The temporal and/or geographic scopes of the data are not comprehensive. As mentioned, CBC data gaps occur in years during which the CBC was not conducted (2010 and 2014). An additional data gap occurs in 1987-1989 when local CBC data were not reported to the Audubon database (T. Rodenkirk, pers. comm., 2015).
In addition, CBC data collection is focused on areas of suitable habitat that are easily accessible, and may be limited by other factors (e.g., volunteer time constraints, volunteer training).
The BBS is conducted along 144 bird observation routes across Oregon. David Ziolkowski (pers. comm., November 24, 2014) explains that these routes are randomly selected to create a sufficiently large sample in order to “represent habitats in proportion to their natural occurrence on the landscape.” He adds that multiple studies have shown that this sample design effectively represents bird habitat, but, since “observational” routes occur along roads, open water habitats are slightly underrepresented while developed habitats are marginally overrepresented.
Data from bird surveys are inherently subject to several “co-varying factors” that may be correlated with survey results (e.g., stylistic and skill differences between observers, survey effort, and habitat change). For example, imagine a hypothetical survey for which participation has increased steadily over time, and the raw data show that sightings have also increased. In this case, assuming that the number of sightings may be correlated with effort, it is difficult to determine what portion of the observed trend is due to a true change in bird abundance because some of the observed difference may have resulted from increased effort.
To account for theses co-varying factors, the BBS processes its data using a Bayesian hierarchical model developed by the USGS (2014). The model calculates a relative bird abundance metric. In some cases, the reliability of model calculations is limited by deficiencies in the data (e.g., small sample size, poor precision, etc.)(USGS n.d.).
Figure 2. Christmas Bird Count data for both survey effort (red) and number of species recorded (yellow) is available from 1990-2013. These variables tend to “track” each other well, meaning that years with high effort tend to correspond to years with high avian diversity. Since the raw data have not been corrected for co-varying trends such as survey effort, it’s difficult to determine how much of an observed abundance trend may be due to true underlying patterns. Data: Audubon 2014
No such model has been developed for the CBC. Consequently, only raw CBC data have been reported. The use of raw data presents a series of concerns and limitations. The raw CBC data have not been adjusted for:
Misidentification- Although volunteers are generally very knowledgeable about local bird species, experts speculate that some data may contain bird species misidentifications (particularly in the 1980s). For example, citizen scientists that participated in the CBC prior to 1982 may have misidentified long-billed curlew, mistaking this species for the whimbrel (Rodenkirk 2012).
The abilities of volunteers- For example, song identification is the primary means for detecting the swamp sparrow; this skill requires advanced birding knowledge (Rodenkirk 2012). Therefore, it’s difficult to determine if an increase in swamp sparrow sightings corresponds to a true abundance trend.
Survey effort- Changes in effort may correspond to changes in some abundance metrics (e.g., species diversity)(Figure 2).
Habitat change-Habitat change within the survey area may result in a redistribution of bird species. For example, the natural succession of forest habitat in an undisturbed area may result in the redistribution of species that require open canopies to areas of less dense vegetation (e.g., clear cuts in the uplands), potentially resulting in a lower raw count even though total abundance may remain unchanged.
Climate variables (e.g., El Niño events)- The arrival and departure of migratory birds to the lower Coos watershed is often a function of climate variables (e.g., year-to-year changes in temperature and precipitation). In years with climatic anomalies, the timing of the CBC may not correspond with the presence of every species that winters in the Coos estuary.
In lieu of statistical modeling, local bird experts were consulted to help distinguish between true abundance trends and observed patterns likely caused by other factors (Rodenkirk 2012; T. Rodenkirk, pers. comm., January 19, 2012; R. Namitz, pers. comm., January 19, 2012).
Consultation with local experts also helps identify known inaccuracies in the Audubon (2014) data. For example, Rodenkirk (pers. comm., 2015) explains that CBC volunteers frequently confuse the short billed curlew with the long billed curlew. Although the short billed curlew is a very rare sighting in the lower Coos watershed, the raw Audubon (2014) data report a high number of sightings on a regular basis. These inaccuracies are not corrected by the Audubon Society prior to publication. To avoid including inaccurate data, the trends from Audubon (2014) have been carefully scrutinized and compared against reliable, unpublished data from local experts (T. Rodenkirk, pers. comm., 2015).
The data may also be subject to technological limitations. For example, counts of double-crested cormorant colonies are conducted by low-altitude aerial photo analyses. In some cases, these methods limit experts’ ability to discern between cormorant species (J. Lawonn, pers. comm., 2014). This report only used data from aerial photos that were positively identified as double-crested cormorants.
Both the BBS and the Catalog of Oregon Seabird Colonies use breeding population counts as proxies for abundance. Experts suggest, however, that the use of breeding populations alone may underestimate bird use of estuarine habitats since breeding population data don’t include non-breeding birds foraging in the estuary (Adrean 2013).
Both CBC raw data and BBS relative abundance data are meant to index overall inter-annual trends in bird abundance (i.e., changes in relative abundance from one year to the next). These data should not be interpreted as estimates of the total population size. Due to methodological differences between these surveys, extreme caution should be taken when making direct comparisons between data sources.
Finally, it should be noted that two sources were used in compiling marbled murrelet habitat data. Both data sources were combined to produce a general idea of marbled murrelet nesting habitat abundance in the project vicinity. Federal critical habitat designation is based on computer modeling of habitat suitability and state habitat designation is based on field observations of marbled murrelet nesting behavior (ODF 2014, USFWS 2011).
