Functional Ecology British Ecological Society

Functional Ecology 2010, 24, 1171-1180 doi: 10.HH/j.1365-2435.2010.01731.x Plant-soil associations in a lower montane tropical forest: physiological acclimation and herbivore- mediated responses to nitrogen addition

Kelly M. Andersen1'3*, Marife D. Corre2, Benjamin L. Turner1 and James W. Bailing13 1 Smithsonian Tropical Research Institute Apartado 0843-03092, Balboa, Ancdn, Panama; 2Buesgen Institute - Soil Science of Tropical and Subtropical Ecosystems, Georg-August University of Goettingen, Buesgenweg 2, D-37077 Goettingen, Germany; and3 Department of Plant Biology, University of Illinois, Urbana, Illinois 61801, USA

Summary 1. Soil nutrients influence plant productivity and community composition in tropical forests. In lower montane tropical forests in western Panama, the distribution of understory palm species over a scale of 1-20 km correlates with differences in soil nitrogen (N). We hypothesized that soil N determines seedling performance in the forest understory, and, may therefore influence species distributions along the soil N gradient. 2. We explored the potential for N availability to generate species-habitat associations through species-specific differences in biomass allocation, photosynthetic capacity, N use-efficiency, and susceptibility to herbivory. Seedlings of nine palm species from two sub-families and four habitat types were transplanted into N-addition and control plots at a low N site. Growth, mortality, biomass allocation, photosynthesis, foliar N content and herbivory were measured over 21 months. 3. Foliar N increased for all species (15-68%) following N addition. Most species showed strong (20-200%) increases in photosynthetic rates with N addition except two species with marginal decreases in photosynthetic rates (5-15%). However, shifts in physiological traits did not increase relative growth rate or change in biomass allocation for any species or N treatment combination. Rather, increased leaf quality contributed to greater levels of herbivory in species associated with soils of intermediate and high inorganic N availability. 4. Thus, potential increases in overall growth with N addition were masked by herbivory, resulting in no apparent growth response with increased N. We suggest that for understory palms, and potentially other montane forest plants, distribution patterns are driven by a combination of physiological and herbivore-mediated responses to soil nutrient availability. Key-words: Arecaceae, fertilization, growth trade-offs, habitat association, Panama, seedling transplant experiment

& Cuevas 1998; Adamek, Corre & Holscher 2009). Narrow Introduction N : P ratios in leaves from tropical montane forest trees Nitrogen (N) and phosphorus (P) can limit net primary pro- suggest N limitation (Tanner, Vitousek & Cuevas 1998; ductivity in terrestrial ecosystems (Vitousek & Howarth Soethe, Lehmann & Engels 2008; Wilcke et al. 2008), 1991; Lebauer & Treseder 2008). For tropical forests, P sup- whereas fertilization studies in montane forests of Hawaii, ply is often assumed to limit productivity on highly weath- Jamaica and Venezuela indicate that tree growth is limited ered soils in the lowlands (Walker & Syers 1976; Vitousek & by either N or N and P together (reviewed by Tanner, Vito- Sanford 1986), whereas at higher elevations, where climate usek & Cuevas 1998). No experiments have examined N lim- is similar to temperate forests and soils may be pedogenically itation of seedling establishment in tropical montane forests. young, evidence for nutrient limitation is varied (Tanner Habitat specialization with respect to soil nutrients is com- et cil. 1990; Vitousek & Farrington 1997; Tanner, Vitousek monly documented, but rarely examined experimentally. Recent experimental studies in lowland tropical forests have "Correspondence author. E-mail: [email protected] suggested that three trade-offs influence habitat specialization © 2010 The Authors. Journal compilation © 2010 British Ecological Society 1172 K. M. Andersen et al.

(Fine, Mesones & Coley 2004; Palmiotto et al. 2004; Baltzer (Geonomateae) palms of the Arecoid subfamily are common et al. 2005). First, habitat associations may be determined by elements of the understory of montane forests, with most spe- a trade-off between investment in aboveground biomass to cies reaching < 5 m in height at reproductive maturity. Cha- achieve a higher relative growth rate and belowground bio- maedoroid and Geonomoid palms share similar vegetative mass to acquire limiting nutrients (Chapin 1980; Aerts, Boot morphologies and habitat requirements, but are estimated to & van der Aart 1991; Wilson & Tilman 1993; Aerts & Chapin have diverged 70 million years ago (Cuenca, Asmussen- 2000; Palmiotto et al. 2004). Secondly, habitat associations Lange & Borchsenius 2008). Given the ecological similarities may be determined by a trade-off between growth and the between the groups, we examined whether there was a phylo- efficiency with which limiting soil resources are used (Chapin genetic component to seedling response to soil N availability. 1980; Chapin, Autumn & Pugnaire 1993; Aerts & Chapin To test whether soil N availability determines palm species 2000; Baltzer et al. 2005). At sites where nutrients are limit- distribution patterns, we compared the response to N addi- ing, resource-use efficiency (RUE) can be increased by tion of palms from the three species-soil groups and from an increasing leaf longevity and decreasing foliar nutrient con- additional group of generalist species that showed no habitat centration. Collectively, these traits enhance nutrient use effi- preference. We conducted a field transplant experiment in the ciency by increasing carbon gain per unit of nutrient uptake forest understory to address the following hypotheses: (i) if (Aerts & Chapin 2000). Thirdly, habitat associations may habitat specialization arises directly from differences in the arise from trade-offs between growth and allocation to biomass allocated to acquire soil N, then species with greater defense against herbivores. The level of investment in anti- allocation to aboveground biomass will have a growth advan- herbivore defense should increase as plant growth rates and tage with N addition; (ii) if habitat specialization is deter- resources available to replace leaf tissue lost to herbivory mined by nutrient use efficiency (NUE), then species with decrease (Coley, Bryant & Chapin 1985). Thus, different sets lower NUE or higher foliar N concentrations and higher pho- of trade-offs between growth and allocation to resource tosynthetic capacity will achieve greater growth with N addi- acquisition, use, or defense may determine species distribution; and (iii) if habitat specialization is mediated by tions along soil gradients. herbivores through a trade-off between allocation to growth We used an experimental approach to explore mechanisms versus defense, then species with the strongest growth underlying observed habitat partitioning of plant species response to N addition will also experience the highest herbiv- across a soil fertility gradient in a lower montane forest in ory rates and/or decreased growth. western Panama. High landscape-level diversity in these forests is associated with variability in soil nutrients. Total Materials and methods inorganic N availability ranges from < 1 to 8 ug N cm-3 within the Fortuna Forest Reserve and surrounding pro- SITE DESCRIPTION tected areas (Andersen, Turner & Dalling 2010). Multivariate analyses of soil nutrients and understory palm communities To examine the effect of soil N availability on seedling performance, identified three palm-soil associations: (i) species associated we conducted a field transplant experiment nested within a larger N with soils developed on rhyolite, characterized by an organic fertilization experiment (the NITROF project; Koehler et al. 2009; Adamek, Corre & Holscher 2009; Corre et al. 2010, in press). The surface horizon and low nutrient availability; (ii) species asso- study area is located within the Quebrada Honda watershed ciated with soils developed on andesite, characterized by min- (8°45'40"N, 82°14'22"W; elevation 1200-1300 m a.s.l.) with soil eral soils of intermediate nutrient availability; and (iii) species developed on rhyolitic tuff at the low end of the N gradient (based on associated with soils developed on porphyritic dacite, charac- total inorganic N) found in Fortuna Reserve (Chiriqui province. terized by an organic surface horizon and high nutrient avail- Republic of Panama; Table 1). The soil is derived from volcanic ash ability (Andersen, Turner & Dalling 2010). Variation in understory palm community composition among sites was most strongly correlated with differences in inorganic N, Table 1. Site environmental characteristics measured in May 2007 although experimental tests are necessary to link species from control and nitrogen addition plots (starting in January 2006) distribution patterns to individual nutrients and to provide Environmental variable Control N addition support for mechanisms underlying palm-soil associations.

