effect of six types of artificial nighttime lights on …

26

Upload: others

Post on 24-Jan-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

EFFECT OF SIX TYPES OF ARTIFICIAL NIGHTTIME LIGHTS ON THE ATTRACTION

OF SUBYEARLING SALMONIDS IN THE NEARSHORE AREA OF SOUTH LAKE

WASHINGTON

FINAL REPORT TO KING COUNTY FLOOD CONTROL DISTRICT

by

Roger A. Tabor U. S. Fish and Wildlife Service, Lacey, Washington

Elizabeth K. Perkin McDaniel University, Westminster, Maryland

David A. Beauchamp U. S. Geological Survey, Seattle, Washington

Lyle L. Britt, Rebecca Haehn NOAA Fisheries, Seattle, Washington

John Green, Tim Robinson RGB Optics, Seattle, Washington

Scott Stolnack1, Daniel W. Lantz King County, Seattle, Washington

and

Zachary J. Moore2

U. S. Fish and Wildlife Service, Lacey, Washington

May 2019

1 Retired 2Present address: King County, Water and Land Resources Division, Seattle, Washington

3

Summary

We conducted field experiments in the nearshore area of Lake Washington in 2017 and

2018 (February – April) to determine the degree that different light sources attract subyearling

salmonids (Oncorhynchus spp.). We tested six lights with different spectral characteristics: four

LED lights, an incandescent light, and a high-pressure sodium light (HPS). Light were adjusted

to produce a desired light intensity of 20 lx at the water surface across a rectangular 4 m

alongshore by 5 m offshore patches. The locations of experimental treatments and no-light

control units were randomly assigned with a 20-m buffer between them. Experimental trials

were conducted along a uniform 124-m shoreline section from February to April and spanned

bottom depths of 0.0-0.8 m to correspond with peak nearshore timing and habitat use of

subyearling salmonids, specifically Chinook salmon (O. tshawytscha) and sockeye salmon (O.

nerka). Light systems were turned on shortly before dusk and remained on through one hour

after the posted astronomical twilight time to begin sampling fish. Fish densities were

determined by beach seining through each experimental treatment and control unit. In

comparison to the control units, all light types attracted subyearling salmonids. In general,

juvenile sockeye salmon showed stronger attraction to the lights than Chinook salmon. Neither

species exhibited detectable differences among light types in either year. In conclusion, resource

managers should consider the potential impacts of artificial nighttime lighting and recognize that

most if not all currently available lights attract juvenile salmonids to some degree and using

other methods to reduce the effects of light pollution may be more appropriate.

4

Introduction

Artificial light at night (ALAN) is a common feature of urban development (Longcore

and Rich 2004), yet how ALAN influences the environment including effects on aquatic systems

is often not well known (Perkin et al. 2011). With increasing urban development and expanding

use of brighter and more energy efficient lighting systems, comes a growing need to understand

how ALAN can affects key components of aquatic systems.

In a review of the methods to reduce the effects of nighttime light pollution, Gaston et al.

(2012) listed five categories: 1) reduce amount of lighting, 2) reduce lighting duration, 3) reduce

light spill to nontargeted areas, 4) reduce light intensity, and 5) change the spectral composition.

Of these five categories, the first four are straightforward and are methods that reduce the overall

amount of light on a location of interest. The fifth category, change in spectral composition, has

often been suggested; however, the spectral response can vary widely between organisms

(Marchesan et al. 2005; Perkin et al. 2011; Gaston et al. 2012; Pawson and Bader 2014).

Research is needed on how individual species response to differing spectra but also on how it

may affect species interactions (Davies et al. 2012; Davies et al. 2013).

In the Pacific Northwest, many salmon stocks have declined, increasing the urgency to

better understand habitat relationships including the impacts of urbanization and associated

ALAN. Because certain juvenile life stages of salmon occupy shallow nearshore habitats, they

may be particularly responsive to ALAN. Research has demonstrated that juvenile salmonids are

attracted to lights which may cause them to be more vulnerable to predators (Tabor et al. 2004a;

Celedonia et al. 2011; Tabor et al. 2017). Within the Puget Sound ecoregion, Chinook salmon

(Oncorhynchus tshawytscha) is currently listed as threatened under the Endangered Species Act

(ESA; Federal Register 64 FR 14208, March 24, 1999), and because they inhabit shallow

nearshore waters (i.e., < 1 m deep) during the fry and parr stages (Tabor et al. 2011) they can be

influenced by ALAN. Recovery efforts have often focused on improving habitat conditions,

while potential effects of ALAN have largely been ignored. Information on the effects of ALAN

will help manage existing and future shoreline development.

