be calculated for each soil (i.e., litter addition, litter removal and control) The estimated %32P remaining in solution can be used to calculate the quantity of phosphate remaining in solution at the end of one hour for the throughfall solution by considering a 10 cm depth and a bulk density of 1.0 g cm3 (bulk density is from Sayer et al. 2005). It follows that the s orption difference between the litter removal and litter addition soils is calculated as 6.2 g P ha1 hr1. Coarse scaling of this hour increment over a 24 h period results in estimates of 0.15 kg P ha1 remaining in solution longer i n the litter addition compared to the litter removal soil (b ut this is not a cumulative standing pool) Considering that the annual P requirements for aboveground biomass at these sites is ~8 kg P ha1 (based on calculations from Sayer et al. 2010), uptake on a daily basis is ~0.022 kg P ha1 day1. Therefore, an increase of 0.15 kg more phosphateP per ha in solution over the course of a day (although not all at the same time) suggests litter addition may have a substantial influence on P cycling and availability. Similar calculations could be done for the litter leachate pulses if the influence of solutionto soil ratio on leachate pulse was available for each leachate. However, the development of this relationship was not included in the study. Although calculations cannot be made, results can still be discussed in context to the calculations for the field litter manipulation treatments. At the 2.15 solutionto soil ratio used in the experiments, the difference in %32P sorbed between the litter removal soil and the litter addition soil
94 was 0.454 while the difference between throughfall and Castilla elastica leachate in the removal soil was 0.303 (and C elastica had the largest effect on decreasing phosphate sorption compared to the control). This suggests that if the effect of C elastica leachate on %32P sorptio n across solution to soil ratios followed a similar relationship to that of the litter addition treatment, scaling the effect of C elastica leachate pulse (at the concentration used in this study) on P remaining in solution on a hectare basis would be abo ut 30% less than that of the field litter manipulation treatments, meaning the leachate pulse effect could make a relevant contribution to phosphorus cycling. Scaling over 24 h would depend on both the rate of leachate inputs and the rate at which the leac hate solution was altered by microbial mineralization of carbon and immobilization of nutrients. Summary This study shows that litter inputs can influence both the magnitude and strength of phosphate sorption in a strongly weathered soil at two levels: long term litter inputs and leachate pulses. Leachate pulses had a greater influence on decreasing soil phosphate sorption in the field litter removal treatment, and little influence in field litter addition soil, showing that field litter manipulations and l eachate pulses interact in a predictable manner. In addition to decreasing the quantity of phosphate sorbed, this study provides evidence that litter inputs can decrease the strength of phosphate sorption. For field litter treatments this was demonstrated by a larger fraction of sorbed 32P that was Bray 1 extracted from the litter addition soil compared to the control and litter removal soil. The possibility that litter pulses of some species can also decrease the strength of phosphate sorption was suggest ed by the results of A. excelsum The results of A. excelsum in the litter addition soil were contrary to the other species, with
95 the A. excelsum leachate pulse showing increased phosphate sorption. A. excelsum had a higher index of aromatic carbon than th e other species and it is therefore hypothesized that A. excelsum may be promoting phosphate sorption through organic metal bonds. This study provides evidence for the role of litter inputs in decreasing soil phosphate sorption and increasing readily extractable phosphate. This work adds to the body of literature demonstrating that organic matter inputs can serve as not only input of nutrients to the soil environment but can also modify phosphorus cycling through indirect effects. While this work focuses on forest litter, results suggest that in low input agricultural systems, litter from crops and agroforestry trees (both longer term inputs and leachate pulses) could have similar effects on phosphorus cycling. The effects of crop residues on phosphate sorpt ion have received a great deal of attention in studies focused on high inputs of inorganic fertilizer (Ohno and Crannell 1996; Hunt et al. 2007) and the results presented here suggest organic matter inputs may play an important role in regulating plant phosphorus availability in low input systems on strongly weathered soils. This work also suggests that changes in forest litter production, which can occur during early forest succession (Ewel 1976) and which may occur as a response to climate change (Nepstad et al. 2004), could have potential feedbacks to phosphorus cycling that extend beyond the quantity of phosphorus returned through litterfall.
96 Table 41. Characteristics of leachate and simulated throughfall. Species A. gauchabele A. excelsum C. elast ica C. peltata F. insipida Throughfall pH 5.06 (0.04) 4.97 (0.07) 6.16 (0.02) 7.05 (0.76) 7.72 (0.11) 5.5 DOC (mg L 1 ) 240.1 (38.9) 1053.6 (76.9) 785.1 (29.4) 533.5 (38.2) 631.7 (51.7) 8.2 Phosphate P 4.80 (0.28) 6.43 (0.48) 6.13 (0.33) 2.95 (0.20) 0.81 (0.18) 0.03 Al 0.001 (0.001) 0.014 (0.013) bdl bdl 0.530 (0.121) Ca 8.35 (1.11) 7.15 (1.88) 127.30 (11.84) 82.51 (18.41) 128.24 (4.25) 0.4 Mg 8.60 (0.82) 112.78 (1.54) 91.63 (10.59) 36.83 (8.12) 65.39 (4.40) 0.24 K 178.53 (8.95) 120.81 (8.94) 381.91 (48.03) 487.93 (23.39) 173.20 (9.88) 1.99 Mn 0.31 (0.06) 0.09 (0.01) 0.17 (0.03) 0.03 (0.01) 0.02 (0.01) Na 0.43 (0.02) 4.04 (0.26) 3.84 (0.75) 2.17 (0.10) 3.92 (0.14) 0.46 Si 0.28 (0.05) 4.67 (0.02) 6.72 (0.68) 8.63 (2.55) 12.74 (0.73) All values, except pH, are mg L1. Phosphate P is reported as molybdate reactive phosphate (MRP). Means and standard deviations (in parentheses) are for the original leachate solutions made using a 1:30 litter to solution ratio and 4 h extract time (n=3, with the exception of throughfall). Leach ates and throughfall were then diluted with deionized water for the sorption experiments (6.75 to 6.50 ml of leachate in 10 ml total solution; in methods). Below detection limit is indicated by bdl.
97 Table 42. Characteristics of the Gigante Leaf Litter Ma nipulation project (GLiMP) soils (0 10 cm), which is referred to as 'field litter manipulation' treatments. Litter treatment Removal Control Addition %C* 2.8 (0.3) 3.2 (0.1) 3.8 (0.5) Total P (mg kg 1 soil) 262.3 (13.2) 285.6 (17.8) 266.4 (28.1 ) Resin P (mg kg 1 soil) 0.47 (0.16) 0.51 (0.14) 0.95 (0.02) pH (in water) 5.04 (0.15) 5.19 (0.31) 5.78 (0.01) Soil for this study was collected in October and November 2008 and %C was measured on the soils that were bulked and used in this study (indic ated by *). All other values are for soils collected in January 2009 from the same GLiMP plots used in this study (BL Turner, unpublished data). Means standard deviations are based on two plots per litter treatment.
