Society of Systematic Biologists

Has the Biological Species Concept Outlived Its Usefulness? Author(s): Paul R. Ehrlich Source: Systematic Zoology, Vol. 10, No. 4 (Dec., 1961), pp. 167-176 Published by: Taylor & Francis, Ltd. for the Society of Systematic Biologists Stable URL: http://www.jstor.org/stable/2411614 Accessed: 04-01-2016 19:22 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/ info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Taylor & Francis, Ltd., Society of Systematic Biologists and Oxford University Press are collaborating with JSTOR to digitize, preserve and extend access to Systematic Zoology. http://www.jstor.org

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions SYMPOSIUM ON PHILOSOPHICAL SYSTEMATICS 167 more than abstract logical forms and catego- REFERENCES ries. They are habits, predispositions,deeply BRIDGMAN,P. W. 1936. The nature of physi- engrained attitudes of aversion and prefer- cal theory. Dover Publications, New York. ence. Moreover, the conviction persists- CLAUSEN, J. 1960. A simple method for the though history shows it to be a hallucination sampling of natural populations. Scottish -that all the questions that the human mind Plant Breeding Sta. Report: 69-73. has asked are questions that can be answered DARWIN, C. 1859. The origin of species. in terms of the alternatives that the questions DEWEY, C. 1957. The influenceof Darwinism themselves present. But in fact intellectual on philosophy. In M. GARDNER(ed.), Great progress usually occurs through sheer aban- essays in science. Pocket Library, Henry donment of questions together with both of Holt, New York. the alternatives they assume-an abandon- HARRINGTON,W. 1954. Flora of Colorado. ment that results from their decreasing vi- PERRY, J. W. 1960. Theory of documentation tality and a change of urgent interest. We do and strategy of searching. Apr. 12 and 13, 1960, Cleveland, Ohio. Contract No. AF 49 not solve them: we get over them. Old ques- (638)-357. (mimeographed). tions are solved by disappearing, evaporating, while new questions corresponding to the NORMAN H. RUSSELL, JR. is Professor of changed attitude of endeavor and preference Botany at Arizona State University, Tempe, take their place . . ." (Dewey, reprint 1957) Arizona.

Has the BiologicalSpecies Concept OutlivedIts Usefulness? PAUL R. EHRLICH

IN THE PRESENT PAPER I would the "modern synthesis" in evolution liketo considerthe following question: would deny that reproductiveisolation is Is the biologicalspecies concept a use- a major factorin the evolutionaryprocess. ful taxonomictool? I do not wish to re- It does, however, seem pertinentto ques- examine the philosophicalquestion of tion (even at the risk of being accused whetheror not therereally are distinct of embracing the heresies of typologyand entitiesin nature which correspondto anti-evolutionism) using this isolation as our idea of "biological species." This a criterion for a taxonomic grouping. problemhas been dealt with fully by Grant (1957:74-75) states Burma (1954), Mayr (1957a), and others. "Accordingto the biologicalspecies concept In additionI will ignorethe relatedques- the species is a populationset apart fromthe tion of the degree to which apparent rest of the livingworld by reproductiveiso- structuringin natureis an artifactcaused latingmechanisms. Its boundariestherefore shouldbe markedby a prominentgap in the by the structuringof the vertebratecen- variationpattern. Good discretespecies are tral nervous systemor by the thought in factfound in all majorplant groups. These patternscharacteristic of Western culture. well-definedspecies, however, constitute only Whateverits degreeof reality,there is a fractionof the populationsin theirrespec- no doubt that in the past the biological tivephylads. The botanistfinds problematical species intermixedin varying proportions species concepthas been usefulin help- withgood species in mostof the familiesand ing to shape our thinkingabout evolu- genera. This is in markedcontrast with the tionaryproblems. No one familiarwith situationwhich the zoologistfinds in higher

