Ecological Applications, 16(1), 2006, pp. 328–338
! 2006 by the Ecological Society of America
QUANTIFYING ULTRAVIOLET RADIATION MORTALITY RISK IN
BLUEGILL LARVAE: EFFECTS OF NEST LOCATION
MARK H. OLSON,1,3 MATTHEW R. COLIP,1 JUSTIN S. GERLACH,1
AND
DAVID L. MITCHELL2
1Department of Biology, Franklin & Marshall College, Lancaster, Pennsylvania 17604 USA
Department of Carcinogenesis, The University of Texas MD Anderson Cancer Center Science Park/Research Division,
P.O. Box 389, Park Road 1C, Smithville, Texas 78957 USA
2
Abstract. Ultraviolet (UV) radiation (280–400 nm) is an increasing threat to aquatic
organisms due to stratospheric ozone depletion and reductions in concentrations of dissolved
organic carbon. Because fish are most vulnerable to UV during the egg and larval stages,
parental spawning site selection can strongly influence mortality risk. We examined the
role of nest location in determining UV-induced mortality risk for bluegill ( Lepomis macrochirus) in Lake Giles, Pennsylvania, USA. In a series of five short-term incubation
experiments, we found that survival of yolk sac larvae across the range of depths at which
bluegill spawn was significantly lower in the presence of ambient-UV levels relative to
larvae that were shielded from UV radiation. In addition, survival decreased as a function
of cumulative UV exposure, as measured by the number of cyclobutane pyrimidine dimers
per megabase DNA in DNA dosimeters. Although UV had the potential to significantly
reduce larval survival, DNA dosimeters placed in bluegill nests concurrently with incubation
experiments indicated that most nests were exposed to relatively low levels of UV. Only
19% of nests had predicted UV-induced mortality greater than 25%. Consequently, current
levels of UV may be an important mortality source at the level of individual nests, but not
at the population level. One reason for the weak predicted effect of UV on bluegill survival
is that many nests were located at depths by which much of the incident UV had been
attenuated. In addition, many of the shallower nests were protected by overhanging trees
or other submerged structures. It is important to note that Lake Giles is highly transparent
and therefore not representative of all lakes in which bluegill are found. Nevertheless, Lake
Giles is a natural system and may be representative of north temperate lakes in the future.
Key words: bluegill; DNA dosimeter; fish larvae; Lepomis macrochirus; mortality; nest location;
quantile regression; ultraviolet radiation.
INTRODUCTION
Levels of ultraviolet (UV) radiation (280–400 nm)
are increasing in many aquatic ecosystems due to processes operating across a range of scales. On a global
scale, stratospheric ozone depletion has caused increases in UV reaching the earth’s surface, particularly at
mid and high latitudes (Madronich et al. 1998, McKenzie et al. 1999). This depletion is expected to continue for several decades because of synergistic effects
of ozone depletion and global climate change (Shindell
et al. 1998). On a local scale, UV exposure in aquatic
systems can increase in response to changes in water
chemistry, particularly concentrations of dissolved organic carbon (DOC). In lakes with high DOC concentrations, UV attenuates in the top few centimeters and
relatively little damaging radiation reaches organisms
living below the water surface (Morris et al. 1995).
However, changes in land use and reductions in runoff
associated with climate change can reduce DOC inputs
and thereby increase UV penetration (Schindler et al.
Manuscript received 18 February 2005; revised 23 May 2005;
accepted 3 June 2005. Corresponding Editor: T. E. Essington.
3
E-mail: [email protected]
1992, Molot et al. 2004). Current and future increases
in UV pose a threat to aquatic organisms because prolonged exposure damages DNA, which can result in
mutations, decreased physiological performance, or
death (Siebeck et al. 1994, Häder and Worrest 1997).
Fish are most susceptible to UV damage during the
egg and larval stages (Hunter et al. 1979, 1981). In
part, this vulnerability is a consequence of high surface-area-to-volume ratios that increase the proportion
of cells at or near the external surface. In addition, eggs
and larvae tend to have low concentrations of photoprotective compounds because the synthesis of pigments such as melanin is induced by exposure to light
(Häkkinen et al. 2002) and mycosporine-like amino
acids can only be acquired through consumption (Mason et al. 1998). Eggs and larvae also have limited
capabilities of behaviorally avoiding UV exposure due
to their reduced mobility (Williamson et al. 1997), and
at least some species cannot detect UV in early developmental stages (Britt et al. 2001).
Because eggs and larvae are especially vulnerable
to UV, parental selection of spawning sites can strongly
influence mortality risk. To reduce risk, fish may spawn
at sites that are shaded or at depths that receive little
328
February 2006
PLATE 1.
QUANTIFYING UV RISK IN BLUEGILL LARVAE
329
Male bluegill guarding a nest of eggs in Lake Giles, Pennsylvania. Photo credit: M. H. Olson.
UV. However, spawning site selection is influenced by
a variety of additional factors including spawning site
fidelity (Miller et al. 2001), habitat quality (Essington
et al. 1998), water temperature (Webster and Eiriksdottir 1976, Huff et al. 2004), and proximity to conspecifics (Gross and MacMillan 1981). Some of these
factors may provide an advantage to fish that spawn in
open areas or at shallow depths. In systems with high
UV transparency, these exposed nests might be at risk
of significant UV-induced mortality.
In this study, we evaluated the effect of UV radiation
on survival of bluegill sunfish (Lepomis macrochirus;
see Plate 1). Bluegill may be highly vulnerable to increases in UV for several reasons. First, bluegill spawn
in May through July when UV levels are at their highest. In addition, bluegill nest in shallow water (typically
around 1 m) and eggs adhere individually to structures
in the nest rather than settling into interstitial spaces
(Avila 1976, Scott and Crossman 1979). Finally, previous laboratory and field experiments have suggested
that ambient levels of UV can be a significant source
of mortality for bluegill (Guttierez-Rodriquez and Williamson 1999). Depending on temperature, eggs incubate for approximately 3–5 days (Scott and Crossman 1979) and larvae remain at the nest for an additional 3–7 days before dispersing (Coleman and Fischer
1991, Garvey et al. 2002). During that time, eggs and
larvae may receive high doses of UV, which could potentially influence survival and recruitment rates from
individual nests and for bluegill populations as a whole.