98 Table 4 3. ANOVA and Tukey multiple c omparison results for the effect of solution (i.e., species leachate and throughfall control) on 32P phosphate sorption to soil. Soil litter treatment Removal Control Addition F value 113.7 18.9 73.0 P value <0.0001 <0.0001 <0.0001 % 32 P sorbed A. gauchabele 99.831 (0.014) 99.686 (0.074) 99.297 (0.031) A. excelsum 99.809 (0.012) 99.733 (0.027) 99.498 (0.013) C. elastica 99.605 (0.026) 99.491 (0.026) 99.131 (0.017) C. peltata 99.803 (0.019) 99.692 (0.032) 99.384 (0.009 ) F. insipida 99.789 (0.004) 99.685 (0.032) 99.364 (0.045) Throughfall 99.908 (0.013) 99.772 (0.010) 99.453 (0.026) Analyses were done within a soil treatment (df = 5, 12). Means of %32P sorbed to soil are included and standard deviations are s hown in parentheses. Asterisks indicate if the species leachate was significantly different from the throughfall control, as determined through the Tukey multiple comparison tests
99 Table 44. Species specific leachate effects on 32P phosphate sorption c ompared to a throughfall control. Soil Solution Diff % 32 P F value P value Cohen's d Litter removal A. gauchabele 0.049 16.0 0.016 3.60 A. excelsum 0.069 30.8 0.005 5.54 C. elastica 0.230 202.2 0.0001 11.15 C. peltata 0.095 53.9 0.002 5.86 F. insipida 0.115 108.8 0.0005 12.32 Control A. gauchabele 0.042 3.6 0.132 0.81 A. excelsum 0.003 0.0 0.899 0.14 C. elastica 0.235 167.4 0.0002 12.07 C. peltata 0.059 5.39 0.082 2.49 F. insipida 0.069 31.5 0.005 2.87 Litter addition A. gauchabele 0.065 5.6 0.076 2.29 A. excelsum 0.134 20.8 0.010 6.58 C. elastica 0.121 3.0 0.157 5.56 C. peltata 0.056 4.1 0.113 2.92 F. insipida 0.104 13.4 0.022 2.84 Dif ferences in the percent of 32P sorbed (diff %32P) were calculated as the mean %32P sorbed in the presence of species specific leachate minus that of throughfall at the y intercept (n=3). The P values indicate if leachate differed significantly from through fall. Degrees of freedom are 1,24 for all analyses. Effect size is reported as Cohen's d, which is the difference between the means (i.e., diff %32P) divided by the pooled standard deviation. The pooled standard deviation was estimated using the %32P sorbed for the samples that did not receive spikes of 31P phosphate.
100 Table 45. Concentration of treatment solutions used in sorption experiments and net DOC sorption to soil. Initial phosphateP in solution Initial DOC in solution Net DOC sorbed to so il (mg g 1 soil) 1 soil) (mg g 1 soil) Litter addition soil Control soil Litter removal soil Albizia gauchabele 6.95 (0.39) 0.36 (0.04) 0.21 (0.04) 0.18 (0.03) 0.10 (0.04) Anacardium excelsum 9.31 (0.69) 1.52 (0.11) 0.53 (0.16) 0.59 (0.11) 0.62 (0.07) Ca stilla elastica 8.88 (0.47) 1.14 (0.04) 0.27 (0.04) 0.36 (0.04) 0.38 (0.02) Cecropia peltata 4.27 (0.30) 0.77 (0.06) 0.07 (0.01) 0.00 (0.07) 0.11 (0.04) Ficus insipida 1.17 (0.26) 0.92 (0.07) 0.14 (0.05) 0.23 (0.06) 0.21 (0.07) Simulated thr oughfall 0.07 0.03 0.25 (0.01) 0.18 (0.01) 0.17 (0.02) Phosphate and dissolved organic carbon (DOC) are expressed on a per gram oven dried soil basis. Means and standard deviations (in parentheses) are based on n=3, with the exception of initial thr oughfall values. Net sorbed DOC is DOC initially in solution minus DOC remaining in supernatant at end of 1 h room temperature incubation.
101 Table 46. Simple linear regressions evaluating initial solution phosphate concentration, pH, ionic strength (based on Ca, K, Mg and Na) and net DOC sorbed as predictors of %32P sorbed to soil. Soil litter treatment Intercepts of %32P sorbed as a function of: Slope r 2 P value Addition Leachate pH 0.051 0.37 0.197 Ca, K, Mg, Na ionic strength 0.006 0.10 0.538 Net DOC sorbed 0.125 0.16 0.427 Control Leachate pH 0.024 0.09 0.553 Ca, K, Mg, Na ionic strength 0.013 0.51 0.112 Net DOC sorbed 0.077 0.07 0.601 Removal Leachate pH 0.028 0.16 0.433 Ca, K, Mg, Na ionic strength 0.010 0.82 0.01 3 Net DOC sorbed 0.141 0.29 0.267 The intercepts from the ANCOVA of %32P sorbed to soil, where initial phosphate in solution was a covariate, were used as the response variable. Net sorbed DOC is DOC initiailly in solution minus DOC remaining in supern atant at end of 1 h incubation. Significant correlations are in bold. All models have 4 degrees of freedom.
102 Figure 41. The effect of field litter additions on phosphate-32P so rption to soil during a 1 h incubation. Bars show standard deviations. Al l soil comparisons are significantly different from one another ( P <0.05) as evaluated using a oneway ANOVA and Tukey multiple comparisons of the samples that did not receive phosphate-31P spikes. Litter additio n soil: y = 99.479 0.036*x; control soil: y = 99.774 0.012*x ; litter remova l soil: y = 99.910 0.009*x
103 Figure 42. Leachate effects on r educing phosphate-32P so rption to soil during a 1 h incubation. See Table 4 for a summary of leachates that significantly influenced phosphate sorption compared to the throughfall control
104 Figure 43. Phosphate-32P in supernatant for leachate relative to the throughfall control, calculated as %32P in leachate supernatant/ %32P in throughfall supernatant.
105 Figure 44. Relationship between leachate effect on phosphate-32P sorption and net DOC sorbed to soil. Black symbols are for litter removal soil, grey symbols show control soil and open symbols represent litter addition soil.
106 Figure 45. The percent of phosphate-32P sorbed that was extractable using Bray 1. Bray 1 solution was added after a 1 h incubation wi th throughfall solution. Bars show standard deviation. Note the x axi s does not extend to zero. T tests of the intercepts showed the litter addition had significantly greater extractable 32P than the litter removal or control soil at the P = 0.05 level.
107 Figure 46. The effect of solution to soil ratios on phosphate sorption for the three field litter manipulation soils. All samples had 7.5 ml total solution and were shaken on high sett ing. Data points show two analytical replicates of each bulked soil. Litter addition soil: y = 100.07 0.186*x, r2=0. 996; control soil: y = 99.99 0.0548*x, r2=0.985; litter removal soil: y = 99.98 0.0261*x, r2=0.977. T tests showed slopes of all soi ls were significantly different at the p = 0.05 level.
108 Figure 47. SUVA wavelength scans of initial leachates (averaged from three replicates). Standard deviations are shown for SUVA at 254 nm.
109 CHAPTER 5 CONCLUSION The work presented in this dissert ation focused on 1) leaf litter nutrient solubility, including predictors of solubility and methodological considerations (Chapters 2 and 3) and 2) the influence of leachate from recently senesced litter on soil phosphate sorption via physiochemical mechanisms (Chapter 4). All work used litter from lowland tropical woody species. Here I provide concluding thoughts on issues not addressed in the main data chapters. I speculate on how the litter solubility results from this study may relate to other systems. In addition, there are a myriad of avenues through which leachate from recently senesced litter may influence ecosystem processes. This work experimentally investigated one avenue (i.e., leachate effects on soil phosphate sorption in a strongly weathered soil; Chapter 4); however, the data collected in this study speak to a number of these other roles, which deserve mention. Litter Solubility a mong Ecos ystems and Life History Strategies This study, which used recently senesced litter from woody species i n a lowland tropical forest, shows that litter elements show differential solubility and some elements (e.g., K, P and Na) are highly soluble, as evaluated through soluble fractions (Chapter 3). Although this study used a lab index and did not directly measure litter leaching in the field, the solubility results, in combination with knowledge of high precipitation in these systems, provide support for the role of leaching in nutrient release during the initial phase of decomposition. It is known that many l owland tropical forests receive high rainfall, either seasonally or throughout the year. Tropical dry forests receive >1 m of precipitation annually, while moist, wet and rain forests have progressively greater precipitation (Holdridge 1967).
