Introduction

The Global Biodiversity Crisis is an important challenge facing humankind, which is largely driven by the degradation and loss of natural habitat (Mokany et al. 2020). Although the Convention on Biological Diversity was signed in 1992 to try to reduce biodiversity loss, the shortage of financial and human resources to comprehensively document, describe and conserve global biodiversity (Barbier et al. 2018) has exacerbated species declines. As a result, some species are assumed to have gone extinct before they have been described (e.g. Costello 2015; Lees and Pimm 2015).

Contributing to this problem is the lack of human resources and time to intensively and effectively sample habitats and describe new taxa from the material collected, resulting in considerable gaps in our knowledge of the taxonomy, distribution, biology and ecology of organisms (e.g. Victor et al. 2015). This situation is particularly evident in countries that do not have adequate financial resources or human capacity to extensively sample and document their fauna and flora (Hochkirch et al. 2021). This problem is exacerbated in certain megadiverse countries such as South Africa, which often have high levels of endemism (e.g. Cowling and Hilton-Taylor 1994; Tolley et al. 2019; Foord et al. 2020) and feature global biodiversity hotspots (Myers et al. 2000; Mittermeier et al. 2011).

One potential solution to this problem is to optimise the available human resources through atlas projects, to more effectively generate biodiversity data and acquisition of specimens for natural history collections, improving the documentation of species occurrences (Robertson et al. 2010; Dippenaar-Schoeman et al. 2015). Data generation for various organismal groups benefits greatly from the involvement of volunteers (Lovell et al. 2010; Lewandowski and Specht 2015), citizen scientists (Theobald et al. 2015; Chandler et al. 2016) or students (e.g. Cardoso et al. 2017; Kiljunen et al. 2020; Soukainen et al. 2020; Supriatna et al. 2020) to assist with field sampling. Further, the extraction of non-target groups as bycatch in surveys can also provide data that is comparable to dedicated surveys of those particular non-target groups (e.g. Oberprieler et al. 2019).

Insects are the largest component of terrestrial biodiversity and play various significant ecological roles in ecosystems (Yang and Gratton 2014). Training undergraduate students in Entomology requires a broad scope and approach to provide adequate knowledge in the core disciplines, e.g. ecology, systematics, physiology and pest management. Considering the importance of the Global Biodiversity Crisis, an understanding of the planet’s biodiversity and the importance of conserving species richness is critical in the education process (Dikmenli 2010; Lamarre et al. 2018). This creates an understanding of natural systems and the organisms occupying them. Skills development in biodiversity and ecology, specifically, can be enhanced by exposing students to various sampling techniques and methodological approaches through field excursions and practicals (e.g. Yli-Panula et al. 2018).

Training in insect biodiversity and ecology equips students to understand the importance of (1) identifying an ecological scenario or problem, (2) planning a protocol to sample adequately and effectively to determine ecological patterns, (3) executing sampling, sorting and identification of material, (4) analysing data using appropriate statistical tests, and (5) preparing a written report to a hypothetical client that provides evidence and solutions to a particular problem or scenario. Further, they gain an understanding of the guilds or functional feeding groups of insects (Kaiser et al. 2009; Buschke and Seaman 2011), and how these organisms utilise the available food resources in the biotopes sampled.

In the current study, the contribution of an annual undergraduate Entomology student field excursion to generating biodiversity data on a relatively well-studied non-target arthropod group, spiders (Arachnida: Araneae), was documented over the course of 5 years. A standardised sampling protocol was used to sample terrestrial arthropods, with spiders as a bycatch taxon, to determine the proportion of the local species richness for the site sampled by each student year-group, and also by each group of students. This way, the success of student groups in contributing biodiversity data could be assessed and compared with more experienced arachnologists, who conducted the sampling in the final year of the study. The data generated through this study contributed towards an ongoing atlas project in South Africa, the South African National Survey of Arachnida (Dippenaar-Schoeman et al. 2015), by surveying in a historically undersampled degree-square of the Free State Province (Foord et al. 2011; Haddad et al. 2013).

By exposing students to such a field trip, they can develop a variety of skills related to ecological surveys and data processing. Further, they can develop an appreciation for the contribution that their sampling may make to generating baseline biodiversity data for an area, even for non-target taxa, as well as the value that specimens collected can make to national atlas programmes and future taxonomic research.

