morphometric patterns among diving beetles (coleoptera...

Morphometric patterns among diving beetles (Coleoptera: Noteridae, Hygrobiidae, and Dytiscidae)

I. Ribera and A.N. Nilsson

Abstract: A study of shape and size in relation to swimming strategies among Hydradephaga, with an emphasis on Dytiscidae, was performed with nearly 1600 adult specimens belonging to the families Noteridae (2 species), Hygrobiidae (1 species), and Dytiscidae (74 species). The data were studied by means of bivariate correlations, scatter plots, and two multivariate statistical methods (factor and cluster analysis). The main morphometric groups found included (i) large to medium-sized, streamlined, wide species with the maximum width in the rear part of the body and the maximum height in the front part, with short tibiae and long tarsi; they are considered to be adapted to high-speed swimming in open waters; (ii) small to medium-sized species with a spherical body and long femora; they are considered to be adapted to manoeuvring in stagnant waters; (iii) small species with a discontinuous outline, a narrow body, and long, slender legs; they are considered to be poor swimmers in running waters; and (iv) small to medium-sized species having, in general, a streamlined, relatively high body with short, wide legs; they are considered to be adapted to crawling among dense vegetation or detritus. Despite the clear relationships between systematics and morphometry, most characters were found to be homoplastic a number of times within the families studied. This supports the adaptive interpretation of the morphometry of the species.

RCsumC : On trouvera ici les rCsultats d'une Ctude de la forme et de la taille en relation avec les stratkgies natatoires chez les HydradCphages, Noteridae (2 espkces), Hygrobiidae (1 espkce) et, plus particulikrement, chez les Dytiscidae (74 espkces). Les donnkes ont CtC analysCes au moyen de corrklations bidimensionnelles, de diagrammes de dispersion et de deux mkthodes statistiques multidimensionnelles (analyse factorielle et analyse des groupements). Les principaux groupes morphomCtriques form& sont : (i) les espkces de taille moyenne ou grande, larges et fusiformes, dont la largeur maximale se trouve la partie arrikre du corps et la hauteur maximale, a la partie avant, aux tibias courts et tarses longs; ces espkces sont considCrCes comme adaptkes a la nage rapide en eau libre; (ii) les espkces de taille petite ou moyenne a corps sphkrique et a femurs longs; ces espkces sont considCrCes comme adaptkes aux eaux stagnantes; (iii) les espkces de petite taille a silhouette irrkgulikre, au corps Ctroit et long, aux pattes minces; ces espkces sont considCrCes comme de mauvais nageurs en eau courante; (iv) les espkces de taille petite ou moyenne, fusiformes, a corps relativement haut, A pattes courtes et larges; ces espkces sont considCrCes comme adaptCes a la reptation dans la vCgCtation dense ou les dktritus. En dCpit des relations Cvidentes entre la morphomktrie et la classification, la plupart des caractkres se sont avCrCs homoplastiques un grand nombre de fois chez les familles CtudiCes, ce qui appuie l'hypothkse selon laquelle la morphomktrie a valeur d'adaptation chez ces espkces. [Traduit par la Rkdaction]

Introduction invasions of the aquatic medium (Lawrence and Newton 1982; Evans 1982, 1985; Nichols 1985; Beutel and Roughley

Noteridae, Hygrobiidae, and Dytiscidae are three families of 1988), that have resulted in the Haliplidae, Gyrinidae, and predacious hydradephagan diving beetles that capture their the group formed by the Noteridae, Amphizoidae, Hygro- prey while actively swimming. agree that the biidae, and Dytiscidae. All these authors agree on the mono- aquatic Ade~haga are polyphyletic, with three Dhyly of each of the three families studied here. In these . - -

Received April 4, 1995. Accepted August 10, 1995. families the adults have comparable shapes over a wide size range. In the fauna of the Pyrenees, the size range is from

I. Ribera1 and A.N. Nilsson. Department of Biology, less than 2 mm in Bidessus minutissirnus (Germar) to more University of Umei, S-901 87 Umei, Sweden (e-mail: than 30 mm in Dytiscus marginalis L. [email protected] .se). The body shape has always been interpreted as an adap- ' Present address: Environmental Sciences Department, The fation to swimming. Although gravity is not an imponant

Scottish Agricultural College, Auchincruive, Ayr factor for small aquatic organisms, shape and size are deter- KA6 5HW, U.K. (e-mail: i. [email protected]). minant characteristics that conform to the drag forces and the

Can. J . Zool. 73: 2343-2360 (1995). Printed in Canada 1 Imprime au Canada

Can. J. Zool. Vol. 73, 1995

energetic cost of swimming (Schmidt-Nielsen 1972). Cooper et al. (1985) found that different hunting strategies adopted by aquatic predators were usually linked with different ways of swimming. Daniel (1984) considered that size and shape were fundamental characteristics that determine the velocity and manoeuvrability of aquatic organisms in predator -prey interactions. Strong selective pressures on these characters could be expected to fit size and shape to the physical environment and the way of life.

The structural adaptations of the adults (more or less developed in most species) have been described by many authors (Guignot 193 1 - 1933; Balfour-Browne 1940, 1959; Hughes 1958; Galewski 197 1 ; Nachtigall 1974; Roughley 1981; Lawrence and Newton 1982; Evans 1982, 1985; for a detailed revision see Ribera and Isart 1991). They include the metacoxal articulation, the continuous body outline without irregularities or protuberances that could produce turbu- lence, and modification of the hind legs (and, to a lesser extent, the midlegs), which produce the main part of the thrust in most Dytiscidae and Noteridae (Natchtigall 1977). The genus Hygrobia has a different style of swimming, the "dog paddle," using all three pairs of legs, like .the Hydro- philidae and Haliplidae. These modifications involve an increase in area and a reduction of discontinuities between the segments. The metatarsal claws change their morphological position and lose their original function and the swimming hairs develop and increase in size. The metatarsi are broader and longer, the metatibiae shorter, and the meta- femora (with the muscles that move the tibia) broader than in terrestrial adephagous beetles. The whole of the leg is flattened in one plane, especially the tibia and tarsus. The swimming hairs are very important in broadening the effective area of thrust (Hughes 1958; Blake 1986).

All the characters discussed above are mainly found in the larger species, whilst the smaller species (especially the Hydroporinae) are more variable in morphology (Nachtigall 1977; Wolfe and Zimmerman 1984). The less modified forms (according to these characters) were considered by some authors to be the most primitive (Guignot 1931 - 1933; Galewski 197 1 ; Nachtigall 1974; Francisco10 1979). In the opinion of the same authors, this evolution has paralleled an increase in size: there would be a single adaptive line (with some variation around a single morphological model) that would go via the small, less modified species (Hydroporinae) to large, strongly modified ones (chiefly the Dytiscinae), the Cybistrini being the most modified of all Dytiscidae world- wide. The possibility that other swimming strategies could be reflected in the shape or the body proportions is seldom considered. Only the work of Wolfe and Zimmerman (1984) and, partly, Nachtigall (1974) suggested possible specializa- tion, not to velocity but to manoeuvrability, by means of a more spherical body. Wolfe and Zimmerman (1984) linked this spherical form with more turbulent lotic habitats, whereas the dorsoventrally flattened species were associated with lentic habitats.

The aim of this study was to describe the morphometric variation within the group and to characterize the main morphometric patterns among species. These morphometric patterns were interpreted as different swimming strategies and, in general, different ways of life. This interpretation was based on theoretical hydrodynamic considerations and on com-

parison with other groups such as birds or fish (Hespenheide 1975; Webb 1984).

Materials and methods

The specimens studied were collected in the Pyrenees and pre-Pyrenees (northwestern Spain), mainly between 1985 and 1990 (for a detailed description of the area, localities, and material collected see Ribera 1992). Three families were studied: Noteridae (2 species), Hygrobiidae (1 species), and Dytiscidae (74 species) (Table 1).

Up to 30 adults of each species were measured (1564 in total). Two groups of 10 individuals (5 of each sex) from two areas distant from one another were selected, plus 10 additional specimens from other localities. When this criterion was not met, all collected individuals were measured. For most species (43, more than 20 specimens were measured, and for only 20 species were fewer than 10 measured (Table Al). Of these species, most belonged to genera represented by other species of similar sizes and shapes. When species represented by few specimens were deleted from the analysis, patterns did not change. In studies of a wide range of taxa, if intrataxon variability is very small relative to inter- taxon variability (as was the case here), sampling a few individuals from each taxon is acceptable (Marcus 1990).

