Each card gives the problem, what the right handling is, and each run's verdict with the reason.
1
Two date systems
Old Mac Excel counts days from 1904, Windows Excel from 1900. The same number in a cell gives dates four years apart. Right: every date lands on the day it was actually sampled (stand 31 = 21 Sep 1973).
Our key was wrong here. It said to read each workbook in its own date setting. But the one "1900" workbook with dates (the swamp soils file) actually holds 1904 numbers; read our way, all 13 of its stands come out four years early. All three runs have correct dates.
2
Old-Mac line endings
One text file (ALLT.DATA) uses old Macintosh line breaks, so a normal read sees the whole file as one line. Right: parse it, or show it duplicates a spreadsheet you already have.
R
right, silentParsed it correctly by character position, but its notes never mention the line-ending issue.
3
Zero cover where the plant was in quadrats
988 ground-layer records have the species in at least one quadrat but a cover of 0 or blank. Right: keep the 0 as trace cover, and flag it.
R
flagged, rightKept the zeros and added a cover_missing_or_trace column.
Our key was wrong here. It called the zeros missing data. The field sheets show a tick with no number meant trace cover, and the authors' own column totals for those species are 0 (false nettle at stand 193: 24 of 25 quadrats, total cover 0).
4
Seen in the plot but in no quadrat
Some species were recorded in the plot but fell in none of the 25 quadrats. Right: keep them as present, not drop them.
R
flagged, rightCover for these: 0.
R-rep
flagged, rightCover for these: blank.
R and R-rep both kept these 425 records but disagree on their cover (0 vs blank). Both are defensible, so neither is scored wrong. This is also the one choice the notes-only rebuilds got wrong for both A and R.
5
Logs and rocks are not plants
Eleven "species" codes are logs, rocks, water and similar. Right: separate them from the plants.
A
flagged, rightMarked 11 entries "not a plant taxon".
R
right, silentSeparated them, but only mentions it in passing in the report, not in its decisions.
6
= means a synonym
DRYOPTERIS AUSTRIACA=spinulosa is the data manager writing two names for one fern, not doubt between two. The spinulosa half names the plant actually growing here (D. carthusiana). Right: keep it at species, and say which half matters.
R
flagged, rightRecorded it in a synonym-note column.
7
/ is a real either-or
LONICERA XYLOSTEUM/TATARICA means the collectors could not tell two honeysuckles apart. Right: keep both candidates; do not pick one.
R-rep
flagged, rightGave these a "species group" rank.
8
A code starting "SP" is not always "sp."
ACSP is Acer spicatum (mountain maple), not "Acer sp." Right: keep it at species rather than applying a prefix rule.
No run fell for it, and no run mentioned it.
9
One plant, two codes
The same shrub is LOXT in the ground-layer file and LOXY in the woody file; eight such pairs. Right: link them, so combining the layers cannot count one plant as two species.
A
silent breakGave the two codes different species IDs and never said so. Combine the layers and the shrub counts twice.
R
flagged, rightLinked them, with the reasoning in its decision log.
R-rep
wrong, flaggedDid not link them, but its notes raise the issue.
10
Stands with no woody data
19 stands have an empty row in the woody file. The authors' own summary sheet marks 15 of them "no trees" (cattail marshes, sedge meadows, young fields). Right: 0 for those 15; missing only for the few genuinely unsampled (47, 49, 87).
R
wrong, flaggedDid not settle it: treated the empty rows as unresolved and asked the collectors which were truly empty.
Our key was wrong here too. It first called all 19 unsampled. Corrected before scoring, after we found the "no trees" notes.
11
Two environment sheets
Site environment is split across two sheets holding different stands (111 together). Right: combine both, and record which sheet each row came from.
12
The column headed Dry is drainage
In the relevé sheet, Dry is the 1 to 6 soil drainage class (1 = very poorly drained). Right: identify it as drainage.
R-rep
wrong, flaggedRead it as a 1 to 5 "dryness scale", marked unconfirmed.
13
Two kinds of pH
Stand pH is one number measured in the plot. Relevé pH is a range copied from the county soil survey for the mapped soil type. Right: keep them apart and say where each came from.
A
not reachedNever read the relevé pH.
14
One code, two plants
The mirror image of row 9. The woody file and the ground-layer file each have their own code list, and six codes mean a different plant in each: COAM is silky dogwood in one and hazelnut in the other. Right: look codes up in the list for the file they came from.
A
flagged, rightFound it. Our own earlier cleaning had missed it, which is why it is not in the key.
R
flagged, rightKept a separate code list per file.
R-rep
silent breakUsed one code list for everything, decided dogwood and hazelnut were the same plant, and merged them without saying so.