Article

pubs.acs.org/est

A Meta-Analysis Evaluating the Relationship between Aquatic

Contaminants and Chironomid Larval Deformities in Laboratory
Studies
Bryant S. Gagliardi,*,† Vincent J. Pettigrove,† Sara M. Long,‡ and Ary A. Hoffmann‡
†
Centre for Aquatic Pollution Identification and Management (CAPIM), BioSciences 4, School of BioSciences, The University of
Melbourne, Melbourne, Victoria 3010 Australia
‡
Centre for Aquatic Pollution Identification and Management (CAPIM), Bio21 Institute and School of BioSciences, The University
of Melbourne, 30 Flemington Road, Parkville, Victoria 3010 Australia
*
S Supporting Information

ABSTRACT: Chironomid larval deformities have been widely investigated as an aquatic

pollution toxicity end point. Field chironomid surveys often show a spatial association
between contaminants and deformities, suggesting contaminants cause deformities. However,
over 40 years of laboratory assays have not been able to confirm this causality. We therefore
conducted a review of the literature and meta-analysis, in order to (A) assess whether trends
across assays indicated dose−response effects, (B) characterize the consistency of results, and
(C) investigate whether experimental issues and publication bias were contributing to
inconsistency and/or reducing confidence in results. The experimental issues we investigated
were extraneous nonchemical laboratory stressors (which may mask or interact with chemical
effects), and mortality (which can confound deformity results). Our meta-analysis of the most
commonly tested chemicals suggested dose−response effects for copper, but not lead or zinc.
However, we also found substantial inconsistency across studies. Both mortality and extraneous stressors were potentially
contributing to this inconsistency, reducing confidence in most published data. We observed no evidence of publication bias. We
conclude that any causal link between contaminants and deformities remains uncertain, and suggest improved experimental and
data reporting procedures to better assess this relationship.

■ INTRODUCTION
Since the early 1970s, it has been observed that chironomid
the most recent chironomid deformity literature review was
published over a decade ago,3 and there is no meta-analysis in the
(Diptera: Chironomidae) larvae collected from polluted fresh- peer-reviewed literature, there is an opportunity to investigate
water sites often show a higher frequency of deformities (of the trends across >40 years of published literature.
mouthparts and/or antennae) than those collected from Meta-analysis can be used to evaluate cause-effect pollution
relatively unpolluted sites.1,2 These spatial associations between toxicity relationships across published studies.8 By analyzing the
chemical contamination and deformities have led to the results of multiple studies, relationships can be evaluated across a
frequently posed hypothesis that aquatic contaminants are causal
broader range of concentrations than a single study, and indicate
to chironomid deformities. Deformities have consequently been
widely investigated as a potential sublethal stress end point in the “overall” trends determined across studies.7 This approach
pollution toxicity studies.3 However, field results are subject to may be particularly useful in investigating dose−response
the influence of a variety of extraneous (i.e., nonchemical) relationships between chemicals and chironomid deformities,
stressors,4 making it difficult to unequivocally attribute observed as the apparent variation between individual studies has
deformities to contaminants. The evidence of chemical causality generated uncertainty as to overall dose−response effects.9
produced in field studies is hence largely “circumstantial”.3,5 Here we conduct a literature review and meta-analysis to explore
To further investigate causality, laboratory studiesusually patterns in the literature on laboratory deformity studies. We first
exposing Chironomus (the standard chironomid test genus)
outline factors that might contribute to inconsistent and/or
larvae to chemicalshave been conducted since the earliest
observations of deformities.1 These studies have yielded unreliable results before considering the specific aims of the
apparently inconsistent results.5 This inconsistency may be study.
indicative of experimental artifacts or experimental design issues,
and reduces confidence in a causal relationship between Received: August 9, 2016
contaminants and deformities.6 Literature reviews6 and meta- Revised: October 18, 2016
analyses7 can serve to evaluate dose−response relationships, and Accepted: October 27, 2016
diagnose issues of concern across published bioassays. Given that Published: October 27, 2016

