Simulating the Advection and Degradation of the Environmental DNA

Subscriber access provided by UNIVERSITY OF TOLEDO LIBRARIES

Environmental Measurements Methods

Simulating the advection and degradation of the environmental DNA of common carp along a river Kei Nukazawa, Yuki Hamasuna, and Yoshihiro Suzuki Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b02293 • Publication Date (Web): 13 Aug 2018 Downloaded from http://pubs.acs.org on August 14, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 42

Environmental Science & Technology

Abstract Art: Conceptual diagram and result of eDNA advection and decay model 61x30mm (300 x 300 DPI)

ACS Paragon Plus Environment


Simulating the advection and degradation of the environmental DNA of common carp along a river Kei Nukazawa1*, Yuki Hamasuna1, Yoshihiro Suzuki1

Corresponding author and proof checker: Kei Nukazawa 1

Department of Civil and Environmental Engineering, Faculty of Engineering, University of

Miyazaki,1-1 Gakuen Kibanadai-Nishi, Miyazaki 889-2192, Japan

*Corresponding author E-mail: [email protected] (KN)

Conflicts of interest: The authors declare no competing ﬁnancial interest.

Short title: Advection and degradation of stream eDNA


Page 2 of 42

Page 3 of 42


Abstract 1

The environmental DNA (eDNA) method is a novel technique for precise and efficient

2

biological surveillance. While eDNA has been widely used to monitor various freshwater

3

organisms, eDNA dynamics in streams remain poorly understood. In this study, we investigated

4

the eDNA dynamics of common carp (Cyprinus carpio) in a forested headwater stream affected

5

by the effluent from a carp farm. We evaluated the longitudinal variation in carp eDNA along a

6

river downstream from the farm and performed a temporal eDNA decay experiment using

7

digital polymerase chain reaction. Based on the resulting decay constants, we built a model to

8

simulate the advection and degradation of eDNA along the studied river. The observed eDNA

9

flux (concentration multiplied by flow rate) decreased exponentially with distance downstream

10

from the farm, and eDNA was detected 3 km downstream of the farm. Although the water

11

temperatures were similar, the eDNA decay constant was lower in autumn than in summer. The

12

simulated eDNA concentration was markedly larger (>10 times) than the observed

13

concentration, suggesting that eDNA removal is accelerated in the stream environment

14

compared to in conventional experimental settings.

15 16



17

Keywords:

18

Advection, eDNA, decay, digital PCR, non-uniform flow

19


Page 4 of 42

Page 5 of 42


20 21

Introduction The environmental DNA (eDNA) technique shows great potential for the

22

biomonitoring of aquatic species as it reduces the effort and cost required for field

23

surveillance1–3. eDNA method also allows the accurate identification of target species based on

24

molecular barcoding4,5 and even has a potential to determine individual numbers or biomass

25

based on the eDNA copy number in lentic and lotic water bodies although these are still under

26

investigation6–8. Researchers have applied eDNA method to various organisms including

27

amphibians9,10, freshwater fishes6,11,12, marine species13–15, reptiles16,17, mollusks18,

28

invertebrates19–21, and terrestrial mammals22,23. Despite the extensive use of this method, eDNA

29

detection remains challenging in lotic systems compared to that in lentic environments as

30

stream eDNA dynamics (e.g., transportation and decay) are not well understood. Consequently,

31

the distribution of source individuals remains unclear in lotic systems even when the molecular

32

markers for the target species are detected.

33

eDNA dynamics in streams and rivers have been investigated for a variety of

34

organisms. In these studies, a “point source” of target species eDNA is generally created or

35

assumed to exist at the uppermost part of the studied river section. Pilliod et al. (2014)24

36

reported that the eDNA released from introduced salamanders was not detected 50 m



37

downstream from the source. In contrast, an investigation of invertebrate eDNA in a river fed

38

by a Swiss lake25 revealed that eDNA could be detected 10 km downstream of the lake, even

39

though the invertebrate species were absent in the river. Jane et al. (2015)26 quantified

40

variations in longitudinal eDNA released from trout at low biomass introduced into the fishless

41

headwaters under different flow rates; eDNA was detected 250 m downstream of the source.

