Comparative riverscape genetics reveals reservoirs of genetic diversity for conservation and restoration of Great Plains fishes

Please download to get full document.

View again

of 17
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Comparative riverscape genetics reveals reservoirs of genetic diversity for conservation and restoration of Great Plains fishes
Document Share
Document Transcript
  Comparative riverscape genetics reveals reservoirs ofgenetic diversity for conservation and restoration ofGreat Plains fishes MEGAN J. OSBORNE,* JOSHUAH S. PERKIN, †  KEITH B. GIDO ‡  and THOMAS F. TURNER** Department of Biology and Museum of Southwestern Biology, University of New Mexico, Albuquerque, NM 87131, USA, † Department of Biology, Tennessee Technological University, 1100 N. Dixie Avenue, Cookeville, TN 38505, USA,  ‡ Division of Biology, Kansas State University, 116 Ackert Hall, Manhattan, KS 66506, USA Abstract We used comparative landscape genetics to examine the relative roles of historicalevents, intrinsic traits and landscape factors in determining the distribution of geneticdiversity of river fishes across the North American Great Plains. Spatial patterns ofdiversity were overlaid on a patch-based graphical model and then compared withinand among three species that co-occurred across five Great Plains watersheds. Speciesdiffering in reproductive strategy (benthic vs. pelagic-spawning) were hypothesized tohave different patterns of genetic diversity, but the overriding factor shaping contem-porary patterns of diversity was the signature of past climates and geological history.Allelic diversity was significantly higher at southern latitudes for   Cyprinella lutrensis and  Hybognathus placitus  , consistent with northward expansion from southern Pleisto-cene refugia. Within the historical context, all species exhibited lowered occupancyand abundance in heavily fragmented and drier upstream reaches, particularly  H. plac-itus  ; a pelagic-spawning species, suggesting rates of extirpation have outpaced lossesof genetic diversity in this species. Within most tributary basins, genetically diversepopulations of each species persisted. Hence, reconnecting genetically diverse popula-tions with those characterized by reduced diversity (regardless of their position withinthe riverine network) would provide populations with greater genetic and demo-graphic resilience. We discuss cases where cross-basin transfer may be appropriate toenhance genetic diversity and mitigate negative effects of climate change. Overall,striking similarities in genetic patterns and in response to fragmentation and dewater-ing suggest a common strategy for genetic resource management in this unique river-ine fish assemblage. Keywords : dendritic landscapes, dispersal constraints, graph theory, habitat fragmentation,landscape genetics, river networks Received 30 July 2014; revision received 6 October 2014; accepted 15 October 2014 Introduction Over the last 150 years, fish communities in rivers of the Great Plains of North America have been radicallyaltered by habitat fragmentation, dewatering and land-use changes that negatively affect fish habitat connectiv-ity, heterogeneity, quality and extent (Hoagstrom  et al. 2011; Perkin & Gido 2011; Perkin  et al.  2014a). Remnantpopulations of once vastly abundant and widely distrib-uted riverine fishes now occur in drier, fragmented hab-itats that have resulted from impoundment anddiversion of surface waters and groundwater pumping(Hoagstrom  et al.  2011; Perkin  et al.  2014a). In response,fish species richness and diversity have declined at localand regional (i.e. within major drainages) scales. None-theless, even though many fish species in Great Plainsrivers are imperilled and there have been numerouslocal extirpations (Hubert and Gordon 2007; Jelks  et al. 2008; Hoagstrom  et al.  2011; Perkin & Gido 2011), Correspondence: Megan J. Osborne, Fax: +1 505 2770304;E-mail: ©  2014 John Wiley & Sons LtdMolecular Ecology (2014) doi: 10.1111/mec.12970  relatively few extinctions have been recorded (the RioGrande Basin is a notable exception; Platania 1991;Hoagstrom  et al.  2010). Previous studies have identifiedlocations spread across the Great Plains that are appar-ent refuges for ecologically (Hoagstrom  et al.  2011) andgenetically diverse endemic taxa (Osborne  et al.  2010,2013). As long as fishes and refuges persist, manage-ment actions could potentially restore preimpact com-munity characteristics in suitable or restored reaches of the Great Plains.A logical course of