110 The role of litter leaching in nutrient cycling, however, may not be limited to high rainfall systems. Leaching of elements from recently senesced litter could be an important phase in litter decomposition in any system that 1) has soluble litter and 2) either receiv es steady or pulsed precipitation, or experiences the accumulation of precipitation in gullies or depressions. Based on precipitation patterns, differential litter element solubility could play a role in nutrient cycling in deserts, where total precipitati on is low but may occur in pulsed events; tundra, where water pulses result from the melting of accumulated snowfall; and temperate forests and prairies, where litter similarly accumulates in the fall and can experience pulses of melting snow. In addition, topography and hindrances to water drainage could create localized areas of litter leaching within a site or landscape. Speculating on how species composition influences litter solubility across ecosystems, and variation within systems, is less straightf orward. At present we do not have an answer to the questions: For a given element, is the potential solubility of recently senesced litter from tundra vegetation similar to that of litter in a lowland tropical moist forest? Is the solubility of an annual herbaceous species similar to that of a woody perennial? One plausible hypothesis is that solubility of recently senesced litter is related to plant life history strategy. Variation in water adsorption by litter is one possible mechanism underlying the relationship. Life history strategies are linked to structural leaf traits, such as leaf toughness and lignin content, with higher structural investments in lower resource environments (e.g., light, water or nutrients; Coley 1983; Berendse 1994; Kitajima and P oorter 2010). In turn, leaf traits have been predicted to relate to rate of water absorption and maximum water holding capacity of litter (Taylor
111 and Parkinson 1988), which may then influence litter solubility (Ibrahima et al. 1995). For example, litter wi th higher lignin content or thicker cuticles from plants adapted to lower resource environments may have lower rates of water absorption or lower total water absorption, and subsequently lower soluble fractions (Figure 5 1). A relationship between life his tory strategy and litter element solubility may also be supported by differences in litter element concentrations. Litter element concentration can be related to life history strategy; for example, evergreen litter has been reported to have lower litter P than deciduous species (Killingbeck 1996). Furthermore, Chapter 3 determined that water soluble element on a mass basis was well correlated with litter element concentration for K, Na, P (Figure 31). Therefore, life history strategies with higher litter element concentrations may show higher solubility on a mass basis for a given element (Figure 51). This effect may be accentuated by differences in structural traits: as outlined above, a decreased investment in structural traits for plants adapted to higher resource environments may lead to increases in the soluble fraction. Litter in higher resource environments may therefore have higher soluble fraction due to structural traits, and higher water soluble element on a litter mass basis due to higher litter element concentrations (Figure 51). The hypothesis that litter element solubility is linked to life history strategy via structural litter traits has been little tested. The tests that do exist may not be robust for three main reasons: 1) studies to date focus on single systems and may not span the range of litter traits nor have sample sizes necessary for uncovering the proposed patterns; 2) litter solubility can be measured in many ways and, for understanding a link between litter traits and field solubility, rates of water soluble element on a litter basis
112 and the soluble fraction during short term extracts may be better metrics than maximum solubility; 3) elements vary in their solubility and studies considering only mass loss without investigating specific elements may be too coarse for determining a relationship. For example, in a study using Mediterranean vegetation, deciduous species showed higher water absorbance than evergreen species, but the overall mass loss did not vary among the life history strategies after a 10 day extract (Ibrahima et al. 1995). It would be interesting to know if the rate of water extracted K, Na and P varied among deciduous and evergreen species. In this study (Chapter 3) using 41 species there was little support for a rel ationship between element solubility and litter lignin content or leaf toughness for most of the elements investigated. The exception was a significant, although weak, negative relationship between soluble carbon fraction and lignin content (Figure 3 2), w hich does provide some support for a relationship between solubility and structural traits. Concerns 2 and 3 from above were addressed because the study evaluated initial, and not exhaustive, solubility of individual elements over a 4 h period (for many el ements 4 h is an index of initial solubility and does not represent exhaustion of soluble elements (Chapter 2); K is an exception). However, this study does not address Concern 3; it only considered litter from one site and focused on litter from woody species. A more robust evaluation of the potential link between litter traits and solubility requires the inclusion of palms and herbaceous vegetation from the same site, in addition to vegetation from other ecosystems. The focus of this study was leaf litter, but when considering of the role of litter solubility in decomposition across systems, the solubility of belowground litter should not be overlooked (Robinson et al. 1999). While fine root biomass often represents <
113 10% of total living biomass (Vogt et al. 1996), it can have turnover times > 2 times per year (Yavitt et al. 2011) and fine root production can be of a similar magnitude to aboveground litterfall (Nadelhoffer and Raich 1992). Furthermore, although the extent of nutrient retranslocation and resulting C:N:P ratios of senesced roots are not well known, a number of studies suggest minimal retranslocation during senescence (Nambiar 1987; Aerts 1990; Gordon and Jackson 2000). Therefore, a reported ratio of C:N:P 450:11:1 for live fine roots (<2 mm i n diameter; Jackson et al. 1997) may also serve as an estimate for the stoichiometry of senesced roots. Comparing this ratio to an average of 3144:45:1 (C:N:P) for leaf litter (Cleveland and Liptzin 2007) suggests root litter may have higher concentrations of N and P than leaf litter. Fine root litter therefore represents a large and concentrated source of nutrients and the role of leaching in decomposition of fine root litter could be significant. Other Roles of Leachate from Recently Senesced Litter in Nu trient C ycling Litter leachate chemistry varies with respect to litter decomposition. For example, leachate from relatively recently senesced litter and decomposing litter is known to vary in carbon chemistry (Hongve et al. 2000; Kalbitz et al. 2003). In a ddition, phosphorus in leachate from recently senesced litter is predominately phosphate (Chapter 2 and 3), while leachate from decomposing material may have a higher percentage of total phosphorus in organic form (Qualls et al. 1991). Despite known trends in leachate chemistries and a longstanding interest in leachate influences on ecosystem processes, there is still considerable debate about when and how leachate from the continuum of recently senesced to highly decomposed litter differs in influence on nutrient and carbon cycling (Kalbitz et al. 2000).
