AN ESSAY ON THE APPLICATION OF THE THEORY OF NATURAL SELECTION IN THE CLASSIFICATION OF ANIMALS AND PLANTS, ILLUSTRATED BY A STUDY OF THE EVOLUTION OF THE WINGS OF INSECTS, AND BY A CONTRIBUTION TO THE CLASSIFICATION OF THE LEPIDOPTERA.

It is now thirty-four years since the publication of Darwin's Origin of Species; and the great war of opinions which had been imminent for some time and which broke forth on the appearance of that work, has been fought to a conclusion. There remains no contest except that of a healthy competition in reaping the fruits of the victory. Naturalists differ in their opinions as to details but the great principle of evolution has been firmly established, and our methods of thought have been revolutionized in consequence.

Notwithstanding this I do not believe that the systematists of today are making as much use of the theory of descent in taxonomic work as they might. We are still busy describing species as if they were immutable entities; and in our descriptions we give little thought to the causes that have determined the forms of organisms. It is true that considerable has been done in the direction of working out the phylogeny of the larger groups, as branches and classes, and to a less extent of orders. But rarely is any effort made to determine the phylogeny of the smaller groups.

Here I believe lies the work of the systematist of the future. The description of a species, genus, family or order, will be considered incomplete until its phylogeny has been determined so far as is possible with the data at hand. We are to care less for the mere discovery of new forms, and more for an understanding of the processes by which new forms have arisen. The object of taxonomy will not be a mere grouping of forms according to similarity of structure. But the systematist will have constantly before him the question: What do these variations of form mean?  With this change in the object of taxonomic work, there will come a change in its methods. It is strange that the change has been so long delayed; for we are really using the same methods that were employed before the establishment of the truth of the theory of natural selection.  What these methods are was indicated by Darwin in the following words:

"Practically, when naturalists are at work, they do not trouble themselves about the physiological value of the characters which they use in defining a group or in allocating any particular species. If they find a character nearly uniform, and common to a great number of forms, and not common to others, they use it as one of high value; if common to some lesser number, they use it as of subordinate value." (Darwin, Origin of Species, pp. 367-368, Am. Edition.)

This statement is about as true today as when Darwin wrote it.  For if one will look through the taxonomic works on zoology or botany he will very seldom find any reference to the functions of organs.  But almost all naturalists now believe that in each epoch of time the forms of existing organisms have been determined by a survival in preceding generations of those individuals whose parts were best fitted to perform their functions.

Does it not follow from this belief that we can confidently expect to gain much help in our efforts to work out the phylogeny of organisms by making a careful study of the functions of their organs, and endeavoring to understand the reasons for the action of natural selection?

I suggest, therefore, that the logical way to go to work to determine the affinities of the members of a group of organisms is first to endeavor to ascertain the structure of the primitive members of this group; and then endeavor to learn in what ways these primitive forms have been modified by natural selection, keeping in mind that in each generation those forms have survived whose parts were best fitted to perform their functions.

Obviously there are certain difficulties in the carrying out of this plan. But the measure of our success in determining the affinities of the organisms studied, will depend largely on our ability to overcome these difficulties.

Among the difficulties encountered is the fact that usually our classification must be based largely on a study of living forms; for in most eases the aid to be derived from Paleontology is comparatively slight.  But although the record presented by fossils is very fragmentary, fortunately there are many living forms which are comparatively slightly specialized.  And these will serve to give an idea of the stem form of the group.

Thus to carry out the plan suggested, the zoologist or botanist, if he is forced to work only with living animals or plants, will select from the group to be studied the most generalized type before him, and then trace out the different ways in which this type has been modified in the more specialized forms.

If the group studied be a large one, the probabilities are that instead of a single primitive type, several generalized forms will be found, each representing more or less approximately the stem form of a distinct line of development; and a comparative study of these different forms will be necessary in order to obtain an idea of the structure of their common ancestor.

But how shall one go to work to select from a large number of forms those that are to be considered the more generalized ? The higher animals and plants are such complex organisms that it is not an easy matter to determine the relative degree of specialization of two distinct forms. The problem is also complicated by the fact that even the more generalized forms may present specializations peculiar to themselves.

Numerous examples will occur to any systematist of forms which as a whole are comparatively generalized, but which in some respects are highly specialized, being, as has been expressed by some writers, "sidewise developed." It is essential that these sidewise developments be not included in our conception of the still more primitive form.

