Journal of ADVANCES IN APPLIED ECOLOGICAL TECHNIQUES Applied Ecology 2000, 37, The adequacy of collecting techniques for estimating 884±893 species richness of grassland invertebrates

VALERIE STANDEN Department of Biological Sciences, Durham University, Durham DH1 3LE, UK

Summary 1. There is still considerable debate about the most e€ective methods of sampling invertebrates in monitoring and assessment programmes. 2. The above-ground invertebrates of a limestone grassland in north-east England were compared between samples from pitfall traps and from a D-vac suction trap combined with a lightweight swish net (SW/DV). 3. Over 14 000 individuals were captured, with similar numbers in the pitfall and SW/DV samples. A total of 480 species of Hemiptera, Coleoptera, Diptera and Araneae was identi®ed and placed into 14 taxa for further analysis. 4. The pitfall sample produced species/specimen curves from which it was possible to estimate species richness for all the Coleoptera and Araneae taxa and the calyp- terate . The SW/DV sample was adequate to estimate the species rich- ness of Hemiptera, most Diptera taxa, herbivorous Coleoptera and Linyphiidae. 5. The proportion of Coleoptera and Araneae taxa that were method-unique was higher in the pitfall sample than the SW/DV sample and vice versa for the Hemi- ptera and Diptera taxa. Nevertheless, a relatively high proportion of method- unique species of most taxa was found in both sample types, indicating that they can each contribute to assessing species assemblages in grasslands. 6. Both pitfall traps and SW/DV samples are needed to estimate species richness in grasslands for all taxa except Heteroptera, Homoptera and Lycosidae. Herbivorous Coleoptera and Linyphiidae were collected in numbers adequate for assessing rich- ness in both sample types, but more specimens were required in one or other sam- ple for the remaining taxa.

Key-words: D-vac technique, monitoring, pitfall traps, rarefaction, sampling, swish net. Journal of Applied Ecology (2000) 37, 884±893

Introduction group. Pitfall traps catch more specimens of the lar- ger carabid species and under-represent smaller spe- A comprehensive comparison of invertebrate taxa, cies (Spence & Niemela 1994; Ulber & Wolf- collecting techniques, vegetation morphs and habi- Schwerin 1995); they also catch more spider males tats does not exist, although comparisons between than females and more species of Erigoninae and some variables have been made. For example, Lycosidae than other spider families (Dinter 1995). Kromp et al. (1995) estimated the density of arthro- Pitfall traps captured a di€erent suite of Hemiptera pod groups using emergence traps, pitfall traps and species compared with sweep netting (Cherrill & a ¯ooding method, and found that the size and Sanderson 1994; Danahar 1998). mobility of taxa in¯uenced estimates of density. Although less structurally diverse than some habi- More frequent are comparisons of di€erent collect- tats, grasslands present their own problems for sam- ing methods used to sample a single taxonomic pling invertebrates. Lu€ (1966) demonstrated signi®cant di€erences between the beetle fauna inha- # 2000 British Correspondence: Valerie Standen (e-mail valerie.stan- biting tussocks and the intervening ground. Spiders Ecological Society [email protected]). tend to be either active hunters over the surface of 885 the ground or web-builders inhabiting di€erent were to establish (i) which collecting method(s) V. Standen parts of plants, and so require speci®c sampling allowed an estimate of species richness of 14 methods (Cherrett 1964). Finally, Denno (1994) Coleoptera, Diptera, Hemiptera and Araneae taxa, demonstrated the in¯uence of grassland structure on and (ii) the extent of species overlap between sam- the density and diversity of leafhoppers. pling methods and consequently whether one or The herb-rich grasslands that have developed over more methods are needed to estimate species rich- outcrops of magnesian limestone in south and east ness of each taxon in the grassland habitat. County Durham, UK, have mostly been destroyed by agricultural intensi®cation and the habitat is now nationally scarce. Thrislington, the location of this Methods study, has been considered an important area for

