Taste of quals: on the relationship between diversification and key innovations

This is the last post on the series “Taste of Quals” with examples of questions and answers given in the qualifying exams of the Ecology, Evolution and Systematics program at the University of Missouri St Louis. In this post you’ll find another answer written by Robbie Hart (find more about him here and here) in which he explains why, when and how key innovations can be associated to evolutionary diversification. The reason why I love this text and think it’s a great read is that even though this was written back in 2010, and since then the ‘omics’ era has overloaded us with tons of data, we are yet to understand several of the relationships between innovations and diversification that Robbie mentions. Enjoy the read!

Describe the relationship between diversification and key innovations

For half a century, rapid evolutionary diversifications have been conceptually linked to key innovations. Especially in the case of adaptive radiations, features associated with the diverging lineage have been held to have a causal effect in the diversification. However, the generality of this phenomenon has come under criticism from empirical, theoretical, and philosophical standpoints. Recent, careful studies have utilized detailed phylogenies and quantitative comparisons with null models, connected proposed innovations to genetic and developmental mechanisms, and examined replicated cases of radiations, to move the concept of the key innovation beyond “plausible suggestions” (1).


In the Simpsonian model of evolution, the essence of diversification is entry into a new adaptive zone, or way of life complete with characteristic adaptive pressures (2). Entry is only possible if three criteria of access are met: physical or geographic access, ecological access (the zone must be empty of competitors), and evolutionary or morphological access. To a large part, Simpson’s definitions have been retained. The first two criteria are now often lumped as ‘ecological opportunity’ (3, 4, 5) in contrast with the final criteria, now often termed a ‘key innovation’. Of the many definitions given for key innovation (Hunter lists seven (6), a small subset), Galis gives one that is in accord with modern ideas of macroevolution: “A key innovation is an innovation which opens up a new character space (or breaks constraints) that potentially allows the occupation of more niches” (7). Key innovations, then, open up adaptive space and expose many characters of an organism to a diverse new suite of selective pressures.

Other definitions accentuate different aspects of key innovations, but there are two that especially contrast with this definition. The first is a definition of key innovations in a phylogenetic context taken to mean any trait responsible for increasing diversification (8). The emphasis is explicitly on tempo rather than mode. Analyses of proposed key innovations differentiate this tempo and various mode definitions more or less (or not at all). The second is the concept of correlated progression, which proposes that coordinated suites of traits are the key features (6). Correlated progression is in clear contrast with the ‘constraint-breaking’ definition of Galis; but not incompatible with the tempo-only definition. At its root, correlated progression is getting at the individuation of key innovations, discussed in more detail below. For now, I’ll say that increasing phylogenetic resolution may ‘smear’ what we think of as one key innovation into several – a concept that has similarities to that of correlated evolution.

Necessary and sufficient conditions for diversification

The biggest potential problem with the concept of key innovations, the idea that they are a necessary and sufficient condition for an adaptive radiation, is easily dispelled. Examples showing the contingency of any causation between key innovations and diversification include the drilling radula in naticid gastropods. The drilling radula is an excellent candidate for a key innovation – it opened the door to many opportunities for specialized predation on shelled bivalves, and was associated with a major diversification of naticid lineages. However, the same feature was found to have evolved previously in the Triassic, and quickly disappeared for the extent of the Jurassic (9). Perhaps the most famous adaptive radiation of all, that of Rift Valley lake cichlids, has been attributed in part to the key innovation of recruiting the pharyngeal jaw for food processing, freeing the oral jaws for diversification (10) – a nice example of a constraint-breaking key innovation. However, Embiotocids also have pharyngeal jaws and Tilapiine cichlids share both structural and behavioral traits with the diverse clades of cichlids, but neither the Embiotocids nor Tilapiine cichlids have undergone a comparable adaptive radation (7).These examples illustrate that key innovations per se are at best necessary but not sufficient to spur diversification. This is the case even when it is combined with a an open niche, as with the Triassic drilling naticids.

Andean lupines, which show record levels of fast speciation and morphological change without identified key innovations or ecological opportunity (3), and plethodontid salamanders, which similarly exhibit an incredible burst of lineage diversification, but without key innovations, particular ecological opportunity, or even significant morphological difference (11), are two examples that bring the necessity of key innovations for rapid diversification into serious doubt. Hunter defends key innovations, however, as important conditions for radiations:
“ ‘environmental challenge’ cannot be sufficient to produce a radiation because, in the absence of [available ecological space ability to use that space via key innovations], any environmental challenge that destroys a species’ habitat is likely to result in extinction not radiation” (12).