Materials and methods

General background

This study was based on data generated as part of an annual field excursion forming part of the 3rd year Entomology curriculum at the University of the Free State in Bloemfontein, South Africa. The excursion was an assessment component contributing to the ENTO3714 module, titled “Advanced Insect Ecology”. The core purposes of the excursion are to expose students to an environment not encountered during their regular practical assignments in the Grassland Biome; to sample arthropods using a standardised protocol and to apply sampling methods (beating, litter sifting and pitfalls) not used during their regular practical sessions (sweep-netting); and to learn various statistical techniques to analyse the datasets generated to evaluate differences in abundance, species richness and composition of functional feeding groups (guilds) in different biotopes.

The current study focused on biotopes in the transition zone between the Grassland and Nama Karoo biomes in the western Free State Province. The Grassland Biome dominates the central plateau of South Africa, and is recognized by a field layer dominated by grasses and forbs, with woody vegetation largely restricted to hillsides and drainage lines (Mucina and Rutherford 2006). The Nama Karoo Biome covers a large area in the semi-arid and arid western half of South Africa, with vegetation dominated by low-growing Asteraceae, Fabaceae and Poaceae, and trees generally rare due to the predominantly shallow soils (Mucina and Rutherford 2006). Previous field excursions exposed students to other biomes, e.g. Fynbos in the Western Cape and Afromontane Forests in the Eastern Cape. The number of groups was determined by the number of students registered for the ENTO3714 module each year, and varied from one to three groups comprising two or three students each.

Study area and period

Field excursions were undertaken at the farm Bankfontein in the Luckhof district in the western Free State Province, South Africa (Fig. 1a), during the March–April recess or over a national long weekend at the end of April, in 2015, 2016 and 2018–2020. During 2017, only two students registered for the ENTO3714 module, so the excursion during that year was combined with the annual Zoology excursion (ZLGY3714 module) to the Gariep Dam in the south-western Free State. The farm is predominantly utilised for sheep farming, although some cattle are also managed on the property.

Fig. 1
figure 1

a Map of South Africa, red star indicating the location of Bankfontein farm in the western Free State Province; b–d Habitats sampled for spiders and other arthropods: b Hillside grassland; c Nama Karoo veld; d Riparian woodland. Map a created on www.simplemappr.net

Full size image

The first biotope sampled was the west-facing slopes of hillsides (referred to as hillside grassland; 30°04.526ʹ S, 24°53.039ʹ E), which are classified as Besemkaree Koppies Shrubland (vegetation unit Gh 4 of Mucina et al. 2006a) that include scattered dolerite formations (Fig. 1b). Woody vegetation was dominated by Searsia burchellii, S. ciliata and S. erosa, with scattered Diospyros lycioides, Ehretia rigida, Grewia occidentalis, S. lancea and Ziziphus mucronata. Common grasses include Aristida, Eragrostis and Tragus spp., Heteropogon contortus and Themeda triandra, with Osteospermum leptolobum the most common karooid bushes. The substrate could be classified as sandy shale.

The second biotope, referred to as Nama Karoo veld (30°04.974ʹ S, 24°54.297ʹ E), was dominant in the lower lying parts of the farm, and was classified as Northern Upper Karoo (vegetation unit NKu 3 of Mucina et al. 2006b). The vegetation at the site sampled was dominated by the karooid bushes Asparagus capensis, Eriocephalus ericoides, E. spinescens, Lycium cinereum, Nenax microphylla and O. leptolobum, and the grasses Aristida congesta, Enneapogon sp. and Eragrostis sp., on a shale substrate, with isolated Senegalia mellifera and Vachellia karroo trees (Fig. 1c).

The third biotope (riparian woodland; 30°04.421ʹ S, 24°53.013ʹ E) was surrounded by dolerite formations and formed part of the Upper Karoo Hardeveld (vegetation unit NKu 2 of Mucina et al. 2006b), but specifically parts restricted to the drainage lines that led to the Vanderkloof Dam on the Orange River (Fig. 1d). Searsia lancea, V. karroo and Z. mucronata were the dominant trees present, but were interspersed with Diospyros austro-africana and Lotononis divaricata shrubs. Grass richness was low in this biotope, but Cynodon dactylon and Eragrostis spp. were very abundant, as were Restionaceae, Asparagus spp. and other herbs.

