Novel is not a synonym of good: Luoto and Nevalainen (2013)

I had not heard of the journal Scientific Reports before Thursday when Gavin kindly promised to spoil my day by referring me to a paper published there. It is an open access journal in the mould of PLoS One, publishing “technically sound, original research papers of interest to specialists within their field”. Scientific Reports is hosted on nature.com: no one would ever embellish their CV by claiming to have published in the non-existent journal Nature Scientific Reports would they?

Gavin pointed me to Luoto & Nevalainen (2013), hereafter LN13. LN13 reports chironomid-inferred lake temperature reconstructions for the last few hundred years from three Alpine tarns. Two reconstructions show the expected positive trend, the third shows a slight negative trend. LN13 interpret this as the result of site specific factors: the lake that apparently cooled is north facing and receives snow-melt later in the year.

LN13 use a novel approach to reconstruct temperatures:

In contrast to the commonly used approach of using fossil Chironomidae (Insecta: Diptera) assemblages in lake sediments to reconstruct summer air temperatures, we here apply a novel approach of using a Chironomidae-based intralake calibration set to reconstruct past changes in summer water temperatures in the three study lakes.

This is not going to end well: Velle et al (2012) tested transfer functions based on intralake calibration sets by developing models based on modern observations from one lake and validating it against modern observations in another lake. Performance was generally much worse than expected from the cross-validation performance. Down-core reconstruction from the intralake transfer functions were inconsistent and suffered from a lack of good modern analogues, even in the focal lake. Velle et al (2012) argued that intralake transfer functions risked violating all of the assumptions of transfer functions. There are particular problems with assuming that the environmental variable being reconstructed is ecologically important when many variables (eg. depth, temperature, oxygen concentration) are highly correlated, and this correlation structure is likely to change over time.

In view of Velle et al (2012), intralake transfer functions are only going to be remotely credible with considerable external validation and good modern analogues. Unfortunately LN13 only minimally validates the transfer function model by comparing core-top reconstructions with measurements, and does not even mention analogue quality. Instead we have to eyeball the chironomid stratigraphies to get a feeling for whether there are good analogues. There is a slight problem – neither the raw fossil nor the modern chironomid data are shown in the paper. This does not seem to be a restriction imposed by the journal – up to eight figures are permitted. Instead we have to read several earlier papers to find and evaluate the data.

  • The calibration set from Moaralmsee is in Luoto (2011), which develops a lake-depth transfer function, not the temperature model shown in LN13.
  • The fossil data from Moaralmsee are from Luoto and Nevalainen (2012). There are no obvious analogue problems with these data, which Luoto and Nevalainen (in press) use to reconstruct lake depth.
  • The fossil data from Oberer Landschitzsee are from Nevalainen & Luoto (2012). The dominant taxa, Micropsectra contracta-type and Sergentia coracina-type, are not in the calibration set, indicating an enormous non-analogue problem.
  • The fossil data from Twenger Almsee are in Luoto and Nevalainen (2013). I don’t have access to this journal.

The non-analogue problems at Oberer Landschitzsee are so severe that any reconstruction from there is a joke. It proves only that provided some taxa are in both the calibration and fossil data sets, the transfer function will generate numeric results. If the Oberer Landschitzsee reconstruction is a joke, the whole paper is a joke.

There are other problems with this paper. The evaluation of the calibration set in LN13 is entirely inadequate. No consideration is given to the confounding variables such as sediment type, oxygen concentration and turbulence, which will be highly correlated with temperature. It is difficult to be sure that temperature is the variable driving downcore changes in the fossil assemblages. Even the authors are not sure: Nevalainen & Luoto (2012) write that changes in the chironomid assemblages at Oberer Landschitzsee “were most likely attributable to reduced hypolimnetic oxygen availability”, but now they reconstruct temperature.

The paper treats stratification as a deux ex machina: whenever there is a discrepancy between the reconstructions and the instrumental temperature record, stratification changes are invoked as the explanation. Of course, stratification changes will affect the reconstruction, but when they are invoked without independent evidence, they explain everything and explain nothing. For example at Oberer Landschitzsee, the lake is inferred to have become stratified in response to warming after 1850, but then become polymictic in response to accelerated warming in recent decades.

Site specific factors like snow-melt are important for chironomid-based palaeoenvironmental reconstructions, so the conclusions of LN13 are not implausible, but are not supported by their data. LN13 is a paper that should never have been published.

References

Luoto, T. P. (2012) Intra-lake patterns of aquatic insect and mite remains. Journal of Paleolimnology, 47, 141–157.

Luoto, T. P. & Nevalainen, L. (2012) Ecological responses of aquatic invertebrates to climate change over the past ~400 years in a climatically ultra-sensitive lake in the Niedere Tauern Alps (Austria)Fundamental and Applied Limnology 181, 169–181.

Luoto, T.P. & Nevalainen, L. (2013) Long-term water temperature reconstructions from mountain lakes with different catchment and morphometric features, Scientific Reports 3, 2488 doi:10.1038/srep02488

Luoto, T. P. & Nevalainen, L.  (2013) Climate-driven limnological changes determine ecological thresholds in an Alpine lakeAquatic Biology 18, 47–58.

Luoto, T. P. & Nevalainen, L.  (In Press) A multiproxy reconstruction of effective precipitation in the central Austrian Alps since the Little Ice Age, Climate Research.

Nevalainen, L. & Luoto, T. P. (2012) Faunal (Chironomidae, Cladocera) responses to post-Little Ice Age climate warming in the high Austrian AlpsJournal of Paleolimnology 48, 711–724.

Velle, G., Telford, R.J., Heiri, O., Kurek, J. & Birks, H.J.B. 2012. Testing intra-site transfer functions: an example using chironomids and water depth. Journal of Paleolimnology 48: 545-558. 10.1007/s10933-012-9630-5

Advertisements

About richard telford

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

2 Responses to Novel is not a synonym of good: Luoto and Nevalainen (2013)

  1. Pingback: Mayflies mandibles: as seen in the IPCC report | Musings on Quantitative Palaeoecology

  2. Pingback: Transfer functions and species coverage | 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