114 The effect of leachate on phosphate sorption, which was investigated in this study (Chapter 4), serves as one example where leachate from decomposing and recently senesced litter may have distinct influenc es on nutrient cycling. Carbon chemistry underpins the numerous competing hypotheses for the effect of DOC on phosphate sorption, and the majority of studies focus on leachate from decomposing litter or leachate from material harvested before senescence (G uppy et al. 2005; Hunt et al. 2007; Chapter 4). Chapter 4 demonstrates that leachate from recently senesced litter can reduce the magnitude and strength of phosphate sorption to strongly weathered soils. If decomposing litter had been used, however, leachate would have likely had higher concentrations of humic and fulvic acid compounds, which have been reported to increase phosphate sorption through organic carbonmetal phosphate bonds (and although sorbed, this phosphate is likely more accessible to plants and microbes than phosphate sorbed to metal oxides; Gerke 2010). Variation in leachate chemistry in water extracts along the continuum of decomposing and recently senesced litter, and the lack of distinction among these leachate sources, may be one reason studies investigating leachate effects on soil phosphate sorption show inconsistent results. While the primary focus of Chapter 4 was on phosphate sorption, the sorption of C in leachate from recently senesced litter was also measured and merits attention independent of phosphorus. Soil sorption of DOC is another example of the need to address the relationship among litter decomposition stage, leachate chemistry and ecosystem processes. The movement of DOC from the forest floor to mineral soil, and the subsequent sorption and turnover time are key to understanding forest carbon budgets (Kalbitz and Kaiser 2008). Carbon turnover times vary with carbon chemistry
115 due to affinity for soil sorption sites and bioavailability to soil microbes (Guggenberger and K aiser 2003). Therefore, differences in DOC chemistry can have significant impacts on carbon storage. Improving our understanding of the link between DOC source and function, and the relative contribution of different sources to DOC fluxes, has been highlig hted as a pressing research objective (Kalbitz et al. 2000). In this study, although the soluble fraction of C during a short extraction was low, the concentration of DOC in leachate solution can be high (Chapter 3). Furthermore, net DOC sorption to soil r epresented up to 30% of the initial leachate DOC during 1 h incubations with a microbial inhibitor (Table 45), although the averages vary greatly among species. The high flux of DOC from recently senesced litter, in addition to the high sorption and appar ent variation among species, argues for the need to consider leachate from recently senesced litter in order to better understand C dynamics in the soil profile. The difference in C bioavailability to microbes is yet another reason for careful consideration of leachate source. Carbon in leachates from recently senesced litter are often more bioavailable than that from decomposing material (Kalbitz et al. 2003) and soil microbes are generally limited by carbon (Zak et al. 1994). Leachates from recently senes ced litter may therefore influence nutrient cycling via microbial growth and respiration through a number of avenues. For example, pulses of C from recently senesced litter may increase the immobilization of phosphorus by soil microbes. In turn, by holding phosphorus in a form that is slowly turned over, microbes may in effect decrease the quantity of phosphorus that is sorbed to soil and act as a reservoir of biologically active phosphorus (Oberson et al. 2001). Studies investigating the partitioning of soil solution phosphorus between microbes and physiochemical soil
116 sorption with the addition of commercial carbon sources have been undertaken in tropical soils (Olander & Vitousek 2004), but the importance of leaf litter leachate on partitioning is an inter esting research direction. As another example, microbial activity stimulated by bioavailable C from recently senesced litter may create anaerobic conditions in soil microsites (Liptzin and Silver 2009). In anaerobic conditions microbial reduction of iron c an lead to the release of carbon and phosphorus from iron oxide sites (Chacon et al. 2006; Dubinsky et al. 2010), leading to increased bioavailability. In summary, this dissertation shows litter elements have differential solubility, and leaching has the potential to serve as a significant avenue for the release of nutrients from recently senesced litter. Furthermore, leachate from recently senesced litter can decrease the magnitude and strength of soil phosphate sorption in a strongly weathered soil, whic h may result in increased phosphate availability to plants and microbes. The results from this study advance our understanding of nutrient cycling in lowland tropical forests and, as outlined in the this chapter, serve as a framework and a motivation for investigating leaching of recently senesced litter in other systems.
117 Figure 51. Conceptual figure for how the two main metrics of litter solubility used in this study (i.e., water soluble element express ed on a litter mass basis and soluble fraction) may vary for species adapted to different resource (e.g., water, irradiance, nutrient) availability Water soluble element is shown on the y axis. The soluble fraction is calculated as water soluble element on a mass basis divided by total litter concentr ation of the respective element (shown on the x axis). If the linear regression is significant and the intercept of water soluble element as a function of total litter element is not significantly different than zero, the slope of the line indicates an average soluble fraction across the species considered A positive intercept would indicate the soluble fraction decreases with increased total element in litter. The 1:1 line shows a soluble fraction of 100 %. Predictions for solubility of species adapted t o a high er and low er resource environment are mapped onto this conceptual figure using relationship A and B, both of which have an intercept of zero. When comparing a lower and higher resource environment, an increase in solubility due to a change in stru ctural traits is hypothesized to result in an inc rease in the soluble fraction. If the change in the structural trait is relatively discrete and the species sampled do not span a large continuous range of the trait, this would result in a shift from B (low er resource environment) to A (higher resource environment) represented by the solid lines As another possibility, resources may influence solubility by increasing the element concentration in litter leading to increased water soluble element. This is r epresented by the dashed lines in A and B. As a final scenario, if there is a negative correlation between the structural trait and litter element concentration (i.e., increased lignin with decreased litter element), this may result in a regression with a negative intercept (not shown).
