Aims and Theoretical Perspectives.
It is generally accepted that Body Mass is the best single index of overall size in birds (Clark, 1979; Rising and Somers, 1989; Freeman and Jackson, 1990; Piersma and Davidson, 1991; Dunning, 1993). Variation in body size is the most common trend observed in the vertebrate evolutionary record (Stanley, 1973; Boucot, 1976; Gingerich, 1985; McKinney and McNamara, 1991). Over the last three decades there has been increasing interest in the biological implications of body size and its ecological, morphological and physiological correlates (e.g. Stanley, 1973; Damuth, 1981; Peters, 1983; Schmidt-Nielsen, 1984; Damuth and MacFadden, 1990; Pennycuick, 1992; Damuth, 1993; Blackburn and Gaston, 1994b, 1999). However, few analyses have included the Accipitridae and a key aim of this study is to investigate the evolutionary significance of Body Mass differentiation in the Aquila - Hieraaetus - Poiemaetus clade and compare and contrast such data with those in other avian taxa.
Most studies of avian size variation that have been undertaken to date have been descriptive and generalist (cf. Blackburn and Gaston, 1999). Furthermore,
comparatively few studies have been phylogenetically based, despite the importance of such approaches (Harvey and Pagel, 1991 ; Harvey, 1996). Whilst descriptive studies represent an essential first step in analysis, what we now require are interpretations. Phyletic size increase, usually codified as ‘Cope’s rule’ (Stanley, 1973; McKinney, 1990), has been a particular concern in evolutionary biology over the last two decades (e.g. Gingerich, 1985; MacFadden, 1986; Gould, 1988; McKinney and McNamara, 1991; Damuth, 1993; Jablonski, 1997). However, phyletic size analyses of avian clades have been rare, largely due to the lack of well-resolved phylogenies.
With the exception of Livezey’s (e.g. 1995a, 1995b, 1996a, 1997a) work on waterfowl, our knowledge of phyletic size trends in birds has been largely confined to anagenetic Body Mass increases associated with loss of volancy in insular taxa (e.g. Livezey, 1993). Indeed, some workers (e.g. Stanley, 1973; Damuth, 1993) have argued that there is no evidence of Cope’s rule in birds. Yet, evidence of phyletic size increase may prove to be as frequent in birds as it is in other vertebrate taxa. In the case of accipitrid raptors, we now have several resolved phylogenies against which this hypothesis can be tested and the Aquila - Hieraaetus - Poiemaetus clade is one example.
This leads to an important theoretical consideration. Stanley (1973), in his restatement of Cope’s rule, argued that, if the founding taxon of a clade originates near the lower limit of a potential size range, then subsequent diversification will lead to a concomitant size increase away from the lower limit. Thus, Stanley argued that phyletic size variation can be interpreted solely in terms of stochastic diversification into vacant space, without recourse to invoking what he termed “fundamental advantages” i.e.
adaptations. This perspective was further codified by Gould (1988) who argued that an increase in size within a clade is a function of origin near a lower size limit. It follows that phyletic size variation may be nothing more than an “increase in variance" rather than adaptive diversification (Gould, 1988). Thus, much debate on phyletic size trends has been rooted in challenges to the ‘adaptationist syndrome’ (cf. Gould and Lewontin, 1979).
However, palaeobiological evidence has been the primary focus of this debate (e.g. Stanley, 1973; Jablonski, 1997). No matter how ingenious reconstructions of the palaeobiology of fossils are (e.g. Guthrie, 1990), our knowledge can never approach that gained from the study of living organisms. The new molecular analyses provide us with phylogenies of extant organisms or, in the words of Harvey et al. (1994), “phylogenies without fossils”. It is evidential that such phylogenetic treatments provide a framework against which the adaptive significance of evolutionary trends in living organisms can be investigated \N\ih a much higher degree of confidence than is possible with fossil-based phylogenies. This is the approach followed below. Furthermore, whilst this study is unquestionably ‘adaptationist’, I believe that some of the perspectives proposed by Stanley and Gould are of value in terms of both phylogenetic and ecological analyses and they are developed below.
Another aspect of size variation that is the subject of much current research is the interspecific relationship between animal abundance and Body Mass. Indeed, such relationships are some of the most frequently cited ecological patterns and birds have figured prominently in several studies (e.g. Nee et a!., 1991; Cotgreave and Harvey,
1992; Cotgreave, 1993; Blackburn and Gaston, 1996a, 1996b, 1997,1999). However, only one causative mechanism for such patterns has generally been considered, namely, the hypothesis that abundances are constrained by energy availability (of. Damuth, 1987, 1993; Blackburn and Gaston, 1999). This hypothesis, originally formulated for terrestrial mammals (Damuth, 1981, 1987), is rooted in the fact that metabolic rate increases with Body Mass to the 0.75 power (Kleiber, 1962) and, therefore, species abundance should decrease with Body Mass to the -0.75 power. The latter relationship has been found to characterize a wide variety of vertebrate taxa and has been termed the “energetic equivalence rule" or EER (Damuth, 1987, 1993; Marquet etal., 1995; Blackburn and Gaston, 1999). In short, small species appear to use as large a proportion of energy within ecosystems as large species (Peters, 1983; Peters and Wassenberg, 1983; Damuth, 1993).
