Black Cohosh Revision

From Community

Zoe Gardner, L Lueck, Erik Barry Erhardt, L E Cracker

"Morphometric analysis of Actaea racemosa L."

Revision of manuscript.

[edit] Revision Plan

I know we're both busy, but if we create a plan of attack and work on it a little bit each week, we might have enough finished for a resubmission in a couple months.

Given their "backlog of high quality papers awaiting publication", does it make sense to make proposed revisions and submit to another journal? Do you have a second choice?

Plan for us as a team as I see it right now:

Decide who will be involved in revision (just us two, or other coauthors as well?).
1. E & Z, with Lorna -- Craker to read
Read reviewer comments and manuscript.
For reviewer comments, consider what the strategy for addressing each comment is and detail a few next actions toward addressing.
1. Assign who will work on each comment
2. Write these details out where we can both review our ideas (we can use my community wiki for this).
Phone meetings between us to put our ideas together (may be enough that we have multiple conversations to discuss).
Agree on an overall plan and delegate next actions between us (probably an obvious way to divide work).
1. Time line for work to be completed in.
As we complete parts, check in with each other and take another set of actions.
All read revised paper, final corrections.
Submission to Journal (same or a different one).
Celebrate!

[edit] Time line

Try to finish revision by 11/24 to give a week to send to Lorna.

[edit] October 29th

Z- Revise table 1 to include pertinent data on populations. DONE -- Note to self: 1 minute on GPS is approx equivalent to 1.15 miles.

Z- Revise section on medicinal uses to be shorter. DONE

Z- Revise discussion section regarding expectation of finding dissecta varieties. DONE

Z- Look to see how other manuscripts represent within and between population variation for individual characteristics.

[edit] Nov 5th

Z- Revise the methodology section to include details on population & plant selection. DONE

Z- Change vulgar to botanical name in all instances. DONE

[edit] Nov 12th

Z- In the results section, add an in-text acknowledgement of the lack of significant results in the T-K analysis tables, noting which tables have lots of variation & help towards separation of populations.

Z- Create suggestions as to how paper to see how it could be condensed (help from Lorna and or Craker)

[edit] November 16

E- work through all comments on my list, update Z with what I have done, send updated text/tables/plots.

Order to complete: 7ai, 7aii, 7b, 8a, 8b, 9a, 9b, ** current progress **, 10a, 10b, 10c, 10d

E- Make a list of which Tukey-Kramer tables should be maintained and which deleted.

Lorna Comments e-mail

Z - 20080107, Talk with advisors regarding dendrogram findings.

All Markup changes accepted on 20080208 version, to start clean. So

[edit] February 2008

E - to do: read paper through, focus on results section; write discussion/whole plant and discussion/relationship among pops; try to make all parts of paper have consistent message (and content); results, decide on TK tables to discuss to remove (consider using a minimum F value to keep); Remove figure 9b and include a sentence to describe what happens if individuals are considered as a replacement

[edit] Editor Cover page

Botanical Journal Of the LINNEAN SOCIETY 21 May 2007.

Dr Z. Gardner Department of Plant, Soil & Insect Sciences, University of Massachusetts, Amherst, MA 01003, U.S.

Editor: Dr. Stephen L. Jury Centre for Plant Diversity & Systematics School of Biological Sciences Plant Science Laboratories The University of Reading, Whiteknights Reading, RG6 6AS, UK

Dear Dr Gardner,

Your paper (MS 2294) has now at last been reviewed and I enclose the report of our referee along with a copy of your manuscript containing annotations.

We currently have a backlog of high quality papers awaiting publication and in this context I am only able to accept papers that are highly recommended by referees. Unfortunately, the referee was unable to recommend publication. I am sorry that on this occasion you were not more successful but I hope the comments will be helpful to you.

With best wishes, Yours sincerely Dr. Stephen L. Jury Editor Enc Phone Fax Web Email +44 (0) 118 378 8169/5090 +44 (0) 118 378 8160 www.herbarium.rdg.ac.uk s.l.jury@reading.ac.uk

[edit] Reviewer Comments

A morphometric analysis of A claea racemo.ca L. (Ranunculaceae)

Z Gardner, L Lueck, E.B. Erhardt, L.E, Craker.

