F — Lae TP ee . 9 ra ~ ow? ? 4 : oe. 9 : - Ron te : Sipe PPO he OF, Z = “ y - “e “ * meat POP OT TOP DETE SAAS 2 > i aot > i Bee hae Segre te Seer Ye : s . 2 saa a eel dn eb nie wh si oe $etre A ~ a : - - 5 wy Pe arn Na PomSoeng ss Sereraee - td APS + E% a be eeu e oe : EP SS ee a ee TT ne he Ane aa naan at : ms ov ge > oe teun a ae erie RR ne Lg ot oO Deal ge Pgs A reseele Paber ae AA rem : adele “are we eS pm a tp te Ser a Suone nw = os Ay pret ae = a ae Pp ine PON = So : y cone ote ae rere PP a ae - ~ ae Sinton - Aw Ae A ate! Rte = eS Z.- tea ~ es a me ——— ~ - ~ aed ~ — >. ~ ad - ~ — > Z ~ — re > - * MUS. COMP. ZOOL LIBRARY OCT 21 1970 HARVARD POSTILLA PEABODY MUSEUM YALE UNIVERSITY NUMBER 146 18 AUGUST 1970 MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY (AMPHIPODA, GAMMARIDAEB), WITH EMPHASIS ON SUBTERRANEAN FORMS JOHN R. HOLSINGER DAVID C. CULVER POSTILLA Published by the Peabody Museum of Natural History, Yale University Postilla includes results of original research on systematic, evolution- ary, morphological, and ecological biology, including paleontology. Syntheses and other theoretical papers based on research are also welcomed. Postilla is intended primarily for papers by the staff of the Peabody Museum or on research using material in this Museum. Editors: Nancy A. Ahlstrom, Elizabeth G. Weinman, Elise K. Kenney. Postilla is published at frequent but irregular intervals. Manuscripts, orders for publications, and all correspondence concerning publications should be directed to: Publications Office Peabody Museum of Natural History New Haven, Conn., 06520, U.S.A. Lists of the publications of the Museum are available from the above office. These include Postilla, Bulletin, Discovery, special publications, and available back numbers of the discontinued journal, Bulletin of the Bingham Oceanographic Collection. All except Discovery are available in exchange for relevant publications of other scientific institutions anywhere in the world. MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY (AMPHIPODA, GAMMARIDAE), WITH EMPHASIS ON SUBTERRANEAN FORMS JOHN R. HOLSINGER Department of Biology, Old Dominion University, Norfolk, Virginia 23508 DAVID C. CULVER Department of Biology, University of Chicago, Chicago, Illinois 60637 (Received November 26, 1969) ABSTRACT Gammarus minus Say is a common amphipod species in springs and caves of limestone areas of the eastern and middle-eastern United States. Samples of populations from the central Appalachians were examined closely and morphological variation between spring and cave populations was analyzed. This species occurs in three mor- phological forms: a spring form, an intermediate cave form and an extreme cave form. The latter form was termed variety tenuipes by some earlier workers but has no nomenclatural validity. In contrast to the spring form, the cave forms show a reduction in eye structure, a change in pigmentation of the integument and a proportionate increase in the length of some of the appendages. It is concluded that G. minus is an extremely vagile and highly variable species that can occupy a variety of habitats, ranging from surface springs to small or large cave systems in certain karst areas. POSTILLA 146: 24 p. 18 AUGUST 1970. (2 POSTILLA INTRODUCTION Gammarus minus Say is a common inhabitant of springs, small spring-fed streams and cave streams in the eastern United States but is most often encountered in springs developed in limestone areas. This species was originally described by Say (1818), but subsequent references to it were few. Shoemaker (1940) carefully redescribed G. minus and gave a thorough literature review of the species up until that time. In the same treatment, Shoemaker (1940) synonymized G. propinquus Hay and G. purpurascens Hay with G. minus and designated a variety which he called tenuipes. Hubricht (1943) added numerous new locality records for this species and discussed the variety tenuipes. The species was again treated systematically by Bousfield (1958), who confirmed the earlier synonymies of Shoemaker and suggested that the variety tenuipes might be a distinct troglobitic (obligatory cavernicole) species. More recently, Minckley and Cole (1963) analyzed the morphological variation in several populations of G. minus from spring-fed streams in north-central Kentucky and compared the ecology and morphology of these populations with sympatric populations of G. bousfieldi Cole and Minckley. Finally, this species was compared biogeographically with some other members of the family Gammaridae in a paper on the biogeography of freshwater amphipods of the southern and central Appalachians by Holsinger (1969). Both Shoemaker (1940) and Hubricht (1943) treated the mor- phological variation in G. minus as it related to cave populations in terms of “varieties.” Presumably, Shoemaker and Hubricht used the term “variety” to signify a unified morphological group that apparently had become restricted to subterranean waters. Whether these workers were using the variety tenuipes to express genetic difference or to express the extreme variation of a plastic phenotype, or both, cannot be determined. Nevertheless, Shoemaker (1940) formally designated several cave populations of G. minus from Greenbrier County, West Virginia as a taxonomic variety under the name fenuipes. This name was later used by Hubricht (1943) and Bousfield (1958), but not necessarily in a nomenclatural sense. Although Shoemaker described tenuipes as a “variety,” he designated a type and created a trinomen. However, it was not clear whether he intended this taxon to be a subspecies or an infra- MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY 3 subspecific group. The International Code of Zoological Nomen- clature (Stoll et al., 1964, Article 45) allows a certain amount of freedom in dealing with “varieties” and “forms” proposed prior to 1961. Therefore, based on our interpretation of the Code, the variety fenuipes is considered to be of infrasubspecific rank and should not be accorded taxonomic validity. While Shoemaker (1940) recognized a single variety (tenuipes) of G. minus, Hubricht (1943) went further and divided the species into three varieties or forms, but he did not treat them in a nomen- clatural manner. The three varieties recognized by Hubricht were: a brown spring form with well-developed eyes and short antennae; a bluish intermediate cave form with slightly reduced eyes and long antennae; and a bluish fragile-bodied cave form with greatly re- duced eyes and long antennae (i.e., tenuipes). To facilitate discussion in this paper, three