Management recommendations for short-lived weeds depend on model structure and explicit characterization of density dependence

1. Multiple modelling techniques are currently used to describe population dynamics of established invasions, where intraspecific competition is likely to reduce survival, growth and/or fecundity, suppressing population growth rate. To date, it remains unanswered whether these modelling techniques produce similar management recommendations for density-dependent weed populations and how to model density dependence to better inform management. 2. We constructed demographic models for a short-lived weed based on data on multiple manipulated densities in a glasshouse and data from the literature using three germination strategies. We compared management recommendations produced by two main modelling techniques for density-dependent weed populations and examined whether periodic matrix population models constructed from different densities without characterization of density-dependent processes (implicit models) produce management recommendations similar to that of the same models with density dependence explicitly characterized and simulated (explicit models). The use of a periodic matrix population model enabled us to target simulated management on either vital rates or entire life stages, and to examine the role of a weed's germination strategy on model outcomes. 3. Management recommendations differed depending on how density dependence was included in demographic models. Explicit models showed that management conducted after the density-dependent process driving population dynamics best curbed density-regulated weed populations, with reductions in seed production having a negligible effect regardless of a germination strategy. By contrast, implicit models constructed from multiple densities produced similar management recommendations for sparse and dense populations, with reductions in survival to a flowering stage, juvenile establishment or seed production leading to the greatest predicted declines in weed density. 4. Our results emphasize the importance of model structure when modelling dynamics of density-dependent weed populations, suggesting that explicit characterization and inclusion of density dependence in population models is often necessary to inform management. As a weed's germination strategy had a minor effect on model outcomes, our findings about explicit and implicit modelling techniques can be generalized across annual plants with non-overlapping generations.