In this paper the reader is presented with the results of a numerical examination of collections of plants from 26 populations and concludes that there is little variation between populations in the characters studied. This conclusion seems correct, from the relatively small amount of relevant information given.

Online Supplement: Resolve - will make a file of data & note that this file is available upon request from ZG (or EBE).

[edit] 0

0 However,

- a the paper lacks information on the criteria used in the selection of populations, on the habitat at each site and on the criteria used in the selection of plants within the site. The reader may suspect that some populations contained so few plants that minimal selection was possible. The methods of selection used are likely to greatly influence the variation found and thus the conclusions reached. ADDED DETAILS ON POPULATIONS IN TABLE 1, GAVE AS MUCH CRITERIA ON PLANT SELECTION AS I COULD FIGURE OUT, MAYBE NEED LORNA OR CRAKERS ADVICE ON HOW TO DESCRIBE BETTER.

similar to a short comment 4

- b The reader is told that not all the values obtained are from the plant part enumerated in the Methods section but are not informed for which plants substitution was necessary. This substitution must have increased the observed variation between plants but there is no discussion of the importance of any increase to the final results. ADDRESSED BY SAYING EXACTLY HOW FREQUENTLY THE SUBSTITUTION WAS NECESSARY.

Methods: Z -

- c The Final conclusion could be fully shown in a much shorter paper.

Shorter: Z - more opportunity for condensing?

- d Is the variation found between populations to be expected from the within population variation?
- e Which populations are most different from the others?
- f Are there ‘way out’ plants? (or were such plants selectively not collected?). The variation that is important to answer such questions is some average of the significance of the variation found across all characters. Figures 8 and 9 are the only information given that addresses this.

Discussion: Z - some questions for Lorna to address

[edit] 1

Short comments:

1 Would it not be better, after the first full naming of this species in the introduction, to use the scientific name by itself? Later you use in various places the full definition, the vulgar name and the Latin name. VULGAR NAMES HAVE BEEN REPLACED BY BOTANICAL NAMES.

[edit] 2

2 Is the question of the medical use of this plant relevant to the information given in the paper? A LITTLE BIT. REVISED TO MORE APPROPRIATE AMOUNT OF INFORMATION ON MEDICINAL USE, WITH EMPHASIS ON CONSERVATION CONCERNS

[edit] 3

3 The purpose of the study is given at the bottom of page 3 and at the start of the conclusions, p20. These are not the same. That in the Conclusion seems most appropriate. THE PURPOSE HAS BEEN STANDARDIZED BASED ON REVIEWER COMMENTS.

[edit] 4

4 How were the populations selected? If ‘known’ localities were visited then the variation found is likely to have been biased, although in what way may not be clear. The introduction suggests that there were very many populations in and around the Appalachian Mountains from which the 26 population were selected. Tl paper states that populations were sampled “throughout the natural range” although no samples were taken from the periphery of the range in Ontario, Missouri, Georgia etc. and thus much of the natural variation might be expected to have been missed. Was the full altitudinal range was sampled. WILL INDICATE HISTORICAL RANGE & POPULATIONS SAMPLED IN FIGURE 7. GREATER DETAIL ON METHODS FOR POPULATION SELECTION HAVE BEEN ADDED.

Table 1 contains little information about the locations: no exact position is given,.no altitude estimated, no environmental information (soil, canopy) included and no estimate of the size of each population or of the area covered by the population. Was the distribution of the species related to any particular species in the canopy of the deciduous woodland? TABLE 1 HAS BEEN EXPENDED TO INCLUDE MORE DETAIL.

[edit] 5

5 On page 4 we have “In each population, plants were selected to provide a representative sampling of the entire population.” Does this mean that an attempt was made to obtain the smallest and largest plants as well as medium sized plants and those with both normal and extreme leaves and inflorescences or does it mean that the extremes of height, leaf structure and inflorescence at each site were ignored? The way samples were taken is likely to greatly affect the variability detected and the selection criteria used need to be carefully stated. It is not clear why a ‘botanist’ will have been able to follow the selection criteria better than any other mortal.

Population selection (similar to 4): Z - methodology, how populations were selected.

