ABSTRACT
Historical hard-rock mine activities have resulted in nearly half a million mining-impacted sites scattered around the US. Compared to conventional remediation, (aided) phytostabilization is generally cost-effective and ecologically productive approach, particularly for large-scale sites. Native species act to maintain higher local biodiversity, providing a foundation for natural ecological succession. Due to heterogeneity of mine waste, revegetation strategies are inconsistent in approach, and to avoid failure scenarios, greenhouse screening studies can identify candidate plants and amendment strategies before scaling up. This greenhouse study aimed to concurrently screen a variety of native species for their potential to revegetate Cu/Pb/Zn mine tailings and develop a high throughput and non-destructive approach utilizing computer vision and image-based phenotyping technologies to quantify plant responses. A total number of 34 species were screened in this study, which included: 5 trees, 8 grasses, and 21 forbs and legumes. Most of the species tested were Missouri native and prairie species. Plants were non-destructively imaged, and 15 shape and color phenotypic attributes were extracted utilizing computer vision techniques of PlantCV. Compared to reference soil, all species tested were negatively impacted by the tailings' characteristics, with lowest tolerance generally observed in tree species. However, significant improvement in plant growth and tolerance generally observed with biosolids addition with biomass surpassing reference soil for most legumes. Accumulation of Cu, Pb, and Zn was below Domestic Animal Toxicity Limits in most species. Statistically robust differences in species responses were observed using phenotypic data, such as area, height, width, color, and 9 other morphological attributes. Correlations with destructive data indicated that area displayed the greatest positive correlation with biomass and color the greatest negative correlation with shoot metals. Computer visualization greatly increased the phenotypic data and offers a breakthrough in rapid, high throughput data collection to project site-specific phytostabilization strategies to efficiently restore mine-impacted sites.