morphological groups, approximately equal to those described by Hubricht (1943), will be assigned Roman numerals and designated as follows: “tenuipes”’ Form I, intermediate cave Form II, and spring Form III. The objectives of this study were: 1) to quantify and describe the morphological variation in central Appalachian populations of G. minus with particular reference to subterranean populations; 2) to clarify the morphological status of the three forms of G. minus; and 3) to demonstrate that the subterranean “variety” fenuipes is an ecophenotype or extreme morphological variant of the more com- mon and widespread spring form. During the course of this study we critically examined 222 collec- tions of G. minus from 179 different localities in the Appalachian region. A breakdown of the locality data follows: three caves, four springs and one spring-run in three counties of northern Alabama; one spring in one county of northwestern Georgia; seven springs and two spring-runs in five counties of southern Indiana; three caves and five springs in five counties of central and eastern Kentucky; two caves, five springs and one spring-run in three counties of central Maryland; five caves, nine springs and three spring-runs in eight counties of central and southern Pennsylvania; three caves, six springs and seven spring-runs in nine counties of central and eastern Tennessee; nine caves, 30 springs and one spring-run in 16 counties of northern and western Virginia; 46 caves, 24 springs and two spring-runs in eight counties in eastern West Virginia. In addition to the above, samples were examined in less detail of G. minus from 4 POSTILLA 40 localities in southern Indiana, Kentucky, southern Ohio and central Tennessee. The present study, therefore, is based primarily on the first 222 samples mentioned above. DISTRIBUTION AND ECOLOGY As presently understood, the range of G. minus s. str. extends from eastern and central Pennsylvania southwestward to northwestern Georgia and northern Alabama and westward to central Tennessee, western Kentucky and south-central Indiana. This species is almost exclusively restricted to areas underlain by carbonate rocks. G. minus may also range from southwestern Illinois across southern Missouri and northern Arkansas to the extreme northeastern corner of Oklahoma (Hubricht and Mackin, 1940; Mackin, 1941; Hu- bricht, 1943; Hubricht, 1959; Minckley and Cole, 1963). How- ever, further critical examination of material from this area, espe- cially from the Ozark Plateau region, is needed before definite determinations can be made. The middle-western material tentative- ly referable to G. minus (G. minus s. lat.) is being studied currently by G. A. Cole (in litt.). Throughout the eastern range of G. minus, the sympatric or syntopic occurrence of this species with any other species of Gammarus is extremely rare; the only notable exception is the sym- patric association of G. minus and G. bousfieldi in Doe Run in Meade Co., Kentucky (Minckley and Cole, 1963). A similar situa- tion of mutually exclusive ranges occurs in England, where over- lapping ranges of species of freshwater Gammarus are uncommon (Hynes, 1955). This situation does not hold true, however, for the western Illinois-eastern Missouri area where there are overlapping ranges of four species of Gammarus: G. troglophilus, G. acheron- dytes, G. pseudolimnaeus and G. minus (s. lat.). Species pairs of Gammarus in this area may consist of G. troglophilus/G. acheron- dytes (restricted to Monroe and St. Clair Counties, Illinois), G. minus (s. lat.)/G. pseudolimnaeus, and occasionally G. minus (s. lat.)/G. troglophilus. The association of G. minus (s. lat.) with G. pseudolimnaeus in southern Illinois springs is rather common. Character displacement may influence the slight but subtle dif- ferences noted for populations of G. minus (s. lat.) from south- western [Illinois and farther west, which have so far complicated MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY 5 the specific assignment of this material. The possibility that char- acter displacement sometimes occurs when G. minus exists syn- topically with other species of Gammarus should be investigated in greater detail. Minckley and Cole (1963) studied this possibility for G. minus /G. bousfieldi in Doe Run, Kentucky, but the evidence for character displacement in G. minus in this particular association was unclear and obscured by other factors (see Discussion and Con- clusions section). G. minus is generally limited in habitat to cave streams, springs and small spring-runs. Larger streams, ponds and even cave pools are seldom inhabited by this species. For example, Spring Creek in Greenbrier County, West Virginia is a large stream, principally derived from spring water. The springs that feed this creek contain large populations of G. minus but the creek itself does not. G. minus probably is excluded from larger streams by a number of ecological factors such as temperature differences, pO», available food, pred- ation, cover and flow rate. It is possible that spring forms of G. minus (Form III) disperse by washing downstream during late winter and spring flooding and subsequently migrate back upstream to the same or a different spring when the water level and flow rate return to normal. We observed a part of this postulated dispersal procedure during Feb- ruary 1967 in Pocahontas County, West Virginia. A large popula- tion of G. minus III occurs in a spring resurgence at the entrance to Overholts Blowing Cave. Observations throughout most of the year revealed that this population is limited to a narrow band ex- tending downstream only a short distance from the cave entrance. However, during a late winter thaw in February 1967, we ob- served a number of individuals more than one-half mile down- stream from the spring. Some of these individuals presumably migrate upstream later and re-populate the spring. Unfortunately, we have no quantitative data to indicate how many individuals are washed downstream during periods of high water or flooding or how many individuals subsequently migrate back upstream to a given spring. There are studies on upstream movement in amphi- pods which point out that not all species are able to migrate. In a paper on upstream movements of Gammarus in Doe Run, Ken- tucky, Minckley (1964) pointed out that G. bousfieldi could mi- grate en masse upstream, while the upstream movement of G. minus was much less obvious. Perhaps, on the contrary, there 6 POSTILLA is no appreciable upstream movement by G. minus to springs after flooding and the ability of this