Less than two months ago, I ended up with a very unexpected invitation to review a classic Coexistence Approach paper submitted to Palaeo^3 — Palaeogeography, Palaeoclimatology, Palaeoecology (PPP). The first time ever that this happened and it’s only six years since we published our first critique (Grimm & Denk 2012).
 |
| When a journal needs impact, it cannot be too picky. |
Naturally, I had to decline to formally act as peer, because
- you shouldn’t work for RELX’ Elsevier for free, in whatever capacity (The Kraken awakes; Online proofing: service for stakeholders or shareholders; Elsevier's Research Data)
- given what we showed and pointed out more recently (Grimm et al. 2016 — after waiting a long time for our brain-child crossing the Forest of Reviews; Grimm & Potts 2016 — open access and 2-month discussion phase but no leading Coexistence Approach disciple dared to partake), any editor should straightaway reject any Coexistence Approach paper (or results).
Normally, the following would dissipate into oblivion, because – as we all know – the peer review process usually is confidential and should remain so to protect us all.
Being out-of-business, and having to fill a blog, I can cross the magic lines that bind professional career scientists to give a glimpse how this process works.
 |
| Googling "Guido Grimm" (Personally, I use DuckDuckGo to search the net, and my homepage is also the top-hit there.) |
Here’s my first mail to the handling editor of the paper, who apparently tried to reach out to me for reviewing but couldn’t "...find an up to date e-mail address".
Dear Xxx,
I strongly applaud your will that CA papers published in P^3 are finally reviewed by independent peers.
Your editorial colleague Tom Algeo has my current email, since I contacted him regarding P^3 data documentation policy triggered by a paper your journal published this year by Worobiec & Gedl (see this post https://researchinpeace.blogspot.fr/2018/01/business-as-usual-ppp-keeps-on.html). I was tempted to write a comment, but neither the authors nor Tom were able to provide me access to the primary data they used. The authors responded that they are "not sure" whether they are allowed sharing the data, and Tom made clear the editorial is not "answerable" to any such request [See Trying to disperse the Impermeable Fog #2]. I also found out that during review, the peer that pointed out (in a very friendly and positive way) problems of the study, was completely ignored.
Let's be clear: the NECLIME syndicate cannot afford to release the data they use for their reconstructions because
a) as far as can be said from the only CA/Palaeoflora study ever published providing the primary tolerance data (Quan et al. 2012, P^3), they are extremely biased
b) it can be shown that all resolved, potentially informative intervals are triggered by taxa that are either misidentified, misinterpreted or are unrepresentative. Astatistically used min-max tolerances don't have the capacity to model climate niches, not today, and, hence, not in the past.
c) one could use the meanwhile substantially corrected tolerance data (stated so in Utescher et al. 2014) on the published taxon lists to evaluate the effect on earlier published reconstructions.
In short, tumble the card house and expose a pretty pseudo-science.
Cheers, Guido
I strongly applaud your will that CA papers published in P^3 are finally reviewed by independent peers.
Your editorial colleague Tom Algeo has my current email, since I contacted him regarding P^3 data documentation policy triggered by a paper your journal published this year by Worobiec & Gedl (see this post https://researchinpeace.blogspot.fr/2018/01/business-as-usual-ppp-keeps-on.html). I was tempted to write a comment, but neither the authors nor Tom were able to provide me access to the primary data they used. The authors responded that they are "not sure" whether they are allowed sharing the data, and Tom made clear the editorial is not "answerable" to any such request [See Trying to disperse the Impermeable Fog #2]. I also found out that during review, the peer that pointed out (in a very friendly and positive way) problems of the study, was completely ignored.
Let's be clear: the NECLIME syndicate cannot afford to release the data they use for their reconstructions because
a) as far as can be said from the only CA/Palaeoflora study ever published providing the primary tolerance data (Quan et al. 2012, P^3), they are extremely biased
b) it can be shown that all resolved, potentially informative intervals are triggered by taxa that are either misidentified, misinterpreted or are unrepresentative. Astatistically used min-max tolerances don't have the capacity to model climate niches, not today, and, hence, not in the past.
c) one could use the meanwhile substantially corrected tolerance data (stated so in Utescher et al. 2014) on the published taxon lists to evaluate the effect on earlier published reconstructions.
In short, tumble the card house and expose a pretty pseudo-science.
Cheers, Guido
An unprecedented act: getting the paper and an invitation to review
I got a prompt and positive answer.
Apparently, this editor really wants to change the modus operandi of incestuous review and unhindered (and unhinged) publication of pseudo-results. About time. Just these days PPP published a paper, where the authors simultaneously reconstruct palaeo-temperature and -altitude using the Coexistence Approach (topic for a future post). Attached to the editor’s response was the submitted paper, which findings I was asked to treat with confidence (here’s the PDF [6 MB], a good example of how-not-to-make-a-palaeoclimatic-study co-authored by not only one but both palaeoclimate experts who invented the Coexistence Approach) and to write “…a paragraph explicitly asking for the supplementary data required”, which would have then been passed to “… the lead-author without attribution to you” A nice gesture, I suppose. However, I never felt a reason to hide behind the Impermeable Fog and would insist on being attributed. Also, I'm a fan of open scientific debate, which can be educative also for others.
He expressed further his hope that “once these data have been received”, I might accept the review request.