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions 168 SYSTEMATIC ZOOLOGY in which clearly circumscribedspe- words,forms having a wide spatial dis- cies are a rule and poorlydefined ones the ex- tribution,hut superficiallynot showing ception." greatgeographic variation, were not con- In a sample survey of some genera of sideredas presentingany problemat the plants, Grant found the percentage of species level-althoughinterfertility may "good" (i.e., distinct or easily delimited) not always existwhere it is assumed. species to range from 0 (California The genera of NorthAmerican butter- Ceanothus) to 100 (North American fliesare groupedbelow (somewhatarbi- Asclepias). trarily)into seven categories,according There seems to be general agreement to the frequencyof "good" species found that the biological species concept has withinthem. It was ratherdifficult, in no meaning whatever when one is deal- manycases, to decideexactly which cate- ing with asexual organisms, and is quite gorywas properfor a givengenus, which limited in its applicability in planfts. in itselfwould cast doubt on the overall However, what of the higher animals utilityof the speciesconcept. If biological where "circumscribedspecies are the rule species were readily recognizable,we and poorly defined ones are the excep- would have no difficultyin classifying tion?" Mayr (1957b) has suggested that generaaccording to the proportionof dis- analyses such as were done by Grant tinctspecies found within them. should be carried out for as many groups Group 1. Generain which the species of plants and animals as possible. As a seem quite distinct,and are considered preliminarystep in considering the use- by most workersto presentno serious fulness of the concept, I have done suc'h problems: Parnassius, Neophasia, Phoe- a survey of a group with which I am bis, Lethe, Euptoieta, Poladryas, Micro- familiar, the Nearctic (Papil- tia, Nymphalis, Vanessa, Anartia, Caly- ionoidea) . copis, Erora, Brephidium, Apodemia. , There are few, if any, groups of equiva- Group2. Generain whichmost species lent size that are as well-knownsystema- seemdistinct, but the status of some forms tically as the butterflies.Vast collections is in doubtat the presenttime: Eurema, of them have been amassed, and the litera- Anthocaris, Euchloe, Euptychia, Erebia, ture is replete with observations on their Ministrymon,Euristrymon, Strymon, Phi- distribution and (to a lesser degree) lotes. their genetics and behavior. If the bio- Group 3. Genera in which many or logical species concept is useable, it most forms present serious problems: should be easily applicable to a well- Papilio, Pieris, Oeneis, Speyeria, Boloria, known, diploid and outcrossing group of Euphydryas, Chlosyne, Phyciodes, Poly- "higher" animals such as the Nearctic gonia, Precis, Limenitis, Anaea, Libythea, butterflies. Chlorostrymon,Satyrium, "Thecla," Cal- In my survey of the butterflies,several lophrys,Lyeaena, Plebejus, Everes. things were done which would tend to Group4. Generain which,at the pres- overemphasize the utility of the biol6gi- ent stateof our knowledge,the biological cal species concept. First of all I was species conceptis virtuallyinapplicable: surveyinga limitedfauna, and it is widely Colias, Coenonympha, Cercyonis. admitted that species are more clearcut Group 5. Generawhich are monobasic: when viewed within a limited geographi- Paramecera, Neominois, Agraulis, DJryas, cal area. Thus the problem of the specific Dryadula, Habrodais, Hypaurotis, Phaeo- distinctness (or lack of same) of the strymon,Chrysophanus, Dolymorpha, Hy- Nearctic representatives of Holarctic postrymon, Panthiades, Glaucopsyche, formswas not considered. In addition the Phaedrotes, Cetastrina. question of possible undetected repro- Group 6. Genera which are not well ductive barriers was ignored. In other representedin the Nearcticregion, but

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions SYMPOSIUM ON PHILOSOPHICAL SYSTEMATICS 169