118 APPENDIX ALPHABETICAL LIST OF SPECIES USED IN CHAPTER 3. Species Code Lifeform Alchornea costaricensis Pax & K. Hoffm. ALCC understory Alseis blackiana Hemsl. ALSB understory Anacardium excelsum (Bertero & Balb. ex Kunt h) Skeels ANAE tree Apeiba membranacea Spruce ex Benth. APEM tree Arrabidaea verrucosa (Standl.) A.H. Gentry ARRV liana Astronium graveolens Jacq. ASTG tree Banisteriopsis cornifolia var maracaybensis (Kunth) C.B. Rob. BANC liana Beilschmiedia pendula (Sw.) Hemsl. BEIP tree Callichlamys latifolia (L. Rich.) K. Schum. CAL1 liana Combretum decandrum Jacq. COMD liana Cydista aequinoctialis var. aequinoctialis Seibert CYDA liana Dipteryx oleifera Benth. DIPP tree Doliocarpus olivaceus Sprague & R.O. W illiams ex Standl. DOLO liana Drypetes standleyi G.L. Webster DRYS tree Faramea occidentalis (L.) A. Rich. FARO understory Guapira standleyana Woodson GUAS tree Guarea guidonia (L.) Sleumer GUA2 midstory Guatteria dumetorum R.E. Fr. GUAD tree Heister ia concinna Standl. HEIC midstory Hiraea reclinata Jacq. HIR1 liana Hirtella triandra Sw. HIRT midstory Hieronyma alchorneoides Allemo HYEL tree Jacaranda copaia subsp spectabilis (Aubl.) D. Don JACC tree Lacmellea panamensis (Woodson) Markgr. LACP t ree Licania platypus (Hemsl.) Fritsch LICP tree Luehea seemannii Triana & Planch. LUE1 tree Maripa panamensis Hemsl. MARP liana Petrea volubilis L. PETA liana Pouteria reticulata subsp reticulata (Engl.) Eyma POUU tree Prionostemma aspera (Lam.) Mier s PRIA liana Protium tenuifolium Engl. PROT tree Pseudolmedia spuria (Sw.) Griseb. PSE2 midstory Quararibea asterolepis Pittier QUA1 tree Sloanea terniflora (Sess & Moc. ex DC.) Standl. SLOT tree Terminalia amazonia (J.F. Gmel.) Exell TERA tree Tetr agastris panamensis (Engl.) Kuntze TET2 tree Trichilia tuberculata (Triana & Planch.) C. DC. TRI3 tree Uncaria tomentosa (Willd. ex Roem & Schult.) DC. UNCT liana Virola sebifera Aubl. VIR1 tree Virola surinamensis (Rol. ex Rottb.) Warb. VIR2 tree Xyl opia macrantha Triana & Planch. XYLM midstory Notes: Data are from the Smithsonian Tropical Research Institute and include species name, the Wright Phenology code and lifeform ( http://striweb.si.edu/esp/tesp/plant_species.htm )
119 LIST OF REFERENCES Aerts R (1990) Nutrient use efficiency in evergreen and deciduous species from heathlands Oecologia 84:391397 AlarcnGutirrez E, Floch C, Ziarelli F, Augur C, Criquet S (2010) Drying rewetting cycles and [gamma] irradiation effects on enzyme activities of di stinct layers from a Quercus ilex L. litter. Soil Biol Biochem 42 : 283 290 Appelt H, Coleman NT, Pratt PF (1975) Interactions between organic compounds, minerals, and ions in volcanic ash derived soils 2. Effects of organic compounds on adsorption of phosphate. Soil Sci Soc Am J 39:628630 Arnold AE, Herre EA (2003) Canopy cover and leaf age affect colonization by tropical fungal endophytes: Ecological pattern and process in Theobroma cacao (Malvaceae). Mycologia 95: 388 398 Attiwill PM (1968) The loss of elements from decomposing litter. Ecology 49: 142145 Attiwill PM, Adams M (1993) Nutrient cycling in forests. New Phytol 124:561582 Austin A, Vitousek P (2000) Precipitation, decomposition and litter decomposability of Metrosideros polymorpha in native forests on Hawai'i. J Ecol 88:129 138 Ball BA, Bradford MA, Hunter MD (2009) Nitrogen and Phosphorus Release from Mixed Litter Layers is Lower than Predicted from Single Species Decay. Ecosystems 12:87 100 Bar Yosef B, Kafkafi U, Rosenberg R, Sposito G (1988) Phosphorus adsorption by kaolinite and montmorillonite: I. Effect of time, ionic strength, and pH. Soil Sci Soc Am J 52:15801585 Barrow NJ, Bowden JW, Posner AM, Quirk JP (1980) Describing the effects of electrolyte on adsorption of phosphate by a variabl e charge surface. Aust J Soil Res 18:395404 BastienHenri S, Park A, Ashton M, Messier C (2010) Biomass distribution among tropical tree species grown under differing regional climates. Forest Ecol Manag 260:403410. Berendse F (1994) Litter decomposabili ty --a neglected component of plant fitness J Ecol 82: 187 190 Berg B, Staaf H (1981) Leaching, accumulation and release of nitrogen in decomposing forest litter. Ecol Bull 33 :168 173 Berstein M, Carroll G (1977) Internal fungi in oldgrowth Douglas fir fol iage. Canadian Journal of Botany 55:6 44653
120 Bhatti J, Comerford N, Johnston C (1998) Influence of oxalate and soil organic matter on sorption and desorption of phosphate onto a spodic horizon. Soil Sci Soc Am J 62:10891095 Bieleski R (1973) Phosphate pool s, phosphate transport, and phosphate availability. Ann Rev Plant Physio 24:225 252 Binkley D, Giardina C (1998) Why do tree species affect soils? The warp and woof of tree soil interactions. Biogeochemistry 42:89106 Borggaard O, RabenLange B, Gimsing A, Strobel B (2005) Influence of humic substances on phosphate adsorption by aluminium and iron oxides. Geoderma 127:270279 Borie F, Zunino H (1983) Organic matter phosphorus associations as a sink in P fixation processes in allophanic soils of Chile. Soil Biol Biochem 15:599603 Bowman WD, Steltzer H, Rosenstiel TN, Cleveland CC and Meier CL (2004) Litter effects of two co occurring alpine species on plant growth, microbial activity and immobilization of nitrogen. Oikos 104 :336344 Bray RH, Kurtz LT (1945) Determination of total, organic, and available forms of phosphorus in soils. Soil Sci 59:3945 Brinson, M (1977) Decomposition and nutrient exchange of litter in an alluvial swamp forest. Ecology 58:601609 Briones M, Ineson P (1996) Decomposition of eucal yptus leaves in litter mixtures. Soil Biol Biochem 28 :13811388 Brooks P, Williams M, Schmidt S (1998) Inorganic nitrogen and microbial biomass dynamics before and during spring snowmelt. Biogeochemistry 43:1 15 Chacon N, Silver WL, Dubinsky EA, Cusack DF (2006) Iron reduction and soil phosphorus solubilization in humid tropical forests soils: The roles of labile carbon pools and an electron shuttle compound. Biogeochemistry 78:6784 Chapin FS III, Barsdate R, Barel D (1978) Phosphorus cycling in Alaskan co astal tundra: a hypothesis for the regulation of nutrient cycling. Oikos 31:189 199 Chen Y, Butler J, Stumm W (1973) Adsorption of phosphate on alumina and kaolinite from dilute aqueous solutions. J C olloid Interf Sci 43:421436 Cleveland CC, Neff JC, Town send AR, Hood E ( 2004) Composition, dynamics, and fate of leached dissolved organic matter in terrestrial ecosystems: Results from a decomposition experiment. Ecosystems 7: 275 285
121 Cleveland CC, Reed S, Townsend AR (2006) Nutrient regulation of organic matt er decomposition in a tropical rain forest. Ecology 87:492 503 Cleveland CC, Liptzin D (2007) C: N: P stoichiometry in soil: is there a Redfield ratio for the microbial biomass? Biogeochemistry 85:235 252 Coley PD (1983) Herbivory and defensive character istics of tree species in a lowland tropical forest Ecol Monogr 53:209233 Cornelissen JHC ( 1996 ) An experimental comparison of leaf decomposition rates in a wide range of temperate plant species and types. J Ecol 84: 573 582 Crews TE, Kitayama K, Fownes J H, Riley RH, Herbert DA, Mueller Dombois D, Vitousek PM (1995) Changes in soil phosphorus fractions and ecosystem dynamics across a long chronosequence in Hawaii. Ecology 76:14071424 Croat T (1978) Flora of Barro Colorado Island. Stanford University Press Stanford, CA. Cusack D, Chou W, Yang W, Harmon ME, Silver WL (2009) Controls on long term root and leaf litter decomposition in neotropical forests. Glob Change Biol 15:13391355 Dalton J, Russell G, Sieling D (1952) Effect of organic matter on phosphat e availability. Soil Science 73:173181 De Mesquita MV, Torrent J (1993) Phosphate sorption as related to mineralogy of a hydrosequence of soils from the cerrado region (Brazil). Geoderma 58:107123 Don A, Kalbitz K ( 2005 ) Amounts and degradability of diss olved organic carbon from foliar litter at different decomposition stages. Soil Biol Biochem 37: 21712179 Duarte S, Pascoal C, Alves A, Correia A, Cssio F (2010) Assessing the dynamic of microbial communities during leaf decomposition in a low order strea m by microscopic and molecular techniques. Microbiol Res 165: 351362 Dubinsky E, Silver WL, Firestone MK (2010) Tropical forest soil microbial communities couple iron and carbon biogeochemistry Ecology 91:26042612 Easterwood G, Sartain J (1990) Clover res idue effectiveness in reducing orthophosphate sorption on ferric hydroxide coated soil. Soil Sci Soc Am J 54:13451350 Edzwald JK, Toensing DC, Leung MCY (1976) Phosphate adsorption reactions with clay minerals. Environ Sci Technol 10:485490 El Hajj Z, Ka vanagh K, Rose C, KanaanAtallah Z. (2004) Nitrogen and carbon dynamics of a foliar biotrophic fungal parasite in fertilized Douglas fir. New Phytol 163: 139147