Morphometric variables The measured characters were chosen to characterize size and shape (Table 2). Some of them have been used in previous studies of the group (Zimmerman and Ludwig 1975; Wolfe and Matta 198 1 ; Shirt and Angus 1992; Bilton 1993). Measurements of structures were made in different dimensions, to reflect shape and not only size (Reyment et al. 1984). Some apparently similar variables were introduced, e.g., maximum body height (MH) and maximum elytron height (EH) so that species with a dorsoventrally symmetrical body may be distinguished from those with a more asymmetrical shape, and maximum width (MW) and pronotum width (PW) so that those with a more rectangular shape may be distinguished from those with a more oval shape.

The measure of body length (TL) did not include the head because it was often extruded in specimens preserved in ethanol, which led to errors in measurement. Some measures were made of the hind legs, which are the main source of thrust. The muscles that move the tibia and tarsus are located in the femur (Nachtigall 1974), so its length and width are probably good indicators of the power of thrust.

The categorical variables were introduced to allow com- parisons of the continuity of the outline and the length of the swimming hairs, which are important features in .the morphology of this group of Coleoptera (see the Introduction).

Measuring methods An image was projected from a binocular microscope with a TV camera to a TV monitor, and measurements were made on the screen with a transparent flexible scale. The deformation introduced by the curvature of the screen was checked against measures taken directly from the microscope. In the centre of the screen the deformation was negligible. The magnifications used with each species were adjusted so that

Ribera and Nilsson

Table 1. Species studied, with codes and numbers.

Species Code

Noteridae 1. Noterus clavicomis (De Geer, 1774) Not. clav 2. Noterus laevis Sturm, 1834 Not. laev

Hygrobiidae 3. Hygrobia h e m n n i (Fabricius, 1775) Hyb. herm

Dytiscidae Copelatinae

4. Copelatus haemorrhoidalis (Fabricius, 1787) Cop. haem Hydroporinae

5. Hyphydrus aubei Ganglbauer, 1892 Hyp. aube 6. Hydrovatus clypealis Sharp, 1876 Hyv.clyp 7. Hydrovatus cuspidatus (Kunze, 18 18) Hyv . cusp 8. Yola bicarinata (Latreille, 1804) Yol.bica 9. Bidessus goudoti (Castelnau, 1834) Bid .goud

10. Bidessus minutissimus (Germar , 1824) Bid. minu 1 1. Hydroglyphus pusillus (Fabricius, 178 1) Hyl . pusi 12. Coelambus conjluens (Fabricius, 1787) Coe. conf 13. Coelambus impressopunctatus (Schaller, 1783) Coe. impr 14. Coelambus marklini (Gyllenhal, 18 13) Coe.mark 15. Hygrotus inaequalis (Fabricius, 1777) Hyt.inae 16. Hydroporus longulus Mulsant, 1860 Hyd. long 17. Hydroporus discretus Fairmaire & Brisout in Fairmaire, 1859 Hyd.disc 18. Hydroporus foveolatus Heer , 1 839 Hyd . fove 19. Hydroporus marginatus (Duftschmid, 1805) Hyd. marg 20. Hydroporus memnonius Nicolai, 1822 Hyd. memn 2 1. Hydroporus nigellus Mannerheim, 1853 Hyd. nige 22. Hydroporus nigrita (Fabricius, 1792) Hyd. nigr 23. Hydroporus nivalis Heer, 1839 Hyd. niva 24. Hydroporus palustris (Linnaeus, 176 1) Hyd. palu 25. Hydroporus planus (Fabricius, 178 1) Hyd.plan 26. Hydroporus pubescens (Gyllenhal , 1808) Hyd.pube 27. Hydroporus tessellatus Drapiez, 18 19 Hyd. tess 28. Hydroporus vagepictus Fairmaire & Laboulbkne, 1854 Hyd . vage 29. Graptodytes bilineatus (Sturm, 1835) Grt.bili 30. Graptodytes jlavipes (Olivier , 1795) Grt . flav 3 1. Graptodytes ignotus (Mulsant, 1861) Grt .igno 32. Graptodytes varius (Aubk , 1 836) Grt .vari 33. Rhithrodytes bimaculatus (Dufour , 1852) * Rhi. bima 34. Metaporus meridionalis (Aubk, 1 836) Met.meri 35. Scarodytes halensis (Fabricius, 1787) Sca. hale 36. Stictonectes epipleuricus (Seidlitz, 1887) Stn. epip 37. Stictonectes lepidus (Olivier, 1795) Stn.lepi 38. Stictonectes optatus (Seidlitz, 1887) Stn.opta 39. Deronectes aubei (Mulsant, 1843) Der .aube 40. Deronectes delarouzei (du Val, 1857) Der .dela 41 . Deronectes moestus (Fairmaire, 1858) Der . moes 42. Deronectes opatrinus (Germar , 1 824) Der . opat 43. Stictotarsus duodecimpustulatus (Fabricius, 1792) Stt.duod 44. Stictotarsus griseostriatus (De Geer , 1774) Stt .gris 45. Nebrioporus canaliculatus (Lacordaire, 1 835) Neb. cana 46. Nebrioporus fabressei (Rkgimbart , 190 1) Neb. fabr 47. Nebrioporus depressus elegans (Panzer , 1 794) Neb.depr 48. Oreodytes davisii (Curtis, 183 1) Ore. davi 49. Oreodytes sanmarkii (C .R. Sahlberg , 1826) Ore. sanm

Colymbetinae 50. Platambus maculatus (Linnaeus, 1758) Pla. macu 5 1. Agabus brunneus (Fabricius, 1798) Aga. brun


Table 1 (concluded).

Species Code

52. Agabus didymus (Olivier, 1795) 53. Agabus guttatus (Paykull, 1798) 54. Agabus biguttatus complex 55. Agabus bipustulatus (Linnaeus, 1767) 56. Agabus chalconatus (Panzer, 1796) 57. Agabus lapponicus (Thomson, 1867) 58. Agabus conspersus (Marsham, 1802) 59. Agabus nebulosus (Forster, 177 1) 60. Agabus labiatus (Brahm, 1790) 6 1. Ilybius meridionalis AubC, 1836 62. Rhantus suturalis (McLeay , 1825) 63. Colymbetes fuscus (Linnaeus, 175 8) 64. Melademu coriacea Castelnau, 1834

Laccophilinae 65. Luccophilus hyalinus (De Geer, 1774) 66. L.accophilus minutus (Linnaeus, 1758) 67. Laccophilus ponticus Sharp, 1882

Dytiscinae 68. Eretes sticticus (Linnaeus, 1767) 69. Hydaticus leander (Rossi, 1790)* 70. Hydaticus seminiger (De Geer , 1774)* 7 1. Graphoderus cinereus (Linnaeus, 1758) 72. Acilius sulcatus (Linnaeus, 1758)* 73. Dytiscus circumjZexus Fabricius, 1801 74. Dytiscus murginalis Linnaeus, 1758 75. Dytiscus pisanus Castelnau, 1 834 76. Dytiscus semisulcatus Miiller, 1776 77. Cybister lateralimurginalis (De Geer , 1774)

Aga.didy Aga.gutt Aga. bigu Aga. bipu Aga. chal Aga. lapp Aga. cons Aga. nebu Aga. labi lly . meri Rha. sutu Col. fusc Me1 . cori

Lap. hy a1 Lap. minu Lap.pont

Ere.stic Hyc. lean Hyc. semi Grh. cine Aci.sulc Dyt.cirf Dy t . marg Dy t . pisa Dyt . semi Cyb. late

*Some specimens were from the Natural History Museum, London.

Table 2. Variables used in the study.

Quantitative Body length, from apex of elytra to front edge of the

pronotum Maximum body width (= maximum width of the elytra) Distance between level of MW and apex of elytra Maximum body height, lateral view Distance between level of MH and apex of elytra Length of head from medium line between posterior side of

eyes to clypeus Length of pronotum medially Maximum width of pronotum Maximum elytral height, lateral view Distance between end of metacoxal process and apex of

elytra Length of metafemur (to medium line of metacoxal process) Maximum width of metafemur Length of metatibia Length of metatarsus, claws excluded

Categorical Angle between head and pronotum (three discrete values:

1, no angle; 2, moderate; 3, pronounced) Angle between pronotum and elytra (same values as HA) Length of swimming hairs of metatarsus (two discrete values:

1, poorly developed; 2, well developed)

only the central area of the screen was used. The actual mag- nification of the image on the screen was calibrated with an objective micrometer during each session against the range of magnifications used in the study. The precision of the measurements on the screen was f 1 mm, which corres- ponded to different real lengths depending upon the magnifi- cation used, f 0.07 rnrn with the lowest (7 x), and f 0.01 mm with the highest (45 x ).