species to maintain itself in springs and headwaters depends on its high reproductive rate. The ques- tion has not yet been answered satisfactorily by either our own or Minckley’s observations, and an investigation on this aspect of the animal’s ecology might provide some interesting data. Nevertheless, Minckley (1963) did point out that very short upstream move- ments of Gammarus of one to two feet per day would assist in maintaining populations in headwaters and springs. G. minus I is represented by populations in two well separated karst areas of the central Appalachian region of Virginia and West Virginia: the Great Savannah karst of south-central Greenbrier County, West Virginia in the Greenbrier Valley (New-Kanawha River drainage); and the Maiden Springs karst of southwestern Tazewell County, Virginia in the headwater region of the Clinch River (upper Tennessee River drainage). These two karst areas are separated by about 76 airline miles and several prominent moun- tains and ridges typical for this part of the Appalachians. Popula- tions of G. minus (Forms II and III) occur in caves and springs - located between and on the sides of these two areas. G. minus | occurs in areas of extensive cave and karst develop- ment further characterized by integrated subsurface drainage. Therefore, we assume that limited dispersal by this species from one cave stream to another can take place through the underground conduits hypothesized to exist between caves in these areas. In the Great Savannah karst, which we have examined most carefully, G. minus | is the most common amphipod species in most cave streams. It is much more common than two other amphipods, Stygonectes emarginatus and S. spinatus, with which it is some- times associated. Although common in cave streams, G. minus I is seldom encountered in rimstone pools located above streams or out of reach of potential flooding by streams. Over 100 rimstone pools were examined in Greenbrier Valley caves and only two of these contained specimens (a total of four) of G. minus. In marked contrast are the two troglobitic species of Stygonectes which were encountered in rimstone and other kinds of drip pools nearly 20 percent of the time and are apparently able to disperse interstitially by phreatic and vadose routes other than streams (Holsinger, 1967, 1969). In comparison with G. minus I, G. minus II is found over a MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY rs much wider part of the central Appalachians and is known pri- marily from caves in Greenbrier, Monroe, Mercer, Randolph, Tucker and Pocahontas Counties in West Virginia; Giles and Taze- well Counties in Virginia; and Fayette County in Pennsylvania. With a few exceptions, this form is comparatively less abundant in cave streams than G. minus I, and usually occurs in caves repre- sentative of small to medium-sized subterranean drainage systems. On the other hand, this form is occasionally found in large cave systems such as Bone-Norman and Friars Hole in Greenbrier County and is rather common in caves of the well-developed karst of southern Pocahontas County (such as Swago Creek and Hills- boro areas). Specimens of Form II also have been collected from caves on the periphery of the Great Savannah karst (viz., Grapevine and Fullers Caves) and of the Maiden Springs karst (viz., Lost Mill Caves). In the caves of Tucker, Randolph and Pocahontas Counties this form is more common (although often not very abundant) than any other amphipod species of this region. In some of these caves it is the only amphipod species recorded. Of further ecological significance is the almost complete absence of G. minus from the cave streams of southwestern Virginia (es- pecially in Lee, Scott and Russell Counties), eastern Tennessee, northwestern Georgia and northern Alabama. Some of the karst, caves and subterranean drainage complexes of this part of the Appalachians are developed to an extent comparable with those of Greenbrier and Tazewell Counties; yet, despite the common oc- currence of G. minus in springs of this region and intensive col- lecting over a ten year period, this species is rarely found in these caves. The same situation is generally true of the caves of the adjacent Interior Low plateau region of southern Indiana, Ken- tucky and central Tennessee where some of the most extensive subterranean drainage systems in the world exist. A partial reason for the scarcity of G. minus in cave streams of these regions is probably its inability to compete successfully with troglobitic species of Crangonyx, C. antennatus and C. packardii (s. lat.). These two species of Crangonyx are distributed over an area ex- tending from southwestern Virginia southwestward to Georgia, across northern Alabama, and north through central Tennessee, central Kentucky and into southern Indiana. In the upper Tennessee River drainage basin, especially in the 8 POSTILLA Powell valley of southwestern Virginia, a concentrated investiga- tion of caves has revealed that C. antennatus is a very common species that inhabits both drip pools and small streams (Holsinger, 1969). The cave-stream macrohabitat of this species appears to be similar to that of G. minus (I and II) of other areas. C. packardii (s. lat.) occupies essentially the same kind of habitat in some of the Interior Low plateau caves, especially those of central Kentucky and southern Indiana. The vagility of these two species of Cran- gonyx, combined with their ability to populate, often in large num- bers, both pools and small streams, is probably indicative of their adaptive success and may be the major reason for the near exclusion of G. minus from caves of the same areas. ANALYSIS OF VARIATION A few minor structural variations were noted that apply to G. minus in general but not to any one morphological form. We will com- ment on these first. Minckley and Cole (1963) studied morphologi- cal variation in several populations of G. minus from northern Kentucky and observed differences in the amount of setation of the first four coxal plates and peduncular segments of the first antenna, occasional absence of calceoli on the second antenna of the male and minor size differences in sexually mature individuals. In gen- eral our studies revealed similar variation. In particular, we found that calceoli were occasionally absent on the second antenna of the male but were usually present in larger males. Like Minckley and Cole (1963), we examined the setal formula given by Bousfield (1958) i.e., peduncular segments 1, 2 and 3 with 1, 3-5 and 1 ventral groups of setae respectively, and found