Novel Tests of the Key Innovation Hypothesis: Adhesive Toepads in Arboreal Lizards.

The evolution of key innovations-unique features that enable a lineage to interact with the environment in a novel way-may drive broad patterns of adaptive diversity. However, traditional tests of the key innovation hypothesis, those which attempt to identify the evolutionary effect of a purported key innovation by comparing patterns of diversity between lineages with and without the key trait, have been challenged on both conceptual and statistical grounds. Here, we explore alternative, untested hypotheses of the key innovation framework. In lizards, adhesive toepad structures increase grip strength on vertical and smooth surfaces such as tree trunks and leaves and have independently evolved multiple times. As such, toepads have been posited as a key innovation for the evolution of arboreality. Leveraging a habitat use dataset applied to a global phylogeny of 2692 lizard species, we estimated multiple origins of toepads in three major clades and more than 100 origins of arboreality widely across the phylogeny. Our results suggest that toepads arise adaptively in arboreal lineages and are subsequently rarely lost while maintaining arboreal ecologies. Padless lineages transition away from arboreality at a higher rate than those with toepads, and high rates of invasion of arboreal niches by non-arboreal padbearing lineages provides further evidence that toepads may be a key to unlocking evolutionary access to the arboreal zone. Our results and analytical framework provide novel insights to understand and evaluate the ecological and evolutionary consequences of key innovations.


Supplemental
. Full dataset frequency distribution of gains and losses of toepads among 500 stochastic maps under ARD and ER (C, D) models -vertical dashed line indicates mean value, and horizontal black bar indicates 95% HPD. See Supplementary Table S3 for values of 95% HPD, minimum, median, mean, and maximum number of estimated changes between states for this reconstruction.
Supplemental Figure 4. Top row: Standard dataset frequency distributions for the gain and loss of toepads as simulated across 500 stochastic maps under an ARD model, and corresponding character mapping across the phylogeny. Bottom row: Standard dataset frequency distributions for the gain and loss of toepads as simulated across 500 stochastic maps under an ER model, and corresponding stochastic character mapping across the phylogeny. Vertical dashed line indicates mean value, and horizontal black bar indicates 95% highest posterior density interval (95% HPD). Grey circles denote padbearing species.
Supplemental Figure 5. Full Relaxed stochastic character mapping for arboreality evolution (nsim = 500) under an ARD model. The root state is fixed ("fossilized") for a non-arboreality probability of 1.00. Grey circles denote padbearing taxa.
Supplemental Figure 6. Full Relaxed stochastic character mapping for arboreality evolution (nsim = 500) under an ER model. The root state is fixed ("fossilized") for a non-arboreality probability of 1.00. Grey circles denote padbearing taxa. Supplemental Figure 7. Full Strict stochastic character mapping for arboreality evolution (nsim = 500) under an ER model. The root state is fixed ("fossilized") for a non-arboreality probability of 1.00. Grey circles denote padbearing taxa. Supplemental Figure 9. Standard Strict stochastic character mapping for arboreality evolution (nsim = 500) under an ER model. The root state is fixed ("fossilized") for a non-arboreality probability of 1.00. Grey circles denote padbearing taxa.
Supplemental Figure 10. Standard Relaxed stochastic character mapping for arboreality evolution (nsim = 500) under an ARD model. The root state is fixed ("fossilized") for a nonarboreality probability of 1.00. Grey circles denote padbearing taxa.
Supplemental Figure 11. Standard Relaxed stochastic character mapping for arboreality evolution (nsim = 500) under an ER model. The root state is fixed ("fossilized") for a nonarboreality probability of 1.00. Grey circles denote padbearing taxa.
Supplemental Figure 15. Full dataset frequency distribution from the multistate reconstruction of transitions among arboreality, semi-arboreality, and non-arboreality states among 500 stochastic maps under an all-rates differing model-vertical dashed line indicates mean value, and horizontal black bar indicates 95% highest posterior density interval (95% HPD).
Supplemental Figure 16. Full dataset frequency distribution from the multistate reconstruction of transitions among arboreality, semi-arboreality, and non-arboreality states among 500 stochastic maps under an symmetric rates model-vertical dashed line indicates mean value, and horizontal black bar indicates 95% highest posterior density interval (95% HPD).
Supplemental Figure 17. Full dataset frequency distribution from the multistate reconstruction of transitions among arboreality, semi-arboreality, and non-arboreality states among 500 stochastic maps under an equal rates model-vertical dashed line indicates mean value, and horizontal black bar indicates 95% highest posterior density interval (95% HPD).
Supplemental Figure 21. Standard dataset frequency distribution from the multistate reconstruction of transitions among arboreality, semi-arboreality, and non-arboreality states among 500 stochastic maps under an all-rates differing model-vertical dashed line indicates mean value, and horizontal black bar indicates 95% highest posterior density interval (95% HPD).
Supplemental Figure 22. Standard dataset frequency distribution from the multistate reconstruction of transitions among arboreality, semi-arboreality, and non-arboreality states among 500 stochastic maps under a SYM model-vertical dashed line indicates mean value, and horizontal black bar indicates 95% highest posterior density interval (95% HPD).
Supplemental Figure 23. Standard dataset frequency distribution from the multistate reconstruction of transitions among arboreality, semi-arboreality, and non-arboreality states among 500 stochastic maps under an ER model-vertical dashed line indicates mean value, and horizontal black bar indicates 95% highest posterior density interval (95% HPD).