plants and invertebrates since the 19th century. By S I T E D E S C R I P T I O N the middle of the 20th century it supported the most extensive stand of primary magnesian limestone Thrislington grassland in east County Durham (NZ grassland in Britain and 10 ha were designated as a 318 228) is 110 m a.s.l. with an annual rainfall of site of special scienti®c interest in 1984. In the early approximately 500 mm. The calcareous plant com- 1970s, an extension to the adjacent dolomite quarry munity that has developed over thin soils derived threatened to destroy the site entirely but, as a result from the magnesian limestone escarpment contains of public concern, part was left undisturbed and 5Á5 species found in northern arctic±alpine grasslands ha of ancient grassland was translocated as intact and in chalk grasslands of southern England. Blue turves to an adjacent, previously quarried, area moor grass Sesleria albicans Kit. ex Schult. and (Park 1989). Translocation took place over 8 years small scabious Scabiosa columbaria L. characterize beginning in 1982, and following completion the site the site, with over 140 vascular species recorded in was designated as a National Nature Reserve. total. A large number of nationally or locally scarce An extensive programme of monitoring plants invertebrates are associated with the grasslands, and invertebrates was initiated at Thrislington in including the Northern Argus butter¯y Aricia artax- 1982. Monitoring an invertebrate community can erxes spp. salmacis Stephens 1831 and the glow reveal how it changes over time. Once the species worm Lampyris noctiluca L. composition is known, it may be possible to target rare species or those that play a key functional role in the maintenance of the community, and design a S A M P L I N G P R O G R A M M E sampling programme to monitor changes in their Two adjacent sites approximately 5 ha in area were abundance. However, in the initial stages the feature sampled, one of which was to be translocated in most often used to assess change is species richness. 1983. The ®rst sample was taken in May 1982, then It is therefore necessary to employ a sampling each following month up to and including Septem- regime which ensures that less abundant species, or ber 1982, using three collecting techniques. The sam- those not easily captured, are recorded because they pling programme was repeated at both sites in 1985 may be the species that are most a€ected by distur- and 1988, giving a total of 15 months of data for bance. each. Macroinvertebrates were counted and identi- Where the aim is to monitor all the invertebrates, ®ed by specialist taxonomists, except for (i) Isopoda several collecting techniques must be used as no sin- (Philoscia muscorum (Scopoli) and Armadillidium gle method will capture from all microhabi- vulgare (Latreille)), which were ubiquitous and tats. The methods chosen for this work were pitfall super-abundant and of little value therefore in traps, a D-vac suction apparatus and a swish net detecting change, and (ii) Mollusca and Lepidop- designed to capture the most active and easily dis- tera, as none of the methods employed was suitable turbed invertebrates. The samples di€er in two for these groups. important respects. First, they represent di€erent parts of the total fauna, pitfall traps collecting mostly epigeal fauna, and swish net and D-vac C O L L E C T I N G T E C H N I Q U E S mostly epiphytic fauna. Secondly, the patterns of accretion of individuals and species will vary Pitfall traps because pitfall traps depend on the animals' activity and collect continuously from an unknown area, Nine plastic cups, 8 cm in diameter, were set without whereas the suction apparatus and swish net collect covers 2 m apart in a 3  3 grid in each site. Ethy- from a known area for brief periods. lene glycol of 80% was used as preservative and the # 2000 British The aim of monitoring invertebrates in this pro- traps were set continuously and emptied monthly Ecological Society gramme was to assess changes in species composi- throughout May to the end of September in each Journal of Applied Ecology, 37, tion and richness over a number of years following year of study. No attempt was made to compensate 884±893 grassland translocation. The objectives of this paper for loss, which was approximately 10%. 