The best fitting model is presented. The SAS Calis (Covariance Analysis of Linear Structural Equations) procedure was utilized to determine the fit of the models. The Calis procedure uses normal theory maximum likelihood procedures to estimate fit, and parameter vectors are estimated iteratively with a nonlinear optimization 
algorithm todoi:10.1371/journal.pone.0047554.t65 and 75 . All necessary permits and permissions were obtained for the described field studies.Chemical AnalysesWe collected and air-dried leaves for saponin and flavan quantification. In preparation for chemical extraction, leaf samples were dried overnight in an oven at low temperature and ground to a coarse powder. We utilized a new isolation and quantification procedure for saponin content [61]. One hundred milligrams of dry leaf powder were measured into a centrifuge tube and compounds were extracted from the leaf material in 30 ml of 80 ethanol with stirring. The samples were then centrifuged and the extracted compounds plus solvent were separated from the leaf material and dried. The process was repeated to extract any remaining compounds from the plant material. The dried samples were then dissolved in 15 ml methanol and defatted by shaking the solution with hexanes. The hexane layer was pipetted-off and the process was repeated. The defatted methanol layer was dried, and the samples were dissolved in 20 ml water. This solution was centrifuged to separate any remaining leaf material from the dissolved sample. C-18 SepPak cartridges (Waters Corp., Massachusetts, USA) were then preconditioned with 15 ml acetone followed by 15 mlVariation in Costs of Terpenoids and PhenolicsFigure 2. Means (SE) of photosynthesis, dark respiration, biomass, and 64849-39-4 supplier carbon-based metabolites. Values are from Pentaclethra macroloba seedlings grown in shade (a, c) or full sunlight (b, d) with and without competition. doi:10.1371/journal.pone.0047554.goptimize a goodness of fit function. Chi-square values are calculated for the maximum likelihood goodness of fit to determine the fit of the models. P-values greater than 0.05 indicate a good fit of the data to the model. We accepted the model with the highest P-value as the best description of the relationships between variables. All analyses were done with SAS 9.1 (SAS Institute Inc. 2003).ResultsPhotosynthesis and biomass 
of the shade-grown plants were highest with competition, but dark respiration was slightly higher without 101043-37-2 biological activity competition (Table 1; Figure 2a). The fertilizer treatment did not have an effect on the response variables. Neither photosynthesis, respiration, nor biomass of plants grown in the sun changed with the competition or fertilizer treatments (Table 1; Figure 2b). The interaction between fertilizer and competition was significant for shade-plant metabolite production (Table 1). Sugars were higher in plants with competition and low or intermediate levels of fertilizer. They were lowest also with competition but with high levels of fertilizer. The two groups of secondary metabolites responded in the opposite direction to increased nitrogen. Flavans were highest in low nitrogen conditions (no competition, low fertilizer) and lowest in conditions of high nitrogen (competitionand high fertilizer levels). In contrast, saponins were highest with competition and high fertilizer levels and lowest without competition and with low fertilizer levels (Figure 2c). Metabolites of sun-grown plants were affected by competition such that s.The best fitting model is presented. The SAS Calis (Covariance Analysis of Linear Structural Equations) procedure was utilized to determine the fit of the models. The Calis procedure uses normal theory maximum likelihood procedures to estimate fit, and parameter vectors are estimated iteratively with a nonlinear optimization algorithm todoi:10.1371/journal.pone.0047554.t65 and 75 . All necessary permits and permissions were obtained for the described field studies.Chemical AnalysesWe collected and air-dried leaves for saponin and flavan quantification. In preparation for chemical extraction, leaf samples were dried overnight in an oven at low temperature and ground to a coarse powder. We utilized a new isolation and quantification procedure for saponin content [61]. One hundred milligrams of dry leaf powder were measured into a centrifuge tube and compounds were extracted from the leaf material in 30 ml of 80 ethanol with stirring. The samples were then centrifuged and the extracted compounds plus solvent were separated from the leaf material and dried. The process was repeated to extract any remaining compounds from the plant material. The dried samples were then dissolved in 15 ml methanol and defatted by shaking the solution with hexanes. The hexane layer was pipetted-off and the process was repeated. The defatted methanol layer was dried, and the samples were dissolved in 20 ml water. This solution was centrifuged to separate any remaining leaf material from the dissolved sample. C-18 SepPak cartridges (Waters Corp., Massachusetts, USA) were then preconditioned with 15 ml acetone followed by 15 mlVariation in Costs of Terpenoids and PhenolicsFigure 2. Means (SE) of photosynthesis, dark respiration, biomass, and carbon-based metabolites. Values are from Pentaclethra macroloba seedlings grown in shade (a, c) or full sunlight (b, d) with and without competition. doi:10.1371/journal.pone.0047554.goptimize a goodness of fit function. Chi-square values are calculated for the maximum likelihood goodness of fit to determine the fit of the models. P-values greater than 0.05 indicate a good fit of the data to the model. We accepted the model with the highest P-value as the best description of the relationships between variables. All analyses were done with SAS 9.1 (SAS Institute Inc. 2003).ResultsPhotosynthesis and biomass of the shade-grown plants were highest with competition, but dark respiration was slightly higher without competition (Table 1; Figure 2a). The fertilizer treatment did not have an effect on the response variables. Neither photosynthesis, respiration, nor biomass of plants grown in the sun changed with the competition or fertilizer treatments (Table 1; Figure 2b). The interaction between fertilizer and competition was significant for shade-plant metabolite production (Table 1). Sugars were higher in plants with competition and low or intermediate levels of fertilizer. They were lowest also with competition but with high levels of fertilizer. The two groups of secondary metabolites responded in the opposite direction to increased nitrogen. Flavans were highest in low nitrogen conditions (no competition, low fertilizer) and lowest in conditions of high nitrogen (competitionand high fertilizer levels). In contrast, saponins were highest with competition and high fertilizer levels and lowest without competition and with low fertilizer levels (Figure 2c). Metabolites of sun-grown plants were affected by competition such that s.