© XXXX American Chemical Society A DOI: 10.1021/acs.est.6b04020

Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

■ SOURCES OF INCONSISTENCY AND

UNRELIABILITY
probability of exhibiting deformities than the killed fraction
(which is typically eliminated from the analysis). A false negative
may hypothetically arise if more sensitive larvaesusceptible to
Results of different assays appear to be inconsistent, whether
both chemically induced deformity and mortalityare killed by
comparing tests of the same chemical (e.g., copper5), or the
the chemical and hence excluded from analysis. This has been
results of different chemical assays.5,9 Results have variously
hypothesized to be the case in uranium16 and lead + zinc17
suggested positive, nonsignificant or even negative associations
exposures, which yielded unexpected negative correlations
between chemical concentrations and deformity frequency/
between chemical concentrations and deformity frequencies.
severity.5 However, neither the “within-chemical” nor the
A false positive may arise if a proportion of the chironomid
“across-chemical” inconsistencies have been quantified. It is population is “naturally” deformed, that is, the apparently
unknown whether results are systematically inconsistent, or if “deformed” morphology is simply a natural phenotype occurring
only a small number of unrepresentative outlier results are giving at a certain frequency within the population. If individuals with
a false impression of inconsistency. this morphology have a greater tolerance to the exposure
While potential causes of inconsistent patterns have been chemical, they will better survive the exposure, hence the
considered, they have rarely been empirically investigated. proportion of “deformed” larvae will increase with chemical
Differences between human observers in their designation of concentration. Such a correlation is actually a spurious
larvae as either “deformed” or “non-deformed” has been association between concentration and deformity frequency, as
demonstrated as one potential source of inconsistency;10 and the chemical has not actually physiologically induced the
the use of different Chironomus species (which can differ in their “deformity”. This possibility (i.e., an association between
chemical sensitivity11) may also contribute. Two additional deformities and mortality) may be suggested by results such as
extraneous experimental factors have been suggested as those of Janssens de Bisthoven et al.,18 who observed a clear
influential to chironomid deformity assays. These two factors, positive deformity-mortality correlation, in this case without
“background stressors” and mortality of larvae, may contribute to observing such a clear chemical-deformity association. A false
inconsistency if they are present at different magnitudes between positive may also arise even if there is no differential chemical
studies. They may also compromise (i.e., reduce confidence in) sensitivity between deformed and nondeformed larvae. For
assay results, by potentially resulting in false positives or false instance, if deformed larvae have a lower survival overall because
negatives. Another factor that can reduce confidence in published they are poorer competitors, an increase in mortality may lead to
results is publication bias. These three issues (background a higher incidence of deformity simply because the level of
stressors, mortality, and publication bias) are discussed hence- competition is reduced, irrespective of any inherent levels of
forth. resistance. Confounding of deformity results by mortality can
Background Stressors. Potential laboratory background/ therefore probably only be avoided by use of lower-than-lethal
extraneous stressors (i.e., those other than the exposure exposure concentrations.
chemical) include inadequate substrates, inbreeding, parental A high mortality rate in any treatment, as well as being the
effects, and unknown stressors in substrates and food.3 result of chemically induced mortality, may be the result of
Experimental evidence supports at least two of these factors background mortality, which occurs independently of chemical
inbreeding depression12 and parental effects13as potential concentration. As laboratory assays typically aim to eliminate all
influences on deformity results. Given that the list of potential extraneous stressors,4 background mortality levels should
background stressors is long,3 the likelihood of them varying normally be low. However, when background mortality occurs,
between studiesand hence contributing to result incon- mortality-susceptible larvae (which may have a higher or lower
sistencyis high. Background stressors may also compromise probability of exhibiting deformity than those surviving) will
results through additive, synergistic or antagonistic interactions usually be excluded from the analysis. In addition to selecting for
with chemical toxicity, to increase/decrease the deformity mortality-resistant larvae, both chemically induced and back-
response,14,15 thereby risking false positive or false negative ground mortality reduce sample size. Both forms of mortality
results. They may also lead to false negative results by “masking” hence also risk false negatives by reducing statistical power and
chemical effects, that is, inducing a high deformity rate in control therefore the capacity of an experiment to detect a deformity
larvae,3 hence reducing the observed effect size between control effect (i.e., type II error).
and exposed larvae.15 Background stressors may also themselves Publication Bias. Another potentially compromising factor
induce a deformity response.12 is publication bias. This occurs when the research that appears in
Mortality. Mortality in assays can fall into two categories: the published literature is systematically unrepresentative of the
mortality induced by the exposure chemical (“chemically- population of completed studies.19 Publication bias can come
induced mortality”), and mortality induced by extraneous about when there is a bias toward publishing positive results that
stressors (“background mortality”). The potentially distortive confirm or substantiate previous findings, or alternatively, when
effects of mortality upon deformity results are complex, and are there is an increased likelihood of publishing negative results as
outlined henceforth. an increasingly critical view of a field develops over time.20 It is a
Although deformities are a sublethal end point, published potential problem within ecotoxicology.21 Publication bias in
deformity assays often expose larvae to lethal chemical chironomid deformity studies has not been investigated
concentrations.5 This has the potential to confound deformity previously. Although it may reduce confidence in published
results, leading to false positive or false negative results. Larvae data,19 publication bias is not likely to contribute to result
that survive a given concentration are likely to have numerous inconsistency.
physiological differences from those killed by that concentration. This study presents a systematic, quantitative analysis of the
“Likelihood of exhibiting deformities” may be one of these literature of laboratory chironomid deformity assays, in order to
differences: the surviving fraction of the population (in which (A) assess whether trends across published assay results are
deformities are typically quantified) may have a greater or lesser indicative of dose−response effects, (B) characterize the
B DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

inconsistency of results, and (C) investigate whether exper- for that concentration. For studies that reported deformity
imental issues and publication bias are present. frequencies at several time points during an experiment (e.g., ref

