1
Widespread co-occurrence of virulent pathogens within California amphibian communities
Appendix A
Site selection and sampling methods – We used the National Wetlands Inventory (NWI, United States Fish and Wildlife Service) to identify wetlands managed by the East Bay Regional Parks, Santa Clara County Parks, Blue OaksRanch UC Reserve, East Bay Municipal Utility District, California State Parks, Midpeninsula Regional Open Space District, Contra Costa Watershed District, and the US Fish and Wildlife Service, which are located in a four county area of the Bay Area of California, USA (Santa Clara, San Mateo, Alameda, and Contra Costa counties). To focus our survey efforts on wetlands that were likely to contain amphibians, we only included sites that were small (<2 ha) and fishless. We selected 85 wetlands from this listfor our field survey. We visited each site once between May and August of 2010 to collect amphibians for pathogen testing. Due to logistical constraints associated with travel between sites, we did not spatially randomize our visitation schedule to wetlands. Instead, wetlands within each park were sampled at similar times (i.e. within a week of each other). At each site, individuals of all species were collected by hand during visual transects conducted along the shoreline. Transects were completed when 100 individuals had been caught or we searched for a combined total of 60-person minutes. The visual transects were supplemented with standardized dipnet sweeps conducted every 15 m around the wetland’s perimeter(Johnson et al., 2002). We also conducted 3-5 habitat-stratified seine hauls through each wetland to ensure we were not missing any additional species. The captured amphibians were stored in the shade in large coolers with ice packs until all observations were completed. For each species excluding the endangered California red-legged frog(Rana draytonii), we randomly selected up to 20 metamorphic individuals for shipment to the University of Colorado for necropsy and evaluation of trematode and ranavirus infection. R. draytonii were only swabbed for Bd (see below) and immediately released.
Bd Swabs– For metamorphic individuals of each non-endangered species at a site, we batch swabbed up to 100 individuals (n ≤ 20 individuals per swab), although this number varied due to the number of captures (see Table A1). Nitrile gloves were changed whenever switching to a new swab. Individuals of the endangered species as well as adults of all species were handled while wearing a new pair of nitrile gloves and individually swabbed to reduce the likelihood of Bd cross contamination.
Necropsy – Collected individuals were shipped overnight to the University of Colorado where they were euthanized by immersion in a 0.5% solution of MS-222. Using a stereo-dissecting microscope, individuals were necropsied to assess infection status with Ribeiroia and echinostomes. During the necropsies, the entire individual was examined for metacercarial cysts including the epithelial tissue, major organ systems, and digestive tract(following Johnson and Hartson, 2009). We collected tissues for ranavirus testingfrom individuals at 43 of the 85 field sites. For each species at each site, a ~2 mm section of the liver and kidney was removed and pooled for all examined individuals and stored in a microcentrifuge tube at -80°C for determination of ranavirus infection status. Although our ranavirus samples represent a pooled sample of individuals for each species at a site, this approach allowed us to reduce the costs associated with processing individual samples.
Bd and ranavirus detection – Pathogen detection was done for both Bd and ranavirus utilizing standard quantitative PCR techniques. DNA from swabs (Bd) and tissue samples (ranavirus) was extracted using a Qiagen DNeasy kit following the included standard protocols. Following the standard method by Boyle et al.(2004), Bd quantification was done using a reagent mix of ABI Taqman 2X master mix, forward primer (900nM), reverse primer (900nM), and MGB probe (125nM). We used 5 µL of DNA template to create a 25µL reaction. A standard curve of 100, 10, 1, and 0.1 zoospore equivalents was used on each plate. Standards were obtained from CSIRO labs in Australia. The thermal protocol was 50˚C (2 min), 95˚C (10 min), followed by 50 cycles of 95˚C (15s), 60˚C (1 min). Ranavirus detection followed Kerby and Storfer (2009). The reagent mix consisted of ABI Taqman 2X mastermix, forward primer (300nM), reverse primer (900nM), and TAMRA probe (250nMl). We combined 2 µL of DNA with 18µL of the above reagents for 20µL reactions. The thermal protocol was 95˚C for 10min followed by 40 cycles of 95˚C (20s), 54˚C (20s), and 72˚C (30s). For both procedures, samples were run in triplicate on an ABI StepOnePlus qPCR machine and considered positive if 2/3 wells amplified. If only one well amplified, the samples were rerun in triplicate and considered positive if any of those wells also amplified.
