Monday 6 June 2016

BirdBox2016: everybody out!

Well, I say everybody, but that's not exactly true.

Of the 9 healthy looking chicks which hatched from the 9 eggs laid, we seemed to lose 4 during the last few days.

Now is the time to take a look at the data and present a few graphs.

We have recorded a certain amount of data for our Maple Tree bird box, but naturally there are some limitations when it comes to interpretation.


This year we have recorded the local atmospheric pressure in addition to the outside temperature and general weather conditions. Measuring the outside temperature is surprisingly difficult, but all we are really interested in here are the general trends, and whether these parameters appear to affect behaviour.

But the prime measurement of interest is a count of the number of times that the bird box entrance sensor is triggered. There are mechanisms in the bird box system to limit false and multiple triggering. Having watched how fast the adults enter and exit, my one concern is whether they could get their whole body past the sensor in under 500ms (in which case they may not always be counted).

During the early stages of box selection and nest building, blue tits often peck around the entrance without entering the box (i.e. they can clock up a high count from the outside). However, once the chicks have hatched, they always fly straight in with food, and straight out again. So it is reasonable to divide the box count by 2 and use this as a measure of the number of "feeds" during this intense feeding stage.

As this basic count is recorded on an hourly basis, we can look at the frequency of feeding throughout the day and attempt to estimate the average number of feeds per chick. Although, having lost 4 chicks along the way, I have to guess which days the chicks died, which is another source of error.

Also note that the system is battery operated, so it "sleeps" if there is no activity after 20 minutes. When it is woken up (as a result of detection at the entrance) it is unable to count activity for about the first 50 seconds. Again, this is not a real problem during the busy feeding stage, as the system only gets a chance to sleep over night.

The main graph

So the main graph is the easy one, which probably reflects fairly accurately what was going on during the final 20 days.

The atmospheric pressure didn't change at a very high rate during these 20 days. And (as in previous years) I can't see any correlation between the "weather" and levels of box activity.

Popular feeding times

I have created a chart based upon daily box counts for each 1 hour segment. The "hatching" days and the final (flight) day are excluded because the adults do sometimes enter the box without food on those days.

As expected, feeding peaks during the first full hour of the day that the parents can gather food

If we add up the total number of calculated feeds over this period, the answer is just over 8000 feeds.

By dividing by the number of chicks (and remember that this started to reduce from 9 around the 1st June) I get the average number of feeds per chick: 1085

Also remember that this figure does not include feeds on the first few days and the last day, which would easily push the result well over 1100 feeds per chick.

It is tempting to try to visualise what 1000 caterpillars would look like. If they were 15mm long and 3mm in diameter, the volume would be over 100 cubic centimetres.   ...some lunchbox!

However, the caterpillars would have been smaller than this during the first few days, and the parents have been supplementing their meals with fat from our fat-ball feeder.

Average feeds per chick

Now its starting to get tricky. It would have been easy if all 9 chicks had survived, but I've had to guess when each of the 4 chicks died, and adjust the denominator accordingly.

The graph shows the feed rate increasing as the chicks get bigger. But its worth noting that the average size/weight of each feed is also increasing, especially during the first 10 days as the chicks are able to swallow larger items.

Peak visits per hour by day

Not sure if this shows anything very useful, but for this graph I've plotted the highest (peak) number of feeds within an unspecified hour.

Just why they are able to gather more food during some hours rather than others is not clear (i.e. is food more abundant or are they just working harder?).

Photo time...

Wow! It's a long way down

Tempting the chick to leave

Too many caterpillars?

I'm outta here!

What are we going to do with ourselves from now until March 2017?


  1. How about build a two sensor entrance hole so that you can differentiate between 'out' and 'in'?

    1. I have already started testing ideas for BirdBox2017, so will certainly give this some thought. Initially it occurs to me that the 2nd sensor will be quite close to the first. So the time difference (as a bird enters or leaves the nest) will be very short.

      This could be quite demanding for the Pi, but would be much easier to program in the Picaxe, where I can run a program in a very tight loop. My existing 8 pin Picaxe would need to be replaced with a 14 pin device to include enough I/O (e.g. additional inputs to read two photo-transistors and additional outputs to provide the Pi with an "in" trigger and an "out" trigger).

  2. I'm working on a MkII that will do this better than my MkI. the problem is that Mk1 is several meters up a wall so playing with it is not really possible. Have a working prototype on the bench though...