■ MATERIALS AND METHODS

Literature Search. We conducted an Internet-based search
27), we reported the total deformity frequency (i.e., that at the
termination of the experiment).
For each exposure treatment, we recorded dose magnitude as
of the research literature (peer-reviewed journal articles and (exposure treatment concentration/control concentration) and
academic text book chapters). In separate searches, we entered response magnitude as (exposure treatment deformity fre-
the search terms “chironom* + deform*” and “chironom* + quency/control deformity frequency), following the meta-
abnorm*” (to capture Chironomidae, chironomid, Chironomus, analysis methods of O’Brien and Keough.7 The exact values of
deformity, deformed, deform, deformation/s, deformities, deformity frequencies that were depicted graphically were not
abnormalities and abnormal) into the search engine Google reported, so were estimated using the plot digitizing software
Scholar and the databases Web of Science and Scopus, and PlotDigitizer (http://plotdigitizer.sourceforge.net/). Control
imported all results using the reference management program chemical concentrations that were below the analytical limits
EndNote. The titles and abstracts of all English-language of detection (LOD) were recorded as 1/2 × LOD,28 which also
publications (published prior to 1 January 2016) were read, allowed a ratio to be taken. Analogously, deformity frequencies of
and each publication investigating the incidence of chironomid 0 were replaced with half the maximum possible frequency in the
larval deformities in laboratory chemical exposure experiments sample given the sample size:
was saved. For each study, we recorded the test taxon used and
the exposure chemical/s. (
freq deformed larvae = 1/2
1
nlarvae in treatment + 1 ) Treatments where
Analysis 1: Meta-Analysis Assessing Potential Dose− LOD was not reported for chemical nondetects, or where
Response Effects. Exposure chemicals analyzed in ≥3 nlarvae in treatment was not reported alongside zero deformity
publications, using Chironomus as a test organism, were assessed frequencies, were therefore eliminated from Analysis 1. Each
in Analysis 1. For each chemical, its concentration in each control data point in this meta-analysis represents the deformity
and exposure treatment for each experiment were recorded. We frequency in a single exposure treatment within a single
also recorded whether experiments involved single or mixed- publication.7
chemical exposures. We recorded concentrations in each For each chemical, we then conducted Spearman’s correlation
analyzed exposure matrix in ppm (i.e., mg/L for water analysis of deformity results to infer whether the overall data
concentrations, mg/kg dry weight for sediment and tissue supported a dose−response relationship, which is a key indicator
concentrations). Although substrate contaminant concentrations of causality.29 For analysis and plotting, we took the ln of each
are likely to be particularly relevant to the sediment-dwelling dose and response value.7 Therefore, positive ln(response)
Chironomus,22 we noted that some studies did not report values corresponded to an increased deformity frequency in
substrate concentrations, meaning we could not assess substrate response to chemical exposure, zero value to no response, and
effects for these studies. For studies that reported both negative ln(response) values corresponded to a reduction in
analytically determined and nominal concentrations, we deformity frequency in response to exposure. As there was
recorded the analytically determined concentration as this is substantial spread in this data (in relation to the range of
the more accurate exposure level. For studies that measured concentrations tested), ln-transforming the x-axis data meant it
concentrations at several time points, we took the average could be plotted on a more meaningful scale. Potential dose−
concentration determined (e.g., refs 23 and 24). We did not response effects for each chemical were inferred by a significant
analyze results in which chemical concentrations in controls and/ (p < 0.05) positive correlation between ln(dose) and ln-
or treatments were unreported. (response).7
Experiments using field-contaminated (as opposed to We also performed a sign test of this data to further test for
laboratory-“spiked”) sediments were excluded from Analysis 1, potential causality. For each treatment in this data set, we
with one exception, Di Veroli et al.23 This was because these recorded the number of positive, zero, and negative ln(response)
studies (except for Di Veroli et al.23) compared deformity values. For each study (with 5 or more exposure treatments) we
frequencies in relation to either “site from which sediment was tested whether the number of positive values was greater than the
sampled” (i.e., control sediment versus polluted site sediment),25 number of nonpositive (i.e., zero plus negative) values. We also
or “concentration of polluted site sediment” (i.e., a concentration performed this analysis at the “contaminant” level, pooling all
gradient of polluted site sediment diluted with uncontaminated study values (n ≥ 5 treatments) for a given chemical. Potential
sediment),26 rather than in relation to a specific contaminant chemical causal effects were inferred where the frequency of
concentration. This made it difficult to analyze any chemical- positive responses was significantly higher (p < 0.05) than
deformity dose−response relationship. nonpositive values.
For each control and exposure treatment, we recorded the Analysis 2: Meta-Analysis Characterizing Inconsis-
deformity frequency. Deformity “severity indices/scores” are tency of Results. To characterize inconsistency of results, we
sometimes reported alongside deformity frequencies. These created a boxplot chart of ln(response) values from Analysis 1,
indices are derived by applying a mathematical formula to separately creating boxplots for each of the most common
deformity frequencies, such that treatments inducing “severe” exposure chemicals. In each case, results were considered
deformities are scored as more deleterious than those inducing inconsistent if they, with the exclusion of potential outliers
relatively “benign” deformities. As this distinction is subjective, [i.e., values outside the range [Quartile 1 − (1.5 × Interquartile
and such indices have been shown to be redundant with range) to Quartile 3 + (1.5 × Interquartile range)], comprised
deformity frequency results,9 such results were excluded from negative, zero and positive values, as implied by “inconsistency”.5
this analysis. For studies that investigated the deformity effects of Analysis 3: Literature Review Investigating Incidence
a chemical concentration across several generations (e.g., refs 12 of Compromising/Inconsistency-Contributing Experi-
and 18), we averaged the deformity frequency across generations mental Factors. We considered evidence of background
C DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

stressor, chemically induced mortality, and background mortality We therefore performed a funnel plot analysis of all treatments
effects in the literature. The presence of these (or, the failure to for which deformity frequencies and sample size (i.e., number of
satisfactorily eliminate their occurrence) was taken as evidence of larvae) was reported. We calculated LOR, that is,
their potential contribution to inconsistency and reduced ndeformed larvae in exposure ÷ n non − deformed larvae in exposure
confidence in the results. This literature review analyzed results
ln( ndeformed larvae in control ÷ nnon − deformed larvae in control ), and its inverse stand-
at the “experiment” level, that is, each statistical analysis of an 38
ard error, for each treatment. We plotted this data, and tested
experimental result. Data for all exposure chemicals was analyzed for evidence of publication bias using Egger’s test for funnel plot
in Analyses 3 and 4. asymmetry.39 Significant asymmetry was inferred if p < 0.05.
As with Analyses 1 and 2, severity score/index results were not We also investigated whether “year of publication” of a study
considered. Analyses of field-contaminated sediments were had a significant association with the magnitude of the deformity
included, however, as Analysis 3 did not require the input of effect size (as LOR) in Chironomus studies. It has been suggested
chemical concentrations. Also included, unlike Analyses 1 and 2, that such a “year effect” can be diagnostic of a tendency to favor
were results that did not necessarily report deformity publication of results that either substantiate or refute previous
frequencies, as long as statistical analysis was included. As per findings, which constitutes a publication bias.20 We conducted
the previous analyses, we also excluded non-Chironomus Spearman’s correlation analysis of the data, and a potential
results.30 publication bias was inferred if a significant (p < 0.05) positive
To quantify the number of experiments which had apparently correlation (indicating a tendency to confirm previous findings)
not eliminated background stressors (“Factor A”), we recorded or negative correlation (indicating a tendency to refute previous
whether the deformity frequency in the control treatment was findings) was observed between year and LOR.20 All statistical
reported and was <10%. We set 10% as the nominal maximum analyses in the present study were conducted in R.40
“acceptable” control deformity frequency as 1) it is the maximum
acceptable control mortality rate for valid chironomid lethality
assays according to the American Society for Testing and
■ RESULTS
Literature Search. Our literature search found 41 studies
Materials International guidelines,31 hence is representative of a investigating chemical effects on chironomid deformity
generally acceptable “control stress” level, 2) it is lower than 16%, frequency/severity in laboratory experiments (Supporting
which is considered an unacceptably high control deformity Information, S1). All studies investigated effects in Chironomus
frequency by some researchers,3 and 3) it is generally concordant spp., with the exception of Dickman and Rygiel,30 which
with field estimates of deformity rates of Chironomus populations investigated effects in Chironomidae. The latter study was
inhabiting relatively unpolluted sites, though these can vary excluded from the following analyses in order to focus on
substantially, from 0%32 to as high as 20%.33 Chironomus.
To quantify the number of experiments which had apparently Analysis 1. The most commonly assayed chemicals were
not satisfactorily eliminated chemically induced mortality copper, lead, cadmium, and zinc (Table 1). Of these, the number
(“Factor B”), we recorded whether the difference in mortality of studies reporting the parameters necessary for correlation
rate between control and treatments was statistically analyzed analysis comprised four copper studies, seven lead studies, and
and was not significant (p > 0.05). To further test for satisfactory three zinc studies.
elimination of chemically induced mortality, and additionally, Potential dose−response effects were inferred for copper
potential background mortality effects, for each analysis we (Figure 1A, ρ = 0.19, p = 0.03), but not for lead (Figure 1B, ρ =
recorded whether the mortality frequency was reported for all 0.07, p = 0.31) or zinc (Figure 1 C, ρ = 0.09, p = 0.61). In the sign
(control and exposure) treatments, and whether it was <20% in test analyses, the frequency of positive responses varied between
each case (“Factor C”). Twenty percent mortality in controls is studies, however it was generally near or higher than 50%
sometimes considered the maximum allowable for a valid (Supporting Information, Table S1). Significant study or
chironomid lethality experiment (e.g. refs 34 and 35), so we contaminant-level effects were not found for any of the metals
nominally set this as an “acceptably low” mortality rate. (p > 0.05), however only one zinc study had a sufficient number
Analysis 4: Meta-Analysis Investigating Publication of treatments for analysis.
Bias As a Compromising Factor. Publication bias can be Analysis 2. After the exclusion of outliers, ln(response)
indicated by an association between sample sizes and outcomes results for copper, lead, and zinc spanned negative, zero and
of experiments. In the absence of publication bias, outcomes of positive values, indicating a high level of inconsistency in
experiments will be independent of sample size, as both large and experimental results (Figure 2).
small experiments, resulting in either positive or negative Analysis 3. Two hundred and eighty-five statistical analyses,
outcomes, will have an equal likelihood of publication. However, from 37 studies, were considered. For Factor A, the majority of
an association between these two variables can indicate analyses appeared to avoid background stressors. However, these
publication bias, often seen as a paucity of small experiments were still potentially common in published experiments: 70 out
resulting in negative outcomes.36,37 This can be inferred by of 285 (i.e., 24.56%) analyses either did not report the deformity
funnel plot analysis; for binomial data this can be investigated by frequency in controls, or reported a control deformity frequency
plotting the log odds ratio (LOR, a measure of effect size) of a of >10%. The control deformity frequency was reported for 282
treatment against its inverse standard error (1/SE, an indicator of out of the 285 analyses; values ranged from 0 to 68.9% (average =
sample size and hence precision; where a high 1/SE value 7.00%).
corresponds to a large sample size).37 In the absence of Mortality was the more common factor potentially contribu-
publication bias, this plot will form a symmetrical “funnel” ting to inconsistency. For Factor B, 263 out of 285 (i.e., 92.28%)
shape around the mean LOR, with an increasing variability in of analyses either did not statistically compare mortality rate
LOR as 1/SE decreases. Asymmetry about this mean may between control and treatments, or performed this comparison
suggest publication bias, for example, a lack of data points with and observed a significant (p < 0.05) effect, indicating
both low log odds-ratio and 1/SE values. nonavoidance of chemically induced mortality. Analysis for
D DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

