ALTtOUOH the functional anatomy of woodpeckers has long been a subject of interest, their internal anatomy has not been used extensively for determining probable phylogenetic relationships within the family. In part this is probably due to the reluctance to use highly adaptive fea- tures in phylogenetic studies because of the likelihood of convergent evolution. Bock (1967) and others have pointed out that adaptiveness in itself does not rule out taxonomic usefulness, and that the highly adaptive features will probably be the ones having conspicuous anatomical modifications, and Bock emphasizes the need for detailed studies of func- tion before using features in studies of phylogeny. Although valuable, functional conclusions are often based on inferences not backed up by experimental data. As any similarity between species is possibly due to functional convergence, I believe what is needed most is detailed study of a number of features in order to distinguish between similarities re- sulting from convergence and those based on phylogentic relationship. Simple structures are not necessarily more primitive and morphological trends are reversible, as Mayr (1955) has pointed out. Individual varia- tion may occur and various investigators may interpret structures dif- ferently. Despite these limitations, speculation concerning phylogeny will continue in the future, and I believe that it should be based on more, rather than fewer anatomical studies. MATERIALS AND METItODS Alcoholic specimens representing 33 genera and 47 species listed in Table 1 were dissected with the aid of a Zeiss operation microscope. Major items studied include the tongue and hyoid apparatus, salivary glands, certain jaw muscles, and the limb muscles. In order to make direct comparisons between species, an approximate representation of relative size was obtained by the use of a cranial length index, listed for each species in Table 1. This was calculated by dividing the distance from the nasal-frontal hinge to the posterior end of the head by the same measure- ment in Dryocopus pileatus. Relative dimensions were calculated by dividing the actual dimension in each case by the appropriate cranial length index. It is hoped that when more specimens have been studied, the quantitative data can be used in a more sophisticated mathematical treatment, but in this analysis measurements were used only in cases of obvious difference. The criticism that too few individuals of a species were used may be valid, but in order to gain a better perspective in this preliminary study, it was decided to examine a large number of features in a representative series of genera and species. All too often in previous works er- roneous conclusions were reached by studying only a few species of a group. Some 65 The Auk, 89: 65-85. January 1972 TABLE 1 LIST OF SPECIES STIJDIED  AND SIJ1VI1VIAR3/4 OF CERTAIN FEATIJRES Species Sartorius origin Iliotib. Cranial post w. Thoracic % il. troc. index % biceps spines anterior Jynx torquilla .59 125 11/.2 41 Picumnus cirrhatus .38 52 3 100 Picumnus innominatus .41 75 3 65 Nesoctites micromegas .53 120 21/..2 53 Sasia ochracea .37 55 3 80 Chrysoptilus melanochloros .76 91 2I/.2 41 Colapres auratus (4) .85 80-97 2-21/.2 36-83 Piculus fiavigula .63 72 la 50 Camperhera permlsta .58 49 2 100 Celeus fiavus .73 50 1I/.2 0 Celeus sp. (trunk only) -- 58 1V2 0 Celeus elegans .76 54 2 15 Micropternus brachyurus .70 72 1 0 Picus vittatus .81 69 2I/.2 56 Dinopium javanense .76 83 21 -- Meiglyptes tristis .61 -- -- -- Mulleripicus pulverulentus 1.19 36 2a 0 Dryocopus pileatus (2) 1.00 63-81 21/.2 51-66 Dryocopus lineatus .82 66 21 39 Asyndesmus lewis .74 40 2 0 Melanerpes erythrocephalus (2) .69 38-52 2 100 Melanerpes Iormicivorus .68 30 2 100 Centurus striatus .67 42 2 100 Centurus carolinus (4) .73 38-47 2-2I/.2 17-67 Centurus pucherani .66 41 2 100 Sphyrapicus varius .55 32 21 45 Sphyraplcus thyroideus .54 16 21/ 100 Trichopicus cactorum .55 52 2 -- Veniliornis passerinus .54 66 2 V2 100 Dendropicos Iuscesens .47 49 21 100 Dendrocopos villosus .67 52 21/.2 100 Dendrocopos pubescens (2) .50 47 2 100 Dendrocopos scalaris .53 25 3 89 Dendrocopos nuttallii .57 67 2 100 Dendrocopos albolarvatus .62 40 21/'2 74 Picoides arcticus .66 68 2 100 Sapheopipo noguchii (trunk only) -- 54 212 100 Xiphidiopicus percussus .66 39 2 66 Mesopicos goertae .66 51 11/.2 84 Thripias namaquus .70 39 21/ 100 Hemicircus concretus .55 105 3 47 Blythipicus rubiginosus .73 74 2/2 80 Chrysocolaptes lucidus .81 61 3 100 Phloeoceastes guatemalensis .87 43 3 100 Phloeoceastes melanoleucos .91 44 3 100 Phloeoceastes leucopogon .83 68 31/.2 100 Campephilus principalis 1.11 49 31 -- Where more than one specimen was examined, the number is shown in parentheses. indication of individual variation was obtained where several specimens were studied, particularly for Colaptes and Centurus. For many of the features to be considered, additional information was available from published works. Descriptions of the salivary glands and hyoid apparatus appear in the works of Leiber (1907), Antony (1920), and Steinbacher (1934, 1957). Information concerning adaptations for climbing and pecking has been obtained from the studies by Burt (1930), Beecher (1953), Bock and Miller (1959), and Spring (1965). For the limb muscles those of Burh Hudson (1937), Lowe (1946), and Hudson and Lanzillotti (1955) were used. The papers of Lucas (1895) and Scharnke (1931) contain data on the tongue and hyoid horns. For comparison with species outside the family, in addition to use of published information I dis- sected other specimens, particularly passerines. The muscle terminology employed by George and Berger (1966) has usually been followed, but for the tracheal and hyoid muscles Leiber's terms have been used in some cases. As most of the structures studied are uniformly developed throughout the family and some show intraspecific variation, of the hundreds of potentially useful features only a few may be of value as phylogenetic indicators. The first part of the dis- cussion that follows considers about 30 features that may have potential value. An attempt has been made to judge whether a feature is "primitive" or "advanced" depending largely on its occurrence in other birds. Occurrence in Jynx or piculets is considered to be supporting evidence of a feature's primitiveness. The next por- tion of the paper considers groups of individual genera on the basis of anatomical similarity and interrelationships. More descriptions and analyses of function es- pecially of woodpecker limb muscles are needed, but this paper discusses only those structures that may be important for phylogenetic considerations within the family. AIVAL3/4SIS Ot' IIVDIVDUAI. FEATURES Latisslmus dorsi.