In contrast, studies of abundance-size relationships within avian communities have shown that larger species often control more energy than smaller species and, therefore, maintain larger populations than expected for their body size (e.g. Brown and Maurer, 1986; Juanes, 1986; Nee at a!., 1991; Cotgreave and Harvey, 1992). This implies that larger species are better at obtaining energy from their environments (Brown and Maurer, 1986; Damuth, 1993). However, herein lies a paradox. If the demonstrated abundance-size relationships for birds hold, then we should expect adaptive phyletic size increase to be a common evolutionary trend in avian clades, yet such trends have not been identified to date (of. Stanley, 1973; Damuth, 1993).
However, several studies have demonstrated that abundance-size relationships in avian raptors differ from those of other birds in that larger raptors breed at lower densities than smaller raptors (Schoener, 1968; Newton, 1979; Juanes, 1986; Kruger, 2000). Thus, in terms of the EER, we should not expect Cope’s rule to apply in the case of raptors. It therefore follows that, if phyletic size increase does characterise raptor clades, then it may be due to Stanley’s (1973) and Gould’s (1988) stochastic mechanisms. It is also the case that most previous studies of avian abundance-size relationships have been focused on communities and guilds rather than clades and, remarkably, no study to date has considered the role of flight adaptation. Thus, a further aim of this study is to investigate abundance-size relationships in the Aquila, Hieraaetus and Poiemaetus clade, relate these to ecological and aerodynamic data and analyse them with respect to current theory.
The investigation of reproduction forms the fundamental basis of population ecology (e.g. Newton, 1979,1989a). It has long been accepted that several species in the clade have very low rates of reproduction (e.g. Amadon, 1964). This is manifest in low clutch sizes and the evolution of both obligate and facultative siblicide, all of which are readily interpretable as brood reduction strategies (Meyburg, 1974; Newton, 1979; Edwards and Collopy, 1983; Gargett, 1990; Forbes and Mock, 1994). Furthermore, several of these taxa are also characterised by relative longevity, extended ontogeny, delayed or deferred reproduction and high parental investment in offspring.
Thus, some members of the clade exemplify K-selection (sensu MacArthur and Wilson, 1967; Wilson, 1975), a key parameter in population ecology. Yet, as Wilson (1975)
has emphasized, “ ...it is possible to discern ascending grades of /(-selected demographic traits and population stability”. The results of previous studies have indicated that some members of the clade exhibit such variation. Therefore, another aim of this study is to investigate how and why members of the clade differ in their degree of /(-selection. Until now /(-selection in accipitrids has not been investigated phylogenetically and neither have defining traits been statistically investigated with respect to other key parameters such as Body Mass. Such investigations are undertaken below.
An important aspect of /(-selection is extended ontogeny (Pianka, 1970; Wilson, 1975) and heterochrony is also a key concern in modern evolutionary studies (Chapter 1 ). in the Aquila, Hieraaetus and Poiemaetus eagles the most obvious manifestation of heterochrony is in the timing of plumage maturity, which is closely related to age at first breeding in accipitrids (Newton, 1979). Given this relationship it is possible to estimate the age at which adulthood is attained for all the taxa considered by reference to plumage sequences obtained from specimens and the literature. The ecological implications of such data are interpreted below.
The potential of ecomorphology has been emphasized in Chapter 1. In this study I am primarily concerned with its implications for our understanding of predation strategy and flight adaptation in the Aquila, Hieraaetus and Poiemaetus eagles. Whilst these eagles are closely related they are ecologically highly differentiated and several are sympatric (Brown and Amadon, 1968; Brown, 1976a, 1976b; Cramp and Simmons, 1979; Brown
Ecomorphology provides insights into how and why this ecological diversification has taken place. Rostral and pedal dimensions are universally accepted as those most closely correlated with predatory biomechanics, i.e. prey handling, in diurnal raptors (Goslow, 1972; Cade, 1982; Jenkins, 1995). Consequently, we should be able to develop and test hypotheses about predation strategy from the analysis of such biometrics. Therefore, another aim of this study is to analyse rostral and pedal differentiation in the clade, relate such data to known predation strategies and make inferences about the predation strategies of taxa for which dietary information is lacking.