By the way, the authors, which include the acting High Lord Mosbrugger and High Wizard Utescher of the (only) scientific syndicate advocating the Coexistence Approach (Mosbrugger & Utescher 1997, paper published in – guess where), NECLIME – suggested five reviewers, all of which are listed as NECLIME members (most research etc funded with German money...), and three of which are frequent co-authors of Utescher (and Mosbrugger, to a lesser degree), two joined Utescher as co-authors of the 2014 paper (thought to be a reply to our 2012 critique; see also Grimm et al. 2016, table 1) and have first-authored papers that we showed to be flawed (see Supplement File ES2 to Grimm & Denk 2012; and Quan et al. 2012, the data we used for Grimm et al. 2016 to demonstrate practical flaws in the application of the Coexistence Approach).
Purest pal review reveals itself, how nice (and fashionable in climate sciences: a recent post regarding an editorial in Global and Planetary Change, 2011 article in Forbes Magazine). Which explains, why an invalidated method is still used. And the cosyness of the Impermeable Fog ensures, it's not too obvious.
 |
| The commonly appied single-blind confidential peer-review. |
But no, I would never act as a formal peer, for the reasons given above. Which should be pretty clear from the post I mentioned in my first mail and the one following-up…
I don't review, but I'm happy to share my opinion
Why would you throw matches at somebody that loves to burn bridges? This was my answer (including a non-formal review of the paper) terminating further correspondence between the editor and me on the topic (additions in [blue]).
Dear Xxx,
I appreciate the offer, but there is really no point in reviewing this paper, you may as well publish it like the many others. Because from an objective point of view it can only be rejected, and the authors encouraged to apply a not fundamentally flawed method. We outlined some options that may put even the Palaeoflora data to better use in Grimm & Potts (2016). But obviously, this is not (and cannot be, regarding the substantial impact generated by NECLIME, more than 10,000 citations since 1997) the long-standing policy of your journal.
Point 1 Using the a (pseudo)quantitative actuo-palaeontological CA on Palaeogene floras — Mosbrugger & Utescher wrote themselves in the 1997 paper that CA will provide best results for the Miocene (Miocene, because unweighted mutual climate range approaches, of which the CA is a subcategory, cannot resolve the Pleistocene fluctuations!) and be ok for the Oligocene but hardly possible for older time periods. Reason being that the Northern Hemisphere witnessed a major taxonomic turnover from the 'greenhouse' phase in the Paleocene/Eocene to the 'ice-house' in the Oligocene and the expansion of the so-called "Arcto-Tertiary elements" (modern-day mostly but not exclusively temperate lineages) to lower latitudes. Some of the Arcto-Tertiary elements, the most famous being the oaks, invaded various niches at lower latitudes (today, a few oaks can be found in the tropics), and let to a drastic decline of the ancient, supposedly warmth-loving so-called "para-tropical" elements. It is hence highly debatable to apply the actuo-palaeontological principle at all. It's pretty unlikely that any genus with e.g. today a MAT tolerance of 14–27 °C (warm temperate to tropical) and another of 8–16 °C (boreal to warm temperate) had the same in the Eocene, which is what the entire study builds on (only using a precision of 0.1 °C). And which we already demonstrated for the Eocene of China (Palaeoflora data provided in Quan et al. 2012).
Regarding data documentation everything should be already in the Electronic Supplement: "The complete floral lists, assigned NLRs and their climatic requirements are given in the Electronic Supplements 1–7." [This would be the second time in 20 years the used data are documented]; it later says: "Floral lists with corresponding NLRs employed in this study and their climatic requirements are made available in the Electronic Supplements 1–7." Not explicitly mentioned is that the climate tolerances are estimated by "careful selection of 4–6 climate stations" (Utescher et al. 2014), rendering them naturally rough estimates which should be represented as such, and not with a pseudo-precision of 0.1 °C and 1 mm precipitation.
Point 2: Lack of validation — Only the original 1997 paper is referenced for reliability/method validation, a paper that used wrong tolerances and four modern-day validation floras, two of which were from the same place (50km apart, Upper Rhine Valley), and one where the CA reconstructed partly erroneous intervals (technically impossible for a mutual climate range approach when using a modern-day flora, but happened because Mosbrugger & Utescher ignored a major climatic border between their modern validation flora, growing in an Mediterranean Csa climate, and the climate station they used for comparison, in a fully humid Cfa climate). Why one can publish results based on a method that so far has only been shown to not work (Klotz 1999, Thompson et al. 2012, our papers) and has never been validated, remains a mystery to me.
Also, as usual, they ignored everything we showed in our papers, just two examples: "Bombacaceae Kunth growing widely in tropics is a warm outlier in the analysis and is only present in two palynofloras." – their interval has been, and apparently, still is wrong, why the often occur as "warm outliers". Another reason is that they conflict with fully temperate taxa found in coexistence. In such a case, everything is possible. A too narrow interval (for the Bombaceae or the temperate taxa they were found with), mixed flora, or a niche shift. Since it cannot be said what it is, both the Bombaceae and the taxa not overlapping with them would need to be eliminated from the reconstruction, or the entire flora not considered (mutual climate range approaches and the actuo-palaeontological principle require that all taxa can and could coexist, if some don't, then something is wrong).
"Larix Mill. represents a cold outlier in the analysis. The taxon is mainly present in pollen records and here interpreted as altitudinal element" falls in the same category. It's also funny, because this is what we pointed out (also here, the at least the MAT intervals were too narrow). It's good to see that they at least now take up some of our points, even though they are not crediting us.