The orientation of the specimens was another important factor to be considered, because most of the measurements were maximum distances and small changes in orientation could lead to very different values. The measurements were taken when the two distal points of the object being measured were focused simultaneously. In Table A 1, for all variables the average value for each species is given.

Transformation of the variables A previous study on sexual dimorphism and geographic variability in size and shape showed that although there were some significant differences in several species, intraspecific differences were always less than interspecific differences, even among congeneric species (Ribera 1992). Therefore, in this study the mean value for each species was used for all the variables.

The distribution of the raw data (given in Table Al) was log-normal. Decimal logarithms were used to normalize the data. The ratios were obtained by the transformation

Ribera and Nilsson

log(measure) - log TL (Mosiman and James 1979; Gilbert and Owen 1990), TL being the longest measure for all the species. In previous studies the transformation used was (measure)/TL or (measure)/(TL + HL), where HL is head length (Ribera and Isart 1991; Ribera 1992). The results were virtually identical, but it seems that the log-transformed ratio overcomes some of the problems with ratios (Mosiman and James 1979; Ricklefs and Travis 1980). Both the logarithms of the raw data and the ratios had a normal distribution (Ribera 1992).

The use of different forms of ratios is quite common, but has been severely criticized (e.g., Atchley et al. 1976; Atchley and Anderson 1978; Humphries et al. 198 1 ; Jackson and Somers 199 1). However, Corruccini (1977), Oxnard (1978), Daly (1985), James and McCulloch (1990) and Marcus (1990), among others, considered that they could be useful for exploratory rather than demonstrative purposes in biological studies, or to answer specific questions.

Multivariate techniques Two main multivariate techniques were used: factor analysis using the correlation matrix of the ratios obtained with the quantitative variables, and cluster analysis (average linkage between groups, UPGMA, method using the squared eucli- dean distance) of the ratios and categorical variables. The extraction method for the factor analysis was the maximum likelihood estimation. No rotation method was used with the factors extracted, to avoid difficulties associated with its justification. The species scores used in Figs. 2 and 3 (Table Al) were computed using the factor score coefficient matrix obtained in the analysis. All the procedures were computed using the statistics package SPSSIPC + .

Results

Allometric relationships among variables For all the quantitative variables the correlation coefficients with the longest measure (log TL) or the length of the femora (log FL) were very high (P < 0.01) (Table 3). Although all the slopes of the regressions were close to 1 (which could be interpreted as indicative of general geometric similarity among the species through the full size range), most were either slightly above or below this value (see the 95% confi- dence intervals in Table 3). The larger species tended to be flattened, with the maximum width in the rear part of the body (slopes of the regressions of log MH, log EH, and log DW with log TL less than l), and the maximum height in the front part (slope of log DH greater than 1). The femora were broader and the tarsi longer than in the smaller species (slopes of the regressions of log FW and log RL with log TL and log FL greater than 1 ; Table 3). The larger species also had a smaller pronotum and head (slope of the regressions of log HL and log PL with log TL less than 1) and the insertion point of the hind legs in a more anterior position (slope of log DM with log TL greater than 1).

In the scatter plots of these regressions three main size groups were observed, the smallest species being the most abundant and variable (Fig. 1). These three groups corres- ponded quite well to the three dytiscid subfamilies Hydro- porinae, Colymbetinae, and Dytiscinae. The gap was bigger between the Hydroporinae and the other two subfamilies, which showed some overlap in size.

Table 3. Allometric regressions between variables.

Y X r* a b SE of b 95% CI for b

Note: Allometric equation: log Y = a + b log X.

Some of the species were wider (Fig. la) and higher (Fig. 1 b) than expected (Hyphydrus aubei Ganglbauer, Hygrotus inaequalis (Fabricus), Yola bicarinata (Latreille), and the genus Hydrovatus). Acilius sulcatus (L.) and Cybis- ter lateralimurginalis (De Geer) were wider, but not higher, while Hygrobia hemuznni (Fabricus) was higher but not wider. Most of the former species also had longer femora than expected (Fig. lc). The shape of the femora of H. hermunni was like those of Hydroporinae, long and thin (Figs. l c and 1 f ) , whereas the genera Luccophilus and Noterus had femora more similar to those of Dytiscinae and Colymbetinae, short and broad (see outlines in Figs. 3 and 4). However, while the species of Luccophilus had shorter tibiae and longer tarsi than expected according to their size (relative to the length of the femur; Figs. Id, le), Noterus species had average dimensions for these two variables. Some species of Dytis- cinae had shorter tibiae (Fig. Id), Eretes sticticus (L.) longer tarsi (Fig. le), and C. lateralimurginalis wider femora than expected (Fig. If).

Factor analysis of the ratios and the categorical variables The correlations between the ratios were lower than those among the raw data; the highest were around 0.70-0.80, with many lower than 0.40 (Table 4). However, the Kaiser- Meyer-Olkin test (used to assess the adequacy of the data for the analysis) gave a value of 0.66, which was considered acceptable (Norusis 1986). Bartlett's test, used to assess the sphericality of the correlation matrix, was also highly significant (P < 0.00 1). High values (P > 0.05) would have meant that there were no significant correlations between variables and the possible factors extracted would have been uninterpretable.

The first three factors accounted for 24.1, 23.7, and 16.5 % of the total variance, respectively, with corresponding eigenvalues of 3.13, 3.07, and 2.14 (the sum of the eigenvalues of all the factors is 13, the number of variables in the analysis). As these percentages were relatively low (64.2% in total), the analysis must be interpreted with cau- tion, although it was still useful as an exploratory tool. Some analyses had been carried out previously to see if excluding

2348 Can. J. Zool. Vol. 73, 1995

Fig. 1. Allometric relationships between variables. (a) Logarithm of body length (log TL) vs. logarithm of maximum body width (log MW). (6) Log TL vs. logarithm of maximum body height (log MH). (c) Log TL vs. logarithm of femur length (log FL). (d) Log FL vs. logarithm of the tibia length (log BL). (e) Log FL vs. logarithm of the tarsus length (log RL). (f) Log FL vs. logarithm of femur width (log FW). Original measures are given in millimetres x lo- ' . *, Hygrobia hermanni; +, Noteridae and Laccophilinae; A , Hydroporinae; 0, Colymbetinae and Copelatinae; m, Dytiscinae. "Hyv," "Lap," and "Not" refer to the genera ~idrovatus, Lnccophilus, and Noterus, respectively. For other species codes see Table 1; note that generic names are abbreviated to the first letter

A

1 . 8 ~ 1 I 1 1 1 1 1

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3 log(body length)

I l . 6 ! I I I 1 I 1 1 4 .6 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6

log(body length)

1.24 I I I I , I I 6 1.6 1.8 2 2.2 2.4 2.6 2.8 3

log (femur length)

1.6.

.F2 A *A R.bima

A A~ Y. bica

A

1 1 a 1 1 I 1

A 1 . ~ 1 9 Hyv A A

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3. log(body length)

log(femur length) log(femur length)

Ribera and Nilsson 2349

Fig. 2. Representation of average species scores for factors 1 and 2 in the factor analysis of the ratios. For the identification numbers of the species see Table 1 . The profiles of the species were traced from photographs.

FACTOR 2

GROUP 2

8

-2.1 1 I I I I I I I I I I I I I I I I -3.6 -3.3 -3.0 -2.7 -2.4 -2.1 -1.8 -1.5 -1.2 -0.9 -0.6 -0.3 -0.0 0.3 0.6 0.9 1.2 1!5 118 2 1

FACTOR 1

apparently redundant variables (i.e., those with high correlations in Table 4) would increase the variance accounted for by the factors. However, none of the combinations tested significantly improved these percentages or altered the main morphometric patterns resulting from the analysis, so the whole set was used, mainly to avoid ad hoc manipulations of the data. Only three factors were considered, owing to the small amount of total variance associated with the others and the difficulty of their biological interpretation.

General trends associated with individual factors Some of the variables had high loadings for more than one factor (Table 5), but for a clearer exposition we first consider them separately. The first factor was mostly associated with the ratios of the pronotum length (rPL), position of the maximum width and height (rDW and rDH, the latter with a negative loading), elytron height (rEH), head length (rHL), and insertion point of the hind legs (rDM, also with a negative loading) (Table 5).

Species at the positive extreme for this factor had a relatively longer pronotum and head (high values of rPL and rHL), the maximum width in a more anterior position (high rDW), the maximum height in a more posterior position (low rDH),

and higher elytra (high rEH) (Figs. 2 and 3). In contrast, species at the negative extreme had a relatively shorter head and pronotum, the maximum width in the rear part of the body, the maximum height in the front part of the body, and more flattened elytra.

