76
cover-type maps, or local vegetation coverage, can serve as a guide for determining priority of survey
efforts. We would caution that, early in the process, deciding a priori that an area cannot support lynx
without detailed local knowledge, is not without risk. If grids are designed as a representative sample of
a particular cover type, sampling within this cover type provides no information about any other cover
type. There will always be this trade-off: tight stratification rules will presumably increase the efficacy,
but they limit the inference. Additionally, this grid-based approach works best and is most efficient in
areas where cover types are reasonably contiguous.
Exactly how you resolve the relative importance of these two properties: efficacy vs inference, and hence
how you stratify your landscape prior to sampling, will largely be a function of local knowledge,
priorities, and vegetation patterns. Two examples may provide insight into this process. In the Superior
NF a question of primary importance is: Do we have any resident populations of lynx? To answer this
question, the Superior will be looking to place grids in those areas where they have the most recent
evidence of lynx occurrence and where the habitat appears to be most suitable. In the Okanogan NF, the
presence of lynx in the area studied by Koehler and Brittell is not in question. The Okanogan, therefore
is placing grids in areas where they have some information that lynx occur, and would like to gather
more data concerning these lynx. They are not, therefore, necessarily placing the surveys in the "best"
areas, as is the Superior, but they are still only surveying in-and-adjacent-to cool wet forest types.
In all cases, we recommend avoiding multiple fine-scale stratification rules, and particularly rules not
supported by scientific data. For instance, there is no evidence, particularly in the summer, that lynx use
specific topographic features preferentially. Lynx telemetry locations are not found adjacent to creeks or
on flatter topography more than expectation within the study areas. Employing such rules in landscape
stratification radically reduces the ability to infer the sample to the landscape (because so much of the
landscape will be outside of the strata) without any direct evidence that the rules will increase the
sampling efficacy.
Working in conjunction with other survey efforts
Hair snagging can be used to compliment other survey methods, such as snow tracking. For instance, if
snow tracks were found in an area, particularly where lynx were thought to be absent, then placing a
grid across the area would potentially validate the snow tracks. If individual DNA identification was
performed on the samples, the addition of a hair survey could provide information concerning the
number of lynx in the area. Using snow tracking as a pre-sampling method to determine grid
placement in no way invalidates the protocol. In many areas this is a very sensible approach.
Non representative placement of transects, or even individual scent stations can sometimes provide
useful information. For instance, if a lynx is known to exist within a specific drainage, one might want to
specifically sample the drainage to try to determine whether the lynx is still present, or to obtain a
