Conservation genetics of three flightless beetle species in southern California
- First Online:
- Cite this article as:
- Caterino, M.S. & Chatzimanolis, S. Conserv Genet (2009) 10: 203. doi:10.1007/s10592-008-9548-7
- 158 Views
Regional scale conservation decisions can be aided by information on the distribution of intraspecific diversity, especially the extent to which patterns are common to multiple species. We compare patterns of intraspecific mitochondrial cytochrome oxidase I (COI) variation among three flightless beetles (Coleoptera: Tenebrionidae: Nyctoporis carinata LeConte; Staphylinidae: Sepedophilus castaneus (Horn); Carabidae: Calathus ruficollis Dejean) in the southern part of the California Floristic Province biodiversity hotspot. All species exhibit moderate to high levels of total variation, ranging from 2% to 10% (maximum uncorrected distance). Most populations of all species exhibit unique haplotypes, but few populations’ haplotypes constitute exclusive clades. Many adjacent pairs of populations show indications of some, though limited, genetic connectedness, due either to gene flow or ancestral polymorphism. However, in most cases this diminishes sharply over greater distances. By both statistical and phylogenetic measures, Sierra Nevadan populations are highly distinct from those in the coast and transverse ranges. Among the latter, the eastern transverse ranges are generally most unique and isolated, with diversity in the western parts of these ranges showing fewer barriers. Otherwise, few measures agree on areas of highest conservation value, and overall patterns tend to be species-specific.
KeywordsColeoptera TenebrionidaeCarabidaeStaphylinidaecytochrome oxidase I
One of the principal goals of biodiversity conservation is to preserve the evolutionary potential of biological entities (Forest et al. 2007), be they species or populations. This forward-looking approach is paradoxically best served by a backward-looking perspective. That is, a species’ past is one of the best predictors of its future. Population genetic, phylogeographic, and phylogenetic methodologies peer progressively further back in time to reconstruct the patterns and processes that have led to the distribution of genetic diversity presently observed in species, and yield insights into the factors that have influenced this distribution. Studies on the conservation genetics of individual species can provide important information applicable to that species’ long term preservation. However, extending such studies to multiple codistributed species can provide information applicable to the conservation of entire biodiversity landscapes. In this paper we compare intraspecific patterns of genetic diversity among three codistributed, flightless beetle (Coleoptera) species in the southern part of the California Floristic Province (CFP).
Understanding the genetic landscape in southern California is particularly urgent, as southern California is one of the most densely populated and rapidly developing areas in the world. The CFP’s designation as a hotspot reflects not only its high levels of diversity and endemism, but the intensity of these anthropogenic threats as well. Southern California is home to more endangered species than nearly any other part of the United States (Dobson et al. 1997). The threats to these and other species take many forms. Urbanization of wildlands in the area has proceeded steadily, with no end in sight (Rundel and King 2001; Syphard et al. 2005). Development has been accompanied by invasive species, pollution, and increased frequency of fire, all of which have had detrimental effects on native biodiversity (Fisher et al. 2002; Seabloom et al. 2006; Schwartz et al. 2006; Syphard et al. 2006). Beyond these direct effects, remaining habitat has been fragmented, which both exacerbates some of the above problems (Suarez et al. 1998; Bolger et al. 2000), and through demographic and population genetic effects, imposes challenges of its own (Wilcox and Murphy 1985). Losses of genetic diversity through fragmentation have already occurred in southern California arthropods (Vandergast et al. 2007). Furthermore, these effects may be manifested at surprisingly low levels of urbanization (Riley et al. 2005). It has also been demonstrated that habitat fragments sufficient to host some of the region’s endangered vertebrates do not necessarily protect insects associated with these habitats (Rubinoff 2001).
On a more positive note, the obvious threats to the unique biodiversity of southern California has motivated a strong and active conservation community. Local, state, federal, and private agencies have come together in ecosystem level planning, most notably through the Natural Community Conservation Planning (NCCP) effort, coordinated by the California Department of Fish and Game (CDFG 2007). The NCCP process aims to incorporate all relevant science in its planning, including information on population genetics of species concerned. The NCCP’s flagship plan, the San Diego Multiple Species Conservation Program, aimed at preserving coastal sage scrub communities, has garnered national attention, paving the way for additional science-based conservation in densely populated southern California.
