Comparing stand-level volume growth and yield predictions from time-explicit, state-space, and simultaneous approaches across complex mixed forest stands in the Acadian Region of North America

Abstract

Stand-level growth and yield models are computationally simple and have moderate demands on input information, but their applications have largely been limited to single-species and even-aged stands. State-space and simultaneous approaches that are not explicitly functions of time potentially extend the applicability of stand-level growth and yield models to mixed-species and/or multi-cohort stands where age (and the period of stand growth that often is derived as the difference in ages) may not be available and/or suitable predictors. The Acadian Forest, dominated by naturally-regenerated and mixed-species stands of primarily multi-cohort structure, offers an ideal framework for a comprehensive assessment of alternative approaches, such as time-explicit, state-space, and simultaneous approaches’ performance in stand-level growth and yield predictions across a wide range of complex forest stands. We conducted a comparative assessment utilizing an extensive database from this region. It was found that the three approaches were highly consistent in providing relatively accurate and largely unbiased stand-level growth and yield predictions. The time-explicit approach had a simplistic form but similar prediction performances comparable to the other two more complex modeling approaches. In comparison, the simultaneous approach, despite being path-invariant, was computationally challenging and offered limited improvements from the other approaches. Our findings showed potential applicability of stand-level growth and yield models beyond single-species and even-aged forests.