Sampling methodology and intensity

Each student group was required to implement a standardised sampling protocol in each of the three biotopes. Considering the paucity and low growth of grasses in the Nama Karoo veld, no sweep-net sampling was undertaken. This contrasts with their regular practical assignments, which rely exclusively on sweep-net sampling. Three methods were employed in each biotope by each group to sample arthropods:

  1. (1)

    10 pitfall traps (6 cm diameter 250 ml bottles) were set out 5 m apart in a line transect. Ethylene glycol (50 ml) was added to each pitfall trap as a preservative, with traps kept open for 4 days (96 h) before being removed. Where multiple student groups sampled in a single year, trapping transects were separated by at least 10 m to minimise interference. Arthropods were sorted from the collected material in white plastic trays and preserved in 70% ethanol in separate vials for each sample;

  2. (2)

    10 leaf litter samples were taken by filling 5 l plastic freezer bags with litter from beneath woody plants. The collected material was sifted over a large white sheet using a round steel sieve with a diameter of 45 cm, height of 10 cm and mesh spacing of 8 mm. Specimens sampled by litter sifting were collected with forceps or clean glass vials and put into separate vials containing 70% ethanol for each sample. For this method, either the author or one senior post-graduate student assisted each student group with collecting specimens from the sheet; and

  3. (3)

    10 beating samples were taken by striking the branches of woody foliage 50 times over a standard sweep-net with a diameter of 40 cm. Each of the 50 beat units comprised striking a single branch at the same point three times before moving onto the next unit. Once 50 units had been completed, the sampled material was transferred to a 5 l plastic freezer bag until sorting in the laboratory. Beat samples were euthanized using toilet paper doused in ethyl acetate, which was added to each bag on return to the field laboratory. This was to ensure that flying insects could not escape during sorting. The material was then emptied into large white plastic trays, and the arthropods removed using forceps and transferred to vials containing 70% ethanol.

The number of samples per habitat and method was determined by the number of student groups on each trip: 30 (2015), 20 (2016), 10 (2018), 30 (2019) and 20 (2020). In all of the years, students were responsible for all sampling and processing of material, except for 2020, when the University of the Free State closed in the week prior to the proposed excursion due to the COVID-19 pandemic. The author and two post-graduate Arachnology students (one Ph.D and one M.Sc level) consequently undertook the sampling before national government imposed a national lockdown on 27 March 2020, and executed collecting for the two student groups that would have undertaken the excursion.

In 2020, we were only interested in extracting spiders from the samples for the purpose of completing this study. We therefore deviated from the methodology above slightly by emptying beat samples into white sorting trays without euthanasia, so that live spiders could be easily recognised and collected while walking on the plant material or walls of the tray. Insects were ignored and not sampled, as students would not have had an opportunity to process the material during lockdown. Consequently, students in the 2020 cohort prepared their excursion reports based on a hypothetical visit to Bankfontein, utilizing the datasets of one of the 2019 groups.

Spider sorting and identification

Normal sorting and identification procedures for arthropods were implemented for each of the 10 method-samples taken by each group. Insects were identified to family level using the keys of Triplehorn and Johnson (2005) and further separated into morphospecies, while arachnids and other non-insect arthropods (woodlice, millipedes and centipedes etc.) were identified by the author. Students were instructed to keep all arachnids and other non-insect arthropods in a vial labelled for each sample taken, so that material could be allocated to the correct sample once identified.

Initial sorting and morphospecies identification of arachnid samples was done in the field laboratory to ensure inclusion of fine resolution data in the student reports, with final identifications performed at the University of the Free State by the author. For the purposes of this study, both adult and immature spiders were identified and tallied. Immature morphospecies were separated on the basis of colouration or morphological attributes until adults were sampled from the same biotope, which were linked to a particular morphospecies and identified. In some species-rich genera (Gnaphosidae: Zelotes and Thomisidae: Thomisus), there were inadequate characters to separate potential immature morphospecies, and these were then pooled and treated as a single morphospecies; adults of these genera were identified and tallied separately (Appendix 1). Adult material was identified to species level using taxonomic literature available from the World Spider Catalog (2021) for all families. In some cases, spiders could not be identified to species level, due to the lack of modern revisions or redescriptions, and otherwise because new species had been discovered. Voucher specimens of adult arachnids have been deposited in the National Collection of Arachnida at the ARC—Plant Health and Protection in Pretoria, South Africa.