Sampling adequacy–Due to differences in sampling protocols, the number of individuals sampled varied across pathogens (Mean±1SE; Bd, 41±3 hosts per site; echinostomes, 16±1; Ribeiroia,16±1; ranavirus, 13±1). Based on rarefaction estimates (Johnson and Buller, 2011; Hartson et al., 2012), ~10 samples from each site are sufficient for macroparasite species detection in these small systems. For both Bd and ranavirus, we used logistic regression to examine the influence of sample sizein the pooled samples (swabs and tissue) on pathogen detection likelihood (0 = not detected, 1 = detected) across the sampled wetlands. Results indicated that detection was not associated with sample size for either pathogen (Bd: χ2=0.6, R2N=0.01, P=0.429; ranavirus:χ2=3.7, R2N=0.11, P=0.06). These results suggest that sample size had minimal influence on pathogen detection among the sampled wetlands. Assuming that our testing for microparasites was perfect (i.e. tests are 100% specific and sensitive), we also usedthebinomial approximation to assess the probability of detection (P) given a sample size N
P = 1 - (1- r)N
where r is the infection prevalence in the population. At an assumed infection prevalence of 5% and mean samples sizes presented above, our probability of detection for Bd and ranavirus are 88% and 49%, respectively. Thus, lack of detections (absences) for ranavirus should be interpreted cautiously, suggesting our results are likely conservative.
Analyses – We used the checkerboard score (C score) to test for patterns of pathogen species co-occurrence in our presence-absence matrix using EcoSim 7 (Gotelli and Entsminger, 2011). Statistical methods used in the evaluation of species co-occurrence patterns have been discussed extensively in the literature (Wilson, 1987; Stone and Roberts, 1990, 1992; Gotelli, 2000). Based on simulation studies, the C score index was superior to other co-occurrence indices in terms of Type I and II error properties (Gotelli, 2000; Gotelli and Entsminger, 2011). In brief, the C score calculates the average amount of co-occurrence between all unique species pairs in the assemblage, which is then compared to a null model that randomizes the species matrix with the constraint thatcolumn and row sums remain fixed (Gotelli, 2000). The observed C score is compared to the mean C score of 5000 iterations of the null model. If the observed C score is significantly less than the simulated score, pathogens co-occur more often than expected based on the null model. We conducted a Mantel test to assess whether similarity in parasite community compositions was correlated with distance between wetlandsusing PASSaGE 2.0 (Rosenberg and Anderson, 2011). We calculated community similarity (Jaccard’s index of similarity) and inter-wetland distance matrices for use in the analysis. Spatial patterns ofpathogen species richness (0-4) at each site were analyzed using Moran’s I under randomization in the R statistical package (package ‘spdep’; R Development Core Team 2008). Global Moran’s I tests for patterns of spatial dependence in the dataset with positive values indicating that sites in close proximity were more likely to share similar values for pathogen richness. To ensure that each site had at least one nearest neighbor, we defined neighborhoods as the square of the minimum distance between sites.
Supplemental results – We divided the wetlands into categories based on the total number of pathogens detected (1, 2, 3, or 4 pathogen communities) to examine how frequently each pathogen was found in each richness category. For wetlands in which only a single pathogen was detected, echinostomes were the most commonly detectedpathogen(74% of ponds) followed by ranavirus (13%; Fig. A1). Bdand Ribeiroia were found in <9% of wetlands containing a single pathogen species. Echinostomes occurred in 92% and 100% of two and three-species communities, respectively. On average, the remaining pathogens were found in 36% and 67% of the two and three-species communities, respectively. The four most frequently observed pathogen communities were echinostomes alone, echinostomes + Ribeiroia, echinostomes + ranavirus + Bd, and echinostomes + Ribeiroia + ranavirus + Bd (Fig. A1). We observed eight of the 11 possible pathogen species combinations.
We encountered six amphibian species during our field sampling (Bufo boreas, Pseudacris regilla, Rana catesbeiana, R. draytonii, Taricha granulosa, T. torosa; see Appendix Table A1 for sample sizes). All four pathogens were detected in populations of B. boreas, P. regilla, and R. catesbeiana(Fig. A2). Due to their endangered status, we only sampled for Bd in populations of R. draytonii, which tested positive at 47% of sampled sites. For T. torosa, we detected Bd and ranavirus infections but not trematode infections. For sites in which multiple species were sampled and infections were detected at the site, the average percentage of species infected was 72, 91, 83, and 83% for Bd, echinostomes, Ribeiroia, and ranavirus, respectively.
