IJMBR 1 (2013) 21-34

Combined analyses of genetic and morphological data indicate more than one species of (Araneae: Araneidae) in South

P. Z. N. Franzini1*, A. S. Dippenaar-Schoeman2,3, K. Yessoufou4 and F. H. Van der Bank1

1Department of Zoology, African Centre for DNA Barcoding, University of Johannesburg, P. O. Box 524, Auckland Park 2006, South Africa. 2ARC-Plant Protection Research Institute, Private Bag X134, Queenswood, Pretoria, 0121, South Africa. 3Department of Zoology and Entomology, University of Pretoria, Pretoria, 0001, South Africa. 4Department of Botany and Plant Biotechnology, African Centre for DNA Barcoding, University of Johannesburg, P. O. Box 524, Auckland Park 2006, South Africa.

Article History ABSTRACT Received 25 February, 2013 Forty-three (43) Cyrtophora species are known throughout the world whereas Received in revised form 02 only is suspected to occur in South Africa. Earlier April, 2013 Accepted 29 April, 2013 investigations of South African representatives of the revealed a wide range of morphological features including colour and abdominal shapes, thus Key words: raising the need for a thorough investigation of species diversity of the genus. Arachnida, Using fresh and museum specimens sampled from 2006 to 2011, we combined Cryptic species, DNA data - the mitochondrial cytochrome c oxidase subunit 1 (CO1) and nuclear Cyrtophora citricola, histone H3 genes - with morphological features of the genitalia to investigate the DNA barcoding. question. Our genetic analysis differentiates between three to 10 potential species in our dataset based on H3 and CO1, respectively. Based on morphological characteristics, we found diverse abdominal colour variations but only two genitalia shapes were distinguishable in the collections; also suggesting that at least two species do occur in South Africa. Interestingly, specimens exhibiting these two distinguishable genitalia shapes also differ in both genetics and morphological colour variation. These specimens correspond Full Length Research Article to C. citricola and a potentially new species from the Malelane region. ©2013 BluePen Journals Ltd. All rights reserved

INTRODUCTION

The correct identification of many species is still specimens (Huber, 2004). Genital characteristics, on problematic due to complex variation in morphology which most taxonomists rely on for the description of (Barrett and Hebert, 2005). occurs in , can however be occasionally misleading, due to the family Araneidae, making it difficult to match the polymorphism in some species (Huber and González, males with females. Identification of immature spiders is 2001; Jocqué, 2002). Given all these constraints, even more problematic due to underdeveloped identifying spiders to species level can be a challenging morphological characters that are useful for species task. discrimination. For example, most identification keys are A potential tool to overcome impediments in mainly based on the morphology of genitalia of adult morphological species identification is provided by DNA barcoding (Hebert et al., 2003a; b; Barrett and Hebert, 2005; Nagoshi et al., 2011; Van der Bank et al., 2012). DNA barcoding approach is increasingly acknowledged *Corresponding author. E-mail: [email protected]. as a tool to identify closely related species (Edwards et

