Sample Phylogeny And Systematics Research Paper. Browse other research paper examples and check the list of research paper topics for more inspiration. If you need a religion research paper written according to all the academic standards, you can always turn to our experienced writers for help. This is how your paper can get an A! Feel free to contact our research paper writing service for professional assistance. We offer high-quality assignments for reasonable rates.
Systematics is the study of the relationships among organisms, whereas phylogenetics is the study of the evolutionary relationships among organisms. In that sense, phylogenetics is a subset of systematics. Historically, before evolution was accepted as the unifying principle in biology, phylogenetics did not exist per se. Instead, organisms were classiﬁed on the basis of their phenotypic similarity.
Academic Writing, Editing, Proofreading, And Problem Solving Services
Get 10% OFF with FALL23 discount code
1. Deﬁning The Questions
One of the main problems with using phenotypic similarity as a basis for classiﬁcation is that organisms evolve to become adapted to the environments in which they live (their niches) so that phenotypic convergence is a fairly common occurrence. For the most part, the longer ago in time that two organisms shared their most recent common ancestor, the more phenotypically dissimilar they will have become. However, phenotypic convergence as a consequence of adaptive evolution, along with rapid divergence as organisms adapt to totally novel environments, means that any classiﬁcation based on phenotypes inevitably confounds phylogenetic relationships and adaptive evolution. Given the evolutionary lability of behavioral characteristics, analysis of the adaptive or functional signiﬁcance of particular traits requires that the phylogenetic relationship is properly deﬁned so that evolutionary changes in behavior can be mapped on to those in niche use.
The advent of cladism as a technique aimed at identifying phylogenetic groupings through phenotypic analysis was a notable ﬁrst attempt to isolate phylogenetic groupings. Cladism, as initiated by Hennig (1966), is, however, based on a model of evolutionary change which excludes the possibility of phenotypic convergence because it assumes that the same character cannot evolve on more than one occasion and that once a character has evolved it cannot revert to its former state. More recently, the ﬁeld of phylogenetic analysis has been revolutionized by the rapid production of gene sequence data. Realistic models of molecular evolution are incorporated into phylogenetic analyses of various regions of the genome to produce phylogenies (evolutionary trees) of ever-increasing accuracy. How those trees will be used to produce classiﬁcations is not clear.
At the moment, recognized groupings (genus, tribe, family, and so on) are being revised so that they describe monophyletic groups (containing all the contemporary descendants of a particular common ancestor). However, the nominal groupings involved diﬀer according to historical precedence: systematists who originally produced the hierarchical classiﬁcations in use today ranged along an axis from lumpers to splitters. The former were less keen to classify species with a particular degree of morphological similarity into diﬀerent higher level taxa than were the latter. As a consequence, even when species within the hierarchical groupings are rearranged to represent monophyletic groupings, the most recent common ancestor for all the species in a particular genus would have been earlier in some orders than the most recent common ancestor for two species separated at the family level in other orders.
The development of this research paper introduces natural groupings, explains how phylogenies are determined, and then shows how phylogenies can be used to understand the functional signiﬁcance of behavioral diﬀerences among taxa.
2. Natural Groupings
It has always been clear to modern humans that there are natural groups of organisms: trees, birds, ﬂowering plants, spiders, and tortoises would be cases in point. It has also been clear, at least in more recent times, that there is a natural hierarchy of groups: there are animals versus plants, but within animals there are vertebrates and invertebrates, and within vertebrates there are birds, reptiles, mammals, and amphibians. The embodiment of the hierarchical natural classiﬁcation was produced by Linnaeus around 1800. No particular mechanism was envisaged for why this natural hierarchical classiﬁcation should exist, but one frequently stated reason for studying it was to provide insight into the mind of the creator. With the beneﬁt of hindsight, insofar as classiﬁcations were based on functional similarity, what were recognized to be the same character could in fact have separate evolutionary origins.
The major sea change occurred in 1859 with the publication of Charles Darwin’s Origin of Species in which he argued that life was created on one or few occasions and that subsequent speciation resulted in the natural hierarchy. Indeed, it is worth remembering that the only ﬁgure in the book is of an evolutionary tree. From that time on, with a few notable lapses, it has been accepted generally by working scientists that natural groupings are a consequence of evolution and that they should be monophyletic. In practice, however, it did not always turn out that way. For example, it is accepted that birds share a more recent common ancestor with crocodiles than either does with lizards. As a consequence, crocodiles should be classiﬁed with birds rather than with lizards. This means that the reptiles do not constitute a monophyletic group, but what is known as a paraphyletic group: a group made up of only those descendants of a common ancestor that resemble it (or each other) phenotypically.