886 Swish samples families. However, several groups were not identi®ed Monitoring to species owing to lack of taxonomic expertise and The swish samples were designed to collect grassland these were excluded: Sciarinea (), 130 ¯ying over the sward and visiting the tallest in¯ores- invertebrates specimens; (), 50 specimens; cences. These often escape the standard sweep-net and Anthomyiinae (acalypterate Schizophora), 400 technique, which is designed to collect amongst the specimens. The Araneae were grouped as Linyphii- grass sward and canopy. The aluminium frame was dae, Lycosidae and `other spider' families. The 0Á4 m in width on a 1Á5-m handle and was swept families included in the non-familial groupings are only over the topmost part of the sward. One sam- listed in Table 1. ple unit was 20 strokes of 0Á75 m width taken at 0Á75-m intervals so that the area swept was 6 m2 per sample within approximately 10 m2. Five sample R A R E F A C T I O N A N D C O L L E C T O R C U R V E S units were collected, giving an area of approximately Although species abundance models such as rarefac- 30 m2 sampled each month. tion (Hurlbert 1971) cannot be used to compare spe- cies richness for samples collected by di€erent D-vac samples methods (Krebs 1989), the species/specimen curves The D-vac used was a standard Dietrick suc- for di€erent taxa collected by the same method were tion sampler (D-Vac Ltd, Ventura, CA, USA) (Die- computed using the BIODIV program (Baev & trick 1961) with a 35-cm diameter intake placed Penev 1995). Using the initial sample, the program onto the ground for 30 s. One sample was four 30- calculates the expected number of species in samples second catches, giving an area of 0Á38 m2 per sam- of individuals taken randomly. The results are ple. Five samples were taken, giving an area of plotted graphically as `collector curves'. approximately 2 m2 sampled each month. Results Combined swish and D-vac samples (SW/DV) Approximately 4000 Hemiptera, Coleoptera and For purposes of comparison with pitfall traps, the Araneae, and 3000 Diptera, individuals were cap- swish net and D-vac samples were combined (SW/ tured, yielding over 480 species. Other taxa identi- DV). Totals for the two methods could not simply ®ed and counted but not considered further were be added together as many of the same species were Acrididae, Panorpidae (one species each), Hyme- found by both methods. Individuals were therefore noptera (20 species), Opiliones (10 species) and Myr- treated as though they had been collected by a single iapoda (eight species). A summary of the four major method and species richness calculated for the com- taxa caught in pitfalls, swish and D-vac samples is bined data set. shown in Table 2. Most species and individuals of Coleoptera and Araneae were found in pitfall trap T A X O N O M I C G R O U P S U S E D I N A N A L Y S E S samples, most Diptera in swish net samples, and most Hemiptera in D-vac samples. Comparisons between collecting methods were based on totals obtained by combining sites, months A B U N D A N C E A N D S P E C I E S R I C H N E S S and years for each technique. Where large numbers W I T H I N E A C H S A M P L E T Y P E of species were involved, a choice must be made as to how to group them to ease analysis and interpre- The pitfall trap, swish net and D-vac samples pro- tation. For preliminary analysis, species were placed duced, respectively, 7080, 2786 and 4216 specimens in broad taxonomic groups. Where these were rela- of all invertebrates. The swish net and D-vac sam- tively uniform in function or behaviour, e.g. Carabi- ples were combined (7002 specimens), giving dae and Staphylinidae, each family was retained as approximately equal numbers of individuals in each a group while others, such as the remaining Coleop- sample. The most speciose taxa in the pitfall sample tera families, were split into two groups with mainly (> 10% total) were Coleoptera herbivores, Liny- herbivorous Coleoptera families in one group phiidae, acalypterate Schizophora and Staphylini- (Coleoptera herbivores) and mainly scavenger, pre- dae, in that order. In the SW/DV sample they were datory, parasitic or detritivore Coleoptera families acalypterate Schizophora, Homoptera and Linyphii- in the other (`other Coleoptera'). The Sternor- dae (Table 1 and Fig. 1). The large proportion of rhyncha (Aphidoidea and Coccoidea) were not iden- species of herbivorous Coleoptera and acalypterate ti®ed, and the remaining Hemiptera were treated as Schizophora in pitfalls was surprising, but the for- the two suborders, Homoptera and Heteroptera. mer taxon contains many root-feeding weevils while The Diptera comprised the largest group and were many of the latter are carrion feeders drawn to # 2000 British divided into the suborders Nematocera, Brachycera decomposing remains of pitfall victims. Ecological Society and Cyclorrhapha, but with the latter split into Collector curves for the pitfall sample showed Journal of Applied Ecology, 37, Aschiza and Schizophora. The Schizophora were that all the Coleoptera and Araneae taxa and the 884±893 divided into the calypterate and acalypterate calypterate Schizophora reached a plateau and were 887 Table 1. Species richness of each lower taxonomic group captured in pitfall, swish net and D-vac samples. Values for SW/ V. Standen DV were calculated from combined raw data for swish net and D-vac methods