Int. J. Mod. Biol. Res. 22

al., 2008; Sun et al., 2012; Van der Bank et al., 2012). features of the abdomen and genitalia. Over the past 10 years, the barcoding approach became a widely used method for accelerating species identification, where traditional failed, for MATERIALS AND METHODS example, when only small fragments of organisms devoid of morphological features are available (Hebert et al., Sampling 2003a; b). It has been acknowledged not only as a tool for species identification, but also important in We undertook a five-year long data collection from 2006 discovering new species (Hebert et al., 2004; Hajibabaei to 2011. Our data collection included 34 specimens of the et al., 2005b). It also assists in controlling biological genus Cyrtophora which are currently housed at the materials at ports of entry (Yassin et al., 2008; National Collection of Arachnida (NCA), Plant Protection Newmaster and Ragupathy, 2009), conservation planning Research Institute (PPRI) of the Agricultural Research (Bickford et al., 2007; Forest et al. 2007), and Council (ARC) in Pretoria, South Africa. Additional surveillance of disease vectors (Nadin-Davis et al., 2012). sampling of specimens was undertaken from sites across Although recent analyses revealed the need for refining South Africa to cover as much as possible the known the approach (Taylor and Harris, 2012), DNA barcoding geographical ranges of the genus (Figure 1). Three remains a complementary or alternative tool to traditional specimens from Zimbabwe, one from Botswana and one taxonomic approaches (Derocles et al., 2012; Sun et al., from Madagascar were also included in the DNA analysis 2012) especially when the preferred method of traditional to allow comparison with the South African specimens. morphology is helpless (Packer et al., 2009). Newly sampled specimens were identified by the In this study, we focus on the genus Cyrtophora Simon, collectors to genus level using a field guide (Dippenaar- 1864 in South Africa. Members of the genus are Schoeman and Van den Berg, 2010) and identification commonly referred to as tent-web spiders due to the was confirmed by the second author. Information on unique shape of their orb-web (Levi, 1997; Dippenaar- localities, GPS coordinates, altitude, habitats, and Schoeman and Van den Berg, 2010). The genus photographs for all specimens (excluding PZNF L1 and comprises 43 species, as well as 9 sub-species, that are L2) is available on the international Barcode of Life wide-spread across the world (Platnick, 2012). Five Database (iBOLD). DNA sequences of the barcode species are found in Africa namely: C. caudata region CO1 are also available on the BOLD system Bösenberg and Lenz, 1895; C. citricola Forsskål, 1775; (www.boldsystems.org) and the sample and process ID C. gemmosa Thorell, 1899; C. larinioides Simon, 1895 are shown in Table 1. In total, 86 specimens were and C. petersi Karsch, 1878 (Platnick, 2012). The included in this study, consisting of 59 females, 8 males diversity of the southern African species is little explored and 19 juveniles. and only the cosmopolitan species C. citricola has so far been reported to occur for instance, in South Africa (Foord et al., 2002; Dippenaar-Schoeman et al., 2005; DNA extraction, amplification and sequencing Dippenaar-Schoeman, 2006). The need for revisions of the genus Cyrtophora has been felt worldwide, leading to Tissues (mainly legs) were sampled and placed in a the multiple initiatives observed in several countries. For microplate with a drop of 100% ethanol. Specimens were example, Yaginuma (1968) revised the Cyrtophora of preserved and handled following the guidelines set in Japan, Tikader (1982) focused on the fauna of India, Levi Hajibabaei et al. (2005a). The microplate was sent to the (1997) revised the species of the Americas and Levy Canadian Centre for DNA Barcoding (CCDB) where the (1998) undertook similar initiatives in Israel. In South CO1 gene was sequenced using standard CCDB Africa, although the genus has been suspected to be protocol (www.dnabarcoding.ca). more diverse than currently thought, no formal proof has DNA extraction, amplification and sequencing of the H3 been provided as yet. gene (and the CO1 gene of PZNF L1 and 2) were done The first world-wide revision of the genus was at the African Centre for DNA Barcoding (ACDB; undertaken by Corey (1993), who studied 19 species and www.acdb.co.za), University of Johannesburg in South noted some important morphological variations in a Africa. QIAprep Spin Miniprep Kit (Qiagen, Switzerland) number of cases. Cyrtophora citricola, the only species protocols and equipment were used for these purposes. currently known in South Africa is one of the species in The polymerase chain reaction (PCR) of the nuclear H3 which colour variation has been reported (Corey, 1993; gene was done using the primer pair H3aF (5’ Edwards, 2006), thus raising the need for in-depth ATGGCTCGTACCAAGCAGACVGC 3’) and H3aR (5’ investigations. The aim of this study was to determine the ATATCCTTRGGCATRATRGTGAC 3’). Each PCR species diversity of the genus Cyrtophora in South Africa. reaction contained 22 µl PCR Master Mix (Thermo To reach this objective, we combined DNA barcode data Scientific, Japan), 1 µl forward primer, 1 µl reverse (CO1) with the DNA nuclear region H3 and morphological primer, 0.5 µl BSA and 0.5 µl DMSO. The volume of DNA

Franzini et al. 23

Figure 1. Sites where Cyrtophora specimens were sampled in South Africa.

template added was calculated according to the Genetic distances between specimens’ DNA concentration of DNA after extraction. The PCR thermal sequences regime started with one cycle at 94°C for 3 min, a further 30 cycles of 94°C for 1 min, 48°C for 1 min and 72°C for Sequences were edited and aligned using the MUltiple 3 min and a final cycle of 72°C for 7 min. PCR products Sequence Comparison by Log Expectation (MUSCLE) were cleaned according to the QIAquick (Qiagen, program (Edgar, 2004). Genetic distances (mean, Switzerland) protocol. minimum, maximum and pairwise distances between Cycle-sequencing was done using 1.5 µl sequencing specimens) were calculated for CO1, H3 and CO1 + H3 buffer, 0.3 µl primer, 0.5 µl DMSO, 0.5µl Big Dye (combined). The standard model of evolution used in Terminator (Applied Biosystems, SA) and 35ng DNA DNA barcoding studies, that is, the Kimura 2-parameter template. The cycle sequencing thermal regime consisted (K2P) model was used for the calculation of genetic of 26 cycles at 96°C for 10 s, 50°C for 5 s and 60°C for 4 distances (Kimura, 1980). The genetic variation within min. The cycle sequencing products were cleaned and DNA regions was also assessed using the Tajima’s K DNA sequencing was done on an ABI 3130x1 genetic index of divergence (Tajima, 1983). This index was analyser. evaluated as the mean number of substitutions between

Int. J. Mod. Biol. Res. 24

Table 1. Sample and BOLD process ID for each specimen studied genetically and morphologically according to the BOLD system.

