Low calibration set counts in LT15

Larocque-Tobler et al (2015) promised that, for their chironomid calibration set

  At least 50 head capsules (Heiri et al., 2001 and Larocque et al., 2001) were mounted.

A cursory examination of the data shows that this is unlikely to be true: many of the calibration set samples have a minimum relative abundance above the 2% expected for a singleton in a count of 50 heads.

The corrigendum admitted as much

Lakes with low numbers of head capsules were included in the transfer function used to create the published temperature reconstruction… The following lakes have now been deleted for the corrected temperature reconstruction: GOR, KOS, LEK, SAL, SZE, SZOS, TRZ, WAS and ZAB.

If a few samples were a few (i.e. can count on fingers of one hand) chironomids short of the promised fifty, I would not object. In the Polish calibration set, nine of the forty-eight lakes (two of the fifty lakes sampled have no chironomid data), are now reported to have fewer chironomids than promised, but how many fewer?

Since most samples should have at least one singleton, we can estimate the probable count sum from the reciprocal of the relative abundance of the least abundant taxa.

This plot1 shows the minimum percentage and the probable count sum for the Polish calibration set, highlighting the lakes that the corrigendum declared to have low counts.

unnamed-chunk-1-1

Two aspects of this figure are striking.

First that some of the probable counts are very low. Lake GOR appears to have a count of just five heads, based on a minimum relative abundance of 20% (the missing 20% might be an unidentified chironomid).

Cricotopus Endochironomus albipennis Microtendipes pedellus
GOR 40 20 20

Second, in addition to the reported nine Polish lakes with counts below fifty head capsules, a further ten have a minimum relative abundance greater than 2%, indicative of a count lower than fifty heads.

What about the Canadian portion of the calibration set? There are 73 Canadian lakes (not 72 as reported in LT15 – the data source, Larocque (2008), also reports 73 lakes). Four of these lakes appear to have a minimum count below fifty.

unnamed-chunk-3-1

This possible undercount in a few samples in Larocque (2008) is interesting but will not materially affect the results of that paper.

There are some additional curious aspects to the sums of the relative abundances for each lake in the calibration set. This sum can be below 100% if rare or unidentified taxa have been omitted. Values slightly above 100% are possible with rounding errors. But I struggle to understand how a sum of 105.1% is possible for Polish lake GIL. The sum of 175% for Canadian Lac H is beyond comprehension.


  1. All code used in this post is available on github 
Advertisements

About richard telford

Ecologist with interests in quantitative methods and palaeoenvironments
This entry was posted in Peer reviewed literature, Uncategorized and tagged , . Bookmark the permalink.

2 Responses to Low calibration set counts in LT15

  1. Pingback: The Humpty Dumpty theory of palaeoecology | Musings on Quantitative Palaeoecology

  2. Pingback: why would anyone not trust the author???? | Musings on Quantitative Palaeoecology

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s