a. .2 :x . v . qhzvsdpbt. 1- a») 2.5.223: Is... .1 v3.3a .21 13:... .. . . < ),7~1r(~{ It! .3153 . i I! a; meet- (.1 . LIBRARY may Michigan State Umversity This is to certify that the dissertation entitled ROOT CARBOHYDRATE STORAGE IN TEMPERATE AND TROPICAL FOREST TREE SEEDLINGS: IMPLICATIONS FOR SPECIES COEXISTENCE presented by MEERA IYER has been accepted towards fulfillment of the requirements for the PhD. degree in Forestry Major Professor’s Signature 2'7 Ago/l I 2005 Date MS U is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN Box to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 2/05 p:lC|RC/DaleDue.indd-p.1 ROOT CARBOHYDRATE STORAGE IN TEMPERATE AND TROPICAL FOREST TREE SEEDLINGS: IMPLICATIONS FOR SPECIES COEXISTENCE By Meera Iyer A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Forestry 2006 ABSTRACT ROOT CARBOHYDRATE STORAGE IN TEMPERATE AND TROPICAL FOREST TREE SEEDLINGS: IMPLICATIONS FOR SPECIES COEXISTENCE By Meera Iyer With the goal of better understanding the functional significance of intra- and inter- specific variation in plant resource allocation, its implications for species growth and survival and hence for community composition, I carried out experiments with tree seedlings from both temperate and tropical forest communities. In Michigan, seedlings of eight common northern temperate species were grown in a greenhouse under two levels of light and nitrogen availability, spanning the range of variation found in northern Michigan forests. In the tropics, I transplanted ~ 3,000 seedlings of five common dry tropical forest tree species into natural gradients of light and soil phosphorus (P) availability in Palo Verde National Park, Costa Rica. Light availability in the plots ranged from ~4% to ~40% full sun, and phosphorus availability, from 15% full sun) ....................................... 88 Figure 4.4 Trade-off between growth under high light availability (30% full sun) and survival under low light availability (5% full sun) as influenced by soil resources. . . . . ...90 xii Figure 4.5 Annual diameter growth of seedlings, given survival, across light levels. . ....93 Figure 5.1 Mean (+ SE) concentrations (mg g") of starch, simple sugars, and TNC in roots of the study species ......................................................................... 109 Figure 5.2 Relationship between root TNC concentration (mg g") and resource availability in four harvests of Dalbergia in Carreta (low soil P) and Arboleda (high soil P) ..................................................................................................... 112 Figure 5.3 Relationship between root TNC concentration (mg g") and resource availability in four harvests of Tabebuia in Carreta (low soil P) and Arboleda (high soil P) ..................................................................................................... 113 Figure 5.4 Relationship between root TNC concentration (mg g") and resource availability in two harvests of Cordia in Carreta (low soil P) and Arboleda (high soil P) ......................................................................................... 114 Figure 5.5 Relationship between root TNC concentration (mg g!) and resource availability in the first harvest of Pachira in Carreta (low soil P) and Arboleda (high 501 P ...................................................................................................... 1 15 Figure 5.6 Relationship between species’ mean diameter growth rates (mm mm"l yr") and mean root TNC concentrations (mg g") at the end of the third wet season ............... 1 17 Figure 5.7 Relationship between species’ survival (%) and mean root TNC concentrations (mg g") at the end of the third wet season ...................................................... 1 l8 xiii Chapter 1 Resource allocation and its implications for species growth and survival Introduction Elucidating mechanisms that maintain species diversity in ecological communities has long been a goal in ecology. Several hypotheses have been put forward to explain coexistence in forests (Chesson 2000). Of these, the idea of niche partitioning has been the focus of much research. This hypothesis considers that differences among species in resource use, coupled with spatial and temporal variations in availability of these resources, can lead to species coexistence (Denslow 1987). An alternative to niche theory, the unified neutral theory of biodiversity, calls into question the central assumptions of niche partitioning and asserts that species are competitively equivalent (Hubbell 2001). Analogous to the neutral theory of molecular evolution (Kimura 1983), the neutral theory of biodiversity states that species diversity and relative abundances are determined by random outcomes of probabilistic demographic processes (Caswell 1976, Hubbell 1979, Hubbe112001). The neutral model successfully predicts species composition patterns in communities (Hubbell 2001). However, the same patterns are also predicted by non-neutral models, and a strict assumption of species equivalence finds little empirical support (Chave 2002, Condit et a1. 2002, Fargione et a1. 2003, Silvertown 2004) and is contrary to the numerous studies that have found associations between species traits, species composition and environmental conditions (e.g., Grime 1979, Tilman 1988, Pacala et a1. 1996, Kobe 1999, Rees et a1. 2001). It is likely that both niche partitioning and stochastic processes contribute to species coexistence (e.g., Tilman 2004), though the relative importance of each in particular communities remains unknown. Light availability is limited in most ecosystems where research on species composition has been carried out, including humid tropical forests and most temperate forests. Hence, the majority of niche partitioning studies have focused on partitioning of light availability, disregarding the role of soil resources, which are also critical to plant growth and survival (Marschner 1985) and, like light availability, vary spatially (Stoyan et a1. 2000, Sollins 1998) and temporally (Lodge et a1. 1994, Campo et al. 1998). Such heterogeneity in soils, together with different species responses to soil resource conditions (Burns and Honkala 1990, Sollins 1998) leads to species segregation along topographic or edaphic gradients in both tropical (Davies 2001) and temperate forests (Host & Pregitzer 1992) and soil-based habitat specialization in both biomes (Stoyan et al. 2000, Hall et al. 2004, Palmiotto et al. 2004). Species differences in performance under diverse soil resource regimes (Kobe 1996, Schreeg et al. 2005) suggest that belowground resources may also be subject to niche partitioning. Furthermore, niche differences could operate on two axes simultaneously because resources allocated towards light-harvesting structures are unavailable for soil-resource capture (King 1993). Differences in species performance under different resource environments is a fundamental tenet of niche theory, which requires that no one species is the best competitor under all resource regimes. In this view, species coexistence is facilitated by trade-offs in species performance under differing environmental conditions. Several studies have demonstrated that across light environments, there is an interspecific trade- off between survivorship under low light availability and growth under high light availability in tropical (Hubbell and Foster 1992, Kobe 1999) and temperate forests (Kobe et al. 1995, Lin et al. 2002). This trade-off facilitates species coexistence in temperate forests (Pacala et al. 1996). Recent research in northern Michigan shows a similar trade-off operates via soil resources so that species that grow rapidly under high fertility survive poorly under conditions of low soil resource availability; this trade-off likely contributes to differences in species composition seen across broad landscape-level gradients of soil resource availability (Schreeg et al. 2005). What are the mechanisms that allow rapid growth under abundant resource availability, but preclude survival in resource-poor environments? The allocation of photosynthate to different functions within the plant could be one mechanism that defines differences in species’ growth and survival across resource environments. Research on allocation has concentrated on the effects of light and soil-resource availability on allocation to light-harvesting structures, i.e., to leaves. For example, at high resource availability, species adapted to those conditions have greater leaf surface area than species adapted to poor sites (Poorter and De Jong 1999, Craine et al. 2001 ), with the increased leaf area brought about by a combination of morphological changes at the leaf (thinner leaves) and whole-plant level (increased fraction of plant mass in leaves). Relatively less research has been focused on the effects of resource availability on allocation to soil nutrient and water harvesting structures, i.e., to roots. Analogous to the approach taken with leaves, allocation to roots has been investigated at the whole plant level by studying root mass fraction (the fraction of whole plant mass that is allocated to roots), and at the morphological level by studying the surface area or length of roots. Thus, most species in low-nutrient environments allocate a greater portion of mass to roots than to shoots, leading to a higher root mass fraction than stem mass fraction (Reynolds and D’Antonio 1996, Aerts and Chapin 2000). Again, species from more fertile habitats, which typically have faster growth rates, usually have higher specific root surface area and specific root length than those from low-soil resource environments when grown under the same conditions (Reich et al. 1998, Craine et a1. 2001, Comas et al. 2002). Thus, fast-growing species often have a greater capacity to capture resources through greater allocation to resource-harvesting structures, and increase allocation to these structures as resource availability increases, though a link between growth rate. habitat association and root morphology is not always apparent (Poorter and Remkes 1990) However, a direct analogy between allocation to leaves and roots is facile because roots, unlike leaves, are often the major site of carbohydrate storage (Loescher et al. 1990). Up to 40% of root mass can consist of nonstructural carbohydrates (Singh and Srivastava 1986, Nguyen et al. 1990, Canham et al. 1999, Newell et al. 2002). Moreover, the failure to distinguish between storage and non-storage tissue could lead to misinterpretations of patterns of allocation since allocation to storage has very different outcomes than allocation to structural growth. Storage in the form of total nonstructural carbohydrates (TNC) can occur when there is asynchrony in carbon supply and demand (Chapin et al. 1990). It can also compete directly with allocation towards resource-harvesting structures, and hence with growth (Chapin et al. 1990). Regardless of the cause of reserve formation, stored carbohydrates can be essential for survival (Kozlowski 1992) especially when there is seasonal variation in resource supply (Chapin et al. 1990). For instance, in temperate climates, TNC allows maintenance respiration over the winter, confers frost resistance and is essential for regrowth following periods of deciduousness (Chapin et al. 1990, Kozlowski 1992). TNC can also play an important role in recovery from herbivory in both tropical (Marquis et al. 1997) and temperate systems (Webb 1981) and may be an important carbon source for fine root growth in later years (Langley et al. 2002). TNC is particularly important in environments experiencing periodic disturbances (Sakai et al. 1997, Iwasa and Kubo 1987) and may allow some species to resprout following disturbance (El Omari et al. 2003, Hoffmann et al. 2004). Allocation to storage can vary with resource environment (Mooney et al. 1995, Gansert and Sprick 1998) and among species (Kobe 1997, Canham et al. 1999, Newell et 2002, Wurth et al. 2005). Previous research has suggested a link between TNC and survival, particularly under low light (Kobe 1997) and is also important in drought tolerance (Kozlowski and Pallardy 2002 ). Allocation to TNC thus appears to promote survival, often at the expense of growth, and is likely to be particularly important in species that emphasize survival over growth, especially under low resource availability. Allocation to carbohydrate storage could thus be an important physiological mechanism underlying the documented trade—off between low-resource survival and high-light growth: species that have an allocation pattern that promotes growth could be expected to have lower allocation to storage, and hence lower survival under low resource environments. Research objectives The objectives of my research are to investigate inter- and intra- specific variation in root biomass allocation patterns, characterizing allocation to stored reserves and to non- storage tissue across gradients of light and nutrient availability. A more physiologically- based understanding of species differences in competitive ability, growth and survival under different resource environments would derive insight into mechanisms determining species composition in forest communities, an overarching goal of my project. To test the generality of these ideas, my research critically examines allocation patterns and their relation to species performance across resource regimes in two different biomes — northern temperate forests of Michigan and dry tropical forests of Costa Rica, the latter considered the most endangered of all tropical habitats. Dissertation outline This dissertation presents an integrated set of greenhouse experiments, field experiments, and modeling to evaluate the role of carbohydrate storage as a basis for ecological differentiation and hence community composition in Michigan and Costa Rica. Chapters 2 and 3 are based on an experiment with seedlings of 8 common northern temperate species that were grown at two light levels and two nitrogen levels in a greenhouse environment for four months. Chapter 2 focuses on intra-specific variation in allocation to root carbohydrate storage or non-storage tissue in response to resource availability, and demonstrates that increases in root mass due to low soil resources or high light are likely driven by TNC accumulation rather than increased allocation to resource-harvesting fine roots, which is how variation in root mass allocation has been previously interpreted. Chapter 3 is focused on interspecific variation in TNC-structural tissue allocation in roots and the implications of differing allocation patterns on species growth and survival. In particular, I found that allocation to TNC could explain the trade-off between growth under high resource availability and survival under low resource availability across northern temperate forest tree species. Chapters 4 and 5 deal with field experiments in a dry tropical forest in Costa Rica, where seedlings of five tree species were transplanted into plots stratified across light and soil resource gradients. Chapter 4 focuses on seedling survival and growth and describes partitioning of soil resource and light availability by the seedlings of these tree species. I found a species trade-off between high light growth and low light survival, but only when soil resource availability is high. Chapter 5 investigates root TNC stores in the five species and discusses its relationships with species growth and survival. Species allocation to storage traded-off with growth only when soil resource availability was low but was positively linked to growth when soil resource availability was high. In all chapters, I used maximum likelihood techniques to generate predictive models of the effects of resource availability on species-specific allocation to root storage or non-storage tissue, and on survival and growth. My research hence attempts an understanding of community organization through critically evaluating species differences in performance and the physiological mechanisms that cause those differences. Chapter 2 Storage of nonstructural carbohydrates drives seedling root mass responses to light and nitrogen Introduction The allocation of limited resources to different functions within a plant underlies species- specific performance across resource environments. Much is known about mass allocation among organs (e.g., Poorter and De Jong 1999, Walters and Reich 1999). For instance, most species respond to decreasing light availability by increasing leaf area through some combination of morphological changes at both the leaf (thinner leaves) and whole- plant levels (increased fraction of plant mass in leaves). In contrast to leaves, species-specific responses of root structure and fiJnction to changing resource environments are only beginning to be understood (Ryser 1998, Comas et al. 2002). However, the allocation of carbon within organs into various compounds affects the growth and functioning of the plant (e.g., Bazzaz et al., 1987). Once again, allocation within leaves has received more attention than roots, especially among herbaceous species. For instance, the principal carbon compounds in leaves of herbaceous species and fast-growing woody species are proteins, accounting for 27%, followed by structural compounds such as cellulose and hemicelluose, and total nonstructural carbohydrates, accounting for 14% each of dry mass (Poorter et al. 1997). In contrast, although storage reserves and concentrations of structural compounds are thought to be higher, and of proteins, organic acids and lipids lower, in roots than in leaves, especially in woody species, these generalizations are based on very few studies (Poorter and Villar 1997). In roots, non-structural carbohydrate storage can constitute between 10 and 40% of total root dry mass (Nguyen et al. 1990, Kobe 1997, Canham et al. 1999). Although any perennial plant organ may serve as a storage reservoir, roots often serve as the primary storage site and have the highest concentrations of TNC (Loescher et al. 1990) _. 10 perhaps because they are less likely to be damaged or destroyed by disturbances such as treefalls or fire. Allocation to storage is especially important when resource availability varies temporally (Chapin et al. 1990). In temperate regions, for instance, where carbon gain for deciduous trees is limited to the growing season, carbohydrate storage is essential for the maintenance of living tissue over winter, for making leaves the following spring (Loescher et al. 1990, Kozlowski 1992) and for fine root growth in later years (Langley et al. 2002). In deciduous trees, carbon reserves generally reach a maximum at the end of the growing season, slowly deplete during the dormant season, rapidly deplete during new leaf and fine root flushes, and rebuild throughout the growing season (Loescher et al. 1990, Kozlowski 1992, Newell et al. 2002, Gaucher et al. 2005). Allocation to storage also varies with resource availability. Stored carbon generally increases with higher light availability (Mooney et al. 1995, Naidu and DeLucia 1997, Gansert and Sprick 1998) and with lower nutrient availability (McDonald et al. 1986, Fichtner et al. 1993, Mooney et al. 1995, Paul and Driscoll 1997). By buffering the environment, stored carbohydrates can also enhance survival, enabling recovery after defoliation due to herbivory or disease (Kobe 1997, Marquis et al. 1997, Canham et al. 1999). Furthermore, there is considerable interspecific variation in allocation to storage (Kobe 1997, Canham et al. 1999, Newell et a1. 2002, Iyer et al.) which may be related to variation in adaptive strategies among species. For example storage may allow shade tolerant species in understory environments opportunistic growth spurts following canopy opening (DeLucia et al. 1998, Gaucher et al. 2005), and disturbance adapted species to resprout following top-kill (Iwasa and Kubo 1987, El Omari et al. 2003). ll Although both storage and non-storage compounds serve multiple functions within the plant, allocation to non-storage tissue implies an increased plant investment in one or all of the following: increased access to resources, via increased allocation to root structural compounds; increased soil resource uptake via increased allocation to compounds involved in metabolism, such as lipids and proteins, or to soluble carbohydrates which are also involved in supporting mycorrhizae; and increased defense via increases in lignins and phenolics (Poorter and Villar 1997). Fine roots are a plant’s interface with soil and enable access to and uptake of nutrients and water. Though most studies have not differentiated between storage and other pools, fine roots involved in nutrient capture typically contain only 4-6% TNC (Pregitzer et al. 2000). In general, fine root production (and mortality) increases with nitrogen availability (Pregitzer et al. 1993, Van Vuuren et al. 1996, Espeleta and Donovan 2002). Because resource availability influences fine root dynamics, which are largely composed of non-storage tissue, allocation to non-storage mass in roots should be strongly influenced by resources. Given that root storage and non-storage tissue vary independently with resources and across species, failing to take stored carbohydrates into account could lead to erroneous conclusions about intra- and inter- specific variation in root characteristics. For example, root morphological metrics typically are normalized by total root mass (and hence include nonstructural carbon pools). However, although they can be mobilized, stored carbohydrates are resources sequestered mostly for future needs, rather than for current use so that the inclusion of stored carbon mass in metrics such as specific root area can lead to biased estimates of allocation to nutrient capture. A more functionally realistic index of allocation to resource capture should differentiate between stored 12 reserves and structural biomass. We define root structural mass as the difference between root mass and carbohydrate stores in the root (Canham et a1. 1999), recognizing that this is a simplistic differentiation given that both storage and non-storage carbon can serve multiple functions in the plant. To test the effects of resource levels on intra-specific variation in dry mass allocation to root storage and structural tissue, we carried out a greenhouse experiment with seedlings of eight temperate tree species. In particular, we tested the hypothesis that effects of light and nitrogen on plant allocation to root mass are primarily driven by changes in allocation to nonstructural carbohydrates rather than by changes in structural root mass that enhance resource harvesting. Because structural root mass alone is a crude metric of the potential to take up soil resources, we tested that the conversion of structural root mass to area becomes more efficient (i.e, results in higher root surface areas) with decreasing nitrogen and increasing light availability. Finally, because most studies on root function have not distinguished between nonstructural and structural carbon pools, we tested the hypothesis that normalizing root morphological traits such as surface area to total root mass rather than structural root mass decreases sensitivity to detecting resource effects because under high light or low N, increased root mass due to TNC accumulation would cancel out any increases in root length or surface area. Materials and methods Species and seed sources Our study species were Acer rubrum L. (red maple), A. saccharum Marsh. (sugar maple), Quercus velutina Lam. (black oak), Q. rubra L. (red oak), Q. alba L.