Sample ID Process ID Sample ID Process ID Sample ID Process ID Sample ID Process ID PZNF-0001 HVDBS286-11 PZNF-0024 HVDBS309-11 PZNF-0046 HVDBS331-11 PZNF-0070 HVDBS355-11 PZNF-0002 HVDBS287-11 PZNF-0025 HVDBS310-11 PZNF-0047 HVDBS332-11 PZNF-0071 HVDBS356-11 PZNF-0003 HVDBS288-11 PZNF-0026 HVDBS311-11 PZNF-0048 HVDBS333-11 PZNF-0072 HVDBS357-11 PZNF-0004 HVDBS289-11 PZNF-0027 HVDBS312-11 PZNF-0049 HVDBS334-11 PZNF-0073 HVDBS358-11 PZNF-0006 HVDBS291-11 PZNF-0028 HVDBS313-11 PZNF-0050 HVDBS335-11 PZNF-0074 HVDBS359-11 PZNF-0007 HVDBS292-11 PZNF-0029 HVDBS314-11 PZNF-0053 HVDBS338-11 PZNF-0077 HVDBS362-11 PZNF-0008 HVDBS293-11 PZNF-0030 HVDBS315-11 PZNF-0054 HVDBS339-11 PZNF-0078 HVDBS363-11 PZNF-0010 HVDBS295-11 PZNF-0031 HVDBS316-11 PZNF-0055 HVDBS340-11 PZNF-0080 HVDBS365-11 PZNF-0011 HVDBS296-11 PZNF-0032 HVDBS317-11 PZNF-0056 HVDBS341-11 PZNF-0081 HVDBS366-11 PZNF-0012 HVDBS297-11 PZNF-0033 HVDBS318-11 PZNF-0057 HVDBS342-11 PZNF-0082 HVDBS367-11 PZNF-0013 HVDBS298-11 PZNF-0034 HVDBS319-11 PZNF-0058 HVDBS343-11 PZNF-0083 HVDBS368-11 PZNF-0014 HVDBS299-11 PZNF-0035 HVDBS320-11 PZNF-0059 HVDBS344-11 PZNF-0084 HVDBS369-11 PZNF-0015 HVDBS300-11 PZNF-0036 HVDBS321-11 PZNF-0060 HVDBS345-11 PZNF-0087 HVDBS372-11 PZNF-0016 HVDBS301-11 PZNF-0037 HVDBS322-11 PZNF-0061 HVDBS346-11 PZNF-0088 HVDBS373-11 PZNF-0017 HVDBS302-11 PZNF-0038 HVDBS323-11 PZNF-0063 HVDBS348-11 PZNF-0089 HVDBS374-11 PZNF-0018 HVDBS303-11 PZNF-0039 HVDBS324-11 PZNF-0064 HVDBS349-11 PZNF-0090 HVDBS375-11 PZNF-0019 HVDBS304-11 PZNF-0040 HVDBS325-11 PZNF-0065 HVDBS350-11 PZNF-0091 HVDBS376-11 PZNF-0020 HVDBS305-11 PZNF-0041 HVDBS326-11 PZNF-0066 HVDBS351-11 PZNF-0092 HVDBS377-11 PZNF-0021 HVDBS306-11 PZNF-0042 HVDBS327-11 PZNF-0067 HVDBS352-11 PZNF-0093 HVDBS378-11 PZNF-0022 HVDBS307-11 PZNF-0044 HVDBS329-11 PZNF-0068 HVDBS353-11 PZNF-0094 HVDBS379-11 PZNF-0023 HVDBS308-11 PZNF-0045 HVDBS330-11 PZNF-0069 HVDBS354-11 PZNF-0095 HVDBS380-11

two sequences. All these analyses were done and Baker, 2001; Saunders and McDevit, 2013) in relative to the threshold used. If two specimens using the R program SPIDER (Brown et al., 2012). our dataset of CO1, H3 and combined regions. are more distant than threshold from each other, Here, MOTUs or genetic species are defined as but both are within threshold of a third, all three groups or clusters of specimens that are within the are contained in a single cluster or MOTU. Finally, DNA barcoding tests of species delimitation identified genetic threshold. Identifying MOTU in a all specimens grouped in each MOTU were matrix of DNA sequences is generally well identified using the function “lapply”, also Species delimitation based on DNA sequences indicated in a situation where the number of implemented in SPIDER program. requires a prior knowledge of a genetic distance morpho-species is unknown (Jones et al., 2011), as a threshold. We determined this genetic cut-off thus providing a proxy for the species richness in using the function "localMinima" implemented in a particular geographic region (Saunders and Tree reconstruction (MP Tree analysis) the R package SPIDER. Based on the threshold McDevit, 2013). This MOTU or "genetic species" distance identified, we evaluated the number of was determined using the function "tclust" also in A phylogenetic tree was reconstructed using MP "molecular operational taxonomic units" (MOTU; the package SPIDER. This function clusters approach (maximum parsimony). Tree Jones et al., 2011) or "genetic species" (Bradley specimens according to their genetic distances assemblages were done using heuristic searches