occasional but slight variation. Finally, our observations revealed that the number of accessory flagellar segments of the first antenna may range up to 5 or 6 in some of the larger males. A slight range variation in number of accessory flagellar segments (from 3 to 6) also was recorded by Minckley and Cole (1963) but the differences between populations were not significant. In his diagnosis of G. minus, Bousfield (1958) gave a range of only 3 to 4 accessory flagellar segments, but his observations apparently were based on a small number of speci- mens. MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY 9 We will next examine in detail those characters found by some earlier workers to differ among the three forms of G. minus. Shoe- maker (1940) listed six characters diagnostic of his variety tenuipes: a) slender gnathopods and pereopods, especially the second joint of the last three pereopods; b) lateral lobes of the head with rounded corners; c) reduced number of spines on the urosome in some spec- imens; d) eyes greatly reduced and occasionally almost absent; e) inner ramus of the third uropod proportionately longer; and f) weaker and more delicate appearance of the whole animal. Hubricht (1943) stated the Forms I and II had longer antennae and a bluish color, whereas Form III had shorter antennae and a brownish color. In order to properly analyze the validity of the “diagnostic” characters given above, we assigned all populations studied in detail to one of the three groups based on habitat. These included: 1) the large, well-integrated cave systems in Greenbrier and Taze- well Countries; 2) other caves, excluding those listed in (1); and 3) springs, including resurgences at cave entrances. Thus, in gen- eral, Habitats 1, 2 and 3 should correspond to G. minus I, Il and III populations, respectively. As will be seen below, a division into habitat groups is essential because any mixed population can be separated into groups based on morphology alone. This, however, does not reflect any bio- logical phenomenon except that populations differ. The amount of degeneration of the compound eye is undoubtedly the most striking difference among populations from the three habitat groups. The various stages of eye degeneration are shown in Figure 1; the extremes between individuals from Habitats 1 and 3 are shown in Figures 2 and 3. In most specimens from Habitat 1 the amount of eye degeneration has reached the extreme shown in Figure 3, and there are no discernible eye facets remaining. Occa- sionally, however, a specimen was found in Habitat 1 with a few facets completely formed. Moreover, a few specimens were found in Habitat 2 populations with as much eye degeneration as those from Habitat | populations. The populations from Habitats 2 and 3 can be separated on the basis of the number of well-formed facets in the compound eye. By using specimens from various localities the following data were ob- tained: Springs (Habitat 3): N = 87, median = 28, range — 14 to 42 10 POSTILLA FIG. 1. Structure of the compound eye in mature males (11.00 mm) from four different populations of Gammarus minus. A) Tawneys Cave, Giles County, Virginia (Form II); B) spring in Washington County, Maryland (Form IID; C) the Hole Cave, Greenbrier County, West Virginia (Form I); D) Bowens Cave, Tazewell County, Virginia (Form JI). MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY — 11 ®~¢ re @a @ +s FIG. 2. Head region of Gammarus minus (8.00 mm male) from Fort Spring, Greenbrier County, West Virginia. Upper, enlargement of compound eye showing individual facets. 12 POSTILLA FIG. 3. Anterior region of Gammarus minus (12.00 mm male) from Bene- dicts Cave, Greenbrier County, West Virginia. Upper, enlargement of com- pound eye showing loss of individual facets. MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY — 13 @aves,(Habitat 2): (N = 62> median = 15, range = 0 to 33. The medians are significantly different (P > 0.99) using the distri- bution-free Mann-Whitney U test. Many cases of asymmetry were noted in the amount of eye de- generation in individuals from all three habitat groups. One par- ticularly striking example was a specimen from Coffmans Cave in Greenbrier County which had approximately 20 countable facets in one eye and none in the other. Despite slight variation and asymmetry noted above, differences in the amount of compound eye degeneration were more clear-cut than any other character investigated. A careful examination of numerous specimens did not support Shoemaker’s contentions that the interantennal lobes of the head are more rounded in Form I animals or that the number of dorsal spines on the urosome is reduced. Hubricht (1943) pointed out that spring populations are brown- ish in color and that cave populations are bluish. Although this observation is generally true, there is sometimes a greater varia- tion in color than implied; spring forms tend to be brownish to brownish-green and occasionally brownish-red, while cave forms may vary from bluish to dull gray, and rarely, to almost colorless. For instance, the population from Linwood Cave (Habitat 2) Pocahontas County is especially light in color with some indi- viduals almost colorless, but morphologically this population is intermediate between Forms II and III. Explanations for the re- duction and/or loss of integumentary pigment in cave crustaceans are still incomplete, although a number of experimental studies have been published on this subject (Baldwin and Beatty, 1942; Beatty, 1942, 1949; Anders, 1956; Maguire, 1961). The integumentary pigment of amphipods is made up of various carotenoid-protein complexes; these animals are unable to synthesize the carotenoids and, therefore, must obtain them from the environment (Beatty, 1949; Maguire, 1961 and papers cited therein). Apparently pigmen- tation in many of the troglobitic forms is genetically controlled; hence exposure of these animals to environmental carotenoids does not cause integumentary pigment to reappear (Vandel, 1964). Trog- loxenes and troglophiles vary in their coloration from surface to cave, so availability of usable carotenoid-containing foods may affect the development of carotenoid-based pigments in their in- tegument (Maguire, 1961). Thus spring forms of G. minus with 14 POSTILLA apparently easy access to usable carotenoids in the form of abun- dant vegetable matter are brownish to brownish-green, while cave populations with apparently limited access to usable carotenoids undergo a reduction in pigmentation and are bluish to gray. If a difference in the availability of dietary carotenoids