(white oak), Prunus serotina Ehrh. (black cherry), F agus grandifolia Ehrh. (American beech) and Betula l3 papyrifera Marsh. (paper birch). Collectively, they encompass a wide range of shade tolerances and associations with soil resource levels (Table 2.1). Seeds for all the species were purchased from Sheffield Seed Company, Locke, NY, USA. Growth media, light and nutrient levels Seeds were stratified and then germinated in perlite. In February 2002, germinants were planted in polyethylene-coated 10.2 cm x 10.2 cm x 27 cm cardboard containers filled with a 10:9:1 (vzvzv) of a silica sandzperlitezfield soil mix. The field soil was obtained from a mesic beech-maple-oak forest near the MSU Tree Research Center, East Lansing. This mixture provided a relatively inert, nutrient poor medium where nutrient additions could be controlled. It also facilitated recovery of fine roots during harvests. The experiment was designed as a 2 x 2 factorial with two levels of light (~2% and ~22% of open-sky light) and nitrogen availability (0.5 mg N 1'I and 50 mg N 1'1 in a modified Hoagland’s solution added every three days). To prevent buildup of salts, all containers were flushed with deionized water weekly. Light levels were designed to mimic endpoints in the range of light conditions from understorey to tree fall gaps encountered in northern lower Michigan forests (Schreeg et al. 2005). Similarly, the high N treatment approximates available nitrogen levels in high fertility moraines in Manistee National Forest in northern lower Michigan (~7 pg N g1 soil; Zak et al. 1986, Kobe, unpublished data). To achieve light levels, we used an inner layer of black shade cloth combined with an outer layer of reflective knitted poly-aluminum shade cloth, the latter used to minimize heat build-up. Temperatures in each shade treatment were monitored with Hobo dataloggers (Onset Computer Corporation, Boume, MA, USA) and were 14 found not to differ (t test, p > 0.95). Mean daytime temperatures over the experiment in the two treatments were 23.63:t0.04 °C and 23.612t0.04 °C in the high and low light treatments respectively. Under high light, we planted 30 - 36 seedlings in each species x nutrient level combination. To compensate for expected higher mortality under low light, we planted 36 - 45 seedlings in each species x nutrient level combination. The experiment began in the first week of February 2002 and continued through July 2002. Root morphology To investigate intra- and inter-specific variation in root storage and structural pools and their ontogeny, we harvested subsets of seedlings at regular intervals to 3 months (Table 2.1). At each harvest, six individuals from each species-nutrient-light combination were harvested. Some harvests could not be carried out due to mortality, especially under low light treatments. To minimize variation among samples due to diurnal patterns in carbohydrate storage, all harvests were initiated 2.5 hours afier sunrise. Seedlings were washed in deionized water, separated into leaves, stems and roots. Fresh roots were scanned and the digital images later analyzed for root length and surface area with WinRhizo (Regent Instruments, Blain, Quebec, Canada). All plant parts were freeze- dried for 2 days and then weighed. Dried roots were pulverized with a ball mill (Kinetic Laboratory Equipment Co., California) before TNC extraction and analysis. 15 mm. - - - >533 :wE “go—SE mm :83 Baum ecohéem Saum no. 34% mm am gauge senescence 6829 be, m< :83 80:82 seamen» ask 2: 8 mm 2 $55 as unease om 9% seam 2:28 3.3m 2: a 3. 3 Ease ems i 32 assesses om #8 Be See a 2: 3w - - £55 32 88285 om ado v.85 senses a 2: $65 hm o. ham—Eva 33 Hugo—25 OB Mao SE? 33 Sufism - on hm - Batch 33 ENS—8 be, 2m 2an Ewem £335.93. .V 2: 2. an .. 03608.52: 5829 SE 2&8 pom £3.33 Lupus .2333 v n N _ 3:223 .333.» 396v «not!— «a ou< DEE: 3.530.. zem 2.3—m .95.: can: :eEEeO 33on .83: Sufism v5 mgm no women 0.8 motowofio 39229 32m £823 we own :05 EB .moESba 8.89.2 :8 6288—9 ocean :05 .23 gnaw—0&8 05 5 com: 860% fin 033—. 16 Table 2.2 Candidate models characterizing functional relationship between root stored reserves or structural biomass (y) and whole plant biomass (x). Models 1-3 were also used to characterize relationships between fine root surface area (y) and structural root mass or total root mass (x). Model Equation Interpretation 1 y = ax Proportional increases (a > O) or decreases (a < O) in TNC (or structural biomass) with whole plant mass (WPM) 2 y = exp(a *x) — 1 Exponential increases (a > 0) or decreases (a < O) in TNC (or structural biomass) with WPM 3 y = a*(x"b) More than proportional increases (b > 1) or less than proportional increases (b < 1) in TNC (or structural biomass) with WPM 4 y = a *exp(—b/x) Increases in TNC (or structural biomass) with WPM follows a figmoidal curve Carbohydrate analyses We used a two-step process to measure TNC. First, we extracted and analyzed soluble sugars from tissues, then analyzed extraction residues for starch. Because soluble sugars usually serve multiple physiological functions in the plant besides storage (Chapin et al. 1990), starch is often considered the most important reserve carbohydrate and is used as the sole indicator of stored reserve status (Kozlowski 1992). We chose a two-step process to enable us to measure levels of soluble sugars and starch separately. We determined TNC in roots of all seedlings harvested. For a subset of the species, we also analyzed TNC levels in stems at the final harvest, enabling us to determine whole plant level concentrations. We did not analyze carbohydrate levels in leaves because we were interested in longer-term stores rather than diurnal pools of sugars (Schnyder 1993). Soluble carbohydrates in a 20mg sample were extracted three times at 75°C using 2ml of 80% ethanol and then centrifuged at 1900 g for 5 minutes. The supematants were collected and diluted to a known volume with deionized water and the concentration of 17 soluble sugars (as glucose equivalents) was measured at 490 nm using a phenol-sulfuric acid colorimetric assay (Dubois et al. 1956). The phenol-sulfuric acid assay does not detect sugar alcohols such as sorbitol, an important soluble carbohydrate in some Prunus sp. (Keller and Loescher 1989). To determine sugar alcohols, black cherry alcohol extracts were dried and resuspended in 1 mL of pyridine containing 30 mg ml‘1 hydroxylamine hydrochloride and B-phenyl-D-glucoside as an internal standard. The suspension was heated at 75°C for l h, derivatized using a combination of hexamethyldisilazane and trifluoroacetic acid (Sweeley et al. 1963) and then analyzed using gas chromatography (Roper et al. 1988). The pellet remaining alter ethanol extraction was dried and quantitatively analyzed for starch. To gelatinize the starch, we added 2 ml of 0.1 M sodium acetate buffer (pH 5) and autoclaved the sample at 125°C for 10 minutes. After cooling, the sample was incubated with 10 units of amyloglucosidase (Roche Diagnostics Corp., Indianapolis, IN, USA) at 55°C for 16 hours. Because sample processing can introduce trace amounts of monomers derived from structural carbohydrates, the extractant was analyzed colorimetrically using glucose-specific trinder reagent (Sigma Chemical Co., St Louis, MO, USA) (Roper et al. 1988). TNC concentration was calculated as the sum of glucose equivalents of soluble sugars and starch measured in each sample. We also calculated pool sizes (concentration x root mass) of stored carbohydrates. Concentration is a good measure of proportional allocation to storage, whereas pool sizes estimate total reserves available for future use on a whole-plant basis (Chapin et al. 1990). For some species such as black cherry and sugar maple, individuals grown under low light were too small for TNC analysis. Individual l8 samples in these treatments were combined to obtain one or more composite samples with sufficient mass for analysis. No paper birch seedlings under low light were available for analysis due to high mortality. Models and data analysis We used a set of candidate models to evaluate how root TNC and non-storage mass (= total root mass — TNC pool) varies as a function of whole plant mass (Table 2.2). A linear model with no constant (Model 1) characterizes a constant fraction of whole plant mass being allocated to root storage (or non-storage mass). Model 2 represents an exponential increase in allocation to storage (or non-storage mass) with an increase in whole plant mass. Model 3 is the commonly used allometric scaling model (N iklas and Enquist 2001, Kobe et al. 2005) where b > 1 implies allocation to root TNC (or structure) increases disproportionately with whole plant mass while b < 1 implies less than proportionate increases with plant size. Model 4, a sigmoidal curve, characterizes a lag in allocation to storage (or non-storage mass) at small sizes with increased allocation as size increases, as has been shown under low light in some species (e.g., Kabeya and Sakai 2003). The a parameter represents the asymptotic TNC (or non-storage) pool over the range of whole plant mass in our experiment and b represents the rate of increase of TNC (or non-storage root mass) with whole plant mass. Model parameters were estimated using maximum likelihood methods (Hilbom and Mangel 1997). We tested effects of light and N on TNC, independent of total plant mass effects, by using dummy variables for discrete light and N treatments. For example, for a linear l9 relationship between root TNC pool and whole plant mass, we used equations of the form: Root INC pool = a 1 *(whole plant mass) *d/ + a2 *(whole plant mass) *dg, where a 1 and a 2 are the estimated parameters for each light or nutrient level, and d, and d; are dummy variables that take on values of 0 or 1, depending on the treatment level. Equations were fitted using the Gauss-Newton method in the non-linear procedure of Systat (SPSS Corp, Chicago, IL, USA). We assumed a normal error distribution, testing this assumption with probability plots and G-tests. We calculated Akaike Information Criteria (AIC) with a correction for small sample size (AICC, Hurvich and Tsai 1989, Bumham and Anderson 2002). The AIC (and AICc) is an estimate of the expected, relative distance between the fitted model and the true, unknown mechanism that generated the data (Bumham and Anderson 2002) and, unlike likelihood ratio tests (LRTs), can be used for selection among non-nested models. Unlike LRTs, because the use of AICc for model selection is not a test of significance but a selection among candidate models for the best approximating model closest to the ‘truth’, no significance values are associated with model selections based on AICc. The model with the lowest AICc is chosen as the best approximating model from the set of candidate models for each relationship. In general, a difference of more than 2 units in the AICc of two models indicates poorer support for one model over the other (Bumham and Anderson 2002). To determine treatment effects, the AICC for models incorporating treatment effects were compared to models not including treatment effects. We also analyzed root TNC concentrations at the final harvest using general linear models and ANOVA. We used whole plant mass as a covariate, checking that the assumption of homogeneity of slopes 20 was valid by verifying that the interaction between whole plant mass and treatment was non-significant. We characterized species’ allocations to nutrient harvesting structures (foraging) as the functional relationships between fine root surface area versus root mass and structural root mass. In contrast to specific root area (i.e. fine root surface area / total root mass), analyzing these data as fiinctional relationships can account for ontogenetic effects and avoids problems inherent in analyzing ratios (Jasienski and Bazzaz 1999, MacFarlane and Kobe, in review). We examined fine root area with respect to both total root mass and structural root mass. We used maximum likelihood methods to estimate parameters for these relationships, using AICc to choose the best approximating model from a set of candidate models. A similar set of models to those presented in Table 2.3 were used to test for these relationships. However, we reasoned that seedlings would allocate to nutrient uptake even in their establishment phase and had no a priori reason to expect lag phases such as in the sigmoidal curve. We therefore tested only situations where the rate of conversion of mass to area remained constant over root structural mass (or root mass; model 1); where conversion efficiency increased exponentially with root structural mass (or root mass; model 2); and where conversion followed the power law (models 3). Light and N effects were tested as described for root TNC and structural pools. Results Allocation to roots The proportion of whole plant mass allocated to roots (root mass fraction, RMF) increased with an increase in light in all species (Figure 2.1) except in black oak under 21 high N levels (Fisher’s LSD, p<0.05). Similarly, RMF decreased with an increase in N levels except for red oak, white oak and black oak in low and high light, and red maple in low light. 0 a 0.3 8) AB ’ 9) BC I High Light 0 6 - I 0.6 -« ‘ 0 Low Light 0.4 — 0.4 - I U 0.2 -« 0.2 - O U 0.0 , , 0.0 T . LN HN LN HN 0.8 0.8 C) 80 E (J) PB 0 6 — i 05 .. 0.4 - 0.4 -i c I .2 0.2 - 0.2 a . E '0- 0.0 . . 0.0 . . 8 LN HN LN HN a 0.8 0.3 E e) RM f) R0 3 u 8 0.6 - 0.6 -l I II 04 — i 0.4 ~ I 02 a i 0.2 - II 0 O 1 I 0.0 r I 0 a LN HN 0 a LN HN . , ‘ 9) SM h) W0 I § 0 0.4 l 0.4 4 0.2 -« n 0.2 -1 II 00 r r 00 T r LN HN LN HN N level Figure 2.1 Root mass fraction of species and treatments at final harvest. LN Low N; I-IN High N. 22 Storage of nonstructural carbohydrates Across all species, root total nonstructural carbohydrate (TNC) pools increased with whole plant mass (Figure 2.2). At final harvest, across all species and treatments, stored reserves accounted for about 25% of root dry mass. However, there was considerable intra- and inter-specific variation so that root TNC reserves ranged from 8% to 45% of total root dry mass (Table 2.3). Because size was used as a variable in our models, we first tested for the effects of size-independent ontogenetic effects on root TNC pools (and also root surface area). Within species and treatments, harvest date did not have an effect on TNC pools or on root surface area that were independent of plant mass except in American beech where storage pools and time of harvest had significant interactions (ANCOVA, a = 0.05; data not shown) so that results for beech need to interpreted with the caveat that the nature of the relationship between root TNC and whole plant mass is not independent of age in this species. As expected, in all species, TNC concentrations generally increased with light and decreased with N availability (Table 2.3). For a given whole plant mass, TNC pool sizes were significantly higher under high light availability than low light availability in all species and were higher under low N than high N in most species (AICC, Figure 2.2), with differences generally increasing with plant mass. In the case of red oak under high light, and black cherry under low light, however, a model that did not incorporate N effects had marginally greater support than a model with N effects (AAICc = 1.217) indicating negligible N effects. Note that in all other species, AAICC, the difference between the AICC of a model with N (or light) effects and a model without these effects was greater 23 than 2, indicating substantial support for N (or light) effects (Bumham and Anderson 2002). We also examined intra- and inter-specific variation in the two components of TNC — starch and soluble sugars (e.g. glucose and sucrose). In most species-treatment combinations, starch is the dominant form of stored carbohydrate. The one exception is shade-intolerant paper birch, which has higher amounts of soluble sugars than starch (Figure 2.3), irrespective of N availability. Excluding paper birch, starch concentrations were between 1.5 and 8 times higher than soluble sugars, although in black cherry, starch concentrations were 15-30 times higher than soluble carbohydrate concentrations including sugar alcohols (Table 2.4). In most cases, differences in TNC among resource treatments were driven by changes in starch; soluble sugar concentrations were similar across treatments. To test the assumption that most TNC is stored in roots, we also measured non- structural carbohydrates in stems in the last harvest. Although most seedlings had higher TNC pools in roots than stems, there was substantial variation among species and treatments in root versus stem allocation of TNC (Table 2.5). Red oak and white oak stored close to 90% of their reserves in their roots while red maple stored approximately 55 — 60% in roots. In most species, the proportion of TNC stored in roots did not vary significantly with treatment except for red oak and sugar maple. In red oak, the proportion of (root+stem) TNC that is in roots, adjusted for whole plant mass, decreased with light availability but not N levels (ANCOVA, p <0.05). In sugar maple, this proportion declined by more than half when both light and N levels were reduced (ANCOVA, p <0.05). 24 Table 2.3 Mean root TNC concentration (mg g") :1: SE, across species and treatments in seedlings at the final harvest. Means followed by different letters are significantly different. Means comparison with Bonferroni corrections ((1 = 0.05). Treatment abbreviations are: HLHN High light, high N; HLLN High light, low N; LLHN Low light, high light; LLLN Low light, low N. HLHN HLLN LLHN LLLN AB 361.4 a 45.6a 386.3 a 20.2at - 175.3 :1: 279° BC 78.8 a 28.621 133.5 a 13.6a 46° 48° BO 332.9 a 19.2’1 359.1 a 20.5a 198.] i 338° 294.3 a 42.9a° PB 100.4 :E 72* 89.2 :1: 144* — - RM 180.3 :1: 14.0" 247.9 a: 13.9a - - R0 328.9 :1: 49.7 a 386.0 :1: 18.7a 238.2 :t 21 .4° 286.7 i 178° SM 202.2 :1: 18.5’1 253.3 :1: 14.5a 752° - W0 368.8 a 22.8°° 449.8 3. 174" 329.5 a 280° 326.2:t 353° * Could not test for mean differences since whole plant biomass was a significant covariate. Table 2.4 Mean concentration (mg g") t SE. of sugar alcohols, total soluble carbohydrates and starch in black cherry seedlings under high light at the final harvest. Treatment abbreviations are explained in Table 2.3 legend. Treatment Sorbitol Myoinositol Soluble Starch carbohydrates HLHN 1.43 :1: 0.2 0.22 :t 0.1 4.92 :h 1.0 73.91 i 27.7 HLLN 1.41 :1: 0.1 0.24 d: 0.1 3.92 :t 0.4 129.57 :1: 13.6 Table 2.5 Mean proportion of total TNC found in roots :t SE, across species and treatments in seedlings 105-106 days old. Treatment abbreviations are explained in Table 2.3 legend. HLHN HLLN LLHN LLLN AB 0.75 3. 0.04 0.83 $0.01 B0 0.89 i002 0.89 :1: 0.02 0.77 a 0.02 0.80 a 0.02 RM 0.62 3.0.03 0.54 a 0.06 R0 0.88 30.031“b 0.94 a 0.0021 0.84 3: 0.02b 0.88 3. 0.02b SM 0.77 4:003a 0.76 21:0.04° 0.35 a 0.02b wo 0.95 3 0.00 0.87 d: 0.01 0.87 a 0.04 25 10° 8MB r110,1 . mac 10‘1 - 10°2 7 10-2 _ / / (3 103 a . V —. HLHN 104 a 103 _ _o. HLLN vv —v— LLHN 10-5 . v LLLN 104 L r 10-6 0.1 1 0.01 c)BO 100 d C 100 - g) , 10.1 “ g V&' (a 10.1“ '//V V E )9 ' ' 1°" ‘ ' O ‘ ' ' I E 1 10 1 .1 _ o O 10'1 a o: (8 O / -2 - QQ/ ' 10-2 ~ 0 10 e 0 / 10'3 7 o/ 9 -3 4 10 . 104 0.01 0.1 1 0-1 1 0.1 . V v v 0.01 ' 1 Whole plant mass (9) Figure 2.2 Root TNC pools (g) as a function of whole plant mass (g). Note that sugar maple and red maple have data for high light only. Lines represent best-fit equations. Treatment abbreviations are explained in Table 2.3 legend. 26 450 450 400 — a) HLHN - Solubles 400 _ D) HLLN 350 _ - Starch 300 - 250 A 200 " ‘T m 150 -' U) 100 _ E v 50 _ C 02 0 _' . T H “ ' .0-0 9 AB BC 80 PB RM R0 SM AB BC BO PB RM R0 SM wo 'E 450 450 a) g 400‘ c)LLHN 400- d)LLLN O 350 - 350 - 0 30° 4 300 n 250 n 250 .l 200 '1 200 1 150 — 150 — 100 - 100 .. 50 -1 5o _ 0 r _I- 0 1- AB BC 30 R0 SM W0 AB BC BO RO SM WO Species Figure 2.3 Soluble sugars and starch concentrations in roots at final harvest. Treatment abbreviations are explained in Table 2.3 legend. 27 300 400 O O a) HLHN O b) HLLN 250 - 350 _ .0. A A O - - [J O . C16 0 O . 150 _ 0 121A . 250 - o O 0 AB Cp 0 .- v ,5“ A? D O 0 BO 0 Cl . I v .m 100 L. O ‘ P3 200 '— I O 'V O) ‘ A RM V E 50 A ‘ I R0 150 g ‘ :1 SM '3 V WO V c o 4 1 1 1 1 100 1 1 1 1 8 0 1 00 200 300 400 500 600 100 200 300 400 500 § 180 300 o c) LLHN d) LLLN O 2 160 - O 1- 250 ~ 8 6'5 140 '- O 200 _ V V 120 — ' I V _ V. 150 O . V 100 L O u I :' o v 0 100 ~ 1 V 80 ~ _ I .v C] 60 - O- 50 ' ' 4o 1 1 1 1 1 o 1 1 1 1 50 1 00 1 50 200 250 300 350 0 100 200 300 400 500 Root TNC concentration (mg 9”) Figure 2.4 Stern and root TNC concentrations in species at final harvest. Treatment abbreviations are explained in Table 2.3 legend. 28 _a) AB b) so 0'1 ” 0.1 - . /o O V _ 001 _ f V V LLLN 001 W J; 1 ' l 01 1 0.01 01 ‘l 10 c)BO d) R0 0 -l I .0 —L —L Non-storage root mass (9) e) RM 0.1 - (125‘ $3 0.01 — // ’ ‘ 0.01 4 0.01 0.1 1 0.1 1 e)WO ,0 1 09% A99 0 0.1 Whole plant mass (9) Figure 2.5 Non-storage root mass as a function of whole plant mass. Note that red maple and sugar maple have data for high light only. Lines represent best-fit equations. Treatment abbreviations are explained in Table 2.3 legend. 