Table 1. Number of Studies Investigating Each Chemical for responses at lower doses. A negative response is where the
Which Chemical Concentrations and Deformity Responses deformity frequency is reduced in response to chemical exposure,
Are Reported meaning the interpretation of these patternsand inference of
dose−response effectsis not straightforward.
chemical number of studies
Some ln(dose) values were <0 (Figure 1), meaning there was a
Pb 10 higher chemical concentration in the control than in the
Cu 5 experiment. Though unexpected, these values are still relevant
Zn 4 to the dose−response analysis. For these values, corresponding
Cd 3 negative ln(response) values were taken as “positive responses”
17α-ethynylestradiol 2 in the sign test analyses and Analysis 2 (see below).
4-nonylphenol 2 Meta-analysis data points can either represent a whole study,
DDT 2 or a specific result within a study.42 In the latter situation, data
Ni 2 points from the same study could be viewed as having a limited
2,4-D 1 independence. As such, the analysis may be viewed as having
210
Pb 1 relatively low statistical power (e.g., one may consider that n = 3
acridone 1 in the zinc correlation analysis). The potential dose−response
As 1 effects supported/not supported by Analysis 1 are therefore
bisphenol A 1 subject to confirmation by further experimentation (see also the
β-sitosterol 1 issues discussed below).
chlorpyrifos 1 Analysis 2. To date, uncertainty about the causality of
Cr 1 chemical contaminants to larval deformities has limited their
dacthal 1 application in pollution monitoring and ecotoxicology; the
DDE 1
consensus view is that laboratory results are too inconsistent to
Di(2-ethylhexyl)phthalate 1
unequivocally demonstrate causal effects of chemicals upon
fenbendazole 1
deformities.9,18,24,41,43−45 Analysis 2 results are consistent with
Hg 1
this perception: a substantial proportion of results suggest no
imidacloprid 1
association, or even a negative association between chemicals and
thiacloprid 1
deformities, even after the exclusion of potential outliers.
tributyltin 1
However, there are methodological differences between
U 1
studies that could weaken the detection of patterns in both
xylene 1
Analyses 1 and 2. These include different researchers scoring
ZnO microparticles 1
deformities, use of different Chironomus test species, inconsistent
ZnO nanoparticles 1
categorization of deformity types, different chemical mixtures/
forms (e.g., metal salts) and concentration ranges, different
Factor C found that a mortality rate of >20% in controls and/or exposure matrices, and differing experimental parameters (e.g.,
exposure treatments, or nonreporting of mortality rates, was also temperature); all of these may contribute to interexperiment
common across analyses. This result was observed for 272 out of variability.5,10,11,23,44,46 Therefore, the inconsistency observed in
285 (i.e., 95.41%) of the analyses, potentially indicating Analysis 2 may be either due to these extraneous factors, or to the
chemically induced and/or background mortality effects. A low natural variability in Chironomus deformity responses to
percentage of analyses met all three criteria for studies without contaminants. If either of these situations is applicable, it
potential compromising effects contributing to inconsistency: 4 suggests that a high frequency of deformities is not necessarily a
out of 285 analyses (1.40%). All of the four “reliable” results reliable indicator of contaminant-induced stress, as deformity
observed no significant association between chemical (β- frequencies do not consistently increase with concentration. In
sitosterol, mercury or lead) concentration and deformity addition, the inconsistency of results may point toward “noise” in
frequencies.41 the data. This may indicate that deformities are responding to
Analysis 4. Four hundred and forty-six treatments (from 17 factors other than contaminant stress, that is, chemicals may not
studies) reported sample size (i.e., the number of larvae be causal to deformities. Less “noisy” data may result from
analyzed). The funnel plot of this data set appeared symmetrical deformity responses to other stimuli, which may suggest
about the overall mean LOR, 0.46 (Supporting Information, nonchemical factors as causal to deformities. We have therefore
Figure S1). This was confirmed by Egger’s test, which observed inferred that the relative frequency of zero and negative chemical-
no significant departure from symmetry (t = −0.99, df =444, p = deformity associations may suggest experimental artifacts (see
0.32). This result indicates no publication bias based upon a Analysis 3) and/or a lack of causality.6 Negative associations can
sample size analysis. Our temporal analysis also revealed no also suggest nonmonotonic relationships,2,47 and these may be
publication bias, with no significant change in effect size detected supported by the frequency of negative ln(response) values at
over time (ρ = 0.09, p = 0.052, Supporting Information, Figure low doses in Analysis 1 (Figure 1). However, this possibility
S2).