The local species richness (LSR) for spiders was determined by incorporating all spider data sampled from Bankfontein by students during the excursions, as well as additional collecting by the author, post-graduate students and other arachnologists on the farm during the period 2014–2021. Prior to this period, no sampling had been undertaken there. The LSR therefore includes all spider species records, irrespective of sampling method. Other sampling methods used by more experienced collectors included ad hoc hand collecting from rocks and logs, sweep-netting, searching in grass tussocks, night collecting, and a single canopy fogging sample. In addition, extra samples were taken at other non-student sites by beating, litter sifting and pitfalls.

Statistical analysis

To assess the effectiveness of the sampling protocol in determining species richness, the number of species per sample was used to calculate the mean ± SD species richness for each year for each method-biotope combination. Values were calculated and graphs prepared in GraphPad InStat version 3.00.

Using PAST version 2.07 (Hammer et al. 2001), sample rarefaction curves were prepared for each method-biotope combination. In preparing these curves, datasets were prepared based on annual accumulation of samples, i.e. 2015 alone (3 groups), 2015 + 2016 (5 groups), 2015 + 2016 + 2018 (6 groups), 2015 + 2016 + 2018 + 2019 (9 groups), and 2015 + 2016 + 2018−2020 (11 groups). This was done to assess whether the rarefaction curves would approach an asymptote. In PAST, sample rarefaction implements the “Mao tau” analytical procedure, with standard errors indicated as 95% confidence intervals on the resulting graphs. The confidence intervals were only indicated for the full dataset, i.e. 11 groups.

To assess the contribution of each year-group in sampling spider species richness, the total number of species sampled per year (all groups pooled) was determined and divided by the LSR. Sample completeness for each year-group was calculated as the ratio of the observed species richness (Sobs) and the Chao1-estimated species richness (SChao1), with the latter calculated as:

$$ S_{Chao1} = S_{obs} + \, \left( {f_{{1}}^{{2}} /{ 2}f_{{2}} } \right) $$

where f1 is the number of singleton species (represented by one individual only) and f2 is the number of doubleton species (represented by two individuals only).

Coverage-based rarefaction and extrapolation was used to assess community richness and sampling effort (Chao and Jost 2012). Sample coverage (Cn) was defined as the proportion of the total number of individuals in a community that belong to the species represented in the sample. By subtracting the sample coverage from 1, the proportion of the community belonging to as yet unsampled species can be calculated, which Chao and Jost (2012) referred to as “coverage deficit”. This metric can be interpreted as the likelihood that a new, previously unsampled species will be collected if the sample were to be increased by one individual (Chao and Jost 2012). Coverage for each year-group was calculated using the following equation:

$$ Cn = { 1 }{-}\frac{{f_{1} }}{n} \left[ {\frac{{\left( {n - 1} \right)f_1}}{{\left( {\left( {n - 1} \right)f_1} \right) + 2f_2}}} \right] $$

where n represents the number of individuals in the sample, and f1 the number of singleton species and f2 the number of doubleton species. Chao and Lee (1992) proposed that coverage values should be at least 0.5 for a sample to be considered representative of the assemblage.

Based on the cumulative species sampled by student groups over the course of the study, the similarity in faunal composition sampled using each method was compared between biotopes using Sørensen’s Quotient of Similarity, SQ = 2j/a + b, where j is the number of species shared between the two compared biotopes, and a and b represent the species richness sampled in each biotope (Magurran 2004). A value closer to 1 indicates a high degree of similarity, and closer to 0 indicates greater uniqueness of the fauna of each biotope.