29 Changes in TNC pools as reported above are the result of changes in both mass allocation between stems and roots and the concentration of labile carbohydrates in those organs. In red oak under high light and high N, for instance, root TNC concentration is approximately double stem TNC concentrations (Figure 2.4), while RMF and stem mass fraction (SMF) are ~ 0.65 (Figure 2.1) and 0.2 respectively. The high proportion of TNC stored in roots in this species, is hence due to both the higher concentrations in roots and the higher root mass fraction. Similarly, the reduction in root TNC pools under low light and N levels in sugar maple is due not only to the decrease in root TNC concentrations by approximately half (Figure 2.4), but also because RMF decreases from 0.5- 0.6 under high light to 0.2 under low light and low N (Figure 2.1) while SMF increases from ~02 to ~04 (data not shown). Total stored reserves in our study species varied more than tenfold, from a low of ~3% of dry mass in paper birch under high light and N availability to ~36% of dry mass in white oak under high light and low N. As in the case of root TNC concentrations, whole plant TNC concentration also increased with increasing light availability and decreasing N availability. Non-storage mass in roots In all species, non-storage mass increased with whole plant mass (Figure 2.5). Similar to root TNC-plant mass relations, structural root mass-plant mass relations varied with N and light treatments, but responses were weaker and seen in fewer species. Under low light, only red oak and American beech showed N treatment effects, both showing weak decreases in structural root mass with increases in N (Figure 2.5). Under high light, 30 American beech, black cherry, red maple, and sugar maple decreased non-storage root mass in response to higher N. Red oak models of non-storage root mass as a function of whole plant mass were different between N treatments (AIC, Figure 2.5) under both high and low light. However, especially under low light, support for a model with N effects was marginal: AAIC between models with and without N effects was only 0.6, indicating that little information is lost if N effects are not specified. Under high light, the difference between high and low N arises from higher variance estimates under high N (0.513) versus low N (0.280) and not the parameter estimates governing the functional relationship between root structural and total plant mass. In black oak and white oak, even though root TNC pools decreased with increased N, N did not influence structural root mass, regardless of light availability (Figure 2.5). Surface area of fine roots In high light, for a given structural root mass, American beech, black oak, red oak and sugar maple had greater fine root surface area under low N than high N. In contrast, red maple had higher root surface area under high than under low N (although there was limited overlap between N treatments in structural root mass). White oak was the only species in which root surface area showed no response to N (Table 2.6). In black cherry, functional relationships were very similar for high and low N although their variances were different. Under low light, only shade intolerant black oak and red oak, generally highly sensitive to light limitation, showed increased allocation to fine root surface area under low N. American beech, black cherry and white oak did not vary root surface area in response to N level under low light. 31 To compare our results on species foraging patterns with conventional metrics of root morphology (such as specific root area), we also analyzed fine root surface area relative to total root mass (Table 2.7). When fine root area of red oak under low light and sugar maple under high light are normalized to total root mass instead of non-storage root mass, neither showed a significant response to N level. However, both increased root surface area per unit non-storage root mass under low N. Thus, normalizing fine root area to total root mass can obscure plant responses to nutrient environments. 32 QNeEKNSE Se 2. Zemaowasnéfi 753.25: 9.852.3an who mm Emmtmmmifaafi 5.5.75.5 03 €386,392: 23 a Emtmseemhmme 2.5: $82.2qu 522.853: :3 2 kaozmmremmeznfifi 25: s: Amoemfiemofia m3 : Emhweomnéfi zfi: 9865.848 $3388.85 63 M: fazemvammefiuéfi 2:4: 2: 6512...: 23933.8: who am pfizmmrtqsmnfimm 75.5 3:62.823 and mm Emtesezuéfi 2:3 9622.83.83 2:. mm Emtmmmenufifi 2.3: Engages who 3 Emmaaénfim: 2:4: 0: 4238.883 «.2 : EmEaEimfi 213:. 55.2.8 massage Se 2 gmrmflénfifi 2:5. $32.38: 23 : Emmavmsznéfi 21:: 632 .629 42 2 2mm18mezu 1 implies allocation to root TNC (or non-storage) increases disproportionately with whole plant mass while b < 1 less than proportionate increases with plant size. Model 4, a sigmoidal curve, characterizes a lag in allocation to storage (or non-storage) under small sizes with increased allocation under large sizes, as has been shown under low light in some species 47 (e.g., Kabeya and Sakai 2003). The a parameter represents the asymptotic TNC (or structural) pool over the range of whole plant mass in our experiment (or structural root mass) while b represents the rate of increase of TNC (or structural root mass) with whole plant mass. Model parameters were estimated using maximum likelihood methods (Hilbom and Mangel 1997). I tested effects of light and N by using dummy variables. For example, for a linear relationship between root TNC pool and whole plant mass, I used equations of the form: Root TNC pool = a 1 *(whole plant mass) *d, + a; *(whole plant mass) *dz, where a I and a; are the estimated parameters for each light or nutrient level, and d, and d 3 are dummy variables that take on values of 0 or 1, depending on the treatment level. Equations were fitted using the Gauss-Newton method in the non-linear procedure of Systat (SPSS Corp, Chicago, IL, USA). I assumed a normal error distribution, testing this assumption with probability plots and G-tests. I calculated Akaike Information Criteria (AIC) with a correction for sample size (AICc, Hurvich and Tsai 1989, Bumham and Anderson 2002). The model with the lowest AICc was chosen as the best approximating model from the set of candidate models for each relationship. The AIC, and hence the AICC, is not a test of significance but provides a method to choose the best possible model from a set of candidate models (Bumham and Anderson 2002). Hence no significance values are reported for model selections using AICc. In general, a difference of more than 2 units in the AICc of two models indicates poorer support for one model over the other (Bumham and Anderson 2002). To determine treatment effects, the AICC for models incorporating treatment effects were compared to models not including treatment effects. 48 To test for species differences, for every treatment, I first developed a global model for the relationship of interest (e.g., root structural mass as a function of whole plant mass) using data from all species, across all harvests. I then compared AICC values of species specific models with species-general models to determine if there were species effects in that treatment. I mapped uncertainty in parameter estimates (95% support) to uncertainty in functional relationships to conservatively assess species differences as regions where 95% support for the fimctional relationships do not overlap (Austin and Hux 2002). I also analyzed root TNC concentrations at the final harvest using ANCOVA, with whole plant mass as a covariate. I characterized species’ allocations to nutrient harvesting structures as the functional relationships between fine root surface area and root mass or structural root mass. Analyzing these data as functional relationships, in contrast to specific root area (i.e. fine root surface area / total root mass), can account for ontogenetic effects and avoids problems inherent in analyzing ratios (Jasienski and Bazzaz 1999, MacFarlane and Kobe, in review). I also examined fine root area with respect to both total root mass and structural root mass. 1 used maximum likelihood methods to estimate parameters for these relationships, using AICc to choose the best approximating model from a set of candidate models. A similar set of models to those presented in Table 2 were used to test for these relationships. However, because I had no a priori reason to expect lag phases in allocation to nutrient uptake in the establishing seedlings, only the first three models in Table 2.2 were tested. Light and N effects were tested as described for root TNC and structural pools. 49 Results Storage of nonstructural carbohydrates Across species and treatments, stored reserves accounted for about 25% of root dry mass at final harvest. However, there was considerable intra- and inter-specific variation, with TNC ranging from 8% to 45% of total root dry mass (Table 3.1). Chapter 2 examines effects of light and N on allocation within species. Here I use the same data to compare allocation among species. As expected, in all species, TNC concentrations generally increased with light and decreased with N availability (Table 3.1). Among species, white oak consistently had the highest TNC, followed by American beech, red oak and black oak. Under high light, black cherry and paper birch had the lowest TNC concentrations under both N levels (Table 3.1, ANOVA, p < 0.05) while under low light, black cherry TNC concentrations were at least an order of magnitude lower than other species. Table 3.1 Mean root TNC concentration (mg g") :t SE, across species and treatments in seedlings at the final harvest. Means followed by different letters are significantly different. Means comparison with Bonferroni corrections ((1 = 0.05). Treatment abbreviations are: HLHN high light, high N; HLLN high light, low N; LLHN low light, high light; LLLN low light, low N. Species abbreviations are: AB American beech; BC black cherry; BO black oak; PB paper birch; RM red maple; R0 red oak; SM sugar maple; W0 white oak. HLHN HLLN LLHN LLLN AB 361.4 :1: 45.621‘ 386.3 :1: 20.23 - 175.3 :t 27.9b BC 78.8 :1: 28.6a 133.5 :1: 13.6a 4.6b 4.8b BO 332.9 i 19.2a 359.1 :1: 20.5a 198.1 a: 33.8b 294.3 3: 42.931“ PB 100.4 :1: 72* 89.2 :t 14.4* - - RM 180.3 :1: 14.08 247.9 :t 13.9a - - R0 328.9 :t 49.7 a 386.0 :1: 18.7a 238.2 :1: 21 .4b 286.7 :t 17.8b SM 202.2 212 18.5a 253.3 :1: 14.5a 75.2b - wo 368.8 :1: 22.8ab 449.8 :1: 17.43 329.5 a 280‘) 326.2i 35.3b * Could not test for mean differences since whole plant biomass was a significant covariate. 50 a) HLHN b) HLLN + 1 - _ - r... /' x" 0.1 — ~ " ¢ + A. _ —O- BC 0'01 fl ‘3 + BO / —VL - RM / v R0 A —U - SM 3 0.001 4 - _. WC 3 . . . . . o 0.1 1 10 0.1 1 10 z 1 [— ‘5 C) LLHN d) LLLN o / °= a l . A/ ' . I l /' 001 - ' 4 We! _ _ / 0.001 / / 0.0001 p 0.01 01 1 0.01 0.1 Whole plant mass (9) Figure 3.1 Variation in root TNC pools with whole plant mass under different light and N levels. Treatment and species abbreviations are as in Table 3.1. 51 Across species and treatments, the oaks consistently had the greatest pools of root TNC at any given whole plant mass (Figure 3.1). Under high light, red oak exhibited a high degree of individual variability in TNC so that patterns of storage in this species were not significantly different from any other species. Under high light and low N, root TNC was higher in white oak than in black oak and sugar maple. Under low light and low N, red oak had lower storage than white oak. The species with the slowest relative growth rate in our study, white oak, had the largest reserves compared to all other species (except red oak) under most resource conditions. White oak had higher allocation to root storage than black oak under all treatments except under low light and N, where their TNC stores were similar. When both light and N were high, American beech had significantly lower TNC at lower whole plant mass, but levels were not different from white oak when whole plant mass was greater than approximately 2g. Under high light and low N, both species had similar TNC storage across the range of whole plant mass. Under high light, regardless of N level, sugar maple had higher carbon stores than red maple. TNC stores in black cherry were similar to the maples (Figure 3.1). White oak and black oak both showed linear or near linear increases in root TNC in all treatments except low light and low N, where black oak had an exponential increase in root TNC allocation (Appendix 1). The allometric scaling function (Model 3) provided best fits for red maple under high light: the species had less than proportional increases in root TNC under low N as total plant mass increased, and more than proportional increase in root TNC under high N. Under high light, allocation patterns in red oak, sugar maple and black cherry were best described by Model 4, as root TNC increased sigmoidally 52 with total plant mass. However, unlike red oak, sugar maple and black cherry both approached asymptotic allocation to TNC pools under high light. a) HLHN ,, b) HLLN V 1 1 .5 - 0.1 ‘1 -1 g 3') '2 fl - RM R0 g / / / - SM E // wo '°" 0.01 r . . . g 0.1 1 10 0.01 0.1 1 § 1 l d LLLN ' 5 c) LLHN Ad ) ,r 1’ ¥ /. C I 2° 01 — ’ — ”jig? / / 0.01 - « ? 0.001 1 r 1 . 1 . 0.01 0.1 1 0.01 0.1 1 Whole plant mass (9) Figure 3.2 Variation in root structural pools with whole plant mass under different light and N levels. Treatment and species abbreviations are as in Table 3.1. 53 Non-T NC root mass Species differences in mass allocation to roots were due largely to differences in root storage (Figure 3.1), and less so to differences in non-TNC root mass (Figure 3.2). In contrast to root storage pools, allocation to structural pools was strikingly similar across species in most treatments. Under all treatments, most species exhibited linear or near- linear responses in allocation to structural root mass with similar slopes in the relationship (Appendix 2). Red maple was the only species to exhibit asymptotic allocation to structural root mass, and this was only in high light and high N, when individuals achieved largest size. Under high light, low N, the structural root mass-whole plant mass relationship for red maple is also near-linear (Figure 3.2a and 3.2b, Appendix 2). Under high light, regardless of N level, sugar maple, red maple, black cherry and American beech did not differ from each other or the oaks in their allocation to structural root mass at a given whole plant mass (Figure 3.2a and 3.2b). Under low light, when N levels were high, American beech and black cherry had similar allocation to structural mass for a given whole plant mass though the lack of overlap in their ranges did not allow a comparison with the oaks (Figure 3.2c). When both light and N levels were low, for a given whole plant mass, allocation to structural root mass was lower in sugar maple than in the oaks (Figure 3.2d). 54 b)HLLN 100 - 10 a 5 r N g // (U 1 1 1 1 a 1 1 § 0.01 0.1 1 10 0.001 0.01 0.1 1 t j§,m,, c)LLHhI ' _ cDLLLN O .1 e O C 1: 104 / - 0.001 0.01 0.1 1 0.01 0.1 1 Non-storage root mass (9) Figure 3.3 Fine root surface area as a function of structural root mass under different light and N levels. Treatment abbreviations are as in Table 3.1. 55 Fine root surface area (cmz) 0.01 0.1 1 Total root mass (9) Figure 3.4 Fine root surface area as a function of total root mass in the HLHN treatment. Treatment and species abbreviations are as in Table 3.1. Surface area of fine roots All species had higher root surface area per unit structural mass under low N levels and higher light levels (Figure 3.3, Appendix 3). Across all treatments, the oaks had similar fine root area per unit structural root mass, which usually was lower than that of other species. When both light and N levels were high, the fast-growing shade-intolerants black cherry and red maple both had higher fine root surface area per unit of structural mass, than did the oaks, American beech and sugar maple. Sugar maple, though lower than red maple in its fine root area per unit structural root mass had higher root area per unit structural mass than the oaks. When light levels were high but N low, American beech 56 and black cherry both had higher fine root surface area per unit structural root mass than the oaks. Sugar maple had non-significantly higher root area per unit structural mass than the oaks. Fine root surface area per unit structural mass was lower in white oak than red and black oak, but non-significantly. To compare our results with conventional metrics of root morphology, I also analyzed fine root surface area as a function of total root mass. For the high light, high nutrient treatment (Appendix 4), American beech has root morphology similar to fast- growing red maple, black cherry and sugar maple, and higher than the oaks (Figure 3.4). Allocation patterns under other treatments are similar to those seen when analyzed with structural root mass (Appendices 3 and 4). Relationships among TNC, growth, and survival Within each treatment, there was an interspecific trade-off between mean growth rates and mean root TNC concentrations at the final harvest (Figure 3.5): species with high growth rates allocated smaller proportions of their root mass to carbon storage while species with large energy reserves accumulated these at the expense of growth rates. For example, in all treatments, white oak consistently had the lowest growth rate of all species but the highest allocation to root TNC. In contrast, red maple and black cherry had the highest rate of biomass accumulation under high light and high N, but the lowest root TNC levels. In all other treatments, black cherry had the highest growth rates and the lowest root TNC levels. Although our experiment did not measure survival, I used survivorship data from an experiment that was conducted at Manistee National Forest (MNF) in northern lower 57 Michigan (Kobe, Kunkle and Walters, in prep.) to test the relationship between TNC measured in this study and survival. Data on survival of field seedlings from July 2001 to October 2002 are for seedlings grown in light levels of ~2 — 45% sun in the high fertility moraines and for 14-27% full sun in the low fertility outwash plains (Kobe, Kunkle and Walters, in prep). Across species, field seedling survival in the moraine is positively correlated with mean root TNC levels under high light, high N (Figure 3.6a; r2 = 0.98, p < 0.01) and lOW light, high N levels (data not shown; 12 = 0.86, p < 0.05). Field survival in the outwash is positively correlated with mean root TNC levels in the high light, low N treatment (Figure 3.6b; r2= 0.69, p < 0.05). Fast-growing, shade-intolerant black cherry had the lowest survival in both moraine and outwash and had the lowest TNC storage under high light under both high and low N, and under low light and low N. The slow- growing shade-intolerant white oak had the highest survival on the fertile moraine and the highest storage under comparable resource levels in the greenhouse. In the low- fertility outwash plain, red oak had the lowest mortality; under comparable resource levels in the greenhouse, red oak had the second-highest storage of all species. Black oak had the second highest survival rate on the outwash plain and had TNC levels a little lower than red oak. The shade-intolerant fast growing red maple had lower TNC and lower survival than its more shade-tolerant, slower growing congener on both the moraine and outwash. 58 I I Y I 250 300 350 400 450 500 0.055 0.026 . 8) HLHN , AB b) HLLN 0050 V 0 130 0.024 ~ 0 O v RM 0045 « 1: R0 I W0 0.022 '1 0.040 - 0.020 4 0.035 « 0.018 - 0.030 « v 5 A m- I § 0.020 I T Y Y I 1' 0.014 T 1* 50 100 150 200 250 300 350 400 100 150 200 £0,018 0014 o 1. 4 d LLLN 00.016 CALLHN 0.012 - ) 0.014 - o 0.010 J 0.012 1 0.010 4 . 0.008 - 0.008 - 0.006 1 0.006 « 0.004 . 0.004 « 0.002 . I 0°02 1 0.0m T 7 1 I I I 1 0.m0 I I 0 50 100 150 200 250 300 350 o 50 100 150 200 250 300 350 Root TNC concentration (mg/g) Figure 3.5 Trade-off between growth rate and root TNC concentrations under different light and N levels. Treatment and species abbreviations are as in Table 3.1. 59 80 7o 0 ac 0 so 70 ~ 17 RM I g 60 4 ' A ' R0 — A g 8 0 5M '. 2g 0 E o I we in 50 ‘ o N 60 _ N I E 13 EU .5 O O 0 4o .. a I O c l D > '- 2 '- 'E ,3, 50 ~ 9 § :1 __ 'E _ 30 —l U) a V a a 32 3 a) 3 40 1 0 a2 20 j ,2 = 093 0 V r2 = 0.69 30 f . r 10 fi . . . fl . j 0 100 200 300 400 100 150 200 250 300 350 400 450 500 HLHN Root TNC concentration (mg/g) HLLN Root TNC concentration (mg/g) Figure 3.6 Species survival in field sites in Manistee National Forest as a function of root TNC concentrations in the greenhouse. Treatment abbreviations are as in Table 3.1 Discussion Nonstructural carbohydrate storage, growth and survival Our results show a strong negative relationship between TNC reserves and growth and a strong positive relationship between allocation to TNC and survival, suggesting that allocation to carbon storage likely underlies the trade-off between high resource growth and low-resource survival observed in the field (Kobe et al. 1995, Lin et al. 2002, Schreeg et al.2005). Across all treatments, species grth rates were strongly negatively correlated with allocation to TNC with slow-growing species associated with infertile sites, such as white oak and black oak, consistently having higher allocation to TNC than fast-growing species such as red maple. Our results highlight that allocation of carbon to storage implies relatively less carbon for growth (i.e., new leaves and structures for acquiring further resources). Such reserve formation at the expense of growth has been 60 shown for some species such as sugarbeet (reviewed in Chapin 1990) and has been suggested in some tree species (Waring and Pitrnan 1985, Kobe 1997). The inverse relationship between TNC and growth rates could also explain the finding that species growth rates and root tissue density are negatively correlated (Ryser 1998). Given the high specific gravity of starch (Neuhaus and Schulte 1996), a primary constituent of TNC, our results suggest that differences in starch might underlie differences in tissue density seen across species from different habitats. Across species, there was a strong positive correlation between the proportions of root TNC and survival in the field, consistent with earlier studies that linked TNC and survival (Kobe 1997, Canham et al. 1999). F ast-growing species such as black cherry and red maple had low root carbon reserves and high mortality in the field while slow- growing species generally associated with infertile sites had high TNC reserves and had high survival in both fertile and infertile sites. Large TNC stores could enhance survival by allowing respiration when maintaining a positive carbon balance is otherwise difficult, such as under seasonal drought. TNC stores are also essential for overwinter survival, frost tolerance and for spring reflush (Loescher et al. 1990, Kozlowski 1992). Increased storage can also enhance seedling survival by facilitating recovery after disturbances (Iwasa and Kubo 1983), and after disease (Kozlowski 1992), herbivory (Marquis et al. 1997) or fire (Hoffman et al. 2003). The relationships between root carbon reserves, growth and survival suggest that allocation to TNC could provide a mechanistic explanation for the species trade-off often documented between growth in high-resource environments and survivorship in low- resource environments (Kitajima 1994, Kobe et al. 1995, Schreeg et al., 2005). Our 61 results show that carbohydrate storage correlates with the ability to survive low resource environments, whether of low-light tolerance in some species, such as sugar maple, or of low-nutrient tolerance in others such as white oak and black oak. Our results are consistent with findings of previous researchers who have suggested that fast-growing species survive poorly under low light because of lower investment in storage (Kitajima 1994) and have shown a correlation between low-light survival and root carbon storage (Kobe 1997). Species often have inversely correlated competitive abilities for different resources (Tilman 1985, Huston and Smith 1987). Thus, high TNC reserves in sugar maple may confer low-light tolerance in this species, generally considered a competitor with respect to nutrient availability (Burns and Honkala 1990). Similarly, larger root carbon stores that appear to confer tolerance of low-nutrient conditions in white oak and black oak do not manifest as low-light tolerance; instead, both black and white oak are relatively shade-intolerant (Burns and Honkala 1990, Abrams 2003). Hence, TNC storage appears to confer stress tolerance, manifesting as either low-light tolerance or low- nutrient tolerance, but not both, at least in the species I studied. Indeed, allometric constraints prevent species from adapting to conditions where both nutrient and light availability are simultaneously low (Peace and Grubb 1982) although plants that can tolerate both low light and nutrient supply may be constrained by slow growth when neither is limiting (Huston and Smith 1987). 