5

Research on the effect of ALAN on juvenile salmonids has primarily focused on the

effect of light intensity (Tabor et al. 2004a; Tabor et al. 2017). In situations where appropriate

nighttime lighting is necessary for safety, security, or other considerations, using a lighting

system with a spectrum that is less sensitive to juvenile salmonids may minimize deleterious

effects. Juvenile Atlantic salmon (Salmo salar) display strong sensitivity for short wavelength

light (blue-rich) (Hawryshyn et al. 2010); therefore, using lights that are blue-rich may be more

harmful than other lights (IDA 2010). Avoiding these types of lights has also been proposed for

other animals (IDA 2010). A particular concern has been that LEDs emit a strong spike at 450-

460 nm (i.e., blue spike; IDA 2010). In 2017 and 2018, we conducted field experiments to test

new LED lights and other conventional light sources to provide better information to resource

managers on how best to reduce the effects of ALAN on juvenile Chinook salmon and other

juvenile salmonids while maintaining appropriate lighting for safety considerations. We

hypothesized that lights with less short wavelength light (e.g., 400-500 nm) would be less

attractive to juvenile salmonids. Our specific objectives were to: 1) compare the response (fish

density) of juvenile Chinook and sockeye salmon to light sources of different spectral

characteristics and no-light controls; and 2) determine whether a seasonal or ontogenetic trend in

response was evident within or between the salmon species during the peak period of

immigration to the lake.

Methods

Study design and experimental treatments.— To assess if subyearling salmonids

displayed stronger positive phototaxis to different light types, we conducted field experiments at

a 124-m long shoreline section in Gene Coulon Memorial Beach Park in Lake Washington

(Figure 1). This shoreline section was the same site used in 2014 by Tabor et al. (2017) to test if

increased light intensity levels increased attraction (positive phototaxis) of subyearling

salmonids. Also, this shoreline section was selected because it had relatively uniform habitat

conditions, minimal direct artificial lighting, was easily accessible, and was located near the

outlet of a major salmon spawning river (Cedar River) so subyearling salmonids would be

6

relatively abundant. Overall substrate types within the study area were mainly composed of

coarse and fine gravel and occasional interspersed cobble. The shoreline slope in this area

ranged from 4.1 to 11.7%.

The 124-m shoreline section was divided into six 4-m long experimental units with a 20-

m buffer section on each side of each unit. Preliminary testing indicated that a 20-m buffer

between experiments was adequate because the amount of light be detected from an adjoining

unit was minuscule. For each experimental trial, the location of each treatment was randomly

selected. Overall, we tested six light types; of these, five were tested in 2017 and three were

tested in 2018 (Figure 2). For the 2017 experimental trials, the five lights tested were LED 5000

K [blue spike], LED 2000 K, LED with yellow filter, high-pressure sodium, and incandescent

and one no-light control. In 2018, we simplified our experimental design to test only three LEDs

that strongly differ in spectral composition (LED 5000 K, LED with yellow filter, and LED with

red filter) plus two randomly-assigned control treatments for a total of five treatments. The 24 m

shoreline section at the south end was not used in 2018. In addition to the randomly-assigned

treatments, we sampled one additional no-lit unit on most nights to determine if subyearling

salmonids were near the lights but just outside of our seining area. The extra unit (termed

“Edge”) was sampled either in between two lit units or adjacent to the last lit treatment at the

south end of our shoreline section.

Experimental trials were conducted in February, March, or April in 2017 and 2018 to

correspond with peak nearshore rearing of subyearling salmonids (Koehler et al. 2006; Tabor et

al. 2011). A total nine experimental trials were conducted in 2017 and eight in 2018 (Table 1).

Water temperatures ranged from 6ºC to 8.5ºC and turbidity was typically less than 3.0 NTU

(Table 1). Ambient light levels ranged from 0.03-0.08 lx on clear, moonless nights, 0.1-0.12 lx

on clear, moonlit nights, and to 0.11-0.25 lx on cloudy nights; similar values as reported from

nearby lower Cedar River (Tabor et al. 2004a). To help control for the potentially confounding

effects of wave action, experimental trials were only conducted on calm nights.

7

FIGURE 1.— Location of the Lake Washington 124-m shoreline section in the north part of Gene Coulon Park (City of Renton) that was used in this study, February-April 2017 and 2018. The red lines represent the north and south boundaries of the study shoreline.

Lake Washington

Study shoreline

8

FIGURE 2.—Spectral comparisons of the six light systems used in this study. The top panel displays the four LED light sources and the bottom panel displays the other two light sources. The 2017 trials included five of the six light systems (all but the red filter LED) and the 2018 trials only included the 5000 K, yellow filter, and red filter LED light systems.

9

TABLE 1. Dates and environmental conditions of experimental trials used to test different light types.

Each light source was mounted in a 40 cm long by 30 cm wide by 20 cm high metal box

with an open bottom (Figure 3). The inside of the box was lined with white foam board to allow

for a more even distribution of light. The light source was mounted on one side of the box and a

piece of foam board was added directly over the light source and covered approximately half of

the box. This kept the light from shining directly on the water and allowed for a more even

distribution of light. Each light source was mounted on the top a 2-m post placed along the

shoreline in the middle of the experimental unit. Prior to the experiments, each light was set up

on a 2-m post in a dark room and the light intensity was adjusted with a dimmer switch to get the

desired light intensity of 20 lx at the floor (i.e., water surface). Dimmer switches were marked to

indicate the proper light intensity level. Light intensities were also measured in the field with an

Extech Instruments light meter (model 401036) to insure the desired light intensity was achieved.

A small generator was used to power the lights. The area illuminated extended approximately 4

m along the shore and 5 m offshore with the highest light intensities being in the center of the

experimental unit and 2 m offshore. The lighted area extended out to where the water depth was

~0.8 m and the maximum light intensity levels occurred where the water depth was ~0.3 m.

