Institutional

AMNH (American Museum of Natural History; New York, New York), AUMNH (Auburn University Museum of Natural History; Auburn, Alabama), CAS (California Academy of Sciences; San Francisco, California), DUB (personal collection of Darrell Ubick; San Francisco, California), ICE (personal collection of Wendell Icenogle; Winchester, California), MCZ (Museum of Comparative Zoology, Harvard; Boston, Massachusetts), MEL (personal collection of Mel Thompson), MNHN (Muséum National D’Histoire Naturelle, Paris), SCW (Personal collection of Scott C. Williams, deposited in the AMNH), UCR (University of California Riverside; Riverside, California).

Quantitative morphological

ANTd: number of teeth on the anterior margin of the female cheliceral fang furrow.

Cl, Cw: carapace length and width (Fig. 11). Carapace length taken along the midline dorsal most posterior position to the anterior front edge of the carapace (chelicerae are not included in length). Carapace width taken at the widest point.

LBl, LBw: labium length and width taken from the longest and widest points, respectively (Fig. 12).

MF1, MT1, MM1, MA1: lengths of male leg I femur, tibia, metatarsus, and tarsus (Fig. 13).

MF4, MT4: length of male leg IV femur and tarsus (Fig. 14) taken from the prolateral aspect.

PTl, PTw: male palpal tibia length and width (Fig. 15).

Bl: palpal bulb length from embolus tip to the bulb base, taken in the ventral plane at its longest point (Fig. 16).

PTLs, TBs: number of female prolateral patella and tibial spines leg III.

STRl, STRw: sternum length and width. Sternum length from the base of the labium to its most posterior point. Width taken across the widest point, usually between legs II and III (Fig. 12).

TSrd, TSp, TSr: number of tibia I spines on the distal most retrolateral, prolateral, and midline retrolateral positions.

Measurement, characterization, and illustration of morphological features

Unique voucher numbers were assigned to all specimens (alphanumeric designations beginning with AP or MY); these data were added to each vial and can be used to cross-reference all images, measurement, and locality data. All measurements are given in millimeters and were made with a Wild M8 or Leica M9.5Z dissecting microscope equipped with an ocular micrometer scale. Appendage measurements, quantitative and meristic, were based on left appendages in the retrolateral (unless otherwise stated) view using the highest magnification possible. Measurements of large structures (e.g., leg article lengths, carapace and sternum dimensions, etc.) are accurate to 0.03–0.015 mm. Measurements of smaller structures (e.g., palpal bulb and labial dimensions) are accurate to 0.0075 mm. Lengths of leg articles were taken from the mid–proximal point of articulation to the mid–distal point of the article (sensu Coyle 1995 Figure 1; Figs 11–16). Leg I article measurements are listed in the species descriptions in the following order: femur, tibia, metatarsus, tarsus and metatarsus ventral excavation (Fig. 13); leg IV measurements are femur and tarsus only (given in that order). Carapace and leg coloration are described semi-quantitatively using Munsell® Color Charts (Windsor, NY) and are given using the color name and color notation (hue value/chroma).

Quantitative measurements are based on a minimum of five individuals of each sex when a sufficient number of specimens were available. When more than five specimens were available individuals were sampled from across the species’ geographical and size distribution (i.e., every attempt was made to select specimens that represented the range of sizes available in the collection). Species descriptions and material examined sections were generated using two simple Python scripts written specifically for generating homogenous species accounts [SPEEDM and MATex (Brewer and Bond (2012) available for download from the Dryad Data Repository at doi: 10.5061/dryad.3b95n)]. Characters for the phylogenetic analysis were scored from the type specimens, with the exception of the quantitative characters, which were scored on the basis of multiple specimens. Outgroup taxa were scored using the euctenizid specimens listed in Bond and Opell (2002) and in Bond and Hedin (2006) as exemplars. Quantitative values were taken from each of these exemplar taxa.

Mating clasper and palpal line drawings were made for some specimens (when needed to further clarify spination patterns) with the aid of a dissecting scope equipped with a camera lucida. Line drawings were scanned as digital images and converted to vector drawing objects in Adobe Illustrator (Adobe Systems Inc.). Digital images of specimens were made using a Visionary Digital Imaging System (Visionary Digital^TM, Richmond, VA) where images were recorded at multiple focal planes and then assembled into a single focused image using the computer program Helicon Focus (Helicon Soft, Ltd., Ukraine). The female genital region was removed from the abdominal wall and tissues dissolved using trypsin; spermathecae were examined and photographed in the manner described above. Habitus illustrations were constructed from whole body images that were bisected, copied, and reflected in Adobe Photoshop (Adobe Systems, Inc.) to produce a roughly symmetrical image; the actual raw image on which the habitus illustration is based has been deposited in Morphbank and its record number noted in the figure legend (value in square [ ] brackets). For scanning electron microscopy, specimens were air-dried and sputter-coated with gold prior to viewing on an FEI Quanta 200 scanning electron microscope.

