POLISH JOURNAL OF ECOLOGY 56 4 557–568 2008 (Pol. J. Ecol.)

Special paper

Werner ULRICH*, Krzysztof SZPILA

Nicolaus Copernicus University in Toruń, Department of Ecology, Gagarina 9, 87-100 Toruń, Poland e-mail: [email protected] (corresponding author)

BODY SIZE DISTRIBUTIONS OF EASTERN EUROPEAN DIPTERA

ABSTRACT: Weight distributions of Eastern 2006, 2007) and by evolutionary trends to- European Diptera (estimated from 7966 species wards larger or smaller body sizes (Orme body length data compiled from Stackelberg et al. 2002, Smith et al. 2004). Hence, SSDs and Nartshuk 1969, 1970) differ from respective might be tools to link evolutionary processes distributions of Coleoptera and Hymenoptera. to ecological patterns (Etienne and Olff Nematoceran size distributions were predomi- 2004, Ulrich 2006, 2007). nantly right skewed while the Brachycera tended to have symmetric and left skewed distributions. Our current knowledge about the eco- Skeweness of size distributions was for Nema- logical implications of animal body sizes tocera positively and for Brachycera negatively stems mostly from studies of vertebrate taxa correlated with genus mean body weight. Genera (cf. Peters 1983, Calder 1984, Schmidt- of smaller mean body weight were significantly Nielsen 1984, Brown 1995, Kozłowski species richer than larger sized genera. Our find- and Gawelczyk 2002, Smith et al. 2004). ings are consistent with an evolutionary model From this work five major patterns emerged: that assumes body size dependent speciation and - Classical niche based models (Hutchin- extinction rates. son and MacArthur 1959, May 1986) and theoretical work based on fractal geometry KEY WORDS: Diptera, body weight, size ra- (Morse et al. 1985) predict the lower weight tios, speciation classes to be most species rich. The empiri- cal evidence rather points to medium size 1. INTRODUCTION classes as being most diverse and therefore to unimodal humped distributions (Dial and The study of animal species body size Marzluff 1988, Brown 1995, Novotny distributions (SSDs) within an ecological and Kindlmann 1996, Kozłowski and context has recently regained interest after Gawelczyk 2002, Smith et al. 2004, Ul- the notion that there are taxon specific dif- rich 2006, 2007). ferences in SSDs that might be explained by - SSDs (based on log body weight or log differential patterns of speciation and ex- body length) appeared to be considerably tinction (Dial and Marzluff 1988, Allen right skewed with more small than large et al. 1999, Knouft and Page 2003, Ulrich bodied species (small and large is here always

jjournalournal 116.indb6.indb 557557 22009-01-13009-01-13 10:58:5510:58:55 558 Werner Ulrich, Krzysztof Szpila

Table 1. Basic entries of the dipteran database. Numbers of families, genera and species included in the analysis of 35 European superfamilies of Diptera. Classification according to Pape (2004), Ooster- broek (2006), and Yeates et al. (2007). Numbers of Species with body Superfamily Families Genera Species length data Anisopodoidea 2 2 9 8 Asiloidea 5 86 345 341 Axymyioidea 1 1 1 1 Bibionoidea 2 5 33 33 Blephariceroidea 1 3 5 5 Carnoidea 6 61 262 261 Chironomoidea 4 155 457 195 Conopoidea 1 11 49 49 Culicoidea 3 11 81 81 Diopsoidea 4 7 35 35 Empidoidea 5 85 910 902 Ephydroidea 5 51 240 239 Hippoboscoidea 3 12 27 26 Lauxanioidea 2 18 113 113 Muscoidea 5 106 845 804 Nemestrinoidea 2 6 18 18 Nerioidea 1 3 12 12 Oestroidea 5 268 804 801 Opomyzoidea 10 38 677 627 Pachyneuroidea 1 1 1 1 Phoroidea 2 20 361 361 Platypezoidea 1 5 23 23 Psychodoidea 1 9 92 91 Ptychopteroidea 1 1 7 7 Scatopsoidea 2 21 50 49 Sciaroidea 7 226 1206 1133 Sciomyzoidea 6 38 131 130 Sphaeroceroidea 4 23 224 223 Stratiomyoidea 2 21 94 93 Syrphoidea 2 82 460 460 Tabanoidea 2 18 130 128 Tephritoidea 7 91 351 347 Tipuloidea 3 51 363 351 Trichoceroidea 1 2 12 12 Xylophagoidea 2 2 6 6 All 111 1540 8434 7966