Note that identification of the geology of the third group of P^lwaler 3 97 ± 010 3 88 ± 011 3 soils as developed on dacite replaces the previous description NH4 (ug N cm" ) 248 ± 019 2 66 ± 0 53 3 (granodiorite; Andersen, Turner & Dalling 2010) following N03 (ug N cm" ) 051 ± 013 101 ± 026 Al (|ig Al cm"3) 144 ± 111 129 ± 22-5 more detailed geological investigation of the site. Ca (ug Ca cm"3) 714 ± 119 101 ± 22-0 Palms are a major component of the understory in tropical K (ug K cm"3) 190 ± 149 28 1 ± 679 forests, modifying micro site conditions for seedling establish- Mg (ug Mg cm"3) 234 ± 351 369 ± 13 1 ment (Farris-Lopez et al. 2004; Wang & Augspurger 2004). P (ug P cm"3) 094 ± 014 1 74 ± 0 57 3 Therefore, information on palm ecology and processes influ- Zn (ug Zn cm" ) 043 ± 007 061 ± 018 Red : Far red ratio 023 ± 003 0 32 ± 002 encing palm species distributions has direct implications for Canopy openness (%) 5 37 ± 0 54 6 55 ± 0 54 understanding regeneration within tropical forests. Both Chamaedorea (tribe Chamaedoreeae) and Geonomoid Values represent mean (n = 4) with standard errors. © 2010 The Authors. Journal compilation © 2010 British Ecological Society, Functional Ecology, 24, 1171-1180 Seedling responses to nitrogen addition in a tropical montane forest 1173 deposits, has a sandy loam texture, and is classified as Alic Hapludand Each garden consisted of two seedlings of each of nine species planted (Koehler et al. 2009). Soils are continually moist, with an average 25 cm apart to avoid shading and belowground competition. Seed- annual rainfall for 2007 and 2008 of 7800 mm year-1 (Andersen, lings had one fully expanded leaf at the time of transplantation and Turner & Dalling 2010) and a mean annual temperature of 19 °C showed no sign of transplant stress and overall mortality rates were (Cavelier, Soils & Jaramillo 1996). The area is classified as lower mon- low throughout the experiment. tane rain forest in the Holdridge life zone system and consists of Seedlings were monitored for survival, leaf number, leaf area and mature forest with a closed canopy with a mean height of 20 m (Ada- leaf damage in bimonthly censuses beginning in February 2006. After mek, Corre & Holscher 2009). Oreomunnea mexicana (Juglandaceae), 20 months, we measured maximum photosynthetic rates at Vochysia guatamalensis (Vochysiaceae), Colpothrinax aphanopetala 400 umol m-2 s-1 on of a subset of individuals for each species x N (Arecaceae) and Podocarpus oleifolius (Podocarpaceae) are common treatment combination. The most recently expanded fully mature leaf canopy species within the study area (Fig. 1). was chosen for physiological and foliar nutrient measurements. Pho-