A key aspect of avian ecomorphology is flight adaptation. The last two decades have witnessed increasing interest in avian flight performance, but the absence of key biometric data for many taxonomic groups has limited research potential (e.g. Andersson and Norberg, 1981; Kerlinger, 1989; Pennycuick, 1989; Mendelsohn ef a/., 1989). In this regard, the eagles of the genera Aquila, Hieraaetus and Poiemaetus are no exception; despite their extensive literature, basic data such as wing areas, wing spans and aspect ratios obtained from adequately sexed samples are largely lacking. For example, Mendelsohn et al. (1989) provided such data for several Aquila, Hieraaetus and Poiemaetus eagles but, unfortunately, they did not differentiate between the sexes. This limits the utility of their data sets (Chapter 4).
As wings and tails are typically closed in standard skin specimens, areas and spans cannot be directly obtained from them. Fortunately, several authorities (Jacksic and Carothers, 1985; Kerlinger, 1989; Mendelsohn et al., 1989; Pennycuick, 1989) have
provided formulae, regression equations and programs via which it is possible to obtain estimates for flight performance calculations from linear biometrics obtained from skin specimens. Thus, a final aim of this study is to investigate flight adaptation in the
Aquila, Hieraaetus and Poiemaetus eagles and relate such data to phylogeny and ecology. Particular emphasis is placed on the role of primary slots (sensu Kerlinger, 1989) and rectricial form; these traits have been largely neglected in previous studies.
Specimens and Data
The Operational Taxonomic Units (sensu Sneath and Sokal, 1973; Bateman, 1994) selected for investigation were the generally accepted biological species in the traditional genera Aquila, Hieraaetus and Poiemaetus (Mayr and Cottrell, 1979; del Hoyo et al., 1994) but account was made of the taxonomic recommendations given in Chapters 3 and 6. These OTUs are listed in Table 2.1. In the case of polytypic species, I selected the best-represented subspecies as the OTU. For each OTU, I lumped data for both immatures and adults. Data from juveniles were excluded from all analyses.
The base phylogram (Figure 1.1) utilised in this analysis is that defined and discussed in Chapter 1, based on the cytochrome b analyses of Seibold (1994), W inkef a/. (1998) and Wink and Sauer-Gürth (2000). In this phylogenetic perspective, the small
Hieraaetus eagles (H. pennatus, H. morphnoides, H. weiskei, H. ayresii and H. kienerii)
together with A. wahlbergi are interpreted as primitive, basal taxa and the larger
Hieraaetus eagles (H. fasciatus and H. spilogaster) are, together with the eagles traditionally assigned to the genera Aquila and Poiemaetus, interpreted as more
derived.
Amongst these more derived taxa, three further clades are apparent: the ‘spotted eagles’ (A. pomarina, A. hastata and A. cianga), the ‘imperial eagles’ (A. rapax, A. heliaca and A. nipalensis) and a third clade comprising H. fasciatus, H. sp/7ogasfer and
A. verreauxii. This phylogenetic hypothesis forms the basis of the cladogenetic interpretations given below. However, traditional generic classifications (cf. Mayr and Cottrell, 1979) are adhered to, pending the now clearly necessary taxonomic revision.
A total of 509 museum skin specimens of the 19 OTUs were measured. The number of specimens measured for each OTU was largely dictated by availability and is summarised in Table 2.1. Forty different external biometrics, as defined in Appendix A, were taken on each specimen using vernier dial callipers and a steel rule. It is increasingly accepted that phenotypes are composed of modular units (e.g. Cheverud, 1996; Gatesy and Dial, 1996; Wagner, 1996; Dassow and Munro, 1999), and the genetical basis of their development and integration is the subject of much current research (e.g. Shubin et al., 1997; Tautz and Schmid, 1998; Lovejoy et al., 1999). Given the basic form of standard museum skins we can, realistically, only investigate certain modular units of the avian phenotype at the macroscopic level. The external biometrics taken during the course of this study were designed to provide measures of the following phenotypic units; head (specifically rostral elements), wing, tail and ped. In the case of accipitrid raptors these units are evidentially those most closely related to aerial and terrestrial locomotion and predation strategy (Hartman, 1961; Goslow, 1972).
The sex of each specimen was assigned on the basis of label data and the protocols given in Chapter 4 and Appendix B. Reversed sexual dimorphism is evident in all the species analysed (see results below and Chapter 4) and each sex of each taxon is therefore morphometrically distinct. In the most dimorphic species, the differences between the sexes are as great as interspecific differences in the less dimorphic species (see results below and Chapter 4). As a consequence, each sex of each species was treated as a discrete ecomorphotype, here termed a sextaxon'.