is the major factor in the development of pigment in G. minus, then any con- comitant genetic influence on pigment production would be masked and very difficult to discern. Both Shoemaker (1940) and Hubricht (1943) pointed out that G. minus I had proportionately longer and more slender appen- dages and hence a more fragile appearance than Form III, but it is probably the proportionately longer pereopods and antennae that give G. minus I the overall appearance of having a more delicate body and more slender appendages. These subjective observations are more apparent than real. Shoemaker (1940) also stated that the second joints (or bases) of the last three pereopods were more narrow than in the surface forms, but this is rare. The widths of the pereopod bases differ only slightly among representatives of the habitat groups as seen in Figure 4. Some real differences among the three morphological groups in the proportionate lengths of certain appendages compared to body lengths suggests a trend toward allometry in the cave forms. To investigate the significance of these differences we made numerous measurements and treated the resulting data statistically. Tables | and 2 give data for ratios of pereopod 7, uropod 3 and antenna | to total length for mature males from a variety of spring and cave populations. The ratios among the three habitat groups were Statistically significant (Mann-Whitney U test) with two excep- tions: a) uropod 3/total length is not significantly different in springs and caves (Habitat 2); b) pereopod 7/total length is only marginally significant (P < 0.10) in springs and caves (Habitat 2). A general regression equation for appendage growth is Y = aX’ + c, where: Y is the dependent variable, i.e., antenna 1, pereopod 7 and uropod 3; b is the coefficient of allometry which equals 1 when no allometry occurs; X is the independent variable, i.e., total length; a is the slope of the regression line; c is the intercept on the ordinate. From this regression there are four ways in which the different ratios of dependent variable to independent variable might arise: MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY | 15 OVERHOLTS SPR. LINWOOD CAVE WwW. VA. Ww. VA. 10.50 mm 11.50 mm 1.00 mm TAWNEYS CAVE BOWENS CAVE VA. VA. “THE HOLE" 10.00 mm 12.00 mm W. VA. — 11.00 mm FIG. 4. Variation in structure of pereopod 7 in mature males from the in- dicated localities. Note difference in length but general similarity in structure of the pereopods of the males from the spring in Maryland (Form III) and the cave in West Virginia (Form I). Note also the slight variation in the shape and armature of the bases of the pereopods of males from different popula- tions. 16 POSTILLA TABLE 1. Variation in the ratio of length of three different appendages (antenna 1, pereopod 7 and uropod 3) to length of body in mature males of 12 populations of Gammarus minus. Mean Mean ratios Cave Localities Sample length Antenna Pereopod Uropod size (mm) 1/body 7/body 3 / body Higginbothams, Greenbrier Co., W.Va. 5 Ie) 0.698 0.485 0.217 Buckeye Creek, Greenbrier Co., W.Va. 5 10.9 0.652 0.461 0.192 Grapevine, Greenbrier Co., W.Va. >) 11.8 0.655 0.489 0.187 Linwood, Pocahontas Co., W.Va. 5 123 0.590 0.426 0.175 Cave Hollow, Tucker Co., W.Va. 5 10.8 0.625 0.452 0.180 Benedicts, Greenbrier Co., W.Va.* 5 10.8 0.750 0.513 0.223 Ludington, Greenbrier Co., W.Va.* 5) 10.4 0.674 0.485 0.188 McClungs, Greenbrier Co., W.Va.* 8 ies 0.720 OU523 0.207 The Hole, Greenbrier Co., W.Va.* 12 1253 0.707 0.494 0.200 Greenbrier, Greenbrier Co., W.Va.* 20 12.0 0.647 0.482 0.209 Hugh Young, Tazewell Co., Va.* 7 11.0 0.749 0.485 0.226 Bowens, Tazewell Co., Va.* 1 11.9 0.793 0.493 0.226 *Indicates Habitat | cave; the remaining populations are from Habitat 2 caves. TABLE 2. Variation in the ratio of length of three different appendages to body length in mature males of 11 spring populations of Gammarus minus. Mean Mean of ratios Spring Localities Sample length Antenna Pereopod Uropod size (mm) 1/body 7/body 3 / body Washington Co., Maryland 20 11.8 0.529 0.426 0.183 Smulton Spr., Centre Co., Pa. 10 10.1 0.576 0.449 0.188 Lancaster Co., Pa. 10 11.6 OS 0.432 0.185 Tazewell Co., Va. 10 10.3 OS72Z 0.443 0.189 Maiden Spr., Tazewell Co., Va. 5) 9.1 0.556 0.416 OTs Spr. at Van, Lee Co., Va. 6 8.7 0.602 0.486 0.195 Windfields Spr., Bath Co., Va. 18 11.8 0.569 0.432 0.185 Sweet Sprs., Monroe Co., W.Va. 10 8.6 0.579 0.424 0.178 Overholt Spr., Pocahontas Co., W.Va. 11 9.6 0.613 0.452 0.190 Cold Spr., Monroe Co., W.Va. 10 9.7 0.586 0.443 0.191 Spr. on Mill Run, Tucker Co., W.Va. 10 alee? 0.570 0.459 0.166 ee ee MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY _ 17 1) Different ratios may reflect different coefficients of allometry. In particular, when 5b is less that 1 the dependent variable is in- creasing with a decreasing rate, and when 5 is greater than 1, the dependent variable is increasing with an increasing rate. 2) a, b and c may be the same for all groups, but when c # O, changing the size of the independent variable, i.e., total length, will change the ratio. 3) Different ratios may reflect different slopes of the linear (b = 1) regression line. 4) Different ratios may reflect different intercepts. In practice, solving for b in the above equation involves taking the logs of both sides which causes c to disappear. In order to find c by the usual methods, b must be considered to be equal to 1. Moreover, we will only consider a when b is assumed to be equal to 1. First, we will consider those dependent variables that show allometric growth. Those that do not show any allometry will be considered in terms of the last three possibilities listed above by using the equation Y = aX + c. TABLE 3. The results of a test for allometry between the three habitat groups of Gammarus minus. Independent Dependent variable variable Habitat b* total length pereopod 7 3 (springs) 0.96 + 0.08 i i: 2 (caves) 0.86 + 0.33 ms ne 1 (large caves) 0.98 + 0.16 e uropod 3 3 (springs) 0:87 22011 se ” 2 (caves) 0.81 + 0.62 es a 1 (large caves) Let OE 1022. é antenna 1 3 (springs) 0.79 + 0.12 és ‘ Fort Spring (3) MOL se 0.12 3 os 2 (caves) 0.81 + 0.31 s Ai Coffmans Cave (2) 0.92 + 0.12 is me 1 (large caves) 0.83 + 0.14 me a Benedict Cave (1) 0.83 + 0.14 outer ramus inner ramus Fort Spring (3) 127 = 0:09 of third | of third Coffmans Cave (2) Ps 0310 uropod uropod Benedict Cave (1) 1.15 2='0:09 1Those habitats labeled 1, 2 and 3 were calculated using only mature males. For Fort Spring, Coffmans Cave and Benedicts Cave, all sizes and both sexes were used. *If b is not significantly different from 1.0 there is no evidence for allometry. 