10

FIGURE 3.—Photographs of light systems used to test the attractive quality of lights with different spectral composition. The top photograph provides an underside view of the light fixture (LED 2000 K light is shown); the small blue box on the upper right side is the dimmer switch. The bottom photograph displays the setup along the shoreline of Gene Coulon Park in Lake Washington, February 22, 2017. The flagging in the water delineates the 4-m shoreline length of the experimental unit. Photo credit: Roger Tabor, USFWS.

11

On each treatment night, light systems were set up and turned on shortly before dusk.

We waited one hour after the posted astronomical twilight time to beginning sampling fish

(Table 1). We only waited one hour to minimize the chance of a sudden change in the weather

conditions (i.e., wind or rain). This short period also minimized the time we would have to

supervise the lights and minimized the time juvenile Chinook salmon and other subyearling

salmonids would be vulnerable to any increased predation in the lighted area. The abundance of

subyearling salmonids in each of the 4-m long shoreline sections was determined from separate

beach seine sets. One beach seine set was deployed through each experimental unit (Figure 3).

Lights remained on during beach seining to minimize changes in fish behavior. The total amount

of time to beach seine all experimental units varied from 45 minutes to one hour in 2017 and 30

to 45 minutes in 2018, therefore the total amount of time each experimental unit was lit after

astronomical twilight varied from one to two hours. We assumed the change in fish abundance

between one and two hours was minor compared to the change during the first hour. Also,

seining was always conducted systematically from the north end of the study area to the south

end to minimize any bias.

To collect fish, we used a small beach seine that was 6 m long and 1.3 m deep with a 1.15

m deep by 1.3-m long bag in the middle. The mesh size in the wings was 8-mm stretch and 4-

mm stretch in the bag. The seine was set offshore and parallel to shore then pulled to shore

towards the two corners of the experimental unit, so the seine encircled nearly the entire lit area

(Figure 4). The net was set in water approximately 1.2 m deep. The additional non-illuminated

area sampled by the beach seine was deeper than that typically used by juvenile Chinook salmon

at night (Tabor et al. 2011). Thus, we believe the vast majority of fish collected were in the

illuminated area. Also, snorkel observations around other lighted areas supported this

assumption (R. Tabor, personal observation). After each beach seine set, all fish were placed in

a bucket and the first five Chinook salmon collected with a small dip net were transferred to

another bucket and later measured for fork length (mm). A subsample of sockeye salmon (O.

nerka) fry were also set aside for length measurements. The remaining fish were identified,

counted, and immediately released. After all experimental units were seined, the fish that were

set aside were aestheticized with MS-222 and measured. After they were allowed to recover,

they were released back to the lake.

12

FIGURE 4.—Photograph of the small beach seine being pulled to shore to capture fish in the vicinity of the

light system (middle of photograph), Gene Coulon Park, Lake Washington. The light system used in this photograph was the LED red filter light. Photo credit: Parker Miles Blohm, KNKX Radio.

Data analysis.--Prior to analyses, the data were log transformed to meet requirements for

homoscedasticity and normality. Data from 2017 and 2018 was analyzed separately, since a

different set of lights were used.

Linear mixed models were used to determine the effects of light treatments (fixed factor)

on density (or catch per set) of juvenile Chinook salmon and sockeye salmon, and the combined

abundance of prickly sculpin (Cottus asper) and coastrange sculpin (C. aleuticus). Because the

first sampling date in 2017 took place before sockeye salmon fry were active in Lake

Washington, it was excluded from the sockeye salmon analysis. Sampling date and water

turbidity were both included as random factors in the model. We only included the random

effects in the final models that resulted in the lowest Akaike’s information criterion. The “nlme”

package (Pinheiro et al. 2018) was used for this analysis.

13

To determine differences between specific light treatments, we used a Tukey HSD

following an ANOVA. All analyses were carried out in R (R Core Team 2018).

Results

In 2017, experimental trials were conducted on nine dates from February 7 to April 11.

During the first five dates (February 6 to March 16), the number of subyearling salmonids

captured ranged from 84 to 465 (14.0 to 66.4 salmonids/set; Appendix A) while in the last four it

ranged from 30 to 49 (4.3 to 7.0 salmonids/set; Appendix B). Because few subyearling

salmonids were collected in the last four trials, our data analysis primarily focused on the first

five dates. In 2018, we conducted eight experimental trials from February 7 to March 5 and the

number of subyearling salmonids ranged from 71 to 313 (14.2 to 62.6 salmonids/set; Appendix

C). In 2017, subyearling salmonids consisted of 47.7% Chinook salmon and 52.3% sockeye

salmon; while in 2018, catch composition of salmonids was 92.2% Chinook salmon and 7.8%

sockeye salmon.

The lengths of juvenile Chinook salmon were generally similar throughout February

trials; whereas, lengths in late March and April trials tended to be slightly larger than in February

trials (Table 2). The February mean length of juvenile Chinook salmon for the four experimental

trails was 40.5 mm FL in 2017 and 40.9 in 2018 while the March mean length was 42.0 mm FL

in 2017 and 43.5 in 2018. April trails were only conducted in 2017 and the mean length of the

two trials was 45.7 mm FL. We did not measure many sockeye salmon but there did not appear

to be a noticeable increase in length in the March and April experimental trials compared to the

February trials (Table 2).