--Pars anterior (Figure 1) takes origin in woodpeckers from the last two cervical spinous processes and in most species also from the first thoracic. Pars posterior, not previously reported in woodpeckers, is present in Jynx, piculets, Campephilus, and Phloeoceastes (Figure 1). Origin is by an aponeurosis deep to sartorius. It inserts on the humerus proximal to that of pars anterior. A humeral anchor similar to that described by George and Berger (1966) is present. The pres- ence of pars posterior without doubt represents the retention of a primitive feature. Rhomboldeus superJicialis.--Origin (Figure 1) is from the last cervical spinous process or first thoracic to mainly the third or sometimes the fourth thoracic spinous process. Probably the area of origin has been reduced in woodpeckers as an origin exists from the last cervical in Jynx, Nesoctites, and at least some passerines. In Jynx, Nesoctites, Camperhera, Centurus carolinus, Sphyraplcus varius, Xiphidiopicus, Mesopicos, Blythipicus (Figure 3), and tlemicircus the muscle has a small insertion on the clavicle as well as the scapula. Rhomboideus proJundus.This muscle is unusual in woodpeckers in that it con- sists of a small anterior superficial portion and a larger posterior deep portion. Origin of the anterior superficial portion usually is from part or all of the last two cervical spinous processes, and insertion is on the dorsal margin of the scapular blade near the bend (Figure 3). In Jynx it has a much broader insertion and is not easily separated from the posterior deep portion. In piculets the origin of the Delt. mai. Hum, TrL dor. ant, post. Rhorn. prof. Figure 1. Phloeoceastes leucopogon. Dorsal view of left shoulder region. Deltoideus major has been reflected to show insertions of latissimus dorsi. posterior deep portion is fleshy from the ilium in addition to that from the thoracic spinous processes. Serratus pro/undus.---Area of origin of this muscle has been reduced as only Jynx has an origin from the last cervical rib; in others it is from two to four cervical transverse processes only. In Dryocopus, Mulleripicus, melanerpines, Sphyrapicus, Dendrocopos scalaris, D. albolarvatus, Xiphidiopicus, Chrysocolaptes, and Phloeo- ceastes origin includes the next to the last cervical vertebra. In others it is from two or three more anterior (Figure 2). Serratus anterior.--Origin in most woodpeckers is from the last cervical rib (Figure 2), but in a few including Jynx, Nesoctites, Sasia, Campethera, Micropternus, Dinopium, Melanerpes /ormicivorus, Centurus pucherani, Xiphidiopicus, and Thripias it is also from the first thoracic rib, and in Picumnus and Mesopicos from the first thoracic only. Origin from two ribs is usual in passerines and other birds. Origin from only one rib is considered more specialized. Serratus posterior.--The origin of this muscle is variable. It is most commonly from thoracic ribs 2, 3, and 4 (Figure 2), but in Jynx, Nesoctites, Sasia, and Picoides it is from ribs 2, 3, 4, and 5, in Picumnus cirrhatus, Mesopicos and Phloeoceastes guatemalensis from ribs 3, 4, and 5, and in Campethera, Celeus (one side only in elegans), Mesopicos, Blythipicus, and Hemicircus (one side only) from ribs 3 and 4. Although this is a rather plastic feature, origin from a greater number of ribs (as in Jynx and others) is considerer primitive. Serratus metapatagialis.--The most common origin is from thoracic rib 2 ventral to serratus posterior (Figure 2). Origin is from ribs 2 and 3 in Jynx, Micropternus, prof. Ser. ant. Ster. troch Serr. post. Figure 2. Centurus carolinus. Thoracic region, left side, showing origins of serratus muscles and sternotrachealis. Mesopicos, and Mulleripicus. In Dinopium the origin is probably from rib 3 only. The origin in Asyndesrnus, Melanerpes erythrocephalus, Sphyrapicus varius, and Trichopicus is from ribs 1 and 2. Again this may be a highly variable feature, but because of the generally more posterior origin in Jynx and other birds, origin from ribs 2 and 3 is considered primitive. Subcoracoideus.--The coracold or ventral head of this muscle, which is small in woodpeckers (Figure 3), was not found in Picumnus and Sasia. A clavicular portion of the dorsal or scapular head is present in Jynx, piculets, and many other members of the family (Figure 3). It was not found in Micropternus, Picus, Dinopium, Chrysocolaptes, Phloeoceastes guatemalensis, P. leucopogon, Celeus elegans, Dryo- copus pileatus, Mulleripicus, and Thripias. Presence of a clavicular portion is con- sidered primitive. Pectoralis pars abdominalis.--The anterior attachment of this derreal component is variable. It may have a direct attachment to the humerus by fleshy fibers (Jynx, Picumnus, probably Sasia, Centurus carolinus, Dendrocopos albolarvatus, Xiphidio- picus, and Phloeoceastes guatemalensis), tendinous and fleshy fibers (Melanerpes Jormicivorus, Asyndesrnus, Sphyrapicus, and Phloeoceastes melanoleucos), or by apo.neurosis (Nesoctites, Chrysoptilus, Colapies, Campethera, Micropternus, Dinopium, Mulleripicus, Dryocopus, Melanerpes erythrocephalus, Centurus striatus, Dendrocopos except albolarvatus, Picoides, Sapheopipo, and Mesopicos). It has an attachment to the main portion of pectoralis a few mm from the attachment of the latter to the humerus in Celeus, Piculus, Trichopic.us, Thripias, Blythipicus, and Chryso- colapies. In Hemicircus it ends deep to the skin in the axilla. On the basis of its fleshy attachment to the humerus in Jynx and piculets, this feature is considered Rhombo prof. ant. supe, L  /Trl. $1/2ap.  . _/ J Rhomb. superf. S err' Pø$"  j /'/Cor. bra. posto Figure 3. Blythipicus rubiginosus. Medial view of muscles having attachments to scapula, coracold and clavicle. The partial origin of scapulotriceps from the clavicle is unusual in woodpeckers. primitive, but its considerable variation in other groups of birds makes its sig- nificance questionable. Flexor digitorum superficialis and flexor digitorum pro/undus.--Melanerpines and sapsuckers have a vinculum between the tendons of these two muscles at the carpometacarpus (Figure 4). Although the tendons may be enclosed by a common sheath in other birds, a vinculum has not been described before and is therefore considered to be a specialized feature. Flexor carpi ulnaris.