■
remains speculative, as ecotoxicological study designs typically
involve too few exposure treatments and replicates for accurately
DISCUSSION diagnosing nonmonotonic/hormetic associations.48
Analysis 1. Our analysis of laboratory results observed Although our study assessed only laboratory results, field
potential dose−response/causal effects for copper, but not lead results may be similarly inconsistent. For example, in an extensive
or zinc. This provides some support for copper as causal to field study by Janssens de Bisthoven et al.,49 investigating 37
deformities. However, in the case of the correlation analysis, the deformity-chemical correlations, 24 were positive and 13 were
pattern appears somewhat driven by a number of negative negative. We also noted that the LOR (i.e., effect size) data in
E DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

Figure 1. Dose−response meta-analysis results for (A) copper, (B) lead, and (C) zinc. Spearman’s correlation analysis suggested dose−response effects
for copper (ρ = 0.19, p = 0.03), but not lead (ρ = 0.07, p = 0.31) or zinc (ρ = 0.09, p = 0.61). Triangular data points represent substrate exposures,
diamond-shaped points are for water exposures, circular points are for tissue concentrations.61,62,63

shows many values at or below x = 0, indicating a substantial

number of zero and negative effects. This variability in this
“general chemical” response may bring into question the
suggestion that deformities are a general contamination indicator
for field pollution assessments.
Analysis 3. We found that background stressor and, more
commonly, mortality effects, were potentially contributing to
result inconsistency, reducing confidence in published data. By
extension, there is a low confidence in the results of our Analyses
1 and 2. The potentially “compromising issues” we investigated
have been suggested as potentially problematic either within
chironomid deformity studies specifically,3,12,16,46 or in ecotox-
icology more broadly.21
A low percentage (1.40%) of analyses met all three criteria for
studies without potential background stressor or mortality
Figure 2. Characterization of result inconsistency: boxplot of effects. All four of these “reliable” results observed no significant
ln(response) data for Cu, Pb, and Zn exposures. Circular data points association between chemical (β-sitosterol, mercury or lead)
represent potential outlier values, gray line is the line of “no chemical
concentration and deformity frequencies.41 The null hypothesis
effect”.
of “no chemical effect” is therefore yet to be satisfactorily rejected
in robust laboratory experimentation. Background stressors in
Analysis 3testing responses across all chemicalsshowed most cases will be indicated by a high deformity frequency in
substantial variability. Figure S2 (Supporting Information) controls.3,14 One background stressor, inbreeding depression,
F DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

