Does the Alaska Game Board Really Care?

There is something similar going on now with the wolf control proposals the Board of Game (BOG) will consider making permanent at its March 2006 meeting. The BOG is trying to build wolf control programs ("houses") without using valid censuses ("tape measures") to estimate moose numbers in each of five different areas covering 50,000 square miles of Alaska. Each area has a moose population estimate provided by the Alaska Department of Fish and Game (ADFG) but only two of the five areas have estimates based in part on censuses. In the other three areas moose numbers were estimated without using a "tape measure." Instead, in the areas lacking censuses, surveys designed to gather moose sex and age ratio data (used for setting hunting regulations) provided the basis for the population estimates.

The problem is that surveys cannot produce accurate estimates of moose numbers. Moose biologists have known this for decades. That is why ADFG spent thousands of dollars and lots of effort starting in the 1970s to develop aerial moose census techniques. They were successful and today there are procedures in place that, under good environmental conditions, using good pilots and experienced observers, result in scientifically valid, accurate estimates of moose numbers.

Why does the BOG need accurate estimates of moose numbers? In order to conduct a wolf control program under the intensive management law the BOG must first find that current moose numbers are below moose population objectives set previously by the BOG. If numbers are at or above objectives, wolf control need not occur. Obviously, the only way to determine where a moose population is in relation to objectives is to census the population. But current censuses are not available in most areas. Rather than postpone wolf control until moose censuses are available, the BOG has elected to build houses without using tape measures by using moose population estimates based on surveys.

The best example of serious problems that occur when good moose census information is lacking comes from the McGrath area. In 2000 a planning team was appointed to examine the issues at McGrath after local residents claimed there were too few moose and requested a wolf control program. ADFG biologists told the team there were about 850 moose in the area but about 3,500 were required to meet local subsistence demand for moose. A good moose census had not been done. The team recommended reducing bears and wolves to re-build the moose population and the BOG approved the plan. In 2001 a good moose census in the area resulted in an estimate of 3,600 moose. Had the census not been done, bears and wolves would have been reduced needlessly. A census revealed four times the number of moose that were thought to be there a year earlier. In fact, enough moose were present for subsistence use the whole time.

Wolf control advocates now claim that we can't wait for moose census data--that moose are so scarce in most areas that wolf control must continue and we can get by with population estimates based on surveys. And they argue that censuses are too expensive. Well, houses can surely be built without using tape measures but builders should know full well that they likely won't pass inspection. In order to meet building code specifications house builders simply must invest in tape measures.

But who will inspect the wolf control programs to ensure that they are up to code? Surely not the BOG. I listened to the entire eight-hour teleconference held by the BOG in January when it approved the emergency regulations that preceded the current proposals. Not once did I hear a BOG member question the basis for any of the data presented by ADFG. Nor did I hear ADFG biologists caution the BOG that moose population estimates based on surveys may be very inaccurate. On that day, the building inspectors were absent, and I'd bet they won't suddenly appear at the BOG meeting in March.
I am including here in edited form some of the comments I recently sent to the BOG on the wolf control proposals. This gets pretty technical, especially if you don't have the proposals handy. But if you are interested in specific problems with the moose population estimates, you will find some of them identified here.

COMMENTS ON PROPOSALS 162-166-MARCH 2006 BOARD MEETING

For each of the 5 proposed implementation plans contained in proposals 162-166 a moose population estimate is provided under part (3)(A)(i) or (3)(A)(ii). These estimates are then often repeated in subsequent parts of each proposal. The estimates are at times based on aerial moose population censuses done at some point in the past. Techniques for these censuses were developed by ADFG starting in the 1970s. When done under good snow, light, and wind conditions with experienced observers they can result in valid, accurate moose population estimates. At other times, when censuses have not been done, moose population estimates are derived from aerial moose sex and age (herd composition) surveys .

The population estimates are often given in very precise terms in the proposals, e.g., 3,959 moose (0.46 moose/mi2; range 2,460-5,494 moose) in Unit 19D-East in 2001 and 4,374 moose (0.5 moose/mi2; range 3,444-5,281moose) in the same area in 2004. In this case, the estimates are based on censuses that allowed the calculation of confidence limits around the mean-these are the upper and lower ranges provided. Typically, the censuses occurred in only a portion of the total land area and were extrapolated to the entire area. For Unit 19D-East, the 2001 census occurred in 5,204 square miles and was extrapolated to an additional 3,309 square miles (where no census was done) to produce an estimate for the entire 8,513 square mile area.

In those cases where the estimates were based on surveys rather than censuses, estimates were given as either finite numbers with no range, e.g., Unit 13, or a range was given with no explanation of how it was derived, e.g., Unit 16B.

Accurate moose population estimates are required in order for the Board to implement intensive management and to adopt predator control programs. These estimates are required to assess whether or not moose population objectives have been met, a preliminary step in the intensive management process. It is therefore important to assess the reliability of the estimates and the quality of the data used to derive them.