best informationwould indicate that they tween the Palaearctic and Nearctic forms belong in Group 1 or Group 2: Nathalis, recentlylumped as vau-album. For many Anteos, Historis, Eumaeus, Leptotes. years the Nearctic form was considered Group 7: Genera which are not well specifically distinct and went under the represented in the Nearctic region, but name j-album. best informationwould indicate that they Erebia is a genus typical of those belong in Group 3 or Group 4: Battus, in which most species seem distifict Graphium, Parides, Enantia, Lycorella, (Group 2). Ignoring the status problem Heliconius, Metamorpha, Hypolimnas, of our members of Holarctic series, and Myscelia, Diaethria, Eunica, Mestra, the possible reproductive incompatibility Hamadryas, Biblis, Dynamine, Marpesia, of individuals from various populations Apatura, Tmolus, Cyanophrys, Atlides, of widely and discontinuously distri- Electrostrymon, Hemiargus, Agriades, buted "species" such as theano, we Zizula, Euselasia, Emesis, Lasaia, Caria. find eight distinct segregates (the'ano, Some of the genera listed under discoidalis, vidleri, callias, disa, rossii, Group 7 may well move into Group 6 youngi, and epipsodea), and one diffi- when we know more about the many cult complex (magdalena-mackinleyensis- tropical groups represented there. I will fasciata-avinoffi). It seems unlikely that confine my further comments to those this complex will ever prove amenable to genera in the firstfour groups. pigeonholing. I would like to discuss a few of the Papilio is a large genus containing genera in greater detail in order to give many taxonomicallyrefractory complexes some idea of the necessarily rather loose (Group 3). Most outstanding in North standards employed in assigning them America is the machaon complex. A quote to the differentcategories. fromRemington (who has been working Nymnphaliswill serve as an example on the elucidation of the genetic relation- of the genera with distinct species ships of formsin this complex) will serve (Group 1). The recognition of four dis- as an illustration of the difficultiesen- tinct segregates (antiopa,1 milberti,cali- countered. He discusses (1958) the fornica,and vau-album) in North Amer- forms of the machaon complex in the ica dates at least back to Morris (1862), Riding Mountain region of Manitoba and and to my knowledge this arrangement tentativelyconcludes: has never been seriously challenged. "1) RidingMountain is the locus of origin Specimens may be assigned to one of the of P. kahli, an isolated,distinctive offshoot four forms with certainty. It should be of P. polyxeneswith black wings, spotted noted, however, that only the "rules" abdomen,and large acentric"pupil." 2) In our removal of the relativelyrecent times P. machaonarrived on listed earlier prevent the plateau,perhaps carried from Alberta in genus from this category. There is con- hay or straw duringthe developmentof the siderable doubt as to the relationship be- National Park or of highwaysor railroads. 3) These two Papilio at firstlacking behav- 1 The continuedpresence of authors'names ioral and otherisolating mechanisms, hybrid- followingthe names of species of North ized ratherfreely; the distinctivegenotype of Americanbutterflies is, in mostcases, a waste P. kahli allows the F1 heterozygotesto show of type. Startingwith this paper I will only some yellow-wingcharacters never seen in cite author'snames wherethere is some con- laboratorycrosses of machaon with true ceivable chance of confusion.It is strongly polyxenes. 4) Since P. kahli and P. avinoffi urged that the trivial names in the forth- belong to separate species,one expects that comingdos Passos checklistbe adopted by isolatingmechanisms are evolvingin Riding all personsinterested in Nearcticbutterflies, Mountainpopulations and thateventually nat- and thatno changesbe made in themexcept ural hybridswill no longerbe produced.Mean- where retentionof a name might lead to while,each speciesmay be incorporatinginto confusion.Citation of authors' names as a its genotypenew adaptivealleles fromother matterof courseshould cease. species (introgression).There is littlebasis

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions 170 SYSTEMATIC ZOOLOGY forregarding kahli as dimorphic,in the sense subspecieswithin a superspecies(Brown, of P. glaucus females.For the present,these 1955), or to be a series of full-fledged Riding Mountainswallowtails may be called P. kahli (or P. polyxenes kahli), P. machaon species (most earlier authors). At the avinoffi,and theirhybrids." momentwe have no real evidenceon the genetic relationshipsof any of these There is no reasonto believethat similar forms,and thus no way (except guess- situationsdo not exist throughoutthe work) of comingto a decisionas long as complex. we insiston employinggenetic standards. Pieris is includedin this categorybe- What can we concludefrom this crude cause of the completeconfusion regard- surveyof the Nearcticbutterfly genera? ing' the status of the protodice-occiden- Above all else we can say that,at least talis-calyce-sisymbrii-beckeriiseries of at the present level of knowledge,the forms. Oeneis is placed here because, prevalenceof the clearlydefined species among other problems,the scope of the is a myth. Even biasing the surveyin "species"chryxus is a matterof differing favorof the distinctnessof species,about taste,and this series of formscannot be one-halfof the generaresist separated with assurance from British partitioning. Columbianforms usually lumped in with This resultis reallynot surprising,as taygete. The White Mountainbutterfly, the very natureof the biologicalspecies Oeneis semidea and its Rocky Mountain definitionmakes its use impossible in relative 0. lucilla are, at present,con- practice. The only test of conspecificity sideredpart of the "species" melissa,al- (or lack of same) is what happenswhen thoughthere is no genetic evidence to two formsoccur togetherin nature,and, support this placement. Similarlyonly even when sympatricorganisms can be a look at the latest butterflybook will studied,partial interbreedingmay make tell one whetheror not the Mt. Katahdin a clear decisionimpossible. Genetic tests ,0. katahdinand its Cordilleran in the laboratory,while yielding valuable relative 0. brucei are in or out of the information,are neverdefinitive. Organ- species polixenes. In Speyeria an array ismswhich are sympatricbut remaindis- of 109 names was reducedby dos Passos tincti-n nature may freelyinterbreed in and Grey(1947) to a groupof 13 "species." the laboratory. Conversely,organisms However,no evidencehas ever been pre- placed togetherin the laboratorymay be sented which would indicate that there unable to interbreedsuccessfully, al- ofthe are 13 in thoughpopulations same organisms actually biological species other Speyeria-although there seems to be approachingeach graduallyin na- generalagreement that most or all of the turemay be quite capable of blendingto- 13 segregatesrepresent related series of gether.Mayr (1957b: 378) pointsout that clusters. this indeed mightbe the case with the The three genera in the "conceptin- laboratory-isolatednorthern and southern applicable"category (Group 4), all have Rana pipiens. Of course,one coulddefine the same thingin common.That is that the biologicalspecies in such a way that no two people seem to be able to agree laboratorytests would be conclusive,but on what are or are not "species" among evenif this were done the amount of work the observed segregates within these involved in clearly delimitinga single genera. Consider,for example, the genus species would be staggering. It seems Coenonympha. One very distinctform, clear that the biologicalspecies is not a haydenii,is universallyconsidered to have practicaltaxonomic category. specificstatus. However,the remaining It mightbe noted parentheticallythat array of Nearcticforms have been con- casual discussionswith other zoological sidered to be conspecific(e.g., Daven- systematists(an ichthyologist,a herpe- port,1941), to be a series of species and tologist,and a hemipteristamong others)