Besides Chinook salmon and sockeye salmon, few other salmonids were captured which

only included 12 cutthroat trout (O. clarkii; mean FL, 168.3 mm; range, 144-197 mm FL) and 3

yearling coho salmon (O. kisutch; mean FL, 117 mm). Sculpins were routinely captured

(Appendices A-C) and consisted of 53% prickly sculpin and 47% coastrange sculpin. We

visually separated sculpin into those less than 75 mm total length (TL) and those equal to or

14

greater than 75 mm TL. Overall, only 18% were equal to or greater than 75 mm TL. Other fish

occasionally captured included sunfish (Lepomis spp.), juvenile smallmouth bass (Micropterus

dolomieu), and juvenile yellow perch (Perca flavascens).

TABLE 2.-- Fork lengths (n, mean, SD) of juvenile Chinook salmon and sockeye salmon from nine experimental trials in 2017 and eight trials in 2018. n = the total number of fish measured for each trial; a maximum of five fish were measured from each experimental unit.

In 2017, all lighted experimental units contained higher densities of subyearling salmon

than the control areas (i.e., unlit areas). Significantly more Chinook salmon were caught at lit

areas than control areas (F = 6.979, df = 6, p = 0.0003). A Tukey HSD revealed that the number

of Chinook salmon caught at the 5000 K LED was the exception which was not significantly

different than the number caught in the no-light control treatment (p = 0.065; Figure 4A). No

differences were detected among the five light types (Tukey HSD tests, P > 0.05). Similarly,

significantly more sockeye salmon were caught at lit areas, including the 5000 K LED, than

15

control areas (F = 13.951, df = 6, p < 0.0001; Figure 4B) and no differences were detected

among the five light types (Tukey HSD tests, P > 0.05).

A

B

FIGURE 4.-Boxplots of the number of subyearling salmonids captured in experimental units during five

trials, Lake Washington, February 7 - March 5, 2017. Significantly fewer juvenile Chinook salmon (A) and sockeye salmon (B) were caught at locations that were not lit than any of the lit sections. The dark band in the middle of the boxplot represents the median value, the bottom of the box is the 25th percentile and the top of the box is the 75th percentile. The whiskers represent up to 1.5 times the interquartile range, and any data beyond that is represented by a dot. Letters above each boxplot denote treatments that are significantly different from one another (Tukey HSD tests). “IN” = incandescent, “HPS” = high pressure sodium, “5k” = 5000 K LED, “2k” = 2000 K LED, “Yellow” = yellow-filtered LED.

In 2018, Chinook densities were significantly lower in the control units than in the

variously lighted treatments (ANOVA F = 20.174, df = 4, p < 0.0001; Tukey HSD, p < 0.05;

Figure 5A). No juvenile sockeye salmon were ever caught at the control areas in 2018. The

number of sockeye salmon caught at 5k LED lights was 4.6 ± 1.1, yellow-filter LED: 6.0 ± 2.2,

16

red-filter LED: 2.3 ± 1.1 (ANOVA F = 13.347, df = 4, p < 0.0001; Figure 5B). However,

because of low replication of the “edge” treatment, there was no statistical difference in the

number of Sockeye salmon captured there and at the red-filtered LED (Tukey HSD, p = 0.29).

Similar to 2017 results, no differences were detected among the three light types for either

Chinook salmon or sockeye salmon (Tukey HSD tests, P > 0.05).

A B

FIGURE 5.-- Boxplots of the number of subyearling salmonids captured in experimental units during eight

trials, Lake Washington, February 6 - March 16, 2018. In 2018, significantly fewer juvenile Chinook salmon (A) were caught at locations that were not lit than any of the lit sections. No juvenile sockeye salmon (B) were caught at the unlit sections at all, while a small number were captured in the different light treatments. Letters above each boxplot denote treatments that are significantly different from one another (Tukey HSD tests). “5k” = 5000 K LED, “Yellow” = yellow-filtered LED, “Red” = red-filtered LED.

17

In 2017, 0.6% (5 of 789) of the total catch of sockeye salmon fry were caught in the

control treatments; whereas, 8.8% (58 of 662) of the total catch of Chinook salmon were caught

in the control treatments. In 2018, no sockeye salmon (0 of 103) were caught in the control

treatments, whereas 10.0% (121 of 1,212) of the Chinook salmon were caught in the control

treatments. On each sample night in 2017 and 2018, the percent of the total Chinook salmon

catch that was caught in the control units was higher than for sockeye salmon fry.

A total of 269 sculpin (prickly sculpin and coastrange sculpin) were captured during the

2017 experiment, representing about 15% of all fish captured in the study that year. In 2018, a

total of 156 sculpin (prickly, coastrange, and unidentified) were caught in the experiment,

representing about 11% of all fish captured in the study. However, there were no significant

differences in the total number of sculpin caught at any of the treatments. Not enough cutthroat

trout or coho salmon were captured to allow for an analysis to be done.

Discussion

We tested a variety of light sources that encompassed a wide range of spectra under

realistic field conditions, and in comparison to the no-light control treatments, both juvenile

Chinook salmon and sockeye salmon appeared to be attracted to all the light treatments, whereas

we were unable to detect any differences in juvenile salmon density among the light sources.