--In most woodpeckers (Figure 4) the insertion of this muscle is by two tendons on the os ulnare, but Phloeoceasteso Campephilus, and 3/4inculum / Fl. dlg. super. Fl. carpi Uln. prof. Figure 4. Cenurus carolinus. Ventral view of distal portion of left wing. \\\11ll Fl. p. et p. d, III  Figure 5. Micropternus brachyurus. Superficial muscles of left thigh and leg. Piculus have only one tendon. The presence of two tendons is considered a primi- tive feature. Sartorius.--Origin (Figure 5) is from one to three thoracic spinous processes and from the ilium (see Table 1). A more limited origin from the spinous processes, which is common in other groups and occurs in Jynx, is considered primitive. The width of origin from the ilium calculated as a percentage of that of the iliotrochan- tericus anterior is also given. Because of considerable intraspecific and intrageneric variation this feature may be of limited value. lliotibialis.--Table 1 gives width at the origin of the posterior portion relative to that of the biceps. As it is wide in Jynx and in passerines, covering the biceps at its origin (as illustrated for Micropternus in Figure 5) a relatively wide muscle is considered primitive. Semitendinosus and semirnembranosus.--The accessory semitendinosus is present in most woodpeckers. It is known to be absent only in Dendrocopos, Dendropicos, Veniliornis, Sapheopipo, Picoides, and Sphyrapicus. Its absence is no doubt a specialized condition. The insertions of semitendinosus and semimembranosus are by distinct aponeuroses along the tibial shaft in most woodpeckers (Figure 6B), but in Jynx, piculets Blythipicus (Figure 6A) and ttemicircus they are united. The latter condition is considered primitive. Acc.  Sernil, Acc.  Semit. B G '  'l//Semim. Figure 6. Medial view of insertions of right semitendinosus and semimem- branosus on tibial shaft. A, Blythipicus rubiginosus with common tendon of inser- tion; B, Xiphidiopicus percussus with separate tendons. sheath Trach. medo trach, ring med. Trach.at. A B C Figure 7. Ventral view of anterior end of trachea to show insertions of trachealis muscle. A, Asyndesmus lewis; B, Picus vittatus; C, Dendrocopos villosus. lschioJemoralis.--In woodpeckers the origin is generally more extensive than in other birds, extending ventrally in many to the pubis. Origin in Jynx from the ischium only is assumed to be primitive. Adductor longus et brevis.--Usually anterior and posterior portions were fairly easily separated, but the two portions show some overlap. In a few species the two bellies are better described as deep and superficial. Many species have attach- ments from this muscle and from the pars media of the gastrocnemius to the meniscus of the knee joint. These were not found in Jynx, Picmnus, Sasia, Mi- cropternus, Dinopium, Veniliornis, Dendropicos, Dendrocopos, Picoides, Sapheopipo, Mesopicos, Chrysocolaptes, Phloeoceastes guatemalensis, and Campephilus. As these features are quite variable, they are of doubtful phylogenetic significance. Peroneus longus.--It is absent in Picumnus and Sasia. The main insertion is on the tibial cartilage and there may be a branch to adjacent fascia, but the latter is difficult to establish with certainty. In some it originates from the fibula. Except that its absence in some piculets is a specialization the form of the muscle appears to be of little value in determining relationships. Sternotrachealis.--In most birds the origin is from the sternocoracoidal process of the sternum (Figure 2), but in many woodpeckers it originates from the first thoracic rib. Sternal origin occurs in Jynx, piculets, Micropternus, Hemicircus, Sphyrapicus, and melanerpines and is assumed to be primitive. Cucullaris pars dermotemporalis.--Origin of this muscle is variable (Figure 9), but in Jynx, piculets (Figure 9A), Sphyrapicus, and most melanerpines (Figure 9B) it is from the posterior orbital rim. This feature, which is also found in some passerines, is considered primitive. Trachealis.--This muscle, also termed tracheolateralis, is attached posteriorly to the first bronchial ring. Anteriorly (Figure 7) it may consist of one to three parts. A medial part (Figure 7A, 7B, 7C), xvhich is present in all the woodpeckers studied, inserts ventrally near the midline on the fused portion of the trachea posterior to the larynx. A thin ventral part may extend anteriorly to the base of the tongue sheath (Figures 7B, 7C). It is present in Jynx, piculets, and several other species. A lateral part when present (Figure 7C) passes lateral to the tracheohyoid muscle and inserts on the trachea close to the insertion of the medial part. It occurs in Jynx, Dendrocopos, Picoides, Mesopicos, Veniliornis, Sapheopipo, and probably Dendropicos. It is also found in some passetines and Leiber (1907) considered it a primitive feature. Tracheohy. Geniorh3/4. Tracheohy. Figure 8. Dorsal view of anterior end of trachea to show attachments of genio- thyroid and tracheohyoid muscles. A, Celeus /lavus (right geniothyroid has been re- moved to show origin of tracheohyoid); B, Dryocopus pileatus; C, Campethera permista. Geniothyroid.--This muscle, first noted by Huber (1821) and discussed and well- illustrated by Leiber (1907), is known only in woodpeckers and is perhaps homol- ogous to the genioglossus of other birds. Origin is from the medial surface of the mandible posterior to that of geniohyoid. In those woodpeckers where it is well- developed a broad band of muscle fibers has one or two points of attachment to the trachea in the vicinity of the tracheohyoid (Figures 8A, 8B, 8C). In other species it is represented only by a few fibers to the floor of the mouth. Without knowing its history outside of the Picidae it is difficult to decide whether a well- developed muscle represents an advanced state or whether it became large early in the evolution of woodpeckers and later regressed. As it is well-developed in Jynx, the latter alternative seems more likely. Tracheohyoid.--The winding of this muscle (which may not be homologous to the tracheohyoid of other birds) around the trachea (Figure 8C) is frequently illustrated, but for the family as a whole it is an uncommon specialization. In other species the muscles of each side meet in the midline (Figure 8A) or inter- digitate (Figure 8B). The coiled condition, which apparently has occurred in- dependently more than once, is considered an adYanced feature. Geniohyoid and hyoid horns.