All of this raises the question – are adaptive radiations even something that we need key innovations to explain? Are they a “thing” at all? Raup, Gould, Schopf and Simberloff published one of the early null models for phylogenetic diversification, and questioned the notion that there deterministic forces like key innovations significantly structured phylogenetic diversity at a macroevolutionary timescale (13). More recent work is often carefully tested against null models (1, 11, 12) , but criticisms on that basis remain (5), including a recent and suggestive model showing that many of the same phylogenetic patterns interpreted as adaptive radiations could in fact be produced by cryptic mass extinctions (14). Of course, sometimes the comparison with a null model finds no differences pervasive enough to indicate key innovations. This is the case for passerine birds (11), and presumably many more unpublished, as suggested by Donoghue (8)).

Conceptual criticisms: individuation, methodology, diversity patterns

Cracraft (15) offers a comprehensive critique of key innovations, citing them for failing on three themes: ontological (non-individuated ‘innovations’), methodological (failure to trace genetic and developmental mechanisms), and empirical (inappropriate comparisons with clade diversity measures). The first point questions whether a proposed evolutionary novelty is in fact a single character with an individuated identity independent of the observer, or whether it is merely a typological construct. Cracraft offers the example of avian flight, a frequently proposed key innovation that certainly offers access to a new adaptive zone, and a host of structures to differentiate and elaborate. However, flight requires a host of characters, stretching back across 50 million years of evolution to Archaeopteryx or before; it’s hard to argue that this amalgamation is a discrete innovation.

Nevertheless, Bond and Oppel could be doing just that when they write “If a key innovation is defined as the appearance of a new capability that facilitates the proliferation of the lineage that possesses it, then one or more characters may contribute to the key innovation. In the case of character complexes, a key innovation is not functional and therefore not present until all of its components are present. Thus, the key innovation appears at the point in a group’s phylogeny where the last of the functionally linked suite of characters appears” (1). Theirs is more a definitional assertion than an argument. Donoghue has a slightly different take on the issue – working with a tempo definition (Bond and Oppel’s was based on apomorphy) he notes that with increasing phylogenetic resolution many traits that researchers formerly lumped as a key innovation are now best understood as “sequences of character change, no one element of which can cleanly be identified as … responsible for shifting diversification rate”(8). Donoghue, pursuing Cracraft’s methodological theme, notes that this shift in what we think of as a key innovation often leads to surprising developmental features as the first steps towards key innovations – for instance, at the root of the compound feature “macrophyllous leaves” we find overtopping growth.

Photo by Alan Cressler.

Yucca moths in yucca flower. Photo by Alan Cressler.

Pellmyr and Krenn’s work on yucca moths (16) elegantly addresses Cracraft’s methodological theme (as well as the ontological and to some extent the empirical). Building off of a large body of prior research, they identify a unique limb (the yucca moth tentacle); associate it with the adaptive radiation of yucca moths, show that it is integrally related to the mechanism driving that radiation, ie. the pollinating co-evolution with yucca1; convincingly argue for the developmental origin of the tentacle as a heterotopic expression of a proboscis element; hypothesize on a genetic basis for the change; and show that it occurred as an abrupt change at the base of the pollinating yucca moth clade.

Finally, Cracraft’s empirical theme explores the difficulty in connecting an evolutionary novelty causally with patterns of species diversity, including problems of: assuming that higher taxa are comparable, making an arbitrary choice of a certain rank of taxa to compare, ignoring counterexamples, correlating a proposed key innovation and number of species, and qualifying clades as “diverse” or “not diverse” rather than quantifying diversity. This final theme has its roots in the very nature of key innovations. They are novelties, unique or at least rare, and this makes it extremely difficult to test causation. “Suppose that we agree wings are a key adaptation of bats. How can we show that they are responsible for there being ca. 870 species as against, say, 87 or 8,700?” (Raikow 1988 qtd in (10)). Cracraft specifically criticizes the comparative- functional argument as based on a correlation driven by an arbitrary choice of taxonomic or phylogenetic level of comparison. This is an issue that shares some conceptual similarity to the concern about null models. Bond and Opell (1) attempt to address the empirical theme by identifying unbalanced bifurcations in a semiautomated, a priori way across a well-resolved tree of all spiders, and only then locating the functional innovations at the nodes that emerge as exceptional; however their basic procedure remains qualitatively what Cracraft criticizes: “to qualify as a key innovation, our analysis requires that a feature: (1) be a synapomorphy; (2) be functionally advantageous; and (3) be capable of facilitation a change or an expansion of adaptive zone [and be associated with one side of an unbalanced bifurcation].”

