Small Mammals: Information about historic and current beaver abundance is available at a statewide level, but this information likely underestimates true beaver abundance because data collection is often restricted to salmon-bearing streams.
Beaver data from the project area are limited and focus entirely on the South Slough Subsystem. It should be noted that the reliability of beaver abundance data is limited by methodological shortcomings. For example, in surveying beaver populations, the presence of beaver activity (e.g., ponds, dams, trails, chews, etc.) is commonly used as a proxy for beaver abundance. However, research shows that these indicators may be poor substitutes for population information because the number of beavers present may not be correlated with the number of lodges, bank dens, or beaver dams in any one area (Swafford et al. 2003). Camera trapping as a beaver surveying method may hold promise, but for Cramer (2010) the camera’s motion sensor was triggered by wind-blown vegetation and not by the presence of beavers.
Large Mammals: Estimating the population of deer, elk, bears, and cougars is difficult due to these animals’ secretive life histories and their densely vegetated habitats (ODFW 2006b, 2008). Love (pers. comm., April 29, 2015) explains that aerial survey methods rely on the visual detection of animals to produce an estimate of herd composition, a method with limitations (e.g., weather and dense cover) that can reduce the effectiveness of the survey effort. He adds that alternative survey methods (e.g., true population counts using ground survey techniques) generally produce less reliable results because the presence of surveyors in the field can influence the behavior of the animals being surveyed.
Bear survey methods involve deploying bacon baits containing a tetracycline biomarker used to generate mark-recapture data (see Background section in the Large Mammals data summary). Since this survey takes place during spring when bears are very active, the baits must be placed at least 5 linear miles apart to ensure that a single bear does not consume multiple baits in any one year.
As a result, the geographic extent of a bear survey area must be expansive to ensure enough bears are marked to produce reliable population estimates with sufficiently small confidence intervals (S. Love, pers. comm., April 29, 2015). For this reason, bear population data are not available for relatively small areas (e.g., the lower Coos watershed).
In addition to spatial limitations, it should be noted that bear survey data rely on the successful harvest of a bear and the removal and processing of one of its pre-molar teeth in a laboratory. Since these processes can take a considerable amount of time, bear population estimates are considered to be a “lagging indicator,” meaning that the current population estimate is designed to indicate trends occurring two years prior (S. Love, pers. comm., April 29, 2015).
Bear surveys also rely on the telltale marks which bears leave on trees to verify that a bear (rather than some other animal) has consumed the tetracycline bait. Although this method is fairly accurate, it’s not 100% reliable since gray foxes (Urocyon cinereoargenteus), which leave similar scratch marks on trees, have been documented eating tetracycline baits intended for bears.
According to ODFW (2006b), generating an estimate of cougar abundance is “not an exact science.” These estimates often rely on computer-generated models whose accuracy relies on reviews of existing literature. Model-generated estimates are typically presented as ranges between two numbers determined by the data’s confidence intervals. For more information about the confidence intervals of statewide cougar estimates, see the Oregon Cougar Management Plan (ODFW 2006b).
Finally, it should be noted that use of alternative metrics (e.g., annual patterns in harvest data, damage reports, etc.) as a proxy for total large mammal populations is subject to a variety of limitations, due to several extraneous variables. For example, variation in harvest rates may be related to food availability rather than true abundance trends (Fieberg et al. 2010, Howe et al. 2010). Similarly, data related to non-hunting conflicts may reflect changes in landscape characteristics, land use, or regulations rather than actual population trends (Merkle et al. 2011, Howe et al. 2010). Harvest data alone are subject to substantial variation related to hunter effort and hunter success, two variables that are at least partially independent of game abundance (e.g., precipitous reductions in the success rate of Oregon bear and cougar hunters in 1994 following the prohibition of dogs and bait as hunting aids).