Thus the Thysanura are regarded as the most generalized of the living Hexapoda.  This would also be the case if of this order only the suborder Collembola were known to us.  In such a case we might conclude from a study of the spring-tails that the primitive Hexapoda possessed a ventral sucker and a caudal spring, and that these organs had been super-ceded by the wings in more specialized forms.  Now we know that while taken as a whole the Collembola are very generalized insects, that so far as the ventral sucker and caudal spring are concerned they do not represent the primitive type of the order, but are sidewise developed.  In both the Cinura and the Collembola we find forms which are clothed with highly specialized scales, scales which rival in complexity of structure those of the Lepidoptera.  Yet no one believes that the primitive Hexapoda were so clothed.  This is another sidewise development.  And the scales of the Lepidoptera, and of the Curculionidae, for example, have arisen independently.

We thus see that although in our efforts to trace out the series of modifications through which a line of organisms has passed we may find forms which appear primitive, we must not expect to find among living forms an exact record of these changes.  Each form studied will represent the tip of a twig which has separated from the main branch.  Fortunately for our purpose we can often find some forms representing twigs that branched off very early and that have not grown very far in their special direction.  In many cases too, forms are found which although highly specialized as regards some of their organs will retain a generalized condition of other organs.  By a comparison of a number of such forms each representing a generalized condition of some of its organs we can get an approximate idea of the common progenitor.

But I repeat, how shall we determine which are the representatives of those short twigs that have undergone but little change, and which are the representatives of branches that have been greatly modified?  The answer to this question is a statement of the method I propose for applying the theory of natural selection to taxonomy more fully than has been done before.

As the structure of a highly organized animal or plant is too complicated to be understood in detail at once, it is suggested that the student begin with the study of a single organ possessed by the members of the group to be classified, and that his studies take the following course: First the variations in form of this organ should be observed, including paleontological evidence if possible; then its function or functions should be determined.  With this knowledge endeavor to determine what was the primitive form of the organ and the various ways in which this primitive form has been modified, keeping in mind the relation of the changes in form of the organ to its function.  In other words endeavor to read the action of natural selection upon the group of organisms as it is recorded in a single organ. The data thus obtained will aid in making a provisional classification of the group.

When this stage has been reached another organ should be selected and its history worked out in a similar way.

The results of the two investigations should then be compared; and where they differ there is indicated the need of renewed study. For if rightly understood the different records of the action of natural selection will not contradict each other. The investigation should be continued by the study of other organs and a correlating of the results obtained until a consistent history of the group has been worked out.

This method differs from that commonly employed in being a constant effort to determine the action of natural selection in the modification of the form of organisms in order to better adapt their parts to perform their function.  Ordinarily little or no attention is devoted to the study of the functions of organs in purely taxonomic works.

If the history of a group be worked out in the manner indicated, the student will feel the need of recording his results in such a way as to indicate the phylogeny of the divisions of the group.  But as the necessities of book making require a linear arrangement of descriptions this is somewhat difficult; for the natural sequence of groups should be represented by constantly branching lines rather than by a single straight line.

It seems to me that the most practicable way of meeting this difficulty is to begin with the description of the most generalized form known, and to follow this with descriptions of forms representing a single line of development, passing successively to more and more specialized forms included in this line. When the treatment of one line of development has been completed take up another line beginning with the most generalized member of that line and clearly indicating in the text that a new start has been made.

Much aid can also be given by a tabular statement of the essential characters of the subordinate groups, using the form of the ordinary analytical table.  An illustration of this is given in Part III of this essay.

In this connection reference should be made to the proper position of degraded forms in a series where an effort is made to represent the natural sequence.  The common practice of assigning such forms the same position that would be assigned to them if their simplicity of structure was the result of a primitive condition seems to me illogical.  An example will make this point clear.  The Hemiptera are doubtless all descended from a common winged ancestor.  The lice, although more simple in structure than most other members of the order, do not represent the form of this ancestor as closely as do the winged members of this order.  They should not, therefore, be placed first in the hemipterous series as is commonly done.  It would represent the facts of nature better to place them last, as forms departing more widely from the primitive type of the order than do the winged forms.  But it should be clearly indicated that although they represent the tip of one of the lines of development that line is a downward bending line.

In attempting to work out the phylogeny of a group of organisms, there will arise, I believe, the necessity of distinguishing between two kinds of characters: first, characters indicating differences in kind of specialization; and second, characters indicating differences in degree of specialization of the same kind.  The former will indicate dichotomous divisions of lines of descent; the latter will merely indicate degrees of divergence from a primitive type.  Thus, to draw an illustration from the following pages, it is shown that there are two distinct ways of uniting the two wings of each side in the Lepidoptera; they may be united by a frenulum (Fig. 22) or the may be united by a jugum (Fig. 27).  These are differences in kind of specialization, and indicate two distinct lines of descent or a dichotomous division of the order. Among those Lepidoptera in which the wings are united by a frenulum great differences occur in the degree to which this organ or a substitute for it is developed; such differences may merely indicate the degree of divergence from a primitive type and may need to be correlated with other characters to indicate dichotomous divisions.