can be eliminated by the use of apomictic parthenogenetic test have not reported mortality data, as potentially mortality-
species;50 or minimized by use of large (>1000) laboratory affected. It is not our aim to criticize the authors of such studies,
breeding population sizes and/or populations comprising field but to emphasize the importance of reporting (and statistically
and/or mixed laboratory sources.14,51 Density/competition testing), mortality frequencies in sublethal ecotoxicological
effects can be eliminated by exposing larvae individually in experiments. By comparison, sublethal chemical57,58 and UV
separate test vessels.31 radiation59 levels in teratogenesis assays have observed dose-
Although experimental evidence suggests mortality as a dependent deformity increases in amphibians.
confounding experimental factor,16,18,46 it was common for The “acceptable” background mortality and control deformity
studies to either not report mortality, fail to test for statistical rates we have applied in this study are nominal, and arguably
differences in mortality between controls and treatments, or arbitrary. However, this argument can be made of any threshold
publish results where a control-treatment mortality difference “acceptable” stress level. Nonetheless, these types of nominal
was evident. For any experimental concentration inducing control values are routinely used as guidelines in ecotoxicol-
significant mortality, it is difficult to determine whether the ogy,22,31,34,35 which can serve as coarse indicators of compro-
observed deformity frequency is due to a direct chemical toxicity mised experimental conditions. More meaningful values might
effect, or an indirect effect of mortality acting to increase/ be those known to have a biological relevance, for example a
decrease deformity frequency within the surviving population.46 mortality rate known to result in a population-level decline, a
The latter is an artifactual/indirect, rather than a causal, chemical deformity severity known to result in an individual-level
association. Mortality effects could similarly be a compromising reduction in fitness, or a deformity frequency detected in a
factor in field studies, but this is difficult to ascertain, given the well-replicated survey of unpolluted field sites.
difficulty associated with quantifying mortality in wild field Additionally, any application of an “acceptable” control
populations. While it is possible that “mortality-prone” and deformity frequency assumes that deformities are a “stress”
“mortality-resistant” larvae have a similar likelihood of exhibiting (i.e., deleterious) response that is induced by environmental
deformities, existing evidence suggests some difference.18 conditions, as opposed to other factors. This is yet to be clearly
As deformities are a sublethal end point, investigations of demonstrated.9 Presently, little is known about the causes,
chemical causality should involve sublethal concentrations,52 nature, and fitness consequences (at either the individual or
over and above them being “environmentally relevant”. There is population level) of chironomid deformities. These are issues
probably little value employing deformities to assess contami- requiring further research, potentially outside the context of
nant-induced stress in field chironomid populations (e.g., in ecotoxicology. There is currently a bias toward chemical analyses,
environmental monitoring programs) until chemical causality is with inbreeding depression12 and temperature25 to our knowl-
established in controlled assays. Avoidance of mortality first edge being the only nonchemical factors considered in the
requires that mortality be quantified in an assay. It is important laboratory. There may be value in investigating other stressors at
that not only survival in all treatments should be comparable, but sublethal levels, for example, disease, starvation, and density
that survival should be to the same larval instar (and statistically effects.
tested for this, as in the β-sitosterol exposures in Vermeulen et Analysis 4. Analysis 4 revealed no evidence of publication
al.41); as chironomid deformity results can also be potentially bias. While our temporal analysis suggested an upward trend in
confounded by instar5 or developmental stage (e.g., if larvae are effects sizes over time, this effect was not statistically significant
differentially lost to metamorphosis between treatments).26 (Figure S2, Supporting Information). The published data
Second, avoidance of mortality requires that any mortality- therefore appear representative of all derived deformity data.
inducing experiment be regarded as “questionable”, and repeated This also suggests that publication bias is not acting to reduce
at lower (in many cases, lower than “environmentally relevant”) confidence in the published data.
chemical exposure concentrations, and that this process be These inferences are limited to laboratory data, and field data
repeated until a mortality difference in control-treatment have not been investigated for publication bias. We additionally
comparisons is not detected, with a reasonable level of statistical note that sample size (number of larvae) was not reported for
power to avoid type II error. In the first instance, lethality- over 200 treatments. This was either due to authors not reporting
inducing chemical concentrations should not be used to the number of larvae chemically exposed/scored for deformities,
investigate a causal relationship between that chemical and a or the number of larvae lost through either mortality,
sublethal stress response. Perhaps only chemical concentrations metamorphosis, or failure to develop to the required instar was
that have a significant mortality impact on a population are not quantified. Sample size is a key determinant of experimental
capable of inducing deformities. However, in that case chemical precision. It is therefore important that, for each treatment, the
effects would be difficult to differentiate from artifactual mortality exact number of deformity-scored larvae is quantified (such as in
effects, except perhaps in cases where the control deformity Arambourou et al.60 and Di Veroli et al.23).
frequency is zero. In summary, our meta-analyses revealed possible copper, but
It is evident that authors have different views on the not lead or zinc dose−response effects upon chironomid
importance of reporting mortality rates. In some cases, mortality deformities. However, we also revealed significant inconsisten-
rates are explicitly quantified, in some they are referenced in text cies across published results. Potential background stressor
without quantification, in some they are quantified and effects and mortality of larvae appeared to contribute to this
statistically compared to control levels (e.g., refs 41 and inconsistency, and to a reduced confidence in most published
53−56), and in some studies they are neither mentioned nor data. Future experiments can better assess chemical-deformity
quantified. The latter situation may sometimes indicate that no associations by eliminating these factors. Uncertainty regarding
mortality has occurred, but in some such cases it is clear that these associations, along with the objective designation of
mortality has occurred, such as studies that investigate effects at “deformity”,10 and the bearing of deformities on individual and
LC50 levels. Because of this uncertainty, we have adopted a population fitness,9 currently limit the application of deformities
conservative approach, where we considered any studies that as an ecotoxicological end point.
G DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

■
*
ASSOCIATED CONTENT
S Supporting Information
(11) Watts, M. M.; Pascoe, D. A comparative study of Chironomus
riparius Meigen and Chironomus tentans Fabricius (Diptera:Chirono-
midae) in aquatic toxicity tests. Arch. Environ. Contam. Toxicol. 2000, 39
The Supporting Information is available free of charge on the (3), 299−306.
ACS Publications website at DOI: 10.1021/acs.est.6b04020. (12) Vogt, C.; Langer-Jaesrich, M.; Elsässer, O.; Schmitt, C.; Van
List of reviewed publications, sign test analyses, and results Dongen, S.; Köhler, H. R.; Oehlmann, J.; Nowak, C. Effects of
of publication bias analyses. References: Supporting inbreeding on mouthpart deformities of Chironomus riparius under
Information S1 List of reviewed publications Table S1 sublethal pesticide exposure. Environ. Toxicol. Chem. 2013, 32 (2), 423−
Sign test analyses for experimental treatments Figure S1: 425.
Funnel plot of experimental treatments for which samples (13) Servia, M. J.; Cobo, F.; Gonzalez, M. A. Incidence and causes of
deformities in recently hatched larvae of Chironomus riparius Meigen,
size (number of larvae) was reported. Figure S2:Scatter-
1804 (Diptera, Chironomidae). Fundam. Appl. Limnol. 2000, 149 (3),
plot of year of publication of each deformity treatment 387−401.
against log odds ratio (PDF) (14) Nowak, C.; Jost, D.; Vogt, C.; Oetken, M.; Schwenk, K.;

■ AUTHOR INFORMATION
Corresponding Author
Oehlmann, J. Consequences of inbreeding and reduced genetic variation
on tolerance to cadmium stress in the midge Chironomus riparius. Aquat.
Toxicol. 2007, 85 (4), 278−284.
(15) Gagliardi, B. S.; Hoffmann, A. A.; Pettigrove, V. J. Inbreeding
*Phone +61383444331; e-mail: [email protected]. depression as a compromising factor in ecotoxicological assays. Integr.
edu.au. Environ. Assess. Manage. 2016, 12 (3), 595−597.
Author Contributions (16) Dias, V.; Vasseur, C.; Bonzom, J. M. Exposure of Chironomus
The manuscript was written through contributions of all authors. riparius larvae to uranium: Effects on survival, development time,
All authors have given approval to the final version of the growth, and mouthpart deformities. Chemosphere 2008, 71 (3), 574−
manuscript. 581.
(17) Martinez, A.; Moore, B. C.; Schaumloffel, J.; Dasgupta, N.
Notes Induction of morphological deformities in Chironomus tentans exposed
The authors declare no competing financial interest.