The incorporation of invertebrates into such integrative planning has been hampered by a dearth of even basic natural history information, let alone detailed data on population genetic relationships. Arthropods, mainly insects, represent an overwhelming fraction of the region’s biodiversity, yet population genetic and phylogeographic relationships have been investigated for only a small number. Most studies that have been done have demonstrated significant genetic divergence over short geographic distances (Landry et al 1999; Seagraves and Pellmyr 2001; Bond 2004; Chatzimanolis and Caterino 2007b), corresponding, in some, to species level divergence (Law and Crespi 2002). In some cases this has been attributed, at least in part, to contemporary environmental factors (Vandergast et al. 2007). Unfortunately, due largely to real distributional differences among these taxa, sampling in previous studies has been varied enough in location and intensity to preclude straightforward synthesis from a regional perspective.
Our goal in this paper is to analyze three species of co-occurring beetles in southern California’s Transverse Ranges from a conservation genetic perspective. In particular we are interested how isolated populations in the major components of these ranges are from one another. This includes analysis of gross distribution of genetic diversity, to discover areas that might harbor unusually high or relictual diversity, as well as analysis of connectedness among regions, to determine how recently populations have been in genetic contact, or indeed, how important ongoing contact through migration may be to maintaining genetic diversity patterns. By examining multiple species together, we can determine the extent to which these patterns are idiosyncratic or shared, offering hope for developing additional multiple species conservation strategies. And by examining these patterns in a phylogenetic context, we can shed light on the historical factors responsible for current patterns, and future prospects for maintaining them.
Materials and methods
This study uses beetles to examine the questions above. Beetles represent the largest single fraction of global biodiversity, and are represented by nearly 8000 described species in California (Caterino 2006). We analyze data from three species: Calathus ruficollis Dejean (Carabidae), Sepedophilus castaneus (Horn) (Staphylinidae), and Nyctoporis carinata (LeConte) (Tenebrionidae). All three of these species are flightless, and their relative immobility should promote geographically-limited patterns of genetic diversity. Detailed phylogeographic analyses of C. ruficollis and S. castaneus have previously been published (Chatzimanolis and Caterino 2007a, b, respectively), but data for N. carinata are newly presented here. In addition we have added 7 individuals of C. ruficollis from the San Gabriel Mts., two of which share a haplotype not previously reported. Data for sampling sites and GenBank Acc. #s (EU037099-EU037184, EU037191) are given in Appendices 1 and 2. Some regrouping of populations in C. ruficollis and S. castaneus has been done for statistical analyses to match sampling for N. carinata so some specific results may differ slightly from those previously presented. Each species is represented by multiple individuals (at least 10 where possible) from each of 10 areas, though not all species were found in all areas (Fig. 1). These areas represent variably discrete subunits of the Transverse Ranges, including, from west to east, the Santa Ynez Mts., Sierra Pelona, San Gabriel Mts., San Bernardino Mts., and San Jacinto Mts. While the eastern-most of these are separated by distinct valleys, corresponding in most cases to major faults, the western ranges merge with each other and with more northern ranges with little in the way of obvious (modern) barriers. These divisions follow major drainages and low passes, and offer contiguous and roughly similar-sized areas for comparison. Two additional areas, the northern and southern Santa Lucia Mountains are not considered to belong among the Transverse Ranges, but they often show phylogeographic relationships to Transverse Range regions (Chatzimanolis and Caterino 2007b), and provide additional context for comparison. In the cases of C. ruficollis and S. castaneus, the sampled localities represent a fairly small portion of their total ranges, which extend in both cases from near the U.S.-Mexico border north along the coast into Oregon and Washington. Our samples of N. carinata do represent most of its supposed range, and include samples of a questionably distinct congener, N. vandykei Blaisdell, in the southwestern Sierra Nevada.
These analyses are all based on sequences of the mitochondrial cytochrome oxidase I gene, which has been widely used in intraspecific studies of insect diversity (Caterino et al. 2000). Specimens were collected into 100% ethanol or directly frozen and stored at −70°C. DNA voucher specimens are deposited at the Santa Barbara Museum of Natural History and their corresponding data are available through the California Beetle Project database at http://www.sbcollections.org/cbp/cbpdatabase1.aspx. Total genomic DNA was extracted using the DNeasy Tissue kit (Qiagen, Valencia, CA). An approximately 826bp fragment of the COI gene was amplified using the primers C1-J-2183 (‘Jerry’: CAACATTTATTTTGATTTTTTGG) and TL2-N-3014 (‘Pat’: TCCAATGCACTAATCTGCCATATTA). PCRs included an initial denaturation of 5 min at 94°C, 35 cycles of: 45 s at 94°C, 30 s at 45°C and 1 min at 72°C; followed by a 2 min final extension at 72°C. PCR products were purified using the QIAquick PCR Purification kit (Qiagen, Valencia, CA) and sequencing was performed by Macrogen, Inc. (Seoul, Korea). All fragments were sequenced in both directions.