It is impracticable to indicate degrees of divergence from the primitive type based on the nature of the frenulum at this stage of the discussion; but another character will serve our purpose well.  In the more generalized Lepidoptera the anal areas of one or of both pairs of wings are furnished with three anal veins; while in more specialized forms the number may be reduced to two or even to one.  But the distinctions indicated by the presence of three, two, and one anal veins in different moths, are merely differences in degree of specialization by reduction of an anal area, and taken alone will not indicate dichotomous divisions.  Thus if we group together all the moths that have retained three anal veins in the hind wings, such a group will contain, not merely the Microlepidoptera, as is commonly stated, but also the more generalized members of several distinct divisions of the Macrolepidoptera.

The fact is, the primitive Lepidoptera evidently possessed at least three anal veins in the hind wings (we will omit the fore wings from the discussion for the time being). In several distinct lines of development within this order the direction of specialization of the anal area of the hind wings has been towards the reduction of the number of veins in this area; but the extent to which this reduction has gone merely indicates the degree of divergence from the primitive type. And so far as this single character is concerned a similar degree of divergence in a similar direction may be possessed by members of widely separated divisions of the order.

But we are not entirely dependent on differences in kind of specialization for indications of dichotomous divisions. Such divisions may be indicated by differences in the order in which specializations take place.

This also can be illustrated by a study of the anal areas of the wings. It is evident that in the primitive Lepidoptera the fore wings as well as the hind wings possessed three anal veins.  And in certain divisions of the order the direction of specialization of the anal area of the fore wings has also been towards a reduction in the number of veins. It will be shown in the concluding part of this essay that in certain divisions of the order the reduction of the anal area of the hind wings has preceded the reduction of the anal area of the fore wings; while in other divisions of the order the reverse is the case.  Here is an indication of a dichotomous division.  Take for example two families of moths, one of which is characterized by the presence of two anal veins in the fore wings and three anal veins in the hind wings; and the other, by three anal veins in the fore wings, and two in the hind wings.  In the former, the specialization by reduction of the anal areas has begun in the fore wings; in the latter, this specialization has begun in the hind wings.  And it is evident that the common progenitor of the two families had three anal veins in both fore and hind wings, and that the difference in the order in which the reduction of the anal areas has begun indicates a dichotomous division.

There will also arise, I believe, in a work of this kind a necessity for distinguishing between the essential characters of a group and those characters which are used by the systematist merely to enable students to recognize members of the group. For it seems to me that the essential characters of a group of organisms do not lie necessarily in the presence or absence of any structure or structures, or in the form of any part or parts of the body of the living members of the group; but rather in the characteristic structure of the progenitor of the group, and in the direction of specialization of the descendants of this progenitor.

Thus, to use again the illustration given above, the Jugatae are essentially characterized as the descendants of those ancient Lepidoptera in which the wings of each side were united by a jugum; and they are also characterized by a tendency towards an equal reduction of the veins of the two pairs of wings.  While the Frenatae are essentially characterized as the descendants of those ancient Lepidoptera in which the wings of each side were united by a frenulum; and they are also characterized by a tendency towards a greater reduction of the veins of the hind wings than of the fore wings, or, in other words, by a tendency towards a cephalization of the powers of flight.  The fact that in many of the Frenatae the frenulum has been lost, does not invalidate in the least the truth of this characterization.  The loss of the frenulum, however, in certain Frenatae renders necessary the use of some other character or characters by the systematists as recognition characters.

The recognition characters are those usually first observed by the investigator, and are those commonly given in taxonomic works.  In many cases these recognition characters are also essential characters, especially in the case of groups that have been thoroughly studied.  But by the taxonomic methods now commonly used search is chiefly made for recognition characters.  The more skilled the systematist the more likely is he to discover and use as recognition characters those that are really essential, although the distinction pointed out here may not be recognized by him.

In the case of those groups where but few or no generalized forms have persisted till this time, the essential characters must to a greater or less extent be inferred.  This is especially true of those characters which refer to the structure of the progenitor of the group.  But the direction of specialization may be shown by a single representative of the group, if it be highly specialized, and we have a clear idea of the essential characters of a larger group including the one under investigation.

