Nightbirds of the SE Corner

 After compiling a post about the Nightbirds of Gippsland I recalled that I had a set of data from Birdata covering my Mallacoota District and Bega Valley Shire.  I will label this area the "SE Corner".  The following snip from Google Earth shows the approximate boundaries of East Gippsland Shire (green boundary), Bega Valley Shire (pink boundary) and Mallacoota District (red boundary).  The Shires also extend into the sea but the wet bit is irrelevant for this project.

So I have "repeated" my analysis from Gippsland to cover the SE Corner.  The summary data from this is covered in additional pages in the Revised Google Sheet on Nightbird in the SE Corners.  I hope the tab titles make it clear which data is what. 

There are many ways of comparing the different sets of information.  From a purist view, the analysis is undoubtedly complicated by Birdata having taking in a proportion of eBird records up to 2021.  (I have no information about why the process hasn't been continued past that year, but suspect shortage of resources is at least partly to blame.)  

Data Access

Before getting to the results and analysis a crucial issue is getting access to the data.  This differs greatly between eBird and Birdata and resolving the various issues has caused the analysis to become rather iterative (a technically nicer description than "going round in circles").

eBird

Getting access to data reported in eBird at the individual record level is quite simple,  Send them a message seeking access for a sensible reason (eg "reporting bird information to the local community") and authorisation for access - valid for a year - is received by email quite soon.

The main issue with eBird is that the smallest area for which data can be downloaded is the equivalent to US County, which in Australia is LGA.  So I can get East Gippsland (~500k records) and Bega Valley (207 k records) Shires but have to get creative to filter data for Mallacoota District.  No conceptual biggy, but a pain.

Birdata

It is possible to get access to summary data (eg a species list for a defined period) for a custom area, such as the Mallacoota District, very easily from the Birdata website.  In making that comment as a general-interest punter I found myself unable to access some data for Powerful Owl and Barking Owl, probably due to their threatened status.  Fortunately an administrator was able to give me what I needed.

However I have found it impossible to get access to unit record data,  similar to that from eBird, other than by involving the staff of BLA.  While the staff are very helpful (and provided the core data for the SE Corner which I have used in this work) they cannot be expected to work "as and when" a single member of Birdlife, such as myself, comes up with bright ideas.  So when I wanted slightly different data I have had to work out how to extract it from that which I had.  Those processes are described below.

I have received some summary data for Gippsland from a colleague with greater experience in handling these data and by getting a trifle creative, have managed to fill the major gaps.  One unusual data item which emerged in this exercise was the very high number of reports of Southern Boobook in the Birdata East Gippsland Region.  This became more intriguing when the reports for East Gippsland Shire were considered: this showed 618 reports of Southern Boobook, which agrees very well with the 630 reports one eBird for this species in the Shire.  This left nearly 1500 records in the relatively small Wellington Shire.  My colleague had already investigated this (emphasis added by me):

"... this is due to repeated 2Ha 20 min surveys over many years at one location near Stratford. There are 1223 records for Boobook at this location where there has been one bird in residence in the garage for 6 years according to a note I found in one of the surveys. Which goes to show how careful we need to be when interpreting bird data surveys."

My overall feeling is that it would be good if BLA could provide more comprehensive data access as is possible with eBird.  However I recognise that this might involve significant resources (and possibly a change in philosophy) within BLA.

Other Systems

The two systems described above are those most widely used by birders but there are a few other systems used of which I describe, and discount for this exercise, a few below.

iNaturalist

An excellent US based system which I use for everything except birds.  The problem for birding purposes is that it is driven by photographs: not bad for flowers, fungi and reptiles but ungood for birds bouncing around in trees or out on the horizon!  No birders that I know use this system.

Government recording systems

Both the NSW and Vic Government wildlife agencies have in-house systems which are mandatory for use by their staff and contractors.  For recreational birders the systems are not user friendly for input, and I haven't tried getting data out for many years.  

The agency executives love them as they have everything under control and avoids conflict with policy objectives.  I don't include any data from these systems below, but suspect it is trivial in number of records compared to the two major birding systems.

Selection and presentation of data

Other than the summary Birdata information for Gippsland (both East Gippsland Shire and the Birdata East Gippsland Region, which also includes Wellington Shire) I also had access to a set of data obtained from BLA which included individual records for Bega Valley Shire plus postcodes 3891 and 3892 (which comprise the Mallacoota District)

Of course the data for Mallacoota District is also in both the Birdata East Gippsland Region/East Gippsland Shire and SE Corner sets.  As the Mallacoota District is not a simple shape for assessing the detailed data in the SE Corner database, I identified the records from this area by reviewing the location names and flagging those in the area.  I did this in a two stage process:

1. Reviewing and marking the names of locations with more than 45 records; and then
2. Sorting that file by latitude and checking the names close to marked records.