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions SYMPOSIUM ON PHILOSOPHICAL SYSTEMATICS 171 lead me to believe that the abundance of Mendelian populations have properties clearly delimited segregates in the higher not possessed by their componentindi- animals has been vastly overestimated. viduals, such groupingsof similar indi- The situation in mammals may well be viduals fromdifferent populations could inferred from the following quote from be considered"unnatural." This does not Hall and Kelson (1959, volume 1, page v): mean,however, that such groupingsare "The taxonomiclimits assigned to anyorder devoid of biologicalinterest. In orderto or genus or intermediatetaxonomic category testthe hypothesisthat numerical analy- reflectout own personaljudgment. For spe- sis will not clusterindividuals by popu- cificversus subspecificstatus of any kind of lation,a pilot studywas undertakenin mammal,we have acceptedthe last-published which the similaritiesof an array of in- opinionthat was supportedby evidenceob- tainedfrom examination of specimens."(my dividualbutterflies were evaluated. Thir- italics) teen male specimensof the genusEuphy- dryas were selected for study-10 from There seems to be no reason why tax- four populationsidentified as belonging onomy at the "species" level should not to the species E. editha,and threefrom be approached in the same way as is the the closelyrelated species E. chalcedona. of higher categories. Quanti- In at least some areas (e.g., the vicinity tative studies of phenetic relationships of San Francisco) populationsof editha will reveal clusters of populations, and and chalcedonacoexist without signs of where it is convenient for purposes of hybridization.The specimens,all selected communicationwe can give formalnames at randomfrom longer series, came from to these clusters. The genetic relation- the followinglocalities: ships of the clusters (i.e., the degree of interbreedingamong them), when known, 3 E. editha (SB600, SB601, SB602): will be important attributes of the clus- San Bruno Hills, San Mateo ters, but will not be involved in category County,California, elevation 600 definition. Using such a system it seems feet,April 3, 1960. obvious that any organisms sufficiently 3 E. chalcedona(CHAL565, CHAL566, distinct to be sympatric without inter- CHAL567): Pinnacles Nationial breeding will fall in distinct clusters, al- Monument,San Benito County, though as in classical taxonomy, poly- California, elevation 1300 feet, morphismwill cause a certain amount of April 26, 1959. difficulty. 2 E. editha (KING8461, KING579): It should be emphasized that the re- KingsCanyon National Park, Cali- moval of degree of isolation as a criterion fornia, elevation 9-10,000 feet, for "membership" in a systematic cate- July 20-21, 1958. gory in no way lessens the evolutionary 3 E. editha (WOOD611, WOOD612, importance of such isolation. It is merely WOOD613): Woodside Experi- recognition of the difficultiesof defining mental Population, San Mateo the necessary level of isolation, and then County,California, elevation 540 of determiningwhether or not that iso- feet,March 22, 1961. lation exists. 2 E. editha (MONT6008,MONT6010): Route 198, 20 road miles east of There is point in the hier- probably a 101,Monterey County, Cali- archy below which grouping organisms Route of their similarity fornia,elevation 2500 feet,April on the basis overall 20, 1960. (rather than imagined phyletic rela- tionships) will result in similar individ- Because of the ready availabilityof a uals fromdifferent Mendelian populations Burroughs220 Algolprogram for comput- being grouped together,rather than indi- ing product-momentcorrelation coeffi- viduals from the same population. Since cientsI decidedto do classicalcorrelation