Evaluation of quantitative morphological characters for diagnosis and phylogenetic analyses

Quantitative morphological features that were determined to have discrete, non-overlapping ranges for individual subsets of species were scored as phylogenetic characters and were used as features for morphological diagnosis of species. The criterion that employs only non-overlapping features limits the number of quantitative features and character states available to our analysis; these non-overlapping characters were chosen from a much larger suite of morphological measurements, many of which lacked discrete non-overlapping ranges (but may have differed statistically). Coyle (e.g., 1994, 1995) has frequently used an analysis of variance method to delineate states of quantitative characters for mygalomorph phylogenetic analyses. Mean values that statistically differ are scored as discrete states, regardless of potentially overlapping ranges. This approach is not uncommon and a number of other authors have likewise proposed methods for scoring overlapping quantitative characters (e.g., Chappill 1989, Goldman 1988, Thiele 1993, Wiens 2001, Goloboff et al. 2006). The use of these quantitative data has received some “conceptual” scrutiny, both negative (Farris 1990) and positive (e.g., Goloboff et al. 2006; see Garcia-Cruz and Sosa 2006 for detailed assessment of various approaches). However, the utility of overlapping quantitative character states has not been adequately tested (but see Hendrixson and Bond 2009). For this reason and others, I am hesitant to use these data as such in this present study. However, acknowledge that the approach I have employed here is not without certain problems. For example, in the case of species represented by only a few specimens, additional collecting could add specimens whose features expand the range of some characters and possibly negate, or change the scoring of said characters.

Morphological characters scored

General morphological and spinning features

1. Thorax: flat = 0; sloping = 1 (see Bond and Beamer 2006; figs 1A, 1C).

2. Carapace pubescence: absent = 0; light = 1; heavy = 2 (Fig. 96).

3. Posterior edge of male carapace: aspinose = 0; with a distinct fringe of heavy spines and/or setae = 1 (Fig. 53).

4. Posterior thorax sclerotization: normal = 0; light = 1. See Bond and Opell (2002) for a detailed explanation of this character and its states.

5. AME and PME: subequal in diameter = 0; AME diameter greater = 1.

6. Eye tubercle: absent = 0; present, low = 1; present, high = 2.

7. Male thoracic groove: transverse = 0; recurved = 1; procurved = 2.

8. Sternal shape (Fig. 17): normal (STRw/STRl > 74.0) = 0; rounded and raised in the ventral plane = 1; long (STRw/STRl < 73.8) = 2.

9. Rastellum: on a distinct process = 0; consisting of large spines not on a process = 1.

10. Rastellar spines: normal = 0; enlarged = 1.

11. Rastellum retrolaterally offset spine (Fig. 189): absent = 0; present = 1.

12. Endite cuspules: restricted to medial posterior aspect = 0 (Fig. 256); widespread = 1.

13. Male labial cuspules: absent = 0; present = 1.

14. Male palpal endite cuspules: absent = 0; present = 1.

15. Labium shape (Fig. 18): wider than long/subquadrate (LBw/LBl > 42.5) = 0; very wide (LBw/LBl < 42.5).

16. Sternal sigilla: large = 0 (Fig. 188); small = 1 (Fig. 256).

17. Sternal sigilla: widely spaced = 0; closely spaced or contiguous = 1 (Fig. 188).

18. Carapace coloration: light = 0 (Figs 88–90); dark = 1 (Figs 79–81).

19. Abdominal color pattern: solid or with solid striping = 0; mottled striping = 1 (Fig. 80).

20. Abdomen coloration: light = 0 (Fig. 105); dark = 1 (Fig. 95).

21. Cheliceral dentition: single promarginal row of large teeth with retromarginal row of small denticles = 0; both margins with larger teeth = 1.