jjournalournal 116.indb6.indb 558558 22009-01-13009-01-13 10:58:5610:58:56 Dipteran body weights 559

used with respect to the mean body size of modal symmetric SSD with an insignificant a given taxon) (Gaston and Blackburn skewness γ = 0.04. The present paper aims 2000, Kozłowski and Gawelczyk 2002, therefore at examining the body size distri- Smith et al. 2004). This pattern is frequently butions of the eastern European Diptera. We explained in terms of intra- (Kozłowski and compiled data from the systematic work of Weiner 1997) and interspecific (Brown et Stackelberg and Nartshuk (1969, 1970) al. 1993) body size optimization or body size comprising 8434 species of Diptera from dependent speciation and extinction rates the European part of the former USSR. This (McKinney 1990, Maurer et al. 1992). work is still the most comprehensive catalog - Mammal size distributions become of Diptera containing body size data. From more symmetrically distributed at small a comparison with similar work on Euro- geographic scales (Bakker and Kelt 2000). pean Hymenoptera (Ulrich 2006), Coleop- Such a pattern implies a selective species as- tera (Ulrich 2006), and world vertebrates sembly caused either by an accumulation of (Smith et al. 2004) we will show that taxon larger species at these scales or by a selective specific size distributions and respective evo- loss of smaller species. lutionary trends exist and that we have to be - The degree to which SSDs are skewed cautious with generalizations of patterns ob- appears to depend on taxonomic level. tained from single taxa. Higher levels were found to have a more pro- nounced skew and therefore a higher pro- 2. MATERIALS AND METHODS portion of small species (Kozłowski and Gawelczyk 2002). This implies a positive The present study is based on the trea- correlation of SSD skew and species richness tise of Eastern European Diptera of Stack- (Ulrich 2006). elberg and Nartshuk (1969, 1970). From - Body size within vertebrate taxa seems this work we compiled a database that con- to be phylogenetically constrained. The tains 8434 species from 1540 genera and 111 study of Smith et al. (2004) on constraints families (Table 1). For 7966 species body on mammalian body size showed that these length data are available (94.4%). The clas- constraints (measured as the coefficient of sification of species above the genus level fol- correlation of congeneric species pairs) are lows Oosterbroek (2006) and Yeates et strongest in medium size classes. al. (2007). The classification into genera is in SSDs of terrestrial invertebrates are much accordance with the Fauna Europea (Pape less studied (Gunnarson 1990, Basset and 2004). The database contains the following Kitching 1991, Novotny and Kindl- taxonomic and morphometric entries: or- mann 1996, Ulrich 2005, 2006, 2007). der, infraorder, suborder, superfamily, fam- Chislenko (1981) published size distributions ily, subfamily, genus, species, minimum, of all major orders and reported for maximum and mean body length, and body nearly all of them symmetric body size distri- weight. butions. Recently, Espadaler and Gomez To ensure comparability to previous (2002) and Ulrich (2006, 2007) published work on insect and vertebrate body size dis- regional SSDs of Iberian ant species and Eu- tributions (Smith et al. 2004, Ulrich 2006, ropean Hymenoptera and Coleoptera, respec- 2007) the present work is based on mean spe- tively. These three studies found hymenopter- cies dry weight W calculated from the arith- an SSDs to differ from and the beetle SSDs to metic mean L of available data on minimum be similar to the vertebrate pattern. and maximum body length using the regres- Beside Coleoptera and Hymenoptera, sion equation of Ganihar (1997) Diptera are the third major holometabolic 2.59 taxon. In Europe about 18,700 species are de- Wmg[ ] 0.032 Lmm [ ] (1) scribed (Pape pers. comm.). Dipteran body size distributions are currently only poorly Of course, in the majority of species the known. Only Chislenko (1981) compiled literature-based mean lengths will only be body length data of the Russian fauna (7331 rough estimates. However, these inaccuracies species). A recalculation of his data gave a bi- are counterbalanced by the large number of