Originally, we had hypothesized that the LED 5000 K light would attract more subyearling

salmonids than the other lights, because it has a blue spike (450-460 nm), and other species of

juvenile salmonids are sensitive to this short wavelength light (Hawryshyn et al. 2010). In 2018,

we also used a LED red-filter light to provide a sharp contrast to the LED 5000 K light, but the

LED red-filter light attracted similar densities of subyearling salmonids as the LED 5000 K or

the LED yellow-filter light. We had expected the red-filter light to attract the fewest fish,

because its longer wavelengths theoretically fall outside the sensitive range for some salmonids.

Perhaps we needed to test lights that had a spectral curve farther to the right (e.g., deep red light).

Stien et al. (2014) found that a deep red light with a peak at 650 nm did not affect the nighttime

swimming depth of Atlantic salmon in net pens; whereas, six shorter wave length lights,

18

including a red light (peak at 620 nm), did affect their swimming depth. In our study, the red-

filter light had spectral peak at 600 nm. Even if a deep red light proves to reduce attraction of

subyearling salmonids, it may be impractical (e.g., cost and light color) for most commercial and

residential applications. Additionally, the lights we used are generally broad-spectrum lights and

additional tests may needed on narrower-spectrum lights such as LPS (low-pressure sodium)

(Davies et al. 2013).

Another possible explanation why we found no differences among light sources is

because we conducted our experiments under field conditions and small differences among light

treatments may have been difficult to detect due to natural variations among experimental units

and trials (i.e., sample dates). An important element in our study design was that good numbers

of juvenile salmonids would be present, and they would be well distributed along our shoreline

section. Patchiness in the distribution of subyearling salmonids could have added additional

noise to the experiments. In addition, several variables such as ambient light, water temperature,

juvenile salmonid size, predator abundance and distribution, and habitat differences can create

variability within and among experimental trials. We were able to conduct five experimental

trials in 2017 and eight in 2018 when juvenile salmonids were relatively abundant, but perhaps a

much larger number of trials may be necessary to detect differences between light sources.

Experiments may also need to be conducted in a laboratory where many of the environmental

variables can be controlled.

The only significant difference among treatments was between no light treatments and

lighted treatments, thus demonstrating that juvenile salmonids were attracted to light regardless

of spectral composition. These results are consistent with previous studies of salmonids in the

Lake Washington system (Tabor et al. 2004a; Celedonia et al. 2011; Tabor et al. 2017) and

elsewhere (McDonald 1960; Nemeth and Anderson 1992; Stien et al. 2014) that have shown that

light intensity is an important factor influencing light attraction. Results of this study and other

studies indicate that the best management strategy is to reduce light intensity whenever possible,

either through reducing the output of lights or by aiming lights away from aquatic environments.

The major concern of ALAN attracting subyearling salmonids is the potential to increase

predation risk (Tabor et al. 2004a; Nightingale et al. 2006). Various piscivores including fishes

and birds are present in and around the nearshore area and may prey on the subyearling

19

salmonids concentrated near artificial lights. Of particular concern are great blue herons (Ardea

herodias) which are large, adaptable predators with a high energy demand (Pitt et al. 1995). It is

unclear if these large birds would feed on small fishes like juvenile Chinook salmon and sockeye

salmon; however, Stickley et al. (1995) found great blue herons preyed heavily on the small fish

Gambusia sp. during moonlit nights. Also, we have observed them feeding at night near lights

where juvenile Chinook salmon are far more abundant than other fish. Additionally, cutthroat

trout, yearling coho salmon, and sculpins were present in the nearshore area and may prey on

subyearling salmonids (Beauchamp et al. 1992; Tabor et al. 2004b). While these predatory

fishes may selectively use lit areas to take advantage of concentrations of juvenile salmon, they

may also avoid shallow, nearshore areas due to the higher risk of predation by wading birds such

as great blue herons (Power 1987). Further research is needed to understand how these complex

interactions between various predators may play out. Recent studies have found increasing

evidence that predator-prey interactions are altered by the addition of ALAN (Davies et al. 2012,

Manfrin et al. 2017), and that broader spectrum lights are more likely to affect a larger number of

species and therefore have potentially greater influence on food webs (Davies et al. 2013).

The ratio of juvenile sockeye salmon caught in the control units compared to lit units was

much lower than for juvenile Chinook salmon (0.6% versus 9.6%). Earlier experiments in Lake

Washington also found the same trend (Tabor et al. 2017). It is unclear if this is due to

differences in habitat use resulting in differential seine catch rates or is due to a stronger

attraction to lights by juvenile sockeye salmon. From January to April, juvenile Chinook salmon

typically inhabit shallow waters less than 1 m deep (Tabor et al. 2011) while the depth selection

of recently immigrating sockeye salmon fry is not well known. However, most sockeye salmon

fry appear to be offshore in deep waters and only a small percent are close to the shore

(Beauchamp et al. 2004). During recent snorkel surveys along a well-lit shoreline, juvenile

Chinook salmon appeared to be in shallower waters than juvenile sockeye salmon (R. Tabor,

personal observations). Chinook salmon may move along the shoreline to lit areas; whereas,

juvenile sockeye salmon may move inshore from deeper areas and be less vulnerable to our

beach seining. Alternatively, their sensitivity to different wavelengths may be quite different

between the two species. Juvenile sockeye salmon typically feed on zooplankton within the

water column while juvenile Chinook salmon appear to feed primarily on emerging chironomids

at the surface (Koehler et al. 2006; Tabor et al. 2011). Because their foraging strategies are

20

different, their spectral sensitivity may be different and thus they may respond differently to

ALAN.

