Untargeted Metabolomics Reveals Fruit Secondary Metabolites Alter Bat Nutrient Absorption

Abstract

The ecological interaction between fleshy fruits and frugivores is influenced by diverse mixtures of secondary metabolites that naturally occur in the fruit pulp. Although some fruit secondary metabolites have a primary role in defending the pulp against antagonistic frugivores, these metabolites also potentially affect mutualistic interactions. The physiological impact of these secondary metabolites on mutualistic frugivores remains largely unexplored. Using a mutualistic fruit bat (Carollia perspicillata), we showed that ingesting four secondary metabolites commonly found in plant tissues affects bat foraging behavior and induces changes in the fecal metabolome. Our behavioral trials showed that the metabolites tested typically deter bats. Our metabolomic surveys suggest that secondary metabolites alter, either by increasing or decreasing, the absorption of essential macronutrients. These behavioral and physiological effects vary based on the specific identity and concentration of the metabolite tested. Our results also suggest that a portion of the secondary metabolites consumed is excreted by the bat intact or slightly modified. By identifying key shifts in the fecal metabolome of a mutualistic frugivore caused by secondary metabolite consumption, this study improves our understanding of the effects of fruit chemistry on frugivore physiology.

Keywords: Carollia; Chemical Ecology; Detoxification; Fecal Metabolome; Frugivores.

Fig. 1

Effect of four secondary metabolites on the ratio of consumption by bats. The ratio was calculated using the total amount of food eaten (g) by each in 30 min divided by the average of control (unsupplemented) food consumed by the same bat. Each data point represents a preference trial, and the shape and color indicate the secondary metabolite identity. Trials were conducted using three different concentrations (a) 0.1, (b) 2, and (c) 3% dry weight. P-values were obtained through a one-sample Wilcoxon test, where a significant P-value indicates that means are significantly different from 1. Values greater than 1 indicate a preference for the treatment compared to the control, while values lower than 1 indicate a preference for the control. A value of 1 would indicate no preference for either treatment or control and, therefore, no deterrence effect of the metabolite. Black points and error bars represent pseudo-medians and 95% confidence intervals computed using the Wilcoxon test. Comparisons between secondary metabolites tested at a given concentration were conducted using GLMMs (Table S2 and S3)

Fig. 2

Relatively high concentrations of secondary metabolites affect the composition of the fecal metabolome. Non-metric multidimensional scaling (NMDS) plots show the effect of three different concentrations, (a) 0.1, (b) 2, and (c) 3% dry weight of four secondary metabolites on the fecal metabolome composition. Each point represents a fecal sample colored by the secondary metabolites ingested. Ellipses represent 95% confidence intervals around group centroids. Due to the limited number of samples, ellipses could not be calculated for 3% (C) eugenol

Fig. 3

Metabolites identified in the random forest and Boruta analyses between the fecal metabolome of bats that ingested four secondary metabolites (piperine, tannic acid, eugenol, and phytol) at 2%. The effect of metabolite identity was calculated using GLMMs. Parameters predicted by the GLMMs are summarized in Table S6. Best match in the NIST library, IUPAC names: A: trimethylsilyl 3-methyl-2-(trimethylsilylamino)-3-trimethylsilylsulfanylbutanoate; B: diethoxy-methyl-octadecylsilane; C: N-(2,6-diethylphenyl)-1,1,1-trifluoromethanesulfonamide; D: trimethylsilyl 2-oxo-3-trimethylsilylpropanoate; E: trimethylsilyl (Z)-octadec-9-enoate. Significance levels are denoted by asterisks, (P < 0.001 = ‘***’, P < 0.01 = ‘**’, P < 0.05 = ‘*’), compared to the control

Fig. 4

Metabolites identified in the random forest and Boruta analyses between the fecal metabolome of bats that ingested four secondary metabolites (piperine, tannic acid, eugenol, and phytol) at 3%. The effect of metabolite identity was calculated using GLMMs. Parameters predicted by the GLMMs are summarized in Table S6. Best match in the NIST library, IUPAC names: A: trimethyl(trimethylsilyloxy)silane; B: trimethyl-[(3Z)-9-trimethylsilyloxyundeca-3,10-dien-6-yn-5-yl]oxysilane; C: hexadec-1-yne; D: (NE)-N-[1-(2,5-dimethoxyphenyl)propan-2-ylidene]hydroxylamine; E: 6-[6-amino-8-(2-aminoethylamino)purin-9-yl]-2-hydroxy-2-oxo-4a,6,7,7a-tetrahydro-4 H-furo[3,2-d][1,3,2]dioxaphosphinin-7-ol; F: N-(2,6-diethylphenyl)-1,1,1-trifluoromethanesulfonamide; G: trimethylsilyl hexadecanoate; H: methyl (9E,15E)-octadeca-9,15-dienoate; I: methyl (E)-dodec-9-enoate; J: tert-butyl-hexadecoxy-dimethylsilane. Significance levels are denoted by asterisks, (P < 0.001 = ‘***’, P < 0.01 = ‘**’, P < 0.05 = ‘*’), compared to the control

Fig. 5

Tentative subclasses of excreted metabolites identified variables distinguishing the fecal metabolome between treatments (piperine, tannic acid, eugenol, and phytol) at 2% and 3% concentrations. Mean Decrease Accuracy is a metric of the importance of each variable in classifying data, showing the reduction in model accuracy when excluding each variable. Higher values indicate more significant importance for accurate classification