Traits are physical quantities and trait measurements are reported with associated units. Units are the fundamental properties of data and without units or careful consideration can cause confusion and delay.  A synthesis of trait data in biology will rely on a common data standard . Cross-fertilization of trait-based approaches will rely on measurements that can be traceable and boils down to one or several of the universal physical standards (Calder 1982) – mass (m), time(t), linear dimensions (l), temperature (T) and the number or amount of a substance (n). Base units have the important property that all other units derive from them. In the Système International d’Unités or SI system, there are seven base units and corresponding physical quantities. This include, meter (m) for length, kilogram (kg) for mass, second (s) for time, kelvin (K) for temperature, ampere (A) for electric current, candela (cd) for luminous intensity, and mole (mol) for the amount of substance. All trait data can be reduced to these dimensions. However, confusion and errors in terms of recording and reporting of the units can underlie trait data and can be propagated through large trait compilations. 

Trait data are unique in that they are necessarily ‘rich in dimensions’.

Because of confusion and ignorance of measurement units, trait data are more prone to errors in recording, translation, and understanding. Indeed, trait data are unique in that they are necessarily ‘rich in dimensions’. That is, trait data involve measurements that can contain multiple SI base units and as a result can be measured and reported in multiple units. So, the same trait can be measured with differing units. Simple measures of size, area, and time are often reported in differing units but all can be related to more fundamental base units. For example, seed size can be measured in length or mass. Whereas the specific leaf area is measured in units of area per unit mass which can be reported as cm2/g or mm2/kg. Photosynthetic rate involves three SI base units e.g. mass per area per time and is often reported in units of umol CO2 cm^-2 sec^-1 or an amount per unit area per unit time. Multiple errors can emerge in reporting these units and in converting between different units. 

“Because of confusion and ignorance of measurement units, trait data are more prone to errors in recording, translation, and understanding. “

Because trait data are often rich in dimensions and associated units, when converting between two different units, it is imperative to make sure that you track the units you are converting and to make sure that the quantities are being converted properly. As an example, if you are converting an area, which is a length squared quantity, you must multiply by the conversion factor twice. Because trait data are dimensionally rich there are numerous potentials for error and error propagation.  In short, a more sincere effort is needed in making sure that units are recorded accurately, preserved accurately through proper metadata, converted properly, and understood correctly. 

Trait-based approaches need to place more careful attention to teaching, documenting, checking, and verifying the units of trait measurements.  While it does not matter what units we use for measuring a given trait or physical quantity, however, units are important when we deal with approximations, uncertainties, and errors.  Suppose we work with a trait measurement where the length scale is typically measured in mm but we have an approximation of 12.5 m to the exact value 12.52 m. The error in this case then 0.02 m. Switching units to mm leads to an error of 200 mm. A study working in mm would report 2 x10^3 as the error, while a study working in m would report 0.02 as the error. Thus, the absolute error is sensitive to the choice of units. As a result, knowing the original measurement units are essential and should motivate the use of relative error: (exact – approximate)/exact), since units then cancel. In the present example, one gets a relative error of 1.6×10−3 regardless of whether the length is measured in m or mm. But how often do we report the original units of measurement?

While these issues have long been known in ecology and similar pleas for a standard have been made (see Calder 1982) these calls have not been fully heard and assimilated by the trait-based community. For example, Calder’s paper has only been cited 6 times since 1982! I believe a recommitment to teaching, assessing, correcting, and standardizing of units could not be more pressing.  

I believe a recommitment to teaching, assessing, correcting, and standardizing of units could not be more pressing.  Thankfully there are several new tools.

There are several new tools to help understand and standardize units. The Parampool  package allows creation of user interfaces with support for units and unit conversion. Parampool supports various type of user interfaces: command-line arguments (option-value pairs), text files, and interactive web pages.

https://github.com/hplgit/parampool

H.P. Langtangen and G.K. Pedersen, Scaling of Differential Equations, Simula SpringerBriefs on Computing 2, DOI 10.1007/978-3-319-32726-6_1

Calder, William A. “A proposal for the standardization of units and symbols in ecology.” Bulletin of the Ecological Society of America 63, no. 1 (1982): 7-10.