Cadmium (Cd) is a biotoxic heavy metal, which is accumulated by plants and animals and thereby enters the human food chain (Gray et al. 2003). The application of phosphate fertilisers has also resulted in the long-term accumulation of Cd in agricultural soils around the world, including New Zealand (NZ). In 1997, the main source of NZ phosphate fertilisers was changed from Nauru island phosphate rocks (450 mg Cd kg-1 P) to a variety of phosphate rocks with lower Cd concentrations, in order to meet more stringent Cd limits in P fertiliser. Following this change, the accumulation of Cd in topsoil samples from the Winchmore research farm (South Island, NZ) was evaluated and was found to have plateaued post-2000 (McDowell, 2012). In this study, stable isotope analysis was used to trace the fate of Cd in Winchmore farm soils in order to determine the cause of the plateau.

The isotope ratio of Cd (δ114/110Cd) was measured in pre-2000 and post-2000 phosphate fertilisers, phosphate rocks, topsoil (0-7.5 cm) and control (unfertilised) subsoil (25-30 cm) samples from the Winchmore site. The analysed topsoil samples were archived samples collected over the period 1959-2015. The isotopic compositions of fertilised topsoils ranged from δ114/110Cd = 0.08 ± 0.03 to δ114/110Cd = 0.27 ± 0.04, which were comparable to pre-2000 fertilisers (δ114/110Cd = 0.10 ± 0.05 to 0.25 ± 0.04) but distinct from the post-2000 fertilisers (δ114/110Cd range of -0.17 ± 0.03 to 0.01 ± 0.05) and control subsoil (δ114/110Cd = -0.33 ± 0.04) (Salmanzadeh et al., 2017). We combined this stable isotope data with Bayesian modelling to estimate the contribution of different sources of Cd. An open source Bayesian isotope mixing model implemented in Matlab (Arendt et al., 2015) was used here with some modifications to estimate the fractional contribution of different sources of Cd through time including pre- and post-2000 fertilisers, and the control soil. The Matlab code of Arendt et al., 2015 was modified to consider only one isotope system (rather than two), and fewer sources. This modelling confirmed the dominant contribution (about 80%) of Nauru-derived (i.e. pre-2000) fertilisers in increasing the Cd concentration in Winchmore soils.

To help constrain the soil Cd mass balance we used an existing model (CadBal) (Roberts and Longhurst, 2005), to estimate residual soil Cd and output fluxes based on known P fertiliser application rates, the initial Cd concentration, farm and soil type, and soil dry bulk density. We incorporated the isotope data into the mass balance expression in order to evaluate the performance of CadBal in estimating the past topsoil Cd accumulation and predicting the future concentrations and isotope ratios of Cd (up to 2030 AD). The results of mass balance modelling confirm that recent applications of phosphate fertilisers have not resulted in an accumulation of Cd during the most recent period, thus Cd removal by either leaching or crop uptake has increased, which is consistent with the modelled isotope data (Figure 1).

We can conclude that it becomes possible to distinguish the sources of Cd within the soil using stable Cd isotopes (Imseng et al., 2018) and that the residual Cd in topsoil at Winchmore still mainly originates from historical phosphate fertilisers (Salmanzadeh et al., 2017). One implication of this finding is that the contemporary applications of phosphate fertiliser are not resulting in further Cd accumulation. We aim to continue our research into Cd fate, mobility, and transformations in the NZ environment by applying Cd isotopes in soils and aquatic environments across the country.

Figure 1. Results of Cd mass balance modelling in CadBal for the period of topsoil fertilisation including a prediction up to the year 2030 AD. (a) Mean concentration of Cd in the dryland treatment of Winchmore long-term irrigation trial (symbols) and the CadBal model (lines) outputs (red symbols = this study- plot 15 of Winchmore site; grey symbols = McDowell study-average of all plots; solid black line = dryland optimized CadBal from McDowell (2012) for all irrigation plots; black dashed line = Plot 15 dryland optimized CadBal-this study, first scenario; blue line = Plot 15 dryland optimized CadBal-this study, second scenario; red line = Plot 15 dryland optimized CadBal-this study, third scenario; red dashed line = Plot 15 dryland optimized CadBal-this study, fourth scenario); (b) Measured and modelled Cd isotope ratios based on CadBal outputs, isotope ratios measured in fertilisers and the fractionation factors of Wiggenhauser, et al. (2016); lines designate modelling scenarios as in (a), red dots are the third scenario with no fractionation (α factor not applied); (c) modeled scenario 3 (solid) and scenario 4 (dashed) isotope ratios in topsoil (red lines), leachate (blue lines) and pasture (green lines).