The ratios of the lengths of the tibiae and tarsi (rBL and rRL, with a negative loading) and the width of the pronotum (rPW), although mainly linked to the third factor, also had relatively high loadings for factor 1 (Table 5). Species at the positive extreme tended to have longer tibiae but shorter tarsi and a wider pronotum than species at the negative extreme (Figs. 2 and 3). The different loadings of the ratios of the pronotum and body widths for this factor (rPW and rMW) were related to the position of the maximum width: species wider anteriorly (at the positive extreme) had a relatively wider pronotum than species wider posteriorly (at the negative extreme), whatever the maximum width of the body. The ratio of the insertion point of the hind legs (rDM) had a more anterior position in this group of species (negative loadings for the three factors). This was probably due to the general elongation of the body in the larger and more flattened species.

It is interesting to note the difference between the ratios

2350 Can. J. Zool. Vol. 73, 1995

Fig. 3. Representation of average species scores for factors 1 and 3 in the factor analysis of the ratios. For the identification numbers of the species see Table 1 . The profiles of the species were traced from photographs.

FACTOR 3 n n . I

FACTOR 1

of the maximum height (rMH) and elytron height (rEH). Although both variables were highly correlated (Table 4), the latter had a closer association with this factor. Species in which the difference between these two variables was small (i.e., more dorsoventrally asymmetrical species, such as those of the genus Hydrovatus or Noterus, in which EH was between 61 and 72% of MH) had higher scores for this factor than species in which this difference was greater (i.e, more symmetrical species, such as A. sulcatus, C. lateralimarginalis, or E. sticticus, in which this proportion was reduced to 41-47%; see Figs. 2 and 3, Table Al).

The differences in the ratios of the head and pronotum lengths, the position of DW and DH, and the length of the tarsi were in accordance with the allometric relationships observed before, because of a general trend according to size: the larger species (mainly Dytiscinae) were all located at the negative extreme, whereas the smaller species, in general, had positive scores (although the smallest species had near-zero values).

The largest loadings for factor 2 were found in the ratios of the maximum width (rMW), femur length (rFL), and maximum body height (rMH). Other variables of less impor-

tance, although with relatively high loadings, were the ratios of the elytron height, head length, and the pronotum width (Table 5).

Species at the positive extreme had a spherical shape (high rMW and rMH) and long femora (high rFL). Species at the negative extreme were more flattened and narrower and had shorter femora (Fig. 2). No size trend was associated with this factor, and near-zero values were found in species in all size groups.

Factor 3 was mainly associated with the morphometry of the hind legs: the ratios of the length of the tibiae and tarsi (rBL and rRL, with negative loadings), femur width (rFW), and pronotum width (which in fact had relatively high loadings for all three factors). The ratios of the femur length (with a negative loading) and maximum width were also related to this factor (Table 5).

Species with short and broad femora (low rFL and high rFW), short tibiae and tarsi (low rBL and rRL), and a relatively wider shape (high rMW and rPW) were located at the positive extreme (Fig. 3). Species at the negative extreme had longer and thinner femora, longer tibiae and tarsi, and a narrower shape. This factor was also not clearly related to size.

Ribera and Nilsson

Fig. 4. Cluster of the standardized mean values of the ratios and the categorical variables using the average linkage between groups. For species codes and numbers see Table 1.

Rescaled Distance Cluster Combine

Stn . lepi Stn . opta Stn . epip Bid. goud Aga . cha 1 Pla . macu Aga . gut t Grt . bili Grt . igno Hyl-pusi Hyd. memn net. meri Hyd. nige Hyd. niva Hyd-disc Hyd. pube Hyd. nigr Hyd .tess Hyd. plan Hyd. long Coe . inpr Coe .mark Hyd. palu Hyd. vage Aga . lapp Aga. labi Stt . gris Grt . f lav Grt . vari Hyd. f ove Aga . bigu Der . aube Der . dela Neb - f abr Ore. sanm Neb. cana Ore. davi Neb. depr Sca -hale Bid. minu Der . roes Stt . duod Hyd. marg Der . opat Dyt . marg Dyt . pisa Dyt . cirf Grh. cine Hyc. semi Col. fusc Rha. sutu Me1 . cori Aga . bipu Aga . nebu Ily .meri Aga . cons Hyc. lean Cop. haem Cyb. late Ere. stic Lap-minu Lap. pont Lap. hyal Not. clav Not. laev Aga . brun Aga .didy Rhi . bima Dyt . semi HYV. clyp Hyv . cusp Hgt . inae Yo1 . bica Coe . conf Hyp. aube Aci . sulc Hyb. herm

Groups of species that factor analysis is an ordination method, so the species The spatial distribution of the species in Figs. 2 and 3 were located on a continuum, without boundaries between reflected the joint effect of more than one factor. Different groups. However, the always subjective delimitation of spe- groups could be recognized, although their composition dif- cies groups was useful for describing the main morphometric fered according to the factors considered. It must be stressed patterns among them. The detailed composition of the groups

2352 Can. J. Zool. Vol. 73, 1995

Table 4. Correlation matrix of the ratios used in the factor analysis.

Note: rX is the ratio of the average value of the variable X (rX = log X - log TL).

Table 5. Factor matrix obtained in the factor analysis of the ratios.

Note: Variables are ordered with respect to their importance to each of factors F1- F3. rX is the ratio of the average value of the variable X (rX =

log X - log TL).

was less important than their general outline, and particular species at the boundaries could always be included or excluded without significantly altering the results.

Group 1: This group includes species with negative scores for factor 1 and near-zero or low negative scores for factors 2 and 3 (Figs. 2 and 3). The most negative scores corres- ponded to A. sulcatus, C. lateralimrgimlis, E. sticticus, three of the four Dytiscus species studied, Colymbetesfuscus (L.), Melademu coriacea Castelnau, Rhantus suturalis (McLeay), and Agabus nebulosus (Forster) (Nos. 72, 77, 68, 74 -76, 63, 64, 62, and 59, respectively). Acilius sulcatus (72) had the most extreme characteristics associated with this factor (already described), but it also had moderately positive scores for factors 2 and 3, mainly because of its wider body (Table A1 ) . Graphoderus cinereus (L.) (7 1) and both Hydati- cus species (69, 70), although included in the main group in Fig. 2, had similar highly positive scores for factor 3 but less negative scores for factor 1 than A. sulcatus. Dytiscus

circumjlexus (73), together with Copelatus haemorrhoidalis (Fabricius) (4), had the most negative value for factor 2, i.e., they had a narrower body and had shorter femora than the rest of the species (Table Al; see also Fig. la).

This group included all the largest species (Dytiscinae plus some Colymbetinae) and the medium-sized R. suturalis and A. nebulosus (Table Al).

Group 2: This group includes species with extreme positive scores for factor 2 and near-zero values for factors 1 and 3. These included mainly H. aubei, H. imequalis, Y. bicarinata, Coelambus conjluens (Fabricius) , Oreodytes sanmurkii (C . R . Sahlberg) , and Bidessus goudoti (Castelnau) (Nos. 5, 15, 8, 12, 49 and 9, respectively; Fig. 2). Hyphydrus aubei illustrated the more extreme characteristics for this factor: a globular body, almost spherical, with long femora. In the scatter plots shown in Fig. 1, which depict the logarithms of the raw data, the same characteristics were observed in the species with extreme values (H. aubei, H. inaequalis, and Y. bicarinata). All these species belong to the Hydroporinae, which includes the smallest species, although H. aubei is one of the bigger species of the subfamily (Table Al).

Group 3: This group includes species with negative scores for factor 3, slightly positive scores for factor 1, and slightly negative scores for factor 2 (Figs. 2 and 3). The most extreme scores were displayed by Deronectes spp., Nebrioporus depressus (Fabricius) , Hydroporus memnonius Nicolai, Stictotarsus spp. , B. minutissimus, Coelambus impressopunctatus (Schaller) , and Metaporus meridionalis (Nos. 47, 40-42, 20, 43, 44, 10, 13, and 34, respectively; Fig. 3). They all had long tibiae and tarsi, long, thin femora, a relatively more rectangular and narrow body, and the maximum width and height in a more central position.