Results

Overall abundance and richness patterns

The LSR for the farm Bankfontein included a total of 242 species, representing 40 families (Table 1; Appendix 1). The most species-rich families were Gnaphosidae (36 spp.), Salticidae (35 spp.), Thomisidae (21 spp.) and Theridiidae (18 spp.). Over the 5 years, students sampled a total of 158 spider species, representing 65.3% of the LSR. The total abundance and number of species sampled by each group in the three biotopes using three methods was considerably variable: 2015, three groups (n = 282, S = 37; n = 353, S = 47; n = 210, S = 48), 2016, two groups (n = 338, S = 46; n = 193, S = 34), 2018, one group (n = 382, S = 60), 2019, three groups (n = 308, S = 47; n = 483, S = 66; n = 358, S = 64) and 2020, two groups (n = 1351, S = 95; n = 1237, S = 99). When the number of species per year were pooled, however, species richness ranged between 57 and 94 for undergraduate student groups (2015–2019). Sampling in the final year of the study (2020), carried out by the author and two post-graduate students, yielded 119 species, representing almost half of the LSR (Table 1).

Table 1 Family diversity of spiders sampled from the farm Bankfontein in central South Africa, indicating the number of species in each family contributing to the local species richness (LSR) and the number of species sampled each year by students (% of LSR in parenthesis)
Full size table

Sample coverage was high in all 5 years (> 0.940), with overall coverage for the cumulative fauna sampled over all 5 years 0.994. Sample completion values were more indicative of the number of groups involved in sampling, being lowest in 2018 when a single group participated in sampling (0.667), and > 0.740 when two or more groups participated (Table 1).

Over the course of the study, total spider abundance and species richness were relatively similar between the three habitats (Table 2). Species richness was slightly higher in the hillside grassland than the other two biotopes, while abundance was highest in the Nama Karoo veld and similar in the riparian woodland and hillside grassland. Sample coverage was above 0.75 for all three biotopes, and was highest in the hillside grassland, with 0.89 (Table 2).

Table 2 Summary of overall spider abundance and species richness patterns sampled over 5 years on the farm Bankfontein in central South Africa, arranged by method and biotope
Full size table

Of the 40 families collected from Bankfontein, seven were only sampled by experienced collectors and not by student groups (Caponiidae, Migidae, Mimetidae, Pholcidae, Selenopidae, Stasimopidae and Tetragnathidae). These families are generally rare and often collected in particular microhabitats, explaining their absence from student samples. Depending on the family and habitat stratum, students collected variable proportions of the local species richness. For the most part, aside from the above exceptions, most families represented by only one or two species were successfully collected by students (Appendix 1). In total, 24 new species (~ 10% local species richness) can be confirmed from Bankfontein from recently revised genera or those currently being researched, although more than that number of species could not be identified with current literature, and may also be undescribed.

Success of sampling methods

The methodology employed by students was successful in sampling almost two-thirds of the LSR over a period of 5 years. The three sampling methods were similarly successful in generating species richness data, with litter sifting yielding the highest richness overall (98 spp.), followed by beating (92 spp.) and pitfall trapping (84 spp.). However, abundance was considerably more skewed: of the 5495 individuals collected, 2570 were collected by beating, 2144 by litter sifting, while pitfall trapping was far less successful (n = 781).

Of the three methods used, pitfall traps sampled the lowest number of species in all biotopes (Fig. 2a–c), while leaf litter sifting (Fig. 2d–f) and foliage beating (Fig. 2g–i) sampled more than twice as many species in all of the biotopes. The Nama Karoo veld produced the largest number of species for sifting and beating, while the hillside grassland had the highest species richness for pitfalls. On an annual accumulation basis, the slopes of the pitfalls had a generally consistent trajectory, suggesting that many additional species could still be sampled by this method. In contrast, the curves for beats and litter sifting increased sharply initially, but had declining slopes with an increasing number of samples, suggesting that the majority of the species in the local species richness in these microhabitats had been sampled. Greater saturation was particularly evident for the beating sampling. However, none of the biotope-method curves approached an asymptote, indicating that more species could be sampled, as supported by the estimated species richness of each biotope (Table 2).