jjournalournal 116.indb6.indb 559559 22009-01-13009-01-13 10:58:5610:58:56 560 Werner Ulrich, Krzysztof Szpila

data points used for the analysis. Body weight the observed mean according to a propor- distributions (in the following the term SSD tional rescaling process (σ2 = μ2) and ac- refers always to the species – body weight cording to a Poisson distribution (σ2 = μ, cf. distribution) are based on ln-transformed Ulrich 2007). It should be emphasized that weights. Skewness γ is computed as in Ul- such a comparison is not a true phyloge- rich (2006, 2007): netic analysis where body size distributions

3 are compared with respect to the underly- n n § w  w · ing phylogenetic tree. Such an analysis is J ¨ i ¸ (n 1)(n  2) ¦¨ V ¸ still impossible due to the low phylogenetic i 1 © w ¹ (2) resolution of the Diptera. To test whether observed body weights are

where wi is the ln-transformed body weight regularly spaced within the observed ranges, of species i and n is the number of species. we computed the ratios of log-transformed

We calculated the standard error of γ accord- body weights (R = ln Wn / ln Wn–1) and raw ing to Tabachnick and Fidell (1996): SE body weights (R = Wn / Wn–1) of subsequent (γ) = (6/n)1/2. species on the rank order of species for all To assess the number of modes of the genera with more than five species (the ra- body weight distributions we used a normal tio test of Strong et al. 1979). We compared kernel density estimator according to Hav- the variance of these ratios with the expecta- licek and Carpenter (2001): tion from a null model, where species body

2 lengths were randomly placed inside the re- 11S §·§·xx spective range (two species were placed at the fx() Expi h ¦ ¨ ¨¸¸ upper and lower observed length or weight, Sh 2S i 1 2 ©¹h ©¹cf. Gotelli and Graves 1996). Respective (3) confidence limits of the null models were ob-

with S being the number of species, xi the tained from 1000 replicates. To compare taxa respective ln transformed body weights, and we calculated the standardized effect size SES h the band width. We used a smooth band- of R as a Z-transformed score [Z = (R–μ)/σ] width according to Silvermann (1986) of to compare the observed index R to the dis- tribution of simulated indices (μ = mean and hS 1.060.2 min(V ; range / 5.36) σ = the standard deviation of the simulated x (4) ratios). SES values below –2.0 or above 2.0 indicate approximate statistical significance with range being the range of ln trans- at the 5% error benchmark. formed body weights. The step width x was Regression slopes refer to reduced major in all cases h/5. Kernel density estimates were axis (RMA) regressions computed with PAST done for all genera with at least five species. (Hammer et al. 2001). Errors are standard To study whether within genus body errors. weights are constrained within upper and lower limits (phylogenetic constraint) we 3. RESULTS followed Smith et al. (2004) and Ulrich (2007) and computed the regressions of ln The eastern European Diptera range from transformed body weights between conge- less than 0.01 mg dry weight in some Leptoc- neric species pairs. For genera having two era spp. (Sphaeroceridae) and Culicoides im- to ten species, all pairs were included; for punctatus (Ceratopogonidae) to nearly 600 larger genera, 20% of all species pairs were mg dry weight in the south-eastern Europe- randomly selected. The coefficient of cor- an Satanas gigas (Asilidae) and about 300 mg relation r is then a measure of how much in the widespread European Tipula maxima body size is constrained within a given tax- (Tipulidae). They span a body weight range on (Smith et al. 2004). We tested r against of more than four orders of magnitude. This two null model approaches and assigned is about one order less than the body weight species body weights within each genus us- range of European Hymenoptera and Cole- ing a normal random distribution around optera (Ulrich 2006, 2007).

jjournalournal 116.indb6.indb 560560 22009-01-13009-01-13 10:58:5610:58:56 Dipteran body weights 561

Fig. 1. Size distributions of all eastern European Diptera and the two suborders and Brachy- cera. Data points denote numbers of species per size class. The number of data points reflects the num- ber of size classes considered. Sizes are given as ln transformed body weights in mg. An asterisk after the skewness (γ) denotes that it is significant at P <0.01.

