A comparison of the ability of PLFA and 16S rRNA gene metabarcoding to resolve soil community change and predict ecosystem functions
Introduction
Soil bacterial community structure is responsive to a myriad of factors, including management practices and global change (Clegg et al., 2003, Evans and Wallenstein, 2014, Fierer et al., 2012, Francisco et al., 2016), and has been linked to ecosystem functioning (Fierer et al., 2012, Orwin et al., 2016, Whitaker et al., 2014). Soil bacterial community structure has traditionally been measured using phospholipid fatty acid (PLFA) profiling, which gives a broad-scale, quantitative overview of the living bacterial community (Willers et al., 2015). However, the development of next-generation sequencing technologies and extensive reference databases for the V3-V4 region of the microbial 16S rRNA gene have allowed information about soil bacterial communities to be obtained at a much finer taxonomic resolution (Smets et al., 2016). Although 16S rRNA gene metabarcoding has a clear advantage where the focus of the study is on the details of bacterial ecology, the relative ability of PLFA profiling and 16S rRNA gene metabarcoding techniques to answer many of the broader questions often asked by ecosystem ecologists, and whether results can be directly compared, is largely untested. Such knowledge would help integrate the wealth of information provided by PLFA profiling-based studies with that gained from metabarcoding approaches.
One key question often asked by ecologists is how communities change in response to a given driver, with community data either analysed using multivariate tests or by calculating metrics to describe the ways in which the community has changed. A previous study showed that PLFA profiling was as sensitive as – and sometimes more sensitive than – DNA fingerprinting techniques for indicating changes in community structure (Ramsey et al., 2006). However, there are few tests of whether the greater detail available from 16S rRNA gene metabarcoding improves sensitivity (but see Cesarz et al., 2013, Chodak et al., 2013, Docherty et al., 2015, Duncan et al., 2016). One major metric used in both PLFA profiling and metabarcoding studies is estimated diversity. Diversity of bacterial PLFAs is generally a function of evenness as most PLFA markers are present in all samples, whereas diversity in metabarcoding reflects both evenness and taxon richness. Other metrics are more commonly used in PLFA profiling than in metabarcoding studies. For example, PLFA-based estimates of the gram-positive:gram-negative bacterial and fungal:bacterial ratios have been linked to community growth strategies (de Vries and Shade, 2013, Treseder et al., 2011, Wardle et al., 2004), and so allow an ecological interpretation of why communities have changed in a particular way (Table 1). A similar approach may help interpret results from 16S rRNA gene metabarcoding data. Three possible metrics currently exist: the Proteobacteria:Acidobacteria ratio, which is thought to reflect the nutrient status of the soil (Smit et al., 2001); the oligotrophic:copiotrophic ratio, which is based on the relative abundance of specific taxa at the phylum and class level thought to belong to each group (Collins et al., 2016; Table 1); and the gram-positive:gram-negative ratio (Treseder et al., 2011, Yuste et al., 2014; Table 1). Although gram staining does not appear to be strongly related to phylogeny, cell envelope type (i.e. monoderm vs diderm), which the gram stain largely reflects, is relatively distinct between phyla (Sutcliffe, 2010, Gupta, 2011). A gram-positive:gram-negative ratio based on cell envelope type, or the relative abundance of each of these groups, could therefore be useful metrics to use. A strong correlation between metrics based on PLFA profiling and 16S rRNA gene metabarcoding would help validate their calculation and allow easier comparison between studies based on different techniques. However, such correlations have not been explored previously.
