Due to a loss of samples, our K uptake and DOC datasets remain incomplete and will therefore not be discussed in the results; the extent of the recovered K uptake and DOC data is presented in Table 3 along with ammonium, nitrate, and phosphate tree replicates. A complementary mixed effects model using individual trees as an error term, in addition to the model formula above, was used to assess the effects of fertilization on nutrient uptake. For this method outliers with a Z-score of 3.5 or greater were removed from each treatment group and the remaining data were analyzed as described for a standard ANOVA except with an error term. This analysis precluded the use of p-scores as the package does not give them. Post-hoc analysis, as described below, is similarly non-applicable to this examination, therefore discussion of differences in results between the mixed effects and ANOVA models will remain limited. A Tukey Honestly Significant Differences post-hoc test was used as a post-hoc examination of the ANOVA results. Tukey’s HSD tests all pairwise comparisons, while controlling the probability of making one or more Type I errors, producing a conservative measure of significant differences between treatment combinations. Tukey’s HSD is used for post-hoc interpretation of significant ANOVA results in order to determine what differences between treatment groups are resulting in ANOVA significance. ANOVA/ANCOVA and Tukey’s post-hoc are both suitable for datasets with unequal variance in sample sizes. Our data was slightly unbalanced in the pseudoreplication across individual trees resulting in slight differences in plot and pseudoreplicate measurements. Tree replicates, and root pseudoreplicates, are shown by the treatment groups in Table 4; with slightly fewer pseudoreplicates evidenced at the 10X concentration in the N- and K-only addition plots.

Tree and plot replication, along with mean uptake rates and standard deviation for each nutrient uptake by treatment,roll bench are shown in Table 5. All analyses conducted in R .Both the full ANCOVA and simplified ANOVA models showed significant effects of an N*K interaction on the uptake rates of ammonium, nitrate, and phosphate. Significant effects on the uptake rates of each nutrient, and DOC exudation, are shown in Table 5 with the results from each the model tested, and an asterisk noting incomplete datasets. Ammonium uptake was significantly affected by an N*K interaction in both models , with model results improving noticeably with the removal of DBH as a covariate. Nitrate uptake was very significantly affected by and N*K interaction in both models , with a marginally significant effect of P addition on uptake in the full model. Phosphate uptake was significantly affected at the P<0.05 level by an N*K interaction in each model. Potassium uptake was significantly affected by an N addition, and N*K interaction effect, while DOC was significantly affected by a P*K interaction; however, as these results are based on incomplete datasets, their significance will be reported in Table 6 but not mentioned in later discussion. P-values for individual predictor variables in the ANCOVA and ANOVA models, along with their model statements, are given in Table 7, Table 8, and Table 9, for ammonium, nitrate, and phosphate respectively. P-values of individual model predictors for K uptake and DOC exudation are presented in Tables 10 and 11 respectively. DBH had no effect on nutrient uptake and model results improved when it was removed as a covariate suggesting that it is not a strong determinant of nutrient uptake; and therefor that seedling studies may be accurate estimations of mature tree uptake rates for Tetragastris panamensis and other tropical species. The DBH of the sampled trees ranged from 84-564 mm, with a mean of 186.36 and a standard deviation of 108.04. A one-sided t-test of mean uptake rates of NH4+ and NO3-, regardless of treatment, showed no significant differences at p=0.05 between the two forms of N. A Tukey Honestly Significant Differences test was used as a conservative post hoc test to explore what differences between treatment groups were potentially responsible for the significant interaction.

The Tukey HSD post-hoc was applied using the simplified model formula using only N and K as predictors of nutrient uptake; this compared the means of all combinations of N and K fertilized plots to each other and corrected for multiple comparisons in order to determine which treatment combinations were significantly different. Treatment combinations are expressed as the absence or presence of either N or K, with each combination of treatments on the left side of the dash being compared to those on the right. The adjusted p-values for Tukey’s HSD are given for each nutrient in Table 12 with an asterisk denoting a significant difference in nutrient uptake between treatment groups. A Tukey HSD plots indicate significant differences between groups as a sideways boxplot where no overlap occurs between the whiskers and the zero line – the 95% confidence interval that a group’s true mean is significantly different from the zero line, the mean of all groups. The relative difference in uptake rates between plots can be assessed from plots of Tukey’s HSD; ammonium uptake was significantly lower in N and K fertilized plots versus uptake in N fertilized as indicated by the non-overlapping boxplot and center line in Figure 1. No other significant differences between individual treatment groups was observed for ammonium, however boxplots of ammonium uptake rates with illustrate the antagonistic relationship between N and K on uptake . Nitrate uptake showed a significant increase in N-only fertilized plots relative to control, and a decrease in N and K fertilized plots compared to N-only fertilized plots . This shows a negative interaction between the presence of N and K on nitrate uptake with single nutrient fertilization producing greater increases in uptake than fertilization with both nutrients . The same pattern of seemingly greater uptake rates with single nutrient addition is observed for phosphate in Figure 5. Phosphate uptake showed a nearly significant decrease in uptake in N and K fertilized plots compared to N-only fertilized plots . There were no significant differences observed between treatments in the Tukey’s HSD for phosphate however the overall N * K interaction effect was significant in the ANOVA model. Although non-significant, a consistent pattern was observed in the differences between treatment groups for all nutrients : notably that the addition of N and K in combination reduces uptake rates compared to N addition alone.