The body weight distributions of the mean skewness gave only for the Nematocera all Diptera and the suborder Nematocera a weak signal of increasing mean skewness (Fig. 1) were significantly (P(γ = 0) < 0.01) with taxonomic level (P = 0.08). right skewed while the respective Brachycera In none of the superfamilies (Fig. 2), fami- distribution did not show a significant skew. lies and genera (data not shown) with at least Three of the 13 Nematocera superfamilies five species was the lowermost or the upper- had a significant right skewed SSD but none most size class the species richest. Nevertheless of the 22 Brachycera superfamilies. In turn, we observed a significant decrease of species none of the Nematocera superfamilies was richness with increasing log mean body weight significantly left skewed but six of the Brachy- of genera (Fig. 3 C). Using species per genus cera superfamilies. Of the 17 superfamilies ratios (S/G) instead of raw data strengthened with more than 100 species (Fig. 2) five had this trend (Fig. 4). S/G ratios of small bodied significantly left and two significantly right genera were on average three to four times skewed SSDs. higher than S/G ratios of larger sized genera. We observed a similar trend for right Variance – mean ratios of genus body skewed SSDs of Nematocera at the family and lengths were fully consistent with a propor- genus level (Fig. 3A) 17 of the 20 Nematocera tional rescaling pattern as predicted by Tay- families with at least five species had a posi- lor’ power law (Taylor 1960). The RMA tive SSD skewness and only three a negative scaling exponent z = 1.99 ± 0.02 does not (Fig. 3A). Of the Brachycera 34 families had significantly deviate from the predicted value a negative SSD skewness and 31 a positive. A of 2 (Fig. 3D). Both ratio tests did not point to Kruskal-Wallis test revealed a significant dif- aggregated or segregated body size distribu- ference in mean skewness between Brachy- tions within genera (P> 0.5). Further, Z stan- cera and Nematocera (P <0.001). Skew was dardized ratios were neither correlated with not correlated with species richness. Of 107 species richness nor with mean body weight nematoceran genera with at least five spe- of genera. Hence, body weight appeared to be cies 43 had left and 58 right skewed SSDs (six randomly distributed with respect to a linear genera had a skew of γ = 0). Of the respective random null model. 295 brachyceran genera 162 had left skewed The differences in SSD skew between and 131 right skewed SSDs (two times γ = 0) Nematocera and Brachycera might be caused (Fig. 3B). A χ2 contingency test pointed to a by a change of skewness along the body size significant difference in skewness between axis. To infer such a trend we used a slid- Brachycera and Nematocera (P = 0.03). ing window technique as in Ulrich (2006, Again skew was not correlated with species 2007) and calculated mean skewness and richness. mean body weight for a sliding window of The above results might indicate a trend 10 genera having more than five species or- towards pronounced skewness in higher dered from smallest to largest. A significant taxa. Separate Kruskal-Wallis test for Nema- positive correlation (Spearman’ r = 0.49, tocera and Brachycera comparing genera, P <0.0001) between mean skewness per ge- families, and superfamilies with respect to nus and mean ln body weight appeared for

jjournalournal 116.indb6.indb 561561 22009-01-13009-01-13 10:58:5610:58:56 562 Werner Ulrich, Krzysztof Szpila

Fig. 2. Size distributions of the superfamilies of eastern European Diptera having more than 100 species. Data points denote numbers of species per size class. The number of data points reflects the number of size classes considered. Sizes are given as ln transformed body weights in mg. An asterisk after the skewness (γ) denotes that it is significant at P <0.01.