Group Abbreviations used in ®gures Total Pitfall Swish D-vac SW/DV

Hemiptera Heteroptera* Het 19 6 11 16 18 Homoptera{ Hom 45 28 21 40 44 Coleoptera Carabidae Car 27 24 3 4 7 Staphylinidae Sta 45 37 7 17 20 Coleoptera herbivores{ ColH 55 46 22 27 31 `Other Coleoptera'x ColS 26 19 7 9 12 Diptera Nematocera{ Nem 17 8 10 8 15 Brachycera** Bra 36 16 25 21 32 Aschiza{{ Asc 28 5 25 9 27 Acalypterate Schizophora{{ Acal 88 38 52 52 74 Calypterate Schizophoraxx Cal 33 18 24 7 27 Araneae Linyphiidae Lin 51 39 19 27 37 Lycosidae Lyc 7 7 1 1 1 `Other spider' families{{ OS 18 16 4 6 9

*Heteroptera: Anthocoridae, Miridae, Nabidae, Neididae, Piesmidae, Rhopalidae, Tingidae. {Homoptera: Cercopidae, Cicadellidae, Cixiidae, Delphacidae. This group is equivalent to Auchenorrhyncha as Aphidoidea and Coccoidea were not identi®ed. {Coleoptera herbivore families: Apionidae, Bruchidae, Byrrhidae, Chrysomelidae, Curculionidae, Elateridae, Hydrophilidae, Nitidulidae. x`Other Coleoptera' families: Cryptophagidae, Lathridiidae, Leiodidae, Scarabaeidae, Silphidae, Coccinellidae, Cantharidae, Lampyridae, Pselaphidae. {Nematocera: Tipulidae, , , Cecidomyidae, , , . **Brachycera: , Empidae, , . {{Aschiza: , , Syrphidae. {{Acalypterate Schizophora: Trypetidae, , , , , , Helomyzidae, , , , , . xxCalypterate Schizophora: , . {{`Other spider' families: Araneidae, Clubionidae, Gnaphosidae, Liocranidae, Mimetidae, Tetragnathidae, Theriidae, Thomisidae, Zoridae.

therefore reasonably well sampled by the pitfall adequate for Homoptera, Heteroptera, Coleoptera sampling programme (Fig. 2a). The Homoptera did herbivores, Brachycera, Aschiza, acalypterate Schi- not form a plateau despite collection of over 600 zophora and Linyphiidae. The Carabidae, Staphyli- individuals. The Heteroptera, Nematocera, Brachy- nidae, `other Coleoptera', Nematocera, calypterate cera, Aschiza and acalypterate Schizophora were Schizophora and `other spider' families were too incomplete due to low numbers. sparse to give adequate collector curves. Thus only Collector curves for the SW/DV sample (Fig. 2b) the Coleoptera herbivores and Linyphiidae were showed that the sampling programme was probably comparable in both samples. Collector curves were

# 2000 British Ecological Society Journal of Applied Ecology, 37, Fig. 1. The proportion of total species in the pitfall sample compared with the SW/DV sample for each taxonomic group. 884±893 The abbreviations used in the diagram are explained in Table 1. 888 Table 2. Species richness, and number of method-unique and singleton species of major taxonomic groups caught in pitfall, Monitoring swish net and D-vac samples. Singleton species were those found only once (a) by each method (method singletons) and (b) by any method (true singletons). Values for SW/DV were calculated from combined raw data for the swish net and D-vac grassland methods invertebrates Totals Pitfall Swish D-vac SW/DV

Hemiptera Individuals 4104 542 938 2624 3562 All species 64 34 32 56 62 Method-unique species 23 1 7 15 30 Singleton species ± a 32 9 10 13 12 Singleton species ± b 12 1 3 8 11

Coleoptera Individuals 3897 2726 586 585 1171 All species 153 126 39 57 70 Method-unique species 100 80 7 13 24 Singleton species ± a 72 37 16 19 24 Singleton species ± b 45 33 6 6 12

Diptera Individuals 2235 498 1112 625 1737 All species 194 85 136 97 175 Method-unique species 94 18 50 26 109 Singleton species ± a 133 38 56 39 55 Singleton species ± b 54 7 30 17 47

Araneae Individuals 3846 3314 150 382 532 All species 76 62 24 34 47 Method-unique species 40 30 4 6 15 Singleton species ± a 31 13 7 11 16 Singleton species ± b 14 9 2 5 7

Total individuals 14082 7080 2786 4216 7002 All species 487 307 231 244 354 (percentage all species caught by each method) (63) (47) (50) (73)

not calculated for Lycosidae because only ®ve speci- bivorous Coleoptera, where more species were mens of one species were captured in the SW/DV found in pitfall traps despite lower abundance sample, while in the pitfall sample over 2000 speci- (Table 1). mens of seven species were caught. The ratio of individuals to species for all taxa was lower for the SW/DV sample (20 : 1) compared with