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions 172 SYSTEMATIC ZOOLOGY and weighted variable group (WVG) (1958) and will not be discussedfurther analyses of the pheneticrelationships of here. Table 1 liststhe 74 charactersused the13 individuals.This procedureis thor- in thestudy, and thenumber of states into oughly outlinedin Sokal and Michener whichthey were divided. Because of the

TABLE 1-CHARACTERS EMPLOYED, WITH NUMBER OF STATES INTO WHICH EACH WAS DIVIDED SHOWN IN RIGHT HAND COLUMN. V = VENTRAL, D - DORSAL, FW = FOREWING, HW = HINDWING.

1. Width of large cream spot in FW discal cell ...... 7 2. Shape of large cream spot in FW discal cell (ratio) . . . 6 3. Width of orange spot in FW discal cell ...... 9 4. Position of proximal cream spot in FW cell M3...... 6 5. Color of proximal spot in FW cell Cu1...... 4 6. Width of cream spot in FW cell Cul...... 6 7. Width of distal red spot in FW cell M.3...... 4 8. Width of proximal cream spot in HW cell M2...... 4 9. Shape of proximal cream spot in HW cell M2 (ratio) ...... 5 10. Amount of orange in proximal cream spot in HW cell M2 . . . 4 11. Width of cream half-moon spot in HW discal cell...... 4 12. Percentage of light scales in proximal orange spot in HW cell M2 . . . 4 13. Separation of proximal orange and white spots in HW cell Cu1 . . . 4 14. Width of proximal orange spot in HW cell M ...... 4 15. Width of distal cream spot in HW cell M3...... 4...... 4 16. Length of forewing...... 5 17. Shape of forewing (ratio) ...... 6 18. Comparison of VFW and VHW patterns...... 3 19. Palpus color ...... 3 20. Color of abdominal dorsum...... 3 21. Color of abdominal venter...... : 3 22. Degree of fusion of half-moonspot in HW discal cell with basal spot. . . 3 23. Shape of distal cream spot in HW cell M3 . . . 3 24. Color of DHW anal margin...... 3 25. Condition of orange marginal spots of DHW ...... 3 26. Condition of dark line in white median band of DHW ...... 3 27. Color pattern of antennae...... 2 28. Shape of abdominal sternum 6...... 2 29. Shape of abdominal tergum 2...... 2 30. Shape of tegula...... 3 31. Form of joining of mesopreepisternumand mesokatepisternum . . . 3 32. Form of dorsal part of pleural suture ...... i.... 2 33. of anterior part of spinasternum...... 2 Shape . 34. Length of middle segment of palpus ...... 6 35. Width of face at tentorial pits ...... 6 36. Width of frontoclypealsclerite between antennae...... 4 37. Ratio of spread of pilifers to face width...... 5 38. Shape of vertex of head ...... 2 39. Height of metaepimeron...... 4 of uncus lobes...... 2 40. Length 3 41. Condition of tegumen just anterior to uncus lobes...... 3...... 42. Form of saccus lobes...... 3 43. Position of tips of valvae ...... 3 44. Shape of juxta ...... 2 45. Shape of process of juxta ...... 2 46. Depth of cleft between teeth of ventral arm of clasp . . . 2 47. Height of inner face of lip of valva ...... 2 48. Form of dorsal arm of clasp ...... 3 49. Length of aedeagus ...... 7 7 50. Height of valva ...... 51. Shape of aedeagus (ratio) ...... 6 52.-64. Characters of forewing venation

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions SYMPOSIUM ON PHILOSOPHICAL SYSTEMATICS 173