122 English, P., Maglothin, A., Keegstra, K., Albersheim, P., 1972. A cell wall degrading endopoly galacturonase secreted by Colletotrichum lindemuthianum Plant Physiology 49, 293. Ewel J (1976) Litter fall and leaf decomposition in a tropical forest succession in eastern Guatemala. J Ecol 64: 293308 Fierer N, Schimel JP (2002) Effects of drying rewet ting frequency on soil carbon and nitrogen transformations. Soil Biol Biochem 34 :777 787 Fioretto A, Papa S, Curcio E, Sorrentino G, Fuggi A ( 2000) Enzyme dynamics on decomposing leaf litter of Cistus incanus and Myrtus communis in a Mediterranean ecosyste m. Soil Biol Biochem 32: 1847 1855 Fioretto A, Papa S, Sorrentino G, Fuggi A (2001) Decomposition of Cistus incanus leaf litter in a Mediterranean maquis ecosystem: mass loss, microbial enzyme activities and nutrient changes. Soil Biol Biochem 33: 311 321 Fi sher, A. M. 2007. Nutrient remobilization during leaf senescence. In: S. Gan (ed) Annual Plant Reviews Vol. 26. Blackwell Publishing Oxford UK. pp 87107 Fontaine S, Mariotti A, Abbadie L (2003) The priming effect of organic matter: a question of microbial competition? Soil Biol Biochem 35 :837 843 Fontes MPF, Weed SB (1996) Phosphate adsorption by clays from Brazilian Oxisols, relationships with specific surface area and mineralogy. Geoderma 72:3751 Fox RL, Searle P (1978) Phosphate adsorption by soils of the tropics. In: Stelly M (ed) Diversity of soils in the tropics. ASA Special Publication no. 34. American Society of Agronomy and Soil Science Society of America, Madison, WI. pp 97119 Froberg M, Kleja DB, Hagedorn F (2007) The contribution of fresh litt er to dissolved organic carbon leached from a coniferous forest floor. Eur J Soil Sci 58:108114 Gallardo A, Merino J (1993) Leaf decomposition in 2 Mediterranean ecosystems of southwest Spain influence of substrate quality. Ecology 74: 152161 Gartner TB Cardon ZG ( 2004) Decomposition dynamics in mixedspecies leaf litter. Oikos 104:230246. Gerke J (2010) Humic (Organic Matter) Al (Fe)Phosphate Complexes: An Underestimated Phosphate Form in Soils and Source of Plant Available Phosphate. Soil Sci 175:41 7 425 Goldberg S, Sposito G (1984) A chemical model of phosphate adsorption by soils: I. Reference oxide minerals. Soil Sci Soc Am J 48:772778
123 Gordon WS, Jackson RB (2000) Nutrient concentrations in fine roots Ecology 81: 275 280 Gosz JR, Likens GE, Borma nn FH (1973) Nutrient release from decomposing leaf and branch litter in the Hubbard Brook Forest, New Hampshire. Ecol Monogr 43: 173 191 Guggenberger G, Kaiser K (2003) Dissolved organic matter in soil: challenging the paradigm of sorptive preservation. Ge oderma 113: 293310 Guppy C, Menzies N, Moody P, Blamey F (2005) Competitive sorption reactions between phosphorus and organic matter in soil: a review. Aust J Soil Res 43:189202 Gustafson F (1943) Decomposition of the leaves of some forest trees under field conditions. Plant Physiol 18:704707 Harter RD (1969) Phosphorus adsorption sites in soils. Soil Sci Soc Am Proc 33:630631 Hautala K, Peuravuori J, Pih laja K (2000) Measurement of aquatic humus content by spectroscopic analyses. Water Res 34:246258 H ttenschwiler S, Jrgensen HB (2010) Carbon quality rather than stoichiometry controls litter decomposition in a tropical rain forest. J Ecol 98:754 763 He LM, Zelazny LW, Baligar VC, Ritchey KD, Martens DC (1997) Ionic strength effects on sulfate and phosphate adsorption on gammaalumina and kaolinite: Triple layer model. Soil Sci Soc Am 61:784793 Hedin L, Brookshire E, Men ge D, Barron A (2009) The nitrogen paradox in tropical forest ecosystems. Ann Rev Ecol Evol Syst 40:613635 Hingston FJ, Atkinson R, Posner AM, Quirk JP (1967) Specific adsorption of anions. Nature 215:14591461 Hobbie S, Vitousek P (2000) Nutrient limitation of decomposition in Hawaiian forests. Ecology 81:18671877 Holdridge LR (1967) Life zone ecology. Tropical Science Center. San Jose, Costa Rica 206 pp. Hongve D, Van Hees P, Lundstrm U (2008) Dissolved components in precipitation water percolated through forest litter. Eur J Soil Sci 51:667677 Hrtensteiner S, Feller U (2002) Nitrogen metabolism and remobilization during senescence. J Exp Bot 53:927937
124 Hunt J, Ohno T, He Z, Honeycutt C, Dail D (2007) Inhibition of phosphorus sorption to goethite, gibbsite, and kaolin by fresh and decomposed organic matter. Biol Fert Soils 44:277288 Hur J, Schlautman MA (2003) Using selected oper ational descriptors to examine the heterogeneity within a bulk humic substance. Envir Sci Tech 37 :880 887 Jackson RB, Mooney HA, Schulze ED (1997) A global budget for fine root biomass, surface area, and nutrient contents 94:73627366 Jaffrain J, Gerard F Meyer M, Ranger J (2007) Assessing the quality of dissolved organic matter in forest soils using ultraviolet absorption spectrophotometry. Soil Sci Soc Am J 71:18511858 Jenny H (1941) Factors of soil formation. Dover Publications. New York. 146 pp. Johnson A, Frizano J, Vann D (2003) Biogeochemical implications of labile phosphorus in forest soils determined by the Hedley fractionation procedure. Oecologia 135:487499 Ibrahima A, Joffre R, Gillon D ( 1995) Changes in litter during the initial leaching phase an experiment on the leaf litter of Mediterranean species. Soil Biol Biochem 27: 931 939 Iyamuremye F, Dick R, Baham J (1996) Organic amendments and phosphorus dynamics: II. Distribution of soil phosphorus fractions. Soil Sci 161:436443 Kafkafi U, Bar Yosef B, Rosenberg R, Sposito G (1988) Phosphorus adsorption by kaolinite and montmorillonite: II. Organic anion competition. Soil Sci Soc Am J 52:15851589 Kaiser K, Kaupenjohann M, Zech W (2001) Sorption of dissolved organic carbon in soils: effects of soil sample storage, soil to solution ratio, and temperature. Geoderma 99:317328 Kalbitz K, Solinger S, Park JH, Michalzik B, Matzner E (2000) Controls on the dynamics of dissolved organic matter in soils: a review Soil Sci 165:277304 Kalbitz K, Schmerwi tz J, Schwesig D, Matzner E ( 2003) Biodegradation of soil derived dissolved organic matter as related to its properties. Geoderma 113: 273291 Kalbitz K, Kaiser K (2008) Contribution of dissolved organic matter to carbon storage in forest mineral soils J P lant Nutr Soil Sc 171:5260 Kaspari M, Yanoviak S, Dudley R, Yuan M, Clay N (2009) Sodium shortage as a constraint on the carbon cycle in an inland tropical rainforest. P Natal Acad Sci 106: 1940519415