22. Pumpkiniform spigots (see Bond and Opell 2002): absent = 0; present = 1.

23. Spigot bases: with invaginations = 0; without invaginations = 1.

Male leg and microstructural characters

24. Tarsus IV length (Fig. 19): short (MA4/MF4 < 60.0) = 0; long (MA4/MF4 > 60) = 1.

25. Tarsus I pseudosegmentation (Fig. 279): absent = 0; present = 1.

26. Tarsus I: straight = 0; curved = 1 (Fig. 308).

27. Tarsus IV pseudosegmentation: absent = 0; present = 1.

28. Tarsus IV: straight = 0; curved = 1.

29. Tarsus I: stout (diameter equal to or greater than metatarsus) = 0; slender (diameter less than metatarsus) = 1.

30. Tarsus ventral spines: absent = 0; present = 1 (Fig. 311).

31. Leg I coloration: uniform = 0; distal 1/2 metatarsus and tarsus light in color = 1.

32. Tibia I length (Fig. 20): short (MTI/MFI < 79.53) = 0; long (MTI/MFI > 79.53).

33. Metatarsus I length (Fig. 21): short (MMI/MFI < 77.5) = 0; long (MMI/MFI > 77.5) = 1.

34. Tarsus I length (Fig. 22): short (MAI/MFI < 49.72) = 0; long (MAI/MFI > 49.72) = 1.

35. Tarsal scopulae: thin = 0; thick = 1.

36. Tarsal scopulae on leg IV: present = 0; absent = 1.

37. Bifid STCI basal tooth: absent = 0; present = 1.

Female leg and microstructural characters

38. Female tarsal scopulae: light = 0; dense = 1.

39. Metatarsus IV preening comb: absent = 0; present = 1 (Figs 45, 62).

40. Metatarsus III preening comb: absent = 0; present = 1.

Secondary sexual and genitalic characters

41. Palpal tibia (Fig. 23): stout (PTw/PTl > 50) = 0; slender (PTw/PTl < 50) = 1.

42. Palpal tibia (Fig. 24): short (PTl/Cl < 36) = 0; long (PTl/Cl > 36) = 1.

43. Palpal tibia spines: long and ventrally positioned = 0 (Fig. 71); short and retrolaterally positioned = 1 (Fig. 278).

44. Palpal tibia megaspines: present = 0 (Figs 341, 349); absent = 1.

45. Male tibia I ventral megaspine: absent = 0; present = 1.

46. Male metatarsus I mating apophysis: absent/non - distinct = 0 (Fig. 333); rectangular = 1 (Figs 149, 160); triangular = 2 (Fig. 192); triangular and hooked = 3 (Fig. 69).

47. Male metatarsus I mating apophysis spine: absent = 0; present = 1 (Fig. 160).

48. Male metatarsus I: straight = 0; anteverted when viewed in retrolateral aspect = 1 (Fig. 266).

49. Male metatarsus I proximal excavation: absent = 0 (Fig. 333); present = 1 (Fig. 107).

50. Embolus serration: absent = 0 (Figs 68, 72); present = 1 (Figs 49, 50, 277).

51. Embolus shape: single bend = 0 (Fig. 126); sigmoidal = 1 (Fig. 72).

52. Embolus: thin = 0 (Fig. 72); stout = 1 (Fig. 277).

53. Embolus shape: cylindrical = 0 (Fig. 72); dorsal - ventrally compressed = 1 (Fig. 277).

54. Sperm duct directly below bulb embolus junction: straight = 0; looped = 1.

55. Tip of embolus: normal, gradual taper = 0; tapers sharply into a very thin terminus = 1 (Fig. 335).

56. Male pedipalp distal prolateral tibial spine: absent = 0; present = 1 (Fig. 271).

57. Palpal bulb (Fig. 25): short (Bl/Cl < 17) = 0; long (Bl/Cl > 17) = 1.

58. Prolateral cymbial lobe: normal = 0; extended = 1.

60. Retrolateral, distal most aspect of cymbium forms a distinct process: no (normal) = 0; yes = 1 (Fig. 278).

61. Retrolateral cymbium spine row: absent = 0; present = 1 (Figs 246, 248).

62. Retrolateral distal tibial spines: absent = 0 (Fig. 165); present = 1 (Fig. 55).

63. Retrolateral distal tibial spines: absent = 0; short = 1 (Fig. 194); long = 2 (Fig. 123).

64. Retrolateral distal spines: absent = 0; arranged distally = 1 (Figs 73–75); offset behind distal margin = 2 (Fig. 167).