TABLE 1 -Continued 52. Distance between tips of R4 and R2...... 5 53. Distance between tips of R2 and M.. 6 54. Distance between tips of M1 and M.. 6 55. Distance between tips of M3 and Cu.. 5 56. Distance between tips of Cu1 and Cu.. 5 57. Distance between tips of Cu2 and 2V ...... , 4 58. Distance from R,-R2crotch to tip of Cu.. 4 59. Length of mdc.. 5 60. Length of ldc.. 6 61. Length of M3-Cu1stalk.. 8 62. Length of M-ldc stalk.. 6 63. Length of free M3...... 5 64. Length of free R4..5 65.-74. Characters of hindwing venation 65. Distance between tips of M1 and M.. 5 66. Distance between tips of M2 and M.. 4 67. Distance between tips of M3 and Cu1 ...... 5 68. Distance between tips of R. and Cu.. 5 69. Distance between M1-RRcrotch and Cu,-CU.2crotch .. 4 70. Length of udc.. 5 71. Length of free M2.. 4 72. Length of free M.. 5 73. Length of M3-Cu1stalk. 4 74. Distance between Cu1-Cu2crotch and tip of 2V. 6 ease of computation2 it was possible to evidenceof repeatedinsertion of a gen- calculate the Q-matrices (specimen x eral size character(especially in view of specimen) for the followingcombinations the slightlylarger average size of the of characters: 1-74; 1-16; 1-27; 28-51; threechalcedona specimens). There was 52-74; 1-27 and 40-74; 1-27 and 40-51; no signof a significantweighting of size- 1-27 and 52-74; odd numbered characters; in the R-matrices,the originaldata, or and even numbered characters. the way the specimensclustered. The R-matrix (character x character) The Q-matrixbased on all 74 characters was also obtained for all characters based is shownin Table 2 (coefficientsgreater on the coded data. However, because the than.25 are significant[P <.011). coding forces us to evaluate each of the A numberof interestingpoints emerge 2701 coefficientsseparately, and because in examiningthe Q-matrices.In eightof of the low number of observations per the nine Q-matrices(including 1-74) the character (13), a second R-matrix was three chalcedona specimenshad higher computed for the first16 characters using correlationsamong themselvesthan any uncoded data and 9 additional specimens one of them had with a non-chalcedona from the same populations. Because of specimen.Interestingly, the onlymatl'ix the large number of measurements used wherethis was not true (28-51) was the as characters, these matrices as well as one containingthe structuralcharacters the raw data were carefullyexamined for (genitalic) normallyconsidered "impor- tant" or "basic" by the old line taxono- 2 The actual preparingand measuringof the-specimens took severaldays. The compu- mists. However,as was expected,with tation of 3601 correlationcoefficients (plus the exceptionof the two Kings Canyon means and standarddeviations of the vari- specimens,there was littletendency for ables) took 20 minutes. Includingpunching the various editha to assort according of the data ontocards the totalcomputational of origin. The results costwas about$25.00. The samecomputations to theirpopulation on a desk calculatorwould take a competent of the WVG search for structurein the operatorabout two yearsof full-timework. matrixare shownin Figure 1. The Kings

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions 174 SYSTEMATIC ZOOLOGY

1 2 10 13 3 9 11 12 7 a 4 5 6 the numericalmethod, even using rela- tivelyunrefined techniques, seems to be quite able to associate what are con- 1000 E E E E E @ g ? y y g g S sidered on other groundsto be related 1000. entities.Indeed, it can do thiswith charac- ters (e.g.,52-74) which,to myknowledge, have neverbeen utilizedin classicaltaxo-

800 nomicstudies. Furthermore,the informa- tioncontent of the matrix,as well as that of the derivedstructure, is vastlygreater than that usually suppliedin a standard taxonomicwork. On the basis of this in-