62 Species variation in allocation to root storage and non-storage mass The eight species included in this study exhibited considerable variation in the amounts of root biomass allocated to storage, ranging from 8 to 45% across species and treatments. These values are similar to those found in 2-year-old field grown seedlings in Canham et al. (1999) for sugar maple and lower for black cherry and red maple. High light root TNC levels in our species were similar to tap root TNC levels in understory red maple, sugar maple and black cherry saplings in DeLucia et a1. (1998). In contrast to storage patterns, these eight species showed remarkably few differences in allocation to structural root mass, consistent with hypothesis 3. There was extensive overlap in species’ allocation to structural root mass in all treatments except when both light and N are low, where sugar maple has lower structural root mass per unit whole plant mass compared to the oaks. This lack of difference in allocation to non- storage mass coupled with the variation in allocation to storage under different resource regimes suggests that patterns relating species root mass allocation with habitats may be driven largely by differences in allocation to root carbohydrate stores, rather than by allocation to root structural mass, which would be used in resource harvesting. Similarly, intra-specific variation in root mass with changes in resource availability is also driven largely by changes in TNC, rather than by changes in structural root mass (Chapter 2). Patterns of carbohydrate allocation could arise from phylogeny and/or seed size since the oaks and American beech, all in the family Fagaceae, often had similar patterns in their responses. However, functional forms of the root TNC-whole plant mass relationships among these four species were different. The most shade-tolerant of the four species, American beech, had more than proportional increases in root TNC with whole 63 plant mass, intermediate tolerant red oak’s showed a sigmoidal response while the least shade tolerant black oak and white oak were linear or near-linear in their responses. Differences were apparent even between black and white oak, the two species that are most similar in their habitats, both being restricted to xeric, infertile sites. Though both had relatively high allocation to TNC, white oak under most conditions allocated more to root carbon reserves than black oak, suggesting the former should have greater shade tolerance and slower growth than the latter, as is the case (this paper, Burns and Honkala 1990, Abrams 2003). Further, allocation patterns were also very different among the two other congeneric species in our study. Of the two maples, the slower-growing, shade- tolerant sugar maple had higher root reserves than its fast growing congener under high light and high N. Sugar maple also had greater structural root tissue per unit whole plant mass than red maple, although red maple was far more efiicient in converting this structural tissue into surface area than was sugar maple. My analysis of fine root surface area as a fiinction of both structural root mass and total root mass showed that conclusions on species’ foraging patterns could be erroneous if root TNC is ignored in calculating specific root area. For instance, when both light and N levels are high, using the conventional metric for root surface area (normalized to total root mass), American beech appeared to allocate far more to foraging than red oak and white oak. Yet, for a given structural root mass, all three species have similar fine root surface area, signifying their conversion efficiencies of structural root mass to surface area are, in fact, similar. Further, the difference in root surface area between the oaks and the other species appears more divergent when analyzed with respect to total root mass than with structural root mass. This difference arises because species that have high root 64 area per total root mass tend to have low root TNC (such as black cherry and red maple) while species that have lower root area per total root mass tend to have higher TNC (such as white oak, red oak and black oak). Hence, normalized to total rather than structural root mass, species differences in fine root area appear exaggerated. Conclusions My results suggest that storage of nonstructural carbohydrates (i.e., TNC) confers stress tolerance, allowing species to survive in environments that inhibit plant growth, such as under low light or low nutrient availability. Increased stress tolerance as a result of TNC storage can thus manifest as either low-light tolerance or as low-nutrient tolerance. Fast- growing species associated with high-resource environments, such as paper birch, typically store low amounts of TNC, suggesting root carbohydrate storage is another trait that differentiates competitors from stress tolerators (Grime 1977). The ability to tolerate low resource environments by allocation to storage at the expense of structural growth provides a mechanistic explanation for the species trade-off between fast growth under high light versus survival in low light (Kitajima 1994, Kobe et al. 1995, Lin et al. 2002) and for the recently documented trade-off between growth under high soil resource availability and survival under low-soil resource availability (Schreeg et al., 2005). Species that allocate more to storage are less competitive in high resource environments than those that allocate more to structural growth, but they have lower mortality in low resource environments. 65 Acknowledgements I thank Wayne Loescher, Zhifang Gao and Mercy Olmstead, Dept of Horticulture, MSU, for their assistance with TNC analysis. I thank Paul Bloese and Randy Klevickas for their support at the Tree Research Center, MSU, and Jeff Eickwort, Jason Vokits and Brent Troost for their help in the greenhouse and laboratory. The manuscript benefited from critiques of an earlier draft by Steve Hart’s graduate discussion group on roots at Northern Arizona University. RKK contributed to the writing of this manuscript while on sabbatical at Northern Arizona University (Biology and Forestry) and The Arboretum at Flagstaff and gratefully acknowledges their support. This work was supported by a grant fiom the McIntire-Stennis Cooperative Forestry Research Program. 66 Chapter 4 Niche partitioning by seedlings in a tropical dry forest: effects of light and soil resource availability 67 Introduction Among the many hypotheses put forward to explain species coexistence in tropical forests, prominent is that of niche partitioning. According to this hypothesis, differences among species in resource use, coupled with spatiotemporal variation in resource availability, can promote species coexistence (Denslow 1987). Because the availability of light is a fundamental determinant of growth and survival in light limited tropical forests, the majority of studies have focused on niche partitioning across light gradients. Several have demonstrated a trade-off between growth under high light and survival under low light (Hubbell and Foster 1992, Kitajima 1994, Kobe 1999); this trade-off has been shown to facilitate tree species coexistence in temperate forests (Pacala et al. 1996). Although evidence for niche partitioning along light gradients has been equivocal, species may simultaneously segregate along soil resource gradients. Tropical forests are highly heterogeneous in soil nutrient and water availability (Sollins 1998). Such spatial heterogeneity in soils leads to species segregation along topographical gradients (Davies 2001) and soil-based habitat specialization (Palmiotto et al. 2004) suggesting that partitioning of belowground resources likely occurs in tropical forests. In temperate forests, a species trade-off between annual grth under high soil resources and survival under low soil resources likely facilitates species coexistence at the landscape level (Schreeg et al. 2005). It is likely similar trade-offs operate in tropical forests. Additionally, niche differences could also operate on two axes simultaneously because resources allocated towards light-harvesting structures (leaves, stems) are unavailable for soil-resource capture (King 1993). 68 In addition to spatial heterogeneity, plants also face temporal variation in resource availability. Occasional droughts are an integral feature of some tropical rainforests (Walsh and Newbery 1999). In tropical dry and moist forests, seasonality in rainfall leads to pronounced changes in both soil nutrient (Lodge et al. 1994, Campo et al. 1998) and moisture levels, possibly with differential effects on species growth and mortality (Delissio and Primack 2003). When there is high seasonal variation in resource availability, species’ trade-offs between growth and survival may also be underpinned by their ability to grow rapidly during the growing season when resources are plentiful versus surviving periods of resource limitation during seasonal drought. Among species in moist tropical forest, growth in the wet season is inversely correlated with species survival in the dry season, especially under low light (Pearson et al. 2003). In this study, I examined growth and survival responses among five co-occurring tree species to variations in light and soil resource availability in a seasonally dry tropical forest. I worked with seedlings, at which stage habitat heterogeneity is postulated to play a more extensive role (Grubb 1977). In particular, the questions I addressed were: 1. How are seedling growth and survival influenced by variation in soil resource levels? 2. Is there a trade-off between species growth rates under high light and survival under low light consistent with their distributions across light environments? 3. How does soil resource availability affect the high light growth — low light survival trade-off? 4. Does seasonal drought have a differential effect on species growth rates and survival? 69 Methods Study site and species The study area is within the Palo Verde National Park (10°21’N, 85°21 ’W), Guanacaste, Costa Rica. This region is classified as tropical dry forest according to the Holdridge (1969) life-zone classification. The annual average temperature is 27.4°C and rainfall 1817 mm, with a pronounced dry season from December to May (Jimenez et al. 2001). The semi-deciduous forests in the 20,000 ha Park have approximately 65 tree, shrub and liana species per 1000 m2 or 0.1 hectare (Gillespie 1999). The species I selected for my study were Astronium graveolens Jacq., Pachira quinata (Jacq.) WS Alverson, Cordia gerascanthus L., Dalbergia retusa Hemsl. and T abebuia rosea (Bertol.) DC. The species have some soil resource affinities (Table 4.1, Eugenio Gonzalez, pers. comm.) but all co-occur as adults in the secondary forests of the Park. I selected these species on the basis of seed availability, and to present a range of seed sizes and growth habits (Table 4.1). Relative to other species in the area, Pachira and T abebuia are fast-growing shade-intolerant species (Kitajima 2002), Astronium and Dalbergia are slow growing late-successional species (Hartshom 1983, Piotto et al. 2004); Cordia is a late-successional species (Opler et al. 1975), whose growth rate is unknown. Of these five species, Cordia is in danger of extinction, while Dalbergia and Astronium are threatened (J imenez-Madrigal 1998). The study species are referred to by their genus names only throughout the remainder of the paper. Experimental methods Based on preliminary soil analysis in March 2000, I chose three sites that differed in soil phosphorus (P) availability to use for a transplant experiment. All sites were in secondary 70 forest areas, within 3-6 km of each other and on Ustorthent soils. The sites had a gradient in soil fertility, and in particular soil P, with site Carreta having the lowest, Ojo de Agua intermediate and Arboleda the highest levels of soil P. Although the sites also had different levels of cation availability (Table 4.2), levels of these nutrients were generally high in all sites and hence were considered unlikely to be limiting to tree growth (Tisdale et al. 1993). Potential N mineralization rates were similar in the two low P sites and higher in the high P site (Table 4.2) but once again, were not expected to be limiting to tree growth in any site. 1 established 140, 70 and 156 transplant plots in Carreta, Ojo de Agua and Arboleda, respectively, stratified across natural gradients in light availability in each site. Within each site, plots also naturally varied in available soil P. Each plot measured 1.5 x 1.5m, with a minimum distance of 1 m between plots. To maintain a consistent light and soil resource environment, vegetation was cleared from the plots and resprouts cut back at intervals. To reduce vertebrate herbivory, all plots were fenced to a height of 1.25m. 71 Table 4.1 Families, mean seed mass and growth habits of the five study species. Species Family Seed Purported Purported soil mass growth, shade- resource 6mg) tolerance associations d Astronium Anacardiacae 22.6a Slow-growing, Intermediate graveolens intermediate fertility shade-tolerant“d Cordia Boraginaceae 44.9 b Intermediate High fertility gerascanthus shade-tolerance, late successional b'd Dalbergia retusa F abaceae 77.4 a Slow-growing, Intermediate- intermediate high fertility shade-tolerant, late successional “‘6 Pachira quinata Bombacaceae 22.4 a Fast-growing, High fertility f shade intolerant, early successional C‘ ,r T abebuia rosea Bignoniaceae 22.6 a F ast-growing, Low fertility shade- intolerant, early successional 8’“ Sources: a: this study; b: Opler et al. 1975; c: Piotto et al. 2004; d: Eugenio Gonzalez, pers. comm.; e: Blair and Perfecto 2004; f: Cordero et al. 2003; g: Kitajima 2002; h: Hubbell 1979. 72 a...» 1 35 8:. 1 3.: 9.8 123 no 8 3mm .3 8 E to a 98 3 Ease mom A22 1 at 8.2: 1 on so 1 we 3 8 SK 5 h 2 3 h 2 3.» may a =8 2335 GR 1 3 $13. 1 8.3 3.: 1 we 2 h 8.: a... a «.2 3 a 3m 38 :3 § Eéfia am: 8282 awe. 8 6.6 56:8 38:. ass 65 a 23%.: :8 Ea maofiafi =9. sagas am: one 1 58 mm a as: m... use... . 2.33.? 183.3% .mémmhamvo 95.3.2 pm.~NHm.oam 1.3642 .v.cflo.mav.¢fl¢.mm.v.chm.m ave—OE a Sdflwwd u NN—flmfig a Ndmiumdm: 9 Racine— . Ehmfladaw a. Wmfiod: a mdflé _. 9.35.? p Ndhwd ~5w< at 2.0 a 8.36%? ~630me u ado—hedge . Ndflmdm . Wmmfinde . bSflNE u vdfimd . oéfifimm nmdflmé 305D 52 =8 a. \z 95 Sega... au 82 m: 0m m =2 5 -58 z 98 1 a8 5:38:82. :25 an A3: v a .<>oz$ 8% 5053 moococobmv ESE—awn Baum—o5 flute..— .mozm 33m 05 5 £8 E gag—8038 Z 3828 v5 hum—BER; ".230 «.9 03:. 73 In June-July 2000, I germinated seed of Dalbergia, T abebuia and Pachira to use in a transplant experiment. I collected Dalbergia and T abebuia seed from the vicinity and purchased Pachira seed from a commercial source that collected seed locally. To promote mycorrhizal inoculation, I germinated seeds in flats containing a mixture of 9:1 sandzfield soil, with soil from each site contributing 1/3 of the field soil component. Each germinant was transferred to a plastic nursery bag filled with soil from the site where the seedling eventually would be transplanted. In addition, I collected new germinants (i.e., seedlings retaining cotyledons) ofAstronium and Cordia from the field and planted them into bags using the same procedure as for the germinated seed. Seedlings of all species were grown at 70% full sun until they were transplanted into the experimental plots over two weeks starting 1 August 2000. Each plot had 2-4 individuals of each species, planted in randomly chosen positions of a 4x4 grid with positions separated by 50 cm. Each plot thus had 10-12 individuals, with some positions in the 4x4 grid remaining vacant. To correlate resource availability with seedling growth and survival, I measured light, soil P and soil moisture in each transplant plot. Resource levels in each site were as in Tables 4.3. I used hemispherical canopy photographs taken at 0.75 m above ground level to estimate light availability in each plot. In November 2000 (end of rainy season), I took one photograph in each quadrant of every plot using a Nikon Coolpix 950 digital camera and an 8 mm fisheye lens (Nikon, NY, USA). The images were analyzed using Gap Light Analyzer (Frazer et al. 1999) to obtain percent full sun light availability. Soil cores (composites of five 10 cm-cores) were taken from each plot in early November 2000 and the samples analyzed for gravimetric water content and for available soil P using Mehlich 3 extractant (Mehlich 1984, Tran and Simard 1993), a strongly 74 acidic extractant commonly used in soil testing laboratories to assess the amount of plant- available soil P (Pote et al. 1999). Because availability of soil nutrients, especially P, can vary temporally and even reverse ranks depending on season (Sollins 1998), I also sampled soils at the end of the dry season in June 2001. Soil P levels in the two seasons were correlated in the high (F091, p < 0.0001) and intermediate (r=0.8l, p < 0.0001) soil P sites. In the Carreta site, dry season soil P levels across all plots were uniformly close to the highest levels seen in the site during the wet season, likely because of reduced uptake by plants. For my analyses, I used soil P levels measured during the growing (i.e., wet) season. To monitor growth and survival, seedlings were censussed and their diameter and height measured over approximately 28 months. Seedlings were censussed in August 2000, November 2000, June 2001, November 2002, May 2002 and November 2002. I used calipers to measure stem diameter ~3 cm above ground level, marking the spot with permanent ink to reduce subsequent measurement error. On 30 August 2000, half of the plots were randomly selected in each site and fertilized by adding 96 mg P kg'I soil in the form of rock phosphate. This fertilization level was based on the highest level of available P measured in the preliminary soil survey. However, subsequent soil sampling ~ 2 months later did not reveal any differences in soil P between the control and fertilized plots, likely due to the loss of P fertilizer following heavy rains shortly after fertilization. Further, there were no differences in growth rates or mortality between seedlings in control and fertilized plots for any species or site. Hence, data from fertilized and control plots were pooled for all analyses. 75 Data analysis Seedling growth rates were calculated using a discrete time compound interest formula: Diameter, = Diameter0.[1+G(res)]t Eq I where Diameter, is seedling diameter (in mm) at time t (in years), Diametero is the initial seedling diameter and G(res) is growth rate as a function of resource availability. Seedling growth was modeled using the Michaelis-Menten function, which has been commonly used to model growth as a fimction of light availability (Pacala et al. 1994, Kobe 1999): A * Light Eq 2 ‘% + Light l where A corresponds with asymptotic growth and S corresponds with the slope of the G(res) = growth function at low light. Effects of soil P and soil moisture were tested by specifying A and S as functions of these resources which enables partitioning soil resource effects to low light or high light growth (Bigelow and Canham 2002, Kobe, in press): . . . . 