65. Retrolateral distal tibial spines: uniform, non-overlapping = 0 (Figs 73–75); uniform = 1 (Fig. 160); absent = 2.

66. Tibia I, 1 - 1 - 1 spination pattern: absent = 0; present = 1 (Fig. 70).

67. TSr (Fig. 26): few (TSr < 10) = 0; many (TSr > 10) = 1.

68. TSp (Fig. 27): few (TSp < 7) = 0; many (TSp > 7) = 1.

69. Spines on prolateral surface of male patella I: few (<7); many large spines (>7; Figs 139, 144).

70. Spermathecal lateral base: absent = 0; present = 1 (Figs 76–78).

71. Secondary spermathecal bulb: absent = 0; present = 1.

72. Median spermathecal stalk: short, approximately as long as wide = 0 (Fig. 76); long, much longer than wide = 1 (Fig. 202).

73. Median spermathecal bulb: large (exceeds diameter of median stalk) = 0 (Fig. 76); small (diameter of bulb and median stalk subequal) = 1 (Fig. 202).

74. Median spermathecal stalk: straight = 0 (Fig. 76); sinuous = 1 (Fig. 203).

Morphological and molecular character matrix construction and phylogenetic analyses

Phylogenetic analyses of Aptostichus relationships were conducted using molecular and morphological data sets employing parsimony, Bayesian, and likelihood optimality criteria. Molecular data sets analyzed include data drawn from previous smaller studies (Bond et al. 2001, Bond and Stockman 2008, Bailey et al. 2010) and from newly collected data. Handling of tissues, DNA preparation, and sequencing followed procedures outlined in Bond and Stockman (2008). Legs were removed from specimens and preserved in RNAlater (Qiagen, Valencia, CA) and stored at −80C; whole specimens were preserved for morphological studies in 80% ethanol. Genomic DNA was extracted using the Qiagen DNeasy Tissue Kit. Standard PCR protocols were used to amplify an approximately 1500–base pair region of the mitochondrial genome spanning the region coding for the 12S rRNA, val-tRNA, and 16S rRNA genes. Amplification products were produced using the primers LR-J-12887 CCGCTCTGAACTCAGATCACGT and SR-N-14612spid AAGACAAGGATTAGATACCCT. After agarose gel verification and purification using ExoSAP-IT (USB, Cleveland, Ohio), products were sequenced on an ABI 3130 automated sequencer (Applied Biosystems, Foster City, CA) directly using the PCR primers and an additional internal sequencing primer (LR-J-13XXXa GGCAAATGATTATGCTACC). Sequence data were edited using the computer program Sequencher (Genecodes, Madison, WI) and then aligned using the software package Muscle (Edgar 2004) with default opening and gap extension penalties; minor adjustments to the alignment were made manually using the computer program Mesquite (Maddison and Maddison 2010). Molecular analyses primarily focused on identification of specimens, particularly juveniles, to species by phylogenetic association; that is, evidence of common ancestry with specimens identified by other means (e.g., morphological or cohesion species criteria). Given the paucity of species for which molecular data were available (less than half), it would be unwise to infer intra-generic relationships among species from these analyses.

Phylogenetic analyses of the molecular data comprised Bayesian and likelihood analyses. For Bayesian analyses the appropriate model of DNA substitution for each of the mtDNA data partitions was chosen using the computer program Kakusan 3 (Tanabe 2007) by Bayesian Information Criterion (BIC). MrBayes ver. 3.1.2 (Ronquist and Huelsenbeck 2003, Huelsenbeck and Ronquist 2001) was used to implement a Bayesian inference of phylogeny using the substitutions models indicated by BIC. Analyses comprised four concurrent Markov Chain Monte Carlo (MCMC) chains run for 40, 000, 000 generations with trees saved every 1, 000 generations. The two independent runs were considered to have converged when the standard deviation of split frequencies value was < 0.01. Topologies were discarded as burn-in following visual inspection in the program Tracer (Rambaut and Drummond 2007); clade posterior probabilities were computed from the remaining trees. The reported likelihood score and post burn-in tree topology was computed using the sump and sumtcommand with the option contype=allcompat, respectively. Likelihood analyses were conducted using the computer program RAxML ver. 7.2.8 (Stamatakis 2006). Parameters employed in the analysis included 1000 random addition sequence replicates with the general time reversible model and gamma model of site heterogeneity (-m = GTRGAMMA, -# = 1000). Branch support values were computed via 1000 non-parametric bootstrap replicates. Bootstrap bipartitions were drawn onto the best tree topology obtained in the previous analysis.

