Material and Methods
We conducted a systematic literature search following the general PRISMA EcoEvo guidelines (O’Dea et al. 2021) to find a representative sample of all published literature testing the impact of microbial inoculation on plant growth in dryland ecosystems. We searched Scopus and ISI Web of Science databases using UNSW Sydney’s institution subscriptions for all studies published before 1st April, 2022 using the search terms: (inocula*) AND (”plant growth” OR ”plant biomass” OR ”plant development”) AND (”dryland*” OR ”arid land*” OR ”semi-arid land*”) AND (“restor*” OR “remediat*” OR “rehabilita*”). Our search resulted in 923 publications for Web of Science and 850 for Scopus. The total number of publications was reduced to 1235 after 538 duplicates were removed (Table S1). We additionally included one study from the author, Dadzie et al. (2022).
We selected studies based on five main eligibility criteria. First, we only considered peer-reviewed primary studies, thus excluding books, review papers, and theses. Second, the study must have been conducted in a dryland ecosystem. We defined drylands as areas with annual precipitation under 600 mm where potential evapotranspiration was greater than the annual precipitation. When studies did not provide climate information, we used location information to obtain approximate climate information within the study duration via cross-checking with historical WorldClim data (version 2.1, 30 seconds resolution ~ 1 km square grid; Fick and Hijmans, 2017). Third, only field studies were considered. Studies that were conducted in pots (mesocosms) and left in the field were not included since the pots provided a microclimatic condition. Fourth, studies must have compared microbially inoculated and non-inoculated plants. We defined microbial inoculation as bacteria and fungi that are purposefully added to plants to improve their growth. Finally, studies must have reported above-ground biomass as a metric for plant growth and development. Studies that reported plant growth with other metrics were not included. For studies that reported plant growth in consecutive years, only the last year of the growing season was considered since we were interested in the overall effect of microbial inoculation on plant growth.
To increase the comparability of our estimates, we ensured that the conditions applied to each treatment were equally applied to the control in each study. For example, when studies reported on the application of fertilizer to the field before planting seeds, we ensured that the controls were given similar fertilizer treatment else they were excluded. In studies that reported both fertilizer application and microbial inoculations as individual treatments, we selected controls that were unfertilized and uninoculated, whereas only treatments that were unfertilized but inoculated with microbial were considered for the response measure. Where there was a crop rotation, we only included data from the first rotation season since subsequent rotations were not receiving direct inoculation and were confounded with plough-back remnant crops.