This is logically sound and appeared to give a sensible answer.  The calculation of the unduplicated numbers of nightbirds ((SE Corner  - Mallacoota =Bega Valley) + East Gippsland Shire = Birdata Sum) is shown in the page "Calc Birdata Total" in the linked Google Sheet.

For the eBird comparison I have simply added the totals for East Gippsland and Bega Valley Shires but have shown the Mallacoota data as I present that in some analyses.

Note that the species are presented in taxonomic order using a code I have developed to combine the order of species as shown by Birdata and eBird.  Hopefully the foolishness (hubris?) that has led to the two different systems will disappear in future as both Birdlife and Cornell have adopted the Avilist taxonomy.

Effort involved in Birding

I have previously investigated, for the Mallacoota District, the resource factors that might affect the birding data collected through citizen science activities.  An obvious finding was that the more effort put in, the more species are found. This will be particularly the case for Nightbirds where more than usual effort is required to go out at night and find the birds (with a possible exception of calling Boobooks).  There are many ways of illustrating attributes of the areas I am canvassing in this exercise.  A selection of these are shown in the page "Attributes of the areas" in the linked Google Sheet.  

It should be noted that this area of enquiry raises many opportunities to become diverted down unexpected avenues of enquiry which I often call 'rabbit burrows'.  There are more potential burrows here than in Watership Down, so I have tried to keep a focus, at the risk of not fully covering some key possibilities.

Clearly East Gippsland is much larger, with about 4 x the area, and 3 x the number of eBird checklists and number of records than Mallacoota District.  

Compared to Bega Valley, East Gippsland is about 3 x the area and the number of eBird checklists and number of records.  A most telling measure is the number of record/observer suggesting to me that there are more resident/regular eBirders in the 'rest of East Gippsland' than Mallacoota and more in both those than Bega Valley.

It wasn't possible for me to access, directly, the number of sampling events or observers for Birdata. For the SE Corner, where I had access to unit record data I felt that taking an event starting at a specific time and date at a specific location was a good indication of a specific sample event and counted the number of those events.  For both number of Birdata records per site and records per event Bega Valley is about 1/3rd the value for the Mallacoota records.  As with the eBird records I feel this indicates a much lower level of resident activity in Bega Valley than in Mallacoota.

A crucial point noted during my analysis was the difference in number of species per survey between (eg) the Birdata information for Bega Valley and the eBird data for Mallacoota.  This is summarised in the page '#Species X survey' in the linked Google sheet.  The differences are quite dramatic:

• For Bega Valley 48% of Birdata surveys report a single species and 72% cover 1 to 5 species.  The equivalent numbers for Mallacoota are 12% and 25% (for eBird) and 7% and 16% (for Birdata);
• The largest category of species counts for Mallacoota in both systems is 11-30 species per survey with 45% of surveys. For Bega that group comprises 14% of the total for Birdata.

For Mallacoota these findings match well with my knowledge of the users of the two systems.

Analysis of results

There are an almost infinite range of ways of analysing these data.  My approach will be to look at three aspects of the data for Nightbirds:

1. Compare the data for the two data systems for the whole project area;
2. Compare the data for the three survey areas looking at both data systems; and
3. Comment on the impact of timing of Surveys for the species .

I have also ranked the top 9 species (omitting Tasmanian Boobook) in ascending order of number of records as shown in the page 'Ranks' in the linked Google Sheet.  This means that the numerically largest rank (in this case 9) is given to the species with the greatest number of records which gives the correct appearance to the charts below..  I think the rather blunt approach of ranks is appropriate given the impact that collection practices might have on the precise values compiled.

Comparison of Data Systems

See page 'Totals both systems' in the linked Google Sheet.

An issue to note is that Birdata has imported some data, up to 2021, from eBird (when some criteria regarding collection protocol are met).  I have not (yet) investigated the impact of that in this project but in another project (looking at Pacific Gulled in the City of Port Phillip) it made a dramatic change to some results.

It is notable that 6 species have the same rank in both systems, and the largest difference is only 2.  I take this to suggest that overall both systems are "telling the same story".

The differences in number are a little more evident and more puzzling.  

A first issue is that there are many more eBird surveys in East Gippsland than Bega Valley but more Birdata Surveys in Bega than Mallacoota. I have attempted to overcome this by presenting reporting rates using number of surveys as the denominator.  