Understanding the response of juvenile salmonids to various spectra of light at different

life stages is critically important. Our results are applicable to shallow, nearshore habitats. In

deeper water, light extinction coefficients play a more substantial role in determining fish

attraction or aversion to predation risk than in the shallow, nearshore environment of this study,

and red light is much more readily absorbed by water than blue light. As juvenile Chinook

salmon grow, they progressively move into deeper waters and by May or June they will move

into the pelagic zone of Lake Washington before migrating out to the ocean (Koehler et al. 2006;

Tabor et al. 2011). Even at a small size, juvenile sockeye salmon are present in the pelagic zone

(Beauchamp et al. 2004). For both species, moving into this open habitat puts them into

increased contact with piscivorous fish (including resident salmonids, northern pikeminnow

[Ptychocheilus oregonensis] and smallmouth bass), as well as lighting installations from a

number of bridges and the Ballard Locks. Earlier tracking results of Chinook salmon smolts

found they are strongly attracted to lights around bridges and other structures (Celedonia et al.

2011). It is unclear how different light spectra may influence the behavior of out-migrating

salmon and their interactions with predators. It is possible that lights with a lower proportion of

blue light may be more important in the pelagic environment, where there is a greater depth of

water to attenuate red wavelengths. Future research will need to address the potential of various

wavelengths of light to cause delays to migration and increased predation of juvenile salmonids

in these deeper water environments.

Acknowledgements

We wish to thank U.S. Fish and Wildlife Service (USFWS) employees and volunteers

Olivia Williams, Sarah Crestol, Nathan Beady, Carlisha Hall, Pat DeHaan, Jeff Johnson, Jennifer

Fields, Matthew Webster, and Jeffery Lee for all their assistance with the field work. An earlier

draft of this report was reviewed by Pat DeHaan (USFWS). We thank the City of Renton staff

including Cailin Hunsaker, Kelly Beymer, Leslie Betlach, Bryce Goldman, Dana Appel, and

Steve Brown for all the logistic support with this project. Funding for this project was made

21

possible by King County Flood Control Board and administered by Kim Harper. The findings

and conclusions in this report are those of the authors and do not necessarily represent the views

of USFWS.

References

Beauchamp, D. A., C.J. Sergeant, M. M. Mazur, J. M. Scheuerell, D. E. Schindler, M. D. Scheuerell, K. L. Fresh, D. E. Seiler, and T. P. Quinn. 2004. Temporal-spatial dynamics of early feeding demand and food supply of sockeye salmon fry in Lake Washington. Transactions of the American Fisheries Society 133:1014-1032.

Beauchamp, D. A., S. A. Vecht, and G. L. Thomas. 1992. Seasonal, diel, and size-related food habits of cutthroat trout in Lake Washington. Northwest Science 66:149-159.

Celedonia, M. T., R. A. Tabor, S. Sanders, S. Damm, D. W. Lantz, T. M. Lee, Z. Li, B. E. Price, W. Gale, and K. Ostrand. 2011. Movement and habitat use of Chinook salmon smolts, northern pikeminnow, and smallmouth bass near the SR 520 bridge, 2008 acoustic tracking study. U.S. Fish and Wildlife Service, Washington Fish and Wildlife Office, Lacey, Washington.

Davies, T. W., J. Bennie, and K. J. Gaston. 2012. Street lighting changes the composition of invertebrate communities. Biology Letters 8:764-767.

Davies, T. W., J. Bennie, R. Inger, N. H. de Ibarra and K. J. Gaston. 2013. Artificial light pollution: are shifting spectral signatures changing the balance of species interactions? Global Change Biology 19:1417–1423.

Gaston, K. J., T. W. Davies, J. Bennie, and J. Hopkins. 2012. Reducing the ecological consequences of night-time light pollution: options and developments. Journal of Applied Ecology 49:1256–1266.

Hawryshyn, C.W., S.D. Ramsden, K.M. Betke, and S. Sabbah. 2010. Spectral and polarization sensitivity of juvenile Atlantic salmon (Salmo salar): phylogenetic considerations. Journal of Experimental Biology 213:3187-3197.

IDA (International Dark-Sky Association). 2010. Visibility, environmental, and astronomical issues associated with blue-rich white outdoor lighting. Tucson, Arizona.

Koehler, M. E., K. L. Fresh, D. A. Beauchamp, J. R. Cordell, C. A. Simenstad, and D. E. Seiler.

22

2006. Diet and bioenergetics of lake-rearing juvenile Chinook salmon in Lake Washington. Transactions of the American Fisheries Society 135:1580-1591.

Longcore, T., and C. Rich. 2004. Ecological light pollution. Frontiers in Ecology and the Environment 2:191-198.