Franzini et al. 25

with 1000 random sequence additions, but keeping only RESULTS 10 trees per replicate to reduce time spent on branch swapping in each replicate. Tree bisection-reconnection Genetic distances was done with all character transformations treated as equally likely, that is, Fitch parsimony (Fitch, 1971). Cytochrome c oxidase subunit 1 dataset Phylogenetic Analysis Using Parsimony (PAUP*) (Swofford, 2003) was used for MP searches and Eighty-six (86) DNA sequences were generated for CO1. bootstrap (BP) resampling. BP values greater than 70% Aligned sequences were made up of 666 base pairs (bp) were considered as strong support (Hillis and Bull, 1993). with the following proportions: A: 28%; C: 14.3%; G: lemniscata (Walckenaer, 1841) (Araneae: 18.3% and T: 39.4%. The pairwise genetic distances vary Araneidae) was used as the out-group due to from 0 (zero) to 0.172 (mean=0.066) between specimens. morphologic similarities to Cyrtophora (Corey, 1993; Tajima’s K index of divergence between CO1 sequences Scharff and Coddington, 1997). Three sequences was 34.92 substitutions between two sequences. Based retrieved from GenBank were included, one for C. on variation within CO1 sequences, the optimised genetic moluccensis (Accession number: FJ607560; Blackledge cut-off for MOTU delimitation was estimated at 0.019 et al., 2009) and one for C. citricola (Accession number: (~2%). Based on this threshold, the barcoding approach DQ973168; Johannesen et al., 2006) for the CO1 matrix of species delimitation clusters all specimens in 10 (marked with an asterisk on Figure 2) and one for C. MOTU or "genetic species". moluccensis (Accession number: FJ607599; Blackledge The MP tree (Figure 2) distinguishes all specimens into et al., 2009) for the H3 matrix (marked with an asterisk on 10 clusters or MOTU (or genetic species) with strong Figure 3). The MP trees were used to determine species support (BP≥90). These MOTU are labelled I to X on the groupings, which were compared with those that resulted tree (Figure 2). The C. moluccensis (FJ607560) from DNA barcoding MOTU delimitation using SPIDER. specimen from GenBank (marked with an asterisk) clustered within the MOTU I (indicated in purple) with strong support (BP=90). However, the C. citricola Morphological characterisation (DQ973168) specimen also retrieved from GenBank grouped separately from the rest of the specimens Specimen examination included in the analysis (Figure 2). Divergence between all individuals averaged 6.68% General body features were studied and the abdomens (range: 0-16.79%; Tables 2 and 3). The distances within photographed for all specimens using a Zeiss KL2500 MOTU ranged from a minimum of 0 to 2.48% in MOTU IX LCD Stereo microscope (Carl Zeiss, SA) fitted with a Carl (Table 2). The individual distances between MOTU Zeiss Axio Camera (Carl Zeiss, SA). Special note was ranged from 2.58% between MOTU IX and X to 16.79% taken of abdominal characters including colour, tubercle between MOTU I and V (Table 3). The mean distance pronunciation and general shape. Male and female within MOTU (0.20%±0.27) was almost 44 times lower genitalia were dissected. The female genitalia were then than the mean distance between MOTU (8.78%±4.28). mounted temporarily on a slide in polyvinyl alcohol solution (PVA) (Walter and Kranz, 2009) and studied under an Olympus SZX7 dissection microscope Nuclear histone H3 genes (Olympus, Japan) with a Leica 2 light source (Leica Microsystems, Holland). Illustrations of the genitalia of For the H3 region, 39 DNA sequences were generated as males and females were drawn at a 6.3x10 magnification only positively identified male and female (no juveniles) freehand or with a drawing tube mounted on a Wild M5 spiders were included. The H3 sequences were stereo microscope (Wild Heerbrugg AG, Switzerland) at generated to allow comparison of results with those of 50× magnification. The external female genitalia were CO1. Aligned sequences were made of 389 bp. Base drawn in ventral and posterior view to show the epigynal composition was A: 27.5%; C: 26.9%; G: 24.2% and T: lip and openings. Illustrations of the male genitalia were 21.4%. The pairwise genetic distances varied from 0 to of the left . The dissected genitalia were placed 0.096 (mean=0.020) between sequences. Tajima’s K in glass vials and stored with the respective specimen. index of divergence between H3 sequences was 2.88 Only the abdominal shape and colouration was studied substitutions between two sequences. The threshold for because the abdomen is the main morphological feature H3 was 0.036 (~3.6%), and the 39 specimens were noticed in the web, especially when the spiders are hiding grouped into three MOTU, also labelled as I and III amongst debris. The abdomen has features, such as the following the labels adopted for CO1. three tubercle pairs characteristic for C. citricola, which The MP tree grouped MOTU I with III, with poor BP are not present in most other species of this genus support (60%). A newly identified MOTU in H3 analysis (Corey, 1993). that we referred to as MOTU XI included all the MOTU

Int. J. Mod. Biol. Res. 26

Figure 2. Maximum parsimony tree showing species orientation of Cyrtophora distinguished by iBOLD analysis of the CO1 gene. Species I is indicated in purple and species III in blue. GENBANK samples are indicated with an asterisk. The out- group is . Numbers below branches indicate BP values.

Franzini et al. 27

Figure 3. Maximum parsimony tree showing species orientation of Cyrtophora distinguished by the H3 gene. Species I is indicated in purple and species III in blue. The GENBANK sample is indicated with an asterisk. Numbers below branches indicate BP values.

Int. J. Mod. Biol. Res. 28

Table 2. Mean, standard deviation and range of intra-specific K2P distances between 10 Cyrtophora species groups generated from BOLD for the CO1 gene.

Clusters/MOTU identified Number of individuals Mean % divergence Standard deviation Range based on genetic distances in each MOTU I 2 0 0 N/A II 1 N/A N/A N/A III 1 N/A N/A N/A IV 37 0.16 ±0.14 0-0.46 V 14 0.15 ±0.32 0-1.08 VI 4 0.69 ±0.44 0-1.39 VII 11 0.54 ±0.24 0-0.92 VIII 7 0 0 N/A IX 6 1.36 ±0.69 0-2.48 X 1 N/A N/A N/A

N/A = Not applicable.

Table 3. Mean, standard deviation and range of inter-specific K2P distances for the 10 Cyrtophora species groups generated from BOLD for the CO1 gene.