125 Kieft TL, Soroker E, Firestone MK ( 1987) Microbial bio mass response to a rapid increase in water potential when dry soil is rewetted. Soil Biol Biochem 19:119126 Killingbeck KT (1996) Nutrients in senesced leaves: keys to the search for potential resorption and resorption proficiency Ecology 77:17161727 Ki tajima K, Mulkey S Wright S (1997) Seasonal leaf phenotypes in the canopy of a tropical dry forest: photosynthetic characteristics and associated traits. Oecologia 109: 490498 Kitajima K, Poorter L ( 2010) Tissue level leaf toughness, but not lamina thickn ess, predicts sapling leaf lifespan and shade tolerance of tropical tree species. New Phyto l 186 :708 721 Lauer M, Blevins D, SierzputowskaGracz H (1989) 31P nuclear magnetic resonance determination of phosphate compartmentation in leaves of reproductive s oybeans (Glycine max L.) as affected by phosphate nutrition. Plant Physiol 89 :13311336 Leigh R, Jones W (1986) Cellular compartmentation in plant nutrition: the selective cytoplasm and the promiscuous vacuole. Pages 249279 in B. Tinker and A. Lauchli edi tors Advances in plant nutrition. Vol. 2.. Praeger, New York. New York, USA. Levesque M, Schnitze.M (1967) Organometallic interactions in soils 6. Preparation and properties of fulvic acidmetal phosphates. Soil Sci 103:183190 Lim P, Kim H, Nam HG (2007) Leaf senescence. Annu Rev Plant Bio 58:115136 Liptzin D, Silver WL (2009) Effects of carbon additions on iron reduction and phosphorus availability in a humid tropical forest soil Soil Biol Biochem 41:16961702 Lopez Hernandez D, Rodriguez G (1986) Competitive Adsorption of Phosphate with Malate and Oxalate by Tropical Soils1. Soil Sci Soc Am J 50:14601462 Lousier J, Parkinson D ( 1978) Chemical element dynamics in decomposing leaf litter. Can J Botany 56:27952812 Lucas PW, Turner IM, Dominy NJ, Yamashi ta N (2000) Mechanical defences to herbivory. Ann Bot London 86:913920 Mack MC, DAntonio CM (2003) The effects of exotic grasses on litter decomposition in a Hawaiian woodland: the importance of indirect effects. Ecosystems 6:723738 Magill AH, Aber JD ( 2000) Dissolved organic carbon and nitrogen relationships in forest litter as affected by nitrogen deposition. Soil Biol Biochem 32: 603613 Mattingly G (1975) Labile phosphate in soils. Soil Sci 119 :369 375
126 McBride M (1994) Enviromental chemistry of soils. Oxford University Press, New York. McBride M (1997) A critique of diffuse double layer models applied to colloid and surface chemistry. Clays Clay Miner 45:598608 McDowell WH, Fisher SG (1976) Autumnal processing of dissolved organic matter in a small wo odland stream ecosystem. Ecology 57: 561 569 McGroddy M, Silver W, Oliveira R (2004) The effect of phosphorus availability on decomposition dynamics in a seasonal lowland Amazonian forest. Ecosystems 7 :172 179 Melillo J, Aber JD, Muratore J (1982) Nitrogen and lignin control of hardwood leaf litter decomposition dynamics. Ecology 63: 621 626 Melin E (1930) Biological decomposition of some types of litter from North American forests. Ecology 11:72 101 Meentemeyer V (1978) Macroclimate and lignin control of lit ter decomposition rates. Ecology 59: 465 472 Meyer JL, Wallace JB, Eggert SL (1998) Leaf litter as a source of dissolved organic carbon in streams. Ecosystems 1: 240 249 Moshi A, Wild A, Greenland D (1974) Effect of organic matter on the charge and phosphate adsorption characteristics of Kikuyu red clay from Kenya. Geoderma 11:275285 Murrmann RP, Peech M (1969) Effect of ph on labile and soluble phosphate in soils. Soil Sci Soc Am Pro 33:205210 Nadelhoffer KJ, Raich JW (1992) Fine root production estimates and belowground carbon allocation in forest ecosystems 73:1139 1147 Nagarajah S, Am Posner J (1970) Competitive adsorption of phosphate with polygalacturonate and other organic anions on kaolinite and oxide surfaces. Nature 228:8385 Nambiar EKS (1987) Do nutrients retranslocate from fine roots? Can J Forest Res 17:913918 Negassa W, Dultz S, Schlichting A, Leinweber P (2008) Influence of specific organic compounds on phosphorus sorption and distribution in a tropical soil. Soil Sci 173:587601
127 Nepstad D, L efebvre P, Lopes da Silva U, Tomasella J, Schlesinger P, Solorzano L, Moutinho P, Ray D, Guerreira Benito J (2004) Amazon drought and its implications for forest flammability and tree growth: A basin wide analysis. Glob Change Biol 10:704717 Nykvist N (19 62) Leaching and decomposition of litter V. Experiments on leaf litter of Alnus glutinosa, Fagus silvatica and Quercus robur Oikos 13: 232 248 Nykvist N (1963) Leaching and decomposition of water soluble organic substances from different types of leaf and needle litter. Studia Forestalia Suecica 3: 1 31 Oberson A, Joner E (2010) Microbial turnover of phosphorus in soil. Pages 133163. in B.L. Turner, E. Frossard, DS Baldwin. Organic phosphorus in the environment edit ors. CABI Pub, Wallingford, UK. Ohno T, Cr annell B (1996) Green and animal manurederived dissolved organic matter effects on phosphorus sorption. J Environ Qual 25:11371143 Olander LP, Vitousek PM (2004) Biological and geochemical sinks for phosphorus in soil from a wet tropical forest Ecosyste ms 7:404419 Ostertag R (2010) Foliar nitrogen and phosphorus accumulation responses after fertilization: an example from nutrient limited Haw aiian forests. Plant Soil 334:8598 Pardo MT, Guadalix ME, Garciagonzalez MT (1992) Effect of pH and background el ectrolyte on p sorption by variable charge soils. Geoderma 54:275 284 Park JH, Matzner E (2003) Controls on the release of dissolved organic carbon and nitrogen from a deciduous forest floor investigated by manipulations of aboveground litter inputs and water flux. Biogeochemistry 66:265286 Parker G, Smith A, Hogan K (1992) Access to the upper forest canopy with a large tower crane. BioScience 42: 664 670 Parsons WFJ, Taylor BR, Parkinson D (1990) Decomposition of aspen ( Populus tremuloides ) leaf litter mod ified by leaching. Can J Forest Res 20: 943 951 Prez Amador M, Abler M, De Rocher E, Thompson D, Van Hoof A, LeBrasseur N, Lers A, Green P ( 2000) Identification of BFN1, a bifunctional nuclease induced during leaf and stem senescence in Arabidopsis. Plant Physiol 122 :169 179 Perrott K (1978) The influence of organic matter extracted from humified clover on the properties of amorphous aluminosilicates. I. Surface charge. Aust J Soil Res 16:327339 Polglase PJ, Comerford NB, Jokela EJ (1992) Leaching of inorg anic phosphorus from litter of southern pine plantations. Soil Sci Soc Am J 56:573577
128 Powers J, Montgomery R, Adair E, Brearley F, Dewalt S, Castanho C, Chave J, Deinert E, Ganzhorn J, Gilbert M, Gonzlez Iturbe J, Bunyavejchewin S, Grau H, Harms KE, Hire math A, Iriarte Vivar S, Manzane E, De Oliveira A, Poorter L, Ramanamanjato J, Salk C, Varela A, Weiblen G, Lerdau M (2009) Decomposition in tropical forests: a pantropical study of the effects of litter type, litter placement and mesofaunal exclusion acr oss a precipitation gradient. J Ecology 97: 801 811 Prescott C, Taylor B, Parsons W, Durall D, Parkinson D ( 1993) Nutrient release from decomposing litter in Rocky Mountain coniferous forests: Influence of nutrient availability. Can J Forest Res 23 :15761586 Qiu S, McComb A, Bell R, Davis J ( 2005) Leaf litter application to a sandy soil modifies phosphorus leaching over the wet season of southwestern Australia. Hydrobiologia 545: 33 43 Qualls RG, Haines BL (1991) Fluxes of dissolved organic nutrients and humi c substances in a deciduous forest. Ecology 72: 254266 R Development Core Team. 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing Vienna, Austria, ISBN 390005107 0, URL http://www.R project.org Redfield A (1958) The biological control of chemical factors in the environment. Am Sci 46:205 221 Robinson CH, Kirkham JB, Littlewood R (1999) Decomposition of root mixtures from high arctic plants: a microcosm study Soil Biol Biochem 31: 11011108 Rodrigues KF (1994) The foliar fungal endophytes of the Amazonian palm Euterpe oleracea Mycologia 86: 376 385 Ryden J, Syers J, McLaughlin J (1977) Effects of ionic strength on chemisorption and potential determining sorption of phosphate by soils. Eur J Soil Sci 28:6271 Sayer E, Tanner E, Lacey A (2006) Effects of litter manipulation on early stage decomposition and mesoarthropod abundance in a tropical moist forest. Forest Ecol Manag 229:285293 Sa yer EJ, Tanner EVJ (2010) Exp erimental investigation of the importance of litterfall in lowland semi evergreen tropical forest nutrient cycling. J Ecol 98:1052 1062 Schoenau J, Bettany J ( 1987) Organic matter leaching as a component of carbon, nitrogen, phosphorus, and sulfur cycles i n a forest, grassland, and gleyed soil. Soil Sci Soc Am J. 51: 646 651 Schuur E (2001) The effect of water on decomposition dynamics in mesic to wet Hawaiian montane forests. Ecosystems 4 :259 273
