(i) Preamble

Many of the issues that we have been discussing in zoogeography so far have been descriptive in nature with few good examples of testing experimentally or statistically predictions from alternative zoogeography hypotheses.

Analytical biogeography is essentially any technique or set of philosophical principles that can be applied to hypothesis testing in zoogeography. We will now focus on a kind of analysis that can test historical hypotheses and our next section will deal with experimental approaches to analytical biogeography.

(ii) Phylogenetic biogeography

This is a marriage of two disciplines: (a) phylogenetics, or the study and execution of historical reconstruction of lineages and their attributes, and (b) biogeography. In combination, therefore, phylogenetic biogeography is the study of historical (evolutionary) groups (taxa) and their interrelationships and the association of such groups with geography as a way of understanding the processes governing the current, past, and future distributions of these groups and their attributes.

Phylogenetic biogeography had its roots in attempts to try and resolve one of the major debates in biogeography: the role of dispersal (movements across barriers) and vicariance (geographic range-splitting) in explaining the distributions of organisms and their attributes (recall the early debates between "extensionists" and early proponents of continental drift).

The figure below shows the distribution of a group of fish (Galaxidae, southern hemisphere smelts). They have a clearly disjunct distribution (even shown by many individual species). These fish are typically diadromous and grow to maturity in freshwater, but spawn in salt marsh or estuarine areas (they can clearly tolerate saltwater for some portion of their life cycle). The young are often planktonic and may drift for long time periods in currents. Consequently, some have argued that their current distribution is due to past (and current) dispersal among the southern continents. Alternatively, others have argued that the groups found on the different continents are ancient lineages whose divergence was driven by vicariance; i.e. the breakup of Gondwanaland. The distribution of green sea turtles describe in out first lecture is another example of these alternative explanations for current biogeographic pattern. Phylogenetic biogeography played (and plays) a role in testing predictions of these alternatives.

The lower figure shows a schematic of how such tests could be done. The upper panel, left, shows a hypothesized phylogeny or "cladogram" (evolutionary relationships) among three taxa (1-3). The lower panel, left, shows the same cladogram, but with the current distributions of the three taxa mapped onto it (i.e. taxon 1 is found in area "A", etc.). This is known as an "area cladogram". The right side of each panel depicts dispersal (upper) or vicariant (lower) hypotheses that could account for the set of evolutionary relationships of the taxa and their current distributions.

The dispersal hypothesis suggests that the ancestor to taxa 1-3 existed only in area A and dispersed to area B. Subsequent isolation resulted in divergence of species 1 and the common ancestor of taxa 2 and 3 in areas A and B, respectively. Finally, the ancestor of taxa 2 and 3 dispersed to area C. Isolation then resulted in the divergence between taxa 2 and 3 (Note that the cladogram could also be explained by dispersal from A à C à B).

The vicariance model, however, suggests a single widespread ancestral lineage (found in areas A, B, and C). Range fragmentation resulted in isolation and the divergence of taxon 1 and the common ancestor of taxa 2 and 3. Subsequent range fragmentation produced areas B and C and drove the divergence between taxon 2 and taxon 3.

(iii) Vicariance biogeography

Vicariance biogeography is that branch of biogeography that considers vicariant events and hypotheses as outlined in the figure above as the major determinants of species distributions and their evolution. This school has the concept of "panbiogeography" formalized by Leon Croizat, an Italian botanist, in the 1950's, at its core. Considered as a "visionary comparative biologist" or part of the "lunatic fringe", Croizat sparked interest by plotting the distributions of related groups of endemic taxa on a map and connecting these areas with lines that he termed "tracks". When this was done repeatedly for unrelated groups of organisms, Croizat noted that the tracks tended to be the same (say between New Zealand/Australia and South America). He called these repeated patterns "generalized tracks" and suggested that they were the footprints of ancestral biotas that were at one time continuous in distribution, but which where now disjunct owing to large scale fragmentation events such as plate tectonics.

Croizat during a collecting trip in the upper Orinoco River.

As you can see from the figure below, Croizat perhaps went a bit overboard by covering the world in these generalized tracks.

A schematic of Croizat's generalized tracks.

Nevertheless, the concept of panbiogeography tended to promote collaboration between earth sciences and biology and so called vicarance or cladistic biogeography.

This school of thought gained prominence in the 1960's especially after evidence for vicariant processes such as continental drift became widely accepted. It is essentially a historical perspective on biogeography (although dispersalist hypotheses can also be historical, dispersal has a clearer relevance to current processes). How can vicariance be tested?