management action would be tofirst address environmental causes of fish imperilment,followed by repopulation of key species in sufficientdensities to sustain them in newly restored habitats(Perkin  et al.  2014a). This strategy necessitates identifica-tion and protection of appropriate reservoirs of geneticdiversity because reintroduction efforts are more suc-cessful and persist longer when repatriates are geneti-cally diverse (Williams 2001). Specifically, geneticallydiverse repatriates may be more likely to adapt to localconditions. When possible, maintenance of historicalpatterns of population structure is also important. Forthese reasons, it is vital to understand how geneticdiversity is distributed on the landscape and the causalmechanisms explaining these patterns. With this knowl-edge, it may be possible to make general predictionsregarding the distribution of genetic diversity that areapplicable across species, which in turn could be usedto guide reintroduction efforts for whole assemblages.The distribution of genetic diversity on the landscapeis determined by at least three broad classes of factors,and these include geological processes, intrinsic traitsand landscape features that include human-mediatedhabitat changes. Geological processes cause drainagerearrangements that isolate or connect river systems,and persistent climatic gradients can shape diversitypatterns similarly across codistributed species over evo-lutionary time (Smith 1981; Richardson & Gold 1995;Waters & Nordt 1995). For example, a divide betweenpreviously connected drainages could produce a similarpattern of genetic structure among co-occurring species(Waters  et al.  2001). In contrast, intrinsic factors such as body size, fecundity and vagility are expected to relateto genetic diversity through their influence on geneticeffective population size and migration (gene flow) (Cri-spo  et al.  2006; Faulks  et al.  2010). Finally, more recentalterations to the landscape itself, such as human-medi-ated fragmentation and river dewatering, may alsoaffect patterns of genetic diversity (Jager  et al.  2001;Turner  et al.  2006; Junge  et al.  2014). For example,impediments to migration can deplete genetic diversitywithin riverine populations (Jager  et al.  2001). Suchalterations can influence genetic diversity throughan interaction with intrinsic features in a mannersomewhat akin to ecological filtering, where constraintsof landscape features depend on traits related to repro-duction and migration that vary among species(e.g. Turner  et al.  2006; Junge  et al.  2014).In this study, we characterized the relative roles of geological and landscape factors in shaping diversityacross species with different life-history strategies. Wequantified genetic diversity at microsatellite loci forthree co-occurring Great Plains river fishes within andamong five major tributary basins of the Great Plains.Two of three target species are members of a reproduc-tive guild of pelagic-spawning fishes (eggs and larvaedrift in the water column) that are highly sensitive tofragmentation and dewatering because of downstreamdisplacement of propagules (Dudley & Platania 2007;Perkin & Gido 2011). The other species is a physiologi-cally tolerant (Matthews & Maness 1979) and abundantfish that employs benthic spawning (deposits eggs in oron substrate). If, despite their ecological differences, allthree species exhibit similar patterns of genetic diversityacross watersheds, we can infer a strong role for histori-cal factors or similar response to landscape modificationthat have uniformly acted upon the riverine fish assem- blage. Alternatively, species with differing ecologicaltraits may respond differently to these factors (McGill et al.  2006). As such, a comparative demographic andpopulation genetic approach, where multiple, codistrib-uted species are studied, may establish whether generalor species-specific patterns best describe overall diver-sity and improve our understanding of where andwhen reintroductions are likely to be successful in theGreat Plains and other systems. Methods Study area The focal area of this study encompassed five majortributary basins of the North American Great Plainslisted from north to south: (i) the Platte between theWyoming  –  Nebraska border and the confluence with theMissouri River; (ii) the Kansas upstream of the conflu-ence with the Missouri River; (iii) the Arkansas betweenLarkin, Kansas and Keystone Reservoir in Oklahoma;(iv) the Canadian between the panhandle of Texas andEufaula Reservoir in Oklahoma; and (v) the