tosynthetic rates (Amla), stomatal conductance, and transpiration rates were assessed in the field using a portable gas exchange system FERTILIZATION EXPERIMENT (LI-6400; LiCor, Lincoln, NE, USA). Instantaneous water use effi- The fertilization experiment consisted of four 40 x 40 m paired con- ciency (WUE) was calculated from Amax and transpiration. Foliar N trol and N addition (N +) plots with 40 m separating paired plots concentration was determined by dry combustion using an elemental and 100 m separating pairs of plots. Nitrogen addition plots received analyzer (Costech Analytical Technologies Inc, Valencia, CA, USA). 125 kg urea-N ha-1 year-1 applied in four intervals per year begin- Foliar P concentration was determined by ignition at 550 °C and ning in January 2006. Gross rates of soil N cycling and microbial bio- extraction in 1 M H2S04, with P detected by molybdate colorimetry mass were initially low and increased in the first year of N addition using a flow injection analyzer (Latchat QuikChem 8500; Hach Ltd., (Corre et al. 2010, in press). Annual fine litterfall also increased dur- Loveland, CO, USA). Photosynthetic N use efficiency (PNUE) was ing the 2 years of N addition compared to the control (Adamek, calculated from AnmA and foliar N concentrations. Photosynthetic P Corre & Holscher 2009). These responses to N addition indicate that use efficiency (PPUE) was calculated from Amm and foliar P concen- canopy trees at the study site were N-limited. trations. Seedlings were harvested to assess relative growth rates, biomass allocation, and foliar nutrient contents. Seedlings were separated into root, shoot, and leaf components to weigh and mea- SEEDLING TRANSPLANT EXPERIMENT sure leaf toughness and leaf area prior to drying. Leaves were digitally scanned to quantify leaf area and herbivory damage using ImageJ We selected nine understory palm species based on their distribution software (Rasband 2008). Leaf toughness was measured as fracture patterns at Fortuna, including at least one Chamaedorea and one toughness using a 516-1000MRP push-pull gauge 'penetrometer' Geonomoid species belonging to each of three palm-soil association (Chatillon/Amtek, Largo, FL, USA). The following growth parame- groups as well as a fourth group of generalists that includes species ters were calculated following Hunt (1982): relative growth rate found across the soil nutrient gradient (Table 2). Seeds were collected (RGR), net assimilation rate (NAR), leaf area ratio (LAR), specific from the Fortuna Reserve in May 2005 and seed mass was recorded leaf area (SLA), root mass ratio (RMR), and leaf mass ratio (LMR). on a subset of 10 seeds per species. Seeds were germinated in a grow- We characterized the environmental conditions of paired common ing house at the Smithsonian Tropical Research Institute Fortuna gardens in May 2007. Hemispherical photographs were taken at the Station in a 50 : 50 mix of washed sand and soil. The soil was center of each experimental garden to measure canopy openness. collected from forests at the NITROF site. Red : far-red ratios were taken above each seedling. Soil chemical In January 2006, we transplanted seedlings into common gardens properties (extractable NH4, N03 P, Ca, K, Mg, Al, Fe) were mea- (75 x 75 cm) in the forest understory (closed canopy) within each of sured in May 2007. Phosphorus and cations were extracted in Meh- four paired control and N addition plots prior to the first fertilization. lich III solution and determined by inductively coupled plasma (ICP) Hence, the experimental design was a split plot with four replicate optical-emission spectrometry on an Optima 2100 spectrometer (Per- blocks, with N treatment as the whole-plot and species as the subplot. kin Elmer Inc., Waltham, MA, USA). Nitrogen was extracted in 2 M KC1 directly in the field and analysed using continuous flow injection colorimetry on a Latchat QuikChem 8500 (Hach Company, Love- land, CO, USA).

DATA ANALYSIS

Growth, biomass allocation, physiological parameters and herbivory were analysed using analysis of variance with N treatment, palm-soil association groups, genera and species as fixed effects, and the blocking structure as a random effect. The blocking structure consisted of N addition treatment nested within four blocks. We tested the main effects and interactions using linear mixed-effect models and conducted post-hoc tests for significant effects after Bonferroni cor- rections (Proc mixed; SAS Institute, Cary, NC, USA). Initial biomass was used as a covariate in the ANCOVA for RGR. Values for RGR and biomass were log transformed and values of proportion leaf area missing were arcsine square root transformed prior to analyses to meet assumptions for ANOVA. Leaf area damage was analysed in two Fig. 1. Photograph of the field site in the Honda watershed. parts to control for non-normality. First, we tested for differences in

Table 2. List of species for the species-soil association groups used in the study. Nomenclature is based on Henderson, Galeano & Bernal (1995), Hodel (1997) and Henderson (2005). Soil associations were based on Andersen, Turner & Dalling (2010). Nutrient (Mehlich III extractable base cations and phosphorus), and inorganic nitrogen availability (2 M KC1 extractable) increases from rhyolitic to andesitic to dacitic soils (Andersen, Turner & Dalling 2010)

Genus Species Code Soil association Growth form

Chamaedoreeae Chamaedorea pinnatifrons (Jacq.) Oerst. CP1 Generalist Arborescent Chamaedorea recurvata Hodel CRT Rhyolitic Arborescent Chamaedorea deckeriana (Klotzsch) Hemsl. CD Andesitic Arborescent Chamaedorea tepejilote Liebm. CT Andesitic Arborescent Chamaedorea woodsoniana L. H. Bailey cw Dacitic Arborescent Geonomateae Geonoma cuneata var. euneata H. Wendl. ex Spruce GC Generalist Decumbent Geonoma cuneata var. gracilis (H. Wendl. ex Spruce) Skov GG Rhyolitic Arborescent Calyptrogyne panamensis var. occidentalis Henderson CAG Andesitic Decumbent Geonoma jussieuana Mart. GJ Dacitic Arborescent

the frequency of plants that encountered herbivory (incidence; P < 0-0001). The small Geonoma cuneata var. gracilis, associ- 0 = no herbivory, 1 = herbivory) among treatments using proc ated with N-poor soils achieved the highest RGR, whereas glimmix (SAS Institute), with a binary distribution and a logit link the large Chamaedorea tepejilote and C. woodsoniana, associ- function. Second, we tested for differences in the mean proportion of ated with N-rich soils, had the lowest RGR. Relative growth leaf area missing for plants with an incidence of herbivory (inci- rate was negatively correlated with seed mass and initial bio- dence = 1) among treatments using the lme function with a power mass (Pearson's r = -0-35, P < 0-0001 and r = -040, variance function structure to fit the model (nlme package; Pinheiro P < 0-0001). Overall survival during the experiment was gen- & Bates 2008) in R (R Core Development Team 2008). A variance function was used to correct for heteroscedasticity among the within erally high, ranging from 80% to 100% survivorship, except group errors (Pinheiro & Bates 2000). for the two species from high nutrient soil, which had survival To examine bivariate relationships we use standardized major axis rates of 69% (Chamaedorea woodsoniana) and 75% (Geono- (SMA), a type II regression technique (Warton et al. 2006). SMA ma jussieana), distributed across both treatments. regression techniques are more appropriate when examining variables with measurement error than standard linear regression (Sokal & Rolf 1995). Differences between slopes for the N addition treatment and BIOMASS ALLOCATION control were tested. If a common slope was detected, differences in Nitrogen addition did not affect on total biomass or bio- intercepts and shifts along the common axis were also tested following mass allocation (Table 3). However, several morphological routines and procedures in the 'smatr' package available in the R traits varied with species-soil association. Rhyolitic species statistical environment. had lower RMR than species from higher nutrient soils (andesitic and dacitic), whereas generalist species did not differ from any of the species-soil groups (F,;, = 3-84, Results P < 005). Dacitic species had lower LMR than rhyolitic and generalist species, whereas andesitic species did not dif- RELATIVE GROWTH RATES fer from any of the species-soil groups (Fi 59 = 306, The understory palms in this study had inherently slow P < 005). Andesitic species had significantly higher SLA growth rates ranging from < 0 50 to 2-45 mg g_1 day-1 and rhyolitic (low nutrient) species had significantly lower