jjournalournal 116.indb6.indb 562562 22009-01-13009-01-13 10:58:5710:58:57 Dipteran body weights 563

Fig. 3. A, B: Skewness of the body size distributions (ln transformed body weights in mg) of dipteran families (A) and genera (B) for all taxa with at least 5 species (S). Given are also the upper and lower 99% confidence limits according to the approximation of Tabachnick and Fidell (1996) (broken lines). Black dots in A refer to Brachycera, open dots to Nematocera. C: Species richness per genus is significantly negatively correlated with ln (mean genus body weight) (r = –0.07; P = 0.006). D: The variance – mean length relationship [σ2 = f(μz)] of central European Diptera has a RMA slope of z = 1.99 ± 0.02 (R2 = 0.92; P <0.0001) (length is used instead of ln body weight to avoid negative values).

Fig. 4. Mean number of species per genus in dependence on mean ln body weight per genus for genera classified into 15 binary body weight classes. The open circle represents the lowest body weight class and is represented by only 9 genera having in total 14 species. It is not included in the regression. Re- gression: slope = –0.74; R2 = 0.82; P <0.001.

jjournalournal 116.indb6.indb 563563 22009-01-13009-01-13 10:58:5710:58:57 564 Werner Ulrich, Krzysztof Szpila Skewness

Mean body weight Skewness

Mean body weight

Fig. 5. Mean skewness of a sliding window of 10 genera shifted along the ranked body weights was for Nematocera (A) significantly positively (Spearman’s r: 0.49 P <0.0001; RMA regression slope: z = 0.25 ± 0.09) and for Brachycera (B) significantly negatively (Spearman’s r: –0.31 P <0.0001; RMA regression slope: z = –0.43 ± 0.08) correlated with mean body weight.

Fig. 6. Regression of body weights of congeneric species pairs for all diperan genera with at least five species. Phylogenetic constraint (coefficient of correlation) r = 0.91; RMA regression slope: 0.99.

jjournalournal 116.indb6.indb 564564 22009-01-13009-01-13 10:58:5710:58:57 Dipteran body weights 565

Nematocera (Fig. 5A). In turn, the mean pected from a random distribution pointing skewness of brachyceran genera showed the to certain phylogenetic constraints on body opposite trend (Fig. 5B) and was significantly size; 4) small bodied genera tended to be spe- negatively correlated with mean genus body cies richer than genera of larger mean size. weight (Spearman’ r = –0.31, P <0.0001). What evolutionary factors shape size Hence, SSDs of large sized nematoceran gen- distributions? McKinney (1990) assumed era had predominately positive and those of that body weight evolution can be modelled brachyceran genera predominately negative by a random walk process and termed this skewness (Fig. 5). the random diffusion model. In such a pro- Body size of Diptera appeared to be high- cess upper or lower boundaries of body size ly constrained when measured by the coef- (modelled as reflecting boundaries) result in ficient of correlation using intragenus pair SSD skews to the opposite direction. If the wise comparisons (Fig. 6). The coefficient boundary model were to apply, the genera of correlation was r = 0.91. The proportional close to these boundaries should exhibit SSD rescaling