Additionally, N appears to consistently,commercial greenhouse supplies although not-significantly, increase nutrient uptake relative to control plots. This suggests an negative relationship between N and K where addition of either nutrient singly is likely to increase nutrient uptake rates while an interaction between N and K depresses uptake of all nutrients. A linear mixed effects model of nutrient uptake showed that solution concentration, along with N and P, were most likely to affect nutrient uptake rates instead of the N*K interaction. ANOVA F-values for the mixed linear effects model of phosphate uptake are shown in Table 13 and suggest that concentration and N were the dominant influences on uptake, while t-values indicate K interactions with N or P may be qualitatively different than other fertilization effects. ANOVA results for NH4+ uptake show that only concentration produced a large effect on NH4+ uptake while no other nutrient appeared to strongly influence uptake . In contrast, NO3- uptake appeared to be much more sensitive to the effects of N and P, and their interactions, than NH4+ . T-values for K interactions are also evidenced for NH4+ and NO3- uptake suggesting that K is producing a universal effect on the uptake of macro-nutrients, however the dynamics and drivers of this potential effect are uncertain.Nutrient solution-depletion methods have examined the uptake of N, as NH4+ and NO3-, more extensively than other nutrients, however, few studies have directly addressed the effects of altered relative N availability on nutrient uptake. Uptake rates of mature in situ spruce and maple trees both showed significantly greater uptake of NH4+ than NO3- in all seasons; with the early growing season showing the lowest uptake rates for NH4+ and NO3-, 0-6 umol/g dw*hr and -5 – 0.5 umol/g dw*hr respectively, while uptake rates in the late fall showed an increase in NH4+ uptake up to 25 umol/g dw*hr . Mature beech trees evidenced uptake rates of 0.25-1.5 umol/g fresh weight * hr in two studies that did not record concurrent NO3- uptake , while mature loblolly pine showed NH4+ uptake rates of 1-4 umol/g dw*hr which were significantly higher than those of NO3- . Seedlings of two maple species exposed to the same solution-depletion method showed uptake rates of NH4+ between 2 and 10 umol/g dw*hr, for external NH4+ concentrations <100 umol/L, while NO3- uptake was consistently lower . While the preference for NH4+ in midlatitude species has been widely observed only one study of NH4+ uptake rates across an N deposition gradient provides evidence for the effect of N availability on uptake rates. Using a labeled isotope modification of the depletion method uptake rates of beech, spruce, pine and birch across an N deposition gradient showed greater N uptake in sites with higher N deposition .

A compilation of reported uptake ranges is shown in Table 16 for comparison with our results. Our data show similar uptake rates to those of NH4+ in mid-latitude species, with the indication that N uptake may be stimulated with N addition, however, our data showed only marginally significant differences in the uptake rates of NH4+ versus those of NO3- for mature Tetragastris panamensis. Demand for N and P have also been inferred by measurements of nutrient concentrations in plant tissues and changes in root biomass. Tropical tree seedlings grown under experimental water and nutrient regimes showed leaf N:C ratios significantly increased in response to low soil water content and N fertilizer addition . In contrast, leaf P:C ratios were only moderately affected by nutrient availability and water content, suggesting greater sensitivity of nutrient uptake to N availability . In the same factorial fertilization experiment this study was conducted in N addition increased fine root biomass in the top five cm .Tissue nutrient concentrations in a tropical pioneer species showed that the transpiration rate of the plant provided a significant control on uptake and varied with N availability, not P . Our results suggest that N availability plays a significant role in the uptake of nutrients, by its interaction with K as well as the seemingly increased uptake rates of all nutrients relative to control and NK plots when added alone. Uptake of NO3-appeared to be more significantly affected by nutrient availability than that of NH4+ , following observations of mid-latitude forests that demand each form of N may be sensitive to soil N conditions. Our data support the interpretation that these forests are at least partially limited by N and that the availability of N may influence nutrient acquisition via effects on transpirational control of nutrient uptake. P uptake studies in the literature are rare for tropical and non-crop species thus interpretation of our results relies on context by other ecological studies inferring nutrient demand and cycling. In our study site FRB increased with P addition but decreased with simultaneous P and N addition . An N by P factorial fertilization experiment in a secondary tropical forest in China found that P addition increased P concentrations throughout the plant, although P concentrations decreased when N and P were added simultaneously . A meta-analysis of root phosphatase production in relation to soil fertility found that N fertilization increased phosphatase exudation, promoting the uptake of P, while P addition suppressed phosphatase production.These studies suggest that addition of P induces a greater demand for nutrients, however specifically not for additional P, an effect which decreases when N is abundant. With a simultaneous increase in root biomass and decrease in phosphatase production, the addition of P appears to promote nutrient acquisition by altering C investment in root tissue instead of phosphatase C-exudation in order to increase long-term passive nutrient acquisition. Our study does not allow a determination of changes in total nutrient uptake due to changes in root biomass, however, our results show that P did not affect the active uptake rates of any nutrients, despite P uptake being significantly affected by the availability of N and K. A pattern of N limitation underlying potential P limitation is also indicated by studies of above ground nutrient and C cycling dynamics.