Manfrin, A., G. Singer, S. Larsen, N. Weiß, R. H. A. van Grunsven, N.-S. Weiß, S. Wohlfahrt, M. T. Monaghan, F. Hölker. 2017. Artificial light at night affects organism flux across ecosystem boundaries and drives community structure in the recipient ecosystem. Frontiers in Environmental Science 5:61.

Marchesan, M., M. Spoto, L. Verginella, and E.A. Ferrero. 2005. Behavioural effects of artificial light on fish species of commercial interest. Fisheries Research 73:171-185.

McDonald, J. 1960. The behaviour of Pacific salmon fry during their downstream migration to freshwater and saltwater nursery areas. Journal of the Fisheries Research Board of Canada 17:655-676.

Nemeth, R.S., and J. J. Anderson. 1992. Response of juvenile coho and Chinook salmon to strobe and mercury vapor lights. North American Journal of Fisheries Management 12:684-692.

Nightingale, B., T. Longcore, and C.A. Simenstad. 2006. Artificial night lighting and fishes. Page 257-276 in C. Rich and T. Longcore, editors. Ecological consequences of artificial night lighting. Island Press, Washington D.C.

Pawson, S. M., and M. K. Bader. 2014. LED lighting increases the ecological impact of light pollution irrespective of color temperature. Ecological Applications 24:1561-1568.

Perkin, E. K., F. Hölker, J. S. Richardson, J. P. Sadler, C. Wolter, and K. Tockner. 2011. The influence of artificial light on stream and riparian ecosystems: questions, challenges, and perspectives. Ecosphere 2(11):122.

Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core Team. 2018. nlme: Linear and Nonliner Mixed Effects Models. R package version 3.1-137.

Pitt, W., D. A. Beauchamp, and M. A. Conover. 1998. Development and application of a

bioenergetics model for estimating losses from predation by great blue herons at fish hatcheries. North American Journal of Fisheries Management 18:52-65.

Power, M. E. 1987. Predator avoidance by grazing fishes in temperate and tropical streams:

importance of stream depth and prey size. Pages 333–351 in W. C. Kerfoot and A. Sih, editors. Predation: direct and indirect impacts on aquatic communities. University Press of New England, Hanover, New Hampshire.

23

R Core Team. 2018. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Stickley, A. R., Jr., J. F. Glahn, J. O. King, and D. T. King. 1995. Impact of great blue heron

depredations on channel catfish farms. Journal of the World Aquaculture Society 26:194-199.

Stien, L. H., J. E. Fosseidengen, M. E. Malm, H. Sveier, T. Torgersen, D. W. Wright, and F. Oppedal. 2014. Low intensity light of different colours modifies Atlantic salmon depth use. Aquacultural Engineering 62:42-48.

Tabor, R. A., A. T. C. Bell, D. W. Lantz, C. N. Gregersen, H. B. Berge, and D. K. Hawkins. 2017. Phototaxic behavior of subyearling salmonids in the nearshore area of two urban lakes in western Washington State. Transactions of the American Fisheries Society 146:753-761.

Tabor, R. A., G. S. Brown, and V. T. Luiting. 2004a. The effect of light intensity on sockeye salmon fry migratory behavior and predation by cottids in the Cedar River, Washington. North American Journal of Fisheries Management 24:128-145.

Tabor, R. A., M. T. Celedonia, F. Mejia, R. M. Piaskowski, D.L. Low, B. Footen, and L. Park. 2004b. Predation of juvenile Chinook salmon by predatory fishes in three areas of the Lake Washington basin. U.S. Fish and Wildlife Service, Western Washington Fish and Wildlife Office, Lacey.

Tabor, R. A., K. L. Fresh, R. M. Piaskowski, H. A. Gearns, and D. B. Hayes. 2011. Habitat use of juvenile Chinook salmon in the nearshore areas of Lake Washington: effects of depth, shoreline development, substrate, and vegetation. North American Journal of Fisheries Management 31:100-713.

24

APPENDIX A.-- Number of fish caught in each experimental trial, February 7- March 16, 2017. Pr. Sc. = prickly sculpin; Co. Sc. = coastrange sculpin; Y-COH = yearling coho salmon; J-BGL = bluegill; and YLP = juvenile yellow perch.

Date Light treatment Unit # (N to S) Chinook Sockeye Cutthroat Pr. Sc. Co. Sc. Other fish7-Feb LED 2000 1 16 2 0 3 2

HPS 2 20 0 0 2 3Incandescent 3 11 1 0 3 0No light 4 15 0 0 4 3LED Yellow filter 5 14 0 0 0 5LED 5000 6 5 0 0 0 2

13-Feb LED 5000 1 12 12 0 2 1LED 2000 2 33 21 0 0 4Incandescent 3 20 30 0 0 2LED Yellow filter 4 19 83 0 4 0HPS 5 12 72 0 1 0No light 6 0 0 0 1 2 1-BGL, 1-YLPNo light-edge 7 2 2 0 3 2

14-Feb LED Yellow filter 1 50 30 0 3 1No light 2 13 2 0 2 4LED 2000 3 23 43 0 0 2Incandescent 4 33 49 0 1 1HPS 5 10 9 0 5 7LED 5000 6 14 7 0 6 1No light-edge 7 5 0 0 6 10