null model approaches in turn gave skews in opposite directions. In Hymenop- r = 0.47 and the Poisson model returned r = tera Ulrich (2006) found a significant nega- 0.67 (not shown). The observed intragenus tive correlation of skewness with mean body variability in body size is therefore much weight per size class. Smaller sized genera lower than expected from both null models. had mostly negative SSD skews, large sized Of the 394 genera with at least five spe- genera positive skews. Hymenopteran SSDs cies and for which it was possible to deter- are therefore in accordance with the random mine the number of modes 34 had unimodal, diffusion model with upper and lower re- 133 bimodal, 193 trimodal, 32 quadrimodal, flecting boundaries (Ulrich 2006). In turn, and 2 quinquemodal SSDs. The number of central European Coleoptera showed a very modes was significantly negatively correlated different picture (Ulrich 2007). Their skew- with species richness (R2 = 0.30, P <0.001). ness – body weight plots were significantly This correlation was even more pronounced U-shaped with smaller and larger sized gen- at the family level (R2 = 0.67, P <0.001). 40% era tending to have a positive SSD skewness. of the families had unimodal and 30% bi- The present findings for eastern European modal SSDs. SSD skewness was at no taxon Diptera reveal a third pattern (Fig. 5). We level correlated with the number of modes. found suborder dependent negative and positive correlations between skewness and 4. DISCUSSION body size. This pattern is not in accordance with the diffusion model. The present study is the third part of the The above dipteran pattern might be analysis of body size distributions of European in accordance with the hypothesis that a . In two previous studies (Ulrich taxon specific speciation/extinction bias 2006, 2007) one of us found hymenopteran exists (Alroy 2000). According to the bias SSDs to be predominantly unimodal and model differential extinction and speciation symmetrical (whole order γ = –0.04) while rates along the body size axis should result coleopteran SSDs were at least above the in body size dependent SSD skews (Knouft genus level mainly right skewed. The cole- and Page 2003). Particularly higher extinc- opteran pattern resembled the widespread tion rates of larger and lower rates of smaller SSD shapes of vertebrates (Kozłowski and species should generate a trend towards left Gawelczyk 2002, Smith et al. 2004). skewed SSDs at larger body sizes. If this Four major findings emerged from the model were to apply, genera of smaller sized present study concerning the SSDs in east- species should be most species rich. This was ern European Diptera: 1) Nematocera and indeed the case (Fig. 3C). Again this pattern Brachycera differ with regard to the pre- differs from that of the Hymenoptera and dominant shapes of their size distributions; Vertebrata, where genera of medium sized 2), size distributions changed in an ordered species are most species rich (Smith et al. manner along the body size axis; 3), intrage- 2004, Ulrich 2006), but also from that of the nus variability in body size was less than ex- Coleoptera, where no correlation between