42

The authors provided useful information on longitudinal DNA patterns in natural moving

43

waters using real-time polymerase chain reaction (qPCR), whereas the observed eDNA

44

concentrations were generally low. Because the previous study27 demonstrated that qPCR

45

generated more variable results of eDNA quantification than droplet digital PCR under lower

46

eDNA concentration, further studies are required to clarify the decreasing trend of stream

47

eDNA. Comparison of the travel distance between the above studies is difficult because of

48

differences in scale and hydrological system (e.g., discharge variability). To address this issue,

49

eDNA flux (i.e., concentration multiplying flow rate) is useful to derive eDNA removal rate

50

along a river and further enables intercomparion of the results.

51

Degradation of eDNA is typically evaluated by monitoring a water body that contains

52

the eDNA of a target species but not individuals of that species, over time. This approach has

53

produced a broad range of eDNA decay patterns in different experimental settings14,28. Various


Page 6 of 42

Page 7 of 42


54

factors affecting eDNA decay have been investigated, including water chemistry28, flow and

55

sediment conditions29 , water temperature30, and microbial abundance30. These studies

56

suggested that ambient abiotic and biotic factors markedly affect eDNA degradation,

57

suggesting that eDNA decay patterns play important roles in stream eDNA dynamics. However,

58

to the best of our knowledge, no study has assessed and compared eDNA decay ratios and in

59

situ longitudinal eDNA dynamics.

60

A few studies have attempted to model eDNA dynamics in rivers. Shogren et al.

61

(2017)31 developed a simple conceptual model for estimating eDNA transport distances over

62

short length scales (approximately 50 m). Sanson and Sassoubre (2017)32 modeled eDNA

63

transport for freshwater mussel species in a creek based on eDNA decay and shedding

64

experiments. Sanson and Sassoubre (2017)32 built a novel eDNA transport model based on a

65

plug-flow reactor and the resulting eDNA decay constant; however, they did not consider the

66

effects of the physical characteristics of the river (i.e., topography, flow rate, and flow velocity).

67

Because the decrease in eDNA depended on the decay constant, spatial changes in hydraulic

68

properties (e.g., velocity) can greatly affect the fate of eDNA released into the river. In addition,

69

the influencing factors (e.g., water temperature) on stream eDNA degradation were not argued.

70

Assuming stream eDNA dynamics are characterized by transportation, deposition, and



71

resuspension along with biotic and abiotic degradation, the use of a hydraulic model should

72

provide a better understanding of eDNA dynamics by evaluating how and to what extent a

73

model can mimic in situ stream eDNA patterns.

74

In this study, we investigated longitudinal stream eDNA patterns using in situ

75

surveillance, in vitro eDNA decay experiments, and in silico numerical simulations of eDNA

76

dynamics along a river. We studied a forested headwater stream subject to effluent from a

77

common carp (Cyprinus carpio) farm but without carp individuals in the stream. This

78

experimental setting allows the assessment of how the eDNA of common carp originating from

79

the farm as a point source decreases/changes as it moves downstream. In combination with the

80

in situ investigation, we sampled stream water at the outlet of the effluent for eDNA decay

81

experiments. Finally, using the decay constants derived from the in vitro experiments, we

82

simulated the dynamics of stream eDNA using a one-dimensional advection–decay model

83

based on the non-uniform flow equation.