We quantified genetic diversity in each population by several measures. To assess raw diversity within each population we calculated nucleotide diversity (equivalent to average uncorrected, or ‘p’, distance among individuals within each population). This captures some phylogenetic depth albeit in a phenetic manner. To measure uniqueness of each population we calculated fraction of unique haplotypes within each area and examined monophyly of each population’s haplotypes. Distinctness of the haplotype complement for each population is measured by the average uncorrected pairwise distance to all other populations.
Connectedness of populations
We investigated demographic trends of population expansion or contraction as an indication of population robustness. Fu’s Fs (Fu 1997) and Tajima’s D (Tajima 1989) were examined in Arlequin for departures from an equilibrium population. We also calculated the exponential growth parameter ‘g’ using Metropolis–Hastings sampling of the genealogy implemented in the program Fluctuate (Kuhner et al. 1998). We used the F81 (Felsenstein 1981) model with empirical (uneven) base frequencies and a ti/tv ratio as estimated by ModelTest (see below), and a Watterson initial estimate of Θ, running 10 short chains of 20,000 steps and 2 long chains of 2,000,000 steps, sampling every 20 or 200 steps, respectively. Fluctuate was also used to estimate the population parameter Θ, which, assuming a roughly constant mutation rate among species and lineages, provides a measure of effective population size.
To assess shared history in a broader sense, we conducted phylogenetic analyses of non-redundant haplotypes of all three species (58 for C. ruficollis, 45 for S. castaneus, and 91 for N. carinata). Phylogenetic trees were estimated using Bayesian methods. The maximum likelihood model for the COI gene was estimated using ModelTestServer version 1.0 (running ModelTest version 3.8) (Posada 2006) using the corrected Akaike Information Criterion (AICc). Bayesian analysis was performed in MrBayes version 3.1.2 (Huelsenbeck and Ronquist 2001) using three runs with 4 × 106 generations, each having four Markov chains, heating equal to 0.2 and sampling every 100th generation. For N. carinata, sequences of Nyctoporis aequicollis and Coelus ciliatus (Tenebrionidae; GenBank Acc.#s, EU037190 and EU032421, respectively) were included as outgroups. Additional details for analyses of C. ruficollis and S. castaneus have been presented previously (Chatzimanolis and Caterino 2007a, b)
Assessing conservation value
Assessing conservation value from population genetic data can take a couple of broad perspectives. Systematists have generally sought phylogenetically distinct, unusually divergent, or unusually diverse species subsegments for prioritization as some sort of evolutionarily significant units (e.g. Moritz 1994) and this has been extended in a comparative context to identify generally significant areas (Calsbeek et al. 2003; Moritz and Faith 1998; Rissler et al. 2006). It is also increasingly possible to discern contemporary demographic patterns as well as recent disruptions of these patterns (Frankham 1995; Vandergast et al. 2007). We primarily take the first approach, focusing on intraspecific diversity and its phylogeographical distribution. We first calculated averages of raw diversity measures directly. However, these numbers are biased in most cases by the absolute value of the more divergent species. We also assigned a rank to each value across populations within each species, with 1 indicating the highest conservation value (indicating greatest diversity or greatest isolation), finally averaging these ranks across species by population.
Nucleotide divergence among haplotypes within populations offers a better indication of their total phylogenetic diversity. These results are similar to simple proportions of unique haplotypes, but are more sensitive to the presence of multiple lineages in populations. The greatest average within population divergence is seen in the southwestern Sierra Nevada. These mountains represent the southernmost extent of the Sierra Nevada, which in general have exhibited disproportionate divergence in all beetles examined (Chatzimanolis and Caterino 2007a, b). While interesting and indicative of a major phylogeographic break, this region lies outside of the Transverse Ranges, and serves here primarily to put the Transverse Ranges into perspective. This is also true to a lesser extent of the Santa Lucias, in the central coast ranges, and we limit our discussions below mainly to parts of the Transverse Ranges. Here the top-ranking within population diversity is seen in the Sierra Pelona, with all species showing moderately high diversity in this region. This region does not, however, contain the highest diversity of any of them individually, and these species maxima are scattered, including the northwestern Transverse Ranges, the Santa Ynez Mountains, and the central Transverse Ranges, for S. castaneus, N. carinata, and C. ruficollis, respectively.
