Christmas Bird Count Mind Numbing Statistics |
|
The Columbia Estuary Christmas bird census was started
in 1979 by David Irons who compiled the count from 1979 to 1985. John Biewener
took over compilation duties for 1986 and 1987. Mike Patterson has been
compiler from 1988 to the present.
Cummulative Average | 116.4 | SD | 9.8 |
Average 1979-1985 | 111.5 | SD | 6.2 |
Average 1986-1987 | 99.5 | ||
Average 1988-2000 | 121.5 | SD | 7.4 |
It's a safe bet that the differences in total species between the three
compilers reflects the advantages of having a compiler/coordinator who
lives in the count circle. Prep work, pre-planning and contacting other
residents is certainly easier for a compiler when all the calls are local.
Comprehensive species total |
|
Species threshold (seen every year) |
|
OBRC Species |
|
In reviewing the data for the Columbia Estuary Count, I removed three records because I could find no details. Black Tern seen in 1979 is undocumented (I'd like to be able to eliminate Fork-tailed Storm-petrel as a possibility). House Wren also submitted in 1979 was not even placed in bold print by the compiler (I am generally disinclined to accept records of House Wren without meticulous documentation.) The Black-chinned Hummingbird seen in 1983 was probably identified correctly, but I am uncomfortable with accepting a record based on a 20 year old memory.
Trends among raptor species. Most notable is the graph documenting the
recovery of Peregrine Falcon and Bald Eagle along the Columbia River and
the rise and fall of White-tailed Kite which I suspect can be linked to
the drought that occurred in the 1980's.
Graph 4. One of the more conspicuous eruptive groups are the seedeaters. Most notable in this group are the Pine Siskins (Pine Swine). The cyclical nature of siskin eruptions shows up well in graph 4a. It is closely correlated to eruptions (in a smaller scale) of other seed dependent species. The only Pine Grosbeak record West of the Cascades occurred during one of the eruptive episodes
Things that make you go, "hmmm."
For any Christmas count, one can expect a minimum threshold
of species that are 100% certain. Beyond this, species counts become less
reliable. Each year species are missed that should have been see. The graph
on the left compares total species that have occurred to years seen. The
theorectical average for a count should be the point at which there is
an even chance of getting or missing a given species on a count. For the
Columbia Estuary Count, the species average is 121 +/- 7 which falls remarkably
close to the perdicted. value.
The probability function fixes the ends of the line generated by the data. It would therefore be expected that as the sample size (number of years the count is held) increases the line would become increasingly curved. This, of course also assumes that there are a finite number of possible species that can be seen on any given count. |
Linear Regression- A mathematical process that averages data points on a graph to determine the best fit for a line drawn through those points.
Coefficient of Correlation (cor)- A mathematical test based on the size of the sample that allows one to judge the reliability of a line created using linear regression. By rule of thumb, the closer this value is to one (1), the more reliable the fit of the line. Larger sample sizes may show a smaller r value and still be judged reliable.
Slope (m)- The ratio of the change in the vertical component to the change in the horizontal of a straight line on a graph. A positive value indicates an increasing trend in the values on the graph when read from left to right. A negative value indicates a decreasing trend.