(Table 3). Nitrogen addition did not affect on growth SLA than dacitic and generalist species (F35\ = 78-36, (Fig. 2a), although growth rates differed among species-soil P < 0-0001). There were no differences in LAR among the associations (f^g = 847, P < 0-0001). Species associated species-soil groups. with N-rich soils (andesitic and dacitic) had significantly There were species level differences in specific leaf area, lower growth rates than species associated with rhyolitic, SLA CFs.93-3 = 13-82, P < 0-0001), and leaf area ratio, LAR N-poor soils (Fig. 2a). The RGR of generalist species did not (^8,94-2 = 481, P < 0-0001). Chamaedorea tepejilote, with differ from any of the species-soil association groups. The large, thin, papery leaves, had the highest SLA (450 cm g_1), rhyolitic and generalist species and one andesitic species, whereas C. recurvata, with very thick, almost succulent leaves, Calyptrogyne panamamensis, naturally occur at the study site, had the lowest SLA (280 cm2 g_1). For most species, root bio- whereas the remaining andesitic and dacitic species are not mass accounted for about 20% of the total plant biomass and found in the forests immediately surrounding the transplant did not significantly differ significantly among species. experiment. Chamaedorea tepejilote had the highest (024) and Geonoma Species level differences in RGR were detectable even after cuneata var. gracilis had the lowest (015) root mass ratios controlling for differences in initial biomass (F%% = 7-71, (Table 3). However, species did differ in allocation to leaf

biomass (i^.ioo = 793, P < 0-0001): Calyptrogyne panam- f 2 ensis allocated 50% of its biomass to leaves, whereas Chamae- \b ? ^ ok dorea tepejilote allocated only 24% of its biomass to leaves. « o s O oc rl o — — Tf Leaf mass ratio was positively correlated with RGR < -f ? '-& K -f s 3 -i (r = 0-57, P < 0-0001). Seed mass was negatively correlated with both LMR (r = -033, P < 00005) and LAR (r = - 0-42, P < 00001) for log-transformed data, but was not o S ° 2 2 P| => (N o o o o o o o o o related to RMR or SLA. s _ ? fS - r'\ rP ^ 3 PHYSIOLOGICAL RESPONSES oc K k" - m K VI 6 2 2 •f < Foliar nutrients c OH In contrast to biomass allocation, foliar N content responded strongly to N addition and differed among species-soil associations, being 30% higher in the N addition treatment com-

pared to the control (F13 = 14-61, P < 005). Daciticspecies a! OO ^D OO had significantly higher foliar N content than species from the o I other soil groups (F = 440, f < 001; Fig. 2b). Foliar N 351 content ranged from 1-7% in G. cuneata var. gracilis to 2-5% a ^ 2 "S "* "* rn cp op in G. jussieunana in the control treatment and from 2-1% in Calyptrogynepanamensis to 3-4% in C. woodsoniana in the N addition treatment. % 9 p? "J- •~< f •»' 2 Foliar N was linked to several important performance ^-3 variables (Fig. 3). Foliar N was related to a decrease in #5 •* O* •§ t/3 9 i, RGR across both N addition and control treatments a! (Fig. 3a), with a common slope of -180 (95% CI: -250 to — o o o = 2'8 2 >£ -1-31). However, seedlings in the N addition treatment had ji ^ a higher RGR at a given foliar N content compared to spe- < cies in the control treatment (P < 00001). Three species (Chamaedorea tepejilote in N addition and control; C. P !% S P •* •* v? £ S woodsonia in control; Geonoma juiessiana in N addition) had

a. £ th13 a a a mean of 20-50% leaf area damage. As a result, these val- U1 ^ r> w m the control and N addition treatments (Fig. 4). However, " u n y 0 n H

Rhyolitic Andesitic Dacitic Generalist Rhyolitic Andesitic Dacitic Generalist

(d)50-

Fig. 2. Mean ± standard error bar (a) relative growth rates (RGR), (b) foliar nitrogen content, (c) area-based maximum photosynthetic rate (^max(area)), and (d) leaf area damage of the four species-soil association groups. Different letters indicate significant differences among species groups (linear mixed-effect model with Bonferroni correc- tions at P < 0.05). Note: 'Variable with a Rhyolitic Andesitic Dacitic Generalist Rhyolitic Andesitic Dacitic Generalist significant difference between control and Species-soil association nitrogen addition plots.

foliar N concentrations, there were no differences in foliar res- Gas exchange piration rates, WUE, PNUE, or PPUE among any of the Understory palm photosynthetic rates were low overall treatment groups. 2 _1 (<2 umol C02 m~ s ), but there were strong effects of N addition on gas exchange rates. Both area- and mass-based PLANT-HERBIVORE INTERACTIONS maximum photosynthetic rates were higher in N addition than control plots (Fu3 = 33 0, f < 005; andFl3 = 29-92, The incidence of herbivory was similar between the N treat- P < 005 respectively). Mean area-based maximum photo- ments with 72% of the plants in the control treatment and synthetic rates (Am!lx (area)) increased from 1-12 umol - 85% of the plants in the N addition treatment having some -2 _1 C02 cm s in the control plants to 163 umol level of herbivory. Andesitic and dacitic species had higher -2 _1 C02 cm s in plants receiving N addition. Dacitic species overall leaf area damage, 29 and 24% respectively, compared had higher area-based photosynthetic rates than species with to generalists and species from N-poor soils with 19 and 16% other soil associations (Fi5&.7 = 3-38, P < 005; Fig. 2c). respectively (f\si = 2-76, P = 005; Fig. 2d). Species dif-