[edit] 6

6 The Morphological Analysis section requires some modification.

Methods: Z - overall writing of methodology section

- a. The diagrams are good and essential to the reader’s understanding.

- b. How frequently were the terminal leaflets of the ‘the main compound leaf unavailable and how similar to the terminal leaflet were the substitute leaflets? Such substitutions seem to invalidate many comparisons. DETAILS PROVIDED IN PAPER.

Terminal leaflets: Z - methodology.

- c. How was the PAR at sites so greatly separated in space measured? Did you calculate PAR from vertical looking hemispherical photographs? LOOKING BACK AT THE DATA, NO ACCURATE INFO ON PAR WAS OBTAINED (DUE TO DIFFERENCES IN CLOUD COVER ON SAMPLING DAYS).

PAR: Z - methodology.

- d. Were the measurements of leaves made on fresh leaves, on dried leaves or on dried leaves refreshed? If you did not use fresh leaves how near were the measurements you made to those made on fresh leaves? I do not understand “10 by 10 to the cm” graph paper”. Is this paper with lines every millimeter? Did you allow for the accuracy of the graph paper at the time of the scan (accuracy of printing and changes due to the dampness of the paper at time of measurement etc): over 6cm any discrepancy might have been significant? Is it right to expect the repeatability of measurements at 150 pixels/inch to be no better than 0.2mm: 10% of the smallest measurements made.

- e. What was the accuracy (repeatability) of measurements with the digital calipers? Did you have problems bringing the flower parts back to exactly their living shape and size with ethanol after they had been pressed and dried?

[edit] 7

7 Table 4.

Units: E - include units in all tables

Table 3 added "number" to frequency counts.
**Table 4 will add units for tables that remain after most are removed.

- a. The accuracy of measurement indicated in these tables seems always to be greater than the possible accuracy of your measurements. Plant height, with three replicates, given to 0.001mm, angles with 20 replicates to 0.01%, lateral leaflet width to 0.0001mm, and so on for all characters. The accuracies given in Table 3 seem usually more sensible (though not degrees).

Accuracy: E - see what data are regarding significant digits

significant digits: degrees to integers and 1 decimal place in std dev.

- b. The results for characters 31-35 seem to be multiplied by 10 — staminodium width at top given to 0.00001mm!

Accuracy: E - see what data are regarding significant digits

Table 4 is correct
**Table 3 needs correction of numerical summaries for characters 31-35.

[edit] 8

8 Figure 5.

- a. The reader will be looking to compare the proportion of each tip type. This comparison across populations would be easier to make if proportions were plotted rather than numbers. The total number of plants examined in the 16 populations could then usefully be placed at the top of each column.

Bar chart: E - how to better present this information

- b. The darker colours are difficult to distinguish in the columns. Use different fills and, possibly, wider columns.

Bar chart: E - how to better present this information

Created a new chart from TipType_plotdata.xls

[edit] 9

9 Variation.

- a. It is often useful to know separately the between population and within population variation. These are confounded in Table 3.

Added a column for the ANOVA F-stat for differences in means.  
This indicates the degree of between population variation compared to within population variation.

Table 3: EZ - Ask Lorna for a way to represent Between/Within variation information

- b. The Tukey-Kramer groupings surely suggest that there is little significant difference between population means except, not infrequently, the largest and the smallest.

Help express within variation for populations: E - include min/max for a measure of within variation for each population; EZ - maybe express to editor that not all tables need to be included, since many show little discriminatory information; EZ - think about cutting most of these tables, include the ones that have most discriminatory information

**Include only those Tukey tables that are referenced in text.
Corrected F statistics (as E found some mistakes he had made).

[edit] 10

10 Page 8 Relationship among populations.

- a. Was the variance of the size characteristics unified before the calculation of Euclidean distance? With size characters the variance is usually related to the mean so that larger Euclidean distances are obtained where the mean is larger. A Log transformation usually reduces this difficulty sufficiently.

Distance: E - Understand what reviewer means by "unified" (standardized?) - already written in methods section that the tree is using population means.; Z remembers using the mean of each character before drawing tree.

