With the third spoon-billed sandpiper on the move, Prof. Rhys Green gives us his insights on why these three birds could lead us to as yet unknown sites, and how that could have important implications for conservation efforts.
From Prof. Rhys Green
The journeys of the three tagged spoonies, ET, HT and CT, may help us to solve a puzzle that is important for the conservation of their species.
The first test of a new method for estimating the total number of wild spoonies in the world was published recently in the journal Oryx. The estimate was based on scan surveys of spoonies in flocks of shorebirds staging on the Jiangsu coast near Shanghai in September-October 2014. This is the area where ET, HU and CT were tagged.
Scan surveys involve recording the proportion of all spoonies seen that have an engraved plastic flag on the leg. The surveys showed that about one in 25 of the adult spoon-billed sandpipers seen had these flags. The flags had been applied to nesting adults thousands of kilometres away at Meinypil’gyno in arctic Russia. About twenty leg-flagged adults were alive in autumn 2014 so, if birds from different breeding areas mix on migration, the world population of adult birds is 25 times 20, or about 500 adults or 250 breeding pairs. Given that the autumn population includes juveniles and one-year olds as well as adults, this result indicates a total population of 600-700 individuals in autumn 2014.
Now here’s the puzzle. If the counts of spoonies for all known wintering sites in China, Thailand, Bangladesh and Myanmar are added together the total comes to about half of the number estimated from the scan survey. Where are the missing birds?
The explanation could be one or both of two things. Maybe there are really more spoonies than have been counted at some of the wintering sites we already know about. That is certainly possible. Some sites, like the Gulf of Mottama where ET is now, are huge and it is difficult to ensure complete coverage on surveys. More exciting is the possibility that there are one or two sites we do not yet know about yet that hold large numbers of wintering birds. Which is it? If there are undiscovered wintering sites, finding them would allow conservation action to protect them there. It is still early enough in the winter that any or all of our three intrepid voyagers could still supply the vital clues before their tags fall off.
This is really exciting. Can’t wait to find out the answers.
Keep up the great work guys!!
Thank you! 🙂
Maybe the basic hypothesis that powered the “new method for estimating the total number of wild spoonies in the world” is fundamentally flawed and there actually ARE far fewer of the species alive on planet Earth than are mathematically extrapolated through the application of the method as described in ORYX.
Rhys has provided the detailed response below – I hope this further clarifies the estimate.
Best wishes, Rebecca
Population estimates are of course only estimates so yes, it’s possible there are actually fewer (or more) spoon-billed sandpipers in the world than the paper concludes. But probably not all that many fewer.
It’s impossible to do these things without some assumptions. These are described in the text of the paper and the impact of each assumption on the accuracy of the estimate is provided in Supplementary Table S1. Only one assumption has the potential to substantially affect the estimate, and that’s the assumption that leg-flagged birds have a similar probability of migrating via Jiangsu to birds without flags. As flagging has mainly been done at one breeding site and migration routes might well vary for birds from different breeding sites, it’s possible this assumption is wrong, as acknowledged in the paper.
However, if the world population is really much lower than our estimate, say 100 pairs rather than over 200, and birds from different breeding areas really do migrate via markedly different routes, then the proportion of adults with flags at many migration and wintering sites would have to be well over 10%. Why? Because the number of flagged adults alive was known quite precisely (21 to 26 birds) at the time. So if there were only 100 pairs or 200 adult individuals, we’d expect the proportion of adults with flags seen at all migration and wintering sites combined to average 10%.
But the proportion was actually 4% at Jiangsu, where there were a lot of adults, and hasn’t been much higher anywhere else where systematic scanning of substantial numbers of adults has been done. Often it is even lower, such as in scans of the large wintering population at the Gulf of Mottama.
All good science is open to being refuted by new data and, as the paper says, our population estimate could be overturned tomorrow by systematic scan surveys at other migration and wintering sites that find substantial numbers of adult spoonies with a high proportion of flagged birds (say 15-20%). But that hasn’t happened so far.
Dear Professor, All,
With apologies for not yet reading the paper, just the abstract and the discussion above (possible to receive a copy please?) I would like to ask how the different ability of observers to detect spoonies and flagged spoonies (and also to age birds correctly) in field conditions was taken into account in the analysis?