C O M P A R I S O N S B E T W E E N S A M P L E T Y P E S the pitfall sample (23 : 1). Much of this di€erence could be attributed to the large numbers of a few Abundance species of Carabidae and Lycosidae in pitfall sam- ples. Carabidae, `other Coleoptera' families, Staphylini- dae, Lycosidae and `other spider' families were rela- tively more abundant in the pitfall sample, while Homoptera, Heteroptera, herbivorous Coleoptera, Singleton species Brachycera, Aschiza, acalypterate Schizophora, Many species were caught once only by a particular calypterate Schizophora and Linyphiidae were rela- method (method singletons) and some species only tively more abundant in the SW/DV sample (Table once by any method (true singletons) (Table 4). In 3). the pitfall sample, over 30% of species of Hetero- ptera, Staphylinidae, `other Coleoptera' families, Brachycera, Aschiza and acalypterate Schizophora Species richness were method singletons. The same comparison for The pitfall traps yielded a total of 307 species and the true singletons showed that, of these, Hetero- # 2000 British the swish net and D-vac sample (SW/DV) 354 spe- ptera, Aschiza and calypterate Schizophora were Ecological Society cies. The relative species richness of each taxon in never found singly in pitfall traps, the di€erence Journal of Applied Ecology, 37, pitfall compared with SW/DV samples was the same indicating that the occurrence of species of these 884±893 as for their numerical abundance except for the her- taxa (and Homoptera and other Diptera taxa to a 889 V. Standen

Fig. 2. Species/specimen or collector curves calculated for each taxonomic group (excluding Lycosidae) based on (a) the pit- fall sample and (b) the SW/DV sample. The abbreviations used in the diagram are explained in Table 1.

lesser extent) as method singletons in the pitfall sam- Comparisons between the proportion of true sin- ple was accidental (Fig. 3a). gleton species in the two sample types showed that, In the SW/DV sample, over 30% of Carabidae, apart from Lycosidae which were never found as herbivorous Coleoptera, `other Coleoptera' families, singletons, a higher proportion of Coleoptera and Nematocera and calypterate Schizophora were Araneae taxa were singletons in the pitfall sample method singletons, whereas only the herbivorous than in the SW/DV sample, while a higher propor- # 2000 British Coleoptera families were substantially less well tion of the Hemiptera and Diptera taxa were single- Ecological Society represented as true singletons, indicating that for tons in the SW/DV sample compared with the Journal of Applied Ecology, 37, most taxa capture by swish net or D-vac was less pitfall sample (Fig. 3c). Less than 5% of Hetero- 884±893 likely to be accidental (Fig. 3b). ptera, Homoptera, Nematocera, Aschiza and calyp- 890 Table 3. The number of specimens of each lower taxonomic acalypterate Schizophora) or, conversely, high Monitoring group captured in the pitfall and the combined swish net uniqueness in pitfall samples (Carabidae, Staphylini- and D-vac samples. Values for the SW/DV sample were grassland dae and Lycosidae). `Other Coleoptera' were rela- calculated from combined raw data for swish net and D- invertebrates vac methods. The composition of non-familial groups is tively highly method-unique in both sample types, shown in Table 1 while Homoptera, herbivorous Coleoptera, calypte- rate Schizophora, Linyphiidae and `other spiders' Pitfall SW/DV showed low uniqueness in both (Fig. 4). As the pro- Heteroptera 21 842 portion of method-unique species may be distorted Homoptera 521 2720 by the inclusion of species found `accidentally' by Carabidae 1466 20 one or other collecting method, the method single- Staphylinidae 632 203 ton species were omitted; their removal increased Coleoptera herbivores 452 923 the proportion of species that were method-unique, `Other Coleoptera' 176 25 Nematocera 58 71 as illustrated in Fig. 4 where the length and direction Brachycera 32 257 of an arrow re¯ects the relative change. The values Aschiza 209 773 for Carabidae, `other Coleoptera' families, Nemato- Calypterate Schizophora 164 402 cera, Brachycera and Aschiza were least reliable as Linyphiidae 382 458 they contained 10 or fewer species in one or other Lycosidae 2438 5 `Other spider' families 494 69 sample type. There are several reasons for lack of method- uniqueness shown by species in these groups. Some contain families with species that inhabit the vegeta- tion but fall to the ground if disturbed and so are terate Schizophora were singletons in pitfall traps, captured in pitfalls as well as by SW/DV technique but the remaining taxa contained a relatively high (e.g. Cicadellidae, Linyphiidae, Elateridae). Another proportion of singletons in both sample types. group are the calypterate Schizophora, which are Species may be found as single specimens because alert, strong ¯iers and might be expected to be they occur at low density or because they are abun- caught only by the swish net. However, many calyp- dant but not eciently caught by either of the col- terate species are also attracted to carrion (e.g. Calli- lecting methods. This indicates that more than one phoridae, blow¯ies) and were caught in pitfall traps. method should be used to collect the rarer species of Similarly, herbivorous Coleoptera include species most taxa. that might be expected to be caught mainly by the SW/DV collecting method from the upper layers of vegetation, but it also includes numerous species of Method-unique species Curculionidae (weevils), many of which are root fee- Many species were found only in one or other sam- ders, and hence some species are captured predomi- ple (Table 4) and, for most taxa, there was a ten- nantly in pitfall traps and colleagues by D-vac or dency for either high uniqueness in SW/DV samples swish net. The `other spiders' group is arbitrary and (Heteroptera, Nematocera, Brachycera, Aschiza and it includes families where most species build webs

