Oecologia (2006) 149: 91–100 DOI 10.1007/s00442-006-0423-2
PL AN T AN I M A L I N T ER A C T IO N S
Tania Brenes-Arguedas · Matthew W. Horton Phyllis D. Coley · John Lokvam · Rachel A. Waddell Beatrice E. Meizoso-O’Meara · Thomas A. Kursar
Contrasting mechanisms of secondary metabolite accumulation during leaf development in two tropical tree species with different leaf expansion strategies Received: 4 November 2005 / Accepted: 20 March 2006 / Published online: 5 May 2006 © Springer-Verlag 2006
Abstract Young leaves of most species experience remarkably higher herbivore attack rates than mature leaves. Considerable theoretical eVort has focused on predicting optimal defense and tradeoVs in defense allocation during leaf expansion. Among others, allocation to secondary chemistry may be dependent on growth constraints. We studied Xavanoid production during leaf development in two species of Inga (Fabaceae: Mimosoideae) with diVerent expansion strategies: Inga goldmanii, a species with slowly expanding young leaves, and Inga umbellifera, a species with fast-expanding young leaves. In these two species, the most abundant and toxic class of defensive compounds is Xavanoids (which include tannins). We measured their concentration by leaf dry weight, their total content per leaf, their HPLC chemical proWle and their toxicity to a generalist herbivore at diVerent expansion levels. Although in both species the Xavanoid concentration decreased with increasing leaf expansion, that decrease was twice as pronounced for I. umbellifera as it was for I. goldmanii. I. umbellifera leaves produced Xavanoids only during the Wrst half of their development while I. goldmanii leaves continued production throughout. The
Communicated by Colin Orians Electronic Supplementary Material Supplementary material is available for this article at http://dx.doi.org/10.1007/s00442-0060423-2 and is accessible for authorized users. T. Brenes-Arguedas · M. W. Horton · P. D. Coley · J. Lokvam R. A. Waddell · B. E. Meizoso-O'Meara · T. A. Kursar Department of Biology, University of Utah, Salt Lake City, UT, USA M. W. Horton Department of Ecology and Evolution, University of Chicago, Chicago, IL, USA T. Brenes-Arguedas (&) 257 S 1400 E, Salt Lake City, UT 84112, USA E-mail:
[email protected] Tel.: +1-801-5817086 Fax: +1-801-5814668
changes in Xavanoid HPLC proWles and toxicity were also more dramatic for I. umbellifera, which had diVerent Xavanoids in young than in mature leaves. Relative to I. umbellifera, I. goldmanii showed smaller changes in both Xavanoid composition and toxicity in the transition from young to mature leaves. These results indicate that, even though young leaves suVer higher rates of attack and are predicted to have better chemical defenses than mature leaves, growth constraints may modulate defense allocation and thus, evolution of defense strategies. Keywords Condensed tannins · Growth-diVerentiation hypothesis · Inga · Plant defense · Optimal defense
Introduction Young leaves of most species experience considerably higher rates of herbivory than mature leaves. Leaves from shade-tolerant tropical trees can suVer up to 70% of their lifetime herbivory during the short period of leaf expansion (Coley and Aide 1991). This is especially impressive considering that the average lifespan for tropical leaves is 2–4 years (Coley 1988). Young leaves are preferred by herbivores because they are tender and nutritious (Feeny 1970; Coley 1983), two traits that are an inevitable consequence of growth. Young leaves are tender because the production of a ligniWed cell wall would prevent cell expansion and they are nutritious because cell expansion and diVerentiation require high nitrogen and water content. Because young leaves are more vulnerable to herbivores and pathogens than mature leaves (Coley 1983), the optimal defense hypothesis (Rhoades 1979) suggests that plants would beneWt by allocating more resources to defend those organs. Indeed, young leaves use a variety of developmental, phenological, biotic, and chemical strategies that function as defenses (Coley and Kursar 1996) and that are exclusive to the expansion stage. However, leaf development is a complex process that
92
probably places both qualitative and quantitative constraints on defense expression. In particular, allocation to secondary chemistry during leaf development may depend on allocation to other types of defenses (Coley and Kursar 1996), on growth constraints (Loomis 1932; Orians and Janzen 1974; Herms and Mattson 1992) or on biochemical constraints (Jones and Hartley 1999). Such constraints during leaf development may be the reason for the remarkable variation in defense among tropical rainforest species. Damage to young leaves varies among tropical species by a factor of 5 (Kursar and Coley 2003) and the rates of damage correlate, among other things, with the expansion rate of the leaves. Tropical species vary in the rates of expansion of their leaves by an order of magnitude (Coley and Kursar 1996) and fast-expanding species suVer higher herbivory rates than slowly expanding species (Kursar and Coley 