jjournalournal 116.indb6.indb 565565 22009-01-13009-01-13 10:58:5810:58:58 566 Werner Ulrich, Krzysztof Szpila

species richness and body size exists (Ul- aries in body size (although higher taxon rich 2007). specific upper and lower boundaries might A surprising result of the present study exist). Other phyla have to be studied to see was the negative correlation of species rich- whether this finding can be generalized. ness per genus (the S/G ratio) with body size In the Hymenoptera and Coleoptera (Fig. 4). Such a pattern does neither occur in within genus variability was found to be the previous studied Hymenopteran and Co- higher in species rich genera but indepen- leoptera (Ulrich 2006, 2007) nor in the Ver- dent of body weight (Ulrich 2006, 2007). tebrata (Smith et al. 2004). Of course, such Hence, body weight distributions in species a pattern might be caused by taxonomists, rich hymenopteran and coleopteran genera who split preferentially large bodied genera appeared to be more clumped than expected into smaller parts. This explanation implies from the null model. The Z-transformed R- fundamental differences in taxonomic prac- values of the ratio test were in both orders tice between coleopteran, hymenopteran, significantly positively correlated with spe- and dipteran taxonomists and seems not cies richness. In the present study we could very likely. In our view, the higher S/G ratios not detect such a pattern for Diptera. Again in smaller bodied genera reflect a real evo- Diptera differed in their size distributions lutionary trend related to a smaller morpho- from the other two large insect orders. logical variability in smaller species. Detailed In summary, the present results again morphological studies have to verify this hy- point to differences in body size distribu- pothesis. tions between major animal taxa. The causes As in the case of Hymenoptera (Ulrich of these differences are not well understood 2006), Coleoptera (Ulrich 2007), and Ver- yet. Hence generalizations of patterns and tebrata (Smith et al. 2004) body size of Dip- models that extrapolate from findings within tera appeared to be highly constrained (r = one taxon should be treated with caution. 0.91; Fig. 6). Both null model approaches gave much lower values of r pointing to a ACKNOWLEDGEMENTS: Miss Hazel Pear- lower variability in body sizes around the son kindly improved our English. Thomas Pape genus means than expected from simple sta- provided information on European species rich- tistical distributions. These and the nearly ness. Paweł Trzciński helped us with the no- identical values for Coleoptera, Hymenop- menclature of Syrphidae. This work was in part tera, and Vertebrata (Ulrich 2006, 2007, supported by grants from the Polish Ministry of Science (KBN, 3 P04F 034 22, 2 P04F 039 29). Smith et al. 2004) imply that the overlap in size between genera is less than expected from proportional rescaling and Poisson null 5. REFERENCES models. The remarkable resemblance of r in- dicates that body sizes of these four very dif- Allen C.R., Forys E.A., Holling, C.S. ferent taxa are constrained in a similar man- 1999 – Body mass patterns predict invasions ner and possibly by similar processes. and extinctions in transforming landscapes As for Hymenoptera and Coleoptera did – Ecosystems, 2: 114–121. Alroy J. 2000 – Understanding the dynamics of the intragenus variability of dipteran body trends within evolving lineages – Paleobiol- sizes resemble a proportional rescaling pro- ogy, 26: 319–329. cess (Fig. 3D). The respective RMA slopes for Bakker V.J., Kelt D.A. 2000 – Scale-depen- Hymenoptera (z = 2.30) and Coleoptera (z = dent patterns in body size distributions of neo- 2.67) are however higher than the dipteran tropical mammals – Ecology, 81: 3530–3547. slope (z = 1.99) and the theoretical expecta- Basset Y., Kitching R.L. 1991 – Species tion of z = 2. Hence, in Coleoptera and Hyme- number, species abundance, and body length noptera, genera of larger bodied species have of arboreal arthropods associated with an an overproportionally larger span of body Australian rainforest tree – Ecol. Entomol. 16: weights than smaller bodied genera. Never- 391–402. theless all three patterns are consistent with Brown J.H. 1995 – Macroecology – Chicago, Univ. Press. speciation/extinction processes without dis- Brown J.H., Marquet P.A., Taper M.L. tinct upper and lower genus specific bound- 1993 – Spatial scaling of species composition:

jjournalournal 116.indb6.indb 566566 22009-01-13009-01-13 10:58:5810:58:58 Dipteran body weights 567