It must be borne in mind, however, that the direction of specialization may undergo marked changes in the course of the history of a single line of development. Thus I feel sure that in the ancient Frenatae the tendency of specialization was towards more rapid flight which tendency resulted in the preservation of the narrower winged forms. But while this tendency has been continued in certain divisions of the group to the present time, so that in these divisions the most highly specialized forms have the narrowest wings (Sphingidae, Zygaenina), there are other divisions in which the tendency has been changed towards a different mode of flight, and has resulted in the preservation of the wider winged forms, and in these divisions the most highly specialized forms are those having the widest wings (Saturniina.)

In recording the results of specialization one is apt to speak as if there were an intelligent directing force which determines the direction of specialization; or as if individuals deliberately chose the way in which they should vary from their progenitors. The fact that we are often able to arrange the members of a group in well defined series, each series culminating in a specialized form towards which the other forms approximate in varying degrees of closeness, leads to the unconscious use of such expressions. It is difficult to keep constantly in mind the extent of the thinning out process that takes place in nature, that the objects of our studies are merely a few fortunate individuals that have withstood tests that have proved fatal to the great majority. Innumerable unfortunate variations perish and leave no record; we see the fortunate ones alone; and the impression is apt to be that there is a definite progression on the part of all. Perhaps the facts of the case can be expressed as follows: The conditions which surround an organism combined with the existing structure of that organism render variations in its offspring in certain definite directions fortunate, while variations in other directions are unfortunate. As the fortunate variations alone are preserved to us the record seems to indicate a strong tendency to vary in definite directions.

In this paper the terms generalized and specialized are used in preference to low and high, which are often loosely used as synonyms of these terms. It should be remembered that lowness or simplicity of structure may be the result of degradation, and hence does not necessarily indicate a primitive or generalized condition. The lice are the lowest of the Hemiptera; but they are by no means the most generalized of the living members of that order.

Professor Hyatt has pointed out* that specialization may take place in two different ways: first, by an addition or complication of parts, specialization by addition; second, by a reduction in the number or in the complexity of parts, specialization by reduction. These expressions are very convenient in indicating the direction of specialization of an organ or set of organs.

Another important principle, first pointed out, I believe by Meyrick, is that "When an organ has wholly disappeared in a genus, other genera which originate as offshoots from this genus cannot regain the organ, although they might develop a substitute for it."

The truth of Meyrick's law, as this last principle may be termed is obvious when we consider that if a part be wholly lost there is nothing for natural selection to act upon in order to reproduce it. And even if a necessity for the organ should again arise and a substitute be developed for it, it is not at all probable that the substitute would resemble the organ so closely as to be mistaken for it.

In the application of Meyrick's law care must be taken that comparison be made only between allied forms, i.e., within what may be termed a single line of descent. I recognize the fact that these expressions are indefinite, but I believe no systematist will have doubt as to my meaning.

Let me state the matter in another way. The loss of an organ is a character that merely indicates a degree of divergence from a primitive type. And so far as any single organ is concerned this stage may be reached in one line of descent very much earlier than in another. In fact the loss of an organ may be correlated in one line of descent with a very generalized condition of other characters; while in another line of descent very highly specialized forms may still possess the organ in question.

A good illustration of this is presented by the condition of the mouth in the Macrolepidoptera. In many moths the mouth parts are wanting, while in other moths and in butterflies the maxillae are very highly specialized. It cannot be concluded from this fact that the mouthless forms are farther removed from the primitive type than are the sphinges and butterflies for example. A study of other structures would not support such a conclusion. We have to do in a case of this kind either with very distinct lines of descent or with a sidewise development.

In the case of the organ selected, the mouth, there comes into play, I believe, a very peculiar principle. For a long time I was greatly puzzled by the many instances in which absence of mouth parts is correlated with a very generalized condition of other structures. The explanation of this phenomenon I now believe to be as follows: Under certain conditions natural selection may tend to change the length of the adult stage. In some cases those individuals that most quickly provide for the perpetuation of the species are the ones that are most likely to have offspring. Under such conditions there would be a shortening of the duration of the adult stage until a point was reached at which it would not be necessary for the insect to take food during the adult stage, and the mouth parts would be lost in this stage.

But this shortening of the duration of the adult stage would also tend to a great degree to remove the species that had acquired it from the struggle for existence in this stage. A species that found it necessary to fly only a few hours or even days in order to provide for the perpetuation of its kind would not offer such an opportunity for the action of natural selection upon the structure of its wings and other organs peculiar to the adult, as would surely occur in a species having a longer period of flight.

* Insecta, page 51. Trans. Lond. Ent. Soc. 1884, page 277.

 Advance to Part II