Friday, July 17, 2015

The first null model war in ecology didn't prevent the second one*

The most exciting advances in science often involve scientific conflict and debate. These can be friendly and cordial exchanges, or they can be acrimonious and personal. Scientists often wed themselves to their ideas and can be quite reluctant to admit that their precious idea was wrong. Students in ecology often learn about some of these classic debates (Clements v. Gleason; Diamond v. Simberloff and Connor), but often other debates fade from our collective memory. Scientific debates are important things to study, they tell us about how scientists function, how they communicate, but more importantly by studying them we are less likely to repeat them! Take for example the debate over species per genus ratios, which happened twice, first in the 1920s, then again in the 1940s. The second debate happened in ignorance of the first, with the same solution being offered!

To understand the importance of testing species-genus ratios we can start with a prediction from Darwin:

As species of the same genus have usually, though by no means invariably, some similarity in habits and constitution, and always in structure, the struggle will generally be more severe between species of the same genus, when they come into competition with each other, than between species of distinct genera (Darwin 1859)

To test this hypotheses, the Swiss botanist, Paul Jaccard (1901) created a ‘generic coefficient’ to describe biogeographical patterns and to measure the effects of competition on diversity. The generic coefficient was a form of the species-genus ratio (S/G), calculated as G/S x 100, and he interpreted a low S/G ratio (or high coefficient) to mean that competition between close relatives was high, and a high ratio (low coefficient) meant that there was a high diversity of ‘ecological conditions’ supporting closely related species in slightly different habitats (Jaccard 1922). At the same time as Jaccard was working on his generic coefficient, the Finnish botanist, Alvar Palmgren, compiled S/G patterns across the Aland Islands and inferred the low S/G values on distant islands to reflect random chance (Palmgren 1921). Over several years, Jaccard and Palmgren had a heated exchange in the literature (across different journals and languages!) about interpreting S/G ratios (e.g., Jaccard 1922, Palmgren 1925). Palmgren’s contention was that the S/G ratios he observed were related to the number of species occurring on the islands –an argument which later work vindicates. A few years after their exchange, another Swiss scientist, Arthur Maillefer, showed that Jaccard’s interpretation was not supported by statistical inference (Maillefer 1928, 1929). Maillefer created what is likely one of the first null model in ecology (Jarvinen 1982). He calculated the expected relationship between Jaccard’s generic coefficient and species richness from ‘chance’ communities that were randomly assembled (Fig. 1 –curve II). Maillefer rightly concluded that since the number of genera increase at a slower rate than richness, the ratio between the two couldn’t be independent of richness.

Jaccard’s generic coefficients plotted by Maillefer showing the relationship between the coefficients (calculated as Genera/Species x 100) and species richness (Maillefer 1929). The four curves depict different scenarios. Curve I shows the maximum values possible, and curve IV is the minimum. Curve III is when coefficients are calculated on sampled values from a flora, which stays on a mean value. Curve II represents the first null model in ecology, where species are randomly sampled (‘hasard’ is translated as chance or luck) and the coefficient was calculated from the random assemblages.

 This example is especially poignant because it foreshadowed another debate 20 years later –and not just in terms of using a null expectation, but that S/G ratios cannot be understood without comparison to the appropriate null. Elton (1946) examined an impressive set of studies to show that small assemblages tended to have low S/G ratios, which he thought indicated competitive interactions. Mirroring the earlier debate, Williams (1947), showed that S/G ratios were not independent of richness and that inferences about competition can only be supported if observed S/G values differed from expected null values. However, the error of inferring competition from S/G ratios without comparing them to null expectations continued into the 1960s (Grant 1966, Moreau 1966), until Dan Simberloff (1970) showed, unambiguously, that, independent of any ecological mechanism, lower S/G are expected on islands with fewer species. Because he compared observationed values to null expectations, Simberloff was able to show that assemblages actually tended to have higher S/G ratios than one would expect by chance (Simberloff 1970). So not only is competition not supported, but the available evidence indicated that perhaps there were more closely related species on islands, which Simberloff took to mean that close relatives prefer the same environments (Simberloff 1970).

Darwin, C. 1859. The origin of the species by means of natural selection. Murray, London.
Elton, C. S. 1946. Competition and the Structure of Ecological Communities. Journal of Animal Ecology 15:54-68.
Grant, P. R. 1966. Ecological Compatibility of Bird Species on Islands. The American Naturalist 100:451-462.
Jaccard, P. 1901. Etude comparative de la distribution florale dans une portion des Alpes et du Jura. Bulletin de la Societe Vaudoise des Sciences Naturelle 37:547-579.
Jaccard, P. 1922. La chorologie selective et sa signification pour la sociologie vegetale. Memoires de la Societe Vaudoise des Sciences Naturelle 2:81-107.
Jarvinen, O. 1982. Species-To-Genus Ratios in Biogeography: A Historical Note. Journal of Biogeography 9:363-370.
Maillefer, A. 1928. Les courbes de Willis: Repar- tition des especes dans les genres de diff6rente etendue. Bulletin de la Societe Vaudoise des Sciences Naturelle 56:617-631.
Maillefer, A. 1929. Le Coefficient générique de P. Jaccard et sa signification. Memoires de la Societe Vaudoise des Sciences Naturelle 3:9-183.
Moreau, R. E. 1966. The bird faunas of Africa and its islands. Academic Press, New York, NY.
Palmgren, A. 1921. Die Entfernung als pflanzengeographischer faktor. Series Acta Societatis pro Fauna et Flora Fennica 49:1-113.
Palmgren, A. 1925. Die Artenzahl als pflanzengeographischer Charakter sowie der Zufall und die säkulare Landhebung als pflanzengeographischer Faktoren. Ein pflanzengeographische Entwurf, basiert auf Material aus dem åländischen Schärenarchipel. Acta Botanica Fennica 1:1-143.
Simberloff, D. S. 1970. Taxonomic Diversity of Island Biotas. Evolution 24:23-47.
Williams, C. B. 1947. The Generic Relations of Species in Small Ecological Communities. Journal of Animal Ecology 16:11-18.

*This text has been modified from a forthcoming book on ecophylogenetics authored by Cadotte and Davies and published by Princeton University Press

1 comment:

Dr. Fox said...

Great post. I never knew any of this pre-1960s history.

Of course, I'd argue that even with an appropriate null model, you can't infer competition from species/genus ratios. Because there's no such thing as an appropriate null model here--one that tells you what the species/genus ratio would be if interspecific competition were somehow turned off while leaving everything else unchanged. Randomizing the observed data definitely isn't the appropriate null, no matter what constraints you place on the randomization.