Date:         Thu, 12 Nov 1998 16:08:40 -0500
Reply-To:     Eran Tomer <etomer@EMORY.EDU>
Sender:       Georgia Birders Online <GABO-L@UGA.CC.UGA.EDU>
From:         Eran Tomer <etomer@EMORY.EDU>
Subject:      GA observer-bias figures
Content-Type: TEXT/PLAIN; charset=US-ASCII

Hello GABO !
 
Our mailing list has recently seen much discussion about bird occurence in
the context of Barn Owls, Georgia vs. Florida birds and straying rarities.
Not surprisingly, the subject of observer coverage has appeared in the
spotlight more than once but always in an ambiguous manner: we know that
it exists, but not its magnitude or direction. To help remedy the
situation I would like to offer a quantitative idea of potential observer
bias in Georgia, specifically concerning non-coastal waterfowl. I will
describe the actual (and lengthy !) procedure only in brief but please
feel free to contact me for exact details.
 
A few months ago I conducted a statistical analysis of DeLorme's Georgia
Atlas & Gazetter.  My goal was to sample a random, statistically -
meaningful proportion of the 2380 grid squares covering the state and
document selected man-made and topographical features present therein.
This data allowed me to estimate the proportion of all Georgia grid
squares containing those features with  99% confidence intervals (or the
plus-minus error range which we can be 99% sure contains the true figure).
I tested my analysis and estimates using several methods and they all
worked perfectly.
 
Three of the selected features were lakes, subjectively classified as
small, medium or large. I defined a lake (as opposed to a pond) as a body
of water at least large enough to provide sufficient takeoff room for a
loon, measuring several known loon-hosting (and non-hosting) lakes to
obtain the rough minimum length (V-E-R-Y ROUGHLY 200-300 yards - the atlas
does not have a scale this fine). Small lakes were almost invariably
present in only one square, with other "lake squares" containing portions
of larger lakes.  Of course, a square could contain more than one lake.
 
My results indicate that at least 40% of Georgia grid squares contain a
lake or a part thereof (minimum 99% confidence limit).  Since the size of
a grid square is known, it is possible to calculate with a simple
algorithm that, in Georgia, one can never be more than 84 miles away (as
the crow flies) from a lake large enough for a loon (again, lower
confidence limit = worst-possible-case scenario).  In terms of actual
numbers, between 1 and 2 thousand such lakes exist in Georgia (closer to
the lower end of the figure).
 
Let us, for the sake of demonstration, presume the following far - from -
realistic assumptions:
 
1.  If a waterfowl is present on any lake, it can always be visually
    located
2.  No waterfowl will be found on a lake or pond too small for a loon
3.  Every lake in the state is fully accessible to birders
 
...sounding optimistic ???
 
Far from it.  Even if all three assumptions were fully true the picture is
quite unappetizing:  birders cover only a miniscule fraction of all
available inland waterfowl habitat every year. If we arbitrarily assume
the number of active birdwatchers in Georgia to be around 650 (my 1997
GOS directory lists under 500 members), there are roughly twice as many
lakes as birders. Now, just try to imagine the situation without the
assumptions...!!!
 
Conclusion: our knowledge of inland waterfowl occurence in Georgia (and
elsewhere !) is based on very few sites in the state and extrapolated
to other areas using ecological knowledge of birds and their habitats.  It
should therefore be fairly accurate for large areas (e.g. physiographic
regions, counties) but much less accurate for any one site, which has
some important consequences. This is even more true for landbirds than
waterbirds as terrestrial habitats are more prevalent in the state.
Exceptions are species with strict requirements for habitats which are
rare in Georgia, e.g. very large lakes.
 
And what does this mean for hapless GABirders seeking their most coveted
birds ?
 
- unless a regular good location is known for one's favorite bird, plan on
much searching.  Even common birds are common only on average.
 
- local checklists are invaluable if constructed responsibly.
 
- it would be very worthwhile to explore new locations rather than visit
and re-visit old ones.  Over the past few years many new hotspots were
discovered in the state and keep being discovered even now.
 
- by observing, documenting and counting local birds we obtain a valuable
scale against which to compare new areas, thus minimizing easily-made
false impressions of them and making extrapolation safer too.
 
Good birding !
 
- Eran Tomer
  Atlanta, GA
  etomer@emory.edu