- b. Dendrograms are one method of squashing multidimensional space into two dimensions. Here complete linkage is used. Complete linkage usually “indicates” groups in the dendrograms even if any ‘groups’ in the data are very ill defined. By contrast dendrograms using single linkage tend not to indicate groups unless these are in the data. The use of group average methods gives intermediate results. One way to get over this is to calculate all three and see how many ‘groups’ are present in all three dendrograms.

I agree with reviewer's comments regarding linkage methods.

dendrograms: E - single vs complete linkage, Z picked method that gave the clearest separation; E - justification for creating the dendrogram in a certain way; E - UPGMA (define in text); Z - Citation (Sneath & Sokal 1973) dendrograms for taxonomy type of stuff; http://pubmlst.org/software/analysis/start/manual/upgma.shtml UPGMA (Unweighted Pair Group Method with Arithmetic Mean) (Sneath & Sokal 1973. Numerical Taxonomy. W.H. Freeman and Company, San Francisco, pp 230-234) is a straightforward method of tree construction. Its original purpose was to construct taxonomic phenograms, which are trees that reflect the phenotypic similarities between operational taxonomic units (OTUs).

After generating dendrograms by many methods, consensus is that there is little structure
using population means.  This is consistent with the univariate findings of the T-K comparisons.
Therefore, I think there is little evidence to conclude that there are clear subpopulations.
Rather, the within variation overwhelms the ability to detect differences between populations,
whether considering characters individually (T-K) or in ensemble (cluster).
I would present the (average, standardized, squared-Euclidean) dendrogram to illustrate.

- c. The production of a dendrogram does not “divide the populations in five groups”: the drawing of horizontal dashed line has been used to do this. Such a line, which need not be straight or horizontal since taxonomically equivalent groups seldom contain the same amount of variation, might have been drawn above (fewer groups) or below (more groups) your dashed line with equal validity as far as the diagram is concerned.

E - Why at 70%? Why one horizontal line? Why horizontal?

I agree that a dendrogram alone does not divide the populations, but a chosen horizontal line
does separate by a degree of similarity, and it does need to be horizontal.
This point is no longer relevant because dendrogram method reveals no subpopulations.

dendrograms groups: E - think about this

- d. The dendrograms may be taken to confirm your statement that the within population variation is not smaller than the between population variation since the samples from each population to seldom come together in your later dendrograms. I would use that to indicate that the ‘groups’ indicated in Figure 6 have no objective reality. However see the comments a-c above.

I agree here.

Figure 6 groupings: E - stronger argument for groupings in figure 6

Comment on paper, "for characteristics with large variability did you change the data to bring the data more towards normality (e.g. log [ ])"

No normality assumption, describe that clustering is based on points in space being joined by closeness... (elaborate)

[edit] 11

11 In the discussion of Ramsey’s study, p9, it is noted that he thought that 16 of the 2000 herbarium specimens he examined might be labeled as dissecta and that thus the present study might have expected to have found four plants that could be so labeled since 509 plants were examined. Surely the herbarium specimens would include few, if any, taken from a single population and one way to consider them is as samples from 2000 populations. Further it is likely that on occasion the plant collected for an herbarium is likely to have been the interesting, that is abnormal, plant in seen in the population. The plants in this paper are not comparable to herbarium specimens being selected using different criteria. DISCUSSION HAS BEEN REVISED TO PROVIDE A MORE REALISTIC EXPECTATION OF FINDINGS REGARDING WILD PLANTS AND HERBARIUM SPECIMENS.

[edit] 12

12 There is no pointer as to where the original information can be found and obtained. In an investigation of this sort different methods, and sometimes different algorithms purporting to use the same method, may give different results and interested readers must be able to access the original data.

Data availability: Z - Online supplement? Who owns the data? UMass, Lorna, Zoe? Under grant (Lorna under German conservation something)?; Z - Does UMass have a standard way of hosting data?; Hierarchy: Journal, UMass, StatAcumen

: ZZ - After discussing with Dr. Craker, he & I agreed that the best thing to do would be to make the data available upon request from the corresponding author as opposed to publishing online. UMass doesn't have any archive for this kind of thing, and it's easiest to maintain one copy that can be easily sent upon request.

Retrieved from "http://statacumen.com/community/wiki/index.php/Black_Cohosh_Revision"