--The long horns and associated geniohyoid muscles that encircle the head and enter the upper bill in Picus, Dinopium, and Colapres (Figure 9C) are also unusual in the Picidae. Most commonly they extend to or near the base of the bill (Figure 9D). In some individuals they deviate to the left rather than to the right. In Dendrocopos villosus, Hemicircus, and Picuranus innominatus (Figure 9A) the horns are wound around the right orbit, an obvious case of convergent evolution. Short hyoid horns are considered primitive, but a secondary reduction in length particularly in sapsuckers must be considered. Nasal gland.--The nasal gland is a compact structure lying adjacent to the nasal cavity in most woodpeckers, but it enters the orbit in Jynx, Picumnus (Figure 9A), and Sasia. This form is similar to that of passerines, and its failure to enter the orbit in other woodpeckers (Figure 9B, 9C, 9D) is assumed a more advanced feature. Maxillary gland.--An unusual and specialized feature in woodpeckers is the large expanded posterior portion that lies in the floor of the orbit (Figure 9), and whicb some authors have confused with the nasal gland. In Jynx the posterior portion Interorb. for. /Max. B  .r)er moterap. " GI. picorum Figure 9. Dissection of right side of head and orbit. A, Picumnus innominatus; B, Asyndesmus lewis; C, Chrysoptilus melanochloros; D, Phloeoceastes guatemalensis. enters the orbit but is not expanded and in Nesoctites it just reaches the pars plana. Mandibular gland.The large portion of this gland, which may be termed glandula picorum, is a specialized feature of woodpeckers (Figure 9). It is large in Jynx, piculets, and many other species. Reduction in size is believed a secondary specialization. Interorbital /oramen.--This foramen (Figures 9B, 9D), present in some wood- peckers, possibly is of phylogenetic significance, but defining an evolutionary trend for it within the family is difficult. It is absent in Jynx, Nesoctites, Colapres, Camperhera, Dinopium, Centurus carolinus, Sphyrapicus, Celeus elegans, Piculus, and Hemicircus. Other /eatures.--The great development of the protractor pterygoideus discussed by Spring (1965) was not studied in detail, but it is much smaller in Jynx. Well- developed in piculets, it probably indicates an early specialization for pecking in that group. The assumption that specialization for climbing has developed more than once within the family is considered with the other evidence. Several features distinguish Jynx from all other Picidae. These include the form of rhomboideus profundus, origin of serratus profundus from the last cervical rib, the limited origin of ischiofemoralis, origin of adductor longus et brevis from ischium and the smaller protractor pterygoideus. In addition, deltoideus minor has a clavicular origin, caudofemoralis has a fleshy as well as tendinous attachment to the pygostyle, and iliacus inserts on the anterior rather than the medial surface of the femur. Three features found only in Jynx and some or all piculets are the soft rectrices, nasal gland in the orbit, and smaller maxillary gland. Most of these features I consider primitive on the basis of their widespread occurrence in other birds, suggesting that Jynx re- sembles most closely the ancestral members of the family. The barbless tongue is probably primitive as Leiber (1907) suggested, but possibly a secondary reduction of spines has occurred. The geniothyroid and tracheohyoid are attached to fused tracheal rings (according to Leiber 5-18) some distance posterior to the larynx. The posterior attachment of tracheohyoid (like its coiling around the trachea in other species) provides a mechanism for greater protrusion of the tongue. Possibly the geniothyroid became well-developed independently and its enlarge- ment in other woodpeckers is not truly a primitive feature. The large glandula picorum, the specialized hyoid apparatus, and lack of modifi- cation of the feet, tail, and jaw muscles for climbing and pecking give evidence that ground feeding preceded the development of other spe- cializations typical of woodpeckers. The morphological evidence clearly justifies recognition of the subfamily Jynginae. Pcu.Ts (PxcuMx^) The considerable morphological variation among the piculets is not surprising in view of their discontinuous distribution. In general they are more primitive than other members of the family, but show some specializations. Primitive features found in all of the piculets studied include the presence of an accessory semitendinosus, the common tendon for semitendinosus and semimembranosus, origin of ischiofemoralis and adductor longus et brevis dorsal to the pubis, the anterior origin of dermotemporalis (Figure 9A), sternal origin of sternotrachealis, large glandula picorum (Figure 9A), and the presence of the ventral part of trachealis. Other primitive features include a clavicular origin for rhomboideus superficialis, the shorter maxillary gland, and the broad iliotibialis in Nesoctites. The nasal gland enters the orbit in Picumnus (Figure 9A) and Sasia. Specialized features in piculets include origin of serratus metapatagialis from second thoracic rib only, origin of sartorius from two or three spinous processes, absence of a lateral part of trachealis, and a poorly developed geniothyroid. Possibly the latter two muscles have developed independently in Jynx and true woodpeckers and their absence in piculets represents the primitive condition. The loss of the coracoid head of subcoracoideus in Picumnus and Sasia and the fleshy insertion of rhomboideus profundus on the ilium do not occur in other members of the family. Origin of serratus anterior from the last cervical and first thoracic rib in Nesoctites and Sasia is primitive, and origin from the first thoracic only in Picumnus is unusual. The origin of ser- ratus posterior shows considerable variation, but the generally more extensive origin probably is primitive. Sasia, which lacks a hind toe, has lost extensor hallucis longus and flexor hallucis brevis. The hyoid horns are relatively short in Picumnus cirrhatus, Sasia, and Nesoctites, but in Picumnus innominatus (Figure 9A) they coil around the right orbital rim and the tracheohyoid makes two turns around the trachea. In cirrhatus the muscles meet in the midline without crossing, in Sasia they cross slightly, and in Nesoctites they interdigitate. The piculets are more like true woodpeckers than Jynx, but they show enough differences from both other groups to justify recognizing the subfamily Picumninae. The fact that the