Excluding the Tasmanian Boobook, for eBird the reporting rates for Nightbird species are higher, often notably, in eBird than Birdata for 7 of the 9 species.  The notably high values are mainly for species with high site fidelity for roosting sites.  I am, and have been for a long time, intrigued by the low reporting rate in eBird for Tawny Frogmouth in Mallacoota.  I can offer no explanation for the relatively high reporting rates for Boobook and Masked Owl in eBird.  Possibly the established sites for those species are well known to visiting eBirders who put in time to visit those sites?

Comparison of Areas

The three areas are Mallacoota District; Rest of East Gippsland; and Bega Valley Shire.

The results for the areas are shown in the page "Totals and Rates SE Corner" in the linked Google Sheet.  There are marked differences between the results from eBird and Birdata.  In view of: 

• the reliance of the Birdata series on records with very few species, 
• the fact that I don't have access to measures of effort for the "Rest of East Gippsland" and 
• the confusion possible with some eBird 

I  will focus on the eBird data in commenting on the three areas in eBird.

As a first step I looked at the the 3 way differences shown in columns J to L in the page.  My first step was to look at the modulus of the differences (ie magnitude of the difference regardless of sign: -1.6 and +1.6 both show the same size of difference)   Adding those differences and summing the result gives:

Mallacoota - Rest of East Gippsland     1.56

Mallacoota - Bega Valley                      2.03

Rest of East Gippsland  - Bega Valley  2.35

Clearly the 2 Victorian areas are much more alike than either is to Bega Valley. 

My second step was to examine the eBird reporting rates for each species shown in columns G to I of the same page.  My basic approach was to identify the highest rates for each species and highlight those cells in yellow.  Where 2 (or all 3) areas were similar I marked the similar ones. In summary:

• Tawny Frogmouth: much higher reporting rate in Rest of East Gippsland than either Mallacoota or Bega Valley.
• White-throated Nightjar: Much higher rate in Mallacoota with the other two areas similar.  The species was reported from 43 localities in Mallacoota with the largest number of reports coming from Watertrust Road area (26 reports) and Shipwreck Creek area(22 reports).
• Australian Owlet-nightjar: all three areas close to a rate of 0.7.  As noted repeat sightings likely once a roosting hollow is identified.
• Sooty Owl: the 2 East Gippsland areas very similar rates, about double that of Bega Valley.  In Mallacoota, Watertrust Road area has most sightings (probably because it is drivable and powerlines make good sites from which to hunt,
• Australian Masked-Owl: the 2 East Gippsland areas very similar rates, about ten times that of Bega Valley.  Again, Watertrust Road is the source of the highest number of reports.
• Eastern Barn Owl: All three areas have - to me - surprisingly low rates with Rest of East Gippsland higher than the other two areas.  This possibly reflects the species preferring open country rather than forest.
• Powerful Owl: similar, modest, rates in all three areas.  Given the size of territory occupied by members of this species modest rates are not a surprise.  
• Barking Owl: Higher, but still low, rate in Bega Valley and not reported at all in Mallacoota.  ABG notes less common in Southern Australia.
• Australian Boobook: Highest rate in Rest of Eat Gippsland but Mallacoota similar(ish).  Much lower rate in Bega Valley.  Reported from many (256) locations in Rest of East Gippsland with a single observations for most localities.  In Mallacoota the localities with most observations are near accommodation (Gipsy Point and the Author's house!)

The author's impression of the habitats and topography of the various areas is that they are broadly similar in nature, with Mallacoota having a larger proportion of wooded country and Bega possibly having a higher proportion of foothills and higher country.  

I have looked at the habitat and prey preferences of the 9 species (excluding the Morepork as there are so few records) as given in the Australian Bird Guide (ABG) with prey preferences for 2 species coming from HANZAB as ABG is silent on those details.  These are shown in a page Habitat and Food in the linked Google Sheet.  I cannot see any reason in there for the differences in reporting rates as shown in the data I have compiled.

Timing of Observations

In my previous post on this topic I included commentary on the timing of observations of Nightbirds and feel this is an important subtopic for trying to understand the differences in reporting of this group of species. 

This section begins with a summary chart of the percentage of total number of events starting in three time ranges.  The choice of these ranges is explained in the previous post.  It should be noted that a fourth 'range' - "Time NS" is shown to account for records using the Historic protocol where start times are not known.

A second chart shows a similar chart but limited to the events in which a Nightbird was recorded.  The difference between the two charts is clear.

A notable point of detail is the higher proportion of daytime observations in the Rest of East Gippsland than either of the other two areas.  It is probable that this is related to the relatively high reporting rate of Tawny Frogmouths (65% of which in the Rest of East Gippsland were seen in the daytime).

I have also computed the correlation coefficients for the series of time observations for Nightbird events.  This gives a very high level of correlation between Mallacoota and Bega Valley and a lower (but at 90% still quite high) between the Rest of East Gippsland and both of the other areas.