Point 3: Impermeable ignorance about fundamental problems of unweighted univariate mutual climate range approaches in general and CA in particular
a) Odd repetition of identical (pseudo)precise values as upper and lower coexistence interval boundaries — Note the same/very similar values are building up many intervals, independent from the time period. The reason for this is that only a very limited set of climate stations are "carefully selected", and often the same for very different taxa with quite different distribution (there's a beautiful figure in our 2016 paper illustrating this bias for CA/Palaeoflora studies). Accordingly, the "estimates" are the same house-numbers than the NECLIME syndicate has been publishing now since nearly two decades in your journal.
b) Lacking any statistical safeguards, CA will always reconstruct subtropical climates — Just compare the reconstructed "coexistence intervals" with those found for the Eocene of China and random subsamples (Grimm et al. 2016). They are finely fitting what we showed and published already: all roads lead to North Carolina (Grimm & Denk 2012), respectively Yunnan, when using East Asian fossil floras (Grimm et al. 2016): the more NLRs are used, the more likely you end up with (summer-)humid subtropical climates; climates in which any plant can (theoretically) coexist. And all mixed floras (from different altitudes, regions) will have an according pseudo-coexistence interval (particularly, since "climatic outliers" are just eliminated, to ensure really any flora will have a coexistence interval, in violation of the very basics of a mutual climate range approach).
c) Misinterpretation of coexistence intervals — by definition, the coexistence interval, which – aside for eventual "climatic outliers" – is the univariate mutual climate range of all NLRs should (if no first-level errors are present) include the real climate value, but it cannot be said whether the real value is closer to the lower or upper boundary. This is even stated in Mosbrugger & Utescher (1997) and has been verified using best-possible tolerance data by Klotz (1999) and us (Grimm & Denk 2012), see also Thompson et al. (2012). But the authors write (because their coexistence intervals mostly overlap across the entire covered time period, see their Figs 7ff): "The onset of pronounced cooling is quite evident from the late Oligocene temperature data, characterized by lower mean values [this is the mid-value between the upper and lower boundary of the coexistence interval] of MAT, CMMT and WMMT". A definite cooling would require a substantial shift in the mutual climate range, the coexistence interval, which is not visible. Instead the paper gives: "MAT (14.8/14–15.6 °C) for the LF 8" as the "lowest" for the Oligocene. Let alone that its biologically impossible, today, to find a genus-level flora providing such precision (less than 1 °C!, even tricky when you have a full species-level data set). This interval (14–16 °C, i.e. subtropical climate in connection with the reported CMT values) stands vs. "the highest values in the early Eocene for MAT (18.2/15.3–21.1 °C) are obtained for four floras: the PF 27, 28, 308 29 and 30 (southwest and central part)", again, subtropical setting (isn't it beautiful, no matter whether Paleocene, Eocene or Oligocene, the Russian Far East was a subtropical paradise). However, the global temperature was about 10 °C higher in the early Eocene than after the Oligocene cooling. Something that obviously had no effect on the Russian Far East: One of the Eocene "highest" intervals (15.3–21.1) overlaps with the Oligocene "lowest" interval (14–15.6 °C), so in the logic of CA and univariate unweighted mutual climate range approaches the hottest Eocene flora may have grown under the same MAT than the coldest Oligocene flora (~ 15 °C) [Explanotory note: the Eocene was globally much hotter than the Oligocene]. The other (18.2–21.1) grew under less than 3 °C hotter conditions (15.6 °C vs. 18.2 °C) or 7 °C (14 °C vs 21.1 °C). So still short from the global average of 10 °C less. Odd in this context, a plant with a minimum MAT tolerance of 18.2 °C must be an exclusively near tropical plant (Cfa, Cwa of southernmost China into full tropics), which even today, is not found coexisting with plants that cannot stand tropical conditions (the only tropical climate station with a Tann < 21.1 is Los Naranjos, Panama, 1200 m a.s.l.) Without having had access to the supplement these narrow(-ish) intervals point towards pseudo-coexistence, i.e. a level one error source, as outlined in Grimm & Potts (2016).
To this end Figs 7ff are revealing, although it is not really clear what the shown values are (I suppose box-plots – some miss the medians, too few data points? – of the coexistence intervals' "mean values" of all floras representing each time slice, comparable to our Road-to-Yunnan-figure of the random subsamples in Grimm et al. 2016?)
d) Theory — Again, and standard in published allegedly peer-reviewed CA/Palaeoflora papers, the theoretical issues with the CA (climate niches are not rectangular, unweighted univariate mutual range approaches cannot be decisive unless biased by first-level errors, how likely is it that the modern distribution can inform a 0.1°C value for a putative ancestor/ancient relative thriving 40 myrs ago) are not addressed at all.
