Digital Hydrography Underestimates Stream Length and Leads to Underestimates of Trout Population Size

Stream length is measured for many fisheries management applications. Characteristics of populations and habitats measured at field sites are commonly generalized to unsampled areas using estimates of stream length or stream network length. There are many ways to measure stream length, but map-based stream length measurements are commonly used in fisheries applications even though they are known to be biased. We evaluated how length of headwater streams in Arizona may be underestimated by the National Hydrography Dataset and how that bias influences streamwide abundance estimates for adult Apache Trout Oncorhynchus apache. As expected, stream lengths measured using National Hydrography Dataset flowlines underestimated true length revealed by National Agricultural Imagery Program imagery on average 11.1% (SD = 4.1%), and this bias was higher in meadow versus forested habitats. The observed bias led to streamwide estimates of adult Apache Trout abundance that were only 88% on average (SD = 5%) of those made with more realistic imagery-based stream measurements. As we have shown, high-resolution imagery, now widely available, can be used to assess and quantify stream length bias, and we conclude that it is important to assess whether this bias has the potential to negatively impact important fishery management decisions.