6000 formationone mightwish to hypothesize that the Kings Canyon populationsplit offfrom the othereditha after an editha- 500l chalcedonasplit, or thatchalcedona repre- sents a rapidlyevolving line originating 400 froma Kings Canyontype stock. A ffrm base is providedfor speculation, but the speculationis not involvedin establish- 300 ing the base. The failureof the techniqueto assort 200 individualsinto groupscorresponding to theirpopulation of originis not disturb- FIG. 1. Diagram of relationships obtained ex- by the weighted variable group method ing. There is nothingto be gainedin (Ln=.045). Ordinate is the correlation coef- tending the taxonomicsystem to this ficientmultiplied by one thousand. level, since the process of gatheringthe samplesautomatically assorts individuals Canyon specimenshad been determined into geographic clusters. However, it by the collectoras editha (on the basis mightbe very interestingevolutionarily of facies and presumedfood plant), but to studythe distributionof frequencies of in themost recent key (Bauer, 1961) they generallysimilar individuals in different run to chalcedonaon the basis of genital populationsas a contrastto the more structure.The diagram of relationships usual procedureof studyingthe distribu- clearlyshows theirposition as a distinct tions of single charactersor genes. also be groupof the editha cluster.It can The changes attendantto the general of the originalma- seen by examination adoptionof a "species attitude"such as the editha specimens, trix that, among is outlinedabove would be fewand salu- KING579 has the highestaverage corre- be chalcedona, tary. Gone fromthe literaturewould lation (.374) with the three are has the thirdhighest the emptydiscussions of what are or and thatKING8461 could (.249). This informationis lost in the not "good" species. Systematists of determiningstructure. The continue to give binomial designations process a genusEuphydryas has longresisted satis- to segregateswhen theybelieved that factorytreatment by classicalmethods-a formalname would be useful. The status thoroughnumerical study would doubt- ofthe segregateswould be determinedby less go a long way towardsthe descrip- theirrelationships with other segregates, tion of the relationshipsof the various the relative positionsbeing determined clustersof populations. by numericalmethods similar to those A numberof tentativeconclusions may now beingused extensivelyfor studies of be drawn fromthis pilot study. First, groupingsat higherlevels.

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions SYMPOSIUM ON PHILOSOPHICAL SYSTEMATICS 175

TABLE 2-Q CORRELATION MATRIX-SPECIMEN X SPECIMEN 2 3 4 5 6 7 8 9 10 11 12 13 SB600 1 .530 .540 .000 -.123 -.090 .311 .101 .499 .472 .423 .423 .376 SB601 2 xxxx .571 -.050 .077 .031 .496 .424 .594 .649 .305 .536 .572 SB602 3 xxxx .392 .190 .416 .443 .365 .640 .549 .479 .606 .498 CHAL565 4 xxxx .603 .642 .229 .378 .220 .103 .248 .160 -.072 CHAL566 5 xxxx .594 .298 .275 .069 .119 -.022 .067 -.115 CHAL567 6 xxxx .219 .468 .369 .172 .400 .245 .171 KING8461 7 xxxx .543 .313 -.425 .103 .520 .330 KING579 8 xxxx .470 .471 .471 .396 .470 WOOD611 9 xxxx .537 .655 .564 .499 WOOD612 10 xxxx .431 .580 .588 WOOD613 11 xxxx .363 .472 MONT6008 12 xxxx .501 MONT6010 13 xxxx

In summary,I thinkthat the biological of their suggestionshave been incorpo- species concept has outlived its useful- rated, but the inclusionof their names ness. The currentrevolution in data proc- here does not necessarilyindicate agree- essingpermits the relaxationof the rigid mentwith the ideas presented.I am also hierarchicsystem long employedto de- indebtedto Mr. RichardC. Rinkelof the scribethe productsof evolution.We may Universityof Kansas, forusing an IBM now modifyour systemto permitmore 650 "Taxon I" programto checkthe calcu- accurateand thusmore useful description lationsinvolved in thesearch for structure of the intricaterelationships of livingor- in the matrix. ganisms. As a step in this directionI This work has been supportedin part suggest that the genetic definitionof by grantsG-9821 and G-14740from the species, never employedin practice,be National Science Foundation. discarded as an ideal. Relationshipsat the lower levels of the taxonomichier- REFERENCES archy should be expressednumerically, in essentiallythe same way as relation- BAUER, D. L. 1961. Tribe . In Ehrlich and Ehrlich, How to know the but- ships of higher categoriesare now ex- terflies. Wm. C. Brown, Dubuque. pressed. It is hoped that systematists,as BROWN, F. M. 1955. Studies of nearctic Coeno- machinemethods relieve them of much nympha tullia (Rhopalocera, Satyridae). of their drudgery,will turn more and Bull. Amer. Mus. Natur. Hist., 105:359-410. more to studyingorganisms in nature. BURMA, B. H. 1954. Reality, existence, and classification: a discussion of the species An afternoonof rigorousfield observa- problem. Madrofno,7:193-209. tion will usually producemore informa- DAVENPORT, D. 1941. The butterflies of the tion of evolutionaryvalue than weeks satyrid genus Coenonympha. Bull. Mus. spentstudying preserved material. Comp. Zool., Harvard, 87:215-349. DOS PASSOS, C. F., and L. P. GREY. 1947. Sys- I would like to thankthe followingin- tematic catalogue of Speyeria (, dividualswho kindlytook theirtime to ) with designations of types read and commenton this manuscript: and fixationsof type localities. Amer. Mus. H. V. Daly, E. G. Linsley, and R. L. Novitates, No. 1370, 30 pp. Usinger,University of California,Berke- GRANT, V. 1957. The plant species in theory and practice. In E. MAYR (ed.), The species ley; R. W. Holm, G. S. Myers,and J. H. problem. Amer. Assoc. Advanc. Sci., Publ. Thomas,Stanford University; A. E. Levi- No. 50:39. ton and C. D. MacNeill,California Acad- HALL, E. R., and K. R. KELSON. 1959. The emy of Sciences; C. L. Remington,Yale mammals of North America. Ronald Press, NationalIn- New York. University;P. H. A. Sneath, MAYR, E. 1957a. Species concepts and defi- stitutefor Medical Research, London; and nitions. In The species problem, Amer. :R. R. Sokal, Universityof Kansas. Many Assoc. Advanc. Sci., Publ. No. 50:1.