We conducted a series of short-term in situ incubation experiments to evaluate larval survival in the
presence and absence of ambient UV in a highly transparent lake in northeastern Pennsylvania, USA. In these
experiments, we also used assays of photodamage in
raw DNA to develop a relationship between UV exposure and larval bluegill survival. This relationship
was then combined with estimates of UV exposure in
bluegill nests to determine the risk of UV-induced mortality across a range of nest locations. Our objective
was to determine whether current levels of UV radiation are a potentially important source of mortality for
bluegill in this transparent ecosystem. Insights from
this study will help us to better understand the potential
impacts of future increases in UV exposure.
METHODS
Study site
We conducted incubation experiments and a nest exposure study in Lake Giles, located at 41!23" N, 75!06"
W in Pike County, Pennsylvania, USA. The lake has
Ecological Applications
Vol. 16, No. 1
MARK H. OLSON ET AL.
330
an elevation of 428 m, a surface area of 48 ha, and
mean and maximum depths of 10.1 and 24 m, respectively. Because Lake Giles has low DOC concentrations (mean of 1.1 mg/L), it is highly transparent. Secchi depths range from 10 to 15 m, and 1% of surface
UV radiation (320 nm) typically reaches a depth of 5
m. Bluegill are the dominant fish species in terms of
abundance and biomass. Other common species are yellow perch (Perca flavescens), pumpkinseed (L. gibbosus), largemouth bass (Micropterus salmoides), and
smallmouth bass (M. dolomieui).
Incubation experiments
We evaluated the effects of ambient-UV exposure on
survival of bluegill larvae in a series of five short-term
incubation experiments in June and July 2004. For each
experiment, we collected yolk sac larvae from a single
nest in Lake Giles the day before beginning an incubation. Larvae were transported in a cooler back to the
laboratory and stocked into UV-transparent plastic containers (whirl-pak bags; Nasco, Fort Atkinson, Wisconsin, USA) filled with 100 mL of filtered (60-#m
mesh) water from Lake Giles. Each container was
stocked with approximately 10 individual larvae. A
subset of containers also received a single dosimeter
for quantification of UV exposure (see below). Afterwards, containers were placed in a cooler filled with
water, which was then placed in a dark incubator and
held overnight at the same temperature as the epilimnion of Lake Giles.
The next morning, we transported experimental containers back to Lake Giles and placed them in pairs at
0.5 m depth intervals from 0.5 to 4.0 m along three
transects (the first round of incubations only went to a
depth of 3.0 m). For each pair of containers, one was
placed in a sleeve made of Aclar (Honeywell, Morris,
New Jersey, USA) and the other was placed in a sleeve
made of Courtgard (CPFilms, Martinsville, Virginia,
USA). Aclar is a clear long-wave-pass plastic that
transmits 100% of photosynthetically active radiation
(PAR) and 98% of UV radiation (295–399 nm). In contrast, Courtgard is a clear long-wave-pass plastic that
transmits no radiation $320 nm and only 9% of UVA
(320–399 nm), but 95% of PAR. To quantify UV exposure levels, we added dosimeters to all experimental
containers placed in either UV transparent sleeves or
the 0.5-m UV blocking sleeves. Each sleeve was attached to a pair of U-shaped wire frames that were
pushed into the sediment until stable. Containers were
intentionally suspended off the lake bottom to reduce
the risk of being covered by sediment. All containers
were deployed in Lake Giles between 10:00 and 13:30
hours on 3 June, 11 June, 17 June, 25 June, and 8 July.
Four days after each experiment began, all containers
were retrieved between 10:00 and 13:30 hours and
placed in a dark cooler filled with water from the epilimnion of Lake Giles. The cooler was then transported
back to the laboratory, and the contents of each con-
tainer was examined under a dissecting microscope to
determine the number of living and dead bluegill larvae. Death was defined as the lack of a visible heart
beat. The proportion of bluegill larvae surviving was
analyzed by analysis of covariance (ANCOVA) with
the presence/absence of UV radiation and experiment
number as categorical variables and depth of incubation
as a continuous variable.
Nest exposure study
Concurrently with incubation experiments, we also
quantified UV exposure in active bluegill nests. We
attached dosimeters to glass slides with two small rubber bands and placed single dosimeters into four separate nests in each of 10 different bluegill colonies (the
fifth round only had eight colonies). All dosimeters
were placed between 09:45 and 13:30 hours except for
the first experiment in which dosimeters were placed
between 14:00 and 16:30 hours. Bluegill colonies were
chosen to represent a range of spawning depths and
locations.
After four days, all dosimeters were retrieved using
snorkel or SCUBA and then immediately wrapped in
foil and placed in a dark cooler to prevent exposure to
direct sunlight. Just before collecting dosimeters, we
measured the depth of each nest to the nearest centimeter.
At the end of the bluegill spawning season, we also
conducted a lake-wide survey of bluegill colony locations. A pair of divers swam the perimeter of the lake
on 8 July, 14 July, and 15 July in search of bluegill
colonies. When a colony was located, each diver made
an independent estimate of the number of nests. Estimates within 10% of one another were averaged. If
estimates were not within 10% of one another, both
divers recounted nests. In every instance, recounts were
within 10% of one another. In addition to counting
nests, we also estimated colony depth from the depths
at the endpoints of the colony’s longest linear dimension and at the endpoints of the longest orthogonal.
DNA dosimeters
Dosimeters consisted of quartz cuvettes (10 mm diameter and 40 mm long) filled with 0.4 mL of DNA
solution (derived from salmon testes and diluted to 100
#g/mL in double distilled water). Each end of the cuvette was sealed with parafilm and a rubber stopper.
When exposed to UV radiation (particularly 280–320
nm), the DNA in cuvettes accumulated molecular damage in the form of various photoproducts. The frequency of the most common photoproduct, a cyclobutane pyrimidine dimer (CPD), was quantified using
radioimmunoassay (RIA). The RIA is a competitive
binding assay between very small amounts of radiolabeled DNA and sample DNA for antisera raised
against UV-irradiated DNA. For the RIA, 50 ng of heatdenatured sample DNA was incubated with 5–10 pg of
poly(dA):poly(dT) (labeled to %5 & 108 cpm/#g by
QUANTIFYING UV RISK IN BLUEGILL LARVAE
February 2006
TABLE 1.
