Spatial Modelling of Miconia calvescens Invasion Risk in North Queensland Rainforests to Refine Surveillance Planning

Abstract

Miconia calvescens is one of the worst invaders of tropical rainforest environments in the Asia‐Pacific region. Introduced as a garden plant in north Queensland Australia, naturalised infestations in the rainforests emerged 30 years ago. The National Tropical Weeds Eradication Program (NTWEP) was established to coordinate national eradication efforts. While the program targets multiple invasive plant species, around 75% of surveillance effort is expended on M. calvescens . This small rainforest tree produces abundant seed, which is dispersed by birds and forms highly persistent soil seedbanks. Significant long‐term search effort is required to detect plants, across variable and challenging terrain. We developed a spatial model of M. calvescens invasion risk to inform and refine decisions on surveillance planning. The model considers environmental factors that support plant establishment and persistence, and dispersal factors that support the introduction of viable seed by birds. The model output is a spatial risk layer showing susceptibility to plant invasion, categorised from very low to very high. Susceptibility was predicted from known locations of mature M. calvescens plants at three time steps (December 2016, 2019 and 2022). Model predictions for 2016 and 2019 were retrospectively evaluated against actual plant detections and search effort by NTWEP in the following 3 years. Results indicate that both models accurately predicted actual detections in areas classified as very high to moderate risk. While most of the model‐predicted high‐risk areas were actually searched, much search effort was also expended in low to ‘no’ risk areas where relatively few plants were found. The model prediction for 2022 was evaluated against current surveillance area coverage targets, demonstrating that current planning effort could be refined by invasion risk. We recommend that our model can inform risk‐based surveillance planning while still maintaining confidence that surveillance coverage is suitably comprehensive for the needs of an eradication program.