EBird over time

 I have just downloaded the last 6 months of eBird data.  There are some interesting charts (if that isn't an oxymoron) a little further down this post after the rants about eBird and ACCESS.

The usual entertainments ensued:

1. Because eBird won't put in a custom polygon feature (as it doesn't assist photographers, it is of no interest to eBird) I have to download the entire East Gippsland dataset (431,692 records) and then extract the District subset (158,800 records - 36.8%);
2. Next step is to clean up the names which I did by loading the dataset to EXCEL (I deleted a few admin fields which I never use and the 2 comments fields, which while useful, I rarely access and doing so reduced the size of the file from 96Mb to 59Mb) and then uploading to ACCESS;
1. First step is to change 'Gray' to 'Grey' (23,717 records);
2. Then convert the stupid Clements taxonomic errors to the the common names used by eBird Australia (since the bogans at Cornell store them all as Clements even though input as Australian).  This affects a further 41,484 records, relating to 43 species.
3. I then ran a check to see which of the remaining names were not already in my reference list.  I can't give exact numbers, but it is mainly because they are  additional species for the Shire - not surprisingly most of them are pelagic species.  The species which caused me the greatest grief was Parkinson's Petrel (Clements) which equates to Black Petrel (every other taxonomy).

I must remember in future to check more closely the format which ACCESS decides to apply to several fields as it makes the data very hard to analyse, and the table is so big my computer doesn't have enough memory to change the record type.  I was aware of the problems following from 'x' in the number observed field (makes it sort as text so 1,10,11, 12, 2, 21  ....) and had deleted all the x's.  However I also noticed:

• 'Number of observers' got imported as dates (1 = 1/1/1900 etc);
• Many other numeric fields were imported as short text - including, to my great surprise, observation count; and
• see above managed to fix up the observation count data.

A few hours of harmless fun.

To the charts!

The first shows the number of records (an observation of a species on a date at a location is a record) per month since January 2014.  As there is a seasonal pattern (more observer effort in the warmer months) I also created a 12 month moving average and plotted a trend line for that.

The trend line has a high value of R2 so is probably indicative of a true trend not just chance.  It is interesting that the average is above the trend from October 2021 to April 2022.  I suspect this has something to do with COVID lockdowns being abandoned.

The second chart shows the number of species recorded per month.  Again a 12 month moving average is shown.

Generally there is a slight upwards eyeball trend (although I didn't calculate .  I note that the Winter plummets are less noticeable from 2019 onwards, which coincides with buying our house at Mallacoota and thus putting in much more time birding in the area, especially in Winter (I think we also did visits in July of 2017 and 2018.

The third chart shows the number of additional taxa per month.  I have plotted two series:

• All taxa which includes 'spuhs' (eg 'tern sp.') and slashes (eg Little/Fairy Tern)
• Species only which excludes spuhs, slashes and domestic taxa.  Where an observation is shown to subspecies level (ISSF in eBird speak) that is included with the parent species.

We keep on adding taxa and are likely to keep doing so as climate change causes species to change their distribution.