There were also significant effects of species-soil group fered in leaf area damage (FS55 = 716, f < 0-0001), ranging (^3,48 = 710, f < 0-0001) and interaction between N addi- from an average of 12% in G. cuneata var. gracilis to 40% in tion treatment and species-soil group (FiAS = 2-80, C. tepejilote across all treatments. Mean leaf area damage was P < 005) on mass-based photosynthetic rates, with dacitic 22% in the N addition treatment compared to 15% in the species showing a greater response to N addition compared control treatment (F\ ,3 = 143, f < 005; arcsine square root to andesitic and rhyolitic species and generalist species show- data). There was a significant interaction between species and ing an intermediate response. Individual species did not differ N addition on leaf area damage (FS55 = 2-83, P < 005). in photosynthetic rates (Table 3). There were significant Geonomajussieana, a species associated with N-rich soils, suf- effects ofN addition treatment CFj,3 = 58-36, P < 001) and fered significantly more herbivory damage in the N addition an interaction between N addition treatment and species-soil treatment (F13 = 10-53, P < 005), whereas other species group (-F3.48 = 4-73, P < 001) on stomatal conductance, did not show significant differences between the N treat- with generalist species showing a greater increase in conduc- ments. tance rates than species with soil associations. Among species, Differences in leaf damage were not related to leaf tough- gas exchange was correlated with foliar N content on both a ness and there were no differences in leaf toughness between mass and area basis (mass-based N Pearson's correlation N treatments or among species-soil groups. However, species coefficient r = 066, P < 0-005; area-based N r = 0-70, differed in leaf toughness (Fs 50 = 504, f < 0001) ranging P < 0-005). Despite increases in photosynthetic capacity and from 30 to 188 g of force to penetrate a 3 mm rod through © 2010 The Authors. Journal compilation © 2010 British Ecological Society, Functional Ecology, 24, 1171-1180 Seedling responses to nitrogen addition in a tropical montane forest 1177

(a) 30 o~ >, 20 _ ^* ffl ~~~0 T3 o

• 1-0

o • a: os CTc GJn 0E O CWc • CTn 02 -

(b) 50 - • o o • 20 - o o u> 03 * * • E 10 - 03 -'o T3 Foliar N (%) o 0) 5 - o Fig. 4. Relationship between foliar nitrogen and phosphorus in seed- o lings growing in control (open symbol) and nitrogen addition (filled 2 - symbol) treatments in a field experiment. Symbols represent species means per treatment. The lines represent the SMA regressions between foliar phosphorus and foliar nitrogen for seedlings in the 80 - nitrogen addition (solid line) and control (dashed line) treatments. CO Foliar phosphorus increased with foliar nitrogen along a common

O) 60 - slope of 0.94(95% CI: 0.59-1.52), with a lower foliar phosphorus at a given foliar nitrogen content for the seedlings in nitrogen addition o E treatment (P < 0.01). Note that the outlier in the control treatment is c 40 Chamaedorea tepejilote, indicated by CT, and is not included in the regression analysis.

20 - i 1 r 25 30 35 Foliar N (%)

Fig. 3. The effect of nitrogen fertilization on the relationship between foliar nitrogen (%) and (a) relative growth rate (RGR), (b) leaf damage and (c) mass-based photosynthetic rate (^max(mass)) for nine species of understory palm. Symbols represent species mean for the nitrogen addition (filled) and control (open) treatments. Note that the two very low RGR values in both control and nitrogen addition treatments are not included in the analysis. the leaf lamina. Furthermore, leaf toughness increased with increasing foliar N content along a common slope of 4-22 (95% CI: 2-76-646) with higher foliar N content at a given 5 10 20 leaf toughness for the plants in the N addition treatments Leaf damage (%) (P < 0-001). Differences in leaf area damage may have masked the Fig. 5. The relationship between leaf area damage (%) and relative anticipated increase in RGR with N addition. Relative growth rate (RGR) for nine species of understory palm growing in growth rate decreased with increased leaf area damage along nitrogen addition (filled) and control (open) treatments. a common slope of -085 (95% CI: -058 to -1-26) for both the control and N addition treatments (Fig. 5). There were no PHYLOGENETIC COMPARISONS differences in elevation between the N treatments or shifts along the common axis. Thus, the higher leaf area damage in Despite morphological and ecological similarities, we the N treatment resulted in RGR values similar to those in detected several differences between Chamaedorea and Geo- the control. nomoid palms. Chamaedorea species had significantly lower