60‘“) = 2:15:12, Light /S +L1ght Eq 3 ,, . G res = A A Light- /c*soil+L1ght Eq 4 When both A and S are substituted to model a soil resource effect on both asymptotic and low light growth rates, the function reduces to a two parameter model of the form: ' t It ' E 5 G(res)=sqbll b 'nght q h/c +Ltght) 76 Soil resource effects were also modeled using modified bivariate Michaelis-Menten functions of the form: A * Light * soil A + Light * [—A— + Soil] Slight Ssoil where Sresource is the slope of the growth curve at the zero level of the subscripted G(res) = resource. Thus, a set of seven equations was used to test functional relationships between resource availability and growth for each species. The basic equation incorporated only the effect of light on growth while the more complex equations added the effects of either soil P or soil moisture, as acting on grth at high light, low light, or both. Equations were fit using the Gauss-Newton method in the non-linear procedure of Systat (SPSS Corp, Chicago, IL). Models were selected on the basis of Akaike’s Information Criteria (Hilbom and Mangel 1997, Burnham and Anderson 2002). The AIC is an estimate of the expected, relative distance between the fitted model and the unknown mechanism that generated the data (Bumham and Anderson 2002). Unlike likelihood ratio tests (LRTs), AIC can be used for selection among non-nested models. The model with the lowest AIC is chosen as the best approximating model from the set of candidate models. In general, AAIC > 2 indicates poorer support for one model over the other (Bumham and Anderson 2002). For Arboleda only, where maximum light availability was < 30% full sun and hence may have precluded asymptotic growth rates, I also modeled the effect of light on growth rate as a linear function: 77 G = M *light + constant Eq 7 Similarly, for the low soil P Carreta site, soil P was also modeled as affecting growth rate in a linear fashion. These linear models were not used in other sites because at higher availabilities, resources were not expected to have linear effects on species growth. For Astronium and Pachira, because of very poor survival beyond the second dry season, I estimated growth models using data from the first 15 months only. Models for other species were based on growth over 28 months. I used maximum likelihood and survival analysis to fit models of mortality as a function of resources. I used an exponential distribution of survival times and an exponential model for the hazard function, resulting in: 4‘1 P(m)=1—e Eq8 ,1 = A.e —(B"Light+C"soil P+D‘soil moisture) Eq 9 A, B, C and D are species-specific parameters estimated from the data (Kobe 1999). Models were fit using program code written in Borland Delphi by Richard Kobe. As before, I used AIC to select the best approximating model. I mapped uncertainty in parameter estimates (95% support) to uncertainty in functional relationships and assessed species differences as regions where 95% support for the functional relationships do not overlap. To evaluate habitat-level species growth and mortality responses to variation in resource availability, 1 first tested if functional relationships were site-specific by comparing AIC values of models that did not differentiate between sites with models that were site-specific. Where AIC revealed site differences, I analyzed site-specific species responses to resource environments. For all species, to explore landscape-level species 78 variation in growth and mortality, I fit models to data that were composited from all three sites. Results Growth Light and soil resources influenced growth rates in all species (Table 4.3). All species showed strong positive growth responses to variation in light. Astronium and Pachira were more responsive to soil P, T abebuia and Cordia to soil moisture, while Dalbergia responded to both soil P and soil moisture (Table 4.4). In Carreta (the site with lowest P), three of five species growth rates responded to available soil P. Models incorporating effects of soil P on growth had lower AIC values (and hence were better fits) in Astronium, Dalbergia and Pachira. In these species, soil P acted to increase diameter growth rates at all light levels, although the form of the function was different in the three species. In Astronium and Pachira, soil P influenced both low-light and asymptotic growth rates (Eq 5) (Table 4.4) while in Dalbergia, a double Michaelis-Menten best characterized the relationship between soil P, light and growth rate. Hence, for all three of these species, light and soil P co-limited growth so that an increase in either light or soil P resulted in an increase in growth rate. In Astronium, Eq 3, characterizing multiplicative soil P effects only on high light growth also had substantial support. In T abebuia, the unmodified Michaelis-Menten and the function modified to include soil moisture alone, or both light and soil moisture effects on growth provided equally good fits: AAIC for all three functions were < 1. In Cordia, light alone determined growth (Table 4.4). 79 : :0 .88 $5. .88 Em: + 938 1 S 2:. an 2.8% m< 58.88 83.3: 8.3 +§138 x: :5 3 5o 32 $85. 38.2 «3803 63.8.8 86.3.3 .33 $31.38 N; as 3 88 a: 3.0 $02.21... $8.88 €3.88 3.8 +33%» and 3.9 3 83 38 8.83.8 A8.o.8o.8 £3 .88 $0.: + 388 c :3 8 8o 38 8.8.53 <0 56 .88 :3 .8: $0.: 1.88 1 5.0 v: :8 :.~ $.85. <0 00 38.8.3 38.8.8 1 $8 8 2.7 a; :3372 m< a: .28 82 .88 $0: 1.3.80 «2 a... a 88 0.3 smarts 23.8.8 23.8.8 «$0.: +£1.03» a :8 a «3 8.4 Ewarw <0 83 .83 an“ an: 30.: +3.88 9: 3o 2 :8 02 Encarta 3.8 .83 $003.: 3.8 + 388 o 38 2 :8 =3. Earta <0 m< A8§_ Ens. A8§_ 3?: 35 was: Ea: €85 toga—a toga—a €258 ton—mam toga tongue. toga toga—am $8 .88 $80 .88 .58 $8 $8 $8 024 m z N z in a. 6 m a < 862 85 am .m as 1 m< 8 $868.88 30< 2. 6.6 1 <0 8 so: 850 1 <0 .8363. 1 <0 Sussex 1 <._ 8.8.88 1 <0 8880 1 00 ..§..E§..< 1 m< gauge 05 go «588 3 E 05 bee Sch 5% win: 080.58 ob? £0008 ehauum 05 33:08:. 608m be 0:83. 05 “Eamon 2335.." noon mo 038m $53235 3508. 8 860% .«o 802 £32» .8255. wean—2 .NV 024 .23 £0008 =u 30 03 23 Au: 353.5800 00 $36808 .2 03m 8388 £1385 «8&3 $3 05 838a? .8620 085—00:— 858082 v.9 033—. 80 and «3-0 3N... .28 1 $5 2 03. a; $302.: a... .85 an: 0.3 .38 38 + 38 :8 So mm 33 ”3 Ema... 3.3 5.3 and .88 381.58» .0 3° .0... £5 .3 $02... <0 0.3 5.. .88 $82.80 1 .3 8 .880 N3 2.. Emmi.» <0 <0 23. 23.8.8 1 50 3.3.3 $5. 8... 7.30.72 m< 23.308 83 «“8 30:13.88 «3 and E :3 $5 Ema... 03.398 2 :0 .88 30.: + 38» o 8.0 E 85 8... 8.8.33 <0 5+ G3 .3: a2 ._ _.8 0.3 .28 $8.3»..Zmiy 1 v2 8 Ed :5 8..” Emmi... <0 <0 3E: 33. 3.3 3.8: 3E: 35. 38.: 3E: toga 0.8095. roams... toga toga toga tongs... toga $8 .\.8 .\.8 .\.8 .\.8 .\.8 $8 $8 0.<< N. z N S 3w .0 u m a w .232 2% dm A388 3. 2.3. 81 2212.3 55.8.8 v2 and ”N 85 88 N +38% 82“ 5... .38 8.2-8 «3»: + «mi» a V2 «N 86 3.: Emfw 5. :3 .88 $2: 3m..~+3.u\§ s... 86 3 So .38 86 38% so... .88 GS .88 331388 c 86 s :3 25 Emii <0 «3+8...» $3 8.3 $3 .:.8 Es .88 . is .518 3... $5 3 m8 :5 v3 Emafi§ so... .88 £2 .38 3.8 +3.83» ”3 m3 2: 86 a3 Ema... a; .28 8mm .88 3m: + «mi» c and a: 25 can $8.... <0 5 3E: 53: 3E: 35 33. 35 Ea: 3E: conga ton—95 comma... tonne... toga—am toga «8&3 #253 $8 $8 $38 $8 $38 $8 $8 $8 024 m z N z 3m a. u m a w .982 35 in 8:58 v... 2%... 82 At very low levels of soil P, species’ growth rates ranked as T abebuia > C ordia > Dalbergia = Pachira (Figure 4.1A). Because of large overlapping confidence intervals on parameter estimates, growth rates in Astronium were not different from other species. With marginal increases in available soil P, growth rates of Astronium and Pachira increased substantially while Dalbergia’s growth rates increased to be equivalent with Cordia’s (Figure 4.18). Growth rate ranks were thus maintained across all light levels but not across nutrient levels. In Ojo de Agua (the site with intermediate soil P), in most species, growth rates depended on soil moisture rather than available soil P. In Cordia, Dalbergia and T abebuia, Eq 5 characterizing linear soil moisture effects on both low- and high-light growth rates provided the lowest AIC, and hence the best fits (Table 4.3) although in all three, other equations modeling combined light and soil moisture effects also provided equal support. In Astronium, growth responded to both soil P and light, with soil P modifying growth rates at low light. Mean growth rates in this species were at least an order of magnitude larger than in other species but significantly different from only Cordia. A model with soil P modifying high light grth also had substantial support according to the AIC (Table 4.4). However, both models had poor fits, with r2 < 0.14. In Pachira too, models incorporating combined light and soil P effects had the lowest AIC values but all models had poor fits, with r2 values ~ 0.05. 83 4 A Carreta, 0.5 mg g'1 P B. Carreta. 2.3 m9 9" P ‘0‘ AS -6- CO A DA -0 PA —. TA o I 0 4o 60 80 ‘T'_ C. Ojo, 5 mg g'1 P, 14.9% moisture D. Ojo. 9.7 mg g‘1 P, 31.7 % moisture >. g 4 - 4 - 2 3’ 3 - 3 « o ‘5 E 2 - 2 - .9 u .. g 1 ‘1 . ’— - - 1 " 'fi 4¥;_ _ _ g 0 Y I 1 I o I I 0 1o 20 30 4o 0 1o 20 30 40 E. Arboleda P. All sites, 5 mg g-1 P, 30% moisture 5 - 0 1b 2b so 0 2b 40 so 3‘0 Light availability (%full sun) Figure 4.1 Models of diameter growth of seedlings (yr") across light levels in the study sites: A) Carreta, with 0.5 mg kg’1 soil P; B) Carreta, with 2.3 mg kg'1 soil P C) Ojo de Agua, with 5 mg kg'1 soil P and 14.9% soil moisture; D) Ojo de Agua, with 9.7 mg kg" soil P and 31.7% soil moisture; E) Arboleda; and F) across all sites, with 5 mg kg'1 soil P and 30% soil moisture. Error bars represent 95% confidence limits at selected light levels. 84 Ranks of mean growth rate changed between low and high light in Ojo de A gua, although overlapping confidence intervals on estimated parameters indicate that these differences were not significantly different (Figures 4.1C and D). Asymptotic growth rates ranked as Astronium > T abebuia > Dalbergia > Cordia although the difference was significant only between Astronium and Cordia. In Arboleda, all species growth rates were determined by light availability alone (Table 4.4). Cordia and T abebuia had a saturating relationship between light and growth rate (Table 4.4). The other species had a linear relationship between light and growth rate, likely because light levels in this site did not exceed 27%. Beyond ~7% light availability, Pachira and Astronium had the highest growth rates, and Cordia the lowest. At lower light levels, T abebuia grew fastest (Figure 4.1E). However, none of these differences were significant (Figure 4.1E). For all species, models that pooled data from all sites had substantially higher AIC values (and hence poorer fits) than site-specific models indicating that the relationship between growth and resources varied with site. However, we pooled data across sites to obtain growth-resource relationships that would approximate species behavior across broader resource variation than that provided by the site-specific models. Data pooled across sites revealed broad differences between the species which in most cases paralleled the site-specific responses. Pooling across sites, for both Astronium and Pachira, a Michaelis-Menten fimction modified to incorporate available soil P effects on asymptotic growth rate had the lowest AIC values (Table 4.5). Dalbergia and T abebuia had growth rates dependent on both light and soil moisture while Cordia’s growth rate was governed by light alone (Table 4.5). At very low soil resource levels, such as at 0.5 85 mg kg'1 P and 15% soil moisture, and very low light availability, Cordia growth rates were higher than T abebuia and Pachira growth rates while all other species growth rates were not significantly different from each other (figure not shown). When light availability was greater than ~lO%, Pachira had significantly lower asymptotic growth rates than all other species. Astronium had nonsignificantly highest growth rates from 0% to 60% fiill sun after which T abebuia growth rates were nonsignificantly higher. With marginal increases in soil resources, grth rates ranks changed so that under high light availability, Astronium = Pachira > T abebuia > Cordia = Dalbergia (Figure 4. 1 F). Because soil moisture acted on low light growth in T abebuia and Dalbergia, further increases in soil moisture led to these species achieving higher low light growth rates than Cordia and Pachira, although they were not significantly different from Astronium growth rates even at 80% soil water availability (figure not shown). Table 4.5 Maximum likelihood parameter estimates and 95% support intervals, sample size (N), coefficients of determination (r2), AIC and AAIC for all models with AAIC <2, relating relative diameter grth rates of species to resource availability across sites. Abbreviations are as in Table 4.4. Sp. Model Param Parameter values N r2 AAIC (95% support limits) AS MM(HL soilP) 4.404 (2.7908018) 57 0.25 -- 0.182 (0135,0230) 2.040(1.855, 2.224) 301 0.45 -- 0.135(0.117,0.153) l.964(1.732, 2.196) 250 0.27 0 0.005 (0.004, 0.007) 3.243(l.860, 4.625) 250 0.27 1.57 0.253 (0.170, 0.337) 0.183 (0.077, 0.288) 1.680 (0.732, 2.628) 49 0.48 -- 0.153(0.131,0.174) 2.720(2.481,2.958) 234 0.53 -- 0.006 (0.005, 0.007) CO MM DA MM (LL soil moisture) Double MM (soil moisture) PA MM (HL soil P) TA MM (LL soil moisture) o agree-32280 Amatmer' 86 Mortality In contrast to growth, where site-specific responses were important, mortality as a function of resource availability did not vary across sites: in all species, models that did not differentiate between sites had far lower values of AIC than site-specific models. In further contrast to growth responses which were sensitive to both light and soil resource availability, light alone affected mortality in all species (Figure 4.2). In all species, mortality decreased with an increase in light availability, except in Pachira where it increased slightly with light (Figures 4.2 and 4.3). Support bands (95%) on the relationship between mortality and light availability for Astronium overlap with those of Pachira indicating that Astronium’s mortality as a function of light was non-significantly lower than Pachira at all levels, although it was significantly higher than the remaining species. From O-40% full sun, corresponding to understory to large gap light conditions, the mature forest species Cordia had significantly lower mortality than all other species. Dalbergia and T abebuia had very similar mortality responses to light availability although from above 65% full sun, such as might be found in abandoned pastures, the slower-growing Dalbergia had lower mortality than the fast-growing Tabebuia. 87 0.8 - 0.6 - 0.4 - 0.2 - Probability of mortality over study period 0.0 4+5 \ W“ \ Ni ii ii :8 \i“i\l\ I T I I ‘I I I ‘l 10 20 30 4O 50 60 70 80 90 Light availability (% full sun) Figure 4.2 Mortality as a function of light availability for the five study species. Abbreviations are as in Table 4.4. A. Low light (< 7.5% full sun) B. High light (>15% full sun) .27 27' 4 ' :47; : ,, 0.8 _ : , {x g " c » ' ‘2’ E 0.0 - — 3 S 'E 0.4 - - 8 2 ‘L 0.2 « . 0.0 a "W“ " ' — —' « 3§§§§«1«m28§§§§$§§8 Months Figure 4.3 Time course of proportion of seedlings surviving under A. Low light (< 7.5% full sun) and B. High light availability (>15% full sun). Shaded portions show the dry season . 88 In all species, mortality was highest over the first dry season (Figure 4.3). Pooling data from all light levels and sites, across species, survival over the first wet season and the first dry season were strongly positively correlated (r=0.89, n=5, p=0.04) but there was no relationship between survival over the second wet season and the second dry season (r=0.43, n=5, p=0.46). These relationships held within higher light levels (>15% full sun). Under low light (< 7.5% full sun), there was also no relationship between first year wet and dry season survival among species (r=0.6l, n=5, p=0.27). The lack of a correlation in low light is likely due to the change in ranks between survival at the end of the first wet season and at the end of the first dry season. In November 2000, before the onset of the dry season, species mortality decreased in the order Cordia < Astronium < T abebuia < Dalbergia _ (U 3 r \ 3:" \ \ \ 0 a) . = 2 “ V O \ \ 2 a 0‘ \ \ V E — — — ' 1 7 \ \ O -— 1 ‘ o 0 O \ g 0 . \ 0 g o V I U V I I I 0 T 1 T r j I T ‘9 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 8 i; c. Arboleda o. All sites V 5 g 0 . 5 1 5 -o— 0.5 mg kg“ P. 15 % H20 9 4 5 \ .9. 10mg kg P". 30% H20 3 4 N -v— 20 mg kg P". 4O % H20 3 i 0 ‘ \ m 3 ‘1 3 _, . \\ E . \ .9 \\ “O 2 “ 2 .1 \ g . 8\\ ' O r: 0 1 r . o 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0,0 0,1 0,2 0,3 04 0,5 0,6 0,7 0.8 Probability of survival under low light Figure 4.4 Trade-off between growth under high light availability (30% fiill sun) and survival under low light availability (5% full sun) as influenced by soil resources in: A. Carreta, with soil P varying (low: r = 0.57, p = 0.31; med: r = -0.03, p = 0.96; high: r = - 0.68, p = 0.21); B. Ojo de Agua with soil moisture varying (low: r = -O.87, p = 0.12; med: = -0.76, p = 0.24; high: r = -0.67, p = 0.33); C. Arboleda (r = -0.99, r = 0.0006); and D. Across all sites with soil P and moisture varying (low: r = -O.7l, p = 0.17; med: r = -0.89. p = 0.04; high: r = -0.92, p = 0.03). Low, medium and high soil resource levels are represented by solid, long and short dashed lines respectively. 90 Across sites, species, and light levels, there was also a strong negative correlation between growth in the first wet season and survival in the first dry season (r=-0.97, n=5, p= 0.0039): Astronium and Pachira had the highest growth rates during the wet season and the lowest survival over the dry season. Conversely, Cordia had the lowest growth rates over the wet season but the highest survival over the dry season. However, contrary to Pearson et al. (2003), we found no relationships between wet season growth in high light and dry season survival in low light for either the first (r=-0.51, n=5, p= 0.38) or the second (r==-0.33, n=3, p= 0.58) wet and dry seasons. I determined an index of performance of the five species based on their growth rates and survival in each site: the product of the probability of survival multiplied by growth rate of each species, representing ‘effective growth’, integrates survivorship and growth into a single metric and hence provides a better indicator of species performance than either taken singly. Figure 4.5 shows effective growth of the species at the three sites and reveals partitioning of both light and soil resources. In all cases, analyzing species performance in this manner revealed patterns different from those shown by growth alone, or mortality alone. In the low soil P Carreta site, at 0.5 mg kg'l P, close to the lowest levels found in the site, Cordia dominates at light levels up to 60% and T abebuia after that. Because of poor survival as well as relatively low growth, Pachira performed worst across all light levels in this site (Figure 4.5A). However, with marginal increases in soil P, performance ranks changed. Cordia retained first rank until 40% full sun, but Astronium was expected to perform better at higher light availability (Figure 4.5 B). With further marginal increases in soil P at this site, Astronium dominated at all light levels. Cordia performance ranked second when light was < 40% but Dalbergia performed 91 better at higher light levels (figure not shown). Astronium also dominated at all light levels in the intermediate fertility site Ojo de Agua when soil resource levels were close to or at the lowest levels found in the site (Figure 4.5C). However, performance ranks again changed with soil resources. When soil P was 9.7 mg kg" and moisture 31.7%, the average levels found in the site, Cordia was expected to outperform all other species when light was below ~15%, but Dalbergia dominated at higher light levels (Figure 4.5D). In the high fertility site Arboleda, Cordia and T abebuia were expected to co- dominate at light levels below ~15%, but Dalbergia performed best at higher light levels (Figure 4.5E). I also looked at species performance on a non-site specific basis (Figure 4.5 F). At this scale too, performance ranks of the species depended on both light availability as well as soil resources. When soil P and moisture were very low, Cordia dominated and Pachira was expected to perform the worst at all light levels. However, with marginal increases in both soil resources, Cordia still performed better than other species below 35% full sun, but Astronium dominated at higher light levels. 92 A. Carreta, 0.5 mg kg’1 P B. Carreta. 2.3 mg kg'1 P 2.0 - _ AS 2.0 -i — - co ~~ DA - 1.5 ‘ __ PA // 1.5 ‘ /// 1.0 '7 // 1.0 a / 0 5 « // i 0 5 00 Tfif 0.0 I I I I 0 20 40 60 80 o 20 40 60 80 C. 010. 5 mg kg-1 P. 14.9% soil moisture o. Ojo, 9.7 mg kg"P, 31.7 % moisture A 1.6 1.6 L— s 1.4 — 1,4 - g 1.2 "' 12 —1 g 1.0 « 1.0 _ 5 0.8 - 0-8 1 a) . _ g 0.6 0.6 :6 0.4 "‘ 0.4 "‘ ‘— __ .__._. a1 0.2 7 0.2 - / *” u / g 0.0 r I I l 00 I I I T E 0 10 20 30 4o 0 10 20 30 40 E Arboleda 2.0 - 2.5 - 1.5 ~ 2.0 - 1.5 4 1.0 - 1.0 - 0.5 ~ 0.5 ~ 0-0 0.0 T . . T 0 35 40 o 20 40 60 80 Light availability (% full sun) Figure 4.5 Effective grth (annual diameter growth of seedlings multiplied by survival), across light levels in A) Carreta, with 0.5 mg kg" soil P; B) Carreta, with 2.3 mg kg" soil P C) Ojo de Agua, with 5 mg kg" soil P and 14.9% soil moisture; D) Ojo de Agua, with 9.7 mg kg" soil P and 31.7% soil moisture; E) Arboleda; and F) across all sites, with 5 mg kg" soil P and 30% soil moisture. 93 Discussion Resource eflects on growth and mortality Our results show clear effects of soil resources and light availability on the growth of seedlings. Astronium and Pachira consistently had higher growth rates under higher available soil P, Dalbergia and T abebuia to soil moisture and Cordia largely to light availability. In contrast to growth, mortality responded to light alone. Based on survival under low light, the mature forest species Cordia appears to be the most shade-tolerant of our species, while Astronium and Pachira are both shade-intolerants. The two most shade- intolerant species consistently had higher growth rates in sites with higher soil P. When both soil nutrient levels and light availability were high, Astronium and Pachira were also the fastest growing of our study species. The nutrient response by these shade intolerant, fast growers suggests Astronium and Pachira are competitors sensu Grime (1979), specializing in resource capture and growth. Higher nutrient uptake in these species would be required to support their higher growth rates (Chapin I991), higher respiration (Bazzaz 1979) and shorter leaf life spans (Reich et al. 1992). However, Astronium is reported to be a slow grower (e.g., Piotto et al. 2004), and the abundance of long-lived Astronium saplings found in the understory (pers. obs.) suggest the species may be a ‘late tolerator’ (Oldeman, 1990, Grubb et al. 1996), requiring high light for establishment but able to persist for long periods in the shade as saplings and adults. Based on species comparisons across sites, the most shade tolerant species, Cordia, can be characterized as a species that responds only to light, although in the intermediate fertility site, Cordia growth rates also increased in response to soil moisture. 94 Based on survival in low light, Dalbergia and T abebuia appear to be intermediate in shade tolerance. T abebuia consistently and Dalbergia largely respond to soil water rather than to nutrient availability. The lack of a growth response in these species, coupled with their lower grth rates under high light and soil resource availability, especially in Cordia, suggest these species may be stress tolerators, sensu Grime (1979). It has been hypothesized that shade tolerants take up P when its supply is greater than demanded by growth rates, using it when conditions allow higher growth rates (Burslem et al. 1995). Such ‘luxury consumption’ of P is an important mechanism by which species tolerate low P environments (Rorison 1968). However, in my species, grth rates in Cordia, Dalbergia and T abebuia do not increase substantially under higher P even when light is not limiting, suggesting that this mechanism does not operate in these species. All species except Pachira had greater survival under higher light availability. Increases in mortality with light, as seen in Pachira, are not uncommon in the tropics (e.g. Nunes et al. 1993, Gerhardt 1996a, Kobe 1999) where higher light can have deleterious consequences due to photoinhibition (Long et al. 1994). Alternatively, higher mortality under higher light availability could be due to an increase in herbivory in higher light environments. Although I fenced all plots to minimize vertebrate herbivory, we did not exclude invertebrate herbivory, which in this forest, is likely higher in higher light environments (Gerhardt I996b). I did not directly measure leaf loss to herbivores, but field observations suggested young leaves in Pachira were more susceptible to herbivory than in other species. 