Table 4. The number of method singleton, true singleton and method-unique species in pitfall samples and the combined swish net and D-vac samples. The composition of non-familial groups is shown in Table 1

Method singleton True singleton Method-unique

P SW/DV P SW/DV P SW/DV

Homoptera 7 7 1 6 1 21 Heteroptera 2 5 0 5 0 14 Carabidae 4 4 4 2 21 3 Staphylinidae 14 5 14 3 28 8 Coleoptera herbivores 12 10 9 3 27 10 `Other Coleoptera' 7 5 6 4 15 8 Nematocera 2 7 0 6 2 9 Brachycera 11 7 2 7 4 26 Aschiza 2 8 0 8 1 25 Acalypterate Schizophora 18 22 5 18 13 56 # 2000 British Calypterate Schizophora 5 13 0 8 6 15 Ecological Society Linyphiidae 9 6 5 6 20 15 Journal of Applied Lycosidae 0 0 0 0 6 0 Ecology, 37, `Other spider' families 4 2 4 1 10 3 884±893 891 V. Standen

Fig. 3. The percentage of species of each taxonomic group that were (a) method or true singletons in the pitfall sample; (b) method or true singletons in the SW/DV sample. (c) The percentage of species in the pitfall sample compared with the per- centage of species in the SW/DV sample that were true singletons. The abbreviations used in the diagram are explained in Table 1.

within the vegetation (e.g. Araneidae) and are enough specimens to estimate species richness of a caught by D-vac, plus others which are nocturnal wide range of taxa in the grassland habitat. The pit- hunters (e.g. Clubionidae) and caught in pitfall fall traps were used to capture the mainly epigeal traps. fraction of the total fauna, and the combined swish net and D-vac method for the epiphytic fraction, resulting in two samples. The questions discussed are (i) which collecting technique(s) could be used to Discussion estimate species richness of each taxon and (ii) given # 2000 British The underlying rationale of the Thrislington sam- that the methods collect from di€erent microhabi- Ecological Society pling programme was to monitor change in species tats, what is the extent of species overlap, is one or Journal of Applied Ecology, 37, richness of above-ground invertebrates. It was other method adequate for each taxon or should 884±893 hoped that the chosen methods would collect both be used? 892 Monitoring grassland invertebrates

Fig. 4. The proportion of species in each lower taxonomic group caught in the combined swish net and D-vac suction meth- ods (SW/DV) that were captured uniquely by those methods plotted against the proportion of species in pitfalls captured only in pitfalls. Open circle ˆ as a proportion of all species in each group; closed circle ˆ as a proportion of all species minus method singletons in each group. The abbreviations used in the diagram are explained in Table 1.