Asian Picumnus innominatus differs somewhat from the South American P. cirrhatus suggests a possible generic separation, but more species should be studied. Nesoctites is perhaps closer to the ancestral members of the Picinae. The strongest evidence for this is the failure of the nasal gland to enter the orbit. On the other hand the maxillary gland is expanded in the orbit in Picumnus, but not in Nesoctites. MICROPTERNU$, MEIGLYPTES, AND CELEUS In these genera the glandula picorum is large, the geniothyroid is well-developed, hyoid horns are short, and the tracheohyoids are simple. Micropternus has several primitive features. The posterior portion of the iliotibialis is broad (Figure 5), sartorius takes origin from the last thoracic spinous process only (Figure 5), sternotrachealis comes from the sternum, and serratus anterior from the first thoracic as well as the last cervical rib. Origin of dermotemporalis was not determined. Genio- thyroid inserts anterior and posterior to the origin of tracheohyoid. Trachealis has ventral and medial parts, but no lateral. Meiglyptes is apparently similar, but the specimen was not dissected completely. Sternotrachealis takes origin from the sternum. Origin of the dermo- temporalis is 4 mm posterior to the orbital rim. Only the medial part of trachealis was found. On the basis of the available evidence Meiglyptes is related to Micropternus. Celeus is in many ways similar to Micropternus, but has fewer primi- tive features. Geniothyroid has only one attachment to the trachea (Figure 8A). Only the medial part of trachealis is present, origin of sternotrachealis is from the first rib, and dermotemporalis originates 5 to 7 mm posterior to the orbital rim. Serratus anterior originates from the last cervical rib only. In the Celeus elegans specimen examined ser- ratus posterior originated from ribs 2, 3, and 4 on one side and from ribs 3 and 4 on the other. In the other two species studied origin was ribs 3 and 4 only. In woodpeckers flexor perforans et perforatus digiti III usually has two heads (Figure 5), but it has only one head in elegans. Insertion of pectoralis abdominalis is to pars thoracicus 3 to 5 mm from the humerus. Origin of sartorius is only slightly more ex- tensive than for Micropternus, and the posterior portion of iliotibialis is narrower. Despite these differences Celeus appears to belong to the same evolutionary line as Microptcrnus. Pcuœus ^N) CAM'ErHEtA Both Piculus and Campethera have a large glandula picorum and a well-developed geniothyroid. In Piculus the origin of sternotrachealis is from the first rib. The origin of dermotemporalis is posterior to the ear opening. Tracheohyoid reaches the base of the right nostril and the tracheohyoids cross and make a three-fourths turn around the trachea. Origin of pectoralis abdominalis is similar to that of Celeus, but that of serratus posterior is thoracic ribs 2, 3, and 4. Origin of sartorius is limited to 1% thoracic spines, and the posterior portion of iliotibialis is of moderate width. Both medial and ventral parts of trachealis are present. The hyoid horns of Camperhera are shorter, but the tracheohyoids (Figure 8C) make a three-fourths turn around the trachea as in Piculus. The geniothyroid has only one attachment anterior to the tracheohyoid (Figure 8C), whereas in Piculus it also has a small posterior attachment. Previous dissection prevented determining the origin of sternotrachealis. Origin of dermotemporalis is dorsal to the ear opening. A primitive fea- ture is the origin of serratus anterior from the last cervical and first thoracic ribs. As in Celeus the origin of serratus posterior is from thoracic ribs 3 and 4, but pectoralis abdominalis has a humeral origin. On the basis of their anatomy Piculus and Camperhera appear to be re- lated and both show similarities to Celeus, but the similar winding of the tracheohyoid around the trachea in Piculus and Campethera may represent convergence in New and Old World lines even though both may have evolved from the same general group. COLAPTES, ?ICUS, AND DINOPIUM These are similar to the genera discussed under the previous heading in having a large glandula picorum and well-developed geniothyroid, but they show greater coiling of the tracheohyoid and extension of the hyoid horns into the upper bill (Figure 9C). Chrysoptilus and Colapres are similar, and this discussion makes no distinction between the two genera. In Colaptes the posterior portion of iliotibialis is broad and the origin of sartorius is from 2 to 21/'2 spinous processes. Pectoralis abdominalis is attached to the humerus by apo- neurosis. Dermotemporalis is attached to the skull posterior to the ear opening (Figure 9C). In Colapres, as in Piculus, Campethera, and Celeus elegans, no interorbital foramen is present, but one was noted in Celeus fiavus. The origins of the serratus muscles are all of the more advanced type. Trachealis has a ventral as well as a medial part (Figure 7B). Few differences were noted between Picus and Colapres. In Pious the posterior portion of iliotibialis is somewhat narrower, subcoracoideus has no clavicular head, and the geniothyroid has an additional attach- ment to the larynx. Dinopium has a similar morphology; the posterior portion of iliotibialis is broad as in Colaptes, but as in Picus the genio- thyroid has an attachment to the larynx and the subcoracoideus has no clavicular head. Dinopium lacks the interorbital foramen as does Coiaptes; specimen damage prevented determining its presence or absence in Picus. Origin of serratus anterior from the first thoracic and last cervical ribs, serratus posterior from ribs 1, 2, 3, and 4 and the origin of serratus metapatagialis from rib 3 are considered primitive features in Dinopium. It has lost the hind toe, but the extensor hallucis longus is well-developed and flexor hallucis brevis is present although very small. Flexor perforans et perforatus digiti III has only one head instead of the usual two. Their similar anatomy suggest that Colaptes, Picus, and Dinopium derive from the same general ancestral group, but such similarities as the coiling of the tracheohyoid and the entrance of the hyoid horns into the bill may be due to convergent evolution. Further study is needed, but I favor the view that the three genera are related and that the hyoid horns have grown into the bill only once. DRYOCOPU$ AND MULLERIPICU$ Dryocopus and Mulleripicus are similar to the three preceding groups in having well-developed geniothyroids (Figure 