Phylogenetic analyses of the morphological data set were performed using PAUP* version 4.0b10 (Swofford 2002). All binary characters were treated as reversible, multistate characters and were treated as unordered, and all characters were initially weighted equally. Heuristic searches were performed using random stepwise addition (1000 replicates) of taxa followed by TBR (tree bisection-reconnection) branch swapping held to one million rearrangements per replicate. Branches with a maximum length of zero were collapsed. The preferred tree topology (presented herein) is based on the search conducted in PAUP* using the “Goloboff Fit Criterion” (Goloboff 1993a, b, c, 1995) with the search parameters described above. Solutions using an array of concavity function constants (k= 3-12) were explored.

For the purposes of evaluating the evolution of habitat type, species were scored for seven habitat type character states: (0) mixed forest and coastal range; (1) chaparral; (2) alpine meadow; (3) desert; (4) coastal dune; (5) mixed redwood; (6) dry steppe. Character scorings were based on personal observations and ecoregion types assessed in the computer program ArcGIS (ESRI, Redlands, CA) using the 2007 EcoRegionsCalifornia07_3 GIS data set (Cleland et al. 2007) downloaded from http://www.fs.fed.us/r5/rsl/projects/gis/data/calcovs/EcoregionsCalifornia07_3.html . Taxa found in multiple habitat types were scored as polymorphic. Character state reconstructions using parsimony were evaluated on the preferred tree topology (Fig. 29) using the computer program MacClade ver. 4.05 (Maddison and Maddison 2000).

Locality data, georeferencing, generation of niche-based distribution models, and conservation status

Latitude and longitude for all collecting localities were recorded in the field using a Garmin® Global Positioning System receiver (Garmin International Ltd., Olathe, KS) using WGS84 map datum. For previously collected specimens (e.g., loaned museum specimens) locality data were georeferenced by hand by finding the approximate locality on United States Geological Survey topographic maps (NAD83 map datum) or Google Earth (WGS84 datum). All georeferenced and field recorded locality data (latitude, longitude, elevation) were crosschecked by hand in Google Earth prior to generating distribution map illustrations and database entry. Distribution maps were constructed using ArcGIS using NAD83 map datum. Because many older collecting labels lack sufficient locality information, many georeferenced values are imprecise and should be used with caution. Data for labels that document only county and/or town information were georeferenced to the approximate geographic center of the locality given. Precision for each georeferenced point is annotated as a superscript in each material examined section of the species’ taxonomy using the confidence value scheme employed by Murphey et al. (2004): 1 = exact coordinates given; 2 = amended exact coordinates (i.e., exact coordinates given but were emended on validation); 3 = public land survey system (or herein geographic place name); 4 = within 1km radius; 5 = within 5km radius; 6 = within 10km radius; 7 = to county or > 10km; 8 = to state; 9 = to project region. Latitude and longitude are recorded to the 4^th decimal place as an indication of the precision in the point assigned by us (i.e., where I have assigned the locality place-holder for the specimen in question), not precision in the recording of the value or to specify the exact point of collection. Actual precision of the record is to be inferred from the numbering system described above. Detailed locality and associated GIS data as supplemental data files in spreadsheet and KML file format can be downloaded online from the Dryad Data Repository at doi: 10.5061/dryad.3b95n.

As an approach to facilitating species discovery and determination, niche-based distribution models (DM’s) were constructed for species for which sufficient locality data were available (> 5 points). Niche-based DM’s provide estimates for the probability of finding a species at a location on the landscape given the set of correlate ecological and climatic parameters used to construct the model. Locality coordinates for each specimen were imported into ArcMap (ESRI, Redlands, CA) and converted into shape files. Following the procedure outlined in Bond and Stockman (2008), DM’s were constructed using environmental layers thought to “likely influence the suitability of the environment” (Phillips et al. 2006) based on our previous analyses of Aptostichus atomarius species complex distributions (see Stockman and Bond 2007, for further justification of layer choice). Seven climatic layers were obtained from the WORLDCLIM data set (Hijmans et al. 2005): annual precipitation, precipitation seasonality, annual maximum temperature, annual minimum temperature, temperature seasonality, and mean precipitation during the driest and wettest quarters. A seventh layer, elevation, was constructed from a mosaic of Digital Elevation Models (DEMs) derived from the National Elevation Dataset (USGS). DEMs were converted to Raster format in ArcMap and resampled from 30-m resolution to 1-km resolution using bilinear interpolation. All seven layers were clipped to the same extent, cell size, and projection. Niche-based (DMs) were created using the computer program Maxent (Phillips et al. 2006). Maxent employs a maximum likelihood method that estimates a species’ distribution that has maximum entropy subject to the constraint that the environmental variables for the predicted distribution must match the empirical average (Elith et al. 2006; Phillips et al. 2006). Parameters for all Maxent analyses used the default values: convergence threshold = 10−5, maximum iterations = 500, regularization multiplier = 1, and auto features selected. Additional larger values of the regularization multiplier were used to ensure that models were not overfitting the data.

