Je veux savoir à propos de…

43018

Federal Register / Vol. 86, No. 148 / Thursday, August 5, 2021 / Rules and Regulations
Harassment Rate The Level B harassment rate or the probability that an encountered bear will experience either incidental or intentional Level B harassment, was calculated using the 20142018 dataset from the LOA database. A binary logistic regression of harassment regressed upon distance to shore was not significant p = 0.65, supporting the use of a single harassment rate for both the coastal and inland zones. However, a binary logistic regression of harassment regressed upon day of the year was significant. This significance held when encounters were binned into
quantile of each probability distribution can be interpreted as the upper limit of the potential harassment rate supported by our dataset i.e., there is a 99 percent chance that given the data the harassment rate is lower than this value. We chose to use 99 percent quantiles of the probability distributions to account for any negative bias that has been introduced into the dataset through unobserved harassment or variability in the interpretation of polar bear behavioral reactions by multiple observers. The final harassment rates were 0.19 during the open-water season and 0.37 during the ice season Figure 5.

either ice or open-water seasons p<0.0015.
We subsequently estimated the harassment rate for each season with a Bayesian probit regression with season as a fixed effect Hooten and Hefley 2019. Model parameters were estimated using 10,000 iterations of a Markov chain Monte Carlo algorithm composed of Gibbs updates implemented in R R
core team 2021, Hooten and Hefley 2019. We used Normal 0,1 priors, which are uninformative on the prior predictive scale Hobbs and Hooten 2015, to generate the distribution of open-water and ice-season marginal posterior predictive probabilities of harassment. The upper 99 percent
season

w

20 +l-----1----:

"C

0.0

0.1
0.2
0.3
0.4
P level B I encounter

ice openwater
0.5

Impact Area As noted above, we have calculated encounter rates depending on the distance from shore and season and take rates depending on season. To properly assess the area of potential impact from the project activities, we must calculate the area affected by project activities to such a degree that harassment is possible. This is sometimes referred to as a zone or area of influence.
Behavioral response rates of polar bears to disturbances are highly variable, and data to support the relationship between distance to bears and disturbance is limited. Dyck and Baydack 2004 found sex-based differences in the frequencies of vigilant bouts of polar bears in the
VerDate Sep<11>2014

17:26 Aug 04, 2021

Jkt 253001

presence of vehicles on the tundra.
However, in their summary of polar bear behavioral response to ice-breaking vessels in the Chukchi Sea, Smultea et al. 2016 found no difference between reactions of males, females with cubs, or females without cubs. During the Services coastal aerial surveys, 99
percent of polar bears that responded in a way that indicated possible Level B
harassment polar bears that were running when detected or began to run or swim in response to the aircraft did so within 1.6 km 1 mi, as measured from the ninetieth percentile horizontal detection distance from the flight line.
Similarly, Andersen and Aars 2008
found that female polar bears with cubs
PO 00000

Frm 00038

Fmt 4701

Sfmt 4700

the most conservative group observed began to walk or run away from approaching snowmobiles at a mean distance of 1,534 m 0.95 mi. Thus, while future research into the reaction of polar bears to anthropogenic disturbance may indicate a different zone of potential impact is appropriate, the current literature supports the use of a 1.6 km 1.0 mi impact area, as it will encompass the vast majority of polar bear harassment events.
Correction Factor While the locations that were used to calculate encounter rates are thought to have constant human occupancy, it is possible that bears may be in the
E:FRFM05AUR2.SGM

05AUR2

ER05AU21.005

khammond on DSKJM1Z7X2PROD with RULES2

Figure 5-Estimated marginal posterior predictive probabilities from the Bayesian pro bit regression of Level B harassment of polar bears on the North Slope of Alaska in the period 2014-2018. Vertical grey lines correspond to the upper 99% quantiles for each distribution, which were used as the estimates of harassment rates.