Group 4: This heterogeneous and not so well delimited group gathered the species with highly positive scores for factor 3 and positive or slightly negative scores for factor 1: Lacco- philus spp., Noterus spp., Hydrovatus spp, Agabus brunneus (F . ) , and Agabus didymus (Olivier) (Nos. 65 - 67, 1, 2, 6, 7, 51, and 52, respectively; Fig. 3). Hydrovatus and Noterus species had the most positive scores for factor 1. The two genera were clearly separated by factor 2 (Fig. 2),

Ribera and Nilsson

mainly because of the relatively longer femora and higher body of Hydrovatus. Moreover, the more elongated shape of the Noterus species resulted in lower ratios of all the variables with positive high loadings for this factor. Laccophilus and Noterus species had similar scores for factors 2 and 3 but different scores for factor 1. This was due, in addition to the differences in the variables with the highest loadings for this factor, to the relatively longer tibiae (Fig. Id) and shorter tarsi (Fig. le) of both Noterus species. As mentioned above, Noterus species also had higher elytra than Laccophilus species. Agabus brunneus and A. didymus had near-zero scores for factor 2, but were located among the species in this group in Fig. 3 (particularly A. brunneus): they had short tarsi and tibiae and short, broad femora (see also Fig. le).

Group 5: This group comprised Hygrobia hermanni (No. 3). This species, the only European member of the Hygrobiidae, had a relatively isolated position in both Figs. 2 and 3. It had the most negative score for factor 3 but a relatively large positive score for factor 2. The most characteristic feature of H. hermanni was its long, thin femora. This is clearly reflected in the scatter plots in Fig. 1. It also had long tarsi and tibiae in relation to body length, although this is not reflected in Figs. Id and l e because they were plotted with the femur length, already much longer than expected. The body of H. hermanni was also higher than expected (Fig. lb), so both characteristics brought it closer to the globular species in group 2 (Fig. 2). However, both groups were clearly separated by factor 3 (Fig. 3), mainly because of its narrower pronotum, thinner femora, and longer tarsi.

Group 6: Most of the species in this group had low scores for all three factors in both Figs. 2 and 3. These were species with average values for the variables studied, without extreme morphometric features. Species in this group belong to the subfamilies Hydroporinae and Colymbetinae, with small or intermediate body sizes.

Hierarchical classifications Two hierarchical classifications were performed using the UPGMA method, one with the average standardized raw data and the other with the average standardized values of the ratios. In both cases the standardized categorical variables were included.

The cluster with the raw data (not shown) resulted in a classification based mainly on size. The first two divisions gathered the largest species (all included in group 1): the Dytiscinae plus M. coriacea and C. fiscus (the largest Colym- betinae) and H. hermanni. The other groups were divided into three main clusters: (1) small species, mostly included in group 3; (2) medium-sized species, mostly included in group 6; and (3) the smallest species. There was thus a primary classification related to the three main size groups, and a secondary one, superimposed upon the first, related to some morphometric characters.

In contrast, the dendrogram of the ratios (Fig. 4) had a primary hierarchy based on the morphometric characteristics, and superimposed upon it a classification based on size. The first divisions separated H. hermanni and A. sulcatus, two species also well characterized by the factor analysis, and then a cluster gathering the species with the most extreme

scores in group 2 above, plus both Hydrovatus species in a rather isolated subcluster. Two species were then separated in individual clusters: Dytiscus semisulcatus Miiller (mainly because of the marked angle between the pronotum and elytra of the specimen studied; see Table Al) and Rhithro- dytes bimaculatus (Dufour). This species had low scores for the three factors studied, but was well characterized in the scatter plots in Figs. l b and If for its extremely flattened body and thin femora. The other species were subsequently divided into four main groups: (i) species in group 4 (except the genus Hydrovatus); (ii) species in group 1 plus some other medium-sized Colymbetinae and C. haemorrhoidalis; (iii) most of the species in group 3; and (iv) species with medium and small body sizes with less extreme scores in groups 2 and 3, and those in group 6.

The cluster of the species included in group 1 was separated into four groups: first C. lateralimarginalis and E. sticticus and then three subgroups, (i) all the Colymbetinae species included plus C. haemorrhoidalis and Hydaticus leander (Rossi) , (ii) G. cinereus and Hydaticus seminiger (De Geer) , and (iii) the three remaining Dytiscus species. This paralleled in some way the distribution of the species along factor 1 in Figs. 2 and 3.

The last large cluster in Fig. 4 was also further separated into several groups: almost all Hydroporus species were grouped together and with other less modified species of Hydroporinae and Colymbetinae with small or medium body sizes. The genus Stictonectes was also well characterized, and clustered with B. goudoti. These four species were close to the species of group 2 in Fig. 2 (all had a rather globular body shape).

The main effect of introducing the categorical variables was the clustering of most of the species included in group 3, together with other species. In this cluster some of the species in group 3 (as defined in the factor analysis) were excluded: H. memnonius, Coelambus impressopunctatus, M. meridionalis, and Stictotarsus griseostriatus (De Geer). Most of the species included also had negative scores for factor 3 and positive scores for factor 1 (i.e., were close to the species in group 3 in Fig. 3), such as Deronectes aubei (Mulsant), Scamdytes hulensis (F.), Hydroporus marginatus (Duftschmid) , or less negative scores for factor 3, such as Oreodytes davisii (Curtis) and Nebrioporus fabressei (Rkgimbart) . However, one species with a positive score for factor 3 (0. sanmarkii) and one with a negative score for factor 1 (Nebrioporus canaliculatus (Lacordaire)) were also included in that cluster (Nos. 45 and 39; Fig. 3).

In addition to the morphometric characteristics already described for group 3, all these species had in common a noncontinuous outline, and most had also short swimming hairs (see Table Al).

Discussion

Interpretation of the observed morphometric patterns Species in the first group, mainly characterized by their extreme negative scores for factor 1, included those habitually considered to be adapted to fast swimming. The most modified species were A. sulcatus, C. lateralimarginalis, and E. sticticus, although A. sulcatus had some additional extreme features, such as a much higher score for factor 3,

2354 Can. J. Zool. Vol. 73, 1995

Fig. 5. Phylogenetic relationships among the studied species, based on information from Ruhnau (1986), Beutel and Roughley (1 988), Bistrom (1988), Wolfe (1988), Bameul (1989), Burmeister (1990), Alarie (199 I), Nilsson and Hilsenhoff (1991), Nilsson and Angus (1992), Fery and Nilsson (1993), and A.N. Nilsson (unpublished analysis of the Agabini). Each species is assigned to one or two of the following morphological groups (see the text): 1, high-speed swimmers; 2, highly manoeuvrable species; 3, poor swimmers in running water; 4, mostly crawlers in microhabitats lacking open water; 5, "dog paddler"; 6, species with average values for the measured variables. Congeneric species belonging to the same morphological groups were pooled under the generic name.

and was clearly separated from the rest of the group in the cluster analysis. The first two species had the best hydrodynamic properties for fast swimming according to Nachtigall (1974) (see the Introduction and outlines in Figs. 2 and 3). In studies of the swimming performance of selected species of Dytiscidae, Natchtigall (1977) found that A. sulcatus had ,the highest relative velocity. Its swimming method, com- bined with its streamlined body, was also energetically highly efficient. These ,three species prefer open waters and can often be seen actively swimming (unpublished observation).

All group 1 species belong to the Colymbetinae and Dytis- cinae, but although all Dytiscinae were included, some Colymbetinae, such as Platambus muculatus (L.) and most of the Agabus species, particularly A. brunneus and A. didymus, belonged to other groups. In the cluster of the ratios (Fig. 4) these species (except A. sulcatus) were also gathered in the same group. Hydaticus seminiger and G. cinereus were clustered together, and were also grouped in Fig. 3 (with H. leander) close to the species in group 4. The differences between A. sulcatus and these latter species were the extreme rear position of the maximum width and the extreme front position of the maximum height in A. sulcatus, which according to Nachtigall (1974 and 1977) are clear adaptations to high-speed swimming in open waters. Both G. cinereus and H. seminiger were found in temporary fens or ponds among dense tufts of Scirpus and Phragmites (Ribera 1992). Similar species of these genera (as well as some Ilybius species) can be found in habitats in which open water is scarce in terms of both time and space, such as fens, peat bogs, and some pingos (Foster 1993; unpublished observations). This suggests a possible adaptation to crawling for these species. Hydrodynamic bodies could be useful both for high-speed swimming in open waters or for crawling among dense vegetation. Evans (1985) suggested that the basic evolutionary habit in the Noteridae, Hygrobiidae, and Dytiscidae seems to be one of free swimming, upon which later specializations have been superimposed. If this is the case, crawling would be an exaptation (sensu Gould and Vrba 1982).