Another way to address the empirical theme is through replicated examples of key innovations driving adaptive radiations. Unfortunately, replicated adaptive radiations themselves are few, in part because of the necessary conditions for geographic replicates. Islands and island- like habitats are promising; but candidate ‘archipelagos’ must be small enough to be replicated2 and at the same time large enough to show any speciation at all3. Caribbean anoles (17), Galapagos snails (4, 18), Mesozoic semionotid lake fish (19), and Hawaiian spiders (20) are rare examples; but are not strongly associated with key innovations (although toe pads and dewlaps have both been suggested for anoles (21)). Rift Valley lake cichlids, are one of the few examples of a replicated adaptive radiation with a fairly well-supported key innovation – the pharyngeal jaw (10).


Even the classic cichlid example, however, demonstrates the highly contingent nature of the connection between diversification and key innovations. The failure to radiate of closely related groups that share the pharyngeal jaw and other innovations with the species-rich groups of cichlids (7); phylogenetic reconstructions that show sexual selection, habitat selection, and trophic diversification as each driving a separate mini-radiation (10); and the possibility that much of the cichlid speciation may be due to peripatric speciation in geologically ephemeral satellite lakes and multiple colonizations from riverine lineages (19, 22) all emphasize this contingency.

Haplochromis (Pundamilia) nyererei, one of the species that is part of the outstanding species pool of Lake Vitoria, which contains more than 500 cichlid taxa.

Haplochromis (Pundamilia) nyererei, one of the species that is part of the outstanding species pool of Lake Vitoria, which contains more than 500 cichlid taxa.

De Queiroz (23) formalizes this contingency by breaking it into three parts: the effect of other taxa; the effect of other traits; and the effect of the environment. ‘Other taxa’ may be seen as more or less equivalent to Simpson’s ‘ecological access’ – one reason that a key innovation could arise and not lead to adaptive radiation is lack of an empty niche. De Queiroz advances image-forming eyes as a possible example of this contingency; the first three clades (vertebrates, arthropods, and cephalopods) in which this innovation emerged all diversified greatly; but it has subsequently evolved 15 times in various taxa without spurring comparable radiations (23). A less hypothetical example is presented in Hawaiian tertragnathid spiders, where ecomorph niches are present on an island or volcano either through immigration or diversification (20). No ecomorph is represented by two sympatric species, so a niche filled by immigration will presumably not spur radiation.

‘Other traits’ gets at individuation and complexes of traits. Functional modules may constrain evolution of their parts (24); and key innovations may often be, or result from, decoupling features from modules to, in effect, add parameters for diversification – this is seen in a number of key innovations from cichlid jaws (7) to avian flight musculature (6). At the same time; features that have a proximal connection to fitness (eg. macrophyllous leaves) may spring from from distal innovations (eg. overtopping growth) (8). Some distal innovations (baupläne) may offer much better ways of ‘solving’ evolutionary ‘problems’ than others – Donoghue gives the example of repeated convergence on ‘tree’ life forms in plants that seem to have not spurred radiations or been particularly efficient until it occurred in lineages with bifacial cambium.

Finally, ‘environment’ is clearly a key contingency; and is often associated with other biogeographic factors that may swamp patterns of adaptive radiation driven by key innovations; through patterns produced by other mechanisms of speciation (as in the possible peripatric speciation of cichlids) (22); ‘tier II’ or ‘tier III’ historical phenomena such as species selection and mass extinction (perhaps responsible for the disappearance of the Tertiary drilling naticids (9)); or environmental changes such as global CO2 change (which affects the ‘keyness’ of the C4 innovation (23)) and climatic cycles (which drove repeated local mass extinctions of radiating semionotid lake fish (19)).

The relationship between diversification and key innovations could, perhaps, be summed up in one word: ‘contingent’. I’m inclined to think of this a more of a ‘profound insight’ than a ‘truism’ (23); and agree with the forecasts of de Queiroz (23), Donoghue (8), and Losos (4); that a careful study of how key innovations can drive diversification is rich ground for evolutionary insight, even lacking a strict deterministic framework.


1 The related Raven-Ehrlich hypothesis of key antagonistic/defensive coevolutionary synchronization driving adaptive radiation through escalation has received somewhat equivocal support (25, 26).

