Riccardo Ciarle, Victoria University of Wellington in Aotearoa New Zealand, discusses his article: Ancestral state reconstruction sheds new light on the loss of divarication hypothesis on New Zealand’s outlying islands
The background
If you step into the New Zealand bush, the first thing you’ll see will be towering podocarps, lofty tree ferns, and a wide array of epiphytes and climbing plants that cover every tree trunk. But in the understory and forest margins you will quickly notice something unusual: dense shrubs that look like tangled wires, with zig-zagging branches, tiny leaves, and small hard-to-reach flowers and fruits tucked inside the mass of branches.

These are divaricate plants: shrubs with tightly interlaced and wide-angled branches bearing clusters of small leaves, that have relatively long internodes compared to leaf size. Even among the many remarkable features of island life, divaricate plants represent one of the most striking examples of convergent evolution on islands. The origin and adaptive significance of divarication has been at the centre of a decades-long debate. One camp suggests it is the product of harsh climates while the other maintains it is a defence mechanism against browsing by moas, the giant avifauna that once roamed New Zealand.
Naturalists noticed that, while widespread on New Zealand’s main islands, divaricate plants are almost absent from New Zealand’s outlying islands. This led to the loss of divarication hypothesis, predicting that, because the divaricate habit is costly to maintain and no moa ever reached any outlying island, plant species should lose their divaricate-related traits after island colonisation.
Previous work tested the loss of divarication hypothesis using a pairwise comparison method, by comparing traits of island endemics to those of their closest relative(s) on the New Zealand “mainland”. This meant two things. First, that most congeneric species were ignored in the process. Secondly and most importantly, that all observed differences were assumed to be the result of island evolution, and that mainland relatives remained relatively unchanged since splitting from island lineages. In other words, it was always assumed that the ancestors of island endemics were divaricate at the time of colonisation.
The problem
Why doubt this assumption? Recent results indicate that divarication is actually a recent feature of the New Zealand flora and evolved after the outlying islands had already formed.
This indicates two possible scenarios: either a divaricate taxon colonised an outlying island and subsequently lost its divaricate features, or a non-divaricate taxon colonised an island and divarication subsequently evolved in its sister taxon on the mainland. In other words, divarication could have been lost on islands or never have been present in the first place.
Using discrete and continuous ancestral state reconstruction, we first tested whether ancestors of island colonisers were actually divaricate at the time of colonisation. Secondly, by measuring leaf size and branching angle, we tested whether island endemics actually evolved toward non-divaricate traits, i.e. larger leaves and smaller branching angles.
What we found
Most island endemics (21 out of 29 taxa) were never divaricate. Nonetheless, when the ancestor was divaricate, divarication was always lost. We also found that significant morphological evolution occurred in both island and mainland lineages, and that island endemics tended to evolve larger leaves and smaller branching angles regardless of whether their ancestor was divaricate.
The big picture
The loss of divarication hypothesis was overall supported, albeit much restricted in its scope. However, the loss of divaricate-related traits in most island endemics suggests that this pattern is better defined as a general trend of island evolution, rather than something restricted to descendants of divaricate species. More importantly, we showed the limits of the pairwise comparison method. Over the years, many aspects of island evolution, such as the island rule or the repeated pattern toward gigantism, have been investigated using a pairwise comparison method, and therefore assuming a lack of significant morphological change in the mainland relatives. While it can be reasonable to assume that island colonisers, exposed to novel selective pressures, will experience higher degrees of change than their mainland relatives, here we showed how ignoring mainland changes can lead to erroneous results, and highlight the need to move past the pairwise comparison method in favour of phylogenetic comparative approaches. Things change on islands, but things change on the mainland too!