Redupstream of Lake Texoma in Oklahoma and Texas(Fig. 1). Together, these cover a large extent of thesouth-central Great Plains. Riverscapes in this regionwere historically connected longitudinally and laterally,characterized by periods of flashy high flows that inun-dated expansive floodplains as well as stochasticdroughts that desiccated broad extents of stream(Dodds  et al.  2004). This natural hydrologic expansion ©  2014 John Wiley & Sons Ltd 2  M. J. OSBORNE  ET AL.  and contraction drove unique adaptations among GreatPlains stream biota, which consequently evolved ecolog-ical characteristics that are intimately linked to the natu-ral variation in hydrology of Great Plains streams(Fausch & Bestgen 1997; Hoagstrom & Turner 2013).During the past century, major forms of landscape alter-ation included conversion of drainage basins into agri-cultural land, construction of instream barriers, and thediversion and extraction of surface and subsurfacewater (Costigan & Daniels 2012), all of which have con-tributed to alterations of stream fish communities insynergistic fashion (Gido  et al.  2010; Hoagstrom  et al. 2011). Study species and life-history contrasts We selected one representative fish from three predomi-nant reproductive guilds. These fishes (and associatedguild) included plains minnow  Hybognathus placitus (pelagic broadcast spawning: releases nonadhesivesemi-buoyant eggs into the pelagic-zone; Taylor &Miller 1990; Platania & Altenbach 1998), emerald shiner Notropis atherinoides  (pelagic-substrate spawning: releasenonadhesive demersal or sinking eggs into pelagiczones; Flittner 1964; Campbell & MacCrimmon 1970)and red shiner  Cyprinella lutrensis  (substrate spawning:deposit adhesive demersal eggs over a variety of sub-strata; Gale 1986; Vives 1993). Although these three fishspecies were historically codistributed across much of the Great Plains, their distributions have shifted because of broad-scale landscape changes that haveoccurred in the region, including massive declines for  H  .  placitus  (Perkin & Gido 2011), declines in some areas but concurrent expansion in others for  N. atherinoides (Pflieger & Grace 1987; Perkin  et al.  2014b) and wide-spread persistence and expansion of   C. lutrensis  (Poulos et al.  2012). We selected the once broadly distributed  H. placitus  as a focal species to guide sampling acrossthe landscape because this species represented the mostsevere declines. Key landscape features and sampling site selection The two major forms of landscape alteration affectingnative and endemic fishes in the Great Plains involvedewatering and habitat fragmentation and associatedhabitat changes (Hoagstrom  et al.  2011). Although thesedrivers are implicated in the decline of a number of species (Gido  et al.  2010), they are also directly relatedto changes in the distribution of stream fish reproduc- 94°W94°W96°W96°W98°W98°W100°W100°W102°W102°W104°W104°W106°W42°N42°N40°N40°N38°N38°N36°N36°N34°N34°NRedCanadian ArkansasKansasPlatte0 100km N. atherinoidesH. placitusC. lutrensis BarriersReservoirsFragmentsStreams 123456789101112131415161718192021232425 26 2728293031323334353637383922 1,000km WesternEdge of High PlainsAquifer  Fig. 1  Map of the study area and sam-pling localities for  H. placitus ,  N. atherino-ides  and  C. lutrensis . Numbers refer tostream fragments and fragment detailsare provided in Table S1. ©  2014 John Wiley & Sons Ltd COMPARATIVE GENETICS OF GREAT PLAINS FISHES  3  tive guilds selected for inclusion in this study (Perkin et al.  2014a). Consequently, we selected sampling sitelocations that incorporated the maximum variation indewatering (measured as the number of days with zeroflow using U.S. Geological Survey stream flow gages)and stream fragment length (measured as the longitudi-nal length of stream between two barriers). As thesealterations have constrained the distribution of   H. plac-itus , our sampling site selection was limited to streamfragments in which  H. placitus  persists and where C. lutrensis  and  N. atherinoides  also co-occur. We priori-tized sampling among 45 large stream fragments acrossthe five tributary basins using data from collectionstaken between 1993 and 2013 (Perkin  et al.  2014a) thatdescribed the probability of occurrence for these fishspecies. We began by targeting stream fragments thatwere most likely to be inhabited by  H. placitus  across allfive tributary basins. During the summer of 2013, wevisited fragments and sampled fishes using seines bytargeting all available habitats at a site for at least 2 hor until a minimum of 30 individuals of each speciescould be collected. Caudal fin clips were collected from  H. placitus ,  N. atherinoides  and  C. lutrensis,  and finclips were stored in 100% ethanol for subsequent DNAisolation.  