Fig. 2
figure 2

Cumulating sample rarefaction curves of spider assemblages sampled using pitfall trapping (a–c), leaf litter sifting (d–f) and foliage beating (g–i) in hillside grassland (a, d, g), Nama Karoo veld (b, e, h) and riparian woodland (c, f, i) on the farm Bankfontein in central South Africa. Curve colours: black (2015; 30 samples), grey (2015–6; 50 samples), purple (2015–6, 2018; 60 samples), green (2015–6, 2018–9; 90 samples) and red (2015–6, 2018–2020; 110 samples). Blue lines indicate 95% confidence intervals for the full set of 110 samples

Full size image

The numerically dominant families differed considerably between sampling methods (Table 3). Over the course of the study, Theridiidae, Oxyopidae, Philodromidae, Araneidae and Thomisidae each represented between 10–20% of the fauna sampled by beating. Almost half of the individuals sampled by litter sifting belonged to Gnaphosidae (48.74%), with Linyphiidae (27.71%) also very abundant. Pitfall traps were also strongly dominated by Gnaphosidae (40.72%) and Linyphiidae (28.17%), and to a lesser extent, Lycosidae (8.07%).

Table 3 Overall percentage spider family abundance sampled in three biotopes over 5 years on the farm Bankfontein in central South Africa using vegetation beating (BT), leaf litter sifting (LL) and pitfall traps (PT)
Full size table

For eight of the nine site-method combinations, the 2020 samples yielded the largest number of species per sample, with the exception of the riparian woodland pitfall traps, which were highest in 2018 (Fig. 3). The pitfall traps and leaf litter sifting samples showed the least variability between years (Fig. 3a–f), but for beating the 2020 samples were considerably more profitable in generating species than the preceding years (Fig. 3g–i).

Fig. 3
figure 3

Mean ± SD of spider species richness per sample collected using pitfall trapping (a–c), leaf litter sifting (d–f) and foliage beating (g–i) in hillside grassland (a, d, g), Nama Karoo veld (b, e, h) and riparian woodland (c, f, i) on the farm Bankfontein in central South Africa. Numbers in square parenthesis indicate the number of samples per year for each method, determined by the number of student groups

Full size image

The distinctness of the pitfall samples was supported by the Sorensen’s Quotient values generated for the three methods, which were markedly lower for pitfalls than the other two methods, indicating a relatively unique fauna in each biotope. Values were highest for beating, indicating considerable overlap between the arboreal spider faunas of the three biotopes (Table 4).

Table 4 Sorensen’s quotient of similarity comparing species composition between habitats for three methods used to sample spiders at Bankfontein farm, South Africa
Full size table

Discussion

Field excursions play a critical role in the development of students in the biological science fields of Zoology, Entomology and Botany, exposing them to the environment and the biodiversity components that they might spend their careers working on. By creating an understanding for the ecological roles of the fauna and flora in ecosystems, students are able to foster an appreciation for the interactions between organisms and the influence of anthropogenic impacts on biodiversity.

In the current study, the contribution of an undergraduate Entomology excursion in generating biodiversity data on a non-target arthropod group, spiders, was investigated. The annual spider species richness sampled by student groups was found to vary between 23.6 and 38.8% of the local species richness (i.e. all species sampled historically at the site). However, the highest value, generated in 2020 (49.2%) could be attributed to sampling having been undertaken by the author, a professional arachnologist with more than 20 years’ experience, and two senior postgraduate students, because of the impacts of the COVID-19 pandemic prohibiting undergraduate excursions. The collecting effort in 2020 was therefore a simulation of two “student” groups, but the results clearly benefited by the considerably greater field experience of the three collectors. Furthermore, sorting spiders from the beat samples while specimens were still alive also had a strong positive effect on their detectability and successful capture, compared to extracting dead spiders from euthanized samples, as was done by student groups.

While it could be suggested that the proportion of the local species richness sampled each year by students was not remarkably high, over the course of the 5 years’ sampling the pooled data represented almost two-thirds of the local species richness (65.3%). Individually, sample completion values for each year ranged between 67.7 and 81.3%, indicating that the standardised protocol (i.e. 10 samples taken with each of three methods in three biotopes by each student group) was very effective in sampling the species that could potentially be sampled by these three methods.

When the Chao1 estimated species richness of Bankfontein based on the student sampling is considered (192 spp.; Table 2), there would still be 50 species forming part of the local species richness pool (242 spp.; Appendix 1) not collected by the three methods. Similarly, the disparity between the proportion of local species richness sampled and the sample completion values of student groups could be attributed to the sampling of rare and ecologically specialised species by more experienced arachnologists, particularly in microhabitats not sampled as part of the protocol (e.g. grass tussocks, on or beneath rocks and logs, etc.), and hand collecting at night, which would have sampled nocturnally active species.