A hypothesized conservation status of all species has been included with each description. I have used the NatureServe ranking scheme–secure, apparently secure, vulnerable, imperiled, critically imperiled, and extinct–to describe perceived status. The designations provided are not based on any formal calculations (see Faber-Langendoen et al. 2009 for explicit criteria) and thus should not be viewed as formal status declarations. Many Aptostichus species are rare in collections and very difficult to collect, consequently parameters like rarity and abundance are impossible to accurately assess at this time. As such, I have based conservation status designations on extent of distribution and apparent threats to habitat (e.g., is the species known only from a highly impacted area?); a brief rationale is provided with each determination. These designations are likely to be very conservative and may belie the imperiled nature of some species. Future studies will seek to formally evaluate conservation status.

Species delimitation and conceptualization

Although often not discussed, any taxonomic revision contains an implicit concept of a species. The general convention within spider taxonomy is a “diagnosable” species concept (Nelson and Platnick 1981, Nixon and Wheeler 1990) wherein populations are delineated as species as a consequence of sharing a set of qualitative, fixed differences. For the vast majority of spider species described these differences are essentially morphological, usually features related to differences in genitalia (male pedipalps, female spermathecae). Alternatively, one could simply think of this as a “morphological” species construct–if populations differ in a set of morphological characters they are delimited as separate species.

As discussed in number of papers related to species delineation in mygalomorph taxa, morphological stasis seems to be more the rule rather than the exception. That is, the prevailing hypothesis is that extreme geographic structuring due to limited dispersal capability may lead to speciation in the absence of morphological or apparent ecological divergence (Bond et al. 2001). As such a number of studies (e.g., Stockman and Bond 2007, Bond and Stockman 2008, Bailey et al. 2010) advocate that an integrative approach to mygalomorph spider species delimitation must be employed if an accurate representation of evolutionary diversity is to be achieved.

Ideally an integrative approach to species delimitation would use data from many sources (e.g., genetic, ecological, and traditional morphological) taken together to formulate species hypotheses. Such an approach serves to more thoroughly document evolutionary diversity whereas an approach that focuses on a single character system (e.g., genitalic features) may overlook species diversity but is easier to implement and is pragmatic both in terms of species documentation and discovery, and subsequent identification by non-specialists. Reflected in the amount of data available for any given set of specimens/populations, the species reported herein generally represent one of three construct classes–morphological, or traditionally delineated species, phylogenetic species, or “cohesion species”. Traditional morphologically delineated species are defined as those populations that represent qualitative differences in phenotype that differ in a discrete manner from other populations groups. Cohesion species follow Templeton’s (1989) concept wherein a species is defined as lineages that are genetically or demographically interchangeable. Cohesion species are those for which genetic and or ecological data has been considered in concert with the distribution of species (see Bond and Stockman 2008) and morphological characteristics. Each species’ account notes the species concept that has been applied. In a number of instances additional phylogenetic information, based on molecular analyses was available and considered as corroborative support for the species hypothesis being put forth. As such I have applied a phylogenetic species concept wherein individuals (populations) share common ancestry in a molecular phylogenetic analysis and thus are mutually exclusive and diagnosably distinct (Cracraft 1989). Operationally, I simply employ the phylogenetic information from this single gene analysis as corrabotive support for hypothesized morphological species when they are recovered as a clade on the gene tree; that is, these data are not necessarily used exclusively to delineate species.

Data resources

The data underpinning the analysis reported in this paper (see below) were deposited on 19 November 2012 in the Dryad Data Repository at doi: 10.5061/dryad.3b95n and at GBIF, the Global Biodiversity Information Facility, http://ipt.pensoft.net/ipt/resource.do?r=aptostichus_locality_data . Images associated with species descriptions have been deposited in Morphbank (http://www.morphbank.net ); Morphbank image record numbers are noted in brackets by each figure in the figure legend.

Summary of taxodiversity

At present the genus Aptostichus comprises 40 species, 33 of which are newly recognized herein. Table 1 summarizes species, type localities, and material available for each species described; >2000 specimens in total were examined. Of these 33 new species, 12 are known only from male specimens; of these, three are described on the basis of a single specimen. Such a sex-based disparity and lack of material for rare taxa has been noted in taxonomic revisions of other mygalomorph groups (e.g., the migid genus Moggridgea O. P. Cambridge, 1875; Griswold 1987) and is not uncommon in systematic works in general (Huber 2003, Lim et al. 2011). As already discussed the number of species will likely increase dramatically as more is learned about this group of spiders; molecular studies to date suggest that “morphological” species, particularly those that are widely distributed, likely disguise a great deal of evolutionary and ecological diversity. Moreover, some underestimation may in part be due to the lack of males collected for populations of unplaced female specimens.

