Sources of uncertainty in stream-associated amphibian ecology and responses to forest management in the Pacific Northwest, USA: A review

Immediate and potentially prolonged impacts of timber harvest on stream-associated amphibians (SAA; giant (Dicamptodon spp.) and torrent (Rhyacotriton spp.) salamanders and coastal tailed frogs (Ascaphus truei)) in the Pacific Northwest, USA, are a concern due to widespread commercial management of forests across their geographic distribution. Evidence suggests that impacts from forest practices on SAA may vary in magnitude and direction, as well as spatially and temporally, as a result of biotic and abiotic factors, including management intensity and frequency. However, past investigations employed a variety of sampling and analytical methods which, when combined with apparent natural variation in SAA populations and habitat conditions, present a challenge to making reliable, unified inference about the relative influences of harvest practices and ecological processes on SAA responses. Nearly all SAA/forest management research has been of short duration, and many studies have occurred on landscapes that were harvested previously, limiting effective comparisons between current and historical distributions of amphibians and immediate vs. prolonged impacts of forest harvest. Also, forest practices have changed dramatically over the last two decades in the Pacific Northwest, especially management of riparian habitats, and information about SAA responses to current management prescriptions are not yet available. Regular, long-term monitoring of SAA responses should be conducted and effects associated with past and current management actions segregated, when possible. Many habitat studies have concentrated on stream features measured at local reach scales (50–150 m) and understanding the scale of potential land management impacts on SAA populations at a landscape level will require implementation of studies at larger spatial and longer temporal scales. Finally, aspects of SAA ecology, including competition, dispersal, predation, and vital rates, are poorly understood. To provide reliable inference, studies should (1) employ rigorous sampling designs (e.g., with temporal and spatial replication) and analytical techniques that permit estimation of detection probability; (2) occur at spatial and temporal scales appropriate to the questions being posed; and (3) be designed to estimate population responses to potential impacts.