84 85

Materials and Methods

86

Study river

87

We studied an upstream river section of the Kaeda River in southwest Japan, which


Page 8 of 42

Page 9 of 42


88

has a catchment of approximately 53.8 km2 and a length of 17.5 km [Fig 1]. The upstream

89

catchment is mostly forested without any residential area. A common carp farm is located at the

90

uppermost stream; this farm intakes river water and drains effluent from aquaculture ponds into

91

the river. At the observation dates of this study, the farm accommodated approximately 100

92

adult individuals in the two ponds. The sampling sites were located at approximately 40 m

93

(hereafter, outlet site), 150 m, 900 m, 1.9 km, and 3.1 km downstream of the location of farm

94

effluent discharge [Fig 1]. We also sampled river water ca. 100 m upstream of the effluent

95

discharge location and a major tributary that converges with the main flow approximately 2.3

96

km downstream of the effluent discharge. A survey conducted by a local fishery organization in

97

2015 indicated that no common carp individuals are found in the studied stretch of river

98

(http://www.miyazaki-ngr.jp/contents/archives/3268). Visual observations confirmed that no

99

carp individuals were present near each site on the dates of the observation.

100 101 102

Water sampling and environmental measurements Water samples were collected in triplicate using plastic bottles (1, 2, and 10 L) in

103

August and November of 2017. The samples were transported on ice in cooler boxes and were

104

used for subsequent filtration within 3 h after sampling. The bottles and cooler boxes were



105

sterilized with 10% bleach for at least 30 min prior to the sampling. According to preliminary

106

surveys, we sampled different water volumes based on the different eDNA concentration at the

107

studied sites. We collected 1 and 10 L of water at the outlet site and the site 3.1 km downstream

108

of the effluent, where high and low eDNA concentrations were expected, respectively. The

109

sample volume was 2 L at all other sites. For the eDNA decay experiment, we collected 20 L of

110

water at the outlet site, where discharged effluent is mixed with river water. We placed 2-L

111

bottles filled with sterilized distilled water in the cooler together with the samples (i.e., cooler

112

blank) to check for contamination during sampling33.

113

Water depth and current velocity were measured using a velocity meter (VR-301,

114

KENEK) to determine the flow rate at each site. A full description of the environmental

115

measurements including the collection of basic water quality data can be found in Table S1.

116 117 118

eDNA decay experiments The water samples collected at the outlet site were moved to a sterilized container and

119

were stirred using a magnetic stirrer (ca. 900 rpm) throughout the experimental period to

120

imitate an environment of flowing water. The decay experiments were performed in the dark to

121

eliminate the effects of ultraviolet radiation. Three 1-L aliquots of water were sampled from the


Page 10 of 42

Page 11 of 42


122

container after 1, 3, 6, 10, 24, 48 h and were used for subsequent filtration. The water

123

temperature was measured using a pendant temperature logger (UA-001-08, onset) every 30

124

min throughout the experiment. The temperatures for the experiments in August and November

125

were 22.11°C ± 1.05°C and 21.35°C ± 2.59°C (mean ± sd.), respectively.

126 127 128

Filtration, DNA extraction, and measurement of double-strand DNA concentration The water samples were filtered using a glass fiber filter with a pore size of 0.7 µm

129

(GE Healthcare Japan, Tokyo). Two filters were used for each sample, which was obtained from

130

the site 3 km downstream of the effluent to avoid clogging. Filter funnels, bases, clamps, and

131

tweezers were sterilized by soaking in 10% bleach for 10 min prior to each filtration process.

132

Filters were stored in a freezer at −20°C until subsequent DNA extraction. DNA was extracted

133

following the protocol proposed in the previous study34 using DNeasy PowerSoil Kit (Qiagen,

134

Hilden, Germany). Eichmiller et al. (2016b)34 reported that PowerSoil Kit exhibited low

135

variation in eDNA quantification for common carp and no detectable inhibition. In brief,

136

tweezers and scissors that had been sterilized by soaking in 10% bleach were used to cut the

137

filters into sizes of 1×3 mm. DNA was then extracted from the small fragments of the filters

138

following the protocol of the kit manufacturer. The concentration of double-strand DNA



139

(dsDNA) in the extracted DNA was measured using a fluorometer (Quantus Fluorometer,

140

Promega, WI). The extracted template DNA solutions were stored in a freezer at −20°C until

141

the subsequent PCR step.

