Field methods. Tropical understory plant communities were studied
across four subtropical humid Atlantic Rainforest remnants in
southeastern Brazil. Twenty-one plots measuring 10 m in diameter
were established at least 10 m from any forest edge, including
trails and streams. Plots were centered on a randomly selected Piper
individual, and all Piper individuals within the plots were marked
and identified to species or morphospecies based on leaf and stem
morphology. All leaves within reach were exhaustively searched for
herbivores. Piper hosts many insect herbivores, including Eois
(Geometridae) caterpillars that specialize on Piper but may feed on
multiple species within the genus, Curculionidae and Chrysomelidae
that specialize on the genus, other Lepidoptera that may be
specialists (e.g., Memphis (Nymphalidae) or generalists (e.g.,
Erebidae) and Orthoptera generalists. Percent herbivory was visually
determined on all Piper individuals, and specialist and generalist
damage were separated as accurately as possible. In general, damage
originating at leaf margins was classified as generalist herbivory;
damage in the center of leaves characteristic of specialist
chrysomelid and curculionid beetles and characteristic ‘windows’
created by Eois were categorized as specialist herbivory. The number
of leaves on each plant was recorded, and ten leaves per species per
plot were measured to determine the leaf area of each species in a
plot. Non-Piper morphospecies with leaves between 50 cm and 2 m
aboveground were counted in each plot to determine understory
richness; morphospecies were differentiated based on leaf and stem
morphology. Canopy openness was measured in the center of the plot
using a convex spherical crown densitometer (Forestry Suppliers,
Model A). Three soil samples from the top 20 cm of soil were
randomly collected from each plot and analyzed for N, C, Ca, K, Mg,
and P in the Analytical Center at the University of São Paulo using
standard protocols (http://ca.iq.usp.br/novo/). C and N were
measured as percentages; Ca, K, Mg, Na, and P were quantified as
mg/kg soil. The sum of bases (Ca, K, Mg, and Na) was tested in
analyses.
Chemical analysis. Immature and mature leaves with and without
herbivory were collected from each Piper individual whenever
possible. Leaves were air dried and brought to the lab for chemical
extraction. Dry leaf samples were weighed into 2.0 mL Eppendorf
tubes (200 mg; midveins were not included). Mature leaves with
herbivory were preferentially extracted because they represented the
most common sampling category (n = 433). However, many samples were
from mature leaves without herbivory (n = 321). Some individuals
only had immature leaves; there were 17 immature leaf samples with
and 24 samples without herbivory.
Steel ball bearings were added to the Eppendorf tubes, samples were
dipped in liquid N2 and vortexed to grind the leaves. MeOH:H2O (1:1;
0.75 mL) and CHCl3 (0.75 mL) were added to the samples, and they
were vortexed for 5 min. Samples were then centrifuged for 10 min at
10,000 RPM. The aqueous and CHCl3 phases were separated by
pipetting, and samples were re-extracted. The MeOH:H2O layer was
dried under N2, and the CHCl3 was evaporated under a fume hood. The
CHCl3 crude extract was prepared for 1H-NMR analysis, this extract
provided clean spectra containing a broad range of secondary
metabolites.
Chloroform extracts were dissolved in 0.8 mL of methanol-d4
(Cambridge Isotope Laboratories Inc. 0.05% v/v TMS), and 1H-NMR
spectra were acquired (400 MHz; 512 scans) on a Varian 400-NMR
spectrometer. 1H-NMR spectra provide both structural and
compositional chemical information. Spectra were processed using
MestReNova v8.1.2 by aligning stacked spectra using the methanol-d4
solvent peak (CD3, 3.31 ppm), automatic global phasing, Whittaker
baseline correction, total area normalization and subsequent average
sum binning from δ 0.5-12 at δ 0.04 intervals. Solvent and internal
standard regions were excised from spectra (-1-0.5, TMS; 3.1-3.5,
CD3OD; δ 4.5-5.1, D2O) before exporting tabular data for statistical
analysis.