331
Physical conditions for the five rounds of incubation experiments.
Temperature (!C)
Date
3–7 June
11–15 June
17–21 June
25–29 June
8–12 July
Starting
Ending
20.0
20.3
22.3
21.6
23.8
19.6
21.0
20.8
22.0
22.9
Cumulative surface
exposure (W/m2)
14 234.1
23 574.6
23 563.0
20 595.9
24 954.9
'
'
'
'
'
66.3
81.1
141.0
22.2
142.7
1% attenuation
depth (m)
Date of
UV depth
profile
3.30
3.28
3.38
3.38
3.41
8 June
15 June
21 June
21 June
21 July
Notes: Starting and ending temperatures were taken at the surface of Lake Giles, Pike County,
Pennsylvania, USA. Cumulative surface exposure for 305 nm (mean ' SE) was calculated from
a ground UV profiling radiometer located at the Lacawac Sanctuary, 20 km west of Lake Giles.
The 1% attenuation depths were estimated from regressions of ln(percentage of surface irradiance for 305 nm) as a function of depth. All regressions had r2 % 0.98.
nick translation with 32P-deoxythymidine triphosphate
[dTTP]) in a total volume of 1 mL 10 mmol/L Tris,
pH 7.8, 150 mmol/L NaCl, 1 mmol/L ethylenediaminetetraacetic acid (EDTA), and 0.15% gelatin (Sigma, St.
Louis, Missouri, USA). Antiserum was added at a dilution that yielded (35% binding to labeled ligand,
and after incubation overnight at 4!C the immune complex was precipitated with goat anti-rabbit immunoglobulin (Calbiochem, San Diego, California, USA)
and carrier serum from non-immunized rabbits (UTMDACC, Science Park/Veterinary Division, Bastrop,
Texas, USA). After centrifugation, the pellet was dissolved in tissue solubilizer (NCS, Amersham, Piscataway, New Jersey, USA), mixed with ScintiSafe (Fisher,
Pittsburgh, Pennsylvania, USA) containing 0.1% glacial acetic acid, and the 32P quantified by liquid scintillation spectrometry. Under these conditions, antibody binding to an unlabeled competitor inhibits antibody binding to the radiolabeled ligand. These details,
as well as those concerning the specificities of the RIAs
and standards used for quantification, are described in
Mitchell (1996, 1999).
To evaluate the potential role of shading in reducing
mortality risk, we compared estimates of UV-induced
mortality based on dosimeters placed in bluegill nests
with UV-induced mortality predicted for the same
depth in the absence of any shading (henceforth full
exposure). We calculated the expected mortality under
full exposure by combining incident measurements of
UV irradiance with estimates of UV attenuation in Lake
TABLE 2. Analysis of covariance of larval bluegill survival
with round of experiment and UV treatment (shielded vs.
ambient) as categorical variables and depth as a covariate.
Source
Round of experiment
UV treatment
Depth
Round & UV
Depth & Round
Depth & UV
Depth & UV & Round
df
F
P
4
1
1
4
4
1
4
8.86
20.38
24.00
0.57
1.05
0.01
1.56
$0.0001
$0.0001
$0.0001
0.66
0.38
0.92
0.19
Notes: Significant sources of variation are in boldface type.
All tests had 199 degrees of freedom in the denominator.
Giles. Incident measurements of UV irradiance at 305
nm were measured continuously with a ground-based
UV radiometer located at the Lacawac Sanctuary, located 20 km west of Lake Giles. Measurements integrated over 15-minute intervals were summed for the
entire time a given dosimeter was deployed in a nest.
Depth profiles of UV attenuation were estimated for
the 305-nm bandwidth using a profiling UV radiometer.
Profiles were usually measured at some point close to
or during a four-day exposure (Table 1). Together, the
attenuation coefficient and cumulative incident-UV irradiance provided an estimate of irradiance levels as a
function of depth, which was converted to a predicted
dosimeter value using the equation y ) 0.0239(x) *
10.24. The slope of this regression is based on data
from Wilhelm et al. (2002), and the intercept represents
the mean frequency of CPD’s/megabase [Mb] DNA for
dosimeters placed in experimental containers placed in
UV blocking sleeves. Finally, the predicted dosimeter
value under full exposure and the observed dosimeter
value for a bluegill nest at the same depth was converted to predicted UV-induced mortality using data
from the incubation experiments (see Results).
RESULTS
Incubation experiments
Physical conditions in Lake Giles, including cumulative surface UV exposure and attenuation depth,
varied among the five rounds of experiments (Table 1).
When all five rounds were analyzed together, analysis
of covariance (ANCOVA) on larval bluegill survival
indicated significant main effects of round, UV treatment, and depth, but no significant interaction terms
(Table 2). After removing the effect of round with a
one-way ANOVA, residuals of survival were consistently higher in the UV-shielded treatment relative to
ambient UV (Fig. 1). In addition, residuals increased
slightly as a function of depth in both treatments
(Fig. 1).
Within the ambient-UV treatment, larval bluegill
survival decreased as a function of increasing UV exposure as measured by the number of CPD’s/Mb DNA
in dosimeters (Fig. 2). After dropping the nonsignifi-
332
MARK H. OLSON ET AL.
Ecological Applications
Vol. 16, No. 1
as a function of a single predictor variable (Cade and
Noon 2003). For the relationship between larval bluegill survival and dosimeter value, the 0.70, 0.80, and
0.90 deciles all declined significantly with increasing
number of CPD’s/Mb DNA (Fig. 3, Table 3). In contrast, all smaller deciles had nonsignificant regressions
(Table 3). Therefore, the upper side of the distribution
of larval bluegill survival rates declined with increasing dosimeter value, whereas the lower side of the distribution did not change.
Nest exposure study
FIG. 1. Residuals (means ' SE) of the proportion of larval
bluegill surviving at the end of four-day incubation experiments as a function of incubation depth in ambient-UV and
UV-shielded treatments in Lake Giles in Pike County, Pennsylvania, USA. Residuals were calculated from a one-way
ANOVA with round of experiment as a categorical variable.