What one would need to do to make this paper publishable — The minimum the authors need to do, is to illustrate the actual shift in a basic statistic manner by providing box-plots of min-max tolerances (see Grimm et al. 2016, Grimm & Potts 2017) for each flora and parameter instead only reporting the (pseudo-)coexistence interval and focussing on the meaningless "mean values", so one can at least assess whether the climate-defining taxa (the "shift" between 18.2 °C, possible lowest for hottest Eocene and 15.6 °C, possible highest coldest Oligocene) are due to exotic elements or not (nothing new, all said, outlined and published in Grimm et al. 2016, Grimm & Potts 2016). I'd be surprised if not the box-plots of reasonable tolerances would reflect the general cooling trend and the overall climate evolution in the Russian Far East (however, we showed, Grimm & Denk, 2012, for China that the temperature gradient between 0 and 1000/1500 m did not change much the position of the box-plots, only when you leave the subtropical lowland forest into the montane fully temperate forest, you see a shift in the distribution of min MAT tolerances, less so for max MAT tolerances). Which is all that can be inferred from taxon lists at this point (given that a genus', generic lineage, climate niche changes over time as much than those of its constituent species): trends! But not quantitative estimates such as the average temperature of the coldest month in the Eocene was 9–10 °C (well, 9.x–10.x °C, but as Utescher et al. 2014 make clear, this precision is just application-related and should not be viewed as such, without specifying at which accuracy they should be viewed.)
Moreover, the authors could make x-y plots of the min-max tolerances for two climatically informative parameters (e.g. CMT vs. MAP; WPwet vs WPdry), to test for pseudo-coexistence. Any flora with no bivariate mutual climate (coexistence) area (100% coexistence is a must, when the basic assumption should be fulfilled) is either a mixed flora or indicates a violation of one of the four basic assumptions of the CA, hence should be eliminated or filtered for non-coexisting taxa.
Rather than providing tables and lengthy texts reporting pseudo-precise (again how can modern distribution inform an interval with a precision of less than 2 °C or less than 10 mm precipitation/month for the Oligocene and older, if the same method fails to resolve the LGM?) and over-accurate (0.1 °C, 1 mm) coexistence intervals, the authors could illustrate the reconstructed intervals in a concise fashion as we did it for Quan et al. (2012): plotting all temperature values together viz all precipitation values, to pinpoint unfitting values/odd combinations. From the provided box-plots Figs 7ff it cannot be deduced how MAT, WMT, and CMT intervals correlate (which they should, but often they do not when it comes to coexistence intervals; note also that Thompson et al. (2012) showed for North America that CMT is the only relevant temperature parameter when it comes to plant distribution, has this been tested for East Asia?) or the precipitation values (side note: "MPwarm" is a bioclimatically useless parameter in the context of mutual climate range approaches, since the warmest month can be a very different one depending whether one is in the tropics, the dry climates or the temperate/boreal climates, again, something we pointed already out in our papers). An according supplement or in-text figure could replace chapters 4.1 and 4.2 and Figs 7ff. But insisting that e.g. "WMMT (27.3/26.6–28.1 °C)" is a quantitative estimate that the warmest month at locality PF26 in the early Eocene was 27.5 °C hot on average (the Lieth et al. climate station data includes 286 stations with a Tmax of 27–28 °C, tropical climates, dry climates, subtropical climates), and the using "mean values" to represent intervals such as "CMMT (8.2/3.8–12.6°)", i.e. “10.4°” or “8.2°” CMT and “10.1°” (64, mostly subtropical, some hot dry climates) or “6.65°” (86 stations, mostly subtropical, some dry climates), mean values for "CMMT (6.6/−0.3–13.6 °C)" (the latter interval matches 975 stations out of 2625, including everything extratropical from the hot deserts to the cosy Cfc of Iceland, i.e. a good deal of the B- and the most of the C-climates between 20–60 °N!) for the hottest Oligocene in the nicely coloured maps, always was and remains just ridiculous.
Beyond what I wrote above (which you may treat as a review and can forward to the authors), I will not formally review this paper for sure, because
But I may be tempted to point out how flawed this study is, once published (provided the Supplement Tables include the information promised), and what really can be said based on the compiled data. It probably will be a very nice example for another post and potential piece for bioRxiv's "conflicting data" category.
If you have the guts, you reject this paper straight-away and any other CA/Palaeoflora paper until the CA-applicants have shown how precise and accurate their reconstructions are at best using modern-day validation floras under realistic circumstances, i.e. genus- and family-level taxon lists and doing rarefication tests (validation process is outlined in Grimm & Potts, 2016, which is open access, and had a 2-month open discussion phase to which all authors of Utescher et al. 2014 were invited, but no-one took up the opportunity). Mainly they should stop using the CA protocol and apply at least a proper unweighted univariate mutual climate range approach: i.e. using the intervals, not the "mean values", and only include floral lists with 100% coexistence.
I'd like to point out that CLAMPP, which is mentioned in the introduction, is validated using now over thousands of modern validation floras, and has undergone permanent upgrades to minimise errors since pretty much the very beginning (which also was rough). So why does CA still get away with three, one of which partly failed?
Or you publish it as-is, as your journal has published many other papers of the CA syndicate (apparently despite what non-syndicate reviewers said, unchanged as in the case of Worobiec & Gedl 2018). If the supplement files include what is promised, it will be an exemplarily well-documented paper for CA/Palaeoflora standards, rarely found, and very valuable in that respect (the always cited Palaeoflora database does not provide any useful access to the used data)
Cheers, Guido
I appreciate the offer, but there is really no point in reviewing this paper, you may as well publish it like the many others. Because from an objective point of view it can only be rejected, and the authors encouraged to apply a not fundamentally flawed method. We outlined some options that may put even the Palaeoflora data to better use in Grimm & Potts (2016). But obviously, this is not (and cannot be, regarding the substantial impact generated by NECLIME, more than 10,000 citations since 1997) the long-standing policy of your journal.