Molecular methods Genomic DNA was extracted from air-dried fin clipsusing standard proteinase-K digestion and standardphenol/chloroform methods (Hillis  et al.  1996).  Hybo- gnathus placitus, N. atherinoides  and  C. lutrensis  wereassayed for variation at nine, seven and eight microsat-ellites, respectively. Microsatellites were amplified as10  l L reactions, containing one  l L diluted DNA, 1XColorless GoTaq  Flexi Buffer, 2 m M  MgCl 2  solution,125  l M  dinucleotide triphosphates (dNTPs), 0.4  l M  of  both forward and reverse primers, and 0.375 units of GoTaq  DNA Polymerase. For  H. placitus,  PCRs wereinitially denatured at 90   C for 2 m, followed by 30cycles of denaturing at 90   C for 30 s, with annealing of 56  ° C ( Ca3/Ca12 ) or 58  ° C ( Ppro126/Ppro132 ,  Nme93/232 )or 49  ° C ( Lco3/Lco6  and  Lco7 ) for 30 s, and extension at72   C for 45 s and ending with a final extension at 72   Cfor 30 m. For  N. atherinoides,  PCRs were the same but with annealing at 59  ° C [ Nme232, Nme93 ,  Nme208 (Gold  et al.  2004),  Ppro126 ,  Ppro132  (Bessert & Ort  ı 2003)and ( Ca12 ) (Dimsoski  et al.  2000)] or 49  ° C  Ca6 (Dimsoski  et al.  2000) for 30 s, and extension at 72  ° Cfor 45 s and ending with a final extension at 72  ° C for30 m. For  C. lutrensis,  PCR conditions were the samewith annealing temperatures of 58  ° C ( Nme232, Nme93 )or 52  ° C for  Nme208 , or 59  ° C ( Ppro126/Ppro132 , Nme174 ,  Ca12 ) or 50  ° C for  Lco6  (Turner  et al.  2004). Foreach sample, one microlitre of PCR product was mixedwith 10  l L of formamide and 0.4  l L of HD400 sizestandard and then denatured at 90  ° C for 5 min. Allsamples were run on an automated ABI 3130 DNAsequencer and analysed with  GENEMAPPER  software(ABI). Data analyses Genetic variability microsatellites.  GENEPOP  version 3.1(Raymond & Rousset 1995) was used to conduct modi-fied exact tests to determine whether the observedgenotype frequencies conformed to Hardy  –  Weinbergexpectations within each collection locality. This pro-gram was also used to conduct the global test for link-age disequilibrium among loci. Sequential Bonferronicorrection (Rice 1989) was applied to account for multi-ple comparisons. As the number of alleles and expectedheterozygosity are dependent on sample size, we useda resampling procedure to calculate diversity measures.The minimum cut-off size was determined by the mini-mum sample size shared across species; hence, siteswith fewer than 20 samples were excluded. Briefly,1000 random subsamples were drawn without replace-ment from each temporal sample. Diversity and 95%CIs were calculated for each locus across subsamplesand a mean was obtained across loci for each statistic[corrected number of alleles ( N  ac ), gene diversity (  H  ec )and heterozygosity (  H  oc )]. This analysis was conductedin the  R  statistical package ( R  script available on request)(R Development Core Team 2008). Diversity statisticswere calculated for each species and stream fragment(Table 1). One-way analysis of variance on ranks wasconducted in  SIGMA PLOT  version 11.4 to examinewhether genetic diversity statistics differed significantly between tributary basins for each species. Genetic response to historical features.  Hoagstrom & Berry(2006) suggested that  H. placitus ,  C. lutrensis  and N. atherinoides  invaded streams of the northern plainsfrom Gulf of Mexico drainages following glacial peri-ods. If this was the case, higher diversity at southernlatitudes is consistent with the prediction that stablepopulations survived  in situ  during Pleistocene climaticoscillations. Hence, to test the prediction that past envi-ronments and geologic events are the overriding factorsshaping how diversity is distributed across the land-scape, we used ordinary least-squares linear regressionof allelic diversity and latitude conducted in  SIGMA PLOT version 11.4.We evaluated population structure for each speciesusing Weir & Cockerham’s (1984) hierarchical  F -statis-tics calculated in  ARLEQUIN  version 3.11 (Excoffier  et al. 2005). We conducted two  AMOVA s, the first to determine ©  2014 John Wiley & Sons Ltd 4  M. J. OSBORNE  ET AL.  whether or not genetic variance could be attributed todifferences between major basins. For this analysis,samples were grouped according to drainage basins (i)Missouri [Platte and Kansas] and (ii) Arkansas-Red[Red, Canadian and Arkansas]). The second  AMOVA  wasused to determine whether there were significant differ-ences in population structure among the five sampledtributary basins (  F CT ), among fragments within tribu-tary basins (  F SC ) and among fragments irrespective of tributary basin (  F ST ). In this case, fragments weregrouped according to tributary basin of srcin (i) Platte,(ii) Kansas, (iii) Arkansas, (iv) Canadian and (iii) RedRiver fragments. Significance was assessed by 1000 bootstrap replicates.In addition to  AMOVA , we assessed the level and nat-ure of population structure for each species in a spatialcontext with the  GENELAND  software package executed inthe  R  environment (Guillot  et al.  2005). This programuses a geographically constrained Bayesian model thataccounts for the spatial position of sampled individualsand their multilocus genotypes. Five replicate runs wereconducted for each species with 500 000 Markov chainMonte Carlo (MCMC) iterations, the Dirichlet model of allele frequencies, no uncertainty in the spatial coordi-nates, maximum rate of the Poisson process was 100and the maximum number of nuclei for the Poisson  –  Voronoi tessellation was 300. We assessed plots of MCMC runs for convergence. The range of   K  s (numberof genetic clusters) to test was set from one to the maxi-mum number of sampled fragments for each speciesand hence ranged from 9 to 10. The run with the high-est log posterior density was selected as the best repre-sentation of the data. An additional five runs wereconducted at the optimal value of   K   determined in theprevious analyses with a burn-in of 5000 to trim theposterior distribution in the post processing step. Foreach species, a heat map was produced of the posteriorprobabilities of locations belonging to specific geneticclusters. Tributary basin boundaries were superimposedon the heat maps to illustrate that population groupsfall clearly within the riverine network. Genetic response to contemporary landscape features.  Toexamine the genetic response to contemporary land-scape features, we used an information-theoreticapproach (Akaike 1973). The response variables weregene diversity, observed heterozygosity and allelicdiversity adjusted to account for differences in samplesize among collections (as described above). The follow-ing determinant (landscape features) variables wereincluded for each river fragment: fragment length, dis-charge and percentage of days with zero flow. Thisanalysis was repeated using the residual values for theresponse variables (  H  ec ,  H  oc  and  N  ac ) after accountingfor the effect of latitude. Fragment length representedthe maximum length (river kilometres) of uninterruptedstream available to fishes between two barriers. Dis-charge represented the average daily discharge valuefor the period 1970  –  2013 based on U.S. Geological Sur-vey stream flow gages distributed among fragments. Table 1  Fragment length (river kms), sample size ( n ) and genetic diversity statistics ( N  ac  —   allelic diversity,  H  ec  —   gene diversityand  H  oc  —   observed heterozygosity) obtained using a resampling approach for each species sampled from river fragments in theGreat Plains shown in Fig. 1Fragment Fragment length  Hybognathus placitus Notropis atherinoides Cyprinella lutrensisn N  ac  H  ec  H  oc  n N  ac  H  ec  H  oc  n N  ac  H  ec  H  oc Platte 37 504 29 8.339 0.730 0.519 26 10.190 0.770 0.701 24 7.867 0.694 0.592Platte 39 198 12  —   0.804* 0.612* 2  — — —   22 8.172 0.692 0.536Kansas 33 332  — — — —   30 9.133 0.711 0.643 32 9.302 0.758 0.678Arkansas 14 528 35 10.380 0.745 0.565 30 11.682 0.842 0.715 30 10.115 0.740 0.659Arkansas 16 292  — — — —   30 10.504 0.839 0.784  — — — —  Arkansas 17E 186 20 8.444 0.658 0.528 20 11.286 0.827 0.738 20 8.767 0.765 0.597Arkansas 17W 186 30 9.453 0.684 0.556 21 10.791 0.848 0.794 30 10.289 0.796 0.973Arkansas 18 251  — — — —   30 11.958 0.849 0.806  — — — —  Canadian 7 793 6  — — — — — — —   51 10.701 0.768 0.649Canadian 8 220 32 9.712 0.711 0.555  — — — —   30 10.846 0.834 0.654Red 1 793 24 10.049 0.803 0.508 28 11.201 0.850 0.749 30 12.441 0.843 0.635Red 3 203 30 10.129 0.795 0.540  — — — —   18 10.375 0.835 0.576Red 4 162  — — — — — — — —   12  —   0.818* 0.740**Collections that were excluded from resampling analysis (due to small sample size) for diversity metrics, raw values for  H  e  and  H  o are provided. ©  2014 John Wiley & Sons Ltd COMPARATIVE GENETICS OF GREAT PLAINS FISHES  5
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks