**4. Discussion**

The results of our study do not support the hypothesis that Neotropical katydid species specialize their diet by host plant. By extracting plant DNA from the digestive systems of katydids, we were able to use DNA barcoding to identify which plant species and/or families were recently consumed. Individual katydids of the same species often had multiple and different plant families in their digestive systems, indicating that katydid species were not specializing on single host plants. In addition, the use of multiple primers sometimes recovered different plant species from the same katydid, providing evidence that katydids would feed on a diversity of plants even at short timescales. Consequently, dietary specialization on a specific host plant is not providing these species with a means of facilitating mate localization in the face of intense predation on signaling males and mate-searching females.

In contrast, the katydid species studied here were consuming a wide variety of plant families (16) and orders (12) comprising over half the orders of plants found on BCI. While many plant families and orders appeared in the katydid diet, some were particularly well-represented. Within the katydid samples that yielded a single identified plant order, 27% of the samples were Laurales and 23% were Fabales. No other plant family or order comprised more than 8% of the samples (Table 1). The abundance of Laurales and Fabales in the diet appears to be consistent with the relative abundance of stems of these plant orders on Barro Colorado Island, although species diversity in several other lineages, such as Gentianales, Myrtales, Rosales, and Malpighiales, is equally high or higher, suggesting that katydids may avoid some lineages of plants in favor of others [47–49]. Furthermore, it should be noted that the katydid specimens were collected primarily in January when certain trees may be in the young leaf stage and hence easier to digest than other species. If this is the case, then our results may in part be dependent on tree phenology. Further sampling at different times of the year in different seasons is certainly warranted.

The dietary composition of katydids provides one-directional information about the height at which katydids are foraging. Because even canopy emergen<sup>t</sup> tree species start as saplings, the presence of canopy species in the katydid diet does not necessarily mean that the species was consumed in the canopy. In contrast, when katydids are eating shrubs and understory trees, it is strong evidence that the species is foraging low in the forest. While sample sizes are small for some katydid species, katydid species that are observed at ground level are in fact consuming understory vegetation (e.g., *Docidocercus gigliotosi* [22]) (Figure 4).

One notable finding was that many plant families were detected in only one year. In part, the year-to-year differences likely reflect the fact that there are many plant families and most were represented with relatively low frequency. However, the fact that some plant families are well-represented in one year and rare or absent in others suggests that there may be times when particular plant families or individual plants are especially palatable. For example, in a given sampling location on a given night, multiple katydids of multiple species had the same plant in their stomach (Table S3), perhaps reflecting a nearby feeding opportunity that attracted multiple species of katydid.

Transient feeding opportunities on particular plants would be consistent with what is known about the palatability and phenology of many tropical plants. Even though leaves may persist for several years, 25–70% of leaf damage occurs in the weeks when leaves are expanding [50,51]. In response to herbivore pressure, tropical plants have evolved a variety of strategies to minimize their window of vulnerability to herbivores [27]. Herbivore evasion strategies can include exceptionally rapid expansion of leaves, delayed greening, and synchronous flushing, strategies that minimize exposure to herbivory by compressing the window when leaves are maximally palatable [52,53].

The possibility of short-term feeding windows on specific plants is also supported by anecdotal field observations during this study, where katydids of multiple species were observed aggregated on a tree a few days before the tree produced substantial and obvious new growth (L. Symes, personal observation).

If katydids do aggregate to exploit transient feeding opportunities, co-localization on food sources might still facilitate mate finding, even for diet generalists. There are several avenues of investigation that could provide information on whether katydids aggregate to feed on plants during leafout. One strategy is to bait traps with plant volatiles such as benzyl nitrile, phenyl acetaldehyde, and/or 2-phenylethanol [54,55] to determine whether these chemical compounds are attractive to katydids, suggesting targeted feeding on vulnerable plants. A second strategy is to deploy long-term acoustic recorders and test for periods of time when a single location on the landscape has an elevated number of katydid calls, reflecting aggregation or one or more katydid species in an area for a short period of time. Understanding whether katydids aggregate on plants that are producing new leaves is important for understanding herbivore pressures on the forest vegetation, patterns of food availability for insect predators, and impacts of habitat patch size on insect populations and foraging effectiveness.

The results of DNA barcoding, when applied to Neotropical forest katydids, helps to define previously unseen connections between plants and herbivores, including connections that occur out of sight in the forest canopy. The generality of the katydid diet is more or less consistent with observations on other species in other locations, but underscores our lack of knowledge of how animals with rare short-range signals find each other in tropical forests. Our observations also sugges<sup>t</sup> several avenues for future research.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/d14020152/s1, Table S1: Primers utilized for DNA barcode amplification; Table S2: List of accessions of DNA barcodes for BCI trees/shrubs/lianas used in BLAST searches; Table S3: List of katydid samples, plant BLAST result, and collection metadata. Figure S1: The phylogenetic distribution of plant species in the diets of katydids on BCI.

**Author Contributions:** Conceptualization, C.M.P. and L.B.S.; methodology, C.M.P. and L.B.S.; formal analysis, C.M.P. and L.B.S.; investigation, C.M.P., N.L.W., S.J.M. and L.B.S.; resources, C.M.P., H.M.t.H., W.J.K. and L.B.S.; data curation, C.M.P. and L.B.S.; writing—original draft preparation, C.M.P. and L.B.S.; writing—review and editing, C.M.P., H.M.t.H., W.J.K. and L.B.S.; visualization, C.M.P. and L.B.S.; supervision, C.M.P. and L.B.S.; project administration, C.M.P. and L.B.S.; funding acquisition, C.M.P., H.M.t.H. and L.B.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** C.M.P. and N.L.W. were funded by Castleton University and the Vermont Biomedical Research Network (VBRN). Research reported in this publication was supported by an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health, gran<sup>t</sup> number P20GM103449. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIGMS or NIH. L.B.S. was supported by the Neukom Institute of Dartmouth College. L.B.S., S.J.M. and H.M.t.H. were supported by funding from the Smithsonian Tropical Research Institute, Dartmouth College, and an Artificial Intelligence for Earth Innovation gran<sup>t</sup> from Microsoft/National Geographic, gran<sup>t</sup> number NG5-57246T-18. W.J.K. was supported by the Smithsonian Institution.

**Institutional Review Board Statement:** Not applicable.

**Data Availability Statement:** Sequence data for chloroplast DNA amplified from gu<sup>t</sup> contents is available in the Supplementary Materials. Reference plant barcodes (accessions found in Table S2) can be found at https://www.ncbi.nlm.nih.gov/genbank/ (accessed on 10 August 2021).

**Acknowledgments:** The authors would like to thank Steier, J., Lehman, K., and Wright, J. for their help in generating the barcode sequence data for shrubs and lianas on Barro Colorado Island.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