8B). Glandula picorum is somewhat smaller in Dryocopus, but is well-developed. Dermotemporalis inserts on the skull posterior to the ear opening. Trachealis has medial and ventral parts, the hyoid horns reach the base of the bill and the tracheohyoids interdigitate (Figure 8B). Iliotibialis is of moderate width and sartorius originates from 2 to 2,4 spinous processes. Only one head of flexor perforans et perforatus digiti III is present in D. lineatus. An interorbital foramen is present. The feet have become more specialized for climbing. Mulleripicus differs little from Dryocopus, but a primitive feature is the origin of serratus metapatagialis from thoracic ribs 2 and 3. Glandula picorum is larger and the posterior portion of iliotibialis is narrower. They resemble Micropternus in having an attach- ment of geniothyroid anterior and posterior to tracheohyoid and a short hind toe, and in Mulleripicus the origin of serratus metapatagialis from thoracic ribs 2 and 3. Unlike Micropternus origin of sternotrachealis is from the first rib and serratus anterior has an origin from the last cervical rib only. MELANERPINE WOODPECKERS AND SPHYRPICU$ In this group and the ones to follow, in contrast to those previously discussed the geniothyroids are poorly developed with only a few fibers to the floor of the mouth, and glandula picorum in general is less well- developed (Figure 9B). The distal horny part of the tongue is usually longer and the foot is more adapted for climbing. Centurus, Melanerpes, Asyndesmus, and Tripsurus are morphologically similar and in this anal- ysis they are usually considered together. The melanerpine woodpeckers have several primitive features including origin of sternotrachealis from the sternum (Figure 2), origin of a dermotemporalis from near the orbital rim (Figure 9B), and the lack of specialization of the tracheohyoids and hyoid horns. Glandula picorum is moderately long and slender. Origin of serratus metapatagialis from the first and second thoracic ribs in some species and the presence of a vinculum (Figure 4) between the tendons of flexor digitorum super- ficialis and flexor digitorum profundus (except in Centurus striatus) are distinctive features. Subcoracoideus has a clavicular head. Serratus anterior originates from the last cervical rib (Figure 2) except in Melanerpes formicivorus and Centurus pucherani in which the origin is the last cervical and first thoracic. Posterior portion of iliotibialis is narrow to moderate in width and the origin of sartorius is from 2 to 2 spinous processes. Origin of dermotemporalis is more posterior in Cen- turus striatus. The interorbital foramen is small in most species (Figure 9B) and absent in the specimens of Centurus carolinus studied. A small ventral part of trachealis was found on one side in Centurus pucherani, but it was absent in the other species (Figure 7A). Tracheohyoids interdigitate markedly in striatus and carolinus, less so in pucherani and Melanerpes formicivorus, and not at all in Asyndesmus and Melanerpes erythrocephalus. The length of the hyoid horns is also variable with the longest reaching the base of the bill. In Trichopicus serratus posterior originates from thoracic ribs 1, 2, and 3 and glandula picorum is larger, but it is otherwise similar. The posterior portion of iliotibialis is relatively wide and the glandula picorum not reduced in size, but the vinculum between the flexor tendons and the origins of serratus posterior and metapatagialis are specialized fea- tures. Sphyrapicus resembles Dendrocopos by the absence of accessory semi- tendinosus and, according to Burt (1930), by the greater folding of the frontal region of the skull, but in other ways it resembles the melanerpines. Unlike Dendrocopos, it has a vinculum between the tendons of flexor digitorum superficialis and flexor digitorum profundus, origin of sterno- trachealis is from the sternum, dermotemporalis originates from the orbital margin, glandula picorum is long and slender, and trachealis has no lateral part. As in most melanerpines the interorbital foramen is absent, and in Sphyrapicus varius serratus metapatagialis takes origin from thoracic ribs 1 and 2. The very short hyoid horns and the origin of the tracheohyoids from the lateral surface of the trachea may represent a secondary reduction. DENDROCOPOS WOODPECKERS In Mesopicos, Dendrocopos, Picoides, Sapheopipo, Dendropicos, Venili- ornis, and Thripias the glandula picorum is short, trachealis has a lateral part (Figure 7C), sternotrachealis is from the first thoracic rib, dermo- temporalis originates dorsal to the ear opening, geniothyroid is poorly developed, and an interorbital foramen is present. The foot has become specialized for climbing. In Dendrocopos accessory semitendinosus is absent. Posterior portion of iliotibialis is narrow to moderately broad, and sartorius originates from 2 to 3 spinous processes. The tracheohyoids extensively interdigitate. Hyoid horns are of moderate length, except in villosus in which they coil around the right orbital margin. I examined only a trunk specimen of Sapheopipo, but found no dif- ferences to support its separation from Dendrocopos. Picoides is similar to Dendrocopos, but the origin of serratus posterior from thoracic ribs 2, 3, 4, and 5 may be evidence for a more basal position. Flexor hallucis brevis is of moderate size, but extensor hallucis longus is weakly de- veloped. Although the loss of the hind toe is not fundamental, this along with the different origin for serratus anterior (if the latter is a consistent difference) favors the separation of the two genera. No distinctive fea- tures were noted for Veniliornis. It may represent a relatively recent offshoot from North American Dendrocopos, or possibly it is more primi- tive. Mesopicos is clearly related to Dendrocopos, but it has several features thought to be primitive. These include the presence of accessory semi- tendinosus, origin of serratus anterior from the first thoracic rib, ser- ratus posterior from ribs 3, 4, and 5, serratus metapatagialis from ribs 2 and 3, and the more limited origin of sartorius from iV., spinous processes. The tracheohyoid muscles interdigitate. Dendropicos closely resembles Dendrocopos, but serratus posterior originates from thoracic ribs 2 and 3 only. No ventral part of trachealis was found and the presence of a lateral part was somewhat doubtful. This point needs further study, but apparently Dendrocopos and Dendropicos are related. The specimen of Thripias studied was