Heavy metals can persist in the environment for decades, yet relatively little is known about how soils respond to exposure over decadal timescales. Acceptable metal levels in soils are often determined by short-term, controlled exposure experiments, where impacts to microbial communities and functional endpoints are measured over weeks, days, or hours (Kuperman and Carreiro, 1997; Beare et al., 1990; Hopkins et al., 1994; Farrell et al., 2009; Chander and Joergensen, 2002; Smolders et al., 2009; Zheng et al., 2017; Pietikäinen and Fritze, 1995; EPA USA, 2007). The acclimation time in these short-term experiments may not be representative of how soil environments respond to contamination events in the long-term, as changes reflect only immediate microbial responses (Smolders et al., 2003). Previous studies have confirmed this, concluding that metal toxicity to soil microbes in freshly spiked soils bears little resemblance to the effects detected in field-contaminated soils over longer timeframes (Renella et al., 2002; Smolders et al., 2004). Often, retrospective observations from locations such as historical mining sites are used as a proxy for longer-term exposure research (Pennanen et al., 1996; Dell’Amico et al., 2008; Berg et al., 2005; Berg et al., 2012; Liu et al., 2005; Oliveira and Pampulha, 2006). However, retrospective observational studies typically only measure metal concentrations at the time of sampling, even though concentrations may have varied considerably since the contamination event began.

One of the few long-term studies that monitored and controlled exposure concentrations found that historical Cu exposure (15-year) was directly associated with tolerance to Cu exposure (Wakelin et al., 2014). However, this study only tested up to 200 mg kg−1 Cu, which is only marginally higher than Cu concentrations naturally found in agricultural soils (NSW EPA, 1997). Other microbial studies have similarly observed community resilience to functional loss (e.g. nitrification) when re-exposed to heavy metals (Smolders et al., 2003; Mertens et al., 2006; Diaz-Ravina et al., 1996; Díaz-Raviña et al., 1994; Kunito et al., 1999), demonstrating microbial ability to adapt and tolerate new, harsher environmental conditions. However, it is unknown whether there is maximum tolerance threshold after which soil microbial communities cease to adapt or build resilience and are irreversibly, functionally impacted in situ.

To address this question, we examined both soil functionality and microbial community structure in a long-term (12-year), controlled, field-based experiment, which tested a wide range of Copper (Cu) concentrations: from 0 to 3310 mg Cu kg−1. Copper was applied to agricultural soils in 22 distinct, randomised plots within a 1 km radius, and concentrations at these sites, along with two control sites, were monitored over 12 years with no amendment or human interference. We combined environmental DNA (eDNA) sequencing with a range of functionality assays (substrate induced respiration (SIR), substrate induced nitrification (SIN), proteolysis, and peptide and amino acid turnover) to analyse and compare community composition and soil function across the different Cu treatments. The functionality assays chosen, assess common soil functions, including carbon and nitrogen turnover. This long-term experimental field site presented a unique opportunity to determine the concentration at which contaminated soils can return to former functionality if contamination ceases (e.g., mining, ore extraction, or biosolid application). We also examined whether conserved functionality reflects previous (non-exposed) community structures or if novel community assemblages are created.