COS 37-5 - Determining the feeding networks of grasshoppers in grasslands of the southeastern United States using DNA barcoding

Tuesday, August 8, 2017: 9:20 AM
C122, Oregon Convention Center
John A. Barone1, Kevin S. Burgess1 and JoVonn G. Hill2, (1)Biology, Columbus State University, Columbus, GA, (2)Biochemistry, Molecular Biology, Entomology and Plant Pathology, Mississippi Entomological Museum, Mississippi State, MS

A significant challenge in building food webs has been determining the diets of mobile or wary animals. With insects, traditional methods have included field observations, feeding trials, and the examination of plant cuticles in feces. Over the last five years, ecologists have begun using DNA barcoding as a new tool to assess the diets of herbivorous insects. To date, most studies have amplified plant DNA from whole body samples and compared the plant DNA to GenBank. However, local plant DNA barcode libraries can increase the accuracy of plant identifications, particularly for generalist herbivores such as grasshoppers.

We are building food webs using DNA barcoding for grasshoppers and their host plants from two southeastern habitats, blackland prairies and longleaf pine savannas. For blackland prairies, we have sequenced two gene regions, rbcL and matK, for over 200 plant species. For longleaf pine savannas, we have sequenced rbcL for over 100 plant species. To determine grasshopper diets, we collected feces from grasshoppers in the field. Plant DNA was extracted from these samples, amplified using PCR, and sequenced. Sequences were edited and then compared to the appropriate DNA barcode library, or when no clear match was present, to GenBank.


We obtained high quality DNA sequences from 114 grasshopper samples (from 24 grasshopper species), including 103 from prairies and 11 from longleaf pine savannas. Of these 88 (77%) contained 0 to 2 ambiguous bases. The remainder had larger numbers of ambiguous bases, either because of degradation of the DNA or the presence of DNA from two or more plant species. Seventy-two samples perfectly matched plant barcodes in one of the libraries. The remaining 16 were likely for plants not included in the barcode libraries. Five of the 11 samples from longleaf pine savanna could not be confidently assigned to species, presumably because the library for the habitat is not close to completion. Overall, we have found that DNA barcoding is a valuable complement to fieldwork in determining the diets of insect herbivores.