© 2010 The Authors. lournal compilation © 2010 British Ecological Society, Functional Ecology, 24, 1171-1180 1178 K. M. Andersen et al. growth rates than Geonomoid species (F, ^ = 46-46, MECHANISMS FOR SOIL-BASED HABITAT ASSOCIA- P < 0-0001) and significantly lower leaf mass ratios TIONS (fi,6i = 19-43, P < 0-0001), but there were no differences in allocation to root biomass between the genera. Chamaedorea Field experiments have provided evidence for several mecha- species had overall lower LAR than Geonomoid species nisms in generating soil-based habitat partitioning in tropical (fi,6i = 12-87, P < 0-001), but Chamaedorea species had a forests, including trade-offs between growth and biomass higher SLA than Geonomoid species (Fi<6i = 10-20, allocation patterns, resource-use efficiency, and herbivore P < 0-005). Chamaedorea species also had higher foliar N defense can influence species-soil associations (Fine, Mesones contents (2-55%) than Geonomoid species (2-23%) & Coley 2004; Palmiotto et al. 2004; Baltzer et al. 2005). Our {Fijsi = 41-34, P < 0-0001). Chamaedorea seedlings had sig- results provide support for both physiologically based and nificantly higher WUE compared to Genomoid seedlings herbivore-mediated responses to soil N addition among (fi,60 = 400, P < 005). There was a significant N addition closely related groups of understory palms. treatment by genus interaction, with Chamaedorea seedlings increasing ^4 (mass) by 27% and Genomoid seedlings max Biomass allocation increasing ^max(mass) by 39%^ 60 = 430, f < 0-05). Simi- larly, there was a significant interaction between N treatment We found no responses in biomass allocation patterns to soil and genus for PNUE, with Chamaedorea increasing PNUE N addition, yet many growth experiments have found shifts with N addition by only 6% compared to the 24% increase in RMR with soil N availability (Tilman & Wedin 1991; Aerts for Genomoid seedlings (F,^o = 5-14, P < 005). Chamae- & Chapin 2000; Walters & Reich 2000). Phenotypic shifts in dorea seedlings had significantly higher levels of herbivory RMR with soil N were particularly common in fast-growing with an average of 28% leaf area missing compared to 16% species (Aerts & Chapin 2000), which may explain why the leaf area missing for Geonomoid species (-Fi,6i = 4-78, slow-growing understory palms in the current study showed P < 0-005). However, the effect of N addition was relatively no shift in biomass allocation patterns with N addition. Alter- mild for Chamaedorea species (12% increase in leaf damage), natively, consistent RMR among control and N addition compared to Geonomoid species (76% increase in leaf dam- plots may indicate that plants were foraging for additional age) (Fl60 = 515, f < 0-05). Chamaedorea species also had soil resources, such as P or water. greater leaf toughness than Geonomid species (Fj 57 = 32-22, Biomass allocation patterns differed among contrasting P < 0-0001) across both treatments. There were no genus species-soil associations. Economic theory suggests that fas- level differences in respiration, conductance or photosyn- ter-growing species from fertile sites should allocate relatively thetic rates on an area-basis. more biomass to leaves compared to slower-growing species from infertile sites (Chapin 1980; Bloom, Chapin & Mooney 1985; Aerts & Chapin 2000). There was a strong positive cor- Discussion relation between LMR and RGR across species and N treat- The extent to which soil N limits productivity in tropical ments in the current experiment. However, species from montane forests remains unclear, in part because few studies nutrient-rich dacitic soils had higher RMR and lower LMR have experimentally manipulated nutrient concentrations in and RGR compared to species from nutrient-poor rhyolitic such ecosystems (Vitousek & Farrington 1997; Tanner, Vito- soils. It is important to note that the rhyolitic species naturally usek & Cuevas 1998; Adamek, Corre & Holscher 2009). Here, occur at the site of the experiment. Therefore, the locally N addition had no direct effect on growth of palm seedlings adapted rhyolitic species may have employed alternative in the shaded understory after 21 months. Rather, strong mechanisms, such as symbiosis with native soil biota, allow- increases in foliar N increased photosynthetic rates, but ing them to maintain a high LMR and, thus, RGR. The dacit- higher quality leaves were more susceptible to herbivory, ic species, on the other hand, may have been nutrient-stressed masking any potential growth response of the seedlings. across both treatments and increased their investment in the Thus, the lack of growth response to N addition may be medi- acquisition of belowground resources at the expense of LMR ated by herbivores or limitation by additional resources. For and RGR. example, low light availability and soil P concentrations may have restricted the response of the seedlings to added N, as Physiological responses has been shown in other studies (Walters & Reich 2000; Palm- iotto et al. 2004; Baltzer et al. 2005). Ratios of foliar N : P We found significant increases in foliar N concentration and were higher for plants in the N addition treatment compared photosynthetic capacity with N addition. However, there to the control, suggesting that P became increasingly limiting. were no differences in potential PNUE among any of the Further experimental manipulations of light and soil P avail- treatments or species groups. Increased photosynthetic ability would be necessary to show limitation from other capacity with foliar N content suggests that at least some of resources. Nonetheless, our results suggest that soil N affects the additional N was invested in N-rich photosynthetic struc- physiological traits and herbivore pressure of understory tures (Field & Mooney 1986; Lambers, Chapin & Pons 1998). plants and, thus, may influence species distributions along soil Reich, Walters & Ellsworth (1997) showed a strong correla- N gradients in this montane forest. tion between foliar N content and maximum photosynthetic © 2010 The Authors. Journal compilation © 2010 British Ecological Society, Functional Ecology, 24, 1171-1180 Seedling responses to nitrogen addition in a tropical montane forest 1179 rates across a wide spectrum of