so poorly preserved in some parts it is discussed with some reservation. No determinations were made for trachealis or tracheohyoids, but for other features such as the loss of the accessory semitendinosus, Thripias is similar to Dendropicos. A primitive feature is the origin of serratus anterior from the first thoracic and last cervical ribs. Origin of pectoralis abdominalis is from pars thoracicus. X1PHIDIOPICUS The Cuban Green Woodpecker is similar to Dendrocopos woodpeckers in having a short glandula picorum, a posterior origin for the dermo- temporalis, poorly developed geniothyroid, interdigitation of the tra- cheohyoids, origin of sternotrachealis from the first rib, and the presence of an interorbital foramen. Like Mesopicos, an accessory semitendinosus is present (Figure 6B). As in melanerpine woodpeckers ventral and lateral parts of trachealis are absent. A primitive feature is the origin of serratus anterior from the last cervical and first thoracic ribs. Perhaps Xiphidiopicus is an isolated remnant of the ancestral stock that gave rise to Dendrocopos and its relatives. No morphological evidence indi- cates that Xiphidiopicus is a recent offshoot from Den&ocop.os or the melanerpines. IVORY-BILLED WOODPECKERS Latissimus dorsi posterior is present in Campephilus and Phloeoceastes leucopogon and was found on one side in melanoleucos. Glandula picorum is short and the distal portion is narrow (Figure 9D). Only the medial part of trachealis is present and the tracheohyoids meet in the midline or interdigitate slightly. Geniothyroids are poorly developed although, P. leucopogon and Chrysocolaptes have a few fibers to the trachea or larynx. The hyoid horns are of moderate length. An interorbital foramen is present (Figure 9D), except that this feature was not examined in Campephilus. Origin of sternotrachealis is peculiar in Campephilus and Phloeoceastes in that in addition to the attachment to the first rib there is a small slip to the wall of the interclavicular air sac. Campephilus, Phloeoceastes, and Chrysocolaptes are similar in that dermotemporalis in- serts posteriorly and is far ventral in position, being immediately dorsal to depressor mandibulae (Figure 9D). Sartorius takes origin from 3 to 31/2 spinous processes, posterior portion of iliotibialis is of moderate width, peroneus longus arises in part directly from the fibula, extensor hallucis longus has an additional attachment to the first phalanx, and flexor perforans et perforatus digiti III has only one head. In Campe- philus and Phloeoceastes the tendon of flexor carpi ulnaris to the os ulnare is not divided. In most features Chrysocolaptes resembles the American ivory-bills. It resembles Blythipicus in the origin of sternotrachealis from the first rib only, the attachment of pectoralis abdominalis to pars thoradcus instead of the humerus, and the insertion of flexor carpi ulnaris. Bly- thipicus (Figure 6A) and Hemicircus have a common tendon of insertion for semitendinosus and semimembranosus, a feature found also in Jynx and piculets, but not in other woodpeckers. Hemicircus has more features in common with Blythipicus than the other ivory-bills, but resembles them in the posterior ventral origin of dermotemporalis, and the origin of sartorius from 3 spinous processes. It differs from the others by the attachment pectoralis abdominalis to skin only, origin of sternotrachealis from sternum, lack of an interorbital foramen, the very broad posterior part of iliotibialis, the coiling of the hyoid horns around the orbit, and the winding of the tracheohyoid around the trachea. Campephilus and Phloeoceastes are clearly related and do not show a close relationship to Dryocopus. The resemblance of Chrysocolaptes to the American ivory-bills is difficult to explain on the basis of convergence. Blythipicus and Hemicircus are less like the American species, but they do have features in common with Chrysocolaptes. Primitive features present in some ivory-bills are the presence of latissimus dorsi posterior, common tendon for semimembranosus and semitendinosus, sternal origin of sternotrachealis, and broad posterior portion of iliotibialis. Superficial resemblances exist between Chrysocolaptes and Dinopium, Hemicircus and Meiglyptes, and Blythipicus and Micropternus, but most of the anatomical evidence does not support these relationships. Flexor perforans et perforatus digiti III has only one head in Dinopium as in Chryso- colaptes, but this was true also for Dryocopus lineatus and Celeus elegans. In such other features as the origin of the serratus muscles, the size of the glandula picorum, the relative development of geniothyroids, tracheo- hyoids, and the hyoid horns, Dinopium and Chrysocolaptes are dissimilar. Hemicircus, Meiglyptes, and Micropternus are alike in having a sternal origin for sternotrachealis, but the two ivory-billed genera Hemicircus and Blythipicus resemble closely neither Meiglyptes nor Micropternus. GENERAL DISCUSSION In his consideration of convergence in sympatric species Cody (1969) compares the genera of ivory-billed woodpeckers Hemicircus, Blythipicus, Campephilus and Phloeoceastes with the "logcock" genera Meiglyptes, Micropternus, Dinopium and Dryocopus which, in spite of striking simi- larities in plumage among species pairs from the two groups, can be distinguished by differences in morphology, behavior, and ecology. Bock (1963) doubts that these similarities of complex plumage patterns can be explained on the basis of convergence, but the evidence obtained in the present study tends to support Cody's view. One argument against the independent origin for the different ivory-bill genera from less specialized ground woodpeckers is that in general the ivory-bills have other primitive features not found in their supposed ancestral groups. Phloeoceastes and Campephilus have a latissimus dorsi posterior and a different origin for sternotrachealis, which seems to rule out an origin from Dryocopus. Chrysocolaptes lacks the specialization of the tracheo- hyoid and hyoid horns of Picus and Dinopium. Blythipicus and Hemi- circus have retained the common tendon for semitendinosus and semi- membranosus, unlike their possible relatives Micropternus and Meiglyptes. If features such as the presence of a latissimus dorsi posterior were present throughout the ivory-bill group their close relationship would be on much firmer ground, but they no doubt represent an earlier radia- tion from the ground woodpeckers and their anatomical diversity is