Considering the divergence among populations, most show considerable distinctness. The average distances among populations of N. carinata are highest overall and individually by population in all cases, with no population averaging closer than 6% to all others. In S. castaneus, most populations are more than 3% divergent. Divergences among populations of C. ruficollis are in line with expectations for a more or less panmictic species, with nearly all exhibiting <1% average divergence. Among Transverse Range areas, the among population divergences averaged across species are largely similar among areas. Each species shows some slight outliers: in both S. castaneus and N. carinata, the San Bernardino and San Jacinto populations are slightly more divergent. In C. ruficollis the Central Transverse Ranges appear most distinctive, although this is due mainly to the presence of a divergent Sierra Nevada haplotype in this population. But in general, within species, the populations exhibit surprisingly constant levels of divergence among one another.
Our primary measure of population connectedness was pairwise FST values (Table 1; Appendix 3). For each population we measured this as the number of significant FST values to all other populations. In S. castaneus and N. carinata, all or nearly all interpopulation comparisons were significant, with, in the latter, only Sierran, Coast Range and, surprisingly, Santa Cruz Island populations showing insignificant separation (this last likely due to low sample size). In C. ruficollis, the most isolated population is that of the Central Transverse Ranges, and this agrees with phylogenetic results which support this as a distinct clade (Chatzimanolis and Caterino 2007a). San Jacinto and San Gabriel populations also appear isolated from most others. The best connected populations appear to be the northwestern Transverse Ranges and Sierra Pelona, both insignificantly isolated from the majority of other populations.
Isolation by distance (IBD) using untransformed FST and geographic distance
Pairwise FST correlation tests using Permute! and deleting all localities not shared by all
Two-group AMOVA analyses for N. carinata populations agreed with C. ruficollis in a strong separation of Sierran (and Tehachapi – not present in C. ruficollis samples) populations from Coast and Transverse Range populations (Fig. 2a). This was somewhat consistent in S. castaneus, where maximum among-group variation was found when separating northern (including Sierran) and southern groups between the Sierra Pelona and San Gabriel mountains (Fig. 2e), splitting the Transverse Ranges between major genetic groups (for further discussion of this break see Chatzimanolis and Caterino 2007b).
There was little consistency among measures supporting any significant trends in population expansion or contraction. If anything there are indications that several populations may have undergone recent expansion. In C. ruficollis Chatzimanolis and Caterino (2007a) attributed apparent expansion in several populations to somewhat anthropophilic tendencies in this species. However, no such habits are obvious in the other species. In almost every area some species shows signs of population expansion, with the southern Santa Lucias, the (adjacent) northwestern Transverse Ranges, Santa Ynez Mts., and San Gabriels indicating expansion by some measure for two of the three species. Fluctuate provided the greatest number of significant results, though not supporting expansion scenarios deemed significant by Fu’s Fs or Tajima’s D in all cases (noteably that indicated for S. castaneus for the southern Santa Lucias), perhaps indicating that the departures from neutrality these latter statistics indicate derive from other factors. From a conservation perspective, population contraction is of particular interest, and Fluctuate indicated significant contraction for the central Transverse Range population of S. castaneus, and the San Gabriel and San Bernardino populations of N. carinata, but for no area for more than a single species. At this point we cannot determine the recency of any of these changes in population size, and would not suggest that any necessarily represent response to anthropogenic factors.
Examining population genetic patterns across multiple codistributed species would ideally reveal consistent patterns of diversity and isolation, pointing toward clear assessments of relative conservation significance of different areas. The purpose of this analysis was not specifically to establish conservation priorities in these three beetle species, but rather to assess the extent to which their distributions of genetic diversity identified similar regions of genetic interest. At a general level, these three flightless beetle species all exhibited remarkable intraspecific diversity, even just within what, for two of them, is a small portion of their total range. While in Calathus ruficollis this diversity shows little geographic structure, indicating considerable recent gene flow, in both Sepedophilus castaneus and Nyctoporis carinata, mitochondrial diversity shows much more geographic structure, and even where populations contain representatives of multiple haplotype clades, there is little evidence of recent contact (in that the putative migrant haplotypes are not represented in ‘source’ areas). The clearest conservation message that can be built on these observations is that the Transverse Ranges are by no means a genetically homogenous block, and all of its subregions exhibit unique genetic variants, some of which are highly divergent.