ecosystems. The increased N concentration and potential photosynthetic capacity. How- photosynthetic rate following N addition suggests that soil N ever, the cost of increased foliar N, or leaf quality, was a cor- limited physiological function of seedlings in the control gar- responding increase in the amount of leaf area consumed by dens at this low N site. However, higher potential photosyn- herbivores. Thus, generalist species and species associated thetic rates did not translate to increased growth rates, with low N soils maintained higher realized growth rates than possibly due to low understory light levels, increasing P limi- species with andesitic or dacitic soil associations. Realized tation and/or increases in herbivory. growth is the growth potential minus the tissue losses (Kitaj- Physiological traits were consistent with species-soil pref- ima 1996). The similar realized growth rates in control and N erences whereby species associated with relatively nutrient- addition treatments, despite higher herbivory in the N addi- rich, dacitic soil exhibited higher foliar N content and tion treatment, suggests that potential growth rates of species potential photosynthetic rates than species with preferences associated with dacitic and andesitic soils were higher in the for low nutrient soils, suggesting that physiological traits N addition treatment. Thus, leaf quality may influence species are important in determining species-soil associations (Cha- distribution patterns indirectly through susceptibility to pin, Autumn & Pugnaire 1993; Aerts & Chapin 2000). Pre- herbivore attack. vious soil nutrient augmentation experiments with tropical woody species have not detected responses in leaf level PHYLOGENETIC CONSTRAINTS physiology with increases in nutrient availability (Tanner, Vitousek & Cuevas 1998; Baltzer et al. 2005). In contrast, Phylogenetic constraints may also influence species distribu- the accumulation of foliar N in response to increased N tion along soil N gradients. Physiological and herbivore- supply may be a key mechanism for understory palms to mediated responses to N addition revealed a phylogenetic persist in low light environments. Increased uptake of N signature regardless of species-soil association. After con- may be advantageous to slow-growing palms, allowing them trolling for differences in species-soil group, Chamaedorea to store N for later use when ephemeral supplies are species consistently had higher foliar N contencentrations, exhausted (Chapin 1980; Lawrence 2001). However, the her- gas exchange rates, and consequently, herbivory, whereas bivory costs associated with high foliar N content suggests Geonoma species, with lower quality leaves, maintained that N-rich leaves may be particularly disadvantageous for higher RGR. Furthermore, Chamaedorea species had palms at low soil N sites and may influence species-soil greater leaf toughness than Geonoma species, indicating the associations. two groups differ in nutrient use and defense strategies. This suggests that Geonoma species, with their more conservative nutrient strategy, should be able to occupy more nutrient- Plant-herbivore interactions poor sites compared to Chamaedorea species within a given Palms are generally slow growing plants that invest heavily soil-based habitat. This pattern may account for the in structural defenses (lignin, fibre, spines) rather than sec- regional distribution of Geonoma and Chamaedorea species ondary chemical defenses (Braker & Chazdon 1993; Coley in Panama. Geonoma reaches its highest species diversity in & Barone 1996). Structural defenses limit the suite of herbi- eastern Panama where soils are generally more weathered vores capable of accessing leaf tissue, and low nutritional and nutrient poor (reflecting a longer time since volcanic quality may also serve as an anti-herbivore defense mecha- activity), whereas Chamaedorea reaches its highest species nism, as well as a means of optimizing metabolism and diversity in western Panama where soils are generally growth rate in low resource environments (Moran & Hamil- younger and more nutrient rich (reflecting more recent vol- ton 1980; Coley & Barone 1996; Campo & Dirzo 2003; Fine canic activity) (Henderson, Galeano & Bernal 1995; Hodel et al. 2006). Both the present study, and a study comparing 1992). herbivory among three Geonomoid species at the La Selva In conclusion, we found that the main response of under- Biological Station, Costa Rica (Braker & Chazdon 1993), story palm seedlings to increased soil N supply was showed large differences among species in the proportion of increased foliar N concentration resulting in higher potential leaf area damaged, indicating herbivore preferences for photosynthetic capacity and increased leaf area damage by species with particular leaf traits. Leaf quality traits, such as herbivores. We found no growth response or biomass allo- cuticular waxes, high leaf water content, high leaf area per cation shifts with N addition, although growth responses of unit mass, and low lignin content, were suggested as pre- the seedlings may have been offset by increased herbivory ferred leaf traits for herbivores at La Selva (Braker & Chaz- with N addition. Alternatively, other soil nutrients, such as don 1993). In the present study, herbivore damage was P, may have limited the growth response to added N. Palm higher in the N addition treatment, in which plants had species associated with low nutrient, rhyolitic soils main- higher foliar N concentrations at a given leaf toughness, tained relatively high growth rates in control and N addition indicating foliar N content was an important trait determin- treatments compared to species from nutrient-rich sites. Our ing herbivore preference. results suggest that leaf quality and susceptibility to herbi- Growth rates of palm seedlings were indirectly affected by vore attack may influence the distribution of palm species, N addition due to increased herbivory in the N addition treat- and likely additional plant species, along soil gradients in ment. The main response to N addition was increased foliar this lower montane forest. © 2010 The Authors. Journal compilation © 2010 British Ecological Society, Functional Ecology, 24, 1171-1180 1180 K. M. Andersen et al.