It is well documented in the technical literature on moose by studies conducted by ADFG and others that survey data cannot be used to reliably and accurately estimate population size. This is because moose herd composition surveys do not employ stratified random sampling nor do they allow calculation of sightability correction factors for moose not observed. Such surveys typically employ limited search times and low searching intensities per unit of habitat, "high grade" the best habitats, and may occur when survey conditions are marginal. They simply cannot be used to estimate moose numbers because they lack a way to estimate precision. Therefore, the moose population "estimates" based on surveys given in proposals 165 (Unit 13) and 166(Unit 16B) are inadequate to assess current moose numbers in relation to intensive management moose population objectives, to calculate moose:predator ratios, to calculate sustained yields, or for any other calculations using moose numbers in these proposals.

For proposals 162 and 164 (Units 19A and 19D-East) moose population estimates are based on aerial censuses done in varying portions of the areas, but even here there are problems in assessing the accuracy of the estimates and the quality of the underlying data. For example, in proposal 164 (Unit 19D-East) I noted above the extrapolation of the 2001 data to an area of 3,309 square miles (38.9% of the total area) that was not censused. This is not scientifically valid. Because no flying was done in this area it is impossible to conclude that the average moose density there was the same as that in the censused area, a necessary assumption to do the extrapolation. The average density in the uncensused area was simply unknown. If it was lower or higher than in the censused area, the population estimate for the entire are may be much too low or much too high. The only valid procedure is to admit that data are lacking for the 3,309 square mile area and that a moose population estimate for the entire 8,513 square miles (the whole of Unit 19D-East) is not available. As a result, there is no valid way to determine whether or not the intensive management moose population objective has been reached.

There are additional problems with moose population estimates in proposal 164 (Unit 19D-East). The 2004 census cited in the proposal was summarized in an ADFG memo. This memo clearly stated that survey conditions (snow cover) deteriorated during the census. As a result the census was terminated before completion and portions of the area censused in 2001 were not flown in 2004. The memo warns that extrapolating the data to the entire area is not warranted, yet proposal 164 cites a moose population estimate of 4,374 for 2004 with no indication that this estimate is of low reliability and questionable accuracy. Later in proposal 164, in part (4)(A), the 2004 estimated moose population range (3,444-5,281) is repeated and compared with the intensive management objective (6,000-8,000 moose), again with no warnings regarding data quality or problems with extrapolation. In part (5)(A) of the proposal, the 2004 range of moose numbers is repeated and it is concluded that the intensive management moose population objective has not been met but the decline in the overall moose population in Unit 19D-East has been stopped. These conclusions are not justified and are not scientifically valid. I emphasize that the 2004 moose census data in Unit 19D-East do not allow an accurate moose population estimate in the 5,204 square mile area censused in 2001, and certainly must not be extrapolated to the entirety of the Unit. Lacking this estimate, the BOG cannot validly conclude that the intensive management moose population objective has not been met.

Even if the problems cited above did not exist, there are other problems with the Unit 19D-East moose population estimates. The moose density estimate for the 2001 census cited in proposal 164 part (3)(A)(ii) of 0.46 moose per square mile differs from the density presented to the BOG in prior years. Previously (November 2001), the density presented to the BOG was 0.43 moose per square mile. This is not merely a typographical error as the total number of moose for all of Unit 19D-East was estimated using the 0.46 statistic. This seemingly small difference (0.43 vs. 0.46) equates to a difference of 256 moose when the extrapolation procedure used in proposal 164 is employed (3661 vs. 3916 moose in all of Unit 19D-East). Given the precision of the estimates this difference seems trivial but it is not trivial from a scientific standpoint. Why has the density (and the total number of moose) changed? Was it an error or were the census data re-analyzed? If the latter, was a new, corrected memo issued? If so, it was not distributed to interested parties. If the data were re-analyzed and a new density was calculated, this raises serious scientific concerns about the reliability of data presented by ADFG, not only for Unit 19D-East but for other units as well.

The problems cited above of improperly extrapolating moose census data in proposal 164 (Unit 19D-East) also apply to proposal 162 (Unit 19A) where aerial moose censuses were conducted in small portions of the total land area. Density estimates were calculated, and these were then extrapolated to the entire area and converted to estimated moose population sizes. In proposal 162 this procedure is described in Part(3)(A)(i). Clearly, the area actually censused in Unit 19A was small in relation to the entire unit and it is not scientifically valid to extrapolate moose density to areas of the unit that were not censused. Furthermore, the areas censused in 2001 and 2005 were different-these were given as "a portion of the Aniak drainage" in March 2001 and "the portion of the unit south of the Kuskokwim River" in February 2005. The conclusion that moose numbers declined between 2001 and 2005 is therefore not scientifically valid. The conclusion in proposal 162 Part(3)(A)(xi) that the moose population in Unit 19A is "well below" the Iintensive management population objective is also unwarranted because an accurate estimate of the current moose population is not available.

Moose population estimates in Units 12 and 20E are presented in proposal 163 Part (3)(A)(i). These are "based on extrapolations from surveys" The surveys were done in 4,630 square miles of 6,600 square miles comprising the whole predator control area. This means that 1,970 square miles (29.8%) were not surveyed. Because no data were available indicating that moose densities were similar in the surveyed and unsurveyed areas it is not scientifically valid to extrapolate densities from one to the other. The conclusion that "The moose population in Units 12 and 20E is well below the intensive management objective" (Part (3)(A)(vii)) is unwarranted because no valid estimate of moose numbers is available for the 6,600 square mile area covered by the implementation plan.