Species in the second group, mainly defined by their extreme positive scores for factor 2, had the characteristics that are usually considered an adaptation for manoeuvrability in other groups (e.g., fish; Webb 1984): they are species with poor horizontal stability, owing to the lack of thin edges in the elytra and the pronotum and the more spherical shape, but they are able to turn in every direction with similar facil- ity, because the drag is almost the same whatever the angle of attack. Their general shape could be compared to that of some reef fishes, adapted to manoeuvring at slow speed in geometrically complex physical environments (Webb 1984). Hespenheide (1975) also suggested a mutually exclusive adaptation in some birds between flying at high speed with low manoeuvrability (e.g . , swifts) and at low speed with high manoeuvrability (e.g., swallows). The former forage high above vegetation (i.e., in the "open" air), while the later

forage near the top of vegetation or near water (i.e., in more complex environments). Some of the species in this group (such as Hyphydrus aubei or Hygrotus inaequalis) were habitually found in vegetated ditches with steep margins (Ribera 1992), a microhabitat shared by similar species of the same genera in other geographical areas (e.g., Hyphydrus ovatus (L.) and Hygrotus quinquelineatus (Zetterstedt) in Scotland and Ireland; unpublished observations).

All the species of this group belong to the Hydroporinae, but they do not have a common origin (Fig. 5). It seems that the spherical body could be a broad plesiomorphic state in some related genera of the group, such as Hyphydrus or Hygrotus, in which most of the species (not studied in this work) have similar shapes. However, other genera were much more variable (e. g . , only one species of Coelambus among the three studied, C. confluens, was included here), with some species showing convergent morphologies that can be interpreted as adaptations to specific environments or swimming behaviour.

Wolfe and Zimmerman (1 984) related globular shape to greater manoeuvrability in lotic environments. According to our results the latter association seems improbable, because the most globular species preferred stagnant waters. Among the species in this group, only 0. sanmurkii (with a lower score) was clearly associated with lotic waters. It was included in this group of species chiefly because of its globular body and long femora. However, when the variables reflecting the continuity of the outline were introduced into the cluster analysis, this species was reclassified as belonging to group 3. Almost all the species habitually found in running waters were gathered in group 3 or in the corresponding cluster in Fig. 4, or had similar scores in Fig. 3 (e.g., Agabus biguttatus-complex or Agabus guttatus (Paykull)) . In these species the typical adaptations of the group were not well developed (though this does not necessarily imply that they are unadapted to aquatic life or that they are more primitive or less evolved than other species in the group): marked angles between the body segments, long slender hind legs, and in most cases short swimming hairs. They also had good attachment mechanisms (e.g., the claws were often well developed; unpublished observation) and tended to swim in short trajectories close to the substratum, avoiding areas with intense flow (Wolfe and Zimmerman 1984). They resemble other beetles habitually considered poor swimmers but well adapted to aquatic life (e.g., large Elmidae or Dryopidae, such as the genera Macronychus or Pomatinus). The species in this group can be considered poor swimmers inhabiting running waters.

All the species in group 3 had small or medium-sized bodies. Large species were usually found in stagnant waters (or in stream pools, such as M. coriacea) (Ribera 1992). It is probable that in fast streams or rivers, even the best swimmers among the large species cannot cope with the drag forces and the current. This relation between size and water veloc-

Ribera and Nilsson

1,2 Noterus 7 3 Hygrobia hermanni 4

5 (2)

I I I 4 Copelatus haerorrhoidalis (l)(4)

9 Bidessus goudoti

10 Bidessus minutissirus

4 1 11 Hydroglyphus pusillus

I 8 Yola bicarinata 5 Hyphydrus aubei

15 Hygrotus inaequalis -1 12 Coelarbus conf luens

't 13 Coelarbus irpressopunctatus

14 Coelarbus rarklini - 34 Hetaporus reridionalis I - 29-32 Graptodytes

33 Fthithrodytes biraculatus

[Siettitia]

36-38 Stictonectes 6 (2)

48 Oreodytes davisii

49 Oreodytes sanrarkii

I 1 43 Stictotarsus duodecirpustulatus 3

44 Stictotarsus griseostriatus (3)(6)

35 Scarodytes halensis 3 (6)

LC 45-47 Nebrioporus 3 (6)

Platarbus raculatus

Agabus chalconatus

Ilybius reridionalis

Agabus labiatus

Agabus lapponicug

Agabus brunneus

Agabus didymus

Agabus bipustulatus

Agabus nebulosus

Agabus conspersus

Agabus guttatus

Agabus biguttatus-complex

1 63 Colymbetes fuscug

62 Fthantus suturalis

64 Heladera coriacea

65-67 Laccophilus

F 68 Eretes sticticus

69-70 Hydaticus

71 Graphoderus cinereus

72 Acilius sulcatus

1- 73-76 Dytiscus

1 77 Cybister lateralimarginalis


ity has also been recognized by Bournaud et al. (1992), based on ecological data from the literature concerning the group.

Bidessus minutissimus was placed among the poor swimmers inhabiting running waters, both in the factor and the cluster analysis (Figs. 3 and 4). It was normally found among gravel or cobbles in streams, but has also been collected when water was pumped from river edges (C. Hernando, personal communication). It belongs to the Bidessini , a monophyletic group as defined by Bistrom (1988) (Fig. 5). The other three species of Bidessini studied were considered to be more or less specialized in manoeuvrability (Y. bicarinata and B. goudoti) or less modified (Hydroglyphus pusillus (Fabricius)). These species were the smallest among those studied (Table Al). It is doubtful that a streamlined body would be of any use to them in open waters, owing to their low Reynolds' numbers and the small drag forces involved. However, in other physical environments such as gravel or dense vegetation, the observed morphometric differences could have some adaptive value.

Hygrobia hermnni was placed near the species in group 2 in Fig. 2, but had the most extreme score for factor 3. The marked angle between its head and pronotum, with the long neck that gives it its characteristic appearance (it was the only species with a value of 3 for the variable HA), con- tributed to its isolated position in the cluster analysis. Because of its less developed swimming method, too, it is considered a poor swimmer. This species has a peculiar life-style. The larvae eat exclusively Tubificidae, and the typical habitats of the adults in the area were those in which these Oligochaeta were abundant: ponds, usually with turbid water and a silt bottom, without fish (Ribera 1992). In laboratory experi- ments, Balfour-Browne (1922) found that adult H. hermnni were unable to capture fast-swimming Ephemeroptera larvae, their prey being only Oligochaeta or very inactive insect larvae. This species could be considered a poor swimmer in stagnant waters, where, because of the turbidity, the ability to manoeuvre seems to provide a useful swimming strategy for locating and handling its rather sessile prey.

The species in the fourth group were mainly characterized by their highly positive scores for factor 3 and their positive or slightly negative scores for factor 1. The two Noterus species studied live by burrowing among detritus and vegetation rafts on the bottom (Balfour-Browne and Balfour-Browne 1940; Galewski 1971). Agabus brunneus, with a similar shape, was found mainly at the edge of small streams, among decaying leaves or under stones (Ribera 1992). It is interesting to note that the sister-group of Hydrovatus is the Methlini (Wolfe 1988), represented in the Iberian Peninsula by Methles cribratellus (Fairmaire). This species is a very poor swimmer, found among detritus or at the base of aquatic vegetation, very close to the bottom or near the edges (unpublished observations). Wolfe (1988) suggested that the acutely pointed and sclerotized apex of the abdomen of Methles species could be used to puncture aquatic plants to obtain trapped air, so these beetles would not have to regularly reach the surface to breathe. Both Hydrovatus species studied are also poor swimmers, usually found among dense vegetation in fens or ponds. All these considerations suggest that these species are rather poor swimmers adapted to crawling among detritus and rotten or dense vegetation. However, it must also be noted that A. didymus, Luccophilus minutus (L.), and L. hyalinus (De Geer), which share some morphological characters, had

wider ecological valencies and can hardly be considered poor swimmers. Agabus didymus and A. brunneus were the only species belonging to the brunneus-group included in the study, so their systematic relatedness has to be considered when interpreting their morphometric features (see below).

Another possible adaptation to a particular environment is that of R. bimculatus, well characterized by the cluster analysis and the scatter plots shown in Fig. 1. It is related to Siettitia (a phreatic genus) (Bameul 1989) and has semi- subterranean habits, having sometimes been collected in caves (Lagar 1974). It has possible shape adaptations to an interstitial microhabitat: a very flattened body, with the hind legs apparently not modified for swimming, and the maximum width in the front part of the body.

Systematic and phylogenetic relationships among groups There are obvious relations between systematics and morphometry in the ordination and classification of the species studied, although the variables measured were chosen with the intention of reflecting adaptive characters, not taxonomic differences between species. Some of the groups of species correspond broadly to the major taxonomic divisions: all the Dytiscinae and some Colymbetinae were included among the fast swimmers (although other species, mainly Colym- betinae, were included among other groups), and the poor swimmers and species with good manoeuvrability were almost exclusively Hydroporinae (Fig. 5). The greater variability among the smaller species could be due to their greater number and complex systematics, but may also occur for biomechanical reasons: the physical constraints of the environment become more severe as size increases (drag forces, Reynolds' number, etc.).