142 143 144

Quantification of common carp eDNA using digital PCR To quantify the eDNA of common carp, we used a digital PCR (dPCR) system

145

(QuantStudio 3D digital PCR system, Applied Biosystems, CA) with previously designed

146

primer and probe sets specific to common carp and targeting mitochondrial cytochrome b12.

147

The specificity of the assay was validated for the study area (Supporting Information). Unlike

148

conventionally used qPCR, which provides a relative DNA concentration, dPCR allows the

149

absolute quantification of target DNA concentration. In past studies, compared to qPCR, dPCR

150

resulted in lower variation in quantified eDNA in goby fishes35 and more stable quantification

151

results under low eDNA concentrations for common carp27. In this study, the reaction mixture

152

for eDNA quantification contained 1× QuantStudio 3D Digital PCR Master Mix (Applied

153

Biosystems, CA), 2 µL DNA temperate solution, 900 nM of each primer (the forward and

154

reverse primers), and 125 nM of Taqman probe. The mixture was dispensed to independent

155

wells of a QuantStudio 3D Digital PCR 20K Chip using a QuantStudio 3D Digital PCR Chip


Page 12 of 42

Page 13 of 42


156

Loader (Applied Biosystems, CA). The endpoint PCR reaction was performed using a thermal

157

cycler (ProFlex, Applied Biosystems, CA). The PCR reactions followed the default protocol of

158

dPCR that comprised polymerase activation at 96°C for 10 min followed by 40 cycles of

159

annealing and extension at 60°C for 2 min, denaturation at 98°C for 30 s, and final extension at

160

60°C for 2 min. The target eDNA concentration was then quantified based on the number of

161

wells determined to be ‘positive’ (i.e., the well has a higher fluorescence intensity than the

162

background) using a QuantStudio 3D digital PCR system and QuantStudio 3D Analysis Suite

163

software (Applied Biosystems, CA). We adopted the software’s default threshold of

164

fluorescence intensity to discriminate positive and negative wells. Note that if the number of

165

positive well is less than 4 and the intensity of the positive wells was as low as negative wells,

166

the positive wells were considered as false. The dPCR procedure was performed in a separate

167

room from the filtration and DNA extraction processes, and none of the instruments were

168

transferred between the rooms.

169 170 171 172

Data analysis We calculated the flux of common carp eDNA by multiplying the eDNA concentration with flow rate for each site. Subsequently, the flux was used to calculate the



Page 14 of 42

173

removal rate via the following formula (F0 – Fi)/ F0, where F0 is the flux at the outlet site and Fi

174

is the flux at the site i. This allowed the evaluation of eDNA abundance loss with distance from

175

the carp farm. Following Thomson et al. (2012a)14, we fit eDNA concentration with elapsed

176

time using the formula dC/dt = −βC, where C is the eDNA concentration (copies/L), β is the

177

decay constant (/h), and t is the elapsed time (h). This non-linear regression model was built

178

using the nls package in R version 3.4.3 (R core team, 2017). We used Pearson’s correlation

179

coefficient to assess the association between the eDNA and dsDNA concentrations.

180 181 182

Advection and decay model for simulating stream eDNA dynamics Using the decay constants derived from the eDNA decay experiments in August and

183

November of 2017, we simulated the one-dimensional advection in eDNA with degradation

184

along the studied river section. We first assumed a simple rectangular flume and computed the

185

water depth h throughout the studied river section using the following formula for

186

one-dimensional non-uniform flow:

187 188

,

(1)

where H is water level (m), Q is flow rate (m3/s), B is river width (m), g is acceleration due to


Page 15 of 42


189

gravity (m/s2), and n is roughness coefficient (n = 0.06). We acquired digital elevation model at

190

a spatial resolution of 5 m from the Geospatial Information Authority of Japan and defined the