Point 1 Using the a (pseudo)quantitative actuo-palaeontological CA on Palaeogene floras — Mosbrugger & Utescher wrote themselves in the 1997 paper that CA will provide best results for the Miocene (Miocene, because unweighted mutual climate range approaches, of which the CA is a subcategory, cannot resolve the Pleistocene fluctuations!) and be ok for the Oligocene but hardly possible for older time periods. Reason being that the Northern Hemisphere witnessed a major taxonomic turnover from the 'greenhouse' phase in the Paleocene/Eocene to the 'ice-house' in the Oligocene and the expansion of the so-called "Arcto-Tertiary elements" (modern-day mostly but not exclusively temperate lineages) to lower latitudes. Some of the Arcto-Tertiary elements, the most famous being the oaks, invaded various niches at lower latitudes (today, a few oaks can be found in the tropics), and let to a drastic decline of the ancient, supposedly warmth-loving so-called "para-tropical" elements. It is hence highly debatable to apply the actuo-palaeontological principle at all. It's pretty unlikely that any genus with e.g. today a MAT tolerance of 14–27 °C (warm temperate to tropical) and another of 8–16 °C (boreal to warm temperate) had the same in the Eocene, which is what the entire study builds on (only using a precision of 0.1 °C). And which we already demonstrated for the Eocene of China (Palaeoflora data provided in Quan et al. 2012).
Regarding data documentation everything should be already in the Electronic Supplement: "The complete floral lists, assigned NLRs and their climatic requirements are given in the Electronic Supplements 1–7." [This would be the second time in 20 years the used data are documented]; it later says: "Floral lists with corresponding NLRs employed in this study and their climatic requirements are made available in the Electronic Supplements 1–7." Not explicitly mentioned is that the climate tolerances are estimated by "careful selection of 4–6 climate stations" (Utescher et al. 2014), rendering them naturally rough estimates which should be represented as such, and not with a pseudo-precision of 0.1 °C and 1 mm precipitation.
Point 2: Lack of validation — Only the original 1997 paper is referenced for reliability/method validation, a paper that used wrong tolerances and four modern-day validation floras, two of which were from the same place (50km apart, Upper Rhine Valley), and one where the CA reconstructed partly erroneous intervals (technically impossible for a mutual climate range approach when using a modern-day flora, but happened because Mosbrugger & Utescher ignored a major climatic border between their modern validation flora, growing in an Mediterranean Csa climate, and the climate station they used for comparison, in a fully humid Cfa climate). Why one can publish results based on a method that so far has only been shown to not work (Klotz 1999, Thompson et al. 2012, our papers) and has never been validated, remains a mystery to me.
Also, as usual, they ignored everything we showed in our papers, just two examples: "Bombacaceae Kunth growing widely in tropics is a warm outlier in the analysis and is only present in two palynofloras." – their interval has been, and apparently, still is wrong, why the often occur as "warm outliers". Another reason is that they conflict with fully temperate taxa found in coexistence. In such a case, everything is possible. A too narrow interval (for the Bombaceae or the temperate taxa they were found with), mixed flora, or a niche shift. Since it cannot be said what it is, both the Bombaceae and the taxa not overlapping with them would need to be eliminated from the reconstruction, or the entire flora not considered (mutual climate range approaches and the actuo-palaeontological principle require that all taxa can and could coexist, if some don't, then something is wrong).
"Larix Mill. represents a cold outlier in the analysis. The taxon is mainly present in pollen records and here interpreted as altitudinal element" falls in the same category. It's also funny, because this is what we pointed out (also here, the at least the MAT intervals were too narrow). It's good to see that they at least now take up some of our points, even though they are not crediting us.
Point 3: Impermeable ignorance about fundamental problems of unweighted univariate mutual climate range approaches in general and CA in particular
a) Odd repetition of identical (pseudo)precise values as upper and lower coexistence interval boundaries — Note the same/very similar values are building up many intervals, independent from the time period. The reason for this is that only a very limited set of climate stations are "carefully selected", and often the same for very different taxa with quite different distribution (there's a beautiful figure in our 2016 paper illustrating this bias for CA/Palaeoflora studies). Accordingly, the "estimates" are the same house-numbers than the NECLIME syndicate has been publishing now since nearly two decades in your journal.
b) Lacking any statistical safeguards, CA will always reconstruct subtropical climates — Just compare the reconstructed "coexistence intervals" with those found for the Eocene of China and random subsamples (Grimm et al. 2016). They are finely fitting what we showed and published already: all roads lead to North Carolina (Grimm & Denk 2012), respectively Yunnan, when using East Asian fossil floras (Grimm et al. 2016): the more NLRs are used, the more likely you end up with (summer-)humid subtropical climates; climates in which any plant can (theoretically) coexist. And all mixed floras (from different altitudes, regions) will have an according pseudo-coexistence interval (particularly, since "climatic outliers" are just eliminated, to ensure really any flora will have a coexistence interval, in violation of the very basics of a mutual climate range approach).