body masses of North American land mam- May R.M. 1986 – The search for patterns in mals – Am. Nat. 142: 573–584. the balance of nature: advances and retreats Calder W.A. 1984 – Size, function, and life his- – Ecology, 67: 1115–1126. tory – Univ. Press., Cambridge. McKinney M.L. 1990 – Trends in body size Chislenko L.L. 1981 – Structure of fauna and evolution (In Evolutionary trends , Ed. K.C. flora as dependent on organismal body size McNamara) – Univ. Arizona press, Tucson, – Univ. Press , Moscow. pp. 75–118. Dial K.P., Marzluff J.M. 1988 – Are the Morse D.R., Lawton J.H., Dodson M.M., smallest organisms the most diverse? – Ecol- Williamson M.H. 1985 – Fractal dimen- ogy, 69: 1620–1624. sion of vegetation and the distribution of ar- Espadaler X., Gomez C. 2002 – The spe- thropod body lengths – Nature, 314: 731–733. cies body-size distribution in Iberian ants Novotny V., Kindlmann P. 1996 – Distribu- is parameter dependent – Vie et Milieu, 52: tion of body sizes in taxa and com- 103–107. munities – Oikos, 75: 75–82. Etienne R.S., Olff H. 2004 – How dispersal Oosterbroek P. 2006 – The European families limitation shapes species – body size distribu- of the Diptera. Identification, diagnosis, biol- tions in local communities – Am. Nat. 163: ogy – KNNV Publishing, Utrecht. 69–83. Orme C.D.L., Quicke D.L.J., Cook J.M., Ganihar S.R. 1997 – Biomass estimates of ter- Purvis A. 2002 – Body size does not pre- restrial arthropods based on body length – dict species richness among the metazoan – J. Biosciences, 22: 219–224. J. Evol. Biol. 15: 235–247. Gaston K.J., Blackburn T.M. 2000 – Pat- Pape T. (Ed.). 2004 – Diptera, . Fauna Eu- tern and process in macroecology – Black- ropaea version 1.1, http:faunaeur.org well, Oxford. Peters R.H. 1983 – The ecological implications Gotelli N.J., Graves G.R. 1996 – Null Mod- of body size – Cambridge, Univ. Press. els in Ecology – Smith. Inst. Press, Washing- Schmidt-Nielsen K. 1984 – Scaling: why is ton D.C. animal size so important? – Univ. Press, Cam- Gunnarsson B. 1990 – Vegetation structure bridge. and the abundance and size distribution Stackelberg A.A., Nartshuk E.P. 1969 (Ed.) of spruce-living spiders – J. Anim. Ecol. 59: – Keys to the of the European part of 743–752. the USSR – Vol V Diptera and Siphonaptera, Hammer O., Harper D.A.T., Ryan P.D. Leningrad, part 1. 2001 – PAST: Palaeontological statistics soft- Stackelberg A.A., Nartshuk E.P. 1970 (Ed) ware package for education and data analysis – Keys to the insects of the European part of – Palaeontologica Electronica, 4: 9 pp. the USSR – Vol V Diptera and Siphonaptera, Havlicek T.D., Carpenter S.R. 2001 – Pe- Leningrad, part 2. lagic species size distributions in lakes: are Silvermann B.W. 1986 – Density estimation they discontinuous? – Limnol. Oceanogr. 46: for statistics and data analysis – Chapman and 1021–1033. Hall, New York. Hutchinson G.E., MacArthur R.H. 1959 Smith F.A. et al. 2004 – Similarity of mam- – A theoretical ecological model of size distri- malian body size across the taxonomic hier- butions among species of – Am. Nat. archy and across space and time – Am. Nat. 93: 117–125. 163: 672–691. Knouft J.H., Page L.M. 2003 – The evolu- Strong D.R., Szyska L.A., Simberloff tion of body size in extant groups of North D.S. 1979 – Tests of community wide char- American freshwater fishes: speciation, size acter displacement against null hypotheses distributions, and Cope’s rule – Am. Nat. 161: – Evolution, 33: 897–913. 413–421. Tabachnick B.G., Fidell L.S. 1996 – Using Kozłowski J., Gawelczyk A.T. 2002 – Why multivariate statistics – 3rd Ed. Harper Col- are species’ body size distributions usually lins, New York. skewed to the right? – Funct. Ecol. 16: 419–432. Taylor L.R. 1960 – Aggregation, variance and Kozłowski J., Weiner J. 1997 – Interspecific the mean – Nature, 332: 721–722. allometries are byproducts of body size opti- Ulrich W. 2005 – Die Hymenopteren einer mization – Am. Nat. 149: 352–380. Wiese auf Kalkgestein: Ökologische Muster Maurer B.A., Brown J.H., Rusler R.D. einer lokalen Tiergemeinschaft – Schriftenr. 1992 – The micro and macro in body size evo- Forschzentr. Waldökosysteme A 195, 203 pp. lution – Evolution, 46: 939–953. (in German).

jjournalournal 116.indb6.indb 567567 22009-01-13009-01-13 10:58:5810:58:58 568 Werner Ulrich, Krzysztof Szpila

Ulrich W. 2006 – Body size distribution of Eu- Yeates D.K., Wiegmann B.M., Courtney ropean Hymenoptera – Oikos, 114: 518–528. G.W., Meier R., Lambkin C., Pape T. Ulrich W. 2007 – Body weight distributions of 2007 – Phylogeny and systematics of Diptera: central European Coleoptera – Eur. J. Ento- Two decades of progress and prospects (In: mol. 104: 769–776. Linnaeus tercentenary: Progress in Inver- tebrate Taxonomy, Eds. Z.–Q. Zhang, W.A. Shear) Zootaxa 1668, pp. 565–590.

Received after revising June 2008

jjournalournal 116.indb6.indb 568568 22009-01-13009-01-13 10:58:5810:58:58