Data concerning morphology and autecology were obtained by specimen examination, or were taken from the literature. In terms of reproduction, I investigated Clutch Size, Siblicide, Incubation Period and Fledging Period using published data and museum oological data. Patterns of heterochrony were investigated by establishing plumage sequences from specimens and the literature. Body Mass data were obtained from specimen labels, published sources and specialists. In the few cases in which Body Masses were unavailable (H. weiskei, both sexes; A. gurneyi, male; H. kienerii, female), they were predicted via least-squares regression of mean Primary 7 Chords on mean Body Masses for the sextaxa in which these variables were established. However, some problems are associated with the use of Body Mass in the species considered and need to be emphasized.
Only a small number of available skin specimens had accurate associated Body Mass data and this largely enforced the use of mean Body Masses. Consequently, most of these data are ‘unassociated’ to the specimens utilised, resulting in the limitations identified by Welsh et al. (1988). They also preclude the derivation of summary
statistics and the performance of tests in most instances. This problem is all too common in ornithology (cf. Livezey, 1992) and will not be readily redressed. When several different Body Mass datasets were available for a particular sextaxon, the “best available sample” (Dunning, 1993) was utilised. The two criteria employed in the selection of a “best available sample” were:
a) Data from a single source that furnished the largest available sample size.
b) Data from a single source that were associated with the most complete summary statistics.
Further data that require comment are measures of abundance/density. In all instances, I measured abundance/density in terms of the number of breeding pairs per km^ extrapolated from sizes of breeding territories obtained from the literature. With the exception of A. verreauxii (see below) I used the minimum recorded size of breeding territory for each species. Many accipitrids typically defend and utilise both a home range and a breeding territory (Newton, 1979). In such instances a home range is exploited both during and outside the breeding season and may contain several alternative breeding territories which are smaller in area and occupied only during the breeding season. However, several of the taxa considered are long distance migrants and do not occupy the same home ranges throughout the year (Cramp and Simmons, 1979; Newton, 1979; del Hoyo et al., 1994). Thus, the only directly comparable interspecific measure of density is breeding territory.
Accipitrids have suffered catastrophic population declines due to direct and indirect anthropogenic influences (cf. Brown, 1976a, 1976b; Newton, 1979) and because of this their population densities are often artificially low. For this reason I selected the minimum rather than the mean or maximum size of breeding territory. However, in the case of A. verreauxii there is a marked dichotomy in the available density data. The best studied population is in the Zimbabwean Matapos Hills, where densities can exceed one breeding pair/10 km^ but, elsewhere in this taxon s range, densities can be as low as one breeding pair/250 km^ (Gargett, 1975,1990; Newton, 1979). Whilst the latter density is exceptionally low, the high density of the Matapos population is due to the unusually high density of hyraxes, which are the primary prey of this specialist raptor (Gargett, 1990). In fact. Hustler (1995) has queried the Matapos eagle densities and argued that they may not be 'normal' as they are much higher than elsewhere in the eagle's range. However, on Hustler's (1995) own admission the available habitats surrounding the Matapos are severely degraded. Gargett and Gargett (1997) countered Hustler's (1995) arguments by emphasizing that concepts of 'normality' are not absolute, as all natural populations are subject to differing levels of temporal scarcity and abundance in relation to changing environmental factors. Indeed, as the Matopos Hills are a protected area, it may be the case that this enclave is one of the few places left where 'natural' eagle densities persist. As a consequence of this debate, I included both high and low density values for A. verreauxii in two separate treatments.
Statistical Analyses and Flight Performance Calculations
All statistical analyses were undertaken on a hand calculator or via the SPSS 10.0 software package (SPSS, 1999a, 1999b). Summary statistics were calculated for each biometric for each sextaxon (Appendix 0). Size variation was investigated via the construction of Body Mass frequency distributions (histograms), spectral plots of selected biometrics and Body Mass mapping on phylograms. In all other analyses (unless otherwise stated) variables were logio transformed to normalise data, stabilize variance and render data homoscedastic (Sokal and Rohlf, 1981). Intraspecific differences between sexes for each of the 40 external biometrics considered were tested for by Oneway ANOVAs and a post hoc test for unequal sample sizes, Student- Newman-Keuls test (Sokal and Rohlf, 1969). Where appropriate, tests for data normality were performed utilising either the Shapiro-Wilk test (samples < 50 cases), or the Kolmogorov-Smirnov-Lilliefors test for larger samples (SPSS, 1999a, 1999b).
Allometry, taken here to be “ the study of size and its consequences” (cf. Gould, 1966), is one of the most appropriate means of investigating interspecific morphometric similarity and differentiation. It has long been accepted that many biological scaling relationships are described by Huxley's (1932) power relationship;
p
Y =oX
and a and P are the parameters of the power equation. Logarithmic transformation