1992b). Leaf expansion rates also correlate with other traits such as delayed greening (Kursar and Coley 1992a; but see Numata et al. 2004), nitrogen content (Kursar and Coley 1991) and synchrony in leaf production (Coley and Kursar 1996). It has been hypothesized that this diversity in defensive traits is the result of trade-oVs between growth and defense (Kursar and Coley 2003; Coley et al. 2005). Some species rely on escaping damage to young leaves through synchronous leaf production and rapid rates of leaf expansion at the expense of chemical defenses (“escape” syndrome), while other species rely on chemically defending themselves against damage at the expense of fast expansion (“defense” syndrome). To test for trade-oVs between growth and defense we must start by using an appropriate measure of plant allocation to defense. For practical purposes, in this study we will focus on biosynthesis of secondary chemicals. Traditionally, studies on young leaves have measured changes in concentrations (e.g., Coley 1983; Turner 1995; Read et al. 2003). Concentrations are useful to describe defense from the ecological point of view, as they describe the dosage relevant to herbivores. However, because leaf biomass also changes throughout development, changes in concentration alone are not an adequate indicator of changes in biosynthesis (Koricheva 1999). A leaf might synthesize secondary chemicals continuously, such that the metabolites accumulate, but unless this increase compensates for the increase in the leaf weight due to expansion and ligniWcation, the actual concentration will decrease. Additionally, leaves undergo changes in the composition of secondary metabolites during development. Changes in composition during leaf expansion can be an adaptive response to prevent the evolution of herbivore resistance (Ruusila et al. 2005), or can be the result of biosynthetic pathways and precursor–product relationships (Salminen et al. 2001, 2004). To better study changes in allocation to defense during leaf development we must integrate changes in metabolites' total content per leaf, composition, concentration, biomass, and bioactivity. Changes in content and composition give information regarding biosynthesis, and changes in concentration and bioactivity link those to defense quality.
We suggest that the inherent diVerences in leaf expansion rates among tropical species should inXuence the accumulation of secondary chemicals during leaf development. The accumulation of secondary chemicals should be lower in fast-expanding species. These species allocate large amounts of resources to growth and should therefore limit the resources allocated to defense. The secondary metabolites in these species probably accumulate at a slow rate producing a decrease in concentration during leaf development. In contrast, slowly expanding species do not require the allocation of as many resources to growth per unit time and can allocate more to secondary metabolite biosynthesis. These species are expected to synthesize more secondary metabolites throughout development such that their concentration would be maintained or increased during expansion. These diVerences in the patterns of allocation, together with potential diVerences in composition, should inXuence concentrations and have distinct implications for plant–herbivore interactions. In this study we compared production of chemical defenses during leaf development in two closely related Neotropical rainforest trees with diVerent expansion strategies in the genus Inga (Fabaceae: Mimosoideae). Coley et al. (2005) described how these two species show alternate defense syndromes: Inga umbellifera is a fast-expanding “escape” species and I. goldmanii, a slowly expanding “defense” species. Here we test explicitly if this diVerence in defense syndromes inXuences the accumulation of secondary metabolites during leaf expansion, as would be expected by the growth-diVerentiation hypothesis. Most of the secondary metabolites with anti-herbivore activity in these two species are procyanidins or condensed tannins (polymers of Xavan-3-ols). Because both species contain both monomers and polymers, we will collectively refer to them as “Xavanoids”. Besides secondary metabolites, I. umbellifera and I. goldmanii express other defensive traits. All Inga species have extra-Xoral nectaries that attract ants only during the expansion stage (Koptur 1984). I. goldmanii has hairs, which are denser in expanding leaves, and I. umbellifera has delayed chloroplast development and high synchrony in leaf Xushing (Coley et al. 2005). While these other defenses are also ecologically relevant and obviously change during leaf expansion, here we will focus only on the secondary metabolites. We asked whether I. umbellifera and I. goldmanii have diVerent patterns of accumulation of Xavanoids during leaf expansion and if the changes occurring during leaf expansion have ecological implications. More speciWcally we measured changes throughout expansion in the concentration, in the total content of the leaf, in the composition and in the toxicity of the Xavanoids.