On a less satisfying note, patterns of the distribution of diversity among species shows little consistency. The most consistent finding separates Sierra Nevadan populations from most (S. castaneus) or all others. Sampled populations in the Coast ranges, the northern and southern Santa Lucia Mts. are less distinctive, but still exhibit relatively high divergence, diversity, and some degree of phylogenetic unity for two of our three species, the northern portions, especially. Among areas within the Transverse ranges, every measure we examined identified a different area as most significant. And for no measure did all the species rank the chosen area unanimously high. For example the most divergent area as assessed by averaging pairwise distances to other populations is the San Jacinto Mts. (Table 1). However, this is not the most divergent area for any of them, individually. For C. ruficollis it is the central Transverse Ranges, for N. carinata the San Bernardinos, and for S. castaneus, the San Gabriels. In fact, every area we recognize in the Transverse Ranges is ranked as most or second-most significant by one or another measure. This might be taken to underscore our previous point, that all areas here protect significant genetic diversity, and none are expendable or interchangeable. This does imply a rather high management burden, but better prioritization with data from additional species remains a possibility.
Connectedness of populations may be an important determinant of long-term viability of species. The occasional influx of individuals and novel genetic variants can buffer a population from inbreeding, stochastic effects and localized selection events (Newman and Tallmon 2001). So whatever insights into migration routes among subregions in the Transverse ranges can be gleaned from population level data are valuable. Significant migration may occur over a broad time scale, ranging from the very recent, as indicated by shared haplotypes, to the historic, as seen in disjunct distributions within haplotype clades. Our main indicator of recent migration is interpopulation FST. By this measure only C. ruficollis shows any indication of significant dispersal, with nearly all adjacent populations showing insignificant FST values. Significant subdivision is only apparent between more distant sites. Given the strong geographic fidelity of the other two species, it is conceivable that some apparent C. ruficollis migration is very recent, i.e., anthropogenic (Chatzimanolis and Caterino 2007a).
Taking a longer view, all species do show signs of historical population connections. Several populations in both S. castaneus and N. carinata comprise haplotypes from multiple clades. In particular the populations in the southern coast ranges (Santa Lucias) and western and northwestern Transverse ranges contain mixtures of haplotypes, suggesting occasional movement between them. On the other hand, although covering a large area, these regions are not separated by very clear topographic or environmental boundaries, and these ‘populations’ may well all be connected, with distance being the main isolating factor. In a few other cases, more distinct signatures of unique migrations may be seen. In particular the appearance of individuals of the San Bernardino clade of N. carinata in the Tehachapi Mts. and in the San Jacintos clearly indicates dispersal. These areas otherwise host highly distinct lineages. While the importance of these unique events to long-term viability is impossible to assess, we should consider these dispersal corridors, across the Antelope Valley in the first case, and San Gorgonio Pass in the second, as potentially significant. The latter in particular, now traversed by a major highway, Interstate 10, is probably no longer viable for a flightless insect.
The clearest message we can distill out of our results is that relatively small areas in southern California can host distinctive and isolated genetic diversity. Two of the three species examined here support this strongly, and many of other organisms studied in detail in the region have as well. Though we hoped our data might provide a clear ranking of areas in terms of their conservation value, this has not been possible, because even in our limited sampling the species disagree on which these would be. While this seems unfortunate in that it does not help set regional conservation priorities, it may (if such a pattern continues to hold) free land managers to focus on regions experiencing the most pressing threats.
We appreciate the assistance of A. Ramsdale, R. Aalbu, K. Will, S. Mulqueen, P. Russell, P. Jump, and I. Foley in collecting or providing specimens, and that of G. Betzholtz in the lab. Several preserves, agencies (and their personnel) generously granted collecting permission and otherwise provided assistance: the California Department of Fish and Game, Los Padres National Forest (M. Freel), the UC Sedgwick Reserve (M. Williams), the Arroyo Hondo Preserve (Land Trust for Santa Barbara County, C. Chapman, J. Iwerks, J. Dunn, and J. Warner), UC Whitaker Forest (R. York), Sequoia National Forest (J. White), Camp Cedar Falls (R. Young), Angeles National Forest and the San Dimas Experimental Forest (M. Oxford), San Bernardino National Forest (M. Lardner, R. Eliason), the UC James Reserve (M. Hamilton), the UC Coal Oil Point Reserve (C. Sandoval), and the UC Santa Cruz Island Reserve (L. Laughrin). This work was supported by the Schlinger Foundation, a bequest from G. Oostertag, and National Science Foundation award DEB0447694 to M. Caterino.