Note that the vicariance model makes a clear prediction about the relationships among the geographic areas, i.e. that area A "split-off" from areas B and C before the latter two areas diverged (geographically) from each other. This prediction can be tested by amassing geological or paleontological data that could inform the investigator about the sequence and dating of geographic range splitting. If the geological sequence of events is the same (or similar) to that predicted by the vicariance model, then that model would be supported as the evolutionary explanation for the divergence and distribution of the three taxa (1-3). This is the great strength, in theory, of the vicariance approach to biogeographic inference; it makes a prediction about a sequence of vicariant events that can be tested through the examination of independent data sets (i.e. geological information). In essence, vicariance biogeography looks for relationships (associations) between the phylogeny of organisms and the geological history of the areas they inhabit.

A key element to the whole approach is to obtain a reliable phylogenetic construction of the taxa in question. Phylogenetic systematics provided a popular (especially in more recent times) philosophy and methodology to obtain cladograms.

(iv) Phylogenetic systematics

  Phylogenetic systematics is a philosophy and methodology for the reconstruction of ancestor-descendant (i.e. evolutionary) relationships amongst a set of taxa. It was developed by a German entomologist, Willi Hennig, in the mid-1960s's. The basic tenants of phylogenetic systematics are:

Features used to reconstruct the evolutionary history of a group of taxa (i.e. their phylogeny) are modification of existing structures. Such features may include any feature of an organisms phenotype, but morphological traits or genetic attributes (DNA sequence variation) are most commonly employed.

The history of the changes in these "character states" reflects the ancestor-descendant relationships of organisms bearing the traits.

The distribution of traits among a set of taxa is used to identify "monophyletic groups" of taxa, i.e. taxa along a branch of a phylogenetic tree that all share a common ancestor (all descended from the same branching point on the tree). Monophyletic groups are also known as "clades" and phylogenetic trees as "cladograms".

Monophyletic groups are defined by the possession of shared derived traits (also known as "synapomorphies"). These shared, derived traits reflect the inheritance of modified ("evolved", "derived", "apomorphic") traits in two or more taxa from their common ancestor. All other taxa outside the monophyletic groups so defined possess the ancestral or "pleisomorphic" traits.

The logic here is that the shared derived traits reflect the pattern of changes from the ancestral condition and, therefore, must also provide a record of ancestor-descendant relationships of taxa bearing those traits.

The figure below depicts a cladogram of 5 taxa based on the analysis of variation in 5 characters, each with two character states (A, A', B, B'……E, E'). The unprimed letters represent the ancestral states, the primed letters represent the derived state of each character. The cross-hatches on the tree represent shared, derived traits, and the horizontal lines above the tree span a series of nested monophyletic groups. Note that some traits are derived, but they are NOT shared between any taxa (e.g. E', D'). They are, therefore, "uninformative" in terms of assigning relationships between two or more taxa (they can't be used to define monophyletic groups). Taxa that are descended from a common ancestral point are also known as "sister taxa" (i.e. taxon 5 is "sister" to the monophyletic group consisting of taxon 3 and 4. Also The monophyletic group of taxa 3-5 are a sister taxon to taxon 2).

Relationships amongst taxa are determined by comparing the number of shared, derived traits in a pairwise fashion. Those taxa that share the most traits are considered the most derived lineage. The table below shows a matrix of the distribution of shared derived traits among 4 "ingroup" taxa (the taxa whose interrelationships are the subject of study) and one "outgroup" taxon (a taxon that is known to have diverged from the common ancestor of all ingroup taxa before any of the ingroup taxa diverged from each other). Choice of the outgroup taxon is critical because it represents a more ancestral lineage than any of the ingroup taxa and is used to determined which of the character states (A, or A') is ancestral (possessed by the outgroup) or derived.

Note that the outgroup does not share any of the derived traits with any of the ingroup taxa. Taxa 3 and 4 form a monophyletic group as they share three derived traits, two of which (D', E') are not found in any other taxon, These are the most derived group and taxon 4 is the most derived single taxon as it possesses a fourth derived trait (F') that is unique to it. Taxa 1 and 2 are also a monophyletic group (defined by trait B'), but they only possess two derived traits in total; therefore, they are closer on the tree to the outgroup taxon.

In essence, this is how the phylogenetic trees are generated by using the phylogenetic systematic approach.

(v) Assumptions and caveats

While the phylogenetic systematic approach is appealing because of its combination of evolutionary philosophy and quantitative assessment of relationships, there are a few points that should be recognized about possible limitations.