c) Misinterpretation of coexistence intervals — by definition, the coexistence interval, which – aside for eventual "climatic outliers" – is the univariate mutual climate range of all NLRs should (if no first-level errors are present) include the real climate value, but it cannot be said whether the real value is closer to the lower or upper boundary. This is even stated in Mosbrugger & Utescher (1997) and has been verified using best-possible tolerance data by Klotz (1999) and us (Grimm & Denk 2012), see also Thompson et al. (2012). But the authors write (because their coexistence intervals mostly overlap across the entire covered time period, see their Figs 7ff): "The onset of pronounced cooling is quite evident from the late Oligocene temperature data, characterized by lower mean values [this is the mid-value between the upper and lower boundary of the coexistence interval] of MAT, CMMT and WMMT". A definite cooling would require a substantial shift in the mutual climate range, the coexistence interval, which is not visible. Instead the paper gives: "MAT (14.8/14–15.6 °C) for the LF 8" as the "lowest" for the Oligocene. Let alone that its biologically impossible, today, to find a genus-level flora providing such precision (less than 1 °C!, even tricky when you have a full species-level data set). This interval (14–16 °C, i.e. subtropical climate in connection with the reported CMT values) stands vs. "the highest values in the early Eocene for MAT (18.2/15.3–21.1 °C) are obtained for four floras: the PF 27, 28, 308 29 and 30 (southwest and central part)", again, subtropical setting (isn't it beautiful, no matter whether Paleocene, Eocene or Oligocene, the Russian Far East was a subtropical paradise). However, the global temperature was about 10 °C higher in the early Eocene than after the Oligocene cooling. Something that obviously had no effect on the Russian Far East: One of the Eocene "highest" intervals (15.3–21.1) overlaps with the Oligocene "lowest" interval (14–15.6 °C), so in the logic of CA and univariate unweighted mutual climate range approaches the hottest Eocene flora may have grown under the same MAT than the coldest Oligocene flora (~ 15 °C) [Explanotory note: the Eocene was globally much hotter than the Oligocene]. The other (18.2–21.1) grew under less than 3 °C hotter conditions (15.6 °C vs. 18.2 °C) or 7 °C (14 °C vs 21.1 °C). So still short from the global average of 10 °C less. Odd in this context, a plant with a minimum MAT tolerance of 18.2 °C must be an exclusively near tropical plant (Cfa, Cwa of southernmost China into full tropics), which even today, is not found coexisting with plants that cannot stand tropical conditions (the only tropical climate station with a Tann < 21.1 is Los Naranjos, Panama, 1200 m a.s.l.) Without having had access to the supplement these narrow(-ish) intervals point towards pseudo-coexistence, i.e. a level one error source, as outlined in Grimm & Potts (2016).
To this end Figs 7ff are revealing, although it is not really clear what the shown values are (I suppose box-plots – some miss the medians, too few data points? – of the coexistence intervals' "mean values" of all floras representing each time slice, comparable to our Road-to-Yunnan-figure of the random subsamples in Grimm et al. 2016?)
d) Theory — Again, and standard in published allegedly peer-reviewed CA/Palaeoflora papers, the theoretical issues with the CA (climate niches are not rectangular, unweighted univariate mutual range approaches cannot be decisive unless biased by first-level errors, how likely is it that the modern distribution can inform a 0.1°C value for a putative ancestor/ancient relative thriving 40 myrs ago) are not addressed at all.
What one would need to do to make this paper publishable — The minimum the authors need to do, is to illustrate the actual shift in a basic statistic manner by providing box-plots of min-max tolerances (see Grimm et al. 2016, Grimm & Potts 2017) for each flora and parameter instead only reporting the (pseudo-)coexistence interval and focussing on the meaningless "mean values", so one can at least assess whether the climate-defining taxa (the "shift" between 18.2 °C, possible lowest for hottest Eocene and 15.6 °C, possible highest coldest Oligocene) are due to exotic elements or not (nothing new, all said, outlined and published in Grimm et al. 2016, Grimm & Potts 2016). I'd be surprised if not the box-plots of reasonable tolerances would reflect the general cooling trend and the overall climate evolution in the Russian Far East (however, we showed, Grimm & Denk, 2012, for China that the temperature gradient between 0 and 1000/1500 m did not change much the position of the box-plots, only when you leave the subtropical lowland forest into the montane fully temperate forest, you see a shift in the distribution of min MAT tolerances, less so for max MAT tolerances). Which is all that can be inferred from taxon lists at this point (given that a genus', generic lineage, climate niche changes over time as much than those of its constituent species): trends! But not quantitative estimates such as the average temperature of the coldest month in the Eocene was 9–10 °C (well, 9.x–10.x °C, but as Utescher et al. 2014 make clear, this precision is just application-related and should not be viewed as such, without specifying at which accuracy they should be viewed.)
Moreover, the authors could make x-y plots of the min-max tolerances for two climatically informative parameters (e.g. CMT vs. MAP; WPwet vs WPdry), to test for pseudo-coexistence. Any flora with no bivariate mutual climate (coexistence) area (100% coexistence is a must, when the basic assumption should be fulfilled) is either a mixed flora or indicates a violation of one of the four basic assumptions of the CA, hence should be eliminated or filtered for non-coexisting taxa.