The genus Aptostichus has diversified within an extensive area that spans the California Floristic Province. Species are found in virtually every habitat type (see discussion of ecological evolution below) including extreme arid desert environments, mesic montane and coastal habitats, to high elevation alpine habitats of the Sierra Nevada Mountains. Without question the genus represents a classical adaptive radiation where lineages have diversified and apparently adapted to inhabit a set of disparate environments, climates, and habitat types. Ecological factors that may influence Aptostichus diversity and distributions include climatic suitability during dispersal, prey type and availability, water and temperature, and soil type (to name a few). Given their close ties to the substrate, as fossorial organisms, parameters associated with burrow architecture and design (depth, thickness of silk lining, trap door design, etc.), it is not surprising that many of these features vary from species to species.

Understanding of Aptostichus ecology and behavior is severely limited at this time. To date approximately 15 of the 40 species have been collected only from pitfall traps thus female burrows have never been observed. Of these about half are known from only a few specimens and thus are quite rare in collections. Moreover, some species like Aptostichus derhamgiulianii and Aptostichus sierra have not been collected since they were first discovered over 40 years ago and at least two species are now presumed extinct (Aptostichus killerdana, Aptostichus lucerne). Alternatively, recent collecting efforts have uncovered a number of new morphologically distinct species (e.g., Aptostichus satleri, Aptostichus isabella, Aptostichus cajalco). Given how narrowly endemic many species are there is likely to be considerable diversity that both awaits discovery but is also threatened due to development and habitat destruction throughout the California Floristic Province biodiversity hotspot.

Aptostichus phylogeny

Figure 28 summarizes the maximum likelihood and Bayesian inference analyses of the mtDNA data set. The data set comprises 337 individual specimens representing 15 of the 40 species documented herein, sequenced for the mitochondrial region spanning the 16S-tRNA-valine-12S genes; of these 206 were newly generated sequences (GenBank Accession numbers: JX103235-JX103440). The matrix is an aligned 1618 base pairs; Kakusan chose the GTRGAMMA model for each of the three partitions. The –ln likelihood value for the best tree from the RAxML analysis is -59821.7255. The Bayesian analysis was run for 40, 000, 000 generations with half of the trees discarded as burnin. The harmonic and arithmetic means of the post-burnin tree topology likelihoods were -60532.81 and -60373.61, respectively. The summary tree (Fig. 28) supports the basal placement of Aptostichus simus, the monophyly of the Atomarius Sibling Species Complex, and general placement of other species into species groups delineated on the basis of the morphological analysis (see below). However, as discussed earlier these data were used principally to place undetermined individuals into species rather than formulate hypotheses of deeper relationships across the tree; that is, many of the internal nodes are not strongly supported. Using the results from this analysis in combination with specimens, particularly males that could be identified to species, many female and juvenile Aptostichus atomarius, Aptostichus stanfordianus, Aptostichus angelinajolieae, Aptostichus icenoglei, and Aptostichus barackobamai specimens were determined. This analysis indicates that Aptostichus stanfordianus likely comprises more than one species (previously noted by Bond and Stockman 2008) and that specimens collected from Madera County may be an undescribed species related the Aptostichus simus (A. sp. Madera, Fig. 28). However, mature males and females are not currently available and thus the status of these specimens remains unanswered.

The morphological matrix comprised 72 characters scored for all 40 Aptostichus species. Figures 17–27 summarize the quantitative character scorings for all taxa but Aptostichus satleri, Aptostichus isabella, and Aptostichus sinnombre. Table 2 summarizes the results for the parsimony analysis based on characters equally weighted (EW) and the implied weighting analyses (IW) using a range of concavity function constants (k=3-12). The EW analysis resulted in > 440, 000 trees comprising 238 steps whereas the implied weights analyses resulted in considerably fewer trees with tree lengths ranging from 238–241 steps.