95 Species trade-ofifs in growth rate and survival I found a strong trade-off between species growth under high light availability and survival under low light availability. Such trade-offs in performance under different light environments have been found in both tropical (Kitajima 1994, Kobe 1999) and temperate trees (Kobe et al. 1995, Lin et al. 2002) and in temperate forests can facilitate species coexistence (Pacala et al. 1996). Furthermore, soil resources influenced the growth-survival trade-off. Because higher levels of available soil P led to higher asymptotic grth rates in fast-growing species only, while survival remained unaffected in all species, higher levels of available soil P led to stronger trade-offs between high light growth and low-light survival. The trade-offs in grth and survival among the species led to clear differences in performance across resource environments (Kobe 1999). As light and soil resource availability change, the dominant species in each environment changes. Significantly, performance ranks of species in our study often did not change with light, but with soil resource availability. Further, our conclusions on species performance would have been very different if we had looked at grth alone, highlighting the importance of looking at mortality and grth simultaneously across gradients in resource availability. While previous studies have found significant differences among species in growth and survival in response to light and nutrient availability, differences in growth and mortality alone cannot be taken as evidence of resource partitioning. Differences in growth and survival can lead to coexistence only if changes in resource environments lead to changes in species' performance ranks (Latham I992, Kobe I999). Kobe (1999) showed light 96 gradient partitioning among four neotropical seedlings. Here, I demonstrate the effects of soil resources as well as light on niche partitioning among tree seedlings in a dry tropical forest. The changes in performance ranks of species (taking both growth and survival into consideration) across spatial heterogeneity in resources provides the prerequisites for species coexistence through resource gradient partitioning in this forest. Drought eflects on growth rate and survival As expected, I found great variation within species in growth and mortality across seasons. Mortality during the first dry season was greatest, between 25 and 80%, which is higher than that reported in a moist forest in Panama (Pearson et al. 2003) but similar to that reported in other dry forest studies (Gerhardt 1996). Seasonal drought had differential effects on mortality in the species. Astronium was more strongly affected by the first dry season, showing disproportionately high mortality compared to growing season mortality, an effect we would have missed in a short-term study. Across all light environments, wet season growth and dry season mortality were strongly negatively correlated. However, unlike an earlier seasonal forest study (Pearson et al. 2003), there was no relationship between wet season grth under high light and dry season survival under low light because of disproportionately low Astronium survival over the first dry season. Experimental limitations First, because my fertilizer treatment to increase soil P did not result in differences in available soil P levels between control and fertilized plots, my study was correlative. My 97 study focused on soil P as the nutrient that most likely limited growth in the study sites. However, levels of other soil nutrients also differed, though nonsignificantly, possibly leading to the differences in species growth among sites that could not be attributed to measured resources. Unmeasured variables such as rockiness could also have led to species growth differences among sites. Second, the expected high mortality during the dry season suggests dry season soil moisture is an important factor that influences survival in these species. However, due to logistical problems (unexpected changes in Costa Rican visa regulations), I was unable to measure soil moisture during the dry season. Third, although the Mehlich 3 extractant used is one of the most widely used extractants for measuring plant available P, chemical-extraction assays may not truly capture P availability to plants (Sollins 1998). Finally, although tropical dry forests are less speciose than rainforests, because my study used only five species, extrapolation to other species must be done cautiously. Conclusions This study demonstrates that seedlings growing in seasonally dry tropical forests experience multiple resource limitation, with their growth rates responding to availability of light, soil P or soil moisture. Light effects on survival combined with soil resource effects on growth led to a segregation of species along a resource axis so that no one species dominated all resource levels. The competitive hierarchy of species changed with both light and soil resource levels, and hence provided the prerequisite for coexistence of seedlings of these species through resource-based niches. 98 Acknowledgements We thank the Ministerio del Ambiente y Energia, Costa Rica for granting permission to carry out the research, and the Organization for Tropical Studies (OTS) for the use of their field station. MI would like to thank Francisco Enrique Granados, Sandra Quiros, Justin Harbison, Lynn Holcomb, Nicole Turner and Mauricio Solis for assistance in the field. This work was supported by grants from OTS and the Taraknath Das Foundation to M1, and partly by grant DEB 0075472 from the National Science Foundation to RKK. 99 Chapter 5 Stored carbohydrates in seedlings in a tropical dry forest: effects of light and soil resource availability 100 Introduction The ability of some species to persist in low-resource environments usually trades off with their ability to grow fast under high resource availability (Kitajima 1994, Kobe et al. 1995, Schreeg et al. 2005). Thus, shade-tolerant species are able to survive low-light environments but are unable to grow as fast as shade-intolerants when light availability is high (Kobe et al. 1995, Lin et al. 2002). For species that maximize survival rather than growth, especially under low resource availability, allocation of resources to defense and to storage (Kitajima 1994, Kobe 1997) rather than grth could enhance chances of survival. Allocation to storage of carbohydrates is generally at the expense of growth (Chapin et al. 1990). Stored carbohydrates in temperate seasonal forests are considered essential for species’ survival and play important roles in respiration, overwinter survival and recovery from defoliation (Chapin et al. 1990, Kozlowski 1992). Storage is particularly important for survival through low-resource availability periods in seasonal environments (Chapin et al. 1990). In temperate forests, shade-tolerant species have higher total nonstructural carbohydrates (TNC) than shade-intolerant species (Kobe 1997), likely allowing them growth spurts in response to light gaps (DeLucia et al. 1998). TNC also enables recovery from herbivory (Chapin et al. 1990, Kozlowski 1992). It is hence likely to be important in tropical forests where herbivory is a leading cause of seedling mortality in low-light environments (Kitajima and Augspurger 1989). TNC has seldom been explored in tropical forests (Tissue and Wright 1995, Marquis et al. 1997) although recent studies have shown substantial variation in storage among species in seasonally dry tropical forests (Newell et al. 2002, Wurth et al. 2005). 101 In addition to species differences, spatial and temporal variation in resource availability can affect the storage of carbohydrates. Storage has been shown to increase with light availability (Mooney et al. 1995, Gansert and Sprick 1998) — although others have shown no light effects on storage (Kobe I997, Canham et al. 1999) — and decrease with nutrient availability (Mooney et al. 1995). These studies suggest nutrient availability and light are both important in the field where trees can experience limitations in several resources that vary simultaneously. In this study, I examined species differences in allocation to root TNC (generally the site of greatest TNC concentrations (Loescher et al. 1990)), and its relationship with growth and survival under varying light and soil resource availability among five co- occurring species in a seasonally dry tropical forest. The questions I addressed were: I. What are the magnitudes of root TNC storage and how do they change with resource availability? 2. What effects does seasonality have on root TNC concentrations? 3. Is TNC positively related to survival and negatively related to growth? Methods Study site and species The study area is in the Palo Verde National Park (10°21 ’N, 85°21 ’W), Guanacaste, Costa Rica, a tropical dry forest according to the Holdridge (1969) life-zone classification. The annual average temperature is 27.4°C and rainfall 1817 mm, with a pronounced dry season from December to May (Jimenez et al. 2001). The semi- deciduous forests in the 20,000 ha Palo Verde Park have approximately 65 tree. shrub and liana species per 1000 m2 or 0.1 hectare (Gillespie 1999). 102 Table 5.1 Families, mean seed mass, leaf phenology and grth habits of the five study species. Specres Family (3:18;; mass Phenology c Growth habit Astronium Anacardiacae 22.6 a Leaf-exchanging; Slow-growing, graveolens leaves emerge intermediate shade- immediately afier tolerant leaf shedding during the dry season. C ordia Boraginaceae 44.9 b ? Intermediate shade- gerascanthus tolerance, late successional Dalbergia Fabaceae 77.4 a Brevideciduous; Slow-growing, retusa photoperiod intermediate shade- increase induces tolerant, late leaf flush end successional March Pachira Bombacaceae 22.4 a Deciduous; Fast-growing, quinata photoperiod shade intolerant, increase induces early successional leaf flush end Apr/early May Tabebuia Bignoniaceae 22.6 a Deciduous; leaf Fast-growing, rosea flush after rainfall shade-intolerant, early successional Sources: a: this study; b: Opler et al. 1975; c: Borchert et al. 2002. I selected five species for the study - Astronium graveolens Jacq., Pachira quinata (Jacq.) WS Alverson, Cordia gerascanthus L., Dalbergia retusa Hemsl. and T abebuia rosea (Bertol.) DC. — based on seed availability and to present a range of seed sizes and growth habits (Table 5.1). Species appear to show some soil resource affinities (Table 4.1, Chapter 4) though all co-occur as adults in the secondary forests of the Park. Astronium is a leaf-exchanging species, Dalbergia is brevideciduous, (Rivera et al. 2002) and all other species are deciduous as adults although the period of leaflessness varies (Table 5.1). Hereafter, species are referred to by their genus names only. 103 Experimental methods I used three sites varying in soil phosphorus (P) availability for my study. All sites were in secondary forest areas and within 3-6 km of each other. The sites provided a natural gradient in soil P with site Carreta having the lowest, Ojo de Agua intermediate and Arboleda the highest levels of soil P. As part of an experiment investigating resource effects on grth and survival in these species (Iyer et al., in prep., Chapter 4), I established I40, 70 and 156 experimental plots, respectively, in each site, stratified across light and soil P availability. Each plot measured 1.5 x 1.5m, with a distance of at least I m between plots. To maintain a consistent light and soil resource environment, vegetation was cleared from the plots and resprouts cut back at intervals. To minimize vertebrate herbivory, all plots were fenced to a height of 1.25m. In June-July 2000, l germinated seed of Dalbergia, T abebuia and Pachira. We collected Dalbergia and T abebuia seed from the vicinity and purchased Pachira seed from a commercial source that collected seed locally. To promote mycorrhizal inoculation, I germinated seeds in flats containing a mixture of 9:1 sandzfield soil, with each site contributing 1/3 of the field soil component. Germinants were transferred to plastic nursery bags with soil from the site where they would eventually be transplanted. I collected new germinants (i.e., seedlings retaining cotyledons) of Astronium and C ordia from the field and transplanted them into bags using the same procedure as for the germinated seed. Seedlings of all species were transplanted into the experimental plots over two weeks starting 1 August 2000. Each plot had 2-4 individuals of each species. and hence 10-12 individuals planted in randomly chosen positions of a 4x4 grid with positions separated by 50 cm, with some grid positions remaining vacant. 104 To test for relationships between allocation and resource availability, I measured light, soil P and soil moisture in each transplant plot as described in Chapter 4. Resource levels in each site are as in Tables 4.2 and 4.3, Chapter 4. On 30 August 2000, half the plots (randomly selected) in each site were fertilized by adding 96 mg P/kg soil in the form of rock phosphate. This fertilization level was based on the highest level of available P measured in the sites during a preliminary soil survey. However, subsequent soil sampling 2.5 months later did not reveal any differences in soil P between the control and fertilized plots. Further, there were also no differences in growth rates, mortality or storage of carbohydrates between seedlings in control and fertilized plots for any species or site. Hence, data from fertilizer and control plots were pooled for all analyses. To investigate intra- and inter-specific variation in root storage, and to determine how allocation responds to seasonality, I harvested seedlings at the end of the first three wet seasons that the seedlings experienced: November 2000, November 2001 and November 2002. We had one harvest at the end of their second dry season in mid-May 2002, a week before the first rainfall of the wet season. Because of the lower number of Astronium and Cordia seedlings available for transplanting at the start of the experiment, we harvested seedlings of Dalbergia, Pachira and T abebuia only. The first three harvests were restricted to the high and low P sites only. Large-scale mortality in Pachira led to very small sample sizes from the second and subsequent harvests. To allow broader species comparisons, I also harvested Cordia seedlings in May 2002. For the final harvest in November 2002, I harvested seedlings of all five species from all three sites. 105 Seedlings to be harvested were randomly chosen. Although I attempted to excavate all seedlings’ root systems completely, because of the difficulty involved in excavating roots from the high-clay soils, we only excavated roots to a depth of about 80cm. Seedlings were washed in tap water, separated into leaves, stems and roots and dried in a plant dryer at ~ 75°C for two days. Dried roots were transported to Michigan State University where they. were pulverized with a ball mill (Kinetic Laboratory Equipment Co., California) before TNC extraction and analysis. Laboratory analyses 1 analyzed TNC in roots of harvested seedlings as described in Chapter 3. Briefly, I used hot alcohol to extract soluble sugars which were then analyzed with a phenol-sulfuric acid assay (Dubois et al. 1956). Extraction residues were enzymatically digested and analyzed with a glucose-specific colorimetric assay (Roper et al. 1998). TNC concentration was calculated as the sum of glucose equivalents of the soluble sugars and starch measured in each sample. Data analysis I developed a set of candidate models to characterize root TNC concentrations for each species, harvest and site combination as functions of light. I used a linear model with a constant to characterize a constant fi'action of whole plant mass being allocated to root storage or structure. TNC concentration = a*Resource + b Eq. 1 106 where Resource is either light availability, soil P or soil moisture. I also modeled storage as a power function: INC concentration = c*(Resource"d) Eq. 2 Eq. 2 is the allometric scaling model used by Niklas and Enquist (2001), where d > 1 implies allocation to root TNC (or structure) increases disproportionately with whole plant mass while cl < 1 less than proportionate increases with plant size. Soil P and soil moisture were modeled as modifying light-driven allocation by substituting for the parameters a and b, or c and d in Equations 1 and 2, respectively, with functions of the form (f‘soil) where soil represents either soil P or soil moisture. Model parameters were estimated using maximum likelihood methods (Hilbom and Mangel 1997). I assumed a normal error distribution, testing this assumption with probability plots and G-tests. Models were selected on the basis of Akaike’s Information Criteria (Hilbom and Mangel I997, Bumham and Anderson 2002), with a correction for sample size (AICc, Hurvich and Tsai 1989, Bumham and Anderson 2002). The model with the lowest AICc is chosen as the best approximating model from the set of candidate models for each relationship. In general, a difference of more than 2 units in the AICc of two models indicates poorer support for one model over the other (Bumham and Anderson 2002). I tested if harvest time affected the relationship between allocation to TNC and resources differently by using dummy variables. For example, to test if harvests two and four differed, for a linear relationship between root TNC and resources, I used equations of the form: Root TNC conc= [A 2*(Resource) + le *vz + [A4*(Resource) + B4] 1324, 107 where A 2, B; and A4, B4 are the estimated parameters for harvests 2 and 4, respectively, and v2 and v. are dummy variables that take on values of 0 or 1. I tested functional differences between harvests by evaluating AAICc between harvest-specific and combined equations. 1 did not model relationships in cases where sample sizes were less than 9. Equations were fit using the Gauss-Newton method in the non-linear procedure of Systat (SPSS Corp, Chicago, IL). Results Across species, two-year old seedlings had root TNC concentrations ranging from 10% to 39% (Table 5.2). In all sites, the fast-growing shade-intolerant species Pachira had the highest root TNC concentration while the mature forest species Cordia had the lowest concentrations. Starch was the dominant form of stored carbohydrate in all species except T abebuia, which had starch concentrations close to zero in earlier harvests with very slight increases in later harvests (Figure 5.1). The effects of whole plant mass on root TNC concentration were not very prevalent. In most species-harvest combinations there was no correlation between the two metrics (data not shown). Root TNC concentration and whole plant mass were correlated only in the following cases: Dalbergia in the first harvest in the low P site (n=15, r=0.73, p=0.0019), and in the first two harvests in the high P site (n=l3, r=0.78, p=0.0015; and n=12, r=0.84, p=0.0007, respectively); T abebuia in the second harvest in the low P site (n=1 1, r=0.77, p=0.0052) and in the final harvest in the high P site (n=9, r=0.81, p=0.0084). 108 CARRETA ARBOLEDA _A.Dalbeia § '9 i 400 .3.“ ~300 §q L200 §- -100 o IIIIIII -0 §_B.Pachira _ 8- l. A” b) 0’13.“ , M r. g , —200 8 § ‘ 1. '5 i1 2. E r 100 g go I‘IAIJ.IIIH;II.IT Fl_iITI¥-IIIIITIII 0 E “a. r: C.Tabebuia _ _ '- 8§i 400g 0 O. o 58' -300 75 0° ma- . 84 l. O (II-IIIIII-UIII § 4 D. Cordia, Astronium .1 C o- m o o N O .0. o Months Figure 5.1 Mean (+ SE) concentrations (mg g ‘1) of starch, simple sugars, and TNC in roots of the study species A. Dalbergia, B. Pachira, C. T abebuia and D. Cordia and Astronium (the latter represented by a filled circle at a single point at the 28th month only) at each harvest in low P Carreta (left column) and high P Arboleda (right column). Dotted lines represent monthly precipitation (mm). 109 Table 5.2 Mean i SE of root carbohydrate concentrations (mg g'1 TNC) at the end of the third wet season (November 2002). Sample sizes are given in parentheses. Letters denote significant differences between species in each site (Tukey’s HSD, p<0.05). Site Astron ium Cordia Dalbergia Pach ira T abebuia Carreta 179.98“ 139.5:t24.5c 259.0i10.7ar396.0a 169.9:i:l 7.5C (2) (15) (15) (2) (15) 01'0 dc l96.9il4.9 ab 101.1i17.8b 271.3i19.6a 2752:8478 224.6:h21.83 Agua (8) (11) (10) (4) (10) Arboleda 216.9i35.0a ° 160343343c 252.3s12.1ab 335.6i28.0a l92.7:i:20.2bc (5) (12) (13) (5) I9) Resource efl'ects across harvests Across species, sites, and harvests, resource availability explained between 16 and 85% of variation in root TNC concentrations (Appendix A.5, Figures 5.2, 5.3 and 5.4). Light explained between 40 and 50% of the variation in root TNC concentrations at the end of the first wet season in most species-site combinations (data not shown). At the end of the second wet season, soil resources acted together with light in influencing root carbohydrate storage (Appendix A, Figures 5.2 and 5.3). However, by the end of the third wet season (fourth harvest), light had an effect on root TNC concentrations only in T abebuia. In the high soil P site, light interacted with soil P to increase root carbon storage in T abebuia (Figure 5.3) while in low P Carreta, light similarly influenced root carbohydrate storage in the second and fourth harvests at the end of the second and third growing seasons respectively: AICc for the relationship was lower when data from the two harvests were pooled than when the two harvests were analyzed separately (Appendix, Figure 5.3). In Cordia, soil P had a very weak negative effect on root TNC concentrations in high P Arboleda, while soil moisture had a positive effect on root TNC concentrations in low P Carreta (Figure 5.4). In Dalbergia, resource availability did not 110 have any effect on root TNC concentration in Arboleda, but soil moisture had a positive effect on root TNC in Carreta (Figure 5.2). Dry season root TNC concentrations were only weakly correlated with resource availability (Appendix A.5, Figures 5.2, 5.3 and 5.4). In Dalbergia, soil moisture had a weak negative effect on root TNC concentrations at the end of the dry season (third harvest) in Arboleda while in Carreta, resource availability did not influence root TNC (Figure 5.2). In both Cordia and T abebuia, variation in resources did not have any effect on root TNC in the high soil P Arboleda. In low P Carreta, increased soil P led to weak increases in root TNC in Cordia while in T abebuia, soil P and light interacted to produce a weak positive effect on root TNC (Figures 5.3 and 5.4). Of the three species for which we measured root TNC concentrations at the end of the dry season as well as at the end of the wet season, Cordia and Dalbergia had higher TNC levels and T abebuia lower TNC concentrations at the end of the dry season than at the end of the rainy season (Figure 5.1); Dalbergia had highest and T abebuia the lowest root TNC concentrations, respectively, in May in both sites. 111 6:23.60 Eamon 28032 3:5 .8 =8 :35 ave—one can E :8 Bo: 80th E SMESQQ .