Symbols denote mean survival over the five rounds of experiments.
cant interaction term from the full ANCOVA model
with round of experiment as a categorical variable and
dosimeter value as a continuous variable, the main effects of round and dosimeter value were both highly
significant (overall model, F5,87 ) 15.83, P $ 0.0001;
main effect of round, F5,87 ) 17.59, P $ 0.0001; main
effect of dosimeter value, F1,87 ) 8.81, P $ 0.005).
Survival was highly variable both within and among
rounds, and this variability appeared to decrease as a
function of dosimeter value (Fig. 2). This pattern can
be examined using quantile regression, which quantifies changes in the distribution of a response variable
Ultraviolet exposure levels in bluegill nests, as measured by dosimeters, varied widely among rounds and
as a function of depth (Fig. 3). An ANCOVA of
ln(dosimeter value) with round of experiment as a categorical variable and ln(depth) as a continuous variable
was significant and included significant main effects
but no interaction term (overall model, F9, 169 ) 23.85,
P $ 0.0001; round, F4, 169 ) 8.38, P $ 0.0001; depth,
F1, 169 ) 114.54, P $ 0.0001; interaction, F4, 169 ) 1.32,
P % 0.25). In addition to individual exposure levels,
variability in exposure levels among nests also decreased as a function of depth (Fig. 3). In nests at depths
of 1–2 m, the coefficient of variation (CV) in dosimeter
values was 76.0. In contrast, the CV for nests 4–5 m
deep was only 32.7.
We used the 0.90 decile regression equation of survival vs. dosimeter value from the incubation experiments (Table 3) to predict potential UV-induced mortality for bluegill larvae located in our exposure study
nests. We used this equation instead of the ordinary
least-squares regression equation to provide a more accurate estimate of the effect of UV on survival, par-
FIG. 2. Proportion of larval bluegill surviving at the end of four-day incubation experiments as a function of UV exposure
(expressed as a dosimeter value, in cyclobutane pyrimidine dimers (CPD’s) per megabase DNA). Each point corresponds to
an experimental unit from the ambient-UV treatment. Symbols denote different rounds of incubation experiments. The three
lines represent significant relationships for the 0.70 (dotted line), 0.80 (dashed line), and 0.90 (solid line) deciles. See Results:
Incubation experiments for details.
QUANTIFYING UV RISK IN BLUEGILL LARVAE
February 2006
333
FIG. 3. Ultraviolet exposure (expressed as a dosimeter value, in cyclobutane pyrimidine dimers (CPD’s) per megabase
DNA) as a function of nest depth. Each point corresponds to an individual bluegill nest. Symbols denote different rounds
of exposure. For all rounds combined, the best-fit exponential curve is y ) 83.2e+0.471(x), n ) 170, r2 ) 0.66.
ticularly for low UV doses (i.e., close to the y-intercept)
where survival is expected to be high. In general, UV
was predicted to be a relatively small source of mortality in most bluegill nests (Fig. 4). Approximately
55% of nests were expected to have mortality of 0.15
or lower. Because the 0.90 decile regression predicts
mortality of 0.08 with no UV exposure, the actual
change in mortality due to UV exposure was minimal.
Nevertheless, UV exposure was a potentially important
source of mortality in some nests (Fig. 4). The highest
predicted mortality was 0.49, and approximately 19%
had predicted mortality of at least 0.25. All nests located !3 m deep had predicted mortality $0.20. In
contrast, shallower nests exhibited a wide range of potential UV-induced mortality. For example, nests lo-
TABLE 3. Results of quantile regression of proportion surviving as a function of dosimeter value for 0.10–0.90 deciles.
Decile
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
Intercept
Slope
Asymptotic
rank score
statistic
0.00
0.1022
0.1278
0.2031
0.3016
0.4365
0.6630
0.8199
0.9223
0.00
+0.00057
+0.00024
+0.00045
+0.00085
+0.00121
+0.00204
+0.00273
+0.00305
0.0004
0.81
0.17
2.05
2.29
1.87
4.84
4.19
3.74
P
0.99
0.37
0.68
0.15
0.13
0.17
0.03
0.04
0.05
Notes: The asymptotic rank score statistic is distributed as
a chi-square statistic and has one degree of freedom (i.e., the
difference in number of parameters between the full and reduced models). Significant relationships appear in boldface
type.
cated in 1–2 m depth had predicted mortality that
ranged from 0.08 to 0.49.
Predicted UV-induced mortality estimated from dosimeters in bluegill nests differed from UV-induced
mortality estimated for full-exposure conditions. These
differences were strongest in shallow nests and decreased as a function of depth (Fig. 5). An ANCOVA
of the ln-transformed difference between predicted
mortality in a bluegill nest and predicted mortality at
the same depth under full exposure with round of experiment as a categorical variable and depth as a continuous variable was significant and included significant main effects but no interaction term (overall model, F9, 169 ) 17.15, P $ 0.0001; round, F4, 169 ) 5.96,
P $ 0.0005; depth, F1, 169 ) 99.29, P $ 0.0001; interaction, F4, 169 ) 1.45, P % 0.10). Under full exposure,
nests $1.0 m deep were predicted to have mean mortality of 0.55. In contrast, mean predicted mortality was
only 0.19 based on actual exposure levels. With increasing depth, the difference in predicted mortality
approached 0, indicating these nests were receiving
close to full exposure conditions. Although differences
in predicted mortality in nests compared to full exposure were strongest for shallow nests, there were still
many nests in this region for which the difference was
close to 0. In some cases, the difference was even $0,
indicating that predicted mortality under actual exposure were higher than under full exposure. Most of
these cases were observed in the experiments that ran
from 17 to 21 June or from 25 to 29 June.
The depths where highest predicted mortality (0–2
m) were observed were also the depths where the majority of bluegill nests were located in Lake Giles (Fig.
6). Approximately 56% of all bluegill colonies had
334
Ecological Applications
Vol. 16, No. 1
MARK H. OLSON ET AL.
FIG. 4. Frequency histogram of predicted UV-induced mortality in bluegill nests. Predicted mortality was calculated for
all five rounds of exposure using dosimeter values from Fig. 3 and the 0.90-decile regression equation from Fig. 2.
mean depths "2.0 m. Because the number of nests in
a colony was not correlated with colony depth (r )
0.16, n ) 46, P % 0.25), the depth distribution of
colonies is also representative of the depth distribution
of individual nests. Although the majority of bluegill
colonies were located in shallow water, approximately
30% were located at depths %4.0 m. The maximum
colony depth we observed was 5.8 m.