Despite the acknowledged fact that the reptiles constitute a paraphyletic group which, if it were to be made monophyletic, should include the birds, there has been little serious eﬀort on the part of biologists to execute this particular taxonomic revision. To do so, would in some sense apparently cost more eﬀort than it is worth. Instead, reptiles without the birds are accepted as a paraphyletic natural group, while palaeontological and other evolutionary analyses should include birds within the group.
3. Identifying Phylogenetic Relationships
There have been many schools of systematics, and most are of merely historical rather than contemporary importance. The major division is into methods that analyze phenetic relationships vs. those that attempt to reconstruct phylogenetic relationships.
The development of phenetic methods became particularly popular in the 1960s and 1970s following the publication of Sokal and Sneath’s (1963) book entitled Principles of Numerical Taxonomy. The book outlined apparently objective methods of character scoring and the production of phenograms (trees that summarized degrees of phenetic similarity), using the then newly and widely available computers. The fatal ﬂaw to the whole enterprise was the absence of a model of phenetic evolution. There is an inﬁnity of so-called clustering algorithms and distance methods, each of which is appropriate for a particular model of phenetic evolution. Unfortunately, there was no objective way to decide which to use. It might be argued that the aim of numerical taxonomy or phenetics is not to determine evolutionary relationships but rather to describe phenetic similarity, but even then there is an inﬁnite number of distance methods and clustering algorithms, with no principled method of deciding on appropriate ones for all circumstances. The exercise foundered.
The use of phylogenetic methods was given a major boost by Hennig’s (1966) book Phylogenetic Systematics. Hennig realised that if the phylogeneticist was willing to make certain assumptions about character evolution, then phylogenetic trees could be constructed from data based on the distribution of character states among extant taxa. If every character evolved only once and evolutionary change was irreversible, then monophyletic groups could be deﬁned on the basis of shared derived characters. Problems arose when diﬀerent characters produced diﬀerent answers so that one character placed a particular species in one group while another character placed it in a diﬀerent group, which meant that at least one character did not evolve according to Hennigian principles. Common sense often prevailed among cladists to distinguish between, for example, characters that might have evolved once and been lost versus those that evolved more than once. However, once cladists began to entertain the possibility that there are diﬀerent models of evolution for diﬀerent characters, then the advent of a model-based statistical approach to phylogenetic reconstruction became inevitable. The problem recurs, just as it did with numerical taxonomy, that for phenetic evolution there is frequently no principled way to distinguish among alternative models of evolution.
In the years immediately before genes were sequenced at such low cost and with such speed, an informative stepping stone between phenetic and phylogenetic analysis occurred with the advent of DNA–DNA hybridization technology. The method was developed and widely applied by C.G. Sibley and co-workers. Radioactively labeled DNA strands about 500 nucleotides long from a focal species were hybridized with similar strands from a number of phylogenetic relatives. The rate of dissociation of these duplexes provided a measure of genetic similarity between the sequences, thereby excluding purely morphological convergence as a factor producing errors in phylogenetic tree reconstruction. Several surprises emerged. For example, among the birds, Australian passerines seemed to be more closely related to each other than they are to passerines from other continents, thus suggesting that previous classiﬁcations based on morphological comparisons had failed to detect considerable convergent morphological evolution. Sibley and Ahlquist obtained representative samples of many thousands of bird species, representing the large majority of tribes. They then subjected the data to DNA–DNA hybridization analysis to produce an estimated phylogeny of the birds based simply on the rate of dissociation of DNA–DNA hybrid duplexes. That phylogeny was used by Sibley and Monroe (1990) to produce a classiﬁcation of birds: categories of rates of duplex dissociation were arbitrarily deﬁned that were associated with each particular taxonomic status. Despite the various analytical diﬃculties of their method, and the arbitrary nature of their classiﬁcation, their procedure was claimed to be both objective and repeatable. As a ﬁrst attempt at a full phylogenetic classiﬁcation for a major taxon, it has probably yet to be improved upon. However, while most of the major revised phylogenetic relationships that Sibley and co-workers revealed have remained intact, more appropriate methods for phylogenetic reconstruction from gene sequence data are now available.
DNA–DNA hybridization methods provide just one measure of the distance separating a pair of sampled individuals. That distance depends on the genetic similarity between those individuals in terms of base sequences. However, the same distance measure can be achieved in many diﬀerent ways: if there are 20 base diﬀerences between individuals 1 and 2, and there are 20 base diﬀerences between individuals 2 and 3, that does not tell us how many diﬀerences there are between 1 and 3—just that there are between 0 and 40. All pairwise DNA–DNA hybridizations would need to be completed, even if we were accepting that DNA diverged at a metronomic rate through evolutionary time.
The production of vast amounts of gene sequence data means that every base pair becomes a character. Once we have a base sequence from a particular gene of a particular individual, it can be compared with that for every other individual for which we have equivalent data. This allows phylogenetic reconstruction using discrete rather than distance measures. What is more, we can describe models of molecular evolution with reasonable accuracy: we may know the relative substitution rates of the ﬁrst, second and third base pairs, and we often have a good idea of transition to transversion ratios (the frequency with which a substitution is between purine and purine or between pyrimidines, vs. between purine and pyrimidine). We also know that some base substitutions are eﬀectively neutral because they result in the same amino acid being coded for at the base-substituted triplet.
Before mentioning the most commonly used methods for discrete data, it should be pointed out that distance measures can be used and are used with gene sequence data. The methods diﬀer in whether they assume branch lengths from the root to the tips of the tree are all equal and adjust the data accordingly (e.g., unweighted group pair method with arithmetic means: UPGMA) or whether they allow diﬀerent rates of evolution in diﬀerent parts of the tree (e.g., the neighbor joining method which often gives a good approximation to the minimum evolution tree). De-spite the fact that information is lost (the evolution of individual base sites cannot be traced down the tree), there is also the problem that approximations are made which often result in biologically impossible branch lengths being reconstructed (see Page and Holmes 1998).
The two most common model-based discrete methods of model based phylogenetic tree reconstruction from gene sequence data involve parsimony and maximum likelihood.
Parsimony was introduced as the so-called minimum evolution distance method by Cavalli-Sforza and Edwards (1967) as a computationally rapid route for approximating maximum likelihood trees. However, as a discrete method for ﬁnding the shortest tree, it is possible to adapt strict parsimony and weight the changes in appropriate ways (e.g., a transition might be four times as frequent as a transversion) and determine that tree with the smallest total of weighted steps. There are several problems with parsimony as a procedure. The ﬁrst is philosophical—parsimony is a minimum criterion by deﬁnition and it may be argued that there is no minimum principle to evolution. The second major problem is that of ‘long branch attraction’ (see Fig. 1). Despite these disadvantages, parsimony remains a commonly-used procedure because of its computational speed.
Maximum likelihood is a very powerful tool in phylogeny reconstruction. The method asks which phylogenetic tree is the most likely one to have given rise to the set of gene sequences that we observe in our sample from the present day, given our speciﬁed model of gene sequence evolution. Indeed, it can go further and be used to estimate the most likely parameter estimates of our evolutionary model. The likelihood approach is increasingly commonly used for the reconstruction of molecular phylogenies, and as computers get ever faster and more powerful, it will probably become the approach of ﬁrst choice.
4. Using Phylogenetic Relationships To Analyze Behavior
When behavior is mapped on to phylogenetic tree structure, at least four types of informative analysis become possible. First, behavior itself may be seen to be responsible for phylogenetic tree structure. For example, species with particular modes of behavior may be more likely to divide into new species than others. Second, by mapping behavioral characteristics on to phylogenies, it is possible to determine which traits are ancestral and which are derived, and to determine which traits have changed most frequently over evolutionary time. Third, behavioral characteristics do not evolve independently of each other or of habitat use. If we are to understand why character covariation occurs, then phylogenetic tree structure must usually be an integral part of the necessary analysis. Fourth, adaptive radiations, such as Darwin’s Galapagos ﬁnches, are well known but when we map behavior on to tree structure, it sometimes becomes apparent that independent adaptive radiations have remarkably similar structures—when the tape of life is replayed, similar arrays of diversity result. In this section, examples of each of these four types of phylogenetically-based behavioral analysis are described.
4.1 Behavior And Speciation
When behavioral traits are mapped on to phylogenies, it is sometimes possible to identify so-called ‘key innovations’ which appear to have resulted in high rates of speciation. For example, cichlid ﬁshes are very speciose and their radiations seem to have coincided with a change in position of a particular muscle attachment which allowed the pharyngeal bones to manipulate individual prey items while still holding them. This released other areas of the jaw, which had been used for manipulating prey, to evolve for other functions. The cichlid radiations in African lakes are indeed characterized by a remarkable variety of feeding mechanisms. However, a problem with this explanation is that there may be other unrecognized characteristics unique to cichlid ﬁshes that may also contribute to their explosive radiations.
A more convincing demonstration of key innovations resulting in increased rates of speciation is when the same innovation appears independently in diﬀerent parts of the phylogenetic tree, and is statistically associated across a number of cases with an increased rate of speciation. One of the best examples of a repeatedly evolved such innovation is the ability of insects to feed on vascular plants, known as phytophagy. There is a wide variety of vascular plants, and whenever phytophagy evolved, there are likely to have been many vacant niches to speciate into. Mitter et al. (1988) identiﬁed 13 nodes in a phylogenetic tree of insects where all the descendant species from one branch are phytophagous, whereas all those down the other branch are not. In 11 of 13 cases there were more extant species descended from the phytophagous branch than the nonphytophagous branch, and in each of those 11 cases there were more than twice as many phytophagous than nonphytophagous species.
A ﬁnal example of behavioral changes apparently causing diﬀerences in rates of speciation comes from an original suggestion made by Darwin (1871). It is known that the females of some bird species actively choose their mate on the basis of phenotype. Darwin suggested that, when populations separate, the precise criteria for mate choice might alter so that if those separated populations came back together, mate choice would have promoted speciation. The most obvious criterion for determining whether females choose their mates by phenotype is whether males diﬀer from females in physical appearance. When passerine birds are scored as sexually dichromatic or not, then it turns out that the repeated evolution of sexual dichromatism tends to be associated with subsequently higher rates of speciation (Barraclough et al. 1995). It had taken more than 100 years before the phylogeny became available so that Darwin’s theory could be tested.
4.2 Character Evolution
Mapping character evolution on to phylogenetic trees can be used to test ideas about how and why characters evolve. In order to carry out this process, it is necessary to have a phylogenetic tree showing the relationships among contemporary species, character states for each species, and a model for character evolution. Since the appropriate model for the evolution of most characters is poorly understood, ancestral character states usually are inferred from either a parsimony or other appropriate minimum evolution criterion.
Gittleman (1981) set out to test the idea that when patterns of parental care evolved among ﬁshes, transitions were gradual and did not involve change by more than one parent. Twenty-one transitions were identiﬁed using parsimony on the phylogenetic tree, all of which were between biparental and maternal care, biparental and paternal care, no care and paternal care, and no care and maternal care. No cases involved the double transitions between no care and biparental care or paternal care and maternal care, a statistically signiﬁcant deﬁciency if all transitions were equally likely.
A more thorough and quantitative analysis of historical patterns in parental care, this time among shorebirds, was reported by Szekely and Reynolds (1995). Again they used parsimony for ancestral character state reconstruction, but in this taxon they found that there had been an overall evolutionary reduction in paternal care, which was sometimes but not always compensated for by an increase in maternal care. The authors argued that this is a response to selection for males to partake in extra-pair matings or even to maintain multiple pair bonds.
How rapidly does behavior evolve in comparison with, say morphology and life history? Haldane (1949) suggested that one appropriate measure of evolution change was the ‘darwin’: a change of one logarithmic unit per million years. Gittleman et al. (1996) mapped the evolutionary change of two behaviors (group size, home range size), two life history characters (gestation length, birth weight), and two morphological characters (adult brain weight, adult body weight) using data for extant mammal species from eight taxa for which there were reasonable phlylogenies. The behaviors were more labile than the morphological characters, which were more labile than the life history characters. Such analyses may well identify characters which, for whatever reason, are the most evolutionarily constrained, leaving others more free to change.
4.3 Correlated Character Evolution
Bringing order to organic diversity is made simpler by two organizing principles: close relatives are more similar to each other than are distant relatives, and organisms share adaptations to common environments. As we have seen when attempting to reconstruct phylogenetic trees, it can be important to distinguish between these two reasons for similarity. However, the two principles cannot necessarily be separated: because close relatives are often adapted to very similar ecological niches, they also share many adaptations in common, whether those adaptations were inherited from a common ancestor or not. If we wish to understand the extent to which similar adaptations are likely to arise in similar circumstances, it is nevertheless important to factor out identity by common descent. What is more, because close relatives share characters which may not be functionally related to each other (e.g., mammals have hair and give birth to live young, whereas birds have feathers and lay eggs), treating species values as independent points in a statistical analysis can result in functionally meaningless signiﬁcant associations between characters as a consequence of pseudo-replication.
4.3.1 Independent contrast analysis. In order to tackle this problem, Felsenstein (1985) devised a method which traced the change in two or more characters on a phylogenetic tree, and deﬁned for each character equivalent evolutionary independent contrasts which could be used for data analysis. Given character states for contemporary species, ancestral character states had to be inferred from a speciﬁed model of evolutionary change. Felsenstein used a Brownian motion model of change to estimate ancestral character states. In essence, all change that occurred between sister taxa evolved independently of change between other sister taxa. The consequence is that diﬀerences between the actual or estimated character states for sister taxa are used as the independent contrasts, which are standardized by branch lengths (variances of contrasts are larger when the time separating taxa has been longer). The direction of comparison is arbitrary, though it is conventional to order the taxa so that all contrasts for the independent variable are positive (or zero). The contrasts for the dependent variable(s) then use the same order of taxa and can take both positive and negative values. The contrasts for the dependent variable(s) are then regressed on those for the independent variables, and conventional statistical techniques (correlation or regression) are used to test for correlated evolutionary change. The null hypothesis is that change is uncorrelated, and signiﬁcant departures from the null hypothesis can indicate either positive or negatively correlated character change. A hypothetical example is worked through in Fig. 2.
Although independent contrast analysis is widely used, statistical conclusions can, in fact, change when diﬀerent models of character change are entertained. For example, Harvey and Purvis (1991) report a study of the evolutionary relationship between a measure of running activity and relative foreleg length (foreleg length corrected for body size) in a group of Anolis lizards. Three models of evolutionary change were considered (Brownian motion, punctuation, and minimum evolution), only the ﬁrst of which produced a signiﬁcant relationship between contrasts for the two variables being studied (p < 0.025 compared with p > 0.10 for the other two models). Fortunately, diagnostics are available to test whether the assumptions of Brownian motion character evolution have been approximately met, given data for a set of contemporary species and a well-resolved phylogeny (Garland et al. 1992). However, when the hypothesis of Brownian motion character evolution is rejected, there remains a paucity of statistically powerful comparative methods.
4.3.2 Sister Taxon Comparisons. When a phylogeny is available, but there is no obvious model of character evolution, then the technique of sister taxon comparison can prove very useful. The idea is to compare pairs of contemporary species, chosen such that each ‘sister pair’ shares a more recent common ancestor than either member does with any member of any other pair. The result is that the diﬀerence between each pair of sister species has evolved independently of the diﬀerence between any other pair, and ancestral character states do not need to be estimated. The loss of assumptions results in fewer possible comparisons: if there are n contemporary species in a sample, so long as the phylogeny is bifurcating, then there are n 1 possible independent comparisons, but at most only n 2 sister comparisons.
Møller and Birkhead (1992) have championed the use of sister taxon comparisons, and illustrated their use by testing the idea that high copulation rates among birds are a ‘paternity insurance’ mechanism. First, they asked whether rates of extra-pair copulation increased with local population density, which they did in all eight possible comparisons, which is a highly signiﬁcant result (n.b. 15 potential comparisons are possible with 16 taxa when ancestral states are considered). Second, they asked whether birds living in colonies had higher rates of intra-pair copulations, which they did in 12 of 13 comparisons.
4.4 Parallel Structure Of Independent Radiations
When a species invades a completely new habitat with unoccupied niches, there is the potential for an adaptive radiation to ensue. But how deterministic is evolution? Stephen Jay Gould suggested that contingency was such an important factor that, if we replayed the tape of life, we should invariably get diﬀerent outcomes. That is, of course, true sensu stricto. However, early comparisons such as that between the placental and marsupial mammal radiations, and the very fact of evolutionary convergence, means that it is not the whole story. Recently, a number of case studies suggest a remarkable degree of similarity when the opportunity for similar radiations has been presented.
Anolis lizards in the Caribbean have invaded many small and a few large islands. On the very smallest islands, there tends to be one intermediate-sized species, but as island size increases large and small forms can coexist. Each of the largest islands supports an independently evolved Anolis radiation, with about seven diﬀerent species occupying a variety of well-deﬁned ecological niches. To the casual human observer there has been remarkable convergence, for example, the species living on twigs from diﬀerent islands are well-nigh indistinguishable. Natural selection has led to the evolution of the same so-called ecomorphs evolving on diﬀerent islands. The twig ecomorphs have a morphology that is adapted to the niche in which they live. Other ecomorphs live in the crowns of trees, on the trunks, on the ground, and so on.
While it may be easy to fool a casual human observer, would the same independently evolved ecomorphs be able to fool each other? That experiment has not been performed on Anolis lizards, but it has in a simpler system involving stickleback ecomorphs. At the end of the Pleistocene, after the glaciers had retreated from British Columbia, a marine stickleback invaded a number of freshwater lakes. In those lakes, parallel speciation occurred: a large benthic ecomorph evolved which feeds on invertebrates from the bottom of the lake, together with a smaller limnetic ecomorph which darts around in the open water living oﬀ plankton. There is strong assortative mating such that the benthic and limnetic ecomorphs will only mate with their own type in the wild, and have become true species. But would opposite sex individuals of the same independently evolved ecomorph from diﬀerent lakes accept each other as mates? The answer is yes, in the laboratory. Rundle et al. (2000) took limnetic and benthic species from three separate lakes. Over 750 mating trials were set up in which individual females were presented with single males of the same or the alternative ecomorph, from the same or a diﬀerent lake. Females were far more likely to spawn with a male of the same ecomorph as themselves, in which case it mattered not at all whether the male was from the same or a diﬀerent lake.
- Barraclough T G, Harvey P H, Nee S 1995 Sexual selection and taxonomic diversity in passerine birds. Proceedings of the Royal Society London, Series B 259: 211–15
- Cavalli-Sforza L L, Edwards A W F 1967 Phylogenetic analysis: Models and estimation procedures. American Journal of Human Genetics 19: 233–57
- Darwin C 1859 On the Origin of Species by Means of Natural Selection, or, The Preservation of Favoured Races in the Struggle for Life. John Murray, London
- Darwin C 1871 The Descent of Man and Selection in Relation to Sex. John Murray, London
- Felsenstein J 1978 Cases in which parsimony or compatibility methods will be positively misleading. Systems in Zoology 28: 49–62
- Felsenstein J 1983 Parsimony in systematics: Biological and statistical issues. Annual Review of Ecology and Systematics 14: 313–33
- Felsenstein J 1985 Phylogenies and the comparative method. American Nature 125: 1–15
- Garland T J 1992 Rate tests for phenotypic evolution using phylogenetically independent contrasts. American Nature 140: 509–19
- Gittleman J L 1981 The phylogeny of parental care in ﬁshes. Animal Behavior 29: 936–41
- Gittleman J L, Anderson C G, Kot M, Luh H-K 1996 Comparative tests of evolutionary lability and rates using molecular phylogenies. In: Harvey P H, Leigh Brown A J, Maynard Smith J, Nee S (eds.) New Uses for New Phylogenies. Oxford University Press, Oxford, UK
- Haldane J B S 1949 Suggestions as to quantitative measurement of rates of evolution. E olution 3: 51–6
- Harvey P H, Purvis A 1991 Comparative methods for explaining adaptation. Nature 351: 619–24
- Hennig W 1966 Phylogenetic Systematics. University of Illinois Press, Urbana, IL
- Mitter C, Farrell B, Wiegmann B 1988 The phylogenetic study of adaptive zones: Has phytophagy promoted diversiﬁcation. American Nature 132: 107–28
- Møller A P, Birkhead T R 1992 A pairwise comparative method as illustrated by copulation frequency in birds. American Nature 139: 644–56
- Page R M, Holmes E C 1998 Molecular Evolution: A Phylogenetic Approach. Blackwell Science, Oxford, UK
- Rundle H D, Nagel L, Boughman J W, Schluter D 2000 Natural selection and parallel speciation in sympatric sticklebacks. Science 287: 306–8
- Sibley C G, Monroe B L 1990 Distribution and Taxonomy of Birds of the World. Yale University Press, New Haven, CT
- Sokal R R, Sneath P H A 1963 Principles of Numerical Taxonomy. Freeman, San Francisco
- Szekely T, Reynolds J D 1995 Evolutionary transitions of parental care in shorebirds. Proceedings of the Royal Society of London, Series B 262: 57–64