Materials and methods Inga umbellifera and I. goldmanii are two shade-tolerant tree species in the Fabaceae (subfamily: Mimosoideae). Leaf samples of these two species were collected on Barro
93
Colorado Island (BCI; 9°09⬘N, 79°51⬘W) in the Republic of Panama, a Weld site administered by the Smithsonian Tropical Research Institute. The forest is moist lowland and receives 2,600 mm of rain during an 8-month rainy season (Croat 1978; Leigh et al. 1996). Both study species are abundant and co-occur across the island. We collected leaves that were as undamaged as possible from Wve I. goldmanii and six I. umbellifera understory saplings. From each individual sapling we collected Wve leaves: three young and two mature. The three young leaves were selected at diVerent stages of expansion: early, intermediate, and late, at approximately 10, 50, and 90% of the average maximum size, respectively. Mature leaves were estimated to be 3 months to 1 year old. The fresh leaves were divided in half by separating opposite leaXets from the rachis. The rachis was discarded. Each half leaf was weighed individually, then one half was macerated in 95% EtOH using a Polytron (Brinkman Instruments, Westbury, N.Y.), and the other half was dried in silica gel and weighed again. Ethanol suspensions were stored at ¡80°C until they were shipped to our laboratory at the University of Utah for extraction and analyses. We used the fresh weight of the macerated leaXets (FWm) and the fresh (FWd) and dry weights of the intact dried leaXets (DWd), to calculate the leaf water content=(FWd¡DWd)/FWd; percentage of dry weight (DW%)=DWd/FWd; the expected dry weight of the macerated leaves (DWm)=FWm£DW%, and the total leaf dry weight (DWt)=(DWm+DWd). We also measured the area of the intact dry leaXets (half leaf) using NIH ImageJ (http:/ /www.rsb.info.nih.gov/ij/). The total leaf area was calculated multiplying leaXets area by 2, assuming the leaf was symmetrical. Finally, we estimated the number of expansion days using time versus area curves for each species.
A 10-l sample was analyzed by HPLC (Hitachi LaChrom Elite system with L-2200 autosampler, L-2130 pump, L-2300 oven and L-2450 diode array detector, Hitachi High Technologies, San Jose, Calif.) using a C-18 column (Omnisphere 5 m, 250 mm£4.6 mm ID; Varian, Palo Alto, Calif.) at 25°C oven temperature. The elution conditions were: eluents, MeOH:H2O; Xow rate 1 ml min¡1 with a linear gradient from 30 to 70% MeOH in 30 min and constant 70% MeOH for 5 min. Peaks observed at 280 nm were integrated and numbered sequentially. To characterize the Xavanoid proWle of I. goldmanii, ca. 2 mg (§0.2 mg) of the dried Xavanoid extract from each leaf sample was acetylated in 0.5 ml of 1:1 acetic anhydride/pyridine mixture to maximize peak resolution. The mixture was left at ambient temperature for 24 h until the solid was dissolved. A 10-l sample was analyzed by HPLC using the same conditions as above except for the elution gradient (linear gradient 60–90% MeOH in 20 min followed by a 3-min wash at 100% MeOH). All peaks observed at 270 nm were integrated and numbered sequentially. To analyze the Xavanoid proWles we calculated the concentration of each compound (or peak) in the leaf=(area)£(calibration constant)£(gFm/gDWm) and the compound total content per leaf=(concentration of compound)£(gDWt). The calibration constant was the average of (sum of areas)¡1 for all early expansion leaves in I. umbellifera, and for all leaves in I. goldmanii. This is a valid approximation to metabolite concentration because we injected puriWed Xavanoids.