DISCUSSION
Current levels of ultraviolet radiation can potentially
be an important mortality source for bluegill in Lake
FIG. 5. Reduction in predicted UV-induced mortality in bluegill nests relative to predicted UV-induced mortality under
full exposure as a function of nest depth. Each point corresponds to an individual bluegill nest. Predicted mortality under
full exposure was calculated using cumulative surface irradiance at 305 nm (measured 20 km west of Lake Giles) and
attenuation coefficients for 305 nm to estimate exposure at a specific nest. This value was then converted to a predicted
dosimeter value using the equation y ) 0.0239(x) * 10.24. For all rounds combined, the best-fit exponential curve is y )
4.46e+2.875(x), n ) 170, r2 ) 0.74.
February 2006
QUANTIFYING UV RISK IN BLUEGILL LARVAE
FIG. 6. Frequency distribution of bluegill colonies as a
function of depth in Lake Giles. Colony depth was estimated
as the mean of the endpoints along the greatest linear dimension and the endpoints of the longest orthogonal.
Giles. Larvae incubated across a range of depths at
which bluegill nests had significantly lower mortality
when shielded from UV than larvae exposed to ambient-UV levels. Despite this sensitivity, our analyses
also suggest that most bluegill nests had relatively low
UV-induced mortality. Only 19% of nests were predicted to have UV-induced mortality of 25% or higher.
Therefore, UV could be a significant source of mortality at the level of individual nests even though it
does not appear to be important in the majority of nests.
The low risk of UV-induced mortality in most nests
was a direct consequence of nest location. Many of the
shallow nests were shaded from harmful levels of UV
radiation for at least part of the day because they were
located under or near overhanging trees or other submerged structures. In addition, nests in deeper colonies
received little UV radiation due to the attenuation of
the shortest and most damaging wavelengths. Palen et
al. (2005) also found that oviposition site played an
important role in reducing UV exposure in two species
of salamanders (Ambystoma spp.). When UV transparency was high, A. gracile reduced exposure by increasing egg depth whereas A. macrodactylum reduced exposure by laying eggs in shaded microhabitats.
Nest locations in Lake Giles are consistent with the
hypothesis that bluegill choose nesting sites to minimize UV exposure, as Huff et al. (2004) concluded for
yellow perch spawning in Lake Giles. Indeed, the typical spawning depth of bluegill is approximately 1 m
(Scott and Crossman 1979), and Gutiérrez-Rodrı́quez
and Williamson (1999) found that no bluegill nests in
nearby low-UV Lake Lacawac were deeper than 2.2 m.
However, the present study was not designed to evaluate factors influencing nest site selection in bluegill.
Alternative explanations such as substrate suitability
have not been examined. In addition, bluegill lack UV
photoreceptors (Hawryshyn et al. 1988), although they
may use visible light as a surrogate. Therefore, we
cannot conclude that bluegill choose to spawn in shaded areas or deep water to avoid UV exposure. Nest site
335
selection in bluegill is influenced by many factors, particularly the location of conspecifics (Neff et al. 2004).
Nevertheless, the location of nests played a key role
in limiting the population-level effects of UV in bluegill.
The presence of overhanging trees and other structures along the shore of Lake Giles played a key role
in reducing UV exposure levels in shallow water, where
the majority of bluegill nested. This finding highlights
another important role of riparian vegetation in the
structure and function of aquatic ecosystems (Naiman
and Décamps 1997, Pace et al. 2004). A common trend
in north temperate lakes is for trees to be removed as
shorelines become more developed (Christensen et al.
1996, Jennings et al. 2003, Scheurell and Schindler
2004). In transparent lakes, this removal could greatly
increase the risk of UV-induced mortality in bluegill,
particularly if suitable spawning habitat is not available
in deeper water as it is in Lake Giles. To help sustain
viable bluegill populations, lake managers may wish
to protect stretches of shoreline near bluegill colonies
to provide a refuge from UV exposure.
We predicted mortality due to UV exposure based
on incubation experiments that ran for four days. Had
the experiment run longer, UV-induced mortality would
probably have been higher, especially in experimental
containers that already had high dosimeter values. With
a longer incubation, experimental larvae would have
been more likely to experience at least one cloudless
day of high UV exposure. In addition, effects of UV
exposure can take up to three days to be expressed in
bluegill larvae (M. H. Olson, unpublished data). Therefore, larvae that received a lethal dose of UV on the
second or third day of an incubation could still have
been scored as alive when the experiment ended. Although longer incubations would likely have produced
stronger UV effects, they would have been less realistic
given that larvae only stay at the nest for 3–7 days
after hatching (Coleman and Fischer 1991, Garvey et
al. 2002), and we did not know the precise hatching
dates of our experimental animals. Furthermore, a longer incubation could potentially have problems with oxygen depletion in our experimental containers, which
did not occur in our four-day incubations (M. H. Olson,
unpublished data). Due to logistic difficulties associated with collecting eggs, we were also limited to conducting incubation experiments with larvae. For many
fishes, UV tolerance increases with age and/or developmental stage (Hunter et al. 1982, Kouwenberg et al.
1999, but see Browman et al. 2003). Consequently,
those nests that already had high predicted mortality
for larvae could have even higher total mortality if eggs
are also considered.
We used the relationship between survival and dosimeter value from the incubation experiments to predict UV-induced mortality in the nest exposure study.
Although this relationship was significant, it was highly
variable for low dosimeter values (Fig. 2). This type
336
MARK H. OLSON ET AL.
of pattern is common in ecological data when the predictor variable is only one of several factors that influence a response variable (Garvey et al. 1998, Cade
and Noon 2003). At high values, the predictor variable
exerts a strong influence on the response variable regardless of the magnitude of other factors. At low values, however, the predictor variable exerts a weak influence, and the response variable is influenced more
strongly by other factors (e.g., handling stress, temperature change) that may be independent of the predictor variable. Quantile regression is a useful technique for dealing with this kind of relationship because
it better represents the increasing influence of the predictor variable on the response variable (Cade and
Noon 2003). By using the 0.90 decile, we could estimate the effect of UV under a ‘‘best case scenario’’
when other factors have a minimal effect on larval
survival.