Literature Cited
1
Boyle DG, Boyle DB, Olsen V, Morgan JAT, and Hyatt AD (2004). Rapid quantitative detection of chytridiomycosis (Batrachochytrium dendrobatidis) in amphibian samples using real-time Taqman PCR assay. Diseases of Aquatic Organisms60:141-148.
Gotelli NJ (2000). Null model analysis of species co-occurrence patterns. Ecology81:2606-2621.
Gotelli NJ, and Entsminger GL (2011). EcoSim: Null models software for ecology. in. Acquired Intelligence Inc. & Kesey-Bear, Jericho, VT.
Hartson RB, Orlofske SA, Melin VE, Dillon RT, and Johnson PTJ (2012). Land use and wetland spatial position jointly determine amphibian parasite communities. Ecohealth__:___-___.
Johnson PTJ, and Buller ID (2011). Parasite competition hidden by correlated coinfection: using surveys and experiments to understand parasite interactions. Ecology92:535-541.
Johnson PTJ, and Hartson RB (2009). All hosts are not equal: explaining differential patterns of malformations in an amphibian community. Journal of Animal Ecology78:191-201.
Johnson PTJ, Lunde KB, Thurman EM, Ritchie EG, Wray SN, Sutherland DR, et al. (2002). Parasite (Ribeiroia ondatrae) infection linked to amphibian malformations in the western United States. Ecological Monographs72:151-168.
Kerby JL, and Storfer A (2009). Combined effects of atrazine and chlorpyrifos on susceptibility of the tiger salamander to Ambystoma tigrinum virus. Ecohealth6:91-98.
RDevelopmentCoreTeam (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL
Rosenberg MS, and Anderson CD (2011). PASSaGE: Pattern Analysis, Spatial Statistics and Geographic Exegesis. Version 2. Methods in Ecology and Evolution2:229-232.
Stone L, and Roberts A (1990). The checkerboard score and species distributions. Oecologia85:74-79.
Stone L, and Roberts A (1992). Competitive exclusion, or species aggregation? An aid in deciding. Oecologia91:419-424.
Wilson JB (1987). Methods for detecting non-randomness in species co-occurrences: a contribution. Oecologia73:579-582.
1
Table A1. Sample size and infection status for species tested for Batrachochytrium dendrobatidis (Bd), Echinostomes, Ribeiroia ondatrae, and ranavirus across 85 wetlands in the East Bay region of California, USA. For each wetland, the total number of individuals that were tested for each species and pathogen is indicated. For Bd and ranavirus, the individual samples were pooled together for testing (n ≤20 individuals per sample). Shaded numbers indicate that at least one sample was positive for the pathogen. Species codes are: BUBO = Bufo boreas, PSRE = Pseudacris regilla, RACA = Rana catesbeiana, RADR = R. draytonii, TAGR = Taricha granulosa, and TATO = T. torosa.
Bd / Echinostomes / Ribeiroia / RanavirusWetland / BUBO / PSRE / RACA / RADR / TAGR / TATO / BUBO / PSRE / RACA / TATO / BUBO / PSRE / RACA / TATO / BUBO / PSRE / RACA / TATO
5 Canyon 1 / 12 / 3 / 1 / 12 / 12 / 10
5 Canyon 2 / 21 / 5 / 11 / 11 / 10
5 Canyon 3 / 14 / 2 / 10 / 10 / 10
5 Canyon 4 / 20 / 10 / 10 / 10
6 MDSP / 40 / 40 / 9 / 11 / 9 / 11
Barn / 1 / 9 / 5 / 1 / 9 / 5 / 5
Beaver / 2 / 14 / 1 / 2 / 14 / 1 / 2 / 4 / 5
Big Lake / 40 / 10 / 10
