Advancing the integration of multi-marker metabarcoding data in dietary analysis of trophic generalists

Abstract

The application of DNA metabarcoding to dietary analysis of trophic generalists requires using multiple markers in order to overcome problems of primer specificity and bias. However, limited attention has been given to the integration of information from multiple markers, particularly when they partly overlap in the taxa amplified, and vary in taxonomic resolution and biases. Here, we test the use of a mix of universal and specific markers, provide criteria to integrate multi-marker metabarcoding data and a python script to implement such criteria and produce a single list of taxa ingested per sample. We then compare the results of dietary analysis based on morphological methods, single markers, and the proposed combination of multiple markers. The study was based on the analysis of 115 faeces from a small passerine, the Black Wheatears (Oenanthe leucura). Morphological analysis detected far fewer plant taxa (12) than either a universal 18S marker (57) or the plant trnL marker (124). This may partly reflect the detection of secondary ingestion by molecular methods. Morphological identification also detected far fewer taxa (23) than when using 18S (91) or the arthropod markers IN16STK (244) and ZBJ (231), though each method missed or underestimated some prey items. Integration of multi-marker data provided far more detailed dietary information than any single marker and estimated higher frequencies of occurrence of all taxa. Overall, our results show the value of integrating data from multiple, taxonomically overlapping markers in an example dietary data set.

Keywords: bird; diet; metabarcoding; morphological identification; overlapping markers; secondary predation.

Figure 1

Number of consumed taxa observed at different taxonomic levels (left) and number of occurrences observed at each taxonomic level (right), during the morphological identification (Morphology), with four individual molecular markers (18S, universal marker; trnL, plant marker; IN16STK and ZBJ, arthropod specific markers) and with the multi‐marker approach, for plants and animals. Note that for the morphological identification, animal fragments were not compared across samples, and therefore the total number of taxa corresponds to the sum of the maximum number of morphotypes detected per family and order [Colour figure can be viewed at http://wileyonlinelibrary.com]

Figure 2

Frequencies of occurrence of each order of plants and animals in the diet of Black Wheatears obtained through morphological and molecular analysis (multi‐marker, dark grey bar, and for each set of primers). The orders highlighted in bold indicate significant differences at univariate tests of Multivariate Generalized Linear Models. ¹indicates orders that only showed significant differences among the molecular markers and morphological identification [Colour figure can be viewed at http://wileyonlinelibrary.com]

Figure 3

Czekanowski's overlap index for plants and animals, between the morphological identification, the several molecular markers and the multi‐marker approach used in Black Wheatear diet analysis [Colour figure can be viewed at http://wileyonlinelibrary.com]