The preferred tree topology is based on the analysis using IW with a concavity function constant of k=7 (Fig. 29). The EW and IW analyses were moderately incongruent with respect to the tree topologies. The EW analysis failed to recover an Aptostichus clade that was monophyletic with respect to Apomastus. A strict consensus of the >440, 000 resulted in a largely unresolved tree (towards the tips) that recovered the Sierra, Simus, and Hesperus clades, or species groups (Fig. 29); the Atomarius clade was paraphyletic with respect to the Hesperus clade. Implied weighting analyses that employed concavity function constants k=3-5 resulted in trees that recovered a monophyletic Aptostichus, Sierra, Simus, and Hesperus clades but like the EW analysis the Atomarius clade was paraphyletic. Tree topology stabilized for IW analyses where k=6-12 recovering a pattern similar to that illustrated in Figure 29 where all four major species groups are monophyletic. These clades are largely delineated on the basis of shared male mating clasper and pedipalp differences (illustrated for each species group in Fig. 29).

Both molecular and morphological data matrices and associated trees (all trees recovered and consensus), formatted as Nexus and Phylip files can be downloaded from the Dryad Data Repository at doi: 10.5061/dryad.3b95n..

Character support of major clades and Aptostichus species groups

Table 3 summarizes the unambiguous character state support for each of the major nodes in the preferred tree topology (Fig. 29). I summarize below the support for only the major nodes in the analysis and formally diagnose the four nominal Aptostichus species groups. At this time it would be premature to overemphasize resolution among all of the terminal relationships within Aptostichus because of the incomplete nature of the data set due to missing taxa and few female representatives of some species.

Four characters (given parenthetically following the description of the state) provide unambiguous support for the monophyly of Aptostichus: a mottled, striped abdominal color pattern (19), distal 1/2 of the male metatarsus I lighter in color (31), extended prolateral cymbial lobe (58), and a cymbium with spines (59). Two characters support the monophyly of the clade that comprises the Sierra and Simus species groups: a short male palpal tibia (42), and an embolus with a single distinct bend (51). Six synapomorphies support the node that unites the Hesperus and Atomarius species groups: a anteverted male metatarsus I (48), a long palpal bulb (57), the presence of a male retrolateral distal tibial spine (62), long male retrolateral distal tibial spines (63), a triangular male mating apophysis (64), and uniform, non-overlapping male retrolateral distal tibial spines (65).

Sierra Species Group.Four species comprise the Sierra species group, which is supported by two synapomorphies: long sternum (8) and a long male metatarsus I (33).

Sierra Species Group.Eight species comprise the Simus species group, the monophyly of which is supported by six synapomorphies: absence of cuspules on male endites (14), male palpal tibia stout (41), male palpal tibia spines short and positioned retrolaterally (43), stout embolus that is dorsal - ventrally compressed (52, 53), and retrolateral, distal most aspect of the cymbium formed as a distinct process (61).

Hesperus Species Group.Thirteen species comprise this diverse species group. The key distinguishing feature of this group is the presence of an offset retrolateral rastellar spine (character 11). Additionally, four other characters support the monophyly of this species group: lighter carapace and abdominal coloration (18, 20) and a long and sinuous spermathecal stalk (72, 74).

Atomarius Species Group.Fifteen species comprise the Atomarius species group, the monophyly of which is supported weakly by three synapomorphies: heavy carapace pubescence (2), dense female tarsal scopulae (38) and a distinct secondary spermathecal bulb (71).

Desert adaptation in Aptostichus

Aptostichus is an ideal group for evaluating changes in spider morphology and behaviors associated with invasions of arid, desert and other habitat types (e.g., coastal dunes). Figure 30 maps habitat type on the preferred tree topology (Fig. 29) using parsimony; alternative optimization criteria were not possible due to the polytomies in the preferred tree. The current phylogeny requires at least three independent derivations of strictly desert habitation for 15 species occurring in three of the four species groups. Additional independent derivations of arid habitat are required if chaparral is classified similarly to desert habitats (i.e., all arid habitats are grouped together).

Main (1978) suggested eight adaptations that may allow trapdoor spiders to survive in very arid habitats: 1) larger body size, 2) deeper burrows, 3) increasing foraging area achieved by burrow rim modifications, 4) differential timing of breeding and dispersal, 5) the tendency of brooding females to plug their burrows, presumably for water retention, 6) aestivation of young in sealed burrows, 7) mature non-brooding females that do not plug burrows and are therefore able to feed sporadically, and 8) increased longevity of females. Concentrated efforts to obtain female specimens and additional natural history data will be required to address these questions more thoroughly for this group. Moreover, a complete molecular phylogeny for the genus will likely go further to resolve relationships among the species and will help to abrogate any confounding issues that morphological characters and attendant homoplasy associated with habitat may have.