Losses in microbial functional diversity reduce the rate of key soil processes.

50 The consequences of microbial functional diversity loss on key ecosystem processes remain 51 debatable due to lack of firm evidence from observational or manipulative experiments for a 52 link between microbial functional diversity and specialized ecosystem functions. Here, we 53 conducted a microcosm experiment to test for a link between multiple microbial functional 54 diversity (nitrifiers, methanotrophs and denitrifiers) and corresponding specialized soil 55 functions (nitrate availability, methane, and nitrous oxide flux) using the dilution-to-extinction 56 approach. We found that reductions in functional microbial diversity led to declines in the rates 57 of specialized soil processes. Additionally, partial correlations provided statistical evidence 58 that the correlations between microbial functional diversity and specialized functions were 59 maintained after accounting for functional gene abundance (qPCR data) and substrate 60 availability. Our analyses further suggested little redundancy in the relationship between 61 microbial functional diversity and specialized ecosystem functions. Our work provides 62 experimental evidence that microbial functional diversity is critical and directly linked to 63 maintaining the rates of specialized soil processes in terrestrial ecosystems..

Here, we used the dilution-to-extinction (e.g. Salonius, 1981;Peter et al., 2011;145 Philippot et al., 2013; Delgado-Baquerizo et al., 2016a) experimental approach on soil from 146 two independent sites to explore the relationship between microbial functional diversity and 147 4 specialized soil processes in terrestrial ecosystems. In this study, we explicitly examine the 148 links between microbial functional diversity (e.g., nitrifiers, methanotrophs and denitrifiers) 149 and the rates of specialized functions (e.g. CH4 flux, NO3 production, and N2O flux). All these 150 specialized functions require specific genes to encode enzymes capable of performing these 151 functions which are limited to relatively few microbial species. We chose these functional 152 groups because they are ubiquitous across the globe; functional genes that catalyse processes 153 are well characterised and studied, and their exact role and mechanisms in carrying out 154 processes are well established. This provides a strong theoretical framework to test the linkages 155 between microbial functional diversity and specialized functions. Additionally, activities of 156 these functional microbial communities play key roles in climate regulation (e.g. greenhouse 157 gas emission and mitigation) and nutrient (N) cycling. We aim to experimentally test the 158 hypothesis that reduction in the microbial functional diversity has proportional impact on the 159 specialized processes in terrestrial ecosystems. We hypothesized that: (a) experimental losses 160 in microbial functional diversity will lead to reductions in specialized soil processes; and (b) 161 given the expected importance of soil microbial functional diversity for key soil processes, the 162 microbial BEF relationship should show little redundancy.

165
Site description 166 We collected soil samples from two sites in Australia with contrasting precipitation regimes -  Table 1.

174
Microcosm preparation 175 Soil samples from each site were sieved to < 2mm and divided in two portions: (1) soil for 176 sterilization, and (2) soil for microbial inoculum and experimental controls (non-sterilized 177 original soils). The first portion was sterilised using a double dose of gamma radiation (50kGy 178 each) at ANSTO Life Sciences facilities, Sydney. Gamma radiation was used as it is known to 179 cause minimal change to the physical and chemical properties of soils when compared with 180 other methods of sterilisation such as autoclaving (Wolf et al., 1989;Lotrario et al., 1995). The 181 dilution-to-extinction approach was used to prepare soil microcosms (Salonius, 1981  For each soil (soils A and B), 5 dilutions were used as the microbial inoculum (20 ml of 187 inoculum for each microcosm) to create a diversity gradient; these dilutions were undiluted 188 (10 0 ); 1/10 dilution (D1); 1/10 3 dilution (D3); 1/10 6 dilution (D6); and 1/10 10 dilution (D10).

189
Microcosms with non-sterilized soil served as references but not included in our statistical

215
Abundance of functional genes 216 The abundances of ammonia-oxidizing archaea (using amoA gene), N2O reducing bacteria 217 (using nosZ gene), and methanotrophs (using pmoA gene) were quantified on a CFX-96 218 thermocycler (Biorad, USA) using primers and conditions described in Table S1. Standard 219 curves were generated using ten-fold serial dilutions of plasmids containing the correct insert  Table S1. The PCR products were purified using the Wizard SV Gel    Testing the relationship between microbial diversity and specialised soil functions. 282 We used two independent approaches to analyse our dataset (a P-value and a non P-value 283 approach). First, we tested for differences in functional diversity and key processes across  This is a non-parametric method. As we did not transform our data, we used the Bray-Curtis 289 distance matrix for these analyses -to reduce the influence of extreme values. Additionally, as 290 an alternative statistical approach, we also used Spearman's correlation analysis to evaluate the 291 correlations between microbial functional diversity and specialized functions. We conducted 292 partial correlation analysis to evaluate any potential influence of abundance of functional gene   312 After a six-week incubation, we measured the abundance of functional genes (N2O reducing 313 bacteria using nosZ gene; methanotrophs using pmoA gene and ammonia oxidising archaea and 314 bacteria using amoA) using qPCR -as a proxy for biomass of functional groups in our soil 315 microcosms from two different sites (Soil A and B). Our results showed that microbial 316 abundance had successfully recovered in all diversity dilution microcosms. As such, we did 317 not detect significant differences for microbial abundance levels across different dilution 318 treatments (PERMANOVA P > 0.05; Fig. 1).

322
Shannon diversity for these functional genes was always positively and significantly related to 323 richness in both soils (P < 0.05).

325
Links between functional diversity and specialized functions 326 We observed significant correlations between the diversity of functional groups and their 327 specialized functions for both soil types using all three models tested (Fig. 3). The values of 328 specialized functions across different dilutions are shown in Fig. S1. These correlations were 329 maintained after using an alternative non-parametric approach (Spearman ; Table S2). was significantly correlated to specialized functions. As NO3concentration is also known to 336 regulate N2O production we conducted further partial correlation analysis using nosZ 337 functional diversity as a predictor of N2O flux controlled by NO3concentrations. Our results 338 showed significant correlations of functional diversity of denitrifiers with N2O flux even after 339 accounting for nitrate production (Table S4). 340 Overall, statistical modelling did not demonstrate functional redundancy in the 341 relationship between microbial functional diversity and soil processes (Table 2). In fact, we 342 8 observed little functional redundancy in our results. Thus, the redundant (logarithmic) 343 relationships were observed only in two cases including the relationship between functional 344 diversity and N2O flux and NO3 production at site A (Table 2). In the rest of the cases -4 out 345 of 6 a proportional loss or not clear functional redundancy was detected (Table 2).    Together, our study provides experimental evidence that, similar to what has been 405 reported for plant functional diversity, microbial functional diversity largely influence 406 important soil processes associated with the production of NO3, and fluxes of N2O and CH4. 407 We also provide evidence that the correlation between functional diversity and specialized 408 functions is robust to any effects from functional gene abundance and substrate availability.

409
Our results further suggest that there is little functional redundancy in the relationship between