18 POSTILLA The results of the analysis of allometry are given in Table 3. The only clear case of positive allometry where b is greater than 1 occurs in the regression of the inner ramus of the third uropod against the outer ramus of the third uropod, but not in the regression of body length and uropod 3. However, positive allometry occurs in all three habitat groups, not just in the Habitat | populations. There is a tendency for antenna | to display slightly negative allometry in all three habitat groups. Therefore, we can conclude that there are no differences in allometry which could explain differences in the appendage to body length ratios. If all the populations are on the same regression line, then the total lengths of the cave populations must be greater than the spring populations, and the intercept c must be less than one or vice versa. Random samples from three populations in Greenbrier County were measured for total length. These populations were from Benedicts Cave (Habitat 1), Coffmans Cave (Habitat 2) and Fort Spring (Habitat 3). A Mann-Whitney U test was used to test for significant differences. The Benedicts Cave population was significantly larger (P > 0.99) than Coffmans Cave population; the latter, in turn, was significantly larger than the Fort Spring popula- tion (P > 0.99). This relationship also holds true if only mature males are considered. If the mean lengths of mature males from springs (Table 2) are compared with mean lengths of mature males from caves of either Habitat | or 2, the differences are significant (P > 0.95 by the Mann-Whitney U test), but differences between the two cave habitat groups are not significant. These results indi- cate that as we go from spring forms to cave forms the total length (independent variable) increases, thus effectively moving up the re- gression line. However, when we consider the actual intercepts of the regression lines (see Table 4) very few of them have negative intercepts. Only in the comparison of the inner and outer rami of the third uropod are all of the intercepts of the three groups nega- tive, and this particular ratio is complicated by allometry. There is no significant pattern toward a higher slope or higher intercept in cave populations as shown in Table 4. However, the slopes of the Habitat 1 populations tend to be higher for pereopod 7, uropod 3 and antenna |. A pattern in the intercepts is much less clear. This pattern may not hold true for spring populations in general. The problem can be seen by comparing the antenna | regression of MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY — 19 TABLE 4. Analysis of the changes in intercept and slope of linear regression where Y = aX +c. Independent Dependent Habitat* a : variable variable Ee ee ee total length pereopod 7 3 (springs) 0.43 + 0.04 as ? in 2 (caves) 0.39 + 0.16 8.33 3 1 (large caves) 0.48 + 0.08 Das id uropod 3 3 (springs) 0.16 + 0.02 239 rs 2 2 (caves) 0.16 + 0.11 3702 Yi 1 (large caves) 0.22+0.05 —1.34 a antenna | 3 (springs) 0.44 + 0.06 13510 23 a Fort Spring (3) 0.58 +0.06 —2.30 2 2 2 (caves) O35 0225533382 x * Coffmans Cave (2) 0.56 + 0.07 5.50 a = 3 (large caves) 0.59+0.18 11.90 » a Benedict Cave (1) 0.45 + 0.08 7.60 outer ramus | inner ramus Fort Spring (3) 0.73 + 0.04 —4.20 of third of third Coffman Cave (2) 0.67+0.05 —2.90 uropod \ uropod Benedict Cave (1) 0.63 +0.04 —2.50 *Those habitats labeled 1, 2 and 3 were calculated using only mature males. For Fort Spring, Coffmans Cave and Benedicts Cave, all sizes and both sexes were used. all spring populations with the Fort Spring population, Habitat 2 populations with the Coffmans Cave population and Habitat 1 populations with the Benedicts Cave population. Although ex- pected, there is little correspondence between these pairs. This is due in part to the fact that only mature males were used for the first member of each pair, therefore resulting in the large standard errors seen in Table 4. It is also possible to spot check for differences in slope in the regression by comparing the ratios of various appendages to total length from two populations with approximately the same range in total lengths. This minimizes the disturbing influences of allometry and the differences due to different intercepts. The results of these comparisons are shown in Table 5, and. as indicated by these data, one almost always finds a significant difference whenever popula- tions from mixed habitats are compared. Whenever two like popu- lations are compared, there are significant differences between these pairs about one-third of the time. 20 POSTILLA TABLE 5. Mann-Whitney U test on the ratios of lengths of antenna 1, pereopod 7 and uropod 3 to total length for a variety of population pairs of Gammarus minus. Population pair and habitat group Antenna | Uropod 3 Pereopod 7 Spr., Washington Co., Md. (3) *The Hole Cave, W.Va. (1) <.001 <.001 <.001 *McClungs Cave, W.Va. (1) Spr., Tucker Co., W.Va. (3) <.001 <.001 <.001 *The Hole Cave, W.Va. (1) Spr., Bath Co., Va. (3) <.001 <.001 <.010 *Bowens Cave, Va. (1) Spr., Bath Co., Va. (3) <.001 <.001 <.0001 *Higginbothams Cave, W.Va. (2) Spr., Bath Co., Va. (3) <.001 <.001 <.001 *Grapevine Cave, W.Va. (2) Spr, Bath Cox Vas (3) <.001 <.005 N.S. *Grapevine Cave, W.Va. (2) Spr., Washington Co., Md. (3) <.001 <.025 N.S. «Spr. Bath’ Cos Vas (G3) Spr., Washington Co., Md. (3) <.001 NS. NS. *The Hole Cave, W.Va. (1) Greenbrier Caverns, W.Va. (1) <.005 N.S. <.050 =Spr.leee Co. Via. (3) Spr., Monroe Co., W.Va. (3) N.S. <.001 N.S. Grapevine Cave, W.Va. (2) Higginbothams Cave, W.Va. (2) N:S. N.S. NS. Cave Hollow Cave, W.Va. (2) Buckeye Creek Cave, W.Va. (2) N.S. N.S. N.S. *Ludington Cave, W.Va. (1) Benedicts Cave, W.Va. (1) <.050 NS. <.005 *Indicates the population with higher ratios. DISCUSSION AND CONCLUSIONS On the basis of the observations and analyses given above, we have concluded that G. minus is a single, highly variable species. The careful examination of a number of diagnostic characters did not reveal a single character that would unequivocally divide this species into separate taxa. The former recognition of a separate and distinct variety tenuipes probably resulted from the failure of earlier workers to examine sufficient material from a wide variety of habitats. As already pointed out, Minckley and Cole (1963) found measur- able variations in populations of G. minus from northern Kentucky. When associated with G. bousfieldi in Doe Run, G. minus differed from the “typical” spring populations and the possibility of char- MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY 21 acter displacement was implied. Morphological variation noted for Doe Run populations associated with beds of Fissidens (an aquatic moss) resulted from their association with the moss, since the same variation was noted in G. minus whether G. bousfieldi was present or absent (Minckley and Cole, 1963). Similarly, some of the varia- tion we observed within populations from the various habitat groups (see Tables 1, 2 and 5) might also have been affected by variation in the nature of the habitat. The springs observed varied: some are fast-flowing like the Overholts Blowing Cave resurgence; some are small, seepage types while others are large, pond-like springs with considerable vegetation (such as Fort Spring). According to Houston (1960), differences in current velocity affect the size at maturity of Gammarus pulex. Similarly, current velocity may also affect the size at maturity of G. minus in both caves and springs, since flow rate in both kinds of habitats is often variable. Other environmental parameters of cave streams such as temperature, quantity of washed-in food and chemistry, are also variable and are often in- fluenced by seasonal changes on the surface. One question still largely unresolved is why populations of G. minus I are restricted to certain caves in two very specific areas. Are certain caves able to select out these extreme types while others are not? We cannot satisfactorily answer this question, especially in view of the fact that G. minus Il occasionally occurs in large cave systems (such as Bone Norman Cave) and some of the G. minus II populations occupy caves immediately adjacent to or even hydrologically related to caves with Form I populations. It should be noted, however, that G. minus I does not inhabit smaller, semi- isolated caves and that the two areas inhabited by G. minus I con- tain large, well-integrated drainage systems. Bowens and Hugh Young Caves, although small in terms of human traverse, are a part of a large, underground drainage complex believed to resurge through Maiden Springs in Tazewell County. The most accessible part of this complex is Fallen Rock Cave which contains five or more miles of traversable passages. In terms of traversable size, however, some of the Great Savannah caves in Greenbrier County are considerably larger. Greenbrier Caverns, for example, has at least 18 miles of explorable passages. It would appear, then, that a large subterranean drainage system is a necessary prerequisite for the development of Form I morphology. Undoubtedly, selection is playing a role here, but exactly how it operates is not clear. 22 POSTILLA The variation in G. minus may be genetic or ecophenotypic, or influenced by both effects. It is apparent that the G. minus I mor- phology is composed of the extremes present in both G. minus II and III populations that are being selected for by certain, as yet undetermined, factors of the subterranean environment. Depending on circumstances, the same morphological change may be ecophen- otypic or genetic; processes similar to those shown by Waddington’s (1956) classical experiment on genetic assimilation of environ- mentally induced change in Drosophila melanogaster may explain some variation in G. minus. If the three forms of G. minus are components of a single, variable species, one must accept the potential for gene exchange between population extremes. The opportunities for gene exchange are probably as great, or even greater, between certain spring and cave populations than they are between widely distributed spring populations. Almost any spring population is semi-isolated and such populations are best regarded as geographic isolates in the sense of Mayr (1963). If we consider physical barriers and drainage patterns in karst areas, migration from a spring into a cave or vice versa is easier to envision than migration from one spring to another. The latter event usually would be limited to the rather circuitous routes of surface streams, while the former could take place more directly. One means by which dispersal might occur between springs has already been suggested, but at best this method is limited and difficult to conceive of as a common event. Environ- mental conditions, rather than isolation, might play a more sig- nificant role in determining the form in a given habitat. One of the problems encountered with G. minus in the central Appalachians was somewhat similar to that which Christiansen and Culver (1968) found in the cave collembolan Pseudosinella hirsuta. With this species, there was a striking parallelism and con- vergence in morphology in geographically isolated, highly cave- adapted populations. We found a similar situation with G. minus in two well-separated and isolated karst areas with similar habitats. The tenuipes form may be a convergent ecotype (see Dobzhansky, 1951) which occurs only under special environmental conditions and in the presence of proper genetic variants. The present study is preliminary and it is obvious that there are a number of uninvestigated aspects pertinent to the overall problem which would provide interesting topics for future studies. MORPHOLOGICAL VARIATION IN GAMMARUS MINUS SAY — 23 ACKNOWLEDGMENTS A part of this study was supported by grants to Holsinger from the National Speleological Society Research Advisory Committee and the Old Dominion University Educational Foundation and by a grant to Culver from the Cave Research Foundation. We are grateful to the West Virginia Association for Cave Studies, Inc. (WVACS) for providing field assistance in Greenbrier County, and to Dr. Thomas E. Bowman for providing specimens from the Smithsonian Institution (United States National Museum). We thank the following persons for their assistance in the field and/or for their contribution of specimens: Roger Baroody, William Big- gers, John and Martha Cooper, Richard Franz, Ronald Larue, Charles Maus, David Newson, John Rutherford and Paul Starr. We are also indebted to a number of landowners who kindly al- lowed us to visit caves on their property. LITERATURE CITED Anders, Fritz. 1956. Uber Ausbildung und Vererbung der Korperfarbe bei Gammarus pulex ssp. subterraneus (Schneider), einer normalweise pig- mentlosen: Hohlenform des gemeinen Bachflohkrebses. Zeit. indukt. Abstammungs-u.Vererbungslehre, 87: 567-579. Baldwin, E. and R. A. Beatty. 1942. The pigmentation of cavernicolous animals I. The pigments of some isopod Crustacea. J. Exp. Biol., 18: 136-143. Beatty, R. A. 1942. The pigmentation of cavernicolous animals II. Carotenoid pigments in the cave environment. J. Exp. Biol., 18: 144-152. 