Thursday, 18 June 2015

eBird and other birding data

I approach the comparison of Bird Atlas vs eBird with trepidation.  However here goes with a few points from my perspective, following comments on birding-aus about unexpected species appearing in eBird listings.  What follows starts off there but gets a good but broader than that, covering a number of thoughts I have had about eBird and other birding data systems. 

Unexpected species in an area.

The appearance of strange species in a small area is also evident from time to time in the Atlas.  When I used to refer to that for output (see below for an explanation of why I use the past tense) I found such things as Black Currawong reported from Goulburn NSW.  An obvious mistype for Pied (or Grey) Currawong but it had got into the data..  Ditto Red-winged Parrot in the same area instead of Red-rumped Parrot.  I have made the latter mistake myself.  Accessing the Birdata (ie Atlas) checklist for postcode 2620 as at 1600 hrs on 18 June 2015 presented a list that included:

  • Red-winged Parrot
  • White-browed Treecreeper
  • Yellow-throated Scrubwren
  • Red-browed Pardalote
All of these are impossible in that area.  There were also a bunch of species (eg White-fronted Honeyeater, Beautiful Firetail) that I have never heard of being in postcode 2620, but may have been reported elsewhere in the ACT once or twice.

I suspect that many of the aberrations in eBird people have reported in the birding-aus thread have been overseas birders making “honest mistakes”.  At least they are entering their lists which I would bet many of the casual visitors would never do under the Atlas approach.  I will confess to making such an error on eBird myself – and received a very polite query from a moderator which cleaned the matter up rather quickly.

As well as direct action by the moderators eBird contains a series of queries so that if a bird is unusual in an area (not sure of the definition of that) or the numbers are unusual for the area and season a prompt is activated seeking supporting information.  To strain a metaphor, not only is the driver awake, but auto pilot is also engaged.  (By way of example if I try to enter Banded Stilt for a site on the South Coast of NSW I have to activate the "rarity list" and then get a prompt to add details.) I’m not sure how the aberrations sneak through this system, if they are errors.

Data capture

In the past I put the majority of my sightings into the Atlas through the COG reporting system, or for sites further from home, BirdInfo.  The demise of the latter facility was what led me to use eBird as a data capture tool and I have been very happy entering from dead trees (aka pen and paper) through a laptop.  In the last two months I have started using BirdLog and found it rather good – but I have still had a pen and paper in my pocket as back-up.  From a first glance the eBird app looks just as useful and the notebook is probably on borrowed time

Data for analysis

 I stopped considering Birdata as a source of material for my analysis or trip planning when I read the message from Paul Sullivan referred to earlier in the thread about how the maps were broken and wouldn’t be fixed as work was going into the new portal (I am surprised the word ‘exciting’ wasn’t in there somewhere).  “If its broke don’t fix it but develop something else! “  Before reading the message I had tried to find out why a bunch of my observations over about 5 years were not showing up on the maps and this seemed to be the explanation.

I now use eBird almost exclusively as my source of data.  If I’m after a species it will tell me a list of broad areas, or Hotspots, where it has been seen, highlighting recent  records.  If I want an area around some town anywhere in the world, using Google Earth and the eBird polygon tool I can generate a list for the precise area I am interested in, not a bland geo-coordinate-based rectangle.  All of which is available at no cost and instantly available.  OK, I might have to express a little surprise at some of the species suggested but any output from a huge dataset is going to contain  a few outliers and has to be looked at critically.


A recent message on birding-aus referred to the use of Hotspots. 

I recently tried to access (using the Microsoft program of that name) data from another source for a popular site in the Canberra area.  Knowing that there was no standard name for the site I did a general search for 2 key words.  To my astonishment this generated 41 site names all of which clearly related to part or all of the site in which I was interested.  Some of the differences were as simple as adding an ‘s’ to one of the words which others used a synonym for one word.  In the past I have found apostrophes are a particular nuisance in causing multiple names to appear for one location.

At least as eBird is implemented in Australia there is moderation of site (ie Hotspot) names making the use of data a little less difficult.

Single Repository

My final point is to beg for some form of agreement on a single repository (there is no need for “an exciting new portal”) through which records of birds and other aspects of the Australian environment can be accessed at a reasonable cost.  

At present I could list at least 6 data sources that should be accessed for information about birds in NSW.  I am quite sure that the average person preparing (eg) an EIS is only going to refer to the data source they have been using for the past x years (or if they are a shonky developer, the one which best suits their interests).  In view of its breadth of coverage of the whole of natural phenomena the Atlas of Living Australia would seem to be the best candidate.  That should be quite independent of how data is collected.

No comments: