Abstract

The use of behavioral traits as characters in reconstructing phylogenies is a relatively recent phenomenon. Early attempts in this regard were fraught with criticisms regarding the ability to accurately identify homology among what appeared to be highly variable behavioral characteristics. As a result, most behavioral evolutionary studies have been restricted to first deriving phylogenies from non-behavioral characters. These phylogenies have then been used to tract the evolution of various behaviors. This early reluctance to use behavior in character matrices of phylogenetic reconstruction has been partially overcome in recent years - largely as a result of the interpretation of Remane’s criteria (for determining homology) in behavioral terms as well as from the development of more vigorous methods for determining phylogenies. Roughly three dozen attempts have been made using cladistic techniques incorporating behavioral traits – nearly a fourth of these have been arthropod studies. It has been shown that for arthropods, many behavioral traits typically show better phylogenetic resolution than do morphological traits. This resolution has been determined using consistency indices for both kinds of traits. Fairly stable phylogenies have resulted from analyses using wasp nest architectural characteristics such as nest pedicel position and structure, relative position of combs, shape of the envelope, position of the entrance, and dome architecture. Among bees, the location of the nest site, and drone cell construction are behavioral traits that have been effectively used. Significant emphasis has been placed in an important study of web construction and design among orb-weaving spiders. Among the Orbiculariae, characteristics such as web hubs, relative length of the web radius, spiral web construction, web stickiness, and silk texture have all been successfully used characteristics. Among water mites, male behavior in presenting spermatophores as well as positioning of spermatophores are characteristics that have been used. The connection between morphology and behavior is clearly evidenced in the Orthoptera. Among the Ensifera there is a strong correlation between singing and the anatomical characters of tegmina and tibial tympana, both of which have been successfully utilized in phylogenetic reconstructions. Though limited in number, these and other examples have shown the potential for using behavior in attempts to reconstruct evolutionary history.

Introduction

Since Darwin's publication (1859) of The Origin of Species by Means of Natural Selection, taxonomists have largely understood similarity among organisms to be due to proximity of descent. With the exception of popular nature writing on animal behavior and anthropomorphic interpretations of the natural world, writers since Darwin have recognized that similar behavioral traits among organisms were probably due to the sharing of similar evolutionary histories. Much of this early literature is comprised of ad hoc explanations of probable evolutionary scenarios and essentially no attention is given to the use of behavioral characteristics in the creation of classifications. During the mid part of this century, as a result of what has become known as the Modern Synthesis in systematic biology (which was essentially a new understanding of evolution based on the combined sciences of genetics, paleontology, and evolutionary biology), scientists began superimposing behavioral traits onto existing phylogenies. These kinds of studies were motivated by the desire to learn evolutionary pathways of different kinds of behavior and not to use the behavioral traits as part of the evidence for actually making evolutionary trees. This procedure required an already determined hypothesis of descent, or phylogeny (most commonly determined from morphological data). Using this method, behavioral traits are aligned with the corresponding taxa on dendrograms (branching diagrams of hypothesized evolutionary history) to see if the behavioral traits had evolved in any sort of non-random array among the various taxa involved. Implicit in this methodology is the confidence of morphological characters and the nearly universal (though usually tacit) suspicion of the usefulness of behavioral traits as tools in resolving evolutionary history. An example of this type of research is the study by Howard Evans (1966) on sand wasps. Evans, after describing in detail the biology of genera of sand wasps (subfamily Nysonninae), builds a hypothetical evolutionary tree from morphological data. After the tree is constructed, the various behavioral traits are then superimposed on to the tree. In this way, Evans suggests that the provisioning behavior in sand wasps proceeds through an evolutionary process beginning with a primitive behavior of mandibular prey carriage followed by construction of cocoons without pores, followed by progressive stages of nest provisioning, to the most advanced stage where wasps oviposit in empty cells.

Most of the literature today (that will not be discussed in this review), discussing issues of behavioral evolution, continues to follow this pattern of first constructing evolutionary trees before considering the evolution of specific behavioral patterns. The only major difference between recent studies and earlier research in this regard is the more sophisticated and repeatable phylogeny-producing algorithms available today that were not available to earlier workers. Relatively few systematic studies have used behavioral characteristics in the actual construction of cladograms. The efforts that have been made are fairly recent and have largely followed the newer developments of phylogenetic analysis as well as the evolutionary understanding gained in recent years of behavioral homology.

Homology and Behavior

Homology between features, since the time of Darwin, has most often been understood to exist when the given features are derived from the same (or corresponding) feature of their common ancestor. Much discussion has ensued over the types of evidence that can be used in determining homology; however, three principle criteria (recognized by A. Remane and translated by Atz, 1970) have become recognized as important in determining homology. These criteria are (1) position (homology results from the same position in comparable systems of structures), (2) special quality (similar structures can be homologous without regard to position if they share several unusual characteristics), and (3) constancy or continuity (dissimilar and differently located structures can be homologous if intermediate connecting forms exist). Atz (1970) has argued that these criteria preclude behavioral traits, making the determination of behavioral homology difficult or impossible in most cases. The key to this criticism is the nature of structure and function. Atz argues that most functions are easily associated with some sort of structural entity as, for example, digestion with enzymes or immunity with antibodies. Behavior, it is argued, is all too often treated as an abstraction without reference to structure making it essentially intractable. Other problems in homologizing behavior include the impossibility of using the fossil record, the supposed high incidence of behavioral convergence, and problems associated with determining behavioral homology among higher categories of organisms. These criticisms have been countered by Wenzel (1992) for all three criteria of Remane. The criterion of position is defined as "similarity in topography" with the example of tail wagging in Tilapia given where the behavioral trait occurs in exactly the same place in courtship ceremonies. The criterion of special quality is considered applicable to behavior if innate complex movements take the same distinct form in the same context. The criterion of constancy or continuity becomes readily applicable to behavior in cases such as are manifest with complex ritual behaviors where many species with intermediate behavioral characteristics are evident.

Kinds of Behavioral Characters

Several kinds of behavioral characters have been used in phylogenetic reconstructions. These characters have included nest architecture (including site location), courtship and territorial behavior, parental care, reproductive behavior, feeding, locomotion, general lifestyles, and types of social systems (Wimberger and de Quieroz, 1996). From these kinds of data, two broader categories have been suggested including the end products of behavioral acts and data gathered directly from observation of behavior. In a study of these two kinds of data in black flies, Stuart and Hunter (1998) compared the final cocoon construction in these aquatic insects with the behavior during the construction process of the cocoons. The later data were taken from video recordings of cocoon construction. Cladograms were generated from both sets of data and compared to a cladogram generated from morphological and cytological data. The tree of end-product behavior (cocoon morphology) showed a high level of internal consistency (meaning very few character reversals or convergences) but showed significant conflict with the morphological/cytological tree. This was explained on the low number of characters used in the cocoon morphology reconstruction. The cladogram generated from actual behavioral patterns, on the other hand, showed nearly complete congruence with the morphological/cytological tree, with only minor internal inconsistencies. It appears that, at least for black flies, behavioral acts are more useful for phylogenetic reconstructions than are end-products of behavior.

The superiority of actual behavior over end-products of behavior in cladistic analyses has been hinted at in a study of gall forming thrips in Australia (Crespi et. al., 1998). In this study, cladograms were generated from four types of data and evaluated both separately and together. One of the data sets included behavioral and morphological characters together while another data set included only galls (end-products of behavior). Of the four data sets compared, the gall information was the least informative in resolving branching patterns. These two cases showing evidence against the utility of end-products of behavior are contrasted with the apparent success of earlier approaches relying on these kinds of data for bee cell formation (Alexander, 1991), spider webs (Coddington, 1990), and wasp nests (Wenzel, 1993).

How Good are Behavioral Characters?

In an effort to evaluate the validity of claims made by Atz (1970) and others, that behavioral traits should not be used as phylogenetic evidence, Wimberger and de Quieroz (1996, using data from de Quieroz and Wimberger, 1993) brought together evidence from several taxonomic studies that had used behavior in determining phylogenies. Wimberger and de Quieroz evaluated these cladograms with and without the behavioral characters included and determined the levels of homoplasy (evolutionary convergences and reversals) in both data sets. Using Consistency and Retention Indices (values that reflect homoplasy) the authors were able to show that in most cases behavioral characters resulted in evolutionary trees with less convergences and reversals than did morphological characters. Among arthropods, the studies evaluated normally included very few behavioral characters (except for a study of polistine wasps) which might be partially the reason for the good results; however, these data do suggest that there is ample justification for further use of behavioral traits in cladistics.

Examples of Behavioral Characters Used in Cladistic Analyses

In an effort to provide a more robust evolutionary explanation to Martin Landauer's classic study of the evolution of the dance behavior in bees of the genus Apis, Alexander (1991) provided an analysis of Apis phylogeny using both morphological and behavioral characters. A total of 20 characters were used in the phylogenetic reconstruction of the six species (including one species group) of bees in the study. Of the 20 characters, two were behavioral. These included location of the nest within (or outside of) a cavity and the presence or absence of a central pore in the capping of drone cells. In Alexander's best phylogeny (of the many produced in standard cladistic studies) the derived character of locating a nest outside of a cavity evolved independently in two branches and the overall phylogeny would have shown greater resolution without the use of this character. On the other hand, the character of capping of drone cells with a central pore turned out to be the only character that resolved the clade for Apis cerana and A. koschevnikovi.

Wenzel (1993) used a total of 21 characters in a study of paper wasp evolution. Of the 21 characters used, all consisted of behavioral characters. The study was an attempt to determine whether or not behavioral ontogeny could be used to determine character polarity; however, in the study, comparisons were made of a phylogeny based on morphological characters with one derived from behavioral characters. Of the two cladograms presented, neither showed much difference in the branching patterns of individual taxa - indicating the utility of behavioral characters in phylogenetic reconstruction in this group.

Among the Coleoptera, only one study has been conducted to date including behavioral characters in cladistic studies (Pinto, 1984). In this phylogenetic study of eupomphine blister beetles, two out of 30 total adult characters used were behavioral - including presence or absence of contact of head appendages during male courtship display; and, the presence or absence of mounting behavior during copulation. Of the two characters, the presence or absence of contact of head appendages during courtship provided the greatest resolution. The presence of this character was a defining synapomorphy (shared derived character) of the genus Phodaga.

Crespi et. al. (1998) compared several types of data to determine the evolution of the soldier caste in gall-forming thrips. These data included character matrices using two separate mitochondrial genes, adult morphology (including behavior), and gall morphology. Of the four sets of data compared, the one using gall morphology provided very little resolution to the phylogenetic construction. The other three together, including the morphological data set (which consisted of one behavioral trait - the presence or absence of males during the initiation of gall formation), provided a fairly consistent phylogeny.

In a study of Orthoptera phylogeny, Gwynne (1995) produced a cladogram using 67 characters of which two were behavioral. The behavioral characters included whether or not the male assumed a top position during copulation and whether or not both sexes faced each other during copulation. Many other characters were used in this study, however, that also have behavioral significance. Singing among the Orthoptera is an example of a behavioral trait that has potential for cladistic studies; and, while singing (or any type of stridulation) was not used in the study, the mechanisms of these sounds were. The presence of a harp, folding of the tegmina, and several other morphological characters pertaining to singing were used. This is a clear case where behavior has significant impact on the evolution of particular structures - structures that are useful in classification. The connection between behavior and structure raises the difficult question of redundancy in character matrices that use both types of characters; while, at the same time providing clear evidence of the potential for weighting certain behavioral characters equally with morphological ones.

Another example of the difficulty in evaluating morphological and behavioral characters together is a study done by Urbani (1989) evaluating the phylogeny and behavioral evolution of ants. Urbani found that most behaviors that where representative of general evolutionary trends among the Formicidae showed a high degree of parallelisms and convergences producing "mosaic" phylogenies of very poor resolution. Two reasons were offered for this situation. First, that eusociality among the ants provides certain constraints on the behavioral repertoire available to any evolving taxon. Second, most behaviors have morphological counterparts that have already been used by systematists to produce phylogenies and the corresponding behavioral traits are generally very difficult to quantify adequately. Urbani was only able to find one behavioral character of use for including in a character matrix in this phylogenetic study of ant subfamilies. This character is the manner in which ants transport other adult ants.

Alison Stuart (1998) in a study of black fly phylogeny used behavioral characters of cocoons (the end-product of behavior) and the characters involved in the actual construction of the cocoon, to determine the relative merits of both types of characters (as noted above). Five characters taken from cocoons were used to reconstruct a black fly phylogeny. These characters included the general shape of the cocoon, the shape of the collar, the presence of fenestrae in the walls of the cocoon, and the presence or absence of an inner cocoon. Similarly, fourteen characters of cocoon construction were used in the subsequent reconstruction. These characters included the manner of constructing the cocoon foundation, of placing front strands, of making the pupation tent, of making wall-floor connections, etc. Of the two types of data, the behaviors involved in cocoon construction provided more highly resolved phylogenies than did the reconstructions based on cocoon morphology alone. The former data also more closely matched a phylogeny based on cytological data.

Heather Proctor (1992) used behavioral characters in her study of sensory exploitation of male mating behavior in water mites. In her character matrix, two out of 28 characters included behavioral traits - including whether or not males fanned over spermatophores and the place of spermatophore deposition (whether on a substrate or on legs). Both of these characters were not recorded (data were not available) for two of the 13 species studied. The character of male fanning provided significant resolution in the Neumania muttkowski/papillator clade; whereas, the character of spermatophore deposition was autapomorphic (meaning it was present in one taxon only - providing no resolution to the tree topology). This was an interesting study in that it included both traditional and more recent approaches to behavioral evolution. After her cladistic analysis using the above characters (including behavioral characters), Proctor proceeded to overlay other behavioral characters (the presence of a net stance and of male courtship trembling) onto the derived cladogram to determine probable evolutionary scenarios for these traits. Two equally parsimonious explanations for the evolution of these characters were determined.

Within the same year (1990) two studies on spider classification were published that used behavioral characters. The first to appear was a study of the evolution of the Class Arachnida by Shultz. Among the 64 characters used to reconstruct his phylogeny of the spiders and their kin, only one behavioral character was included, this was the presence or absence of female grasping behavior (of the male opisthosoma) during mating. Shultz indicated in his discussion that other elaborate courtship behaviors were present in transfer of spermatophores but did not include these traits in his study. The other study was conducted by Coddington of orbweaver classification. Coddington used a total of 61 characters to reconstruct the phylogeny of the Orbiculariae. Of these characters, 49 were morphological characters and 12 (his Characters 50 through 61) were behavioral. The later were comprised of characters involved with web construction - including frame construction, radius of the web, hub construction, temporary and sticky spiral construction, leg supports and manipulation of silk. In the final phylogenies, none of the behavioral characters showed evolutionary convergence of reversal among any of the branches (indicating a high degree of resolution). Most of the behavioral characters were useful in resolving the clades of more derived taxa. Lacking in Coddington's study is a discussion of the behavioral characters used, in contrast to his discussion of morphological characters.

Conclusions from Extant Studies

The few phylogenetic studies that have been done incorporating behavioral traits into character matrices have shown the potential for these kinds of characters in evolutionary reconstructions. In most of the studies done to date, greater resolution has been achieved with the use of behavioral traits than would have been possible without them. It is important to note, however, that with only a few exceptions, most of the behavioral characters used comprised very low proportions of the total array of characters in the individual studies - even in cases were many more behavioral traits were available for use. This phenomenon probably reflects two underlying tendencies of systematists - the hesitancy to use characters that have not been proven, and the habit of not using characters that produce high levels of homoplasy. These tendencies, however are not unique to behavioral characters. All new types of data must be similarly proven, whether they be the use of genitalia (which characters appeared as new data nearly three generations ago) or the use of the more recent molecular data. The successes and failures of each new study will inevitably lead to sounder phylogenies based on behavioral characters.
 

References

Alexander, B.A. 1991. Phylogenetic analysis of the genus Apis (Hymenoptera: Apidae). Annals of the Entomological Society of America. 84(2): 137-149.

Atz, J. W. 1970. The application of the idea of homology to behavior. In, Development and Evolution of Behavior, ed. L. R. Aronson, E. Tolbach, D. S. Lehrman, J. S. Rosenblatt, pp. 53-74. San Fransisco: Freeman.

Coddington, J. A. 1990. Cladistics and spider classification: araneomorph phylogeny and the monophyly of orbweavers (Araneae: Araneomorphae; Orbiculariae). Acta Zoologica Fennica 190: 75-87.

Crespi, B.J. et. al. 1998. Phylogenetics of social behavior in Australian gall-forming thrips: vidence from mitochondrial DNA sequence, adult morphology and behavior, and gall morphology. Molecular Phylogenetics and Evolution 9: 163-180.

Darwin, C. 1859. The Origin of Species by Means of Natural Selection. John Murray, London.

De Quieroz, A, and P. H. Wimberger. 1993. The usefulness of behavior for phylogeny estimation: levels of homoplasy in behavioral and morphological characters. Evolution 47(1): 46-60.

Evans, H.E. 1966. The comparative ethology and evolution of sand wasps. Harvard University Press. 526 pp.

Gwynne, D.T. 1995. Phylogeny of the Ensifera (Orthoptera): A hypothesis supporting multiple origins of acoustical signaling, complex spermatophores and maternal care in crickets, katydids, and Weta. Journal of Orthoptera Research 4: 203-218.

Pinto, J. D. 1984. Cladistic and phenetic estimates of relationship among genera of eupomphine blister beetles (Coleoptera: Meloidae). Systematic Entomology 9: 165-182.

Proctor, H. C. 1992. Sensory exploitation and the evolution of male mating behavior: a cladistic test using water mites (Acari: Parasitengona). Animal Behavior 44: 745-752.

Shultz, J.W. 1990. Evolutionary morphology and phylogeny of Arachnida. Cladistics 6:1-38.

Stuart, A.E. and F.F. Hunter, 1998. End-products of behaviour versus behavioural characters: a phylogenetic investigation of pupal cocoon construction and form in some North American black flies (Diptera: Simuliidae). Systematic Entomology 23: 387-398.

Urbani, C. B. 1989. Phylogeny and behavioral evolution in ants, with a discussion of the role of behavior in evolutionary processes. Ethology, Ecology and Evolution 1: 137-168.

Wenzel, J.W. 1992. Behavioral homology and phylogeny. Annual Review of Ecology and Systematics 23: 361-381.

Wenzel, J.W. 1993. Application of the biogenetic law to behavioral ontogeny: a test using nest architecture in paper wasps. Journal of Evolutionary Biology 6: 229-247.

Wimberger, P. H. and A. De Quieroz. 1996. Comparing behavioral and morphological characters as indicators of phylogeny. In, Phylogenies and the Comparative Method in Animal Behavior, ed. E. P. Martins, pp. 206-233. Oxford University Press.