W H I C H C O L L E C T I N G T E C H N I Q U E ( S ) C O U L D Taxa sampled adequately by the SW/DV method B E U S E D T O E S T I M A T E S P E C I E S R I C H N E S S ? The SW/DV sample abundance was adequate to estimate species richness of Heteroptera, Homo- ptera, Brachycera, Aschiza and acalypterate Schizo- Taxa sampled adequately by the pitfall method phora. The Heteroptera and Homoptera are either The pitfall sample collected sucient numbers of all phytophagous or predatory, clinging to vegetation the Coleoptera and Araneae groups to estimate spe- within the sward, and are captured most e€ectively cies richness. Apart from the herbivorous Coleop- by D-vac suction method. Most Diptera tend to tera, most of the taxa contained predatory species escape capture by D-vac either because they visit that hunt actively on the surface, while the `other only the tallest in¯orescences [e.g. hover¯ies (Syrphi- Coleoptera' group contained scavengers that are dae ± Aschiza)] or because they ¯y readily when dis- attracted to the decomposing invertebrates in the turbed, but apart from the calypterate Schizophora pitfall traps (e.g. Silphidae). The pitfall sample also they were well represented in the swish net samples. gave a reasonable estimate of species richness for It was also possible to estimate species richness for the calypterate Schizophora, which also were herbivorous Coleoptera and Linyphiidae in the SW/ attracted to carrion in the traps. The pitfall sample DV sample. The method was inadequate for the was not adequate for estimating species richness of remaining Coleoptera and Araneae taxa and the the remaining Diptera taxa or the Hemiptera. Nematocera and calypterate Schizophora. Pitfall traps capture animals from an unknown The D-vac technique catches species living on the area and therefore do not measure absolute density vegetation (epiphytic) and also many which live on and, because they rely on the animals' behaviour, the ground surface. Using the D-vac suction sam- they do not measure relative density either. For pler, it is possible to relate numbers captured to the example, they over-represent larger carabid species area sampled (Du€ey 1980) and the method was (Spence & Niemela 1994; Ulber & Wolf-Schwerin found to be the most successful for Auchenorhyncha 1995) and male spiders (Dinter 1995), the same spe- (ToÈ rmaÈ la 1982). The new swish net technique also cies are caught di€erentially on di€erent sites collects from a known area and the combination of (Briggs 1961) and di€erent species are caught di€er- D-vac and swish net could be recommended for esti- entially on the same site (Halsall & Wratten 1988). mating the density and species richness of the epi- However, they capture large numbers of individuals phytic fraction of Hemiptera and Diptera (apart continuously. Despite their de®ciencies, pitfalls from Nematocera and calypterate Schizophora), and could be recommended as a collecting technique sui- of the herbivorous Coleoptera and the Linyphiidae. table for monitoring the presence or yearly ¯uctua- tions of the epigeal fraction of the Coleoptera and S P E C I E S O V E R L A P B E T W E E N S A M P L E S Araneae and, providing the sample is large enough, # 2000 British their species richness also. The calypterate Schizo- The Coleoptera herbivores and the Linyphiidae Ecological Society phora were also well represented in the pitfall sam- were e€ectively sampled in both microhabitats, and Journal of Applied Ecology, 37, ple, but special techniques are usually employed for the Nematocera in neither. The remaining taxa were 884±893 this group. sampled suciently well to estimate species richness 893 in one or other microhabitat and the question then References V. Standen arises as to whether the fraction of the fauna not Baev, P.V. & Penev, L.D. (1995) BIODIV, Program for Cal- e€ectively sampled is a subsample of the other and culating Biological Diversity Parameters, Similarity, can be ignored, or whether it is signi®cant and, if so, Niche Overlap and Cluster Analysis, Version 5.1. Pen- how best to sample it. soft, So®a, Moscow. Briggs, J.B. (1961) A comparison of pitfall trapping and The proportion of method-unique Heteroptera soil sampling in assessing the populations of two spe- and Homoptera in the pitfall sample and of Lycosi- cies of ground beetles (Coleoptera, Carabidae). Report dae in the SW/DV sample was very low, and for of East Malling Research Station for 1960, 108±112. most purposes non-epiphytic Hemiptera and non- Cherrett, J.M. (1964) The distribution of spiders on the Moor House National Nature Reserve, Westmoreland. epigeal Lycosidae could be ignored. However, Journal of Ecology, 33, 27±48. although a higher proportion of Coleoptera and Cherrill, A.J. & Sanderson, R.A. (1994) A comparison of Araneae was method-unique