2 Continent-scale comparisons are of course limited in number.

3 One of the most intriguing patterns to emerge from studies of island speciation is a threshold island area, below which no inferred speciation events are seen (18).

Works cited

1. Bond JE, Opell BD (1998) Testing Adaptive Radiation and Key Innovation Hypotheses in Spiders. Evolution 52:403.

2. Simpson G (1953) Major Features of Evolution (Columbia University Press, New York).

3. Hughes C, Eastwood R (2006) Island radiation on a continental scale: exceptional rates of plant diversification after uplift of the Andes. Proceedings of the National Academy of Sciences of the United States of America 103:10334-9.

4. Losos JB (2010) Adaptive Radiation, Ecological Opportunity, and Evolutionary Determinism. The American naturalist 175.

5. Masters J, Rayner R (1998) Key innovations? Trends in Ecology & Evolution 13:281.

6. Hunter JP (1998) Key innovations and the ecology of macroevolution. Trends in Ecology and Evolution 13:31-36.

7. Galis F (2001) in Character Concept of Evolutionary Biology, Wagner GP, pp. 581-605.

8. Donoghue MJ (2005) Key innovations, convergence, and success: macroevolutionary lessons from plant phylogeny. Paleobiology 31:77-93.

9. Fürsich FT, Jablonski D (1984) Late Triassic Naticid Drillholes: Carnivorous gastropods gain a major adaptation but fail to radiate. Science 224:78-80.

10. Danley PD, Kocher TD (2001) Speciation in rapidly diverging systems: lessons from Lake Malawi. Molecular ecology 10:1075-86.

11. Kozak KH, Weisrock DW, Larson A (2006) Rapid lineage accumulation in a non- adaptive radiation: phylogenetic analysis of diversification rates in eastern North American woodland salamanders (Plethodontidae: Plethodon). Philosophical transactions of the Royal Society of London. Series B, Biological sciences 273:539-46.

12.  Ricklefs RE (2003) Global diversification rates of passerine birds. Proceedings. Biological sciences / The Royal Society 270:2285-91.

13. Raup DM, Gould SJ, Schopf TJ, Simberloff DS (1973) Stochastic models of phylogeny and the evolution of diversity. Journal of Geology 81:525-542.

14. Crisp MD, Cook LG (2009) Explosive radiation or cryptic mass extinction? Interpreting signatures in molecular phylogenies. Evolution 63:2257-65.

15. Cracraft J (1990) in Evolutionary Innovations, Nitecki MH, pp. 21-44.

16. Pellmyr O, Krenn HW (2002) Origin of a complex key Innovation in an Obligate Insect-Plant Mutualism. Science 99:5498-5502.

17. Losos JB (1998) Contingency and Determinism in Replicated Adaptive Radiations of Island Lizards. Science 279:2115-2118.

18. Losos JB, Parent CE (2010) in The Theory of Island Biogeography Revisited, Ricklefs RE, Losos JB (Princeton University Press, Princeton), pp. 416-438.

19.  McCune AR, Thomson KS, Olsen PE (1984) in Evolution of Fish Species Flocks, Echelle A, Kornfield I, pp. 27-44.

20. Gillespie R (2004) Community Assembly Through Adaptive Radiation in Hawaiian Spiders. Science 303:356-359.

21. Jackman T, Losos JB, Larson A, de Queiroz K (2000) in Molecular Evolution and Adaptive Radiation, Givnish TJ, Sytsma KJ (Cambridge University Press), pp. 535-557.

22. Genner MJ et al. (2007) Evolution of a cichlid fish in a Lake Malawi satellite lake. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 274:2249-57.

23. De Queiroz A (2002) Contingent predictability in evolution: key traits and diversification. Systematic biology 51:917–929.

24. Wagner GP, Pavlicev M, Cheverud JM (2007) The road to modularity. Nature reviews. Genetics 8:921-31.

25. Agrawal AA, Lajeunesse MJ, Fishbein M (2008) Evolution of latex and its constituent defensive chemistry in milkweeds (Asclepias): a phylogenetic test of plant defense escalation. Entomologia Experimentalis et Applicata 128:126-138.

26. Berenbaum MR, Favret C, Schuler MA (1996) On defining “key innovations” in an adaptive radiation: cytochrome p450S and Papilionidae. The American Naturalist 148:S139-S155.

Hope that's you now, after having read all these posts on how to nail your quals!

Hope that’s you now, after having read all these posts on how to nail your quals!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s