Rather than providing tables and lengthy texts reporting pseudo-precise (again how can modern distribution inform an interval with a precision of less than 2 °C or less than 10 mm precipitation/month for the Oligocene and older, if the same method fails to resolve the LGM?) and over-accurate (0.1 °C, 1 mm) coexistence intervals, the authors could illustrate the reconstructed intervals in a concise fashion as we did it for Quan et al. (2012): plotting all temperature values together viz all precipitation values, to pinpoint unfitting values/odd combinations. From the provided box-plots Figs 7ff it cannot be deduced how MAT, WMT, and CMT intervals correlate (which they should, but often they do not when it comes to coexistence intervals; note also that Thompson et al. (2012) showed for North America that CMT is the only relevant temperature parameter when it comes to plant distribution, has this been tested for East Asia?) or the precipitation values (side note: "MPwarm" is a bioclimatically useless parameter in the context of mutual climate range approaches, since the warmest month can be a very different one depending whether one is in the tropics, the dry climates or the temperate/boreal climates, again, something we pointed already out in our papers). An according supplement or in-text figure could replace chapters 4.1 and 4.2 and Figs 7ff. But insisting that e.g. "WMMT (27.3/26.6–28.1 °C)" is a quantitative estimate that the warmest month at locality PF26 in the early Eocene was 27.5 °C hot on average (the Lieth et al. climate station data includes 286 stations with a Tmax of 27–28 °C, tropical climates, dry climates, subtropical climates), and the using "mean values" to represent intervals such as "CMMT (8.2/3.8–12.6°)", i.e. “10.4°” or “8.2°” CMT and “10.1°” (64, mostly subtropical, some hot dry climates) or “6.65°” (86 stations, mostly subtropical, some dry climates), mean values for "CMMT (6.6/−0.3–13.6 °C)" (the latter interval matches 975 stations out of 2625, including everything extratropical from the hot deserts to the cosy Cfc of Iceland, i.e. a good deal of the B- and the most of the C-climates between 20–60 °N!) for the hottest Oligocene in the nicely coloured maps, always was and remains just ridiculous.
Beyond what I wrote above (which you may treat as a review and can forward to the authors), I will not formally review this paper for sure, because
- it'll be an utter waste of time. Everything done here, has been shown by us to be fundamentally flawed. Our critiques stand unchallenged. And this paper, like Worobiec & Gedl (2018) [or the recently non-NECLIME paper published by Wu et al. 2018], does not even address the most basic problems with the CA.
- I don't work for RELX without being paid for it (or any other journal enforcing and hiding behind peer review confidentiality).
But I may be tempted to point out how flawed this study is, once published (provided the Supplement Tables include the information promised), and what really can be said based on the compiled data. It probably will be a very nice example for another post and potential piece for bioRxiv's "conflicting data" category.
If you have the guts, you reject this paper straight-away and any other CA/Palaeoflora paper until the CA-applicants have shown how precise and accurate their reconstructions are at best using modern-day validation floras under realistic circumstances, i.e. genus- and family-level taxon lists and doing rarefication tests (validation process is outlined in Grimm & Potts, 2016, which is open access, and had a 2-month open discussion phase to which all authors of Utescher et al. 2014 were invited, but no-one took up the opportunity). Mainly they should stop using the CA protocol and apply at least a proper unweighted univariate mutual climate range approach: i.e. using the intervals, not the "mean values", and only include floral lists with 100% coexistence.
I'd like to point out that CLAMPP, which is mentioned in the introduction, is validated using now over thousands of modern validation floras, and has undergone permanent upgrades to minimise errors since pretty much the very beginning (which also was rough). So why does CA still get away with three, one of which partly failed?
Or you publish it as-is, as your journal has published many other papers of the CA syndicate (apparently despite what non-syndicate reviewers said, unchanged as in the case of Worobiec & Gedl 2018). If the supplement files include what is promised, it will be an exemplarily well-documented paper for CA/Palaeoflora standards, rarely found, and very valuable in that respect (the always cited Palaeoflora database does not provide any useful access to the used data)
Cheers, Guido
So far, the paper is not published. But I guess, it’s just a matter of time.
From the tribal wisdoms of the Dakota Indians, passed on from generation to
generation:
When you discover that you are riding a dead horse,
the best strategy is to dismount.
When you discover that you are riding a dead horse,
the best strategy is to dismount.
But not if the dead (actually, still-born; see Klotz 1999) horse is called Coexistence Approach and you can pull some strings during confidential peer review.
Update 16/7/2018 – old dogs can learn new tricks
The bushdrums have been telling me that the paper will not be published. At least, not in PPP. Apparently, they decided to change their modus operandi for the last two decade (closing both eyes when it came to CA/Palaeoflora papers and their extremely poor documentation) and decided that full access to the Palaeoflora Database would be a pre-requisite. Utescher et al. reacted as they did earlier (Tang et al. 2015) when being asked to release/document the primary data: the paper I got to review and another one in the pipe were retracted. Still fearing the light of day, my shadowy friends?
Enforce and encourage open data remains the only safeguard against pseudo-science.
Should be self-evident.
Further reading...
Riding against the Windmills of Palaeo^3 – a comment to Worobiec & Gedl (2018) and the authors and editorial response to my data request illuminating how deep PPP sits in the pocket of NECLIME's wide mantle.
For further examples how confidential peer review (including possible pal review) protects poor science and dubious publication ethics check out the #FightTheFog and bad science flagged posts.
And in case you feel the need to join the (so far) desolate troop (two dozen so far) charging the Windwills of Confidential Peer Review ensuring the Impermeable Fog protects unethical publication procedures, sign up for my still running petition to force all journals to employ peer review transparence.