1949. The pigmentation of cavernicolous animals III. The carotenoid pigments of some amphipod Crustacea. J. Exp. Biol., 26: 125—130. Bousfield, E. L. 1958. Fresh-water amphipod crustaceans of glaciated North America. Can. Field-Natur., 72(2): 55-113. Christiansen, K. A. and D. C. Culver. 1968. Geographic variation and evolu- tion in Pseudosinella hirsuta. Evolution, 22: 237-255. Dobzhansky, Theodosius. 1951. Genetics and the origin of species. Columbia Univ. Press, New York, 364 p. Holsinger, J. R. 1967. Systematics, speciation, and distribution of the sub- terranean amphipod genus Stygonectes (Gammaridae). U.S. Nat. Mus. Bulls 2595 1761p: 1969. Biogeography of the freshwater amphipod crustaceans (Gam- maridae) of the central and southern Appalachians pp. 19-50. Iii les, (Gy Holt [ed.] The distributional history of the biota of the southern Ap- palachians. Part I: Invertebrates. Virginia Polytechnic Inst. Press, Blacks- burg, Va. : Houston, K. 1960. Size of Gammarus pulex in fast- and slow-running streams. Report Dauntsey’s School Nat. Hist. Soc., 1959-1961, pp. 11-13. 24 POSTILLA Hubricht, L. 1943. Studies on the Nearctic freshwater Amphipoda. III: Notes on the freshwater Amphipoda of eastern United States with description of ten new species. Amer. Midl. Nat., 29: 683-712. 1959. Malacostraca (Amphipoda), pp. 876-878. Jn W. T. Edmondson [ed.]| Ward and Whipple’s freshwater biology. John Wiley and Sons, New York. Hubricht, L. and J. G. Mackin. 1940. Descriptions of nine new species of fresh-water amphipod crustaceans with notes and new localities for other species. Amer. Midl. Nat., 23: 187-218. Hynes, H. B. N. 1955. Distribution of some fresh-water Amphipoda in Britain. Verh. int. Ver. Limnol., 12: 620-628. Mackin, J. G. 1941. A key to the species of Amphipoda of Oklahoma. Proc. Oklahoma Acad. Sci., 21: 29-30. Maquire, B. 1961. Regressive evoiution in cave animals and its mechanism. Texas J. Sci., 13: 363-370. Minckley, W. L. 1964. Upstream movements of Gammarus (Amphipoda) in Doe Run, Meade County, Kentucky. Ecology, 45(1): 195-197. Minckley, W. L. and G. A. Cole. 1963. Ecological and morphological studies on gammarid amphipods (Gammarus spp.) in spring-fed streams of northern Kentucky. Occasional Papers of the C. C. Adams Center for Ecol. Stud., 10: 1-35. Mayr, E. 1963. Animal species and evolution. Belknap Press, Cambridge, Mass., 797 p. Say. T. 1818. An account of the Crustacea of the United States. J. Acad. Natur. Sci. Philadelphia, 1: 374-401. Shoemaker, C. R. 1940. Notes on the amphipod Gammarus minus Say and description of a new variety, Gammarus minus var. tenuipes. J. Wash- ington Acad. Sci., 30: 388-394. Stoll, N. R. (Chairman), R. Ph. Dollfus, J. Forest, N. D. Riley, C. W. Sabrosky, C. W. Wright, R. V. Melville (Sec’y.) 1964. International code of zoological nomenclature. International Trust for Zoological Nomenclature, London, 176 p. Vandel, A. 1964. Biospéologie. Gauthier-Villars, Paris, 619 p. Waddington, C. H. 1956. Genetic assimilation of the bithorax phenotype. Evolution, 10: 1-13. ‘ REVIEW si YLE FORM TTEE ABSTRACT NOMENCLATURE ILLUSTRATIONS FOOTNOTES TABLES REFERENCES AUTHOR'S COPIES PROOF COPYRIGHT INFORMATION FOR AUTHORS The Publications Committee of the Peabody Museum of Natural History reviews and approves manuscripts for publication. Papers will be published in approximately the order in which they are accepted; delays may result if manuscript or illustrations are not in proper form. To facilitate review, the original and one carbon or xerox copy of the typescript and figures should be submitted. The author should keep a copy. Authors of biological papers should follow the Style Manual for Biological Journals, Second Edition (Amer. Inst. Biol. Sci.). Authors of paleontological manuscripts may choose to follow the Sugges- tions to Authors of the Reports of the U.S. Geological Survey, Fifth Edition (U.S. Govt. Printing Office). Maximum size is 80 printed pages including illustrations (= about 100 manuscript pages including illustrations). Manuscripts must be typewritten, with wide margins, on one side of good quality 814 x 11” paper. Double space everything. Do not underline anything except genera and species. The editors reserve the right to adjust style and form for conformity. Should be precise and short. Title should include pertinent key words which will facilitate computerized listings. Names of new taxa are not to be given in the title. The paper must begin with an abstract. Authors must submit com- pleted BioAbstract forms; these can be obtained from the Postilla editors in advance of submission of the manuscripts. Follow the International Codes of Zoological and Botanical Nomen- clature. Must be planned for reduction to 4 x 614” (to allow for running head and two-line caption). If illustration must go sideways on page, reduction should be to 34 x 6°%4”. All illustrations should be called “Figures” and numbered in arabic, with letters for parts within one page. It is the author’s responsibility to see that illustra- tions are properly lettered and mounted. Captions should be typed double-spaced on a separate page. Should not be used, with rare exceptions. If unavoidable, type double-spaced on a separate page. Should be numbered in arabic. Each must be typed on a separate page. Horizontal rules should be drawn lightly in pencil; vertical rules must not be used. Tables are expensive to set and correct; cost may be lowered and errors prevented if author submits tables typed with electric typewriter for photographic reproduction. The style manuals mentioned above must be followed for form and for abbreviations of periodicals. Double space. Each author receives 50 free copies of his Postilla. Additional copies may be ordered at cost by author when he returns galley proof. All copies have covers. Author receives galley proof and manuscript for checking printer’s errors, but extensive revision cannot be made on the galley proof. Corrected galley proof and manuscript must be returned to editors within seven days. Any issue of Postilla will be copyrighted by Peabody Museum of Natural History only if its author specifically requests it. Harvard MCZ Libra HDIOUNVULI NIQUE 3 2044 en es aah = < _— paleysy eerter maple Wed cae seme = pote, Sa Aone eens lah a A eww eo ot SP PE RTS ete eg ee ea reary ee SO PPE LE ELT UPD VAG efod ae? gto eats r Srp ie Oak wiry et We" =* : poe ene ye Ie ey PS Pe Oey ord? eae See At ttt tad Oe ek ernie ee ey wer Ve ae ae De Verena ONES eee BSS paries ee ee Pees we eevee va eerTeerrs s ap wee Fa ee ee ddd Se ee oe oe ee