Showing papers on "Nycticorax published in 1987"

Phylogeny of Herons Estimated from DNA-DNA Hybridization Data

Abstract: -Genetic distances derived by hybridizing single-copy DNAs of 31 heron species (or subspecies) and 1 ibis species are summarized as ATm values. From these distances, a phylogeny is estimated and distributional properties of DNA hybridization data are computed. I found that the distinction between night and day herons is primarily adaptive, not genealogical; Syrigma is closely related to Egretta; Bubulcus and Casmerodius are closely related to Ardea, but Egretta is not; bitterns are the sister taxon of the day and night herons; Cochlearius and Tigrisoma are each others' closest relatives and together form the sister group of the rest of the ardeids; and the rate of single-copy DNA evolution differs in different heron lineages. Received 18 April 1986, accepted 3 September 1986. MOST taxonomists agree that herons belong in a family of their own, the Ardeidae, but there is considerable disagreement concerning the intrafamilial relationships of these birds. Over the last 100 years, the number of recognized species in the Ardeidae has varied from 60 to 93 and the number of genera from 15 to 35 (Sharpe 1898, Reichenow 1913). Although herons are usually divided into four groups (day herons, night herons, tiger herons, and bitterns), species are moved back and forth among these groups with each revision of the classification. A fifth group is sometimes considered necessary to accommodate the enigmatic Boatbilled Heron (Cochlearius cochlearius). The continual problems of determining most heron relationships derive primarily from the fact that adaptive changes within the limits of the ardeids' wading-piscivorous Bauplan are difficult to interpret. Herons are constrained to have long bills, legs, and necks, and this constraint has induced a family history rife with parallel and convergent evolution. To develop a hypothesis of heron phylogeny without having to interpret tracked morphological or behavioral characters, I used DNADNA hybridization to compare taxa. The logic of DNA hybridization has been reviewed by Sibley and Ahlquist (e.g. 1983), Benveniste (1985), and others. The technique operates under the assumption that the genetic relatedness I Present address: Tiburon Center for Environmental Studies, San Francisco State University, P.O. Box 855, Tiburon, California 94920 USA. of organisms is reflected in the similarity of their DNA base pair sequences. This similarity can be measured by hybridizing strands of DNA from different species and measuring the bonding strength of these hybrids. The poorer the bonding strength, the more distantly related the organisms. The advantages of DNA hybridization are that it is objective and it accounts for historically informative characteristics encoded in the DNA that are not necessarily expressed physically. Such previously unmeasurable genetic features include pseudogenes (e.g. the obsolete genes coding for tooth structure in birds; Kollar and Fisher 1980) and regulatory genes. Estimating and testing phylogenies.-DNA hybridization produces distance data, and the most appropriate method for clustering such data is least-squares regression (Sheldon in press). Templeton (1985) pointed out that tree-building algorithms based on procedures like least squares simply provide estimates of phylogenies. Alternative estimates require testing by statistical methods before one phylogeny can be accepted as better than others. Unfortunately, statistical methods for testing alternative phylogenetic hypotheses have not been established. Templeton (1985), for example, introduced the delta Q-test, but Saitou (1986) argued that this test is inadequate for differentiating topologies, and Fitch (1986) argued that it assumes evolutionary rate constancy. Although statistical procedures can be used to test for different evolutionary rates in different lineages (e.g. Felsenstein 1984) and, in special situations, can resolve multifurcations 97 The Auk 104: 97-108. January 1987 This content downloaded from 207.46.13.172 on Sat, 15 Oct 2016 05:00:21 UTC All use subject to http://about.jstor.org/terms 98 FREDERICK H. SHELDON [Auk, Vol. 104 TABLE 1. List of the taxa used in DNA hybridization comparisons and the number of separate purifications made of the DNAs of those taxa. Asterisks indicate taxa that were radio-labeled. No. of preparaSpecies tions Syrigma sibilatrix (Whistling Heron)* 1 Ardea herodias (Great Blue Heron)* 5 A. cocoi (Cocoi Heron) 2 A. pacifica (White-necked Heron) 1 A. melanocephala (Black-headed Heron) 2 A. sumatrana (Great-billed Heron) 1 Casmerodius albus egretta [Great Egret (North America)]* 6 C. a. modestus [Great Egret (Australasia)] 1 Bubulcus ibis (Cattle Egret)* 4 Egretta vinaceigula (Slatey Egret) 1 E. tricolor (Tricolored Heron) 4 E. intermedia (Intermediate Egret) 2 E. novaehollandiae (White-faced Heron) 2 E. caerulea (Little Blue Heron)* 