22-Feb No light 1 2 0 0 1 7LED 5000 2 14 25 0 5 3HPS 3 16 7 0 7 4LED Yellow filter 4 15 29 0 2 3 1 Y-COHIncandescent 5 23 18 0 8 2LED 2000 6 33 13 1 5 0No light-edge 7 7 0 0 5 1

16-Mar LED 2000 1 41 56 0 6 0No light 2 1 0 0 4 2HPS 3 19 6 0 3 0LED Yellow filter 4 28 119 0 1 0LED 5000 5 39 121 0 3 1Incandescent 6 20 12 0 2 3No light-edge 7 2 1 2 1 0

Number of Fish Caught

25

APPENDIX B.-- Number of fish caught in each experimental trial, March 30 – April 11, 2017. Pr. Sc. = prickly sculpin; Co. Sc. = coastrange sculpin; Y-COH = yearling coho salmon; and YLP = juvenile yellow perch.

Date Light treatment Unit # (N to S) Chinook Sockeye Cutthroat Pr. Sc. Co. Sc. Other fish30-Mar Incandescent 1 12 3 0 3 0

No light 2 0 0 0 6 6HPS 3 8 1 0 2 1LED 2000 4 3 1 0 2 2LED Yellow filter 5 2 2 0 1 0LED 5000 6 0 6 0 1 0No light-edge 7 0 0 1 1 0

3-Apr LED Red filter 1 2 2 0 2 2HPS 2 7 1 0 4 1LED Yellow filter 3 3 2 0 0 3Incandescent 4 2 1 0 1 2LED 2000 5 15 0 0 6 2No light 6 0 0 1 2 6 1-YLPNo light-edge 7 0 0 1 0 2

10-Apr LED 2000 1 11 0 0 0 2HPS 2 20 0 0 2 1LED Red filter 3 7 0 0 1 1No light 4 2 0 0 3 1LED Yellow filter 5 3 0 0 0 0LED 5000 6 2 0 0 3 1No light-edge 7 4 0 0 1 1

11-Apr LED Red filter 1 5 0 0 0 1LED 2000 2 10 0 0 0 0No light 3 5 0 0 1 4 1 Y-COHIncandescent 4 2 0 0 0 2LED Yellow filter 5 7 0 0 2 1LED 5000 6 1 0 0 1 1No light-edge 7 0 0 1 2 2

Number of Fish Caught

26

APPENDIX C.-- Number of fish caught in each experimental trial in 2018. Pr. Sc. = prickly sculpin; Co. Sc. = coastrange sculpin; Un. Sc. = unidentified juvenile sculpin; Y-COH = yearling coho salmon; J-SMB = juvenile smallmouth bass; J-SUN = juvenile sunfish; and J-YLP = juvenile yellow perch.

Date Light treatment Unit # (N to S) Chinook Sockeye Cutthroat Pr. Sc. Co. Sc. Un. Sc. Other fish7-Feb LED Red filter 1 105 6 0 2 1 2

No light 2 22 0 0 1 2 2LED Yellow filter 3 108 16 0 0 2 2No light 4 3 0 1 6 0 0LED 5000 5 51 2 0 0 3 1 1 J-SMB

12-Feb No light 1 7 0 1 5 1 1LED Red filter 2 83 8 0 0 1 0 1 J-SMBLED 5000 3 84 4 0 0 0 0No light 4 4 0 0 3 1 2LED Yellow filter 5 37 13 0 2 1 2No light-edge 6 5 0 0 3 1 3

13-Feb LED Yellow filter 1 71 9 0 5 3 1LED 5000 2 43 2 0 1 1 1 1 J-SMBNo light 3 13 0 0 1 3 2No light 4 6 0 0 0 3 3LED Red filter 5 31 1 0 0 3 0No light-edge 6 5 0 0 0 1 0

15-Feb No light 1 2 0 0 2 2 9 1 J-SMBLED Yellow filter 2 29 0 0 0 0 0LED Red filter 3 28 1 0 1 4 2 1 J-SUNLED 5000 4 46 11 0 3 2 0No light 5 0 0 0 3 2 2No light-edge 6 3 0 0 2 1 1

22-Feb No light 1 6 0 1 2 2 0No light 2 4 0 0 2 1 2 2 J-YLPLED Yellow filter 3 61 2 0 0 0 0LED Red filter 4 45 0 0 1 1 0LED 5000 5 42 6 0 0 0 0No light-edge 6 3 0 2 2 1 0

26-Feb LED 5000 1 43 2 0 1 0 0No light 2 3 0 0 0 0 0 1 J-YLPNo light 3 8 0 0 1 1 3LED Yellow filter 4 5 1 0 0 2 0 1 Y-COHLED Red filter 5 9 0 0 0 1 2

1-Mar LED Red filter 1 7 1 0 2 0 0LED 5000 2 59 6 0 0 1 0No light 3 4 0 0 1 0 2LED Yellow filter 4 54 7 0 0 0 1 1 J-YLPNo light 5 4 0 0 0 0 0

5-Mar No light 1 17 0 0 1 0 3LED 5000 2 23 4 0 0 1 1 1 J-SMBLED Red filter 3 12 1 0 0 2 0LED Yellow filter 4 15 0 0 0 0 2No light 5 2 0 0 0 1 3

Number of Fish Caught