in the pitfall sample, sweep net and pitfall samples of moorland Hemiptera: and of Hemiptera and Diptera taxa in the SW/DV evidence for vertical strati®cation within vegetation. Entomologist, 113, 70±81. sample, there was still a relatively high proportion Danahar, G.W. (1998) The in¯uence of vegetation and of species of 11 taxa that were method-unique in microclimate on the structure of chalk grassland leafhop- both sample types. This indicates that di€erent spe- per communities. PhD Thesis. University of Sussex, cies assemblages characterize the two microhabitats. Brighton, UK. Denno, R.F. (1994) In¯uence of habitat structure on abun- The observation is supported by the relatively high dance and diversity of planthoppers. Planthoppers, proportion of singletons of eight taxa found in both Their Ecology and Management (eds R.F. Denno & samples. The fraction of the fauna of each taxon T.J. Perfect), pp. 140±159. Chapman & Hall, New that was not collected in sucient numbers to esti- York, NY. Dietrick, E.J. (1961) An improved backpack motor fan for mate species richness should not be disregarded. suction sampling insect populations. Journal of Eco- Either more specimens should be collected by the nomic Entomology, 54, 394±395. pitfall or SW/DV method or special techniques Dinter, A. (1995) Estimation of epigeic spider population should be employed to sample the non-epiphytic densities using an intensive D-vac sampling technique and comparison with pitfall trap catches in winter Nematocera, Brachycera, Aschiza and acalypterate wheat. Natural Enemies in Arable Land (eds Schizophora, and the non-epigeal Carabidae, Sta- S. Toft & W. Riedel), Acta Jutlandica, 70, 23±33. phylinidae, `other Coleoptera', calypterate Schizo- Du€ey, E. (1980) The eciency of the Dietrick vacuum phora and `other spiders'. sampler (D-vac) for invertebrate population studies in di€erent types of grassland. Bulletin of Ecology, 11, The total number of species found in the samples 421±431. (Table 1) was reasonably accurate for herbivorous Halsall, N.B. & Wratten, S.D. (1988) The eciency of pit- Coleoptera and Linyphiidae because adequate speci- fall trapping for polyphagous predatory Carabidae. mens were collected in both habitats, and for Het- Ecological Entomology, 13, 293±299. Hurlbert, S.H. (1971) The non-concept of species diversity: eroptera, Homoptera and Lycosidae because they a critique and alternative parameters. Ecology, 52, were more or less con®ned to one or other habitat. 577±566. For the remaining taxa, the total species was an Krebs, C.J. (1989) Ecological Methodology. Harper Collins, underestimate of the number actually present New York, NY. Kromp, B., P¯uÈ gel, Ch., Hradetzky, R. & Idinger, J. because not all microhabitats within the grassland (1995) Estimating bene®cial arthropod densities using were adequately sampled. The results presented for emergence traps, pitfall traps and the ¯ooding method Thrislington grassland are based on a single sample in organic ®elds (Vienna, Austria). Arthropod Natural and cannot be extrapolated to other grasslands, still Enemies in Arable Land (eds S. Toft & W. Riedel), Acta Jutlandica, 70, 87±100. less to other habitats, but they con®rm the suitabil- Lu€, M.L. (1966) The abundance and diversity of the bee- ity of pitfall traps for sampling the epigeal fraction tle fauna of grass tussocks. Journal of Animal Ecology, of the Coleoptera and Araneae taxa, and of suction 35, 189±208. and netting methods for the epiphytic fraction of Park, D.G. (1989) Relocating magnesian limestone grass- land. Biological Habitat Reconstruction (ed. Hemiptera and most Diptera taxa. G.P. Buckley), pp. 264±279. Bellhaven, London. Spence, J.R. & Niemela, J.K. (1994) Sampling carabid assemblages with pitfall traps: the madness and the Acknowledgements method. Canadian Entomologist, 126, 881±894. ToÈ rmaÈ la, T. (1982) Evaluation of ®ve methods of sampling The data set was collated and made available for ®eld layer , particularly the leafhopper com- analysis by Dr David Sheppard of English Nature, munity in grasslands. Annales Entomologicae Fennici, 48, 1±16. Peterborough. I am grateful to Dr Sheppard for per- Ulber, B. & Wolf-Schwerin, G. (1995) A comparison of mission to analyse the data and for his willing help pitfall trap catches and absolute density estimates of and encouragement. A British Ecological Society carabid beetles in oilseed rape ®elds. Arthropod Natural Small Project Grant funded the recording of the raw Enemies in Arable Land (eds S. Toft & W. Riedel), # 2000 British Acta Jutlandica, 70, 77±86. Ecological Society data as an electronic database. I am also indebted to Journal of Applied three referees who suggested improvements to the Ecology, 37, original manuscript. Received 27 March 1999; revision received 10 March 2000 884±893