Blue North / 20 / 20 / 20
Blue Oak / 40 / 11 / 11
Bob's Pond / 40 / 2 / 10 / 10
Deer Valley / 40 / 9 / 9
Dowdy / 20 / 10 / 10 / 10
DR02 / 70 / 7 / 11 / 11 / 10
DR04 / 20 / 6 / 13 / 10 / 10 / 10
DVPND002 / 40 / 10 / 10
DVPND003 / 40 / 2 / 10 / 10
Edwards / 40 / 10 / 10
Frog / 40 / 40 / 10 / 10 / 26 / 10 / 26 / 10
GDPND004 / 14 / 2 / 10 / 10 / 10
GDPND005 / 20 / 2 / 1 / 11 / 11 / 10
GDPND006 / 20 / 10 / 10 / 10
GDPND008 / 40 / 23 / 23
GDPND009 / 20 / 9 / 9
GDPND013 / 20 / 10 / 10 / 10
GDPND014 / 20 / 10 / 10 / 10
GDPND015 / 46 / 5 / 10 / 10 / 10
Heron / 40 / 1 / 18 / 1 / 18
Hidden / 40 / 100 / 2 / 14 / 34 / 14 / 34 / 10
Kammerer / 50 / 50 / 3 / 12 / 20 / 2 / 12 / 20 / 2
Kevin's / 65 / 15 / 29 / 21 / 29 / 21 / 18 / 17
Kidney / 40 / 10 / 10 / 10
Lower Leakey / 2 / 42 / 8 / 8
Mallard / 40 / 9 / 9
MTPND014 / 30 / 7 / 18 / 18
Mud 102 / 40 / 2 / 10 / 10
Mud 105 / 40 / 4 / 10 / 10
Mud 20 / 10 / 1 / 10 / 13 / 10 / 13 / 10
Mud 23 / 20 / 7 / 11 / 11
Mud 34 / 20 / 1 / 10 / 10 / 10
Mud 35 / 20 / 4 / 10 / 10 / 10
Mud 46 / 8 / 5 / 10 / 10 / 18
Mud 85 / 14 / 4 / 11 / 11 / 10
Mud 98 / 40 / 10 / 10
Murky Bullfrog / 45 / 65 / 15 / 20 / 27 / 10 / 20 / 27 / 10 / 10 / 18 / 17
North / 40 / 10 / 10
OHPND007 / 40 / 10 / 10
OHPND008 / 40 / 12 / 12
OHPND021 / 40 / 40 / 9 / 11 / 9 / 11
OHPND026 / 50 / 6 / 10 / 10
OHPND027 / 40 / 10 / 10
OHPND032 / 80 / 2 / 10 / 10
OHPND034 / 40 / 13 / 13
OHPND047 / 80 / 10 / 10
OHPND048 / 40 / 2 / 10 / 10
Plano Oasis / 20 / 10 / 10 / 10
Poison Oak / 40 / 10 / 10
PRNTH1 / 20 / 3 / 1 / 18 / 18 / 15
PRNTH2 / 41 / 5 / 14 / 5 / 14 / 10 / 10
PRNTH3 / 45 / 27 / 27 / 18
PRPND001 / 10 / 10 / 10
PRPND002 / 40 / 20 / 8 / 10 / 8 / 10 / 9 / 10
PRPND003 / 5 / 3 / 3 / 1 / 1
PRPND004 / 5 / 2 / 12 / 7 / 2 / 12 / 7 / 9 / 10
PRPND007 / 45 / 40 / 23 / 19 / 23 / 19 / 10 / 10
PRPND008 / 21 / 20 / 18 / 20 / 18 / 20 / 8 / 12
PRTY03 / 12 / 10 / 10
Pumpkin / 40 / 10 / 10
Quick / 20 / 20 / 1 / 1 / 10 / 30 / 10 / 30 / 10 / 18
Rainy Day / 30 / 19 / 19 / 10
Ropeswing / 40 / 10 / 10
Sheep / 20 / 4 / 1 / 24 / 24 / 10
TGIF / 20 / 16 / 16 / 10
Toad / 50 / 40 / 4 / 5 / 9 / 5 / 9
Trout / 40 / 40 / 10 / 10 / 10 / 10
Turtle / 20 / 1 / 20 / 20 / 10
Upper Chase / 2 / 40 / 10 / 10 / 10 / 10
Upper Leakey / 40 / 40 / 10 / 13 / 10 / 13
VPPND001 / 10 / 2 / 11 / 11 / 10
VPPND003 / 12 / 9 / 9 / 10
VPPND005 / 17 / 10 / 10 / 10
VPPND006 / 20 / 2 / 9 / 9 / 10
Washburn / 40 / 9 / 9
Windmill / 26 / 10 / 10
Yerba Buena / 40 / 10 / 10
Figure legends
Figure A1. Number of wetlands testing positive according to each possible combination of Batrachochytrium dendrobatidis (Bd), ranavirus (Rv), Ribeiroia ondatrae (Ro) and echinostomes(E).
Figure A2. Number of sampled ponds in which Batrachochytrium dendrobatidis (Bd), ranavirus, Ribeiroia ondatrae, and echinostomeinfections were detected for each tested species. Species codes are: BUBO = Bufo boreas, PSRE = Pseudacris regilla, RACA = Rana catesbeiana, RADR = R. draytonii, TAGR = Taricha granulosa, and TATO = T. torosa. Numbers above each bar indicate the number of samples that were tested while ‘nt’ indicates that the species was not tested for the pathogen.