191

grids of the studied river using the river data from the Ministry of Land, Infrastructure,

192

Transport, and Tourism (MLIT). This physical-based model was chosen to better simulate

193

stream eDNA dynamics because it could reflect complex characteristics of the river (i.e.,

194

topography and flow velocity) and be potentially useful for integrating spatially heterogeneous

195

biotic and abiotic processes (e.g., retention and resuspension). We partitioned the studied river

196

section into two parts: before the convergence of the tributary (0 to 2.3 km downstream of the

197

outlet site) and after the convergence of the tributary (2.3 to 3.1 km downstream of the effluent).

198

A constant flow rate and river width was assigned to each section based on the observed data.

199

The current velocity u (m/s) for each grid was derived from the continuity equation.

200 201

202

To simulate eDNA dynamics load into river, the decay constant of eDNA was integrated to the advection equation as follows:

,

(2)

203

where t is time (sec), and x is spatial resolution (m). We computed the advection and

204

degradation independently for August and November using the mean eDNA concentration at

205

the outlet site as constant input to the uppermost grid (i.e., the grid of the outlet site). First, we



206

used the decay constants derived from the eDNA decay experiments (i.e., β = 0.086 for August

207

and β = 0.017 for November). In subsequent simulations, we set different decay constants (β =

208

0.5, 1, 2, 4 and 10) to understand how variation in the decay constant affects the simulation

209

results. The model accuracy for each decay constant was evaluated based on the root mean

210

square error (RMSE). The evaluations were based on the simulated eDNA concentration 16 h

211

after start of the experiment; the eDNA concentration showed a plateau at this time at the site

212

3.1 km downstream of the effluent (Fig S1).

213

We derived decay constants so that the simulated eDNA concentration agrees with

214

the concentration at the site 1.9 km downstream of the effluent and predicted a detectable

215

distance of the eDNA downstream in response to different initial eDNA concentration and

216

filtration volume. For this purpose, we defined the thresholds of eDNA concentration which

217

determine success or failure of detection as 20, 40, 100, 200, 1,000, and 4,000 copies/L for the

218

filtration volume of 10,000, 5,000, 2,000, 1,000, 200, and 50 mL, respectively. These threshold

219

concentrations approximately correspond to two positive wells in the dPCR system because a

220

negative sample (i.e., without target eDNA) could result in this extent of the positive well

221

number. The eDNA concentration at the 3.1 km downstream site was not used for calibration

222

because the sampling volume at this site was different from those at the other sites.


Page 16 of 42

Page 17 of 42


223 224

Results and Discussion

225

Water temperature was 22.0–23.0°C in August and 13.6–14.4°C in November (Table

226

S1). The measured pH and Electric conductivity (EC) values indicated only small differences in

227

water quality among the study sites (Table S1). The discharge observed in August (0.063 to

228

0.283 m3/s) was slightly lower than that observed in November (0.115~0.409 m3/s; Table S1).

229

No common carp eDNA was detected in the negative control samples in August or

230

November, indicating that no contamination occurred during the field surveys, filtration, DNA

231

extraction, and dPCR process. Carp eDNA was also undetected in all samples taken from

232

upstream of the carp farm and the tributary in both August and November. This indicates that

233

eDNA inputs other than the effluent from the carp farm were negligible in the present study.

234 235 236

Longitudinal profile of stream environmental DNA The concentration and flux of carp eDNA decreased exponentially with distance

237

downstream from the carp farm in both August and November (Fig. 2). Notably, the removal

238

rate of eDNA increased dramatically with the distance from the farm until 900 m downstream

239

(removal rate = 0.73 both in August and November; Table S2); as distance downstream from



240

the farm increased beyond 900 m, the removal rate increased more gradually. Earlier works

241

investigated longitudinal eDNA concentration to understand the fate of eDNA released into

242

rivers26. However, river water containing DNA may be diluted by water from tributaries and

243

subsurface flow, complicating the interpretation of trends in eDNA along the river.