Species Number of specimen Mean % divergence Standard deviation Range I 2 15.22 ±1.01 14.30-16.79 II 1 10.23 ±1.46 8.97-15.70 III 1 5.92 ±3.73 3.16-15.11 IV 37 8.04 ±4.01 3.01-14.30 V 14 11.14 ±4.38 3.11-16.79 VI 4 6.25 ±4.38 3.01-15.19 VII 11 7.47 ±4.06 4.54-16.26 VIII 7 9.58 ±3.28 3.11-15.13 IX 6 7.04 ±4.29 2.58-15.57 X 1 6.26 ±4.11 2.58-15.67

IV-X initially distinguished in the CO1 analysis. This and B (Figure 4). MOTU XI did not receive strong support (70%). The MP tree grouped all the specimens in a way similar C. moluccensis (FJ607599) specimen from GenBank to that of the CO1 analysis (Figure 4), except for MOTU II was also included in the H3 analysis. However, unlike the and X (which were not analysed for the H3 gene). The CO1 analysis, this specimen was separated from all the combined analysis resulted in 100% bootstrap values for analysed specimens (Figure 3). MOTU V and VII as in the CO1 analysis. The support decreased for MOTU IV (97 to 86%), VI (99 to 97%) and VIII (100 to 88%), and increased for MOTU I (90 to Combined 100%) and IX (90 to 97%). The number of MOTU indicated by the individual gene The combined DNA matrix included 39 sequences for analysis ranged from three (for H3) to 10 (for CO1) and which both CO1 and H3 sequences are available. The two for the combined analysis. In an attempt to search for aligned sequences are 1035 bp long composed of A: convergence between DNA and morphological data, we 27.7%; C: 18.8%; G: 20.50% and T: 33%. The pairwise also analysed the morphological characteristics of the genetic distances vary from 0 to 0.139 (mean=0.057). An abdominal and genital morphology of all the specimens. optimised threshold of 0.053 (~5.3%) was obtained for In spiders, as in many other invertebrates, the genital the combined genes, resulting in a total of three MOTU morphology is used as the main distinguishing identified. MOTU I (purple) was once again separated morphological character in alpha taxonomy (Huber, from all the other specimens. However, the MOTU XI that 2004). When comparing the variation in genital structure was isolated in H3 analysis was distinguished into two to the results of genetic analysis, we found evidence for separate groups that was labelled, MOTU XI groups A discrepancy between the two analyses (morphology and

Franzini et al. 29

Figure 4. Maximum parsimony tree showing species orientation of Cyrtophora distinguished by the combined analysis of both the CO1 and H3 genes determined by SPIDER (species I, XI(A) and XI(B)) and MP tree analysis (species I, III-IX) in comparison with genitalia assessment. Species I is indicated in purple and species III in blue. Species XI is separated into groups A and B.

Int. J. Mod. Biol. Res. 30

Figure 5. Abdomen of specimens of C. citricola (a-j) and Cyrtophora species I(k-l) [Location: Hermanus, Western Cape (a); Worcester, Western Cape (b, d); Randburg, Gauteng (e), Little Leigh, Limpopo (g); Makwassie, North West (f); Brandfort District, Free State (c, h-i); Kimberley, Northern Cape (j) and Malelane, Mpumalanga (k-l)].

genetic). The analysis of the genitalia indicated two citricola. In the abdomens, some dark forms exhibit light possible morphological categorisations of which one tear-shaped markings over the tubercles (Figures 5g and corresponds to the features of C. citricola and the other h). one corresponding to a potential new species that we The small sample size and uneven covering of identified as Cyrtophora species I. This Cyrtophora sampling efforts precludes us from linking the variation in species I corresponds to MOTU I. Abdominal variation abdominal colour patterns and genetic variation to the was considerable within C. citricola especially in terms of locality where specimens were collected. For example, the abdominal shape and pronunciation of tubercles, the tear-shaped markings were only recorded for some which can be influenced by immaturity, pregnancy and individuals from Little Leigh (Figure 5g) and Venetia food. Limpopo Reserve in Limpopo and Brandfort District in the Free State (Figures 5c, h and i) while in two other specimens collected from Venetia Limpopo Reserve and Cyrtophora citricola Brandfort District, this feature was absent (Figure 5c). All epigynes studied had an epigynal lip and cup- Thirty-seven (37) specimens, studied with both genes, shaped structures exhibiting concave inner margins are identified morphologically (using genitalia) as C. (Figure 6). The male were distinct in having a

Franzini et al. 31

Figure 6. Illustrations of female epigynes of C. citricola (a-b) and Cyrtophora species I (c-d). Vertical view: a and c; posterior view: b and d (Location: Venetia, Limpopo (a-b) and Malelane, Mpumalanga (c-d)). CD, coiled ducts; CS, cup-shaped structure; L, epigynal lip.

median apophysis and well developed conductor (Figure bifurculation at the posterior end, and a white band along 7). the anterior border of the abdomen (Figures 5k and l). The shape of abdomen is variable in C. citricola. In The female epigyne has a distinct spoon-shaped lip females, shape ranges from elongated to more oval with (Figures 6c and d). This species has no tubercle pairs on slight to pronounced bifurculation posteriorly and one to the abdomen, with the abdomen being thinner and longer three pairs of slight to pronounced tubercles. Colour than that of C. citricola. A single, anterior median tubercle varies from dark to light and intermediate. Some dark is present on the abdomen which was absent in C. forms exhibit light tear-shaped markings over the citricola specimens. The abdomen lacked the distinct tubercles (Figures 5g and h). Epigynes varied slightly in markings of C. citricola (Figure 5). The epigyne is more size but overall they were consistent in defining features flattened than that of C. citricola, and has a more such as the cup-shaped structures and short epigynal lip. pronounced spoon-like lip with the shape of the cup-like structure different from C. citricola (Figure 6).