The importance of phylogeny in interpreting size and shape has been widely recognized, and is one of the drawbacks of ecomorphological studies (e.g . , Dobson 1985 ; Miles and Dunham 1993). Only with detailed knowledge of the phylogeny of a group is it possible to distinguish between evolutionary convergence and homologous characters. The existence of convergent characters is an indication of their adaptive value, whereas homologous ones do not allow direct adaptive inter- pretations. Size and shape are habitually considered to be subjected to fast, strong selection (Gould 1971), something that would contribute to the difficulty of this distinction.

Broad ecomorphological studies, with interfamilial com- parisons, are generally greatly biased by phylogeny (Douglas and Matthews 1992). In this study, although three families were included, most of the species belonged to the Dytiscidae. Each of the three families is habitually considered to be monophyletic, and among the Dytiscidae, the four subfamilies are also considered to be monophyletic (Ruhnau 1986; Wolfe 1988; Nilsson and Larson 1990; Alarie 1991) (Fig. 5). The three non-dytiscid species were well characterized by some of their morphometric features, especially those referring to the hind legs, although, depending on the characters studied, they were associated with other species of Dytiscidae.

The morphometric features depicted here can be considered homoplastic a number of times within the studied families of the Hydradephaga (summarized in Fig. 5). This strongly adds to their value, in terms of swimming behaviour and adequacy, in adaptation the physical structure of the microhabitat. However, it must also be considered that there are many factors other than the swimming strategy which

Ribera and Nilsson

determine the morphometry of the group. In some species the femora are modified to achieve different functions, such as in the genus Laccophilus, which has a stridulatory apparatus between the metacoxa and the femur. The species of this genus also use their hind legs to "jump" when disturbed or placed backwards. The hind legs are habitually used for purposes other than swimming, such as cleaning the back of the elytra or spreading glandular secretions. Perhaps in some cases the association between a slender third pair of legs and a globular body is simply a mechanical necessity for achiev- ing access to the surface of the elytra. Although individual explanations could be found for the results obtained for some of the species, the aim of this work was to describe the patterns among the group as a whole.

Acknowledgements

We are greatly indebted to G.N. Foster (Ayr) and two anony- mous reviewers for their most useful comments on the original manuscript. We also thank J.A. Rkgil, M.A. Valle, and J. Isart for their help during this study. This research was partly funded by a grant to I.R. at the Centro de Investi- gaci6n y Desarrollo de Barcelona, Consejo Superior de Investigaciones Cientificas.

References

Alarie, Y. 1991. Primary setae and pores on the cephalic capsule and head appendages of larval Hydroporinae (Coleoptera: Dytis- cidae: Hydroporinae). Can. J. Zool. 69: 2255 - 2265.

Atchley, W .R., and Anderson, D. 1978. Ratios and the statistical analysis of biological data. Syst. Zool. 27: 7 1 -77.

Atchley, W .R., Gaskins, C.R., and Anderson, D. 1976. Statistical properties of ratios. I. Empirical results. Syst. Zool. 25: 137- 148.

Balfour-Browne, F. 1922. The life-history of the water beetle Pelobius tardus Herbst. Proc. Zool. Soc. Lond. 1922. 79 - 97.

Balfour-Browne, F. 1940. British water beetles. Vol. 1. The Ray Society, London.

Balfour-Browne, F. 1959. British water beetles. Vol. 2. The Ray Society, London.

Balfour-Browne, F., and Balfour-Browne, J. 1940. An outline of the habits of the water-beetle, Noterus capricornis Herbst (Coleopt.). Proc. R. Entomol. Soc. Lond. Ser. A, 15: 105 - 112.

Bameul, F. 1989. Description de Rhithrodytes, nouveau genre d'Hydroporinae d'Europe et d'Afrique du Nord: analyse phylogknktique et biogdographie (Coleoptera: Dytiscidae). Ann. Soc. Entomol. Fr. 25: 481 -503.

Beutel, R.G., and Roughley, R.E. 1988. On the systematic position of the family Gyrinidae (Coleoptera: Adephaga). Z. Zool. Syst. Evolutionsforsch. 26: 380 -400.

Bilton, D.T. 1993. A morphometric study of the diving beetle Hydroporus glabriusculus (Coleoptera, Dytiscidae) in Western Europe, including a comparison of morphological and genetic divergence patterns. Zool . Anz. 231 : 1 1 1 - 124.

Bistrom, 0 . 1988. Generic review of the Bidessini (Col., Dytisci- dae). Acta Zool. Fenn. 182: 1-41.

Blake, R.W. 1986. Hydrodynamics of swimming in the water boat- man Cenocorixa bi$da. Can. J. Zool. 64: 1606 - 16 13.

Bournaud, M., Richoux, P., and Usseglio-Polatera, P. 1992. An approach to the synthesis of qualitative ecological information from aquatic Coleoptera communities. Regul. Rivers Res. Manage. 7: 165 - 180.

Burmeister, E.-G. 1990. On the systematic position of Amphizoi- dae, emphasizing features of the female genital organs (Insecta: Coleoptera: Adephaga) . Quaest . Entomol . 26: 245 - 272.

Cooper, S.D., Smith, D.W., and Bence, J.R. 1985. Prey selection by freshwater predators with different foraging strategies. Can. J. Fish. Aquat. Sci. 42: 1720-1732.

Corruccini, R. S. 1977. Correlation properties of morphometric ratios. Syst. Zool. 26: 21 1-214.

Daly , H.V. 1985. Insect morphometrics. Annu. Rev. Entomol. 30: 415 -438.

Daniel, T .L. 1984. Unsteady aspects of aquatic locomotion. Am. Zool. 24: 121 -134.

Dobson, F. S. 1985. The use of phylogeny in behavior and ecology. Evolution, 39: 1384- 1388.

Douglas, M.E., and Matthews, W.J. 1992. Does morphology predict ecology? Hypothesis testing within a freshwater stream fish assemblage. Oikos, 65: 2 13 - 224.

Evans, M .E.G. 1982. Early evolution of the Adephaga-some locomotor speculations. Coleopt . Bull. 36: 597 - 607.

Evans, M . E. G. 1985. Hydradephagan comparative morphology and evolution: some locomotor features and their possible phylogenetic implications. Proc. Acad. Nat. Sci. Phila. 137: 172-181.

Fery, H., and Nilsson, A.N. 1993. A revision of the Agabus chalconatus- and erichsoni- groups (Coleoptera: Dytiscidae), with a proposed phylogeny. Entomol. Scand. 24: 79- 109.

Foster, G.N. 1993. Pingo fens, water beetles and site evaluation. Antenna, 17: 184 - 195.

Franciscolo, M.E. 1979. Fauna d'Italia. Vol. XIV. Coleoptera Haliplidae, Hygrobiidae, Gyrinidae, Dytiscidae. Edizioni Calderini, Bologna, Italy.

Galewski, K. 1971. A study on morphobiotic adaptations of Euro- pean species of the Dytiscidae (Coleoptera) . Pol. Pismo Entomol. 41: 487 -702.

Gilbert, F.S., and Owen, J. 1990. Size, shape, competition, and community structure in hoverflies (Diptera: Syrphidae). J. Anim. Ecol. 59: 21 -39.

Gould, S.J. 197 1. Geometric similarity in allometric growth: a contribution to the problem of scaling in the evolution of size. Am. Nat. 105: 113-132.

Gould, S.J., and Vrba, E.S. 1982. Exaptation-a missing term in the science of form. Paleobiology, 8: 4- 15.

Guignot, F. 193 1 - 1933. Les Hydrocanthares de France. Miscellanea Entomologica, Toulouse, France.

Hespenheide, H. A. 1975. Prey characteristics and predator niche width. In Ecology and evolution of communities. Edited by M.L. Cody and J.N. Diamont. Harvard University Press, Cambridge, Mass. pp. 158 - 180.

Hughes, G.M. 1958. The co-ordination of insect movements. 11. Swimming in Dytiscus, Hydrophilus and a dragonfly nymph. Exp. Biol. 35: 567-583.

Humphries, J.H., Bookstein, F.L., Chernoff, B., Smith, G.R., Elder, R.L., and Poss, S.G. 198 1. Multivariate discrimination by shape in relation to size. Syst. Zool. 30: 291 -308.

Jackson, D.A., and Somers, K.M. 1991. The spectre of "spurious" correlations. Oecologia, 86: 147 - 15 1.

James, F.C., and McCulloch, C.E. 1990. Multivariate analysis in ecology and systematics: panacea or Pandora's box? Annu. Rev. Ecol. Syst. 21: 129-166.