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions 176 SYSTEMATIC ZOOLOGY

1957b. Difficultiesand importance of the tions of Lepidoptera. Proc. Tenth Inter- biological species concept. In- The species nation. Congr. Entomology, 2:787-805. problem. Amer. Assoc. Advanc. Sci., Publ. SOKAL,R. R., and C. D. MICHENER. 1958. A No. 50; 371. statistical method for evaluating systematic MICHENER,C. D., and R. R. SOKAL. 1957. A relationships. Univ. Kansas Sci. Bull., 38: quantitative approach to a problem in classi- 1409-1438. fication. Evolution, 11:130-162. MORRIS,J. G. 1862. Synopsis of the'described PAUL R. EHRLICH is Assistant Professor Lepidoptera of North America. Smithsonian of Biological Sciences and Curator of Ento- Miscell. Collect. mological Collections at Stanford University, REMINGTON, C. L. 1958. Genetics of popula- Stanford, California.

PheneticClassification and Typology HOWELL V. DALY

FUTURE HISTORIANS will doubt- Simpson (1961:48) has recently ap- less note two technologicalbreak- praised the phenetic methods of Sokal throughswhich have alteredthe course and Michener (1958) and Cain and Har- of systematicbiology: the microscope, rison (1958) as quantifications of typo- supplementingthe human eye in an logical procedures. The identificationof age of data-collection,and the electronic numerical taxonomywith typologywould computer,supplementing the brain in an be of little consequence were it not that age of data-processing.Taxonomists have classical typology has gained the conno- now agreedthat the firststep towardbio- tations of metaphysical,dogmatic, or anti- logical classificationis the gatheringof evolutionary thinking. To understand all possible informationon the various Simpson's meaning, I would like to quote organisms. This objective is well sum- the following (1961:46-47): "The basic marized by Hennig's concept of the concept of typologyis this: every natural "holomorph"(see Kiriakoff,1959). group of organisms, hence every natural The differencein opinionsas to how taxon in classification,has an invariant, such informationis to be utilized stems generalized or idealized pattern shared from the time of Linnaeus. Two ap- by all members of the group." In the proaches to classificationare currently terms of Aristotelian logic at least one recognized (Cain and Harrison,1960): character or fundamentummust be com- the pheneticand the phyletic. Classifi- mon to every memberof the group. Beck- cation is the orderingof objects into ner (1959) has referredto this concept as criteria. "monotypic." His use of this word shduld groups on the basis of certain not be confusedwith its customarymean- In a pheneticclassification of organisms, ing among taxonomists, that is, a taxon the groupsare gauged on an estimateof Tj,1 with but one Tj. The conceptual com- the degree of overall similarity. The panion to a monotypic group has been groupsof a phyleticclassification of or- posed by Beckner as a "polytypic" group ganisms are gauged on the relativere- or aggregation. Simpson (1961:94) has cencyof commonancestry. Both are evo- paraphrased the terms of such an aggre- lutionary;the phyleticis concernedwith gation as follows: of divergenceand the phenetic the time "1. Each individualhas a large but unspeci- systemis concernedwith the amountof fiednumber of a set of propertiesoccur- divergence. ringin the aggregateas a whole.

This content downloaded from 132.248.28.28 on Mon, 04 Jan 2016 19:22:34 UTC All use subject to JSTOR Terms and Conditions