«o 535: Sea 5 $5333 08:82 98 Arm wEv scumbag—So 021—182 50389 amnmcoua—om N6 952% $8 2:865 :8 8:... =2 8 8.8.9.8 an: 83 =8 .8 558.98 29.. cm on om mv ov mm on mwom mu cm 9 or n ma cm 3 or Fwd u N. wmd u N. mud n N. O T O 1 o 1 9. 9: mm 1 n. Ft 1111 1..\. 0 DE N . \1\1 110.1 F-.. ”Weill-I18? o 1 1 i o a m a o o I F... 58.1“. . r. . .1 O H O m. e o o a m. o 282an Fe. as 2: 1 a «note: fl Isuzu: 0 Sc 5303.... .m m 83 =8 8 23%.: =8 83 =2 .8 £33.26 so: 83 =8 8 88.98 22.. m Eunsoufiafioueemu on m. 9 m 0 82888882 m r F . . . p . i p h p p _ p p p p p P . p O mu «no u N. m3 1 w. 86 .1. a. M W $3 1 23%.: =8 1 m Ill 1.1.. '1 6. 8 o 1 00 o . .88 u 22%.: =8 1 v 39cm... N 30201 H —. «moan: cow m Em: mu ON 9. or m . b b F b 1 P F L P p h L1 - F b o 8... u N- r .I on o o a $3 1 23%... =2... 1 1 1 8. 79383.1... 0 fl 111 o o O 11 . r .1 09. 1 .1 11 1 .11 o c o o a $8 1 23%... :8 f 1 8m 0 o o 0 F1.... a... 2: u n. 1 1 SN c an... 1 1 . 1 1 com o v .82»... 1 N. N .85... 3 c 1 a. F .35... . Sm <0m40m1< m :19. 9... 32.3.... =8 .5» =2 é £3335 .5... .5» .3. 5 £32.96 .5... v n N F 3083 ow 2.: .82888889 — b P . n _ n b n — - u! h F 1—1 p n o C . I . u g 5 o 1 N. o o 1 ow o N. 1 J .0 o 1 o 1 8. o c on. $3 1 235.2: :8 \ 1 \ 1 com 1 1 8w 0 a l 8... u N. 1 1 8n m 3022.. O v+~ 3333.. 7.32” n c.3222 _. “moan... . 0mm (Emmy—<0 .< (1'5 6w) uonenuaouoo ON J. 1003 113 A. CARRETA 40° ‘ Harvest3 J Harvest4 o 300~ - 3'99... - . ‘7" 200 5.5 ‘ .4 — a) . P=0.5 mg kg . O) . . O E, 100. . — '3 . ,_._ 12:0.25 2 .g . r =O.57 E 0 I I I I I I I I I I E 10 15 20 25 so 22 24 26 23 so 32 34 § Light availability (% full sun) Soil moisture (%) 8 B. ARBOLEDA ‘2’ _Harvest4 '1— ‘5 O m -—1 ' 12:0.33 q 0 O _ o o O 20 40 60 80 100120140160 Soil phosphorus (mg kg'1) Figure 5.4 Relationship between root TNC concentration (mg g") and resource availability in two harvests of Cordia in Carreta (low soil P) and Arboleda (high soil P). Lines represent best-fit equations. 114 /\.(3/\FU?IET7\ 600 Harvest 1 0 $ 500 -* . 400 - . 300 6 200 ‘7" 100 O) E’ o E I I T If T I I I '13 10 20 30 4o 50 60 7o 80 E Light availability (% full sun) § 0 B. ARBOLEDA 0 .Harvest 1 0 350 - o ‘ = 0.43 a? o i 300 - ' -1 250 - P=100 mg kg 0 200 a____ ______ .4 -1 150-. . o P=25m9k9 100 I I I I I I I I I 6 81012141618202224 Light availability (% full sun) Figure 5.5. Relationship between root TNC concentration (mg g") and resource availability in the first harvest of Pachira in Carreta (low soil P) and Arboleda (high soil P). Lines represent best-fit equations. 115 TNC, growth and survival Species’ mean root TNC concentrations at the end of the experiment and species’ mean diameter growth rates over the experiment were negatively correlated in the low P site Carreta, positively correlated in the high P site Arboleda, and were not correlated in the intermediate soil P site Ojo de Agua (Figure 5.6). The lack of a consistently significant correlations could arise from the small number of species in our study. However, under low light (<10% full sun), there is no relationship between growth and TNC in any site, while under high light (>15% full sun), species’ mean root TNC concentrations and species’ mean diameter growth rates are positively correlated only in the high soil P site (n = 5,p = 0.87, p = 0.05). Contrary to expectations, species’ mean root TNC concentrations at the final harvest and species’ survival over the length of the experiment were not positively correlated: there were weak nonsignificant, negative correlations between the two variables in all three sites (Figure 5.7). Analyzing these relationships under low and high light availability also consistently showed weak, negative correlations between mean root carbohydrate stores in species in November 2002 and survival until that time (data not shown). Species’ mean root TNC concentrations at the end of the second wet season (November 2001) and survival through the following dry season were negatively correlated in low P Carreta (n=3, r=-O.99, p=0.07) and also with survival through the next wet season (n=3, r=-0.99, p=0.09). In the high P Arboleda, there was a weak trend 116 towards a positive correlation between wet season TNC and survival through the following year (n=3, r=0.87, p=0.33). Diameter growth rate (mm mrn'1 yr'1) 1.2 1 1.0- 0.8 - 0.6 s 0.4 - A. CARRETA A. IU§I>O AS CO DA PA TA -O.93, p=0.02 Cl I I I I I I I 2.0 - 1.8 — 1.6 - 1.4 ~ 1.2 . 1.0 J B. OJO DE AGUA D C A r = 0.34. p=0.54 I I I I I 1 1.4 1.2 - 1.0 4 0.8 d 0.6 - C. ARBOLEDA A C] r = 0.88, p=0.05 0.4 I I I I I I I 100 150 200 250 300 350 400 Root TNC concentration (mg g'1) Figure 5.6 Relationship between species’ mean diameter growth rates (mm mm'1 yr") and mean root TNC concentrations (mg g") at the end of the third wet season in A. Carreta (low soil P); B. Ojo de Agua (intermediate soil P); and C. Arboleda (high soil P). 117 80 A. CARRETA . AS A A CO 60 4 ‘ ‘ DA I U PA 40 _ I TA 20 - = -O.51, p=0.34 D O 0 I I I r I f I A B. 0J0 DE AGUA o\° 80 A ‘5' . E 60 r ‘ 'c a) 9' 4o - a) a: 5 20 a . D g = -o.54, p=0.35 E 0 I I I I I I I 3 <0 50 - C.ARBOLEDA 4o - A 30 - A 20 “ . 10 ~ 0 r = -O.49, p=0.41 O O 100 150 200 250 300 350 400 Root TNC concentration (mg g'1) Figure 5.7 Relationship between species’ survival (%) and mean root TNC concentrations (mg g") at the end of the third wet season in A. Low soil P Carreta, B. Intermediate soil P Ojo de Agua, and C. High soil P Arboleda. 118 Discussion All species had substantial proportions of stored carbohydrates in their roots, ranging from ~10 to 40% at the final harvest. These levels are comparable with or higher than those reported for some moist tropical forest tree species (Newell et al. 2002) and rainforest shrubs (Marquis et al. 1997) and are generally higher than those found in other studies in moist (6-8% in Wurth et al. 2005) and seasonal forests (4-6% in roots of plantation species; Latt et al. 2001). Some of these differences may be because Wurth et al. (2005) sampled only 10-15 mm diameter roots, likely leading to their result of lower TNC concentrations. Temperate forest seedlings also have comparable levels of root TNC concentrations (e.g., Canham et al. 1999, Chapter 2). Starch was the predominant form of stored carbohydrate in four of the five species, as is the case in most species (Kozlowski 1992). However, T abebuia roots had very low levels of starch. The negligible levels of starch in this species are likely due to water deficits which cause decreased starch and often, an increase in soluble sugars (Kramer 1983), which maintain osmotic balance and turgor under water stress (Ritchie I982). Temporal changes There were contrasting temporal patterns in TNC concentration among species. Contrary to expectations, mean root TNC concentrations in Dalbergia increased substantially over the dry season (from November 2001 through May 2002). Although I do not have data to directly examine dry season changes in Cordia root TNC concentration, for both Cordia and Dalbergia, TNC root concentrations were higher at the end of the dry season (May 119 2002) than at the end of the wet growing season (November 2002), suggesting that TNC pools are depleted during what is considered to be the growing season and build up during the dry season. It is possible that because of infrequent sampling, I missed further TNC concentration increases at the end of the rainy season in November 2001 and 2002. However, the magnitude of the increase in TNC suggests this is unlikely. Increases in TNC during seasonal droughts have been reported in other seasonal tropical forest species (Newell et al. 2002, Latt et al. 2001, Wurth et al. 2005). It is thus more likely that the increase in TNC is brought about by a cessation of grth but not photosynthesis in these species (Kitajima et al. 1997, Wurth et al. 2005). This is consistent with the link between leaf phenology and dry season root TNC in my study species. Of the two species for which dry season root TNC was higher, Dalbergia, is a brevideciduous species that refoliates in the middle of the dry season (Rivera et al. 2002). I do not have leaf phenology data for Cordia, but it was observed to have leaves at the end of the dry season. On the other hand, T abebuia had minimum TNC concentration at the end of the dry (or dormant) season. This species sheds leaves during the early dry season and refoliates only at the start of the rainy season (Borchert et al. 2002, pers. obs.). However, its fine root growth commences before the onset of the rainy season (pers. obs.). Thus, decreased TNC concentrations likely result from maintenance respiration and fine root growth during the dry season. The temporal changes in TNC concentrations with resource availability highlight the importance of taking not only season but also development stage into consideration. All species in their first year had TNC concentrations that responded to light availability. However, by the third wet season, concentrations were not linked to light availability in 120 most species-site combinations. In general, studies with seedlings less than two years old have found a positive relationship between TNC concentrations and light availability (e.g., Mooney et al. 1995, Naidu and DeLucia 1997, Gansert and Sprick 1998, Chapter 2), while those with older seedlings or saplings have found no correlations with light availability (e.g., Kobe 1997, Canham et a1. 1999). It is possible that the lack of a relationship with light in the older seedlings was due to changes in light levels. However, the lack of a light effect in older seedlings could also reflect the ‘ghost of TNC accumulations and withdrawals past’. The combination of earlier allocation to TNC and of repeated TNC withdrawals - possibly resulting from herbivory, refoliation and drought stresses — likely leads to the absence of current light effects on point measurements of root TNC concentrations in older seedlings. Relationships among TNC, survival and growth A trade-off between low-resource survival and high-resource growth (Kitajima 1994, Kobe et al. 1995, Lin et al. 2002, Schreeg et al. 2005) implies differences in allocation to different fimctions within the plants, such as allocation towards more fine root mass, or increased nutrient uptake rates to increase growth rates, or allocation towards greater defense and storage to increase survival (Kobe 1997). I had thus expected to see a trade- off between growth and allocation to TNC across species, because limited resources would have to be allocated towards either grth or survival (Chapter 3). However, a negative correlation between grth and TNC is evident only in the low P site (Carreta) where most species appear to be more limited by soil resources than by carbon (Chapter 3), as is often the case when environmental conditions are adverse (Komer 2003). The 121 limitation on growth but not photosynthesis could lead to the negative growth-TNC relationship seen in this site as photosynthate accumulates and is stored rather than used towards new resource-harvesting structures. An alternative mechanism by which TNC and growth rates might be inversely correlated would be if TNC withdrawals and growth rates are linked. Changes in TNC and leaf phenology are not always tightly linked in tropical forests (e.g., Newell et al. 2002, Wurth et al. 2005), although this likely varies with grth rates and nutrient availability. Slow-growing species adapted to low fertility sites depend on reserves for structural growth to a larger extent than do fast-growers from high fertility sites (Steinlein et al. 1993) although it is unclear whether this association is due to a species’ inherent growth rate or effects of the environment or both. If growth rates determine the extent of withdrawal, we could expect to see a negative relationship between TNC concentrations and growth. The positive relationship between growth rates and storage in the high P site suggests that where growth is not limited by soil resources, larger TNC stores allow greater structural growth, which in turn allow for greater photosynthetic income and more photosynthate to allocate to both storage and non-storage, leading to a positive correlation between grth and TNC stores. The positive relationship between TNC and growth at the high P site could also arise from higher carbon incomes that enable greater allocation to both storage and non-storage pools. However, the lack of a prevalent effect of whole plant mass on root TNC concentrations argues against this interpretation. Contrary to expectation, there was not a positive correlation between TNC concentrations and survival across species. TNC pool sizes may be better predictors of 122 TNC available for future mobilization (Chapin et al. 1990) and thus, better predictors of survival. However, pool size also was not correlated with survival in any of the sites (data not shown). Species survival and root TNC concentrations also may be uncorrelated if root TNC stores are only a small fraction of available TNC pools. In adult trees in a moist forest, Wurth et al. (2005) estimated that belowground TNC pools represent less than 20% of total available TNC reserves. However, root TNC stores in our species likely represent a larger proportion of whole plant TNC pools than suggested by Wurth et al. (2005), who by measuring only coarse root TNC concentrations (and not main root axes), probably underestimated belowground TNC pools since larger diameter and lower order roots have higher storage (Singh and Srivastava 1986, Kosola et al. 2002). The tendency for TNC and survival to be negatively correlated and for growth and TNC to be positively correlated under higher P suggests that species that grow bigger faster are better able to tolerate drought and herbivore pressure than those that stay small. maintain osmotic balance and/or conserve resources. Larger plants would have longer, deeper roots that would enable them to avoid severe water stress during the dry season by allowing them access to the moist subsoil (Borchert 2000). Lower susceptibility to drought in larger plants has been shown in some other tropical forests (Cao 2000, Poorter and Hayashida-Oliver 2000). The low soil resource availability in Carreta inhibits growth more than in Arboleda so that Carreta seedlings would take longer to escape size- dependent mortality, leading to the weak negative relationship between TNC concentrations and survival over the next year. However, in Arboleda, soil resources are more available and hence individuals can grow larger quicker and hence evade drought 123 stress, leading to the tendency for TNC concentrations at the end of the second wet season to be positively correlated with survival over the following year in Arboleda. Conclusions There was substantial allocation to carbohydrate storage in all species studied in this seasonally dry tropical forest, with some seedlings having ~40% of root mass as TNC. Storage in first year seedlings varied with light availability but in older seedlings of most species, was influenced by soil resources alone, likely because withdrawals over the previous seasons obscured resource-TNC relationships. Two of three species showed an increase in dry-season root TNC concentrations, indicating that growth was apparently more limited than photosynthesis by the seasonal drought. Across species, growth and TNC were negatively correlated in the low fertility site and positively correlated in the high fertility site, but, contrary to expectations, TNC and survival were not correlated. Acknowledgements I thank the Ministerio del Ambiente y Energia, Costa Rica for granting permission to can out the research, and the Organization for Tropical Studies (OTS) for the use of their field station. I would like to thank Francisco Enrique Granados, Sandra Quiros, Justin Harbison, Lynn Holcomb, Nicole Turner and Mauricio Solis for assistance in the field. This work was supported by grants fi‘om OTS and the Taraknath Das Foundation to M1, and partly by a grant from the National Science Foundation (NSF DEB 0075472) to RKK. 124 Appendices 125 126 -- am: .29 £8 :3 8 as n : 23.: -- 83 .29 :8 2.: 8 as n : 2:3 -- $2.89 8.: 8.8 N» as u : 2.3: -- :88 .29 8.: 8.: 2 _ i craves u : 2:9: 03 -- 88.89 8.: :8 a : - 3393 n : 75.: £3.89 8.: an: .39 on: 8.: » Sega u : 75:: $2 .89 8.: 8:8 .29 8.: 8.: a 33:8...“ n: 2:8: 2» 82.: :9 8.: 3.8.39 2.: 8.: a» 3,3... l : 75.: 8389 8.» 68.8.89 8.: 8.8 8 33% u : 7:3: @389 :8 3..» £9 a.» £8 3 39:5... u : E o: a: .89 a»: 98 .89 8.: 8.: : 33...» u m 2.3:: :88 .89 8.: God .89 8.: 8.: N: 3,3...a .i. : 7:3: 2: :8.» .z. s 8.» GS 49 2.8 8.8 2 33% n : 2.5: -- o S .89 2.: Rd 2 : - 3993 n a 7:3: .. a; .29 :8 £8 2 _ - 3998 u : 2.3: .. A88 .89 3.: 8.: S as u a 25:: 0: 688.89 :88 $88.: .8889 88.: 8.: a» 33:5...» n » .3 a: .39 8.: G»... .89 8.: :8 8 33%....“ u : 21:: am.» .89 E»: :2 .29 8.: 8.: a 399% u : 25: um 38 .89 8.: an: .88-: 83 2.8 a» _ - Fens... u m 73.: .. 98.: .889 :88 s8 2 : - Q4393 n : 7:34 a: .8». c 8.: an: .29 8.: 8.: 3 33% l : 2.5: :2 .39 8.: a; .89 2.: 8.8 N» 33.... n : 2:15 m< 6 :8 i a 6 :8 .s N: z .082 “Bases a» .z 32 E»: 33 l 2.5.. Hz :»_: .2»: so: i 75.: “2 32 4:»: :»E .. 2.5: m2 :»E .2»: :9: 1 7:3: 25:8 a as 28.92% Eases: .xwo BE? l 03 ”238 awn» l Em inc :3: l OM moans. we: I 2% £8 :85 I Om ”Page :85 l um 5002. 585:2 l m< 826:8 mm as 823325: 33on 6.88 :33 20:3 :23 07E. goo.— maua—o: £308 :8 A V 5:358:83 no 356508 28 G6 on? 338% £3235 tongs» $3 was $3838 BEES—a 3258—: 83882 a 52.2— < a»: .89 88 3.8 88988 .8 3. 33..» u : A: .. €8.89 :8 88 8 .3 u : d: 03 -- :88.» :9 28 88 8 _ I 3.8me u : 2.3: 2 2 .89 8.: 888 9.9 2.8 88 8 3.8..» u : 23: 88.2.89 8: 8.8.89 88 88 .2 5.5....N u .8 z::: 2» .. :88 .89 88 88 v» 8 n : 23: 88 .89 88 3.8 .89 88 88 8 33...» u : 2:3 .. 3.8 .89 88 88 8 5. u : 2.5: 32 .89 8.: 8:8 .89 88 88 8 33..» "8 2:4: 0: .. :88 ._ »9 8.8 888 2 _ n 3.996 u : 2.5: :88 .89 :8 88 9.9 88 88 2 33:8...» n : 2:9: 2: .. A88 .89 88 888 8» 8 u : .5 .. 5.8.89 88 88 8» 8 u 8. a: o: €88.89 :8 58.889 88 $8 8 3.8...» u .8 .5 888.29 8: 8.8 .89 88 88 8 33..» n: 2.3: a: .889 82 38.89 88 88 2 33..» n .8 z::: um 28.» .89 8.: $88 .89 88 88 8» 33..» u : 75.: .. 88.29 :8 88 8: 8 n : 2:3 :8: .89 888 a: .89 8.: 88 8 23-985.» u a 23: .. 688 ._ 89 88 88 8 8 u : 7:5: :8. U 88 u 8 6 88 .8 N: z :82 88:82.: .8» ._ 88523.9 5 m8 2: 8888383: E2588 EB 33on 838 :83 2083 8:5 238 88 3:325» mamas—2 2882: :8 A 9 58:58.68: mo 8:82.830 88 9c 08» moi—58m £3.68: teams» 883 88 888883 3808389 uoofififi~ 828882 N 8.52. < 127 128 35 ad. .3: ”.2 as: .3 _ : M22 .3 u h 3 75.: a. 75.: who 83.2: 3“ 8.9: 42:5: van» mm Eaqumqm o; a; $532-: ”.2 63%.: «no 82“.”.8: 32 33..“ n .A a 2.3: $5 $3 .2: 2: 89° .38 33 A22 .35: 32 SE...“ n h a 25: :5 who 32 a: 2: as: .8. : mm; 33 £8: 38 3.3.... u .5 a 2.3.. «2 9.8 .m: ”.3 3.3 $2: 3.: 5. u a ma 2:3 23 AZ: .2: 3m Gd: 6%: 3.2 a. n h an 2.3: m3 3.: .N. : «.2 SN: .99: 33 .3 u m a 25: om “3 8.2 .3: 3 9.3 4.2.: mean 5. u .A _ _ 25: 8... 2.2 .2: 3 $3. $18 an... Sim 4.8: 3a 3.3.... u .5 2 ES: :2 mg 8.3. .3: q: $.ch .32: S: .a u a 2 24.3 85 9.2 .2: 3 :3 .2: “no 25$ .3: «.5 33.... u x 2 75.3 85 2.3 .33 3a 8%: .98: 2: 5 u a : 2.3: v2 86% .5. E: 95 352 6.2: E: “a u h 2 ES: om 86 a: .2: 3 89° .85-: ad 83 £82-: ca 32:.“ u m a 2...: a. 2:5 85 A2: .1: 2: 83¢ .33 ”.30. 5. u .n a 2.5: a... 2.: .38 EN 8.»: 48: 3mm 5. u a M: 25: on ”3 8.”. 4.: 3 8.2m {as 5.: a. u a an 24.3 a 2:3 2:. G. S 4.: EN :3 .38 who 9.8.“ .NS: 3% 33..“ u a E 2.5: :3 3:. .3: «.2 Ed .9: and A3: .33: 22 33..“ u a .8 25m 5. 98%... :3: Arcana $3: 925% $3: q. 85E> a a .082 2 “Subway 86am ._ 522:3 5 mm 0.3 823395;“ E0535 98 8325 .3 v.38 88 1358.5 :5 $83 GE: v “82523 $00.. 26 mo 8.5 3&3,“ mafia—2 £258 mama—“€899“ “won 05 .8: moo—35> 2.2m: Erma—58% mo mafia—boon: AZ: mean 038% £3253: .8 toga $3 :5 33838 coca—8E 8:832 n 5.52:: 129 85 9.8 23: 92 RS .83 one 2.8 .23 3:. 3.5....w u 2 a. 2...: a 22.: 2:. Sam 2.: «.2 9.2. .36 2.8 .a u 2 cm 2.32 S5 Gfi .2: 32 6.3 .22: 3o .8 u 2 t 2222 03 Rd 3.8 .5. : NE 33 an: :3 $2: 2.3: «.5 3,3...“ u 2 a 2.32 a. 2222 :2 So 9.: .9: tom 3.5 .35: E: .3 u A a. 2.2.: a 22.3 :3 82 : .2: Km 8.2: .25 «.8 § u 2 an 2.2.22 «3 9.2: .v. : 93 3% .22 «.3 a. u 2 E 2222 02 «no :2 .9: a.“ 222. .22: an: a. u 2 2 22.22 85 2.2 .2: 3 $3 .38 85 8.3 .22: 38 33..“ u .2 2 2222 22 $8 2.2. 6.3 3: a: .35-: «no 222 .23 32 33... u a : 2.5.2 23 8.: .o. E 3 63 «we and a. a .2: a: 3,.ka u 2 2 22.: Rd 8.3 .22-: 22 9.2: .3: 2.8 5. u 2 : 2.32 Rd G .3 .35 $2. .3 u 2 2 22.22 Om 23 a; .2: 3 and .23-: .25 GE .23 Sm 33.... u a a 2.2.2 2 22.5 33 8.8 .3: n: 23 .38 m; 3.22 «.2: «SN 33.... u 2 a 2.52 :3 98 .23 Non 8.22 .3»: 3m 5. u a a 22.2 02 who 82. .9: Z 3.32 .63.: 3.8 § u 2 on 2...: a. 22.: a; 2.9. 2.: EN E? .23: 85 2:2 .54: 2: 3.5..“ u 2 a 2.52 25 3mm 2.: 2: ES .38 a... 6.5 is: 32 33..“ u 2 «N 22.2 9.. 9223 23% 992% 22: €223 .23: x 8§E> a a 282 2 285.22 860% A 285%? 