A little room to wonder

Species Records & Field Notes

A species record links an observation or specimen with information about what was identified, when and where. Keeping uncertainty and record identity visible prevents a copied entry becoming extra evidence.

How can four rows contain fewer than four independent observations?

A European honey bee visits a purple flower, with some legs and body surfaces hidden.
This existing photograph has a named species and credited source. Its partly hidden features show why an image-based record should retain identification references rather than pretending every feature was directly counted.John Severns (Severnjc), via Wikimedia Commons. Public domain by author dedication. No edits. · Source · Public domain — author dedication

A record needs context

An observation becomes more useful when its date, mapped place, method and identification basis stay with it. A species name without these details cannot answer every question. An identification may be tentative, and a count may not have been made even when an organism was noticed.

Biodiversity databases bring together records from many sources. The Atlas of Living Australia explains that automatic checks cannot identify every error and that original data providers remain important. A record appearing in a large collection is a starting point for checking, not a guarantee that every field is correct.

Worked example: separate copies from observations

Consider four invented paper records. R1 says Plant A, grid B2, 1 April, three individuals, identification confirmed for the example. R2 repeats R1 with the same original observation identifier. R3 says Plant A?, B2, 8 April, four individuals, identification uncertain. R4 says Plant A, B3, 8 April, count not recorded, identification confirmed.

R2 is a copy of the R1 observation, so these rows describe three distinct observation events, not four. Two events have confirmed Plant A labels; R3 remains uncertain. Only R1 supplies a confirmed individual count. Treating the blank in R4 as zero or adding the uncertain four would manufacture a different dataset.

Correct visibly rather than erase the trail

If a label is revised, keep the original record identifier, the correction and its reason connected. Otherwise an apparent change in distribution may be an editing change. A date of upload is also not necessarily the date of the observation.

Confidence is not a substitute for evidence. Useful notes say what supports the identification and what cannot be seen. An unresolved label can remain broader, such as Plant A-like form in this invented exercise, instead of being forced into a specific species name.

Useful notes do not need private details

A paper practice record can use an invented grid, fictional date and made-up plant label. It does not need an observer’s real name, a home address or an exact wildlife location. The purpose here is to reason about data quality, not to submit records to an online service.

A tidy record helps another reader see the limits of a conclusion. Three different events can still be too few to establish a population trend. Before comparing totals, check duplicates, missing fields, changes in method and the confidence of each identification.

Only if you feel like trying

Clean a fictional record set

  1. Write R1 to R4 from the supplied example. Draw a link between rows that refer to the same original observation.
  2. Mark identification confidence and missing counts separately. State how many distinct events and confirmed labels remain.
  3. Write a short correction note for R3 if its identification later changes to Plant B. Keep the earlier label and reason for revision visible.

Optional paper data practice. Do not upload records, identify real locations, contact an observer or collect organisms for this activity.