Remember that a cladogram is simply an hypothesis about evolutionary relationships that is only as good as the data that generated it and the assumptions of the method. Any cladogram depicted is usually one of a family of possible alternatives.

Reversals of character states could confound evolutionary reconstruction. If a character changes for A --> A’, it’s at least possible that it could also change back from A’ --> A. If such "reversals" are common (as they might be with some DNA sequences) then groups that are not presently part of a monophyletic group, may actually be descended from a common ancestor. For instance, it the figure above, it’s possible that taxon 4 (with character state B) actually is part of a monophyletic group along with taxon 1 and taxon 2 if it also possessed B’ at one point (inherited from the common ancestor of 1, 2, and 4), but subsequently reverted back to state B by reverse mutation (in the case of DNA sequence data).

Parallel evolution. If a character state is obtained in two or more lineages independently (e.g. by two or more independent mutation) and not by inheritance from a common ancestor then this is known as parallel or convergent evolution. In such cases, the character in question obviously would give a misleading impression of common ancestry and confounds phylogenetic reconstruction. Think of paired fins in whales and fish. Clearly similar in outward appearance, but paired fins in these two lineages obviously did not arise from inheritance from a common ancestor, but through parallel evolution via (presumably) selection for features that enhance movement and orientation ina liquid medium.

Phylogenetic systematics assumes that reversals and parallel evolution are rare events in the data set at hand. The extent of such complicating factors can be estimated by various algorithms and decisions made about such possible cases. In general, alternative trees recognizing reversals or parallel evolution can be constructed (i.e. as a family of tree arrangements) and compared with each other and trees with no reversals or parallel evolution. Generally speaking, the trees that account for the interrelationships amongst taxa and which involve the fewest number of evolutionary changes (e.g. say DNA mutations) is the tree to be preferred under the common criterion of maximum parsimony. Simply stated, parsimony assumes that evolution always operates in the simplest way; if one evolutionary transition from one morphological structure to another requires four changes to bone structure and another only three changes, the simpler (three) model is considered the "most parsimonious" and most likely.

It is, indeed, arguable whether a tree with 27 mutations is really more likely than one with 28 mutations, and whether in fact evolution does always proceed along the "easiest" path, but that is the generally accepted criterion for evaluating alternative trees. The strength of the phylogenetic systematic approach invoking parsimony is that the assumptions and criteria for evaluating alternatives are "up front" and objective.

(vi) Applications of phylogenetic systematics in vicariance biogeography

(1) Area cladograms, continental drift and southern hemisphere midges

Lars Brundin (see Brundin 1988), a Swedish entomologist, conducted a phylogenetic reconstruction of southern hemisphere chironomid midges using morphological characters. The distribution (dark shading) of the various taxa he examined is shown below. Note their distribution at the southern margins of the continents and on several oceanic islands. 

 Shown below is the phylogenetic reconstruction of 8 genera of the midges. Brundin then mapped the distributions of each genus directly on the phylogeny and noted a striking pattern. First, the South African genera were basal on the tree ("SFA") implying they represent the most ancient divergence. In all other genera, taxa from New Zealand ("NZ") are more distantly related to genera from South America (SAm) and Australia (Au) which are most closely related to each other. The pattern of relationships amongst genera is very similar to the inferred sequence of breakup of the various components of Gonwanaland (evidence for which was amassing at the same time, but independently from Brundin’s work). Africa split from South America, Antarctica, Australia and New Zealand during the earliest stages of continental drift. New Zealand is considered to have separated from SAm and Au about 80 million years ago. These latter continents maintained a longer connection via Antarctica. The pattern of generic relationships, therefore, is exactly what one would predict of vicariance (via continental drift) drove the divergences among genera.

Given the evidence for the importance of vicariance discussed above, we should observe similar patterns of divergence in other (unrelated to midges) taxa that are distributed over a similar area. The cladograms below show generally concordant patterns of taxon relationships. Marsupials, birds, fish, reptiles and amphibians all tend to show closer similarity between taxa from Australia and New Guinea (which share the same continental plate) than to those from North America, South America, and (in some cases) Europe. Again, this pattern is consistent with the order of break-up of Pangaea and continents within Laurasia and Gondwanaland. These are known as "consensus cladograms" and strongly suggest that common historical factors (e.g. vicariant events acting on many taxa simultaneously) influence divergence within independently evolving taxa (with very different ecological characteristics, [e.g. dispersal capabilities and behaviour]). Note, however, the somewhat ambiguous affinities of North American fossil marsupial groups (they are aligned both with South America and Europe). Such ambiguity may stem from the influence of dispersal between S. and N. America (via the Central American landbridge).