129 Scott N, Binkley D (1997) Foliage litter quality and annual net N mineralization: comparison across North American forest sites. Oecologia 111:151 159 Sharpley AN, Tiessen H, Cole CV (1987) Soil phosphorus forms extracted by soil tests as a function of pedogenesis Soil Sci Soc Am J 51: 362365 Sharpley AN, Smith S (1989) Mineralization and leaching of phosphorus from soil incubated with surface applied and incorporated crop residue. 18:101105 Sibanda H, Young S (1989) The effect of humus acids and soil heating on the availability of phosphate in oxiderich tropical soils. In: Proctor J (ed) Mineral Nutrients in Tropical Forest and Savanna Ecosystems. British Ecological Society, Special publication no. 9. Blackwell Scientific, Oxford. pp 7183 Sinclair T, Vadez V (2002) Physiological traits for crop yield improvement in low N and P environments. Plant Soil 245 : 1 15 Sparks DL (1995) Environmental soil chemistry. Academic Press, San Diego, California. 267 pp. Staaf H, Berg B (1982) Accumulation and release of plant nutrients in decomposing Scots pine needle litter. Long term decomposition in a Scots pine forest II. Can J Botany 60:15611568 Subbarao G, Ito O, Berry W, Wheeler R (2003) Sodium A functional plant nutrient. Cr Rev Plant Sci 22 : 391 416 Swenson RM, Cole CV, Sieling DH (1949) Fixation of phosphate by iron and aluminum and replacement by organic and inorganic ions. Soil Science 67:322 Swift M, RussellSmith A, Perfect T (1981) Decomposition and mineral nutrient dynamics of plant litter in a regenerating bush fallow in subhumid tropical Nigeria. J Eco l 69 : 981 995 Taylor BR, Parkinson D (1988) Patterns of water absorption and leaching in pine and aspen leaf litter. Soil Biol Biochem 20:257258 Taylor C, Bariola P, Delcardayre S, Raines R, Green P ( 1993) RNS2: a senescenceassociated RNase of Arabidopsis that div erged from the S RNases before speciation. P Natl Acad Sci USA 90: 51185122 Taylor BR, Brlocher F ( 1996) Variable effects of air drying on leaching losses from tree leaf litter. Hydrobiologia 325: 273 182 Thomas H, Stoddart J ( 1980) Leaf senescence. Annu R ev Plant Physiol 31: 83111 Thompson J, Taylor C, Wang T ( 2000 ) Altered membrane lipase expression delays leaf senescence. Biochem Soc T 28 : 775 777
130 Thurman EM ( 1985) Organic geochemistry of natural waters. Kluwer Academic Publishers. Hingham, MA, USA. Tripl er C, Kaushal S, Likens GE, Walter M (2006) Patterns in potassium dynamics in forest ecosystems. Ecol Lett 9 :451 466 Tukey HB ( 1970) The leaching of substances from plants. Annual Review of Plant Physiology 21: 305 324 Turner BL, Haygarth PM ( 2001) Biogeoch emistry Phosphorus solubilization in rewetted soils. Nature 411: 258258 Uehara G, Gillman GP (1981) The mineralogy, chemistry, and physics of tropical soils with variable charge clays. Westview Press, Boulder, Colorado. 170 pp. Van Breemen N (1993) Soils as biotic constructs favouring net primary productivity. Geoderma 57:183211 Van Riemsdijk WH, Boumans LJM, Dehaan FAM (1984) Phosphate sorption by soils 1. A model for phosphate reaction with metal oxides in soil. Soil Sci Soc Am J 48:537541 Vicre M, Fa rrant JM, Driouich A ( 2004) Insights into the cellular mechanisms of desiccation tolerance among angiosperm resurrection plant species. Plant Cell E nviron 27: 1329 1340 Vitousek PM (1984) Litterfall, nutrient cycling, and nutrient limitation in tropical for ests. Ecology 65: 285 298 Vitousek PM (1998) Foliar and litter nutrients, nutrient resorption and decomposition in Hawaiian Metrosideros polymorpha. Ecosystems 1:401407 Vitousek PM, Hobbie S (2000) Heterotrophic nitrogen fixation in decomposing litter: Pat terns and regulation. Ecology 81:23662376 Vogt KA, Vogt DJ, Palmiotto PA, Boon P, OHara J, Asbjornsen H (1996) Review of root dynamics in forest ecosystems grouped by climate, climatic forest type and species Plant Soil 187:159219 Walker T, Syers J (19 76) Fate of phosphorus during pedogenesis. Geoderma 15:119 Wetzel RG ( 1992) Gradient dominated ecosystems sources and regulatory functions of dissolved organic matter in freshwater ecosystems. Hydrobiologia 229: 181 198 White R, Taylor A (1977) Effect o f pH on phosphate adsorption and isotopic exchange in acid soils at low and high additions of soluble phosphate. Eur J Soil Sci 28:4861
131 Wieder R, Wright S ( 1995) Tropical forest litter dynamics and dry season irrigation on Barro Colorado Island, Panama. E cology 76: 19711979 Wieder WR, Cleveland CC, Townsend AR ( 2008) Tropical tree species composition affects the oxidation of dissolved organic matter from litter. Biogeochemistry 88: 127 138 Wieder W, Cleveland CC, Townsend AR ( 2009 ) Controls over leaf litter decomposition in wet tropical forests. Ecology 90: 33333341 Worsfold P, Gimbert L, Mankasingh U, Omaka O, Hanrahan G, Gardolinski P, Haygarth P M, Turner BL, KeithRoach M, McKelvie I ( 2005) Sampling, sample treatment and quality assurance issues for the determination of phosphorus species in natural waters and soils. Talanta 66: 273 293 Wright S, Cornejo F (1990) Seasonal drought and leaf fall in a tropical forest. Ecology 71:11651175 Wright S, Yavitt JB, Wurzburger N, Turner BL, Tanner E, Sayer E, Santiago L, Kaspari M, Hedin L, Harms K, Garcia M, Corre M in press Potassium, phosphorus or nitrogen limit root allocation, tree growth and litter production in a lowland tropical forest. Ecology Yavitt JB, Fahey TJ (1986) Litter decay and leaching from the f orest floor in Pinus contorta (Lodgepole pine) ecosystems. J Ecol 74: 525 545 Yavitt JB, Harms KE, Garcia MN, Mirabello MJ, Wright JS (2011) Soil fertility and fine root dynamics in response to 4 years of nutrient (N, P, K) fertilization in a lowland tropic al moist forest, Panama. Austral Ecol 36:433445 Yuan T (1980) Adsorption of phosphate and water extractable soil organic material by synthetic aluminum silicates and acid soils. Soil Sci. Soc. Am. J 44:951 955 Zak DR, Tilman D, Parmenter RR, Rice CW, Fish er FM, Vose J, Milchunas D, Martin CW (1994) Plant production and soil microorganisms in latesuccessional ecosystems: a continental scale study Ecology 75:23332347 Zimmermann A, Wilcke W, Elsenbeer H (2007) Spatial and temporal patterns of throughfall q uantity and quality in a tropical montane forest in Ecuador. J Hydrol 343:8096
132 BIOGRAPHICAL SKETCH Laura holds a BS in chemistry from Saint Marys College, Notre Dame, Indiana and an MS granted through the Ecology, Evolutionary Biology and Behavior prog ram at Michigan State University. Between her MS and her PhD she spent two years as a Peace Corps volunteer in the Andes of Ecuador where she worked as an agroforestry extensionist and small business consultant to local womens groups. She also worked as a biological technician on a marsh restoration project with the National Park Service and served as a proctor for the Research Experiences for Undergraduates program at Harvard Forest. While completing her dissertation, Laura was a coinstructor for Gener al Ecology, a teaching assistant for the course Tropical Cropping Systems, a research assistant on a biodiversity and ecosystem function project and a science tutor for UF athletes.