Ecologists are also increasingly interested in the role that bacterial communities play in driving ecosystem functioning (e.g. Grigulis et al., 2013, Orwin et al., 2016, Whitaker et al., 2014). Although metrics based on both 16S rRNA gene metabarcoding and PLFA profiling have been linked to ecosystem functions such as respiration (Collins et al., 2016, Whitaker et al., 2014), nitrogen cycling (Philippot et al., 2013), and responses to disturbance (Orwin et al., 2016), no study that we are aware of has compared their relative ability to explain variation in functioning. Thus it is currently unknown how well PLFA profiling- and metabarcoding-based metrics correlate with each other, and what level of taxonomic resolution is required to detect community change and predict ecosystem functions. Here we use a broad land-use gradient to assess the commonalities and differences between PLFA profiling and 16S rRNA gene metabarcoding results. Specifically, we (i) test which technique is more sensitive to change in bacterial community composition across land uses, (ii) test whether similar metrics calculated from PLFA profiling and 16S rRNA gene metabarcoding data are correlated with each other, and (iii) examine the relative ability of each technique to predict a range of ecosystem functions (carbon and nitrogen cycling, and responses to disturbance).
Section snippets
Field site
This study utilises data from two previously published studies (Orwin et al., 2016, Wood et al., submitted) where both PLFA profiling and 16S rRNA gene metabarcoding techniques were applied to the same soil samples. Samples were collected from sites located in the Wairau River catchment in Marlborough, New Zealand (41°30′S, 173°50′E). Six 20 × 20-m plots were sampled in each of five land-use types: natural forest (dominated by Lophozonia menziesii (Hook.f.) Heenan et Smissen and Fuscospora
Ability to distinguish land-use types
Bacterial PLFA profiling and metabarcoding data were able to differentiate the bacterial communities in several of the land-use types examined here (Fig. 1). Natural forests had higher relative abundances of some of the PLFA markers for gram negative bacteria (cy-19:0, and 18:1ω7c) and the general marker 15:0, whereas grasslands and vineyards had higher relative abundances of the PLFAs associated with gram positive bacteria. At the phylum level, Acidobacteria showed the highest relative
Discussion
Our findings suggest that results from PLFA profiling and 16S rRNA gene metabarcoding are broadly comparable: both techniques showed a similar pattern of community differences among the land-use types, metrics based on each technique were mostly well correlated, and they generally explained a similar amount of variation in a wide variety of ecosystem functions. This provides a degree of validation of both methods, and suggests that results from PLFA profiling studies should be broadly
Acknowledgements
We thank private landowners, the Marlborough District Council, and the Department of Conservation for assisting with access to the sample locations, and Chris Morse, Ella Hayman, Hamish Maule, Brian Karl, Sarah Kruis, Kev Drew, Geoff Walls, Karen Bonner and Nic Bolstridge for technical support and field assistance. This project was funded by the New Zealand Ministry of Business, Innovation and Environment through the Smart Idea funding mechanism (contract C09X1205).
References (49)
- et al.
Review and re-analysis of domain-specific 16S primers
Journal of Microbiological Methods
(2003) - et al.
Soil carbon quality and nitrogen fertilization structure bacterial communities with predictable responses of major bacterial phyla
Applied Soil Ecology
(2014) - et al.
Roots from beech (Fagus sylvatica L.) and ash (Fraxinus excelsior L.) differentially affect soil microorganisms and carbon dynamics
Soil Biology & Biochemistry
(2013) - et al.
Diversity of microorganisms from forest soils differently polluted with heavy metals
Applied Soil Ecology
(2013) - et al.
The impact of grassland management regime on the community structure of selected bacterial groups in soils
Fems Microbiology Ecology
(2003) - et al.
Nitrogen losses from two grassland soils with different fungal biomass
Soil Biology & Biochemistry
(2011) - et al.
Detection of short-term cropping system-induced changes to soil bacterial communities differs among four molecular characterization methods
Soil Biology & Biochemistry
(2016) - et al.
European scale analysis of phospholipid fatty acid composition of soils to establish operating ranges
Applied Soil Ecology
(2016) - et al.
Use and misuse of PLFA measurements in soils
Soil Biology & Biochemistry
(2011) - et al.
Do shifts in life strategies explain microbial community responses to increasing nitrogen in tundra soil?
Soil Biology & Biochemistry
(2016)