A look into a rumen

Roe Deer (Capreolus capreolus)

Dietary choices are central to our understanding of ecology and evolution. Still, many aspects of food choice have been hampered by time consuming procedures and methodological problems. Faster and cheaper methods, such as DNA metabarcoding, have therefore been widely adopted.

Traditional visual techniques based on morphology for diet analysis are often limited in their ability to identify all items found in a stomach. Unrecognizable content has to be ignored which can shift relative proportions within analysis. This happens mostly when digestion is well advanced, leaving researchers without any characters that would enable visual identification. It is well known that this can lead to biased results in studies of processes such as food web interactions and energy flow through an ecosystem. DNA barcoding has been suggested as a tool to overcome these issues and over the last 6-7 years a number of studies has been published all of which showed the increase of resolution by utilizing molecular techniques. 

A new study by Swedish researchers looked at the use of metabarcoding to assess ungulate diet through rumen content. They were interested in the diet of Fallow Deer (Dama dama) and Roe Deer (Capreolus capreolus).  In addition they set out to empirically compare metabarcoding to visual identifications. Not surprisingly, metabarcoding provided higher yield and precision of results although  they found that the methods were comparable, but they did not completely overlap. Sometimes the DNA method failed to identify food items that were found macroscopically...

When the colleagues applied both sets of data they found some very encouraging things.  In niche overlap analyses, presence/absence data showed that both methods came to very similar conclusions. When using the sequence count data and macroscopic weight, niche overlap was lower than when using presence-absence data yet tended to increase when using DNA compared to macroscopy. Nevertheless, the significant positive correlation between macroscopic quantity and number of DNA sequences counted from the same plant group give support for the use of metabarcoding to quantify plants in the rumen. 

Again, very promising evidence for the power of DNA barcoding in the context of dietary analysis. It would be very interesting to see how the results of this study correlate with an analysis of fecal samples. As a non-invasive method would increase sample size and effort quite extensively. The other suggestion I have is to use more than just one marker (here trnL). Include ITS or rbcL as well which might boost the metabarcoding success even more.