Cyrtophora (Malelane) species I DISCUSSION All three genetic analyses (CO1, H3 and combined) clustered the two female specimens from Malelane in a Cyrtophora specimens have frequently been recorded single MOTU, which is clearly distinguishable from all during zoological surveys in South Africa. Due to the wide other MOTU on all three MP trees with strong support range of morphological variation in colour and shape, the (BP>90) (Figures 2, 3 and 4). However, the MP tree question was asked whether they all belong to the reconstructed using H3 gene did cluster a Madagascan cosmopolitan species C. citricola. This question has specimen with Cyrtophora species I with poor support never been investigated in the country. Our analysis of (BP=60). both CO1 and H3 genes showed different results to what The two female specimens, sampled from Malelane in has previously been thought, indicating that the South the Mpumalanga Province, have long and narrow African representatives of the genus Cyrtophora might be abdomens with an anterior median tubercle and slight more diverse than currently thought. To test this, we

Int. J. Mod. Biol. Res. 32

Figure 7. Illustrations of the male pedipalp of a specimen of C. citricola from Roodepoort, Gauteng: a) prolateral, b) retrolateral and c) ventral views. Key: E, Embolus; MA, medium apophysis; PC, paracymbium; C, cymbium.

included in our analysis CO1 DNA sequences of C. al., 2008). The latter two possibilities could account for citricola and C. moluccensis that we retrieved from our results given our high level of confidence in the GenBank; we also included in our H3 matrix, H3 DNA morphological identification of C. citricola. sequences of C. moluccensis also retrieved from In order to successfully distinguish between species GenBank. using genetic approach, the genetic distance between In the CO1 analysis, C. moluccensis grouped with species, that is, the inter-specific distance should be MOTU I whereas the same species was separated from significantly greater than the intra-specific distance. the rest of the specimens, including MOTU I in the H3 Comparing genetic distances between MOTU, our proxy analysis (Figure 3). Moreover, C. citricola was completely for morpho-species, we found that the mean inter-specific separated from all the specimens analysed for CO1 in the distances were at least five times greater than the present study, which were assumed to belong to this corresponding mean intra-specific distance. This result same species. There are three possible reasons for this indicates that the number of MOTU ranging between 3 finding: Either C. citricola was not represented amongst and 10, might truly reflect the pattern of species diversity the specimens collected and does not occur in South of the genus Citricola in South Africa. In other words, our Africa or CO1 region is not informative enough for analysis of genetic diversity indicates that there is likely species delineation in this group. It is also possible that more than one species of Citricola occurring in South the GenBank specimen was initially misidentified. Some Africa. Furthermore, a recent study on the spider genus studies have previously raised this issue of wrong Pimus Chamberlin, 1947 (Family ) reported labelling of DNA sequences on GenBank (e.g. Meier et mean inter-specific K2P distances ranging between 6.16

Franzini et al. 33

and 14.83% (Keith, 2010). In the present study, the range Foundation (NRF South Africa) and in part by the of genetic distances between MOTU of Citricola (5.92- Government of Canada through Genome Canada and 15.22%) is comparable to that reported by Keith (2010), the Ontario Genomics Institute (2008-OGI-ICI-03). further suggesting that MOTU might be a good surrogate Researchers of the African Centre for DNA Barcoding of morpho-species of Citricola, especially given that the especially Michelle van der Bank, Olivier Maurin, nodes subtending these MOTU on the tree were Barnabas Daru, Theresa Sethusa and Jephris Gere generally well supported. Nevertheless, it remains provided assistance during laboratory experiments. Petro possible that the pattern of genetic variation might be Marais, The Spider Club of Southern Africa, Robin Lyle, influenced by incomplete lineage sorting and geographic Collin Schoeman and Matjaž Kuntner provided us with structure (Madisson and Knowles, 2006). relevant literature, advice and technical supports. The To further investigate species delimitation within the following collaborators and organisations generously genus, we also considered a morphological angle, helped with collection of specimens used for this study: analysing the genitalia of mature specimens to assess Agricultural Research Council, National Museum of whether the differentiation between MOTU