Some of the scatter in Fig. 2 may also reflect biological responses to variation in UV exposure that is
not revealed in dosimeters. Photodamage in dosimeters
accrues as a function of total UV dose and is independent of dose rate (Wilhelm et al. 2002). Because bluegill utilize both nucleotide excision repair and photoenzymatic repair to reverse DNA photodamage (M.
H. Olson and D. L. Mitchell, unpublished manuscript),
reciprocity does not hold and the consequences of UV
exposure depend on UV dose rate rather than just total
dose (Grad et al. 2001). At low dose rates, bluegill can
repair DNA damage as it occurs. However, a high dose
rate can cause damage that exceeds repair rates and
results in death. Therefore, variation in dose rates could
cause larval bluegill survival to differ among experimental units that had similar dosimeter values.
We also observed wide variation in larval bluegill
survival among rounds of our incubation experiments.
Because no interaction terms were significant, these
differences were consistent across all depths and in the
presence or absence of ambient UV. Therefore, differences in UV exposure, dose rate, attenuation depth, or
temperature fluctuation (which would be most pronounced at shallow depths and therefore could explain
the significant main effect of depth) are not likely explanations. Instead, this result may be caused by variation in the vigor of experimental larvae due to differences in factors such as age, condition, or genotype.
These differences would be observed among rounds of
incubations, but would be relatively small within
rounds because all experimental animals in a given
round were collected from a single nest.
In our analysis of the reduction in predicted UVinduced mortality, we found that some nests measured
on 17–21 June and 25–29 June actually had higher
mortality predicted from dosimeters in the nests than
expected with full exposure (Fig. 5). This discrepancy
may be due to the fact that mortality under full exposure
was derived from incident-UV levels measured 20 km
away from Lake Giles. Alternatively, errors in either
Ecological Applications
Vol. 16, No. 1
the intercept or the slope of the regression used to
convert UV dose to dosimeter values could have reduced our predicted mortality for full exposure. If we
did underestimate predicted mortality for full exposure,
we would expect even larger reductions in predicted
mortality, particularly for shallower nests.
The findings reported in this study are specific to
Lake Giles, Pennsylvania. Lake Giles is one of the more
transparent lakes in North America with a 1% UV305
attenuation depth on the order of 3–4 m in June. Lake
Giles is therefore atypical of lakes in which bluegill
live. For comparison, nearby Lake Lacawac has DOC
concentrations of 4–5 mg/L and a 1% UV305 attenuation
depth of only 0.53 m (Huff et al. 2004). Because much
of the incident UV will be absorbed by DOC in the top
few centimeters, UV is not likely to be a significant
source of mortality in any bluegill nests in Lake Lacawac or other moderate to high DOC systems. Lesser
et al. (2001) reached a similar conclusion regarding
UV-induced mortality risk to spotted salamanders (Ambystoma maculatum). Nevertheless, Lake Giles is a natural system and may be representative of north temperate lakes in the future. Continued depletion of
stratospheric ozone combined with global climate
change may lead to increases in UV exposure over large
parts of the earth’s surface (Madronich et al. 1998,
Shindell et al. 1998). At the scale of individual lakes,
UV penetration can increase in response to droughts
(Yan et al. 1996), acidification (Schindler et al. 1996,
Williamson et al. 1996), or reductions in DOC inputs
(Schindler et al. 1992, Williamson et al. 1996). Each
of these factors could increase in frequency or severity
over time due to global climate change, accumulation
of nitric acid in our environment, and land use changes,
respectively. In addition, invasive Dreissenid mussels
continue to spread in inland lakes throughout North
America (Kraft et al. 2002). Once established, they can
reduce algal concentrations and consequently increase
light penetration (Caraco et al. 1997, Idrisi et al. 2001).
If some or all of these changes were to occur within a
lake, UV exposure may increase to levels observed in
Lake Giles.
Because of their spawning behavior, spawning time,
and low UV tolerance, bluegill may be among the most
vulnerable fishes to increasing UV exposure. However,
several studies have found that current ambient-UV
levels can have strong effects on early life stages of
other fish species, including yellow perch (Huff et al.
2004), pike (Esox lucius, Häkkinen and Oikari 2004;
Galaxias maculates, Battini et al. 2000), and sockeye
salmon (Oncorhynchus nerka, Flamarique and Harrower 1999). Through its effect on egg and larval survival,
UV could influence recruitment and ultimately population dynamics of these recreationally or commercially
important species. In addition, UV could also potentially influence food webs from the top down through
its effect on fish populations.
February 2006
QUANTIFYING UV RISK IN BLUEGILL LARVAE
ACKNOWLEDGMENTS
We thank Bruce Hargreaves for incident UV measurements,
Craig Williamson for UV attenuation profiles in Lake Giles,
Jackie Nicolai for field assistance, and Sandi Connelly for
the steady supply of dosimeters. Access to Lake Giles was
provided by the Blooming Grove Hunting and Fishing Club.
The comments and suggestions of Janet Fischer and Craig
Williamson improved this manuscript. This research was supported by NSF grant DEC-IRCEB-0210972 to M. H. Olson
and D. L. Mitchell and Hackman Summer Scholar Fellowships to M. R. Colip and J. S. Gerlach.
LITERATURE CITED
Avila, V. L. 1976. Field-study of nesting-behavior of male
bluegill sunfish (Lepomis macrochirus Rafinesque). American Midland Naturalist 96:195–206.
Battini, M., V. Rocco, M. Lozada, B. Tartarotti, and H. E.
Zagarese. 2000. Effects of ultraviolet radiation on the eggs
of landlocked Galaxias maculates (Galaxiidae, Pisces) in
northwestern Patagonia. Freshwater Biology 44:547–552.
Britt, L. L., E. R. Loew, and W. N. McFarland. 2001. Visual
pigments in the early life stages of Pacific northwest marine
fishes. Journal of Experimental Biology 204:2581–2587.