Although the three methods sampled quite similar spider species richness, the abundance of spiders sampled by pitfall traps was approximately three times lower than by litter sifting or beating. Creating a more comparable sampling protocol by increasing the numbers of individuals sampled by pitfalls to a level similar to the other methods could follow either of three approaches: (1) the number of pitfalls per biotope could be tripled, (2) considerably larger pitfall traps could be employed, to sample a larger number of rare species (see Work et al. 2002), or (3) drift fences could be added to the trapping setup, which in the case of spiders, dramatically increases pitfall catches (Boetzl et al. 2018).

Two of these proposed changes present their own problems. Although increasing the pitfall trap size to 15 cm might approximately triple the spider catch (Work et al. 2002), making it more comparable, digging such large holes would certainly be more time-consuming, which in the context of a time-restricted field excursion may not be practical. Using a ground augur to rapidly dig holes that fit the 6 cm bottles is a lot more feasible, and even tripling the sampling effort may take less time than digging large holes. In either case, increasing the sampling effort to generate comparable numbers of individuals would considerably boost the species richness of pitfall samples, probably to levels far higher than the other two methods, as suggested by the sample rarefaction curves and Sørensen’s Quotient values. A drawback of fences is that they show a strong bias to sampling highly active spider species, which skews abundance data and assemblage composition (Brennan et al. 2005). It would therefore seem that tripling the sampling effort for pitfalls of the same size used in this study would be the simplest approach to generate abundance data comparable to beating and litter sifting in future surveys.

It is informative that despite pitfalls sampling far fewer individuals than the other two methods, and approximately only half the number of species than the other methods in each biotope, that overall pitfall species richness was quite similar to them (Table 2). This suggests that the litter and foliage assemblages each show considerable overlap between biotopes, while pitfalls generate far more distinctive assemblages and more unique species per biotope, which is supported by the Sørensen’s similarity values, species richness estimators, and slopes of the sample rarefaction curves.

Patterns in the faunistic composition of the assemblages sampled are quite consistent with other studies in semi-arid South Africa. Gnaphosidae are frequently the most abundant and/or species richness family of ground-dwelling (e.g. Lotz et al. 1991; Haddad and Dippenaar-Schoeman 2002, 2005; Haddad et al. 2015; Haddad and Butler 2018) and litter-dwelling spiders (Butler and Haddad 2011; Haddad et al. 2019) in grasslands and Nama Karoo biotopes in central South Africa, and were strongly dominant in both respects here (Appendix 1; Table 3). In contrast, vegetation sampled by beating did not have a strongly dominant family, but rather several moderately dominant families (10–20% of abundance), which is generally similar to patterns seen in two grassland studies (Fourie et al. 2013; Neethling & Haddad 2013), where Araneidae, Philodromidae, Salticidae, Theridiidae and Thomisidae were the most abundant and species-rich groups.

In conclusion, this study demonstrated the value of utilizing student groups for sampling arthropods using a standardised protocol to generate biodiversity data, supporting recent studies on spiders (e.g. Cardoso et al. 2017; Kiljunen et al. 2020; Soukainen et al. 2020). The data generated can be utilised to elucidate patterns in biodiversity between different biotopes and sampling years, and also facilitates comparison of the efficacy of different sampling methods by students. It is worthwhile pointing out to students that despite the hard work and effort required to successfully complete sampling and processing of the collected material, that the data generated has broader application value to the biological and conservation community. Further, such training may foster an interest in future participation in citizen science, expanding their impact in the generation of biodiversity data (Theobald et al. 2015; Chandler et al. 2016).

These results also highlight the need to use opportunities such as student excursions, BioBlitz field trips or international volunteer programmes to address the biodiversity deficit in developing countries across the globe, but particularly in under-resourced Africa. This approach need not be restricted to arthropods such as spiders (although any order could potentially benefit greatly from such data and specimen generation), but also to any group of animals and plants that students may potentially be exposed to during their undergraduate or graduate training. The basic principle of developing a suitable standardised sampling protocol for use on a particular organismal group would set the foundation for successful generation and use of biodiversity data for conservation and management purposes.