Lagar, A. 1974. Estudio morfologico del Graptodytes bimaculatus (Dufour) (Col. : Dytiscidae). Misc. Zool. 3: 37 -40.

Lawrence, J.F., and Newton, A.F. 1982. Evolution and classification of beetles. Annu. Rev. Ecol. Syst. 13: 261 -290.

Marcus, L.F. 1990. Traditional morphometrics. In Proceedings of the Michigan Morphometrics Workshop, 16-28 May 1988, Ann Arbor. Edited by F.J. Rohlf and F.L. Bookstein. Spec. Publ. No. 2, University of Michigan Museum of Zoology, Ann Arbor. pp. 77- 122.

Miles, D.B., and Dunham, A.E. 1993. Historical perspectives in ecology and evolutionary biology: the use of phylogenetic comparative analysis. Annu. Rev. Ecol. Syst. 24: 587 -619.

Can. J. Zool. Vol. 73. 1995

Mosiman, J.E., and James, F.C. 1979. New statistical methods for allometry with application to Florida Red-winged Blackbirds. Evolution, 33: 444 -459.

Nachtigall, W. 1974. Locomotion: mechanics and hydrodynamics of swimming in aquatic insects. In The physiology of Insecta. Edited by M. Rockstein. Vol. 3. 2nd ed. Academic Press, New York. pp. 381 -432.

Nachtigall, W. 1977. Swimming mechanics and energetics of locomotion of variously sized water-beetles Dytiscidae, body length 2 to 35 mm. In Scale effects in animal locomotion. Edited by T. J. Pedley . Academic Press, London. pp. 269 - 283.

Nichols, S.W. 1985. Omophron and the origin of Hydradephaga (Insecta: Coleoptera Adephaga). Proc. Acad. Nat. Sci. Phila. 137: 182-201.

Nilsson, A.N., and Angus, R.B. 1992. A reclassification of the Deronectes-group of genera (Coleoptera: Dytiscidae) based on a phylogenetic study. Entomol. Scand. 23: 275 - 288.

Nilsson, A.N., and Hilsenhoff, W.L. 1991. Review of the first- instar larvae of Colymbetini (Coleoptera: Dytiscidae), with a key to genera and phylogenetic analysis. Entomol. Scand. 22: 35 -44.

Nilsson, A.N., and Larson, D.J. 1990. A review of the Agabus afinis group (Coleoptera: Dytiscidae), with the description of a new species from Siberia and a proposed phylogeny. Syst. Entomol. 15: 227 - 239.

Norusis, M.J. 1986. SPSSIPC + for the IBM PCIXTIAT. SPSS Inc., Chicago.

Oxnard, C. E. 1978. One biologist's view of morphometrics. Annu. Rev. Ecol. Syst. 9: 219-241.

Reyment, R.H., Blackith, R. E., and Campbell, N. A. 1984. Mul- tivariate morphometrics. 2nd ed. Academic Press, New York.

Ribera, I. 1992. Estudio de 10s Hydradephaga (Coleoptera) del Pirineo y Prepirineo: morfometria y ecologia. Col.lecci6 de Tesis Doctorals Microfitxades No. 1729, Publications Universitat de Barcelona, Barcelona, Spain.

Ribera, I., and Isart, J. 199 1 . Morphometric study of the Dytiscidae (Coleoptera, Adephaga) from the Pyrenaic and Prepyrenaic mountains. In Advances in coleopterology . Edited by M . Zunino, X. Belles, and M. Blas. European Association of Coleopterology , Turin, Italy. pp. 233 -248.

Ricklefs, R.E., and Travis, J. 1980. A morphological approach to the study of avian community organization. Auk, 97: 321 -338.

Roughley , R. E. 198 1 . Trachypachidae and Hydradephaga (Coleop- tera): a monophyletic unit? Pan-Pac. Entomol. 57: 273 -285.

Ruhnau, S. 1986. Phylogenetic relations within the Hydradephaga (Coleoptera) using larval and pupal characters. Entomol. Basil. 11: 231-271.

Schmidt-Nielsen, K. 1972. Locomotion: energy cost of swimming, flying and running. Science (Washington, D.C.), 177: 222 -228.

Shirt, D.B., and Angus, R.B. 1992. A revision of the Nearctic water beetles related to Potamonectes depressus (Fabricius) (Coleoptera, Dytiscidae). Coleopt. Bull. 46: 109 - 14 1 .

Webb, P. W. 1984. Form and function in fish swimming. Sci. Am. 251: 58-68.

Wolfe, G.W. 1988. A phylogenetic investigation of Hydrovatus, Methlini and other plesiotypic Hydroporines (Coleoptera: Dytis- cidae). Psyche, 95: 327-346.

Wolfe, G.W., and Matta, J.F. 1981. Notes on nomenclature and classification of Hydroporus subgenera with the description of a new genus of Hydroporini (Coleoptera: Dytiscidae). Pan-Pac. Entomol. 57: 149 - 175.

Wolfe, G. W., and Zimmerman, J.R. 1984. Sensilla, punctation, reticulation and body shape in the Hydroporinae (Coleoptera, Dytiscidae). Int. Insect Morphol, Embryol, 13: 373 - 387.

Zimmerman, J .R., and Ludwig, J .A. 1975. Multiple discriminant analysis of geographical variation in the aquatic beetle Rhantus gutticollis (Say) (Dytiscidae). Syst. Zool. 24: 63 - 7 1 .

Ap

pe

nd

ix

Tab

le A

l. M

ean

valu

e of

the

cha

ract

ers

mea

sure

d fo

r ea

ch s

peci

es.

3

Spec

ies

Q

code

N

T

L

MW

D

W

MH

D

H

HL

P

L

PW

EH

D

M

FL

FW

B

L

RL

H

A

PA

SL

F1

F

2 F

3 5 V

) V

)

1.

Not

.cla

v 30

42

0 24

3 25

8 15

6 23

3 64

11

9 22

6 10

6 17

6 12

2 36

90

10

7 1

1 2

2.01

-1

.20

1.40

2

Not

.lae

v H

y b. h

erm

C

op. h

aem

H

yp . a

ube

Hyv

. cly

p H

yv .c

usp

Yo1

. bic

a B

id. g

oud

Bid

.min

u H

yl .p

usi

Coe

.con

f C

oe. i

mpr

C

oe . m

ark

Hyt

. inae

H

yd. l

ong

Hyd

.dis

c H

yd . f

ove

Hyd

. mar

g H

yd. m

emn

Hyd

. nig

e H

yd. n

igr

Hyd

. niv

a H

yd . p

alu

Hyd

.pla

n H

yd.p

ube

Hyd

. tes

s H

yd. v

age

Grt

. bili

G

rt.f

lav

Grt

.igno

G

rt.v

ari

Rhi

. bim

a M

et. m

eri

Sca.

hale

St

n.ep

ip

Stn.

lepi

St

n.op

ta

Der

. aub

e D

er.d

ela

Der

. moe

s

Tab

le A

1 (c

oncl

ude

d).

Spec

ies

code

N

T

L

MW

D

W

MH

D

H

HL

P

L

PW

E

H

DM

F

L

FW

B

L

RL

H

A

PA

SL

F

1 F

2 F

3

Der

. opa

t St

t.duo

d St

t.gri

s N

eb.c

ana

Neb

. fab

r N

eb .d

epr

0re.

davi

O

re. s

anm

Pl

a. m

acu

Aga

. bru

n A

ga . d

idy

Aga

. gut

t A

ga.b

igu

Aga

. bip

u A

ga.c

ha1

Aga

.lapp

A

ga .c

ons

Aga

. neb

u A

ga.la

bi

Ily . m

eri

Rha

.sut

u C

ol . f

usc

Mel

. cor

i L

ap. h

yal

Lap

. min

u L

ap.p

ont

Ere

. sti

c H

yc.le

an

Hyc

. sem

i G

rh.c

ine

Aci

.sul

c D

yt . c

irf

Dyt

.mar

g D

yt.p

isa

Dyt

. sem

i C

yb.la

te

Not

e: F

or s

peci

es c

odes

see

Tab

le 1

and

for

def

initi

ons

of c

hara

cter

s (v

aria

bles

) se

e T

able

2.

All

mea

sure

men

ts a

re g

iven

in

mill

imet

res

x N

is

the

num

ber

of s

peci

men

s m

easu

red;

F

1 - F

3 ar

e av

erag

e sp

ecie

s sc

ores

for

fac

tors

1 - 3

(in

the

fact

or a

naly

sis

of t

he r

atio

s).

morphometric patterns among diving beetles (coleoptera...

Documents