244

Furthermore, because of differences in discharge, eDNA concentration probably varies between

245

rivers even when species abundance is equal; this can make it difficult to determine the

246

individual number or biomass of the target species. In contrast, our study successfully excluded

247

the effect of eDNA dilution using the flow rate observed at each sampling site and evaluated the

248

increasing trend of eDNA removal rate.

249

In this study, eDNA was detected at the site 3.1 km downstream of the carp farm in all

250

cases (mean ± se. = 110.11 ± 55.17 copies/L in August; 131.77 ± 17.42 copies/L in November),

251

suggesting that eDNA originating from the carp farm is transported over 3 km distance along

252

the river. In earlier studies, eDNA of target salamander and trout species was not detected or

253

was found in very small concentrations 50–250 m downstream the source24,26. However, these

254

studies were based on the introduction of exotic species, resulting in smaller individual number

255

and biomass compared with that in this study. Deiner et al. (2014)25 detected the eDNA of

256

invertebrate species 10 km downstream from the source (a natural lake that harbors the target


Page 18 of 42

Page 19 of 42


257

species). Taken together, whereas the scale of hydrological system differed between these

258

studies, transport distance of eDNA could depend on the initial quantity of eDNA loading,

259

which could be dependent on population size or biomass of the species source.

260

The mean eDNA flux at the outlet site was significantly lower in November (mean ±

261

se. = 412,474 ± 78,491 copies/s) than in August (653,309 ± 40,648 copies/s) (t-test, P < 0.05).

262

However, the eDNA removal with the downstream distance from the source was more

263

moderate in November compared to in August (Table S2). Fukumoto et al. (2015)36 found that

264

among seasons, amphibian eDNA in a temperate climate was most detectable in the winter.

265

They attributed this finding to the lower water temperature in the winter. Thus, water

266

temperature might have contributed to the different patterns in eDNA concentration observed in

267

August and November.

268 269 270

eDNA decay experiments In both August and November, the concentration of common carp eDNA decreased

271

over time (Fig. 3). While the initial eDNA concentration at the outlet site was significantly

272

lower in November than in August (as mentioned above), the eDNA concentration after 48 h

273

was lower in August (mean ± sd. = 834.04 ± 202.24 copies/L) than in November (1,351.26 ±



274

378.21 copies/L). Based on non-linear regression, we derived the following decay constants: β

275

= 0.086 in August (P < 0.01) and β = 0.017 in November (P < 0.01). These parameters indicate

276

that eDNA degraded faster in August than in November, even though the water temperatures

277

were similar. Earlier studies found that ambient biotic factors (microbial activity and

278

extracellular enzyme) and abiotic factors (water temperature, ultraviolet radiation, and water

279

quality) play important roles in eDNA decay32,37. Previous eDNA decay experiments conducted

280

in similar settings as this study found that the temporal eDNA degradation for common carp

281

suppressed as water temperature decreased28,30,38. In these studies, the river/lake water samples

282

were incubated at different temperatures. Thus, the inconsistent results between these past

283

studies and the present study may be explained by seasonal differences in the driving factors of

284

eDNA degradation in stream water (e.g., the effects of the microbial community).

285

Many studies involving eDNA decay experiments speculated on the effects of

286

microbial activity on eDNA degradation34,39. Tsuji et al. (2017)30 found an insignificant

287

negative correlation between fish eDNA and microbial abundance in the similar eDNA decay

288

experiment. Lance et al. (2017)40 reported that microbial abundance strongly affected eDNA

289

degradation. Earlier studies demonstrated that dsDNA is a good substitute for bacterial density

290

in natural water samples, including humic water with high background fluorescence41. In this


Page 20 of 42

Page 21 of 42


291

study, the dsDNA concentration clearly increased over time during the experiments compared

292

with the concentration in the river samples (

Simulating the Advection and Degradation of the Environmental DNA

Recommend Documents