Measures of Effort
From the outset, the greatest challenge facing the BBA has been to achieve a relatively
uniform level of geographical coverage in a timely manner with a limited pool of
skilled, dedicated observers. The inability to meet this goal at the initial statewide
scale was the principal motivating factor in the decision taken in 1994 to bring
a small, mostly Seattle-based nucleus of committed observers to bear on a manageable,
county-by-county project. By reducing the scale we would have a reasonable chance
of surveying every one of the atlas blocks in the chosen counties, but only if we
also took into account the mountain of data collected in the earlier phase (1987–1993).
King County
The effort began with King County, which had already received substantial coverage.
However, there was a great imbalance between the lowlands of the western half of
the county—the most densely populated region of Washington—and the mountainous,
largely inaccessible eastern half (see Figure 1).
 Figure 1 — King County Party Hours 1987–1993
Whereas 40 percent (n=55) of the atlas blocks in the western
half had received more than five hours of survey effort, only 14 percent (n=19)
of those in the eastern half had benefited from a comparable effort. In the western
half, 27 percent of blocks (n=37) had never been visited;
in the eastern half this figure was 69 percent (n=94). Therefore,
our first objective was to concentrate on those blocks that had been the least well
surveyed. The results can be seen by comparing Figures 1 and 2, showing the geographical
distribution of party hours per atlas block 1987–1993 and 1994–2000. Figure 3 shows the party hours aggregated over the entire 14-year period.
 Figure 2 — King County Party Hours 1994–2000
 Figure 3 — King County Party Hours 1987–2000
Kittitas County
At the end of the 1993 atlas season the situation across the Cascades in Kittitas
County was less promising. Seventy percent of the county's 283 atlas blocks had
received 1.5 party hours, or less, of observer effort, while the minority of blocks
that had been better surveyed were largely concentrated along the Yakima River,
the Kittitas Valley, and main roads (see Figure 4). Most blocks in the shrub-steppe zone and in higher-elevation forests remained untouched.
 Figure 4 — Kittitas County Party Hours 1987–1993
Once again, survey hours during the second atlas phase were directed toward the
inadequately surveyed blocks (Figure 5), resulting in relatively uniform coverage
on an hours-per-block basis when aggregated over the whole 14-year period (Figure
6).
 Figure 5 — Kittitas County Party Hours 1994–2000
 Figure 6 — Kittitas County Party Hours 1987–2000
Kitsap County
From 1987 through 1993 Kitsap County received 538 party hours of atlas surveying,
for an impressive average of 8.4 hours per atlas block on a county-wide basis.
However, a closer look reveals that 514 of those party hours were devoted to just
11 atlas blocks, each of which received an average of nearly 47 hours of effort.
Fifty of the county's remaining 53 atlas blocks had never been visited. Figures
7, 8, and 9 show how party hours in the second phase were targeted to compensate
for the imbalance of effort in the first phase.
 Figure 7 — Kitsap County Party Hours 1987–1993
 Figure 8 — Kitsap County Party Hours 1994–2000
 Figure 9 — Kitsap County Party Hours 1994–2000
Island County
Island County was scantily covered through 1993, with 70 party hours of observer
effort (see Figure 10).
 Figure 10 — Island County Party Hours 1987–1993
At 208 square miles, Island represents just four percent of the total four-county
land area. It was a late addition to the second-phase project, after the rest of
the counties had been finished. Fortunately we were able to rely more heavily on
local residents as skilled observers in Island County than we were in the other
three counties. For all of these reasons it was decided to survey every block as
if no earlier work had been done (see Figure 11), even though the first-phase records
were also included in the final database (Figure 12).
 Figure 11 — Island County Party Hours 1994–2002
 Figure 12 — Island County Party Hours 1987–2002
When is a Block “Done”?
The fact that a certain species is found to be breeding on some atlas blocks but
not on others has no meaning unless controls have been put in place to ensure some
degree of uniformity of survey effort. This was a particularly critical concern
for the present project because the limited number of available field observers
had to be deployed with maximum efficiency. After some experimentation, we settled
on three standards of measurement to aid in determining whether a block had received
adequate attention or whether it needed more work: number of species detected, percentage
of species detected at higher evidence levels (PRobable and COnfirmed), and number
of party hours of survey effort. As fieldwork progressed a scheme was developed
to monitor these factors so that observers could be dispatched to any atlas block
judged deficient in coverage.
All else being equal, the raw number of species detected is not in itself
a useful measure of survey effectiveness, since some atlas blocks naturally support
a greater biodiversity than others. We therefore attempted to devise a criterion
based on the species that “ought to” be present on a particular block.
For King and Kittitas Counties we compiled a block-by-block list of expected species
derived from the predicted distributions of the Washington Gap report (Smith et
al. 1997). The minimum target for a “passing” score was set at 50 percent
of the expected number of species. For Kitsap and Island Counties, which are much
more uniform in habitat, we went instead with a list of the 20-plus species most
commonly detected in each county after one full year of second-phase surveying,
establishing a threshold of 90 percent of these species for a “passing”
score.
The percentage of species detected at higher evidence levels is a straightforward
calculation: {(number of species PR + CO) / (total number of species)}
× 100. The “passing” score was set at 50 percent.
Of the three measures, the number of party hours of survey effort is certainly
the simplest to tabulate. Five party hours was considered to be a “passing”
score.
The three criteria correlate closely. Hours of survey effort drives the other two,
yet the latter add unique value as interpretive tools. An informal sampling of our
data shows the curve of species detected climbing steeply for about the first four
or five party hours on a block, and leveling significantly by hour six or seven.
The number of detections at higher levels of breeding evidence, however, continues
to climb at about the same rate for nine or more hours before starting to level
off. It could be said that the number of species detected is principally a spatial
measure, reflecting quality of coverage of all the habitats across the map of a
block. The proportion of higher evidence-level detections, on the other hand, is
principally a temporal measure, reflecting quality of coverage across the breeding-season
calendar. Repeated visits are necessary in order to intersect with the differing
breeding phenologies of the birds inhabiting a block.
In deciding which blocks to declare “done” and which would need further
work, all three measures were taken into account in a rolling review process. A
poor score in just one category, or a marginally sub-par score in any two, virtually
guaranteed that a team would be sent to a block for another look. However, ease
of access was also considered, as was uniformity of habitat. A block of flat, open
agricultural land amply served by roads can be better covered in a couple of hours
than can a remote, roadless, high-elevation block in as many days. Consideration
of how best to allocate scarce survey resources entered into all of our determinations,
resulting inevitably in tradeoffs.
We could not hope to bring all blocks up to the level of the initial project target
of 20 or 30 hours of effort (Mattocks 1988). Instead, our more modest goal was to
bring essentially every block up to a uniform minimum survey standard so as to ensure
a greater comparability of results across the four-county map. Figure 13 illustrates
the degree to which this goal has been successfully met.
 Figure 13 — Total Party Hours (All Years)
|