Metabarcoding soil and leaf-litter

Today I came across a new very interesting paper published in Ecological Indicators. An international group of researchers used established metabarcoding protocols to compare soil and leaf-litter samples with malaise trap samples in southern China and to compare leaf-litter samples with canopy-fogging and morphologically identified spider samples in central Vietnam.

For those that are not so familiar with the term metabarcoding here a text fragment from my most successful blog post ever (sic!): Metabarcoding is a rapid method of biodiversity assessment that combines two technologies: DNA based identification and high-throughput DNA sequencing. It uses universal PCR primers to mass-amplify DNA Barcodes from mass collections of organisms or from environmental DNA. The PCR product is sent to a next generation sequencer and the result is a wealth of DNA sequences. Such sequence collections are auditable, because sites can be sampled by independent parties, or samples can be split, and analysed by certified entities following a standardized protocol. They can also be verified by fieldwork to confirm the presence or absence of particular species. These metabarcode data sets are taxonomically more comprehensive, many times quicker to produce, and less reliant on taxonomic expertise.

However, the leading question in this publication is focusing on the sampling technology rather than the actual analysis in laboratory: Is it possible to substitute ground-level (soil or leaf litter) samples for aboveground samples when conducting biodiversity surveys? The colleagues argue that a soil or leaf-litter sample can be collected in minutes, whereas an aboveground sample, such as from Malaise traps or canopy fogging, can require days to set up and run, during which time they are subject to theft, damage, and deliberate contamination.

It is no secret that we run a lot of malaise traps of the course of Canada's summer seasons and we had only very few instances of damage due to other animals (damage caused by a bear) or vandalism (one case during our school project). However, as reported in the paper there are examples were traps have been damaged by larger animals or uniformed people. New to me was the deliberate adulteration of environmental surveys mentioned. Another disillusioning fact to think about and frankly it makes me angry to think that brilliant minds have to waste their time designing experiments and monitoring tools to counteract potential fraud and vandalism. 

On the other hand it is great to learn that the world we didn't really dare to enter (soil and leaf litter samples) in our efforts to catalog the life on our planet one barcode at a time, can be explored just as effectively as the one literally just centimeters above, which we target with our hundreds of malaise traps:

Here we show that while the taxonomic compositions of soil and leaf-litter samples are very different from aboveground samples, both types of samples provide similar ecological information, in terms of ranking sites by species richness and differentiating sites by beta diversity. In fact, leaf-litter samples appear to be as or more powerful than Malaise-trap and canopy-fogging samples at detecting habitat differences. We propose that metabarcoded leaf-litter and soil samples be widely tested as a candidate method for rapid environmental monitoring in terrestrial ecosystems.

The authors are still cautious with respect to the prospects of metabarcoding but the technology has made substantial progress in the last few years. Some of the issues described already sound like mere technicalities that can be overcome with some creativity and smart testing. 

I think it is necessary to mention that they used different marker systems for the different samples. For the ground-level samples they took 18S and for anything above ground they used COI which could introduce a bias in the analysis but the researchers took this (at least partially) into account:

We caution that the higher prevalence of nematodes and annelids in the ground-level samples could reasonably be attributed to the different genetic markers used; the 18S primers were designed to amplify across the Metazoa, whereas our COI primers are only known to amplify successfully across the Arthropoda (Our COI primers cannot be used to amplify from soil and leaf-litter samples because >99% of returned OTUs are bacterial). Regardless, the taxonomic compositions of the metabarcode datasets are consistent with the microhabitats from which the samples were collected. Soil and leaf-litter microhabitats are indeed highly species-rich in spiders, mites, centipedes, millipedes, roundworms, and ringed worms, whereas canopy-fogging and Malaise-trap samples do capture mostly insects.

I have to admit that I disagree with the authors a little as it is my hope that efforts will be made to develop functional COI primers to amplify DNA Barcodes for all metazoan groups (only very few groups will remain to be a problem) which would allow us to match undetermined metabarcoding samples to reference specimens.