Extractions The leaves macerated and suspended in EtOH were Wltered through Whatman Wlter paper (no. 1) and reextracted with 50% EtOH. The Wltrate was concentrated on a rotary evaporator at 30°C and resolubilized in water for fractionation. The water-suspended Wltrate was Wrst defatted by water-dichloromethane liquid–liquid extraction. The defatted water suspension was then placed in a 2-ml octadecyl silica reverse phase (ODS) column where the extract was fractionated between water and 90% MeOH. The water fraction contained organic acids, sugars, and amino acids, which in previous studies (Coley et al. 2005) have not shown any bioactivity. The 90% MeOH fraction contained the Xavanoids. These were dried and weighed. From the Xavanoid weight in the macerated leaves (Fm), we calculated the concentration of Xavanoids in the leaf=(gFm)/(gDWm), and the total content of Xavanoids per leaf=(concentration)£(gDWt).
All statistical analyses were done using SAS software version 8.1. DiVerences in Xavanoid concentration and content among expansion stages and between species were tested using a linear mixed-eVects design: species(expansion£sapling) (SAS proc mixed). Sapling was a random eVect, because we collected one complete expansion sequence per sapling. Within-species changes during leaf expansion were evaluated using a randomblock design: expansion£sapling for each species independently (SAS proc glm). Sapling was the blocking factor. The secondary compound proWle was analyzed by measuring the changes in content per leaf of the multiple peaks. Because each peak (individual Xavanoid) is probably biosynthetically related to other peaks, they cannot be analyzed independently of each other. We used principal component analysis (PCA) (SAS, proc princomp) to Wnd groups of peaks that covaried during expansion and thus, could be analyzed together.
Flavanoid composition
Toxicity bioassays
To characterize the Xavanoid proWle of I. umbellifera, ca. 2 mg (§0.2 mg) of dried I. umbellifera Xavanoid fraction of each sample was re-dissolved in 1 ml of 70% MeOH.
The plant material used for the feeding trials consisted of young (5–80% of full expansion) and mature leaves of the two study species collected from many understory
Statistical analyses
94
trees throughout BCI. The fresh leaves were macerated Wrst in a Waring blender and then a Polytron in about 4 ml of 95% EtOH per gram fresh weight of leaf. Macerated leaves were processed and assayed in our laboratories at the University of Utah. We did bioassays on crude extracts and on puriWed Xavanoids. Crude extracts were the combination of successive extractions of the insoluble leaf materials with 80% EtOH, 70% acetone, dichloromethane and water. The Xavanoids were the dried 70% MeOH fraction of the ODS fractionation described above. Toxicity was tested using feeding trials on the larvae of Heliothis virescens (Lepidoptera: Noctuidae), a generalist herbivore of tropical origin. We measured toxicity as growth inhibition 50 (GI50): the concentration at which larvae grow 50% as large as larvae reared on control diets. GI50 was calculated by measuring larval growth after 8 days of feeding on an artiWcial agar diet with diVerent concentrations of plant extracts. Diet preparation and assay methods are described in detail in Coley et al. (2005). The larval growth relative to the control (GRC) is a function of the extract concentration (C) and approximates a sigmoidal dose-response curve of b ¡1 the form: GRC={k [1+(C GI¡1 50 ) ] }, where k (upper limit), GI50 (inXection point) and slope (b) are the parameters to be calculated. The model was Wt to the data using weighted non-linear regression (SAS, procnlin), where the weight was the inverse of the SE of the GRC. To statistically test the diVerences between pairs of GI50s we used the SEs of the estimates to calculate a t-value for the diVerence between two parameter estimates.
Results Inga goldmanii leaves take twice as long to expand as those of I. umbellifera, and at maturity they are 5 times as big (Table 1). Water content increased slightly during expansion and decreased again at the last stage of matu-
ration for both species, probably a consequence of ligniWcation and toughening. Flavanoid extracts The concentration of extracted Xavanoids averaged for all expansion stages was the same for both species (I. umbellifera, 20§2%; I. goldmanii, 22§2%; mixed-eVects design, species eVect: F=0.31, P=0.59), but it decreased signiWcantly with increasing leaf expansion (expansion eVect: F=19.27, P