For some context, I was part of some of the surveys at Jiangsu in both October 2013 and 2014 and have spent a lot of time looking for and at the species here in Korea and elsewhere. In Jiangsu, repeatedly, different observers looking at the same flock of birds were able to detect different numbers of spoonies (perhaps on a few occasions detecting some – even when no spoonies were present!); and some observers were also able to detect flags, when others struggled to or could not. This is the messy reality of fieldwork of course. Sometimes it was the weather conditions or the optics used perhaps that complicated things; sometimes perhaps just the eyes or the relevant experience of the observers, but I think few people there would argue that not all observers could detect spoonies or flags equally. So even if the assumption was made that somehow all the spoonies that were present were indeed observed and accurately counted (and that none were double-counted, even though many of the birds were counted at different tide states by different teams of people doing the best they could in often quite difficult conditions: how to tell all unmarked individuals apart?) – then what assumptions were made explicitly about differences in the ability of individual observers to detect flags? This would seem to be a very difficult challenge, not just with spoonies but to a lesser degree even with some larger flagged shorebird species, especially as experience also suggests that the rate of detectability for even the same observer might vary. I do not think I was alone in concluding that flags are rather harder to see on fat birds that are fluffed up in cold conditions (more likely in October, when many are also extremely hard to age properly) than in September, when it is hotter and many birds are still rather slim.
As written above: estimates are only that. And of course, there might be more spoonies remaining than even the proposed method suggest. Let us hope so. However, if the likely bias caused by flag-detectability and accuracy/inaccuracy of counts has not been properly allowed for, would this method not tend to inflate remaining numbers considerably instead?
It might already have been factored in somehow, but if not, perhaps those with unfettered access to the flagging database (closed to most of us) can confirm or refute ]whether a higher proportion of flagged birds tend to be detected at smaller sites with fewer spoonies (eg the Geum Estuary in Korea) and a lower proportion have been detected at much larger sites (like Jiangsu and especially the Gulf of Mottama)?
Your insights on the above would be greatly appreciated.
Thank you in advance.
Nial had not read the paper when he wrote his comment, so it is not surprising that there are some misunderstandings of the principle of the method used. It was certainly not assumed that “somehow all the spoonies that were present [at Jiangsu in 2014] were indeed observed and accurately counted …and that none were double-counted”. The method does not involve any counting, so Nial’s concern that “accuracy/inaccuracy of counts has not been properly allowed for” is misplaced. Instead, the method requires that (1) the PROPORTION of spoonies at the site that were flagged (rather than not flagged) and (2) the PROPORTION of spoonies at the site that were not young of the year were both measured in an unbiased way. These two estimates of proportion were made, not by counting birds to determine total numbers, but by scanning flocks and looking at and classifying each spoonie found carefully to determine (a) whether it is flagged or not, and (b) whether it is a young of the year or not. For estimating the two proportions, it is not necessary that each individual bird is recorded once and only once: only that the chance of a flagged bird being recorded as such is the same as for an unflagged bird. Indeed, it is the fact that it is impractical to ensure that each individual bird is recorded just once, especially for unmarked birds, that makes our method necessary. Think about estimating the PROPORTION of balls that are black out of one hundred unseen black and white balls inside a cloth bag. You could do this accurately by taking out one ball at a time, writing down whether it is black or not, and then replacing it in the bag. Do that enough times and you can get a good estimate of the true proportion of balls that are black. It doesn’t matter that you sometimes pull out the same ball again: indeed, you would get a more accurate estimate by doing one thousand ball checks, so each ball is looked at ten times on average, than if you did ten or one hundred ball checks. The same principle applies to the spoony scan surveys. The two estimated proportions [(1) and (2)] are then used together to calculate the proportion of old (AHY) birds that were flagged. Finally, the separately determined number of flagged AHY birds estimated to be alive anywhere in the world in September 2014 is divided by that proportion to give the world population estimate.
Whether the two proportions we needed could be estimated in the field without bias was something we thought about carefully in designing the fieldwork. We only used data from the September survey in 2014 when the weather conditions were good and the birds were in the middle of moult and so were not carrying a lot of fat. We did not use data from large roosting flocks when flags can be difficult to see, but surveyed on the rising and falling tides when birds were spread out feeding. The scan surveys were done at three locations spread out along the Jiangsu coast and so probably sampled variation there might be among sites. The data were collected by a group of five experienced observers who assessed, for each spoony sighting, whether the bird was flagged or not. They had to be careful to only score spoonies seen well enough to be able to determine this. If the bird was too far away, had fluffed feathers covering the legs or flew away too early to know whether it was flagged or not, that record was not used. Therefore, because of the way the scan surveys were done, we do not think that there was “likely bias caused by flag-detectability”, as suggested by Nial.
Similar rules were required when scanning birds to classify their age though, for that, even more stringent quality-control standards were needed. Ageing is easier in September than later in the autumn, but even so, data were used only from the two observers who were confident that they could age all birds.