Acknowledgements Henderson, A. (2005) A multivariate study of Calyptrogyne (Palmae). System- atic Botany, 30,60-83. We thank Edvelio (Bady) Garcia and Arturo Morris for field assistance, Jona- Henderson, A., Galeano, G. & Bernal, R. (1995) Field guide to the palms of the than Ogrodnik for laboratory assistance, and Tania Brenes-Argued as and Americas. Princeton University Press, Princeton, NJ. Michael McCoy for statistical advice with herbivory data. Carol Augspurger, Hodel, DR. (1992) Chamaedorea palms: the species and their cultivation. Noelle Beckman, Jeff Dawson, Mike Dietz, Laura Schreeg, Edmund Tanner, Allen Press, Lawrence, KS. Hodel, DR. (1997) Two new species of Chamaedorea (Arecaceae). Novon, 1, Joe Wright and two anonymous reviewers provided valuable comments on pre- 35-37. vious versions of the manuscript. This research was supported by an National Hunt, R. (1982) Plant Growth Curves: The Functional Approach to Plant Growth Science Foundation Dissertation Completion Grant, a Smithsonian Institute Analysis. Edward Arnold Limited, London. Predoctoral Fellowship, and funding from the Program in Ecology and Evolu- Kitajima, K. (1996) Ecophysiology of tropical tree seedlings. Tropical Forest tionary Biology of the University of Illinois - Champaign/Urbana to KMA. Plant Ecophysiology (eds S.S. Mulkey, R.L. Chazdon & A.P. Smith), The NITROF project is funded by the Robert Bosch Foundation (Germany). pp. 559-597. Chapman & Hall, New York, NY. Permission to sample in the For tuna Forest Reserve was kindly granted by the Koehler, B., Corre, M.D., Veldkamp, E., Wullart, H. & Wright, S.J. (2009) Smithsonian Tropical Research Institute, the Panamanian National Environ- Immediate and long-term nitrogen oxide emissions from tropical forest soils mental Authority (ANAM), and ENEL Fortuna. exposed to elevated nitrogen input. Global Change Biology, 15, 2049-2066. Lambers, H., Chapin, F.S. Ill & Pons, T.L. (1998) Plant Physiological Ecology. Springer-Verlag, New York. Lawrence, D. (2001) Nitrogen and phosphorus enhance growth and luxury References consumption of four secondary forest tree species in Borneo. Journal of Tropical Ecology, 17, 859-869. Adamek, M., Corre, M.D. & Holscher, D. (2009) Early effect of elevated N Lebauer, D.S. & Treseder, K.K. (2008) Nitrogen limitation of net primary input on above-ground net primary production of a lower montane rain productivity in terrestrial ecosystems is globally distributed. Ecology, 89, forest, Panama. Journal of Tropical Ecology, 25, 637-647. 371-379. Aerts, R., Boot, R.G.A. & van der Aart, P.J.M. (1991) The relation between Moran, N.A. & Hamilton, D.W. (1980) Low nutritive quality as a defense above- and belowground biomass allocation patterns and competitive abil- against herbivores. Journal of Theoretical Biology, 86,247-254. ity. Oecologia, 87, 551-559. Palmiotto, P.A., Davies, S.J., Vogt, K.A., Ashton, M.S., Vogt, D.J. & Ashton, Aerts, R. & Chapin, F.S. Ill (2000) The mineral nutrition of wild plants revis- P.S. (2004) Soil-related habitat specialization in dipterocarp rain forest tree ited: a re-evaluation of processes and patterns. Advances in Ecological species in Borneo. Journal of Ecology, 92, 609-623. Research, 30, 1-67. Pinheiro, J.C. & Bates, D.M. (2000) Mixed-effects Models in S and S-PLUS. Andersen, K.M., Turner, B.L. & Bailing, J.W. (2010) Soil-based habitat parti- Springer, New York. tioning in understory palms in lower montane forests. Journal of Biogeogra- Rasband, W.S. (1997-2008) ImageJ. U.S. National Institutes of Health, Beth- pAy, 37,278-289. esda, Maryland, USA, http://rsb.info.nih.gov/ij/. Baltzer, J.L., Thomas, S.C., Nilus, R. & Burslem, D.F.R.P. (2005) Edaphic R Development Core Team (2008) R: A language and environmental for statistic specialization in tropical trees: physiological correlates and responses to reci- computing. R Foundation for Statistical Computing, Vienna, Austria. procal transplantation. Ecology, 86,3063-3077. Version 2.8.1 edition. Bloom, A.J., Chapin, F.S. Ill & Mooney, H.A. (1985) Resource limitation in Reich, P.B., Walters, M.B. & Ellsworth, D.S. (1997) From tropics to tundra: plants—an economic analogy. Annual Review of Ecology and Systematics, global convergence in plant functioning. Proceedings of the National Acad- 16, 363-392. emy of Sciences of the United States of America, 94, 13730-13734. Braker, E. & Chazdon, R. (1993) Ecological, behavioural and nutritional Soethe, N., Lehmann, J. & Engels, C. (2008) Nutrient availability at different factors influencing use of palms as host plants by a Neotropical forest grass- altitudes in a tropical montane forest in Ecuador. Journal of Tropical Ecol- hopper. Journal of Tropical Ecology, 9, 183-197. ogy, 24, 397-406. Campo, J. & Dirzo, R. (2003) Leaf quality and herbivory responses to soil Sokal, R.R. & Rolf, F.J. (1995) Biometry. W. H. Freeman and Company, New nutrient addition in secondary tropical dry forest of Yucatan, Mexico. Jour- York. nal of Tropical Ecology, 19, 525-530. Tanner, E.V.J., Vitousek, P.M. & Cuevas, E. (1998) Experimental investigation Cavelier, J., Solis, D. & Jaramillo, M.A. (1996) Fog interception in montane of nutrient limitation of forest growth on wet tropical mountains. Ecology, forest across the Central Cordillera of Panama. Journal of Tropical Ecology, 79,10-22. 12,357-369. Tanner, E.V.J., Kapos, V., Freskos, S., Healey, J.R. & Theobald, A.M. (1990) Chapin, F.S. Ill (1980) The mineral nutrition of wild plants. Annual Review of Nitrogen and phosphorus fertilization of Jamaican montane forest trees. Ecology and Systematics, 11,233-260. Journal of Tropical Ecology, 6,231-238. Chapin, F.S. Ill, Autumn, K. & Pugnaire, F. (1993) Evolution of suites of Tilman, D. & Wedin, D. (1991) Plant traits and resource reduction for five traits in response to environmental stress. The American Naturalist, 142, grasses growing on a nitrogen gradient. Ecology, 72, 685-700. S78-S92. Vitousek, P.M. & Farrington, H. (1997) Nutrient limitation and soil develop- Coley, P.D. & Barone, J.A. (1996) Herbivory and plant defenses in tropical ment: experimental test of a biogeochemical theory. Biogeochemistry, 37,63-75. forests. Annual Review of Ecology and Systematics, 27,305-335. Vitousek, P.M. & Howarth, R.W. (1991) N limitation on land and in the sea - Coley, P.D., Bryant, J. & Chapin, F.S. Ill (1985) Resource availability and how can it occur. Biogeochemistry, 13,87-115. plant antiherbivore defense. Science, 230, 895-899. Vitousek, P.M. & Sanford, R.L. (1986) Nutrient cycling in moist tropical forest. Corre, M.D., Veldkamp, E., Arnold, J. & Wright, S.J. (2010) Impact of elevated Annual Review of Ecology and Systematics, 17, 137-167. nitrogen input on soil nitrogen cycling and losses in old-growth lowland and Walker, T.W. & Syers, J.K. (1976) Fate of phosphorus during pedogenesis. montane forests in Panama. Ecology, 91,1715-1729. Geoderma, 15, 1-19. Cuenca, A., Asmussen-Lange, C.B. & Borchsenius, F. (2008) A dated phytog- Walters, M.B. & Reich, P.B. (2000) Seed size, nitrogen supply and growth rate eny of the palm tribe Chamaedoreeae supports Eocene dispersal between affect tree seedling survival in deep shade. Ecology, 81,1887-1901. Africa, North and South America. Molecular Phylogenetics and Evolution, Wang, Y.H. & Augspurger, C. (2004) Dwarf palm and cyclanths strongly 46, 760-775. reduce Neotropical seedling recruitment. Oikos, 107,619-633. Farris-Lopez, K., Denslow, J., Moser, B. & Passmore, H. (2004) Influence of a Warton, D., Wright, I., Falster, D. & Westoby, M. (2006) Bivariate line-fitting common palm, Oenocarpus mapora, on seedling establishment in a tropical methods for allometry. Biological Reviews, 81, 259. moist forest in Panama. Journal of Tropical Ecology, 20,429^438. Wilcke, W., Oelmann, Y., Schmitt, A., Valarezo, C, Zech, W. & Horneier, J. Field, C. & Mooney, H. (1986) The photosynthesis-nitrogen relationship in (2008) Soil properties and tree growth along an altitudinal transect in Ecu- wild plants. On the Economy of Form and Function (ed. T. Givnish), pp. 25- adorian tropical montane forest. Journal of Plant Nutrition and Soil Science, 55. Cambridge University Press, Cambridge, UK. 171,220-230. Fine, P.V.A., Mesones, I. & Coley, P.D. (2004) Herbivores promote habitat Wilson, S.D. & Tilman, D. (1993) Plant competition and resource availability specialization by trees in Amazonian forests. Science, 305, 663-665. in response to disturbance and fertilization. Ecology, 74, 599-611. Fine, P.V.A., Miller, Z., Mesones, I., Irazuzta, S., Appel, H., Stevens, M., Saaksjarvi, I., Schultz, J. & Coley, P.D. (2006) The growth-defense trade-off and habitat specialization by plants in Amazonian forests. Ecology, 87, 150- Received 11 January 2010; accepted 5 May 2010 162. Handling Editor: Matthew Turnhull