Friday, 10 November 2017

How many records in the database are unreliable?

I am currently working my way through the latest maps and histograms prepared for an updated atlas. It is a very slow process because so many of the species accounts require changing in the light of recent data. However, there are some consistent problems and one or two that give rise to uncertainty about the data. So, here are a few examples:

Eristalis rupium

This is mainly an upland species that does occur in lower-lying areas in Scotland. In this case there is an obvious doubtful record from Norfolk. This record comes from the NBN and thus has not been previouslyassessed or checked by us. This is a fairly common problem and we should be able to fix it. If there is a specimen or photograph and we can examine it we can test the veracity of the record. If there is no means of corroboration we will mark as 'requires confirmation' and it will not appear on the maps. Likewise, there is a record from around Hull that requires further investigation - I suspect this too is wrong and that somehow it has slipped through the net of basic checking when we import data.
Figure 1. Distribution of Eristalis rupium in un-filtered data. Record for the Norfolk coast seems highly implausible.

Melangyna quadrimaculata

This is a very early spring species that is quite difficult to identify and might be mistaken for a Cheilosia in the variabilis group. Also, recent experience has shown that dark specimens of Leucozona laternaria are sometimes run to this species. So, the phenology histogram needs to be treated with care: are later records of larvae or adults? If the latter then the chances are that the determination is incorrect and, again unless there is a way of corroborating the data the records will have to be marked as 'requires verification'.
Figure 2. Phenology plot for Melangyna quadrimaculata with outliers in July and August that are almost certainly misidentifications

Melanogaster aerosa

The last two examples are really quite simple, but there are many more complicated ones where there is uncertainty. The data may be right, but they are likely to be badly wrong. This is exemplified by the histogram for Melanogaster aerosa which is fiendishly difficult to separate from M. hirtella. As far as I know, I have never seen it, despite checking many thousands of Melanogaster! I can be certain I've not seen the female, but males are a bit more problematic. Thus, what can be made of the phenology histogram? Alan Stubbs reported in his monograph that M. aerosa (Figure 3) tends to occur later in the year than M. hirtella (Figure 4), and yet we see a substantial overlap and almost two generations, one in April/May and the other in July and August. I suspect a substantial level of mis-identification. The big question is 'how to resolve this one'? I think we are going to have to look very carefully at who the records come from and probably to exclude data from all but those who we have a reasonable knowledge of their ability. This sounds a bit harsh, but clearly there is something adrift with the data here. There are a number of other species where we will have to look equally carefully.

Figure 3. Phenology histogram for Melanogaster aerosa
Figure 4. Phenology plot for Melanogaster hirtella
Platycheirus immarginatus

In our experience, this is a strictly coastal species and there is confusion between females of this species and females of P. perpallidus. So the likely approach is to produce a map based solely upon records of males by known reliable recorders.

Figure 5. Distribution map for Platycheirus immarginatus based on raw data.

No comments:

Post a Comment