has a Bloemfontein, Ditsong Museum of Natural History, The morphological support. We found two broad Spider Club of Southern Africa, Charles Haddad, Astri morphological details in our collection, of which one Leroy, Roger Bills, Victor Hamilton-Atwell and Peter corresponds to epigynal characteristics consistent with Stephen. the morphological features of C. citricola (Corey, 1993; Levi, 1997; 2002), thus confirming the existence of C. citricola in South Africa. This finding casts doubt on the REFERENCES identification of the GenBank sequences of C. citricola. The second category of morphological features concerns Adams D. C., Berns C. M., Kozak K. H. & Wiens J. J. (2009). Are rates epigynal lip and cup-shape structures, and these features of species diversification correlated with rates of morphological evolution? Proc. R. Soc. B 276:2729-2738. were found only in our MOTU I. Overall, we found Barrett R. D. H. & Hebert P. D. N. (2005). Identifying spiders through congruence between genetic and morphological analyses DNA barcodes. Can. J. Zool. 83:481-491. for only two MOTU, suggesting confidently that at least Bickford D., Lohman D. J., Sodhi N. S., Ng P. K. L. & Meier R. (2007). two species can be distinguished, C. citricola and a Cryptic species as a window on diversity and conservation. Trends Ecol. Evolut. 22(3):148-155. potentially new species corresponding to MOTU I. An Blackledge T. A., Scharff N., Coddington J. A., Szuts T., Wenzel J. W., important question remains however: Why do we observe Hayashi C. Y. & Agnarsson I. (2009). Reconstructing web evolution a higher number of MOTU that cannot be distinguished and spider diversification in the molecular era. Proc. Natl. Acad. Sci. morphologically? One explanation could be linked to an U.S.A. 106(13):5229-5234. Bond J. E., Hedin M. C., Ramirez M. G. & Opell B. D. (2001). Deep incomplete lineage sorting. An alternative hypothesis molecular divergence in the absence of morphological and ecological could also be linked to the differences in rate of genetic change in the Californian coastal dune endemic trapdoor spider changes versus morphological differentiation. For Aptostichus simus. Mol. Ecol. 10: 899-910. instance, a rapid genetic variation can occur with no or Bradley R. D. & Baker R. J. (2001). A test of the genetic species concept: Cytochrome-b sequences and mammals. J. Mammalogy 82: little morphological changes (Adams et al., 2009) or vice- 960-973. versa (Wiens and Penkrot, 2002; Collar et al., 2010). Brown S. D. J., Collins R. A., Boyer S., Lefort M. C., Malumbres-Olarte Such evidence of contrasting patterns between genetic J., Vink C. J. & Cruickshank R. H. (2012). Spider: An R package for and morphological rate variation has been revealed in the analysis of species identity and evolution, with particular reference to DNA barcoding. Mol. Ecol. Resour. 12(3):562-565. earlier studies of species delimitation in several groups Collar D. C., Schulte J. A., O’meara B. C. & Losos J. B. (2010). Habitat including spiders (Bond et al., 2001). Clearly, there is a use affects morphological diversification in dragon lizards. J. Evol. need for further analysis to investigate whether there are Biol. 23:1033-1049. morphological changes in other organs different from the Corey D. T. (1993). A revision of the spider genus Cyrtophora (Araneidae). PhD Thesis, Southern Illinois University, United States commonly used genitalia that can corroborate the genetic of America. variation indicated by the isolation of MOTU. Pending Derocles S. A. P., Plantegenest M., Simon J. C., Taberlet P. & Le Ralec these additional analyses the current study provides A. (2012). A universal method for the detection and identification of evidence that the genus Cyrtophora is more diverse than Aphidiinae parasitoids within their aphid hosts. Mol. Ecol. Resour. 12:634-645. currently thought in South Africa. A thorough combination Dippenaar-Schoeman A. S. & Van den Berg A. (2010). Spiders of the of genetic variation, morphological and ecological Kalahari. Plant Protection Research Institute Handbook No. 18. information is urgently required for a better assessment Agricultural Research Council. Pretoria. of species richness of Cyrtophora in Southern Africa. Dippenaar-Schoeman A. S. (2006). New records of 43 spider species from the Mountain Zebra National Park, South Africa (Arachnida: Araneae). Koedoe 49(2):23-28. Dippenaar-Schoeman A. S., van der Walt A. E., de Jager M., Le Roux ACKNOWLEDGEMENTS E. & van den Berg A. (2005). The spiders of the Swartberg Nature Reserve in South Africa (Arachnida: Araneae). Koedoe 48(1):77-86. Edgar R. C. (2004). MUSCLE: Multiple sequence alignment with high This study was funded by the National Research accuracy and high throughput. Nucleic Acids Res. 32:1792-1797.