Browman, H. I., R. D. Vetter, C. A. Rodriquez, J. J. Cullen,
R. F. Davis, E. Lynn, and J.-F. St. Pierre. 2003. Ultraviolet
(280–400 nm)-induced DNA damage in the eggs and larvae
of Calanus finmarchicus G. (Copepoda) and Atlantic cod
(Gadus morhua). Photochemistry and Photobiology 77:
397–404.
Cade, B. S., and B. R. Noon. 2003. A gentle introduction to
quantile regression for ecologists. Frontiers in Ecology and
the Environment 8:412–420.
Caraco, N. F., J. J. Cole, P. A. Raymond, D. L. Strayer, M.
L. Pace, S. E. G. Findlay, and D. T. Fischer. 1997. Zebra
mussel invasion in a large, turbid river: phytoplankton response to increased grazing. Ecology 78:588–602.
Christensen, D. L., B. R. Herwig, D. E. Schindler, and S. R.
Carpenter. 1996. Impacts of lakeshore residential development on coarse woody debris in north temperate lakes.
Ecological Applications 6:1143–1149.
Coleman, R. M., and R. U. Fischer. 1991. Brood size, male
fanning effort and the energetics of a non-shareable parental investment in bluegill sunfish, Lepomis macrochirus
(Teleostei, Centrarchidae). Ethology 87:177–188.
Essington, T. E., P. W. Sorensen, and D. G. Paron. 1998. High
rate of redd superimposition by brook trout (Salvelinus fontinalis) and brown trout (Salmo trutta) in a Minnesota
stream cannot be explained by habitat availability alone.
Canadian Journal of Fisheries and Aquatic Sciences 55:
2310–2316.
Flamarique, I. N., and W. L. Harrower. 1999. Mortality of
sockeye salmon raised under light backgrounds of different
spectral composition. Environmental Biology of Fishes 55:
279–293.
Garvey, J. E., T. P. Herra, and W. C. Leggett. 2002. Protracted
reproduction in sunfish: the temporal dimension in fish recruitment revisited. Ecological Applications 12:194–205.
Garvey, J. E., E. A. Marschall, and R. A. Wright. 1998. From
star charts to stoneflies: detecting relationships in continuous bivariate data. Ecology 79:442–447.
Grad, G., C. E. Williamson, and D. M. Karapelou. 2001.
Zooplankton survival and reproduction responses to damaging UV radiation: a test of reciprocity and photoenzymatic repair. Limnology and Oceanography 46:584–591.
Gross, M. R., and A. M. MacMillan. 1981. Predation and
the evolution of colonial nesting in bluegill sunfish (Lepomis macrochirus). Behavioral Ecology and Sociobiology
8:163–174.
Gutiérrez-Rodrı́quez, C., and C. E. Williamson. 1999. Influence of solar ultraviolet radiation on early life-history stag-
337
es of the bluegill sunfish, Lepomis macrochirus. Environmental Biology of Fishes 55:307–319.
Häder, D.-P., and R. C. Worrest. 1997. Consequences of the
effects of increased solar ultraviolet radiation on aquatic
ecosystems. Pages 11–30 in D.-P. Häder, editor. The effects
of ozone depletion on aquatic ecosystems. Academic Press,
San Diego, California, USA.
Häkkinen, J., and A. Oikari. 2004. A field methodology to
study effects of UV radiation on fish larvae. Water Research
38:2891–2897.
Häkkinen, J., E. Vehniäinen, O. Ylönen, J. Heikkilä, M. Soimasua, J. Karuole, A. Oikari, and J. Karjalainen. 2002.
The effects of increasing UV-B radiation on pigmentation,
growth and survival of coregonid embryos and larvae. Environmental Biology of Fishes 64:451–459.
Hawryshyn, C. W., M. G. Arnold, W. N. McFarland, and E.
R. Loew. 1988. Aspects of color vision in bluegill sunfish
(Lepomis macrochirus): ecological and evolutionary relevance. Journal of Comparative Physiology A: Sensory,
Neural, and Behavioral Physiology 164:107–116.
Huff, D. D., G. G. Grad, and C. E. Williamson. 2004. Environmental constraints on spawning depth of yellow perch:
the roles of low temperature and high solar ultraviolet radiation. Transactions of the American Fisheries Society
133:718–726.
Hunter, J. R., S. E. Kaupp, and J. H. Taylor. 1981. Assessment
of the effects of UV radiation on marine fish larvae. Pages
549–597 in J. Caulkins, editor. The role of solar ultraviolet
radiation in marine ecosystems. Plenum, New York, New
York, USA.
Hunter, J. R., S. E. Kaupp, and J. H. Taylor. 1982. Effects
of solar and artificial ultraviolet-b radiation on larval northern anchovy, Engraulis mordax. Photochemistry and Photobiology 34:477–486.
Hunter, J. R., J. H. Taylor, and H. G. Moser. 1979. Effect of
ultraviolet irradiation on eggs and larvae of the northern
anchovy, Engraulis mordax, and the pacific mackerel,
Scomber japonicus, during the embryonic stage. Photochemistry and Photobiology 29:325–338.
Idrisi, N., E. L. Mills, L. G. Rudstam, and D. J. Stewart.
2001. Impact of zebra mussels (Dreissena polymorpha) on
the pelagic lower trophic levels of Oneida Lake, New York.
Canadian Journal of Fisheries and Aquatic Sciences 58:
1430–1441.
Jennings, M. J., E. E. Emmons, G. R. Hatzenbeler, C. Edwards, and M. A. Bozek. 2003. Is littoral habitat affected
by residential development and land use in watersheds of
Wisconsin lakes? Lake and Reservoir Management 19:
272–279.
Kouwenberg, J. H. M., H. I. Browman, J. J. Cullen, R. F.
Davis, J.-F. St-Pierre, and J. A. Runge. 1999. Biological
weighting of ultraviolet (280–400 nm) induced mortality
in marine zooplankton and fish. I. Atlantic cod (Gadus
morhua) eggs. Marine Biology 134:269–284.
Kraft, C. E., P. J. Sullivan, A. Y. Karatayev, Y. E. Burlakova,
J. C. Nekola, L. E. Johnson, and D. K. Padilla. 2002. Landscape patterns of an aquatic invader: assessing dispersal
extent from spatial distributions. Ecological Applications
12:749–759.