therefore to be expected. In summary convergence in plumage patterns, although remarkable, is probably more easily accepted than the opposite view, which demands a great deal of structural convergence. General morphological similarity is evident in Micropternus, Meiglyptes, Dryocopus, Mulleripicus, Celeus, Piculus, Camperhera, Picus, Colapres, and Dinopium, and despite certain specializations of the hyoid apparatus, all appear to be related at least distantly. In forms specialized for a more arboreal habitat, the hind toe is small or absent. A major question is whether the great resemblance of Picus to Colaptes indicates close relationship or is the result of convergent evolution in the Old and New Worlds. Certainly more species, particularly of Picus, need to be studied. Burt (1930) and Howell (1952) concluded that Sphyrapicus is a derivative of Dendrocopos, but I agree with Short and Morony (1970) that sapsuckers are more closely allied to the melanerpine woodpeckers. Possibly the melanerpines and pied woodpeckers had a common origin, but in any case the evidence strongly favors a closer relationship of Sphyrapicus to the melanerpines than to Dendrocopos. Goodwin (1968) favors the position that all the sub-Saharan African woodpeckers are more closely related to one another than to non-African forms. His alternative that they are polyphyletic is supported by the anatomical evidence, but I suggest that Mesopicos, Thripias, and Dendro- picos are related to Dendrocopos, whereas Campethera is closer to Picus. Goodwin concludes that similarities in plumage of certain Dendropicos and Veniliornis species are due to convergence, but the anatomical evi- dence supports a relationship of the latter to Dendrocopos also. Geo- colapres was not studied, but its relationship to other African forms or to Picus or Colaptes could probably be determined by its internal morphology. ACKNOWLEDGMENTS This study would not have been possible without the most generous help given to me by the curators of several institutions, and others, in the form of specimens for dissection. The aid of P. J. K. Burton (British Museum), W. H. Elder (University of Missouri-Columbia), R. F. Johnston (University of Kansas), C. G. Sib]ey (Yale University), and especially R. L. Zusi (U.S. National Museum) is most gratefully acknowledged. Specimens were also most kindly provided by I. C. Adams, L. Saxe, and L. W. Spring. Waiter J. Bock read the paper critically and made constructive comments. SUMMARY An anatomical study of 33 genera and 47 species of woodpeckers was carried out to learn more about interrelationships and the evolutionary history of the Picidae. About 30 features of possible phylogenic significance are described. The wrynecks and piculets possess many primitive features and on the basis of their anatomy can be distinguished from the Picinae. Some primitive members of the latter subfamily have retained adaptations for ground feeding, with well-developed salivary glands and probing tongues with little specialization for pecking in more solid wood. An early radiation of the ivory-billed woodpecker group apparently occurred with the development of specializations for climbing and pecking. Dryocopus and Mulleripicus are not directly related to the ivory-bills, but probably evolved more recently from the "ground woodpeckers." Picus and Colaptes with their highly specialized tongue and hyoid apparatus are probably closely related and represent terminal branches in the radiation of the "ground woodpeckers." The melanerpine woodpeckers and sapsuckers are related and comprise another group that has evolved a climbing foot and specialized feeding apparatus, although retaining many primitive features. The pied woodpeckers comprise another group whose members also have more arboreal habits and a bill more highly developed for pecking. LITERATURE CITED ANTONY, M. 1920. lber die Speicheldriisen der ViSgel. Zool. Jahrb., 41: 547-660. BEECER, W. J. 1963. Feeding adaptations and systematics in the avian order Piciformes. J. Washington Acad. Sci., 43: 293-299. Boca<, W. J. 1963. Evolution and phylogeny in morphologically uniform groups. Amer. Naturalist, 97: 265-285. Boca<, W. J. 1967. The use of adaptive characteristics in arian classification. Proc. 14th Intern. Ornithol. Congr., 61-74. Boca<, W. J., AND W. DEW. MILLER. 1959. The scansorial foot of the woodpeckers, with comments on the evolution of perching and climbing feet in birds. Amer. Mus. Novitates, No. 1931: 1-45. BURT, W. H. 1930. Adaptive modifications in the woodpeckers. Univ. California Publ. Zool., 32: 455-524. Co)3/4, M. L. 1969. Convergent characteristics in sympatric species: a possible relation to interspecific competition and aggression. Condor, 71: 222-239. GEoRos, J. C., ^m) A. J. Bsaosa. 1966. Avian myology. New York, Academic Press. GooDwI, D. 1968. Notes on woodpeckers (Picidae). Bull. Brit. Mus. (Nat. Hist.) Zool., 17: 1-44. HOWELL, T. R. 1952. Natural history and differentiation in the Yellow-bellied Sapsucker. Condor, 54: 237-282. HuBsa, V.A. 1821. De lingua et osse hyoideo Pici viridis. Innaug. Diss., Stuttgart. Hu)so, G. E. 1937. Studies on the muscles of the pelvic appendage in birds. Amer. Midi. Naturalist, 18: 1-108. Htmso, G. E., ^m> P. J. L.aZXLLOTZX. 1955. Gross anatomy of the wing muscles in the family Corvidae. Amer. Midi. Naturalist, 53: 1-44. Lma, A. 1907. Vergleichende Anatomie der Spechtzunge. Zoologica, 20 (Heft 51): 1-79. Lows, P. R. 1946. On the systematic position of the woodpeckers (Pici), honey- guides (Indicator), hoopoes and others. Ibis, 88: 103-127. LJc^s, F. A. 1895. The tongues of woodpeckers. U.S. Dept. Agr., Div. Ornithol. Mamm., Bull. 7: 35-45. MAYR, E. 1955. Comments on some recent studies of song bird phylogeny. Wilson Bull, 67: 33-44. Scxax, H. 1931. BeitrSge zur Morphologie und Entwicklungsgeschichte der Zunge der Trochilidae, Meliphagidae und Picidae. J. Ornithol., 79: 425-491. Saoar, L. L., A J. J. Moao3/4. 1970. A second hybrid Williamson's X Red- naped Sapsucker and an evolutionary history of sapsuckers. Condor, 72: 310- 315. Seaxc, L. W. 1965. Climbing and pecking adaptations in some North American woodpeckers. Condor, 67: 457-488. SxIcasa, J. 1934. Untersuchungen fiber den Zungenapparat indischer Spechte. J. Ornithol., 82: 399-408. Szx^cxm, J. 1957. Erber den Zungenapparat einiger neotropischer Spechte. Senck. Biol., 38: 259-270. Department o] Anatomy, University of Missouri-Columbia, School of Medicine, Columbia, Missouri 65201. Accepted 1 December 1970.