5 mm as 8233053 22253.: ES 86on Am: $3: 88 39 98 $83 22:: v 280233 308 can we 38 oomEm mafia—2 3308 2588228292 “mun 05 8.: 8055., 222m: 22352280: mo $230808 AZ: 85m £923. £8058“: 28 2033 $3 :5 83850 coca—8E 8:22:82 v 5.52:3. References 130 Abrams MD (2003) Where has all the white oak gone? BioScience 53:927-939 Aerts R, Chapin FS III (2000) The mineral nutrition of wild plants revisited: a re- evaluation of processes and patterns. Adv Ecol Res. 30: 1-67 Austin PC, Hux JE (2002) A brief note on overlapping confidence intervals. J Vasc Surg 36: 194-195 Bazzaz FA (1979) The physiological ecology of plant succession. Ann Rev Ecol System 102351—371 Bigelow SW, Canham CD (2002) Community organization of tree species along soil resource gradients in a north-eastem USA forest. J Ecol 90: 188-200 Blair BC, Perfecto l (2004) Successional status and root foraging for phosphorus in seven tropical tree species. Can J For Res 34:1 l28-1 135 Bloom, AJ, Chapin F S III, Mooney HA (1985) Resource limitation in plants - an economic analogy. Ann Rev Ecol System 16:363-392 Borchert R (2000) Organismic and environmental controls of bud growth in tropical trees. In Viemont JD, Crabbe J (eds.) Dormancy in plants: from whole plant behavior to cellular control. CAB International, Wallingford, UK. 385 pp Borchert R, Rivera G, Hagnauer W (2002) Modification of vegetative phenology in a tropical semi-deciduous forest by abnormal drought and rain. Biotroppica 34:27- 39 Bradley RL, Fyles J W (1995) Growth of paper birch (Betula papyrifera) seedlings increases soil available C and microbial acquisition of soil-nutrients. Soil Biol Biochem 27: 1565-1571 Brouwer H. 1983. Functional equilibrium: sense or nonsense? Netherlands J Agri Sci 31:335-348 Bullock SH (1992) Seasonal differences in nonstructural carbohydrates in two dioecious monsoon-climate trees. Biotropica 24: 140-145 Bumham, KP, Anderson DR (2002) Model selection and multimodel inference: a practical information-theoretic approach. Springer, New York, USA Bums RM, Honkala BH (eds) (1990) Silvics of North America, Agriculture handbook 654. US Department of Agriculture, Forest Service, Washington, DC. Burslem DFRP, Grubb PJ, Turner IM (1995) responses to nutrient addition among shade- tolerant seedlings of lowland tropical rain forest in Singapore J Ecol 83:113-122 13] Campo J, Jaramillo, VJ, Maass, JM (1998) Pulses of soil phosphorus availability in a Mexican tropical dry forest: effects of seasonality and levels of wetting. Oecologia 115:167-172 Canham CD, Berkowitz AR, Kelly VR, Lovett GM, Ollinger SV, Schurr J (1996) Biomass allocation and multiple resource limitation in tree seedlings. Can J For Res 26: 1521-1530 Canham CD, Kobe RK, Latty EF, Chazdon RL (1999) Interspecific and intraspecific variation in tree seedling survival: effects of allocation to roots versus carbohydrate reserves. Oecologia 121:1-1 l Cao K-F (2000) Water relations and gas exchange of tropical saplings during a prolonged drought in a Bomean heath forest, with reference to root architecture. J Trop Ecol 16: 101—1 16 Ceccon E, Huante P, Campo J (2003) Effects of nitrogen and phosphorus fertilization on the survival and recruitment of seedlings of dominant tree species in two abandoned tropical dry forests in Yucatan, Mexico. For Ecol Manage 182:387- 402 Chapin FS (1991) Integrated responses of plants to stress. BioScience 41 129-36 Chapin F S III, Schulze ED, Mooney HA (1990) The ecology and economics of storage in plants. Annu Rev Ecol Syst 212423-447 Comas LH, Bouma TJ, Eissenstat DM (2002) Linking root traits to potential growth rate in six temperate tree species. Oecologia 132:34-43 Coomes DA, Grubb PJ (2000) Impacts of root competition in forests and woodlands: a theoretical framework and review of experiments. Ecol Monogr 70:171-207 Cordero LDP, Kanninen M, Arias LAU (2003) Stand grth scenarios fro Bombacopsis quinata plantations in Costa Rica For Ecol Manage 174:345-352 Craine JM, F roehle J, Tilman DG, Wedin DA, Chapin F S (2001) The relationships among root and leaf traits of 76 grassland species and relative abundance along fertility and disturbance gradients Oikos 93:274-285 Davies SJ (2001) Tree mortality and growth in 1] sympatric Macaranga species in Borneo. Ecology 82:920-932 Delissio LJ, Primack RB (2003) The impact of drought on the population dynamics of canopy-tree seedlings in an aseasonal Malaysian rain forest. J Ecol 19: 489-500 132 DeLucia EH, Sipe TW, Herrick J, Maherali H (1998) Sapling biomass allocation and growth in the understory of a deciduous hardwood forest. Am J Bot 85: 955-963 Denslow J S (1987) Tropical rainforest gaps and tree species diversity. Ann Rev Ecol System 18:431-451 Dubois, M, Gilles KA, Hamilton JK, Rebers PA, Smith F (1956) Colorimetric method for determination of sugars and related substances. Anal Chem 282350-356 El Omari B, Aranda X, Verdaguer D, Pascual G, Fleck I (2003) Resource mobilization in Quercus ilex L. resprouts. Plant Soil 252:349-357 Espeleta, JF, Donovan LA (2002) Fine root demography and morphology in response to soil resources availability among xeric and mesic sandhill tree species. Funct Ecol 6: 113-121 F arrar JF, Jones DL (2000) The control of carbon acquisition by roots. New Phytol 147:43-53 Fichtner K, Quick WP, Schulze ED, Mooney HA, Rodermel SR, Bogorad L, Stitt M (1993) Decreased ribulose-l,S-bisphosphate carboxylase-oxygenase in transgenic tobacco transformed with antisense rch. 5. Relationship between photosynthetic rate, storage strategy, biomass allocation and vegetative plant-growth at 3 different nitrogen supplies. Planta 190: 1-9 Finzi AC, Canham CD (2000) Sapling growth in response to light and nitrogen availability in a southern New England forest. For Ecol Manage 13 l : 153-165 Fransen B, de Kroon H, Berendse F (1998) Root morphological plasticity and nutrient acquisition of perennial grass species from habitats of different nutrient availability. Oecologia 1 15:351—358 Frazer, GW, Canham CD and KP Lertzman. 1999. Gap light Analyzer (GLA): Imaging software to extract canopy structure and gap light transmission indices from true- color fish-eye photographs. Simon Fraser University, Burnaby, British Columbia, and the Institute of Ecosystem Studies, Millbrook, New York. Gansert, D, Sprick W (1998) Storage and mobilization of nonstructural carbohydrates and biomass development of beech seedlings (Fagus sylvatica L.) under different light regimes. Trees 12:247-257 Gaucher C, Gougeon S, Mauffette Y, Messier C (2005) Seasonal variation in biomass and carbohydrate partitioning of sugar maple (Acer saccharum) and yellow birch (Betual alleghaniensis) seedlings. Tree Phys 25:93-100 133 Gerhardt K (1996a) Effects of root competition and canopy openness on survival and growth of tree seedlings in a tropical dry forest. For Ecol Manage 82:33-48 Gerhardt K (1996b) Germination and development of sown mahogany (Swietenia macrophylla King) in secondary tropical dry forest habitats in Costa Rica. J Trop Ecol 12:275-289 Gillespie TW (1999) Life history characteristics and rarity of woody plants in tropical dry forest fragments of Central America. J Trop Ecol 15:637-649 Godbold DL, Bernston GM, Bazzaz FA (1997) Growth and mycorrhizal colonization of three North American tree species under elevated atmospheric C02. New Phytol 137: 433-440 Graham JH, Duncan LW, Eissenstat DM (1997) Carbohydrate allocation patterns in citrus genotypes as affected by phosphorus nutrition, mycorrhizal colonization and mycorrhizal dependency. New Phytol 135: 335-343 Grime JP (1979) Plant strategies and vegetation processes. Wiley, New York Grime JP (1977) Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. Am Nat 11 1: 1169-1194 Grubb PJ, Lee WG, Kollmann J, Bastow Wilson J (1996) Interaction of irradiance and soil nutrient-supply on growth of seedlings of ten European tall-shrub species and Fagus sylvatica. J Ecol 84:827—840 Hall J S, McKenna JJ, Ashton PMS, Gregore TB (2004) Habitat characterizations underestimate the role of edaphic factors controlling the distribution of Entandrophragma. Ecology 85 :2 1 7 1-21 83. Hartshorn, GS. and L]. Poveda. 1983. Plants: checklist of trees. Pp 158-183. In: D.H. Janzen (ed.). Costa Rican Natural History. University of Chicago Press: Chicago. Heilemier H, Freund M, Steinlein T, Schulze ED, Monson RK (1994) The influence of nitrogen availability on carbon and nitrogen storage in the biennial Cirsium vulgare (Savi) Ten. 1. Storage capacity in relation to resource acquisition, allocation and recycling. Plant Cell Environ 17: 1125-1131 Hilbom, R, Mangel M (1997) The ecological detective: confronting models with data. Princeton University Press, Princeton, NJ, USA Hoffmann WA, Orthen B, Do Nascimento PKV (2003) Comparative fire ecology of tropical savanna and forest trees. Funct Ecol 17:720-726 134 Holdridge, L. R. 1969. Life zone ecology. Centro Cientifico Tropical, San José, Costa Rica Host GE, Pregitzer KS (1992) Geomorphic influences on ground-flora and overstory composition in upland forests of northwestern lower Michigan. Can J For Res 22: 1547-1555 Huante, P., E. Rincon, and I. Acosta. 1995. Nutrient availability and growth rate of 34 woody species from a tropical deciduous forest in Mexico. Functional Ecology 92849-858. Hubbell SP (1979) Tree dispersion, abundance and diversity in a tropical dry forest. Science 203: 1299-1309 Hubbell SP (2001) The unified neutral theory of biodiversity and biogeography. Princeton University Press, N.J., USA Hubbell SP, Foster RB (1992) Short-term dynamics of a neotropical forest: why ecological research matters to tropical conservation and management. Oikos 63:48—61 Hurvich, CM, Tsai CL (1989) Regression and time series model selection in small samples. Biometrika 76:297-307 Huston M, Smith T (1987) Plant succession: Life history and competition. Am Nat 130:168-198 Iwasa Y, Kubo T (1997) Optimal size of storage for recovery after unpredictable disturbances. Evol Ecol 11: 41-65 Jasienski M, Bazzaz FA (1999) The fallacy of ratios and the testability of models in biology Oikos 84:321-326 Jimenez J, Gonzalez E, Vega-Mateo J (2001) Perspectives for the Integrated Management of the Tempisque River Basin, Costa Rica. Organization for Tropical Studies, Costa Rica. J imenez-Madrigal O (1998) Arboles maderables en peligro de extinction en Costa Rica. Inbio, San Jose. Kabeya D, Sakai S (2003) The role of roots and cotyledons as storage organs in early stages of establishment in Quercus crispula: a quantitative analysis of the nonstructural carbohydrate in cotyledons and roots. Ann Bot 92: 537-545 135 Kaelke CM, Kruger EL, Reich PB (2001) Trade-offs in seedlings survival, growth, and physiology among hardwood species of contrasting successional status along a light-availability gradient. Can J For Res 31: 1602-1616 Keller JD, Loescher WH (1989)Nonstructura1 carbohydrate partitioning in perennial parts of sweet cherry. J Am Soc Hort Sci 114: 969-975 King DA (1993) A model analysis of the influence of root and foliage allocation on forest production and competition between trees. Tree Phys 12:119-135. Kitajima K (1994) Relative importance of photosynthetic traits and allocation patterns as correlates of seedling tolerance of 13 tropical trees. Oecologia 98:419-428 Kitajima K (2002) Do shade-tolerant tropical tree seedlings depend longer on seed reserves? Functional grth analysis of three Bignoniaceae species. Funct Ecol 16:433-444 Kitajima K, Augspurger CK (1989) Seed and seedling ecology of a monocarpic tropical tree, Tachigalia versicolor. Ecology 70:1102-1114 Kitajima K, Mulkey SS, Wright SJ (1997) Seasonal leaf phenotypes in the canopy of a tropical dry forest: Photosynthetic characteristics and associated traits. Oecologia 109:490—498 Kobe RK (1996) Intraspecific variation in sapling mortality and growth predicts geographic variation in forest composition. Ecol Monogr 66: 1 81-201 Kobe RK (1997) Carbohydrate allocation to storage as a basis of interspecific variation in sapling survivorship and growth. Oikos 80: 226-233 Kobe RK (1999) Light gradient partitioning among tropical tree species through differential seedling mortality and growth. Ecology 80: 1 87-201 Kobe RK, Pacala SW, Silander JA Jr, Canham CD (1995) Juvenile tree survivorship as a component of shade tolerance. Ecol Appl 5:517-532 Kobe RK, Lepczyk CA, Iyer M (2005) Resorption efficiency decreases with increasing green leaf nutrients in a global data set. Ecology 86:2780-2792 Kosola K, Dickmann DI, Paul EA, Parry D (2001) Repeated insect defoliation effects on growth, nitrogen acquisition, carbohydrates, and root demography of poplars. Oecologia 129:69-74 Kozlowski ”IT, Pallardy SG (2002) Acclimation and adaptive responses of woody plants to environmental stresses. Bot Rev 68:270—334 Kozlowski TI‘ (1992) Carbohydrate sources and sinks in woody plants. Bot Rev 58: 1 07- 222 136 Langley JA, Drake BG, Hungate BA (2002) Extensive belowground carbon storage supports roots and mycorrhizae in regenerating scrub oaks. Oecologia 131 :542-548 Latham RE (1992) Co-occurring tree species change rank in seedling performance with resources varied experimentally. Ecology 73:2129-2144 Latt CR, Nair PKR, Kang BT (2001) Reserve carbohydrate levels in the boles and structural roots of five multipurpose tree species in a seasonally dry tropical climate. For Ecol Manage 146:145—158 Leahy MJ, Pregitzer KS (2003) A comparison of Presettlement and present-day forests in northeastern lower Michigan. Am Midl Nat. 147:71-89 Lewis SL, Tanner EVL (2000) Effects of above- and belowground competition on growth and survival of rainforest tree seedlings. Ecology 81: 2525-2538 Lin J, Harcombe PA, Fulton MR, Hall RW (2002) Sapling growth and survivorship as a function of light in a mesic forest of southeast Texas, USA. Oecologia 1322428- 435 Lodge DJ, McDowell WH, McSwiney CP (1994) The importance of nutrient pulses in tropical forests. Trends Ecol Evol 9:384-3 87 Loescher WH, McCamant T, Keller JD (1990) Carbohydrate reserves, translocation and storage in woody plant roots. Hort Sci 25:274-281 Marquis, RJ, Newell EA,Villegas AC(l997) Non-structural carbohydrate accumulation and use in an understory rain-forest shrub and relevance for the impact of leaf herbivory. F unct Ecol 11:636-643 McDonald AJS, Ericsson A, Lohammer T (1986) Dependence of starch storage on nutrient availability and photon flux density in small birch (Betula pendula Roth). Plant Cell Environ 9: 433-438 Mehlich A (1984) Mehlich 3 soil test extractant: a modification of Mehlich 2 extractant. Commun Soil Sci Plant Anal. 15:1409-1416 Miyanishi K, Kellman N (1986) The role of fire in recruitment of 2 neotropical savanna shrubs, Miconia albicans and Clidemia sericea. Biotropica 18:224-230 Mooney, HA, K Fichtner, Schulze ED (1995) Growth, photosynthesis and storage of carbohydrates and nitrogen in Phaseolus lunatus in relation to resource availability. Oecologia 104: 17-23 137 Naidu S, DeLucia EH (1997) Growth, allocation and water relations of shade-grown Quercus rubra L saplings exposed to a late-season canopy gap. Ann Bot. 80: 335- 344 Neuhaus HE, Schulte N (1996) Starch degradation in chloroplasts isolated from C-3 or CAM (crassulacean acid metabolism)-induced Mesembryanthemum crystallinum L. Biochem J 318:945-953 Newell EA, Mulkey SS, Wright SJ (2002) Seasonal patterns of carbohydrate storage in four tropical tree species. Oecologia 131:333-342 Nguyen PV, Dickmann DI, Pregitzer KS, Hendrick R (1990) Late-season changes in allocation of starch and sugar to shoots, coarse roots and fine roots in two hybrid poplar clones. Tree Physiol 7:95-105 Nicotra AB, Babicka N, Westoby M (2002) Seedling root anatomy and morphology: an examination of ecological differentiation with rainfall using phylogenetically independent contrasts. Oecologia 130: 136-145 Niklas KJ, Enquist BJ (2001) Invariant scaling relationships for interspecific plant biomass production rates and body size. Proc Natl Acad Sci 98: 2922-2927 Nunes MA, Ramalho JDC, Dias MA (1993) Effect of nitrogen supply on the photosynthetic performance of leaves from coffee plants exposed to bright light. J Exp Bot 44:893-899 Oldeman RAA (1990) Forests — Elements of Silvology, Springer-Verlag, Berlin Opler PA, Baker HG, Frankie GW (1975) Reproductive biology of some Costa Rican Cordia species (Boraginaceae) Biotropica 7:234-247 Pacala SW, Canham CD, Saponara J, Silander Jr JA, Kobe RK, Ribbens E (1996) Forest models defined by field-measurements — estimation, error analysis and dynamics. Ecol Monogr 6621—43 Palmiotto P, Davies SJ, Vogt, KA, Ashton MS, Vogt DJ, Ashton PS (2004) Soil-related habitat specialization in dipterocarp rain forest tree species in Borneo. J Eco 92:609-623 Paul M, Driscoll SP (1997) Sugar repression of photosynthesis during N deficiency. Plant Cell Environ 20: 110-116 Paul M], Stitt M (1993) Effects of nitrogen and phosphorus deficiencies on levels of carbohydrates, respiratory enzymes and metabolites in seedlings of tobacco and their response to exogenous sucrose. Plant Cell Environ 16: 1047—1057 138 Peace WJH, Grubb PJ (1982) Interaction of light and mineral nutrient supply in the growth of Impatiens parviflora. New Phytol 90: 127-150 Pearson TRH, Burslem DFRP, Goeriz RE, Dalling J W (2003) Regeneration niche partitioning in neotropical pioneers: effects of gap size, seasonal drought and herbivory on growth and survival. Oecologia 137:456-465 Piotto D, Viquez, Montagnini F, Kanninen M (2004) Pure and mixed forest plantations with native species of the dry tropics of Costa Rica: a comparison of growth and productivity. For Ecol Manage 190:359-372 Poorter H, De Jong R (1999) A comparison of specific leaf area, chemical composition and leaf construction costs of field plants from 15 habitats differing in productivity New Phytol 143: 163-176 Poorter H, Nagel O (2000) The role of biomass allocation in the growth response of plants to different levels of light, C02, nutrients and water: a quantitative review. Aust J Plant Physiol 27: 595-607 Poorter H, Remkes C (1990) Leaf area ratio and net assimilation rate of 24 wild species differing in relative grth rate. Oecologia 83:553-559 Poorter L, Hayashida-Oliver Y (2000) Effects of seasonal drought on gap and understory seedlings in a Bolivian moist forest. J Trop Ecol 16:481—498 Pote DH, Daniel TC, Nichols DJ, Moore PA, Miller DM, Edwards DR (1999) Seasonal and soil-drying effects on runoff phosphorus relationships to soil phosphorus. Soil Sci Soc Am J 63:1006-1012 Pregitzer KS, Hendrick RL, Fogel R (1993) The demography of fine roots in response to patches of water and nitrogen. New Phytol 125: 575 — 580 Reich PB, Ellsworth DS, Walters MB, Vose JM, Gresham C, Bowman WD (1999) Generality of leaf trait relationships: a test across six biomes. Ecology 80: 1955- 1969 Reich PB, Walters MB, Ellsworth DS (1992) Leaf lifespan in relation to leaf, plant and stand characteristics among diverse ecosystems. Ecol Monogr 62:365—3 92 Reich PB, Walters MB, Tjoelker MG, Vanderklein D, Duschena C (1998) Photosynthesis and respiration rates depend on leaf and root morphology and nitrogen concentration in nine boreal tree species differing in relative growth rate. Funct Ecol 12:395-405 Reynolds, HL, D’Antonio, C (1996) The ecological significance of plasticity in root weight ratio in response to nitrogen: Opinion. Plant Soil 185: 75-97 139 Ritchie GA (1982) Carbohydrate reserves and root growth potential in Douglas fir seedlings before and after cold storage. Can J For Res 12: 905-912 Rivera G, Elliott S, Caldas LS, Nicolossi G, Coradin VTR, Borchert R (2002) Increasing day-length induces spring flushing of tropical dry forest trees in the absence of rain. Trees 16:445-456 Roper, TR, Keller JD, Loescher WH, Rom CR (1988) Photosynthesis and carbohydrate partitioning in sweet cherry: fruiting effects. Physiol Plant 72:42-47 Rorison [H (1968) The response to phosphorus of some ecologically distinct plant species. 1. Growth rates and phosphorus absorption. New Phytol. 67: 913-9223 Ryser P (1998) Intra- and interspecific variation in root length, root turnover and the underlying parameters. In H Lambers, H Poorter and MMIVan Vuuren (eds). Inherent variation in plant growth: Physiological mechanisms and ecological consequences. Backhuys, Leiden, The Netherlands. pp 441-465 Sakai A, Sakai S, Akiyama F (1997) Do sprouting tree species on erosion-prone sites carry large reserves of resources? Ann Bot 79:625-630 Same BI, Robson AD, Abbott LK (1983) Phosphorus, soluble carbohydrates and endomycorrhizal infection. Soil Biol Biochem 15:593-597 Scheible WR, Gonzalez-Fontes G, Lauerer M, Muller-Rober B, Caboche M, Stitt M (1997) Nitrate acts as a signal to induce organic acid metabolism and repress starch metabolism in tobacco. Plant Cell 9:783-798 Schnyder H (1993) The role of carbohydrate storage and redistribution in the source-sink relationships of wheat and barley during grain filling — a review. New Phytol 123:233-245 Schreeg LA, Kobe RK, Walters MB (2005) Tree seedling growth, survival and morphology in response to landscape-level variation in soil resource availability in northern Michigan. Can J For Res 35:263-273 Singh KP, Srivastava SK (1986) Seasonal variation in the biomass and non-structural carbohydrate content of fine roots of teak (Tectona grandis L. f.) plantations in a dry tropical region. Tree Phys 1:31-36 Smith, D (1969) Removing and analyzing total nonstructural carbohydrates from plant tissue. Research Report 41, College of Agricultural and Life Sciences, University of Wisconsin, USA 140 Sollins P (1998) Factors influencing species composition in tropical lowland rain forest: does soil matter? Ecology 79: 23—30 Steinlein T, Heilmeier H, Schulze ED (1993) Nitrogen and carbohydrate storage in biennials originating from habitats of different resource availability. Oecologia 93:374-382 Stitt M, Krapp A (1999) The interaction between elevated carbon dioxide and nitrogen nutrition: the physiological and molecular background. Plant Cell Environ 22: 583-621 Stoyan H, De-Polli H, Bohm S, Robertson GP, Paul EA (2000) Spatial heterogeneity of soil respiration and related properties at the plant scale. Plant Soil 222:203-214 Sweeley CC, Bentley R, Makita M, Wells WW (1963) Gas-liquid chromatography of trimethylsilyl derivatives of sugars and related substances. J Am Chem Soc 85:2497-2507 Terashima 1, Evans JR (1988) Effects of light and nitrogen nutrition on the organization of the photosynthetic apparatus in spinach. Plant Cell Physiol 29: 143-155 Tilman D (1985) The resource ration hypothesis of plant succession. Am Nat 1252827- 852 Tisdale SL, Nelson WL, Beaton JD, havlin JL (1993) Soil fertility and fertilizers. Prentice-Hall, New Delhi, India. Tissue DT, Wright SJ (1995) Effect of seasonal water availability on phenology and the annual shoot carbohydrate cycle of tropical forest shrubs. Funct Ecol 9:518—527 Tran TS, Simard RR (1993) Mehlich III extractable elements. In MR Carter (Ed). Soil sampling and methods of analysis. Lewis publishers, Ottawa, Canada. pp 43-49 Treseder KK, Allen MF (2000) Mycorrhizal fungi have a potential role in soil carbon storage under elevated C02 and nitrogen deposition. New Phytol 147: 189-200 Truax B, Lambert F, Gagnon D, Chevrier N (1994) Nitrate reductase and glutamine- synthetase activities in relation to grth and nitrogen assimilation in red oak and red ash seedlings — effects of N forms, N concentration and light intensity. Trees 9: 12-18 Van Vuuren MMI, Robinson D, Griffiths BS (1996) Nutrient inflow and root proliferation during the exploitation of a temporally and spatially discrete source of nitrogen in soil. Plant Soil 178: 185-192 141 Vitousek PM (1984) Litterfall, nutrient cycling and nutrient limitation in tropical forests. Ecology 652285-298 Vitousek PM, Howarth (1991) Nitrogen limitation on land and in the sea: How can it occur? Biogeochemistry 13:87-11 Walsh RPD, Newbery DM (1999) The ecoclimatology of Danum, Sabah, in the context of the world’s rain forest regions, with particular reference to dry periods and their impact. Phil Trans Roy Soc Lond Ser B 354: 1 869-1 883 Walters MB, Reich PB (2000) Seed size, nitrogen supply, and growth rate affect tree seedling survival in deep shade. Ecology 81:1887-1901 Walters MB, Reich PB (1996) Are shade tolerance, survival, and growth linked? Low light and nitrogen effects on hardwood seedlings. Ecology 77: 841-853 Walters MB, Reich PB (1999) Low-light carbon balance and shade tolerance in the seedlings of woody plants: do winter deciduous and broad-leaved evergreen species differ? New Phytol 143: 143-154 Waring RH, Pitrnan GB (1985) modifying lodgepole pine stands to change susceptibility to mountain pine-beetle attack. Ecology 66: 889-897 Webb WL (1981) Relation of starch content to conifer mortality and growth loss after defoliation by the Douglas-fir tussock moth. For Sci 27:224-232 Wurth MKR, Pelaez-Riedl, Wright SJ, Komer (2005) Non-structural carbohydrate pools in a tropical forest. Oecologia 143:] 1-24 Zak DR, Pregitzer KS, Host GE (1986) Landscape variation in nitrogen mineralization and nitrification. Can J For Res 16:1258-1263 142 ""l.