4 E. thula (Snowy Egret)* 5 E. garzetta (Little Egret) 2 E. sacra (Eastern Reef Egret) 1 Ardeola grayii (Indian Pond Heron) 1 Butorides striatus virescens [Green-backed Heron (North America)]* 5 B. s. striatus [Green-backed Heron (South America)] 1 B. s. javanicus [Green-backed Heron (Southeast Asia)] 2 Nycticorax violaceus (Yellow-crowned Night-Heron)* 3 N. nycticorax (Black-crowned Night-Heron)* 10 N. caledonicus (Nankeen Night-Heron) 1 Cochlearius cochlearius (Boat-billed Heron)* 5 Tigrisoma lineatum (Rufescent Tiger-Heron)* 3 Ixobrychus exilis (Least Bittern)* 1 I. minutus (Little Bittern) 1 I. cinnamomeus (Cinnamon Bittern) 3 Botaurus lentiginosus (American Bittern)* 5 B. stellaris (Palearctic Bittern) 1 Plegadis falcinellus (Glossy Ibis)* 1 (e.g. Fitch 1986), none so far is useful in evaluating alternative phylogenies consisting of more than four taxa. The best way to evaluate different phylogenies is by the consensus tree method, which seeks corroboration among alternative phylogenies. Lanyon's (1985) Jackknife Strict-Consensus Tree (JST) algorithm was chosen to evaluate trees derived from subsets of the data in this study. The JST method has one additional feature. By comparing subsets of a single distance matrix, it provides an intuitive indication of the additivity and independence of the data. MATERIALS AND METHODS Biochemistry.-The methods used to prepare hybrids were essentially those of Sibley and Ahiquist (1981). Further detail is provided in Sheldon (1986). Briefly, high molecular weight DNAs were extracted from bird erythrocytes and tissues and analyzed spectrophotometrically for protein contamination. The DNAs were then sheared by sonification, yielding fragments with an average length of 400-500 base pairs, as determined by agarose gel electrophoresis. Single-copy DNA fragments were recovered by hydroxyapatite chromatography and radioactively labeled with 125I. The labeled DNAs were mixed with unlabeled driver DNAs in a ratio of 1:400, boiled, and incubated at 60?C to a Cot exceeding 15,000. These conditions permitted hybrids to form between DNA sequences differing in base pair complementarity by a maximum of 25-30%. The hybrids were then fractionated thermally at 2.5?C increments in lots of 25, from 550 to 95?C. Each 25-hybrid lot (= 1 experiment), contained at least one homoduplex control hybrid, comprising label and driver DNAs prepared from the same sample of purified DNA. The radioactivity eluted at each of the 17 fractionation temperatures, representing the amount of DNA that had dissociated to single-stranded form at that temperature, was counted in a gamma spectrometer and constituted a raw datum. Reciprocal comparisons, involving ca. 940 hybrids, were made among 13 species of heron and 1 species of ibis (Table 1). About 300 one-way comparisons were made using labeled DNAs from the same 13 heron species and driver DNAs from 18 additional heron taxa. Another ca. 130 hybrids were produced to determine genetic distances within species. In planning the reciprocal comparisons, an effort was made to hybridize the DNA of each of the 14 labeled species at least 5 times with the driver DNA of each of those species to produce 10 hybrids per pair. This was not always possible, however, because of their availability and supply. Data analysis.-The 23,000 heron raw data are available to any person who sends six formatted, IBMPC disks in a self-addressed, stamped container. Methods of data analysis differed from those of Sibley and Ahlquist (e.g. 1981, 1983) in that ATm was used as the measure of genetic distance, instead of AT50H, and clustering was performed by least squares. The logic behind the use of ATm is discussed in the Results and Discussion and that of data correction in Sheldon (in press). To calculate Tm, the count recorded at each temperature from 62.50 to 95?C was normalized to a percentage of the total counts in that range, and a cumulative frequency distribution was constructed. Tm This content downloaded from 207.46.13.172 on Sat, 15 Oct 2016 05:00:21 UTC All use subject to http://about.jstor.org/terms January 1987] Heron Phylogeny 99 equaled the temperature at which 50% of the counts was recorded, extrapolated by linear regression. Genetic distances (zvTm's) were calculated by subtracting heteroduplex Tm values from the homoduplex Tm of the same experiment. Interspecific distances were summarized in lists (Tables 2-15). The average distances between the 14 labeled taxa were calculated by multiplying the values in the lists by sample size, adding reciprocal products, and dividing the sum by the total number of observations. [A matrix of average distances was published by Sheldon (in press).] From these average distances, trees were drawn using the least-squares option of the programs "Fitch" and "Kitsch" in J. Felsenstein's phylogenetic computer package, PHYLIP (version 2.8). The relative quality of the fit of these trees was judged from the residual sum of