Post by kl9 on Oct 11, 2017 8:19:14 GMT -6
I sent an email to the DNR asking for an explanation on the survey or lack thereof for the WMA acquisition (in the sense it was one-sided) as well as the large margin of error in the wolf estimate. This is the response I got for the latter. Will post the response for the former when I get it.
Subject: RE: MN DNR Deer Email Updates
Hi Kaleb:
Regarding your question #2, below is an answer I provided to a reporter with the same question. It's a bit long-winded, but hopefully it explains the answer effectively. If you have any additional questions, let me know.
John
John Erb
Furbearer/Wolf Research Scientist
Minnesota DNR
Forest Wildlife Populations and Research Group
1201 E. Hwy 2
Grand Rapids, MN 55744
218-328-8875
"The margin of error in this case is driven by variability in the pack and territory sizes for the packs we had radio-collared. Since we don’t have access to data on packs we didn’t radio-collar, and can’t afford to collar all/most packs, we try our best to sample a ‘reasonable’ number, and cover the range of habitats/conditions in which wolves occur. But even if the ones we collar are ‘reasonably’ representative, there is still inter-pack (biological) variability as well as sampling variability stemming from the random element to what wolf packs we in fact caught (sampled). So we use the data we do have in a method called bootstrapping to account for the potential that had we by chance caught and collared different packs in our sampling, we might have gotten a different combination of packs with different averages. We randomly ‘re-sample’ (with replacement) the data (packs) we do have 10,000 times, each time getting an estimate of average pack and territory size, and calculating a population estimate each time. Some estimates, by chance, will be based on our packs that were mostly on the smaller end of pack size (and/or the larger end of territory size) in our sample, and therefore lead to lower population estimates. And vice versa. The ‘hundreds’ margin of error is just what it turned out to yield based on the variability we see in our pack/territory data. If all the packs we monitored were 'identical' there would essentially be no margin of error. The more variability in the packs, the higher the margin of error. The margin of error of ~ +/- 500 has been fairly similar through time. There will always be among-pack variability, and we could likely reduce the margin of error some by collaring more packs, but as with everything there are cost-benefit considerations".
Subject: RE: MN DNR Deer Email Updates
Hi Kaleb:
Regarding your question #2, below is an answer I provided to a reporter with the same question. It's a bit long-winded, but hopefully it explains the answer effectively. If you have any additional questions, let me know.
John
John Erb
Furbearer/Wolf Research Scientist
Minnesota DNR
Forest Wildlife Populations and Research Group
1201 E. Hwy 2
Grand Rapids, MN 55744
218-328-8875
"The margin of error in this case is driven by variability in the pack and territory sizes for the packs we had radio-collared. Since we don’t have access to data on packs we didn’t radio-collar, and can’t afford to collar all/most packs, we try our best to sample a ‘reasonable’ number, and cover the range of habitats/conditions in which wolves occur. But even if the ones we collar are ‘reasonably’ representative, there is still inter-pack (biological) variability as well as sampling variability stemming from the random element to what wolf packs we in fact caught (sampled). So we use the data we do have in a method called bootstrapping to account for the potential that had we by chance caught and collared different packs in our sampling, we might have gotten a different combination of packs with different averages. We randomly ‘re-sample’ (with replacement) the data (packs) we do have 10,000 times, each time getting an estimate of average pack and territory size, and calculating a population estimate each time. Some estimates, by chance, will be based on our packs that were mostly on the smaller end of pack size (and/or the larger end of territory size) in our sample, and therefore lead to lower population estimates. And vice versa. The ‘hundreds’ margin of error is just what it turned out to yield based on the variability we see in our pack/territory data. If all the packs we monitored were 'identical' there would essentially be no margin of error. The more variability in the packs, the higher the margin of error. The margin of error of ~ +/- 500 has been fairly similar through time. There will always be among-pack variability, and we could likely reduce the margin of error some by collaring more packs, but as with everything there are cost-benefit considerations".