Lesser, M. P., S. L. Turtle, J. H. Farrell, and C. W. Walker.
2001. Exposure to ultraviolet radiation (290–400 nm)
causes oxidative stress, DNA damage, and expression of
p53/p73 in laboratory experiments on embryos of the spotted salamander, Ambystoma maculatum. Physiological and
Biochemical Zoology 74:733–741.
Madronich, S., R. L. McKenzie, L. O. Bjorn, and M. M.
Caldwell. 1998. Changes in biologically active ultraviolet
radiation reaching the Earth’s surface. Journal of Photochemistry and Photobiology B: Biology 46:5–19.
338
MARK H. OLSON ET AL.
Mason, D. S., F. Schafer, J. M. Schick, and W. C. Dunlap.
1998. Ultraviolet radiation-absorbing mycosporine-like
amino acids (MAAs) are acquired from their diet by medaka fish (Oryzias latipes) but not by SKH-1 hairless mice.
Comparative Biochemistry and Physiology A, Molecular
and Integrative Physiology 120:587–598.
McKenzie, R., B. Connor, and G. Bodeker. 1999. Increased
summertime UV radiation in New Zealand in response to
ozone loss. Science 285:1709–1711.
Miller, L. M., L. Kallemeyn, and W. Senanan. 2001. Spawning-site and natal-side fidelity by northern pike in a large
lake: mark-recapture and genetic evidence. Transactions of
the American Fisheries Society 130:307–316.
Mitchell, D. L. 1996. Radioimmunoassay of DNA damaged
by ultraviolet light. Pages 73–83 in G. Pfeifer, editor. Technologies for detection of DNA damage and mutations. Plenum, New York, New York, USA.
Mitchell, D. L. 1999. Quantification of DNA photoproducts
in mammalian cell DNA using radioimmunoassay. Pages
165–175 in D. S. Henderson, editor. Methods in molecular
biology: DNA repair protocols. Humana, Totowa, New Jersey, USA.
Molot, L. A., et al. 2004. Risk analysis of dissolved organic
matter-mediated ultraviolet B exposure in Canadian inland
waters. Canadian Journal of Fisheries and Aquatic Sciences
61:2511–2521.
Morris, D. P., H. Zagarese, C. E. Williamson, E. G. Balseiro,
B. R. Hargreaves, B. Modenuttie, R. E. Moeller, and C.
Queimalinos. 1995. The attenuation of solar UV radiation
in lakes and the role of dissolved organic carbon. Limnology and Oceanography 42:239–249.
Naiman, R. J., and H. Décamps. 1997. The ecology of interfaces: riparian zones. Annual Review of Ecology and
Systematics 28:621–658.
Neff, B. D., L. M. Cargnelli, and I. M. Côté. 2004. Solitary
nesting as an alternative breeding tactic in colonial nesting
bluegill sunfish (Lepomis macrochirus). Behavioral Ecology and Sociobiology 56:381–387.
Pace, M. L., J. J. Cole, S. R. Carpenter, J. F. Kitchell, J. R.
Hodgson, M. C. Van de Bogert, D. L. Bade, E. S. Kritzberg,
and D. Bastviken. 2004. Whole-lake carbon-13 additions
reveal terrestrial support of aquatic food webs. Nature 427:
240–243.
Palen, W. J., C. E. Williamson, A. A. Clauser, and D. E.
Schindler. 2005. Impact of UV-B exposure on amphibian
embryos: linking species physiology and oviposition be-
Ecological Applications
Vol. 16, No. 1
havior. Proceedings of the Royal Society of London B 272:
1227–1234.
Scheuerell, M. D., and D. E. Schindler. 2004. Changes in the
spatial distribution of fishes along a residential development gradient. Ecosystems 7:98–106.
Schindler, D. W., S. E. Bayley, P. J. Curtis, B. R. Parker, M.
P. Stainton, and C. A. Kelly. 1992. Natural and man-caused
factors affecting the abundance and cycling of dissolved
organic substances in Precambrian shield lakes. Hydrobiologia 229:1–21.
Schindler, D. W., P. J. Curtis, B. R. Parker, and M. P. Stainton.
1996. Consequences of warming and lake acidification for
UV-B penetration in North American boreal lakes. Nature
379:705–708.
Scott, W. B., and E. J. Crossman. 1979. Freshwater fishes of
Canada. Bulletin 184. Fisheries Research Board of Canada,
Ottawa, Ontario, Canada.
Shindell, D. T., D. Rind, and P. Lonergan. 1998. Increased
polar stratospheric ozone losses and delayed eventual recovery owing to increasing greenhouse-gas concentrations.
Nature 392:589–592.
Siebeck, O., T. L. Vail, C. E. Williamson, R. Vetter, D. Hessen,
H. Zagarese, E. Little, E. Balseiro, B. Modenutti, J. Seva,
and A. Shumate. 1994. Impact of UV-B radiation on zooplankton and fish in pelagic freshwater ecosystems. Archiv
für Hydrobiologie Beihefte Ergebnisse der Limnologie 43:
101–114.
Webster, D. A., and G. Eiriksdottir. 1976. Upwelling water
as a factor influencing choice of spawning sites by brook
trout (Salvelinus fontinalis). Transactions of the American
Fisheries Society 1054:416–421.
Wilhelm, S. W., W. H. Jeffrey, C. A. Suttle, and D. L. Mitchell. 2002. Estimation of biologically damaging UV levels
in marine surface waters with DNA and viral dosimeters.
Photochemistry and Photobiology 76:268–273.
Williamson, C. E., S. E. Metgar, P. A. Lovera, and R. E.
Moeller. 1997. Solar ultraviolet radiation and the spawning
habitat of yellow perch, Perca flavescens. Ecological Applications 7:1017–1023.
Williamson, C. E., R. S. Stemberger, D. P. Morris, T. M. Frost,
and S. G. Paulsen. 1996. Ultraviolet radiation in North
American lakes: attenuation estimates from DOC measurements and implications for plankton communities.
Limnology and Oceanography 41:1024–1034.
Yan, N. D., W. Keller, N. M. Scully, D. R. S. Lean, and P. J.
Dillon. 1996. Increased UV-B penetration in a lake owing
to drought-induced acidification. Nature 381:141–143.