|
Author and year |
Focus of study |
Main findings |
| |
|
Faanes and Bystrak 1981 |
Differences in hearing ability and observer expertise |
In general, well-trained observers are comparable in their ability, whereas
observers that are inadequately trained in bird identification provide significantly
different BBS results than do qualified observers |
|
Scott and Ramsey 1981 |
Effects of multiple species on accurate detection |
Observers that recorded fewer species counted more individuals than did observers
that recorded all species in the same area |
|
Bart and Schoultz 1984 |
Effects of multiple individuals on accurate detection |
In bird song simulations, as the number of singing birds increased from one
to four, the proportion of individuals recorded declined by up to 50% |
|
Emlen and DeJong 1992 |
Differences in hearing ability |
Comparisons of audiograms of people’s hearing to spectograms of birdsong
demonstrate that older people have large deficiencies in hearing and thus older
surveyors provide incomplete and biased information to the BBS |
|
Link and Sauer 1994 |
Analytical methods bias and survey design |
The usual method of BBS population trend analysis (i.e., taking the logarithm of
bird counts in a regression analysis) works well for abundant species, but not
for uncommon species; a new estimating equation makes trend estimates
of uncommon species more accurate |
|
Sauer et al. 1994 |
Change in observer skill over time |
Failure to recognize changes in observer skill over time results in
overly optimistic estimates of bird populations |
|
Bart et al. 1995 |
Representativeness of roadside habitat |
There was very low bias in estimates of habitat change (< 1%)
along roads compared to away from roads, although there was significantly less
forest cover near roads than in the surrounding region |
|
Hanowski and Niemi 1995 |
Differential bird incidence along roads |
On average, on-road point counts recorded 2.5 more species and 3.5 more
individuals than did nearby off-road point counts; 20 species were more abundant
on roads, whereas five species were more abundant off-road |
|
Hutto et al. 1995 |
Differential bird incidence along roads |
Very few species were recorded at only on- or off-road point counts, but
the mean species richness at a given point count was significantly greater at
on-road points |
|
Keller and Fuller 1995 |
Differential bird incidence along roads |
On-road point counts recorded more edge species, but not lower numbers of
interior forest species than did off-road point counts; more individuals and species
were recorded at on-road counts because of the higher number of edge species |
|
Rotenberry and Knick 1995 |
Differential bird incidence along roads |
In shrubsteppe and grassland, only one species was differentially abundant
at either on- or off-road points, suggesting that roads do not create as
significant a habitat discontinuity in grassland habitats as in forested
habitats |
|
James et al. 1996 |
Change in observer skill over time |
Nonlinear regressions are ideal for BBS trend analysis because they
require few assumptions about a population curve through time, they produce
population estimates for which the statistical significance can be tested, and they
allow the inclusion of bias covariates in the analysis |
|
Kendall et al. 1996 |
Change in observer skill over time |
Failure to recognize changes in observer skill over time results in
overly optimistic estimates of bird populations |
|
Link and Sauer 1998 |
Analytical methods bias and survey design |
Bias, such as differences in BBS observers, is inevitable in surveys and
must be taken into account to ensure credible results; an effective way to
account for BBS bias is to include it as a covariate in the trend
analysis |
|
Keller and Scallan 1999 |
Representativeness of roadside habitat |
Land cover changes near BBS routes generally also occurred away from
routes, although significantly more urban cover occurred along routes than in
the surrounding landscape in Maryland, but not in Ohio |
|
Bart et al. 2003 |
Analytical methods bias and survey design |
A linear model of population trend analysis that is design based, not
model based, shows very little bias, unlike, at least in some cases, the
estimating equation approach |
|
Bart et al. 2004a |
Analytical methods bias and survey design |
Current bias in the BBS is 0.008%; if the number of routes is increased to
5106 (i.e., by 40%), the bias will be decreased to 0.003% |
|
Bart et al. 2004b |
Analytical methods: reply to Sauer et al. (2004) |
Observer effects can be accounted for before performing the trend
estimation analysis; sometimes it is unnecessary to account for observer effects,
and including these effects may even result in greater bias |
|
Lawler and O’Connor 2004 |
Sampling bias of large-scale environments |
High elevations and arid regions are underrepresented by the BBS, whereas
northeastern deciduous forest is overrepresented; however, when the area of
comparison is narrowed to BBS-defined physiogeographic regions and U.S. states,
the differences are smaller |
|
Sauer et al. 2004 |
Analytical methods: critique of Bart et al. (2003) |
Bart et al. (2003)’s design-based analysis does not control for factors
that influence bird detection such as observer effects; the analysis
consequently incurs significant bias in trend estimation |
|
Bart et al. 2005 |
Analytical methods bias and survey design |
Increasing the number of BBS routes in the Pacific Northwest region of the
U.S. and Canada would increase the number of species covered and decrease bias
|
|
Sauer et al. 2005b |
Analytical methods: critique of Bart et al. (2004a) |
Bart et al.’s (2004a) analysis has three flaws: their view of the uses
of BBS data is overly simplistic, their model incorporates poorly supported
bias estimates and is therefore statistically weak, and their trend analysis
is flawed for several reasons |
|
Francis et al. 2005 |
Analytical methods: reply to Sauer et al. (2005a) |
The authors acknowledge that the BBS should meet multiple objectives, but
they reaffirm that estimating bird population trends is of fundamental
importance; the authors reiterate that efforts to reduce bias, to recognize that
all bias cannot be eliminated, and to increase the number of routes would
positively influence the BBS |
|
Betts et al. 2007 |
Representativeness of roadside habitat |
The roadside land cover sampling bias of the BBS may prevent the detection
of population changes in forest-based bird species |
|
Harris and Haskell results herein |
Representativeness of roadside habitat |
Roadside surveys in Tennessee give a biased representation of land cover in
the region; these biases change over time and distort simulated bird population
trends |