BB eye: Getting our lists in order?

Published on 12 June 2017 in Editorials

By Jeremy Greenwood

Many of us will share Mark Holling’s frustration with frequent changes to the order in which bird families and species are placed in lists (Holling 2016). I agree that the changes come about because taxonomists want to use the ‘best’ list, so that the sequence is changed in the light of new opinions about the avian evolutionary tree. But the problem lies deeper than this, for even if we knew for sure the tree in all its details there would still be an astronomically large number of possible ways of expressing this as a simple list, all of them equally valid.

The sequences with which we are presented are based on two criteria: placing closely related species close together and placing more primitive species before more advanced ones. Unfortunately these are incompatible because species that are not closely related may be equally primitive, just as two branches on opposite sides of a tree may be equally close to the ground. Lists will therefore differ according to which criterion is given priority. We could get around this problem by abandoning ‘primitiveness’, which is in any case a dubious concept in that all species have been evolving for the same span of time. However, the bigger, and insoluble, problem is that the criterion ‘closely related near to one another’ does not produce unequivocal linear sequences. Think of the tips of all the branches on a tree. We may wish to list them in a sequence with the closest together on the tree being neighbours on the list. But this is impossible, for we are then trying to compress a three-dimensional array into a one-dimensional list: given that we have several thousand tips (species) on the avian evolutionary tree, the number of possible ways of arranging them is incomprehensibly large. That the evolutionary tree is not three-dimensional but multi-dimensional exacerbates the problem.

Fig. 1. The possible sequences arising from a phylogeny of four species (identified by symbols). A phylogenetic tree is shown on the left. Column A to the right of that shows the four species resulting from the three evolutionary splits (numbered). Seven other orderings of the four species (B–H) are all equally congruent with the tree, differing from each other merely in the arbitrary way in which the branches at the three splits are flipped on the diagram.

Modern phylogenetic trees, such as the simple one in fig. 1, show the evolutionary branching across the page, with the species lying at the ends of the branches in a neat order down the page. Surely that provides a robust sequence – star, triangle, square, circle (fig. 1, column A)? Unfortunately it does not, because the order in which the species resulting from a split are placed is arbitrary: branching point 2 could be expressed either as it is or it could equally be flipped so that circle comes above square, as in column B; a flip at point 3 (but not at 2) produces sequence C; flips at both 2 and 3 produce sequence D; and a flip at point 1 produces four further sequences (E–H), depending on the flips at 2 and 3. Eight equally justifiable sequences for only four species! Considering an evolutionary tree of all bird species together, we are faced with a bewildering number of possible arrangements, with no reason to choose one rather than another. When some evolutionary splits are into more than two species, the problem becomes even greater.

The only way to overcome this problem is to use unequivocal, if arbitrary, criteria to produce the sequence. Moreau (1961a,b) and Lack (1968) suggested basing the sequence on the standard order of the alphabet. That is, using scientific names, the orders are placed in alphabetical order of their names, as are the families within each order, the genera within each family, and the species within each genus. Groupings above the ordinal level (such as subclass and superorder) or between the other taxonomic levels (such as subfamily and subgenus) could be used to give closer reflection of the phylogenetic tree should taxonomists feel the picture to be securely enough known. Established taxonomists (e.g. Bock 1968) have argued against this proposal but none has addressed the fundamental impossibility of reducing a multidimensional array to an agreed linear sequence when there is an astronomically large number of possible linear sequences.

A more biologically based system, promoted for birds by Winkler et al. (2015), would be at each point of split to place the branch that has the smaller number of descendant species above the branch that has the larger number.

An argument used against ordering based on the alphabet or on numbers of descendant species is that changes in taxonomy – such as the merging of two families into one, or the splitting of a genus into several – will affect the sequence. But this applies to any conceivable system, as history shows for the traditional method of deciding on sequences (Lack 1968; Holling 2016). I have a vague memory of hearing a lecture in the 1960s that showed that if the alphabetic method had been used for European birds during the previous half-century there would have been greater stability than had been delivered by the various sequences based on traditional practices. Is it not time for the IOC to put the fundamental methodology to the test by repeating this comparison, this time including also the method of ‘the less speciose branch comes first’?

Even if the method that would have provided the greatest stability is chosen, there will still be changes consequent both on genuine changes in knowledge and changes in fashion with respect to lumping and splitting taxa. Currently the IOC adjusts its list at quarterly intervals (Gill & Donsker 2017). Wouldn’t it be better for ornithology if the IOC were to switch to providing a World List that was changed every quarter of a century, alongside more frequent publication of taxonomic decisions (to be taken into account in the next revision of the list)? The list could be used by publishers of handbooks and field guides, the updated information by taxonomists.

Jeremy Greenwood is a former Director of both the BTO and British Birds, and has a particular interest in ornithological statistics.

References

Bock, W. J. 1968. The sequence of bird lists. Ibis 110: 368.

Gill, F., & Donsker, D. (eds.) 2017. IOC World Bird List (v 7.1). doi:10.14344/IOC.ML.7.1

Holling, M. 2016. Getting your lists in order. Brit. Birds 109: 2–3.

Lack, D. 1968. The sequence in European bird lists. Ibis 110: 107–113.

Moreau, R. 1961a. In what order should birds be listed? Bird Notes 30: 18–32.

–– 1961b. Taxonomic realism. Proc. Zool. Soc. Lond. 137: 623–626.

Winkler, D. W., Billerman, S. M., & Lovette, I. J. 2015. Bird Families of the World: an invitation to the spectacular diversity of birds. Lynx Edicions, Barcelona.