Estimation of Moose Abundance using the Geospatial Population Estimator combined with a Sightability Model on Togiak National Wildlife Refuge products

Moose (Alces alces gigas) have recently expanded their range into the boundaries of Togiak National Wildlife Refuge and BLM Goodnews Block (TGK; Aderman et al. 2011). Fewer than 30 moose were recorded in Game Management Unit (GMU) 17A from 1980-1990, however, 84 moose were recorded in 1994 and the total count of moose has since increased to a total of 1,166 individuals counted in 2011 (Aderman et al. 2011). Continued monitoring of moose is essential for developing regulatory proposals and recommendations to the Federal Subsistence Board/Alaska Board of Game (Aderman et al. 2011). Additionally, monitoring moose abundance on Togiak is important to fulfil two of the purposes for which Togiak Refuge was created; restoring large mammal populations and providing opportunities for continued subsistence uses by local residents (Aderman et al 2011).

The abundance of moose has been monitored on TGK from 1995-2011 using a census, or complete count, and Gasaway style sample units (SU; Gasaway et al. 1986), which varied in size because they were delineated by topographic features (Fig. 1). Total counts, or censuses, are based on the assumption that all individuals are counted. Although this assumption is unlikely to be met when wildlife populations are surveyed from aircraft, the TGK moose census were assumed to have consistent bias through time, because moose are highly visible in open habitats when snow conditions are adequate. However, few moose surveys were conducted during the 4-year period from 2012-2015, primarily because lack of snow made it difficult to detect moose. Poor snow conditions are known to lower the sightability (the percentage of moose seen in the study area), yet, for most regions of Alaska, the variation in moose sightability during suboptimal conditions has not yet been quantified. Because climate change is affecting snow cover (Park et al. 2012), and it is unclear when there will be adequate snowfall to conduct moose surveys with consistently high sightability on TGK, biologists and statisticians concluded that an alternative approach to monitoring moose on TGK was necessary.

There are two types of visibility bias that occur during aerial moose surveys: 1) perception bias occurs when moose are not detected because of environmental conditions such as fatigue, lighting, snow cover, etc. (Marsh and Sinclair 1989) and 2.) Availability bias occurs when the animal is not visible and not available to be detected (e.g., hidden by vegetation or tree wells; Marsh and Sinclair 1989). To correct for the bias known to occur during aerial surveys for estimating moose abundance, the Alaska Department of Fish and Game (ADFG) has estimated sightability (Caughly 1974 in Gasaway et al. 1986). Gasaway et al. (1986) partitioned sightability into two components: the observed SCFO (observed by surveying moose at a standard search 4-6 minutes/mile2 and immediately resurveying the same area at a speed of 12 minutes/mile2) and constant SCFc (a constant value determined from radio-collared moose. Gasaway et al. (1986) referred to a constant sightability correction factor (SCFc), which is a value determined from radio-collared moose, to estimate sightability for a survey. The SCFc can correct for both detection bias and availability bias, whereas the SCFO only quantifies perception bias.

In addition to a SCFc for a single survey, sightability models will be useful to predict sightability on future surveys using covariates known to affect sightability of moose (habitat, snow cover, and search time) similar to models developed by Christ (2011).