摘要：
Aims To determine the detectability of a global weedy perennial weed Hypochaeris radicata and its relationship with five common observer, species and environmental variables.
Methods Trained independent observers conducted time-limited repeat surveys of H. radicata during autumn in an endangered grassy box-gum woodland ecosystem in south-east Australia. Single-species single-season site-occupancy modelling was used to determine if detectability of H. radicata was altered by five covariates, observer, litter height, grazing, maximum plant height and flowering state.
Important findings Detectability for H. radicata varied significantly with observer, litter height, plant maximum height and flowering state, but not with grazing. Despite significant observer-specific variation, there was a consistent increase in detectability with plant height and when plants are in flower for all observers. Detectability generally decreased as litter height increases. Perfect or constant detection rates cannot be assumed in plant surveys, even for easily recognizable plants in simple survey conditions. Understanding how detectability is influenced by common survey variables can help improve the efficacy of plant monitoring programs by quantifying the extent of uncertainty in inferences made from survey data, or by determining optimal survey conditions to increase the reliability of collected data. For plants with traits similar to H. radicata, surveying when most plants are at maximum height or in flower, increasing search intensity when litter levels are high and minimizing observer-related heterogeneity are potentially simple and effective ways to reduce detection errors. We speculate that detection rates may be lower, more variable and involve additional covariates when surveying during the peak flowering spring season with the presence of more warm season and taller annual species.