Int. J. Mod. Biol. Res. 34

Edwards D., Horn A., Taylor D., Savolainen V. & Hawkins J. A. (2008). Meier R., Zhang G. & Ali F. (2008). The use of mean instead of smallest DNA barcoding of a large genus, Aspalathus L. (Fabaceae). Taxon interspecific distances exaggerates the size of the “barcoding gap” 57(4):1317-1327. and leads to misidentification. Syst. Biol. 57(5):809-813. Edwards G. B. (2006). Cyrtophora citricola (Araneae: Araneidae), a Nadin-Davis S. A., Guerrero E., Kimberley K. M. & Feng Y. (2012). DNA colonial tent web orb weaver established in Florida. Entomology barcoding facilitates bat species identification for improved Circular 411:1-4. surveillance of bat-associated rabies across Canada. The Open Zool. Fitch W. M. (1971). Towards defining the course of evolution: Minimum J. 5(Suppl 1-M5):27-37. change for a specific tree topology. Syst. Zool. 20:406-416. Nagoshi R. N., Brambila J. & Meagher R. L. (2011). Use of DNA Foord S. H., Dippenaar-Schoeman A. S. & van der Merwe M. (2002). A barcodes to identify invasive armyworm Spodoptera species in check list of the spider fauna of the Western Soutpansberg, South Florida. J. Insect Sci. 11:154, 1-11. Africa (Arachnida: Araneae). Koedoe 45(2):35-43. Newmaster S. G. & Ragupathy S. (2009). Testing plant barcoding in a Forest F., Grenyer R., Rouget M., Davies T. J., Cowling R. M., Faith D. sister species complex of pantropical Acacia (Mimosoideae, P., Balmford A., Manning J. C., Proches S., Van der Bank M., Fabaceae). Mol. Ecol. Resour. 9:172-180. Reeves G., Hedderson T. A. J. & Savolainen V. (2007). Preserving Packer L., Gibbs J., Sheffield C. & Hanner R. (2009). DNA barcoding the evolutionary potential of floras in biodiversity hotspots. Nature and the mediocrity of morphology. Mol. Ecol. Resour 9:42-50. 445:757-760. Platnick N. I. (2012). The world spider catalogue. Version 11.5. Hajibabaei M., deWaard J. R., Ivanova N. V., Ratnasingham S., Dooh American Museum of Natural History. Accessed 18 April 2012. R. T., Kirk S. L., Mackie P. M. & Hebert P. D. N. (2005a). Critical Available online at http://research.amnh.org/iz/spiders/catalog doi: factors for assembling a high volume of DNA barcodes. Phil. Trans. 10.5531/DB.IZ.0001. R. Soc. B 360:1959-1967. Saunders G. W. & McDevit D. C. (2013). DNA barcoding unmasks Hajibabaei M., Janzen D. H., Burns J. M., Hallwachs W. & Hebert P. D. overlooked diversity improving knowledge on the composition and N (2005b). DNA barcodes distinguish species of tropical Lepidoptera. origins of the Churchill algal flora. BMC Ecology 13:9. Proc. Natl. Acad. Sci. 103(4):968-971. Scharff N. & Coddington J. A. (1997). A phylogenetic analysis of the Hebert P. D. N., Cywinska A., Ball S. L. & DeWaard J. R. (2003a). orb-weaving spider family Araneidae (Arachnida: Araneae). Zool. J. Biological identifications through DNA barcodes. Proc. R. Soc. 270: Linn. Soc. 120:355-434. 313-321. Sun X. Q., Zhu Y. J., Guo J. L., Peng B., Bai M. M. & Hang Y. Y. (2012). Hebert P. D. N., Penton E. H., Burns J. M., Janzen D. H. & Hallwachs DNA barcoding the Dioscorea in China, a vital group in the evolution W. (2004). Ten species in one: DNA barcoding reveals cryptic of Monocotyledon: use of matK gene for species discrimination. PLoS species in the Neotropical skipper butterfly Astraptes fulgerator. Proc. ONE 7(2):e32057. Natl. Acad. Sci. 101(41): 14812-14817. Swofford D. L. (2003). PAUP*: Phylogenetic analysis using parsimony (* Hebert P. D. N., Ratnasingham S. & DeWaard J. R. (2003b). Barcoding and other methods). Version 4.0b10. Sinauer. Sunderland, life: cytochrome c oxidase subunit 1 divergences among Massachusetts. closely related species. Proc. R. Soc. 270: S96-S99. Tajima F. (1983). Evolutionary relationship of DNA sequences in finite Hillis D. M. & Bull J. J. (1993). An empirical test of bootstrapping as a populations. Genetics 105:437-460. method for assessing confidence in phylogenetic analysis. Syst. Biol. Taylor H. R. & Harris W. E. (2012). An emergent science on the brink of 42:182-192. irrelevance: a review of the past 8 years of DNA barcoding. Mol. Ecol. Huber B. A. & González A. P. (2001). Female genital dimorphism in a Resour. 12(3):377-388. spider (Araneae: Pholcidae). J. Zool. London 255:301-304. Tikader B. K. (1982). Family Araneidae (=Argiopidae), typical Huber B. A. (2004). The significance of copulatory structures in spider orbweavers. Fauna India (Araneae) 2: 1-293. systematics. In: J. Schult (Ed.). Biosemiotik-praktische Anwendung Van der Bank F. H., Greenfield R., Daru B. & Yessoufou K. (2012). DNA und Konsequenzen für die Einzelwissenschaften. VWB Verlag, barcoding reveals micro-evolutionary changes and river system-level Berlin. Pp. 89-100. phylogeographic resolution of African silver catfish, Schilbe Jocqué R. (2002). Genitalic polymorphism - A challenge for taxonomy. intermedius (Actinopterygii: Siluriformes: Schilbeidae) from seven J. Arachnol. 30:298-306. populations across different African river systems. Acta Ichtyol. Pisc. Johannesen J., Lubin Y., Smith D. R., Bilde T. & Schneider J. M. 42(4):307-320. (2006). Direct submission. Institute of Zoology, Department of Walter D. E. & Krantz G. W. (2009). Collecting, Rearing, and Preparing Ecology, University of Mainz, Germany. Specimens; a Manual of Acarology. 3rd Ed. Texas Tech University Jones M., Ghoorah A. & Blaxter M. (2011). jMOTU and Taxonerator: Press. Texas. Turning DNA barcode sequences into annotated operational Wiens J. J. & Penkrot T. A. (2002). Delimiting species using DNA and taxonomic units. PLoS ONE 6: e19259. morphological variation and discordant species limits in spiny lizards Keith R. (2010). Species limits, molecular phylogenetics and historical (Sceloporus). Syst. Biol. 51(1):69-91. biogeography of the California spider genus Pimus (Araneae: Yaginuma T. (1968). The spider genus Cyrtophora of Japan. Amaurobiidae). MSc Thesis, San Diego State University, United Kansaishizenkagaku 20:34-38. States of America. Yassin A., Capy P., Madi-Ravazzi L, Ogereau D. & David J. R. (2008). Kimura M. (1980). A simple method for estimating evolutionary rates of DNA barcode discovers two cryptic species and two geographical base substitutions through comparative studies of nucleotide radiations in the invasive Drosophilid Zaprionus indianus. Mol. Ecol. sequences. J. Mol. Evolut. 16:111-120. Resour. 8: 491-501. Levi H. W. (1997). The American orb weavers of the genera Mecynogea, , Kapogea and Cyrtophora (Araneae: Araneidae). Bull. Mus. Comp. Zool. 155(5):215-255. Levi H. W. (2002). Keys to the genera of araneid orbweavers (Araneae: Araneidae) of the Americas. J. Arachnol. 30:527-562. Levy G. (1998). Twelve genera of orb-weaver spiders (Araneae: Araneidae) from Israel. Israel J. Zool. 43:311-365. Madisson W. P. & Knowles L. L. (2006). Inferring phylogeny despite incomplete lineage sorting. Syst. Biol. 55(1):21-30.