Our 2012 [PDF] and 2016 [PDF] rebuttal letters for the papers published in Review of Palaeobotany & Palynology, another Elsevier journal with strictly confidential peer review. PS We suggested for both paper the two inventors of the Coexistence Approach, and invited all authors of Utescher et al. (2014) to used the 2-month open discussion phase for our second paper in 2016 submitted to Climates of the Past, in which we deal with the theory. 2012 (well 2011) one did write an 'anonymous' report, 2016 one apparently accepted but didn't return a report, and no-one used the open discussion phase of the Climates of the Past paper (non-invited reviews cannot be done anonymously...only the invited one, but they are also documented).
... and links.
The Coexistence Approach's home and castle: www.neclime.de
The arcane Palaeoflora Database, often cited, rarely documented what of these data (most of which are not publicly accessible and what is publicly accessible may be outdated; Anonymous in pers. comm. 2011; see also Utescher et al. 2014) were actually used in a paper (see Grimm et al. 2016 for discrepancies between data and results reported in Quan et al. 2012) – number of PPP editors having found this page but not looked in the data: 1.
World map of the Köppen-Geiger climate classification (Kottek et al. 2006; Rubel et al. 2016) provided as open data by the Climate Change & Infectious Diseases Group at the Veterinary-medicinal University in Vienna — a must for all palaeoclimate researchers using fossil plants as proxy data. Wladimir Köppen was (among other things) a botanist.
Also useful is to check out the Walter climate charts, a free-to-access collection of climate data from more than 2000 stations can be e.g. found here: https://www.climate-charts.com/index.html
Our online supplements for the 2012 [ZIP-archive] and 2016 papers include quite useful bits, e.g. floral lists that could be used to validate and test mutual climate range approaches and an EXCEL including the station data provided on CD by Lieth et al. (1999).
Cited literature
Grimm GW, Bouchal JM, Denk T, Potts AJ. 2016. Fables and foibles: a critical analysis of the Palaeoflora database and the Coexistence Approach for palaeoclimate reconstruction. Review of Palaeobotany and Palynology 233:216–235.
Grimm GW, Denk T. 2012. Reliability and resolution of the coexistence approach — A revalidation using modern-day data. Review of Palaeobotany and Palynology 172:33–47.
Grimm GW, Potts AJ. 2016. Fallacies and fantasies: the theoretical underpinnings of the Coexistence Approach for palaeoclimate reconstruction. Climates of the Past 12:611–622.
Kottek M, Grieser J, Beck C, Rudolf B, Rubel F. 2006. World map of the Köppen-Geiger climate classification updated. Meteorologische Zeitschrift 15:259-263.
Klotz S. 1999. Neue Methoden der Klimarekonstruktion - angewendet auf quartäre Pollensequenzen der französischen Alpen. Tübingen: Institut & Museum für Geologie & Paläontologie [now: Institute for Geosciences], Eberhard Karls University. [Ph.D. thesis supervised by V. Mosbrugger showing, among other interesting things, that CA was the worst of all option available back then; frequently cited for the opposite in NECLIME papers, since in German and hardly accessible.]
Lieth H, Berlekamp J, Fuest S, Riediger S. 1999. Climate Diagram World Atlas on CD [the data for the Northern Hemisphere are included in ES7 to Grimm & Denk 2012 as EXCEL multi-sheet file]
Mosbrugger V, Utescher T. 1997. The coexistence approach -- a method for quantitative reconstructions of Tertiary terrestrial palaeoclimate data using plant fossils. Palaeogeography, Palaeoclimatology, Palaeoecology 134:61-86.
Quan C, Liu Y-SC, Utescher T. 2012. Eocene monsoon prevalence over China: A paleobotanical perspective. Palaeogeography, Palaeoclimatology, Palaeoecology 365-366:302–311.
Rubel F, Brugger K, Haslinger K, Auer I. 2016. The climate of the European Alps: Shift of very high resolution Köppen-Geiger climate zones 1800–2100. Meteorologische Zeitschrift DOI:10.1127/metz/2016/0816.
Thompson RS, Anderson KH, Pelltier RT, Strickland LE, Bartlein PJ, Shafer SL. 2012. Quantitative estimation of climatic parameters from vegetation data in North America by the mutual climatic range technique. Quaternary Science Reviews 51:18–39.
Utescher T, Bruch AA, Erdei B, François I, Ivanov D, Jacques FMB, Kern AK, Liu Y-SC, Mosbrugger V, Spicer RA. 2014. The Coexistence Approach—Theoretical background and practical considerations of using plant fossils for climate quantification. Palaeogeography, Palaeoclimatology, Palaeoecology 410:58–73.
Worobiec E, Ged P. 2018. Upper Eocene palynoflora from Łukowa (SE Poland) and its palaeoenvironmental context. Palaeogeography, Palaeoclimatology, Palaeoecology DOI:10.1016/j.palaeo.2017.12.019.
Wu J, Zhang K, Xu Y, Wang G, Garzione CN, Eiler J, Leloup PH, Sorrel P, Mahéo G. 2018. Paleoelevations in the Jianchuan Basin of the southeastern Tibetan Plateau based on stable isotope and pollen grain analyses. Palaeogeography, Palaeoclimatology, Palaeoecology DOI:10.1016/j.palaeo.2018.03.030.
No comments:
Post a Comment
Enter your comment ...