■
to zinc- and lead-spiked sediments. Environ. Toxicol. Chem. 2001, 20
(11), 2475−81.
ACKNOWLEDGMENTS (18) Janssens De Bisthoven, L.; Postma, J.; Vermeulen, A.; Goemans,
We thank Dr Philippa Griffin for assistance with the layout of G.; Ollevier, F. Morphological deformities in Chironomus riparius
Figure 1, and four anonymous reviewers for helpful comments Meigen larvae after exposure to cadmium over several generations.
Water, Air, Soil Pollut. 2001, 129 (1−4), 167−179.
on the manuscript. This research was funded by the Centre for
(19) Rothstein, H. R.; Sutton, A. J.; Borenstein, M. Publication Bias in
Aquatic Pollution Identification and Management, The Uni- Meta-Analysis: Prevention, Assessment and Adjustments; John Wiley &
versity of Melbourne.

■
Sons: West Sussex, 2006.
(20) Jennions, M. D.; Møller, A. P. Relationships fade with time: A
REFERENCES meta-analysis of temporal trends in publication in ecology and
(1) Hamilton, A. L.; Saether, O. A. The occurrence of characteristic evolution. Proc. R. Soc. London, Ser. B 2002, 269 (1486), 43−48.
deformities in the chironomid larvae of several Canadian lakes. Can. (21) Wandall, B.; Hansson, S. O.; Rudén, C. Bias in toxicology. Arch.
Entomol. 1971, 103 (03), 363−368. Toxicol. 2007, 81 (9), 605−617.
(2) Warwick, W. F. Morphological abnormalities in Chironomidae (22) OECD.Test No. 219: Sediment-water chironomid toxicity using
(Diptera) larvae as measures of toxic stress in freshwater ecosystems: spiked water; OECD Publishing: Paris, 2004.
Indexing antennal deformities in Chironomus Meigen. Can. J. Fish. Aquat. (23) Di Veroli, A.; Goretti, E.; Paumen, M. L.; Kraak, M. H. S.;
Sci. 1985, 42 (12), 1881−1914. Admiraal, W. Induction of mouthpart deformities in chironomid larvae
(3) Janssens de Bisthoven, L.; Gerhardt, A. A critical appraisal of exposed to contaminated sediments. Environ. Pollut. 2012, 166, 212−
morphological deformities in midge larvae (Insecta, Nematocera: 217.
Chironomidae): Clues for biomonitoring the sediment quality of (24) Arambourou, H.; Beisel, J. N.; Branchu, P.; Debat, V. Patterns of
freshwater ecosystems. Trends Comp. Biochem. Physiol. 2005, 11, 27−36. fluctuating asymmetry and shape variation in Chironomus riparius
(4) Chapman, J. The role of ecotoxicity testing in assessing water (Diptera, Chironomidae) exposed to nonylphenol or lead. PLoS One
quality. Aust. J. Ecol. 1995, 20 (1), 20−27. 2012, 7 (11), e48844.
(5) Vermeulen, A. C. Elaborating chironomid deformities as (25) den Besten, P. J.; Naber, A.; Grootelaar, E. M. M.; van de Guchte,
bioindicators of toxic sediment stress: the potential application of C. In situ bioassays with Chironomus riparius: Laboratory-field
mixture toxicity concepts. Ann. Zool. Fenn. 1995, 32 (3), 265−285.
comparisons of sediment toxicity and effects during wintering. Aquat.
(6) Harshman, L. G.; Hoffmann, A. A. Laboratory selection
Ecosyst. Health Manage. 2003, 6 (2), 217−228.
experiments using Drosophila: what do they really tell us? Trends Ecol.
(26) Hudson, L. A.; Ciborowski, J. J. H. Teratogenic and genotoxic
Evolut. 2000, 15 (1), 32−36.
(7) O’Brien, A. L.; Keough, M. J. Ecological responses to responses of larval Chironomus salinarius group (Diptera: Chironomi-
contamination: A meta-analysis of experimental marine studies. Environ. dae) to contaminated sediment. Environ. Toxicol. Chem. 1996, 15 (8),
Pollut. 2014, 195 (0), 185−191. 1375−1381.
(8) Hose, G. C.; Murray, B. R.; Park, M. L.; Kelaher, B. P.; Figueira, W. (27) Martinez, E. A.; Moore, B. C.; Schaumloffel, J.; Dasgupta, N.
F. A meta-analysis comparing the toxicity of sediments in the laboratory Morphological abnormalities in Chironomus tentans exposed to
and in situ. Environ. Toxicol. Chem. 2006, 25 (4), 1148−1152. cadmiumand copper-spiked sediments. Ecotoxicol. Environ. Saf.
(9) Hämäläinen, H. Critical appraisal of the indexes of chironomid 2003, 55 (2), 204−212.
larval deformities and their use in bioindication. Ann. Zool. Fenn. 1999, (28) Barmuta, L. A.; Maher, W. A.; Batley, G. E. Guidelines for water
179−186. quality monitoring and reporting. Australas. J. Ecotoxicol. 2001, 7, 199−
(10) Salmelin, J.; Vuori, K. M.; Hämäläinen, H. Inconsistency in the 208.
analysis of morphological deformities in Chironomidae (Insecta: (29) Hill, A. B. The Environment and disease: Association or
Diptera) larvae. Environ. Toxicol. Chem. 2015, 34 (8), 1891−1898. causation? President’s Address. Proc. R. Soc. Med. 1965, 58, 295−300.

H DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology Article

(30) Dickman, M.; Rygiel, G. Chironomid larval deformity frequencies, (50) Gagliardi, B. S.; Long, S. M.; Pettigrove, V. J.; Hoffmann, A. A.
mortality, and diversity in heavy-metal contaminated sediments of a The parthenogenetic cosmopolitan chironomid, Paratanytarsus grimmii,
Canadian riverine wetland. Environ. Int. 1996, 22 (6), 693−703. as a new standard test species for ecotoxicology: Culturing methodology
(31) ASTM. E729 Standard Guide for Conducting Acute Toxicity Tests and sensitivity to aqueous pollutants. Bull. Environ. Contam. Toxicol.
on Test Materials with Fishes, Macroinvertebrates, and Amphibians; ASTM 2015, 95 (3), 350−356.
International:West Conshohocken, PA, 2007. (51) Brown, A. R.; Hosken, D. J.; Balloux, F.; Bickley, L. K.; LePage, G.;
(32) Bird, G. A. Use of chironomid deformities to assess environmental Owen, S. F.; Hetheridge, M. J.; Tyler, C. R. Genetic variation, inbreeding
degradation in the Yamaska River, Quebec. Environ. Monit. Assess. 1994, and chemical exposurecombined effects in wildlife and critical
30 (2), 163−175. considerations for ecotoxicology. Philos. Trans. R. Soc., B 2009, 364
(33) Jeyasingham, K.; Ling, N. Seasonal influence on head capsule (1534), 3377−3390.
deformities in Chironomus zealandicus (Hudson) (Diptera: Chironomi- (52) Abdel-Moneim, A.; Moreira-Santos, M.; Ribeiro, R. A short-term
dae). Hydrobiologia 2000, 427 (1), 75−82. sublethal toxicity assay with zebra fish based on preying rate and its
(34) Béchard, K.; Gillis, P.; Wood, C. Acute toxicity of waterborne Cd, integration with mortality. Chemosphere 2015, 120, 568−574.
Cu, Pb, Ni, and Zn to first-instar Chironomus riparius larvae. Arch. (53) Meregalli, G.; Ollevier, F. Exposure of Chironomus riparius larvae
Environ. Contam. Toxicol. 2008, 54 (3), 454−459. to 17α-ethynylestradiol: Effects on survival and mouthpart deformities.
(35) Burton, G. A.; Norberg-King, T. J.; Benoit, D. A.; Ankley, G. T.; Sci. Total Environ. 2001, 269 (1−3), 157−161.
Winger, P. V.; Kubitz, J.; Lazorchak, J. M.; Smith, M. E.; Call, D. J.; (54) Meregalli, G.; Pluymers, L.; Ollevier, F. Induction of mouthpart
Ingersoll, C. G. Interlaboratory study of precision: Hyalella azteca and deformities in Chironomus riparius larvae exposed to 4-n-nonylphenol.
Chironomus tentans freshwater sediment toxicity assays. Environ. Toxicol. Environ. Pollut. 2000, 111 (2), 241−246.
Chem. 1996, 15 (8), 1335−1343. (55) Bird, G. A.; Schwartz, W. J.; Joseph, D. L. The effect of 210Pb and
(36) Palmer, A. R. Quasireplication and the contract of error: Lessons stable lead on the induction of menta deformities in Chironomus tentans
from sex ratios, heritabilities and fluctuating asymmetry. Annu. Rev. Ecol. larvae and on their growth and survival. Environ. Toxicol. Chem. 1995, 14
Syst. 2000, 31, 441−480. (12), 2125−2130.
(37) Palmer, A. R. Detecting publication bias in meta-analyses: a case (56) Janssens De Bisthoven, L.; Vermeulen, A.; Ollevier, F.
study of fluctuating asymmetry and sexual selection. Am. Nat. 1999, 154 Experimental induction of morphological deformities in Chironomus
(2), 220−233. riparius larvae by chronic exposure to copper and lead. Arch. Environ.
(38) Woodward, M. Epidemiology: Study Design and Data Analysis; Contam. Toxicol. 1998, 35 (2), 248−256.
CRC press: Boca Raton, FL, 2013. (57) Bonfanti, P.; Colombo, A.; Orsi, F.; Nizzetto, I.; Andrioletti, M.;
(39) Egger, M.; Smith, G. D.; Schneider, M.; Minder, C. Bias in meta- Bacchetta, R.; Mantecca, P.; Fascio, U.; Vailati, G.; Vismara, C.
analysis detected by a simple, graphical test. BMJ. 1997, 315 (7109), Comparative teratogenicity of chlorpyrifos and malathion on Xenopus
629−634. laevis development. Aquat. Toxicol. 2004, 70 (3), 189−200.
(40) R Core Development Team. R: A Language and Environment for (58) Greulich, K.; Pflugmacher, S. Uptake and effects on detoxication
Statistical Computing. R Foundation for Statistical Computing, Vienna, enzymes of cypermethrin in embryos and tadpoles of amphibians. Arch.
Austria, 2015; ISBN 3-900051-07-0, http://www.R-project.org. Environ. Contam. Toxicol. 2004, 47 (4), 489−495.
(41) Vermeulen, A. C.; Dall, P.; Lindegaard, C.; Olliver, F.; Goddeeris, (59) Ankley, G. T.; Diamond, S. A.; Tietge, J. E.; Holcombe, G. W.;
B. R. Exposure of Chironomus riparius larvae (Diptera) to lead, mercury Jensen, K. M.; DeFoe, D. L.; Peterson, R. Assessment of the risk of solar
and ß-sitosterol: Effect on mouthpart deformation and moulting. ultraviolet radiation to amphibians. I. Dose-dependent induction of
Chemosphere 2000, 41, 45−55. hindlimb malformations in the northern leopard frog (Rana pipiens).
(42) Rosenthal, R., Parametric measures of effect size. In The Environ. Sci. Technol. 2002, 36 (13), 2853−2858.
Handbook of Research Synthesis; Cooper, H.; Hedges, L., Eds.; Russel (60) Arambourou, H.; Gismondi, E.; Branchu, P.; Beisel, J. N.
Sage Foundation, 1994; pp 231−244. Biochemical and morphological responses in Chironomus riparius
(43) Arambourou, H.; Beisel, J.-N.; Branchu, P.; Debat, V. Exposure to (Diptera, Chironomidae) larvae exposed to lead-spiked sediment.
sediments from polluted rivers has limited phenotypic effects on larvae Environ. Toxicol. Chem. 2013, 32 (11), 2558−2564.
and adults of Chironomus riparius. Sci. Total Environ. 2014, 484, 92−101. (61) Kosalwat, P.; Knight, A. W. Chronic toxicity of copper to a partial
(44) Langer-Jaesrich, M.; Koehler, H. R.; Gerhardt, A. Can mouth part life-cycle of the midge. Arch. Environ. Contam. Toxicol. 1987, 16 (3),
deformities of Chironomus riparius serve as indicators for water and 283−290.
sediment pollution? A laboratory approach. J. Soils Sediments 2010, 10 (62) Romanenko, V. D.; Goncharova, M. T.; Konovets, I. M.; Kipnis, L.
(3), 414−422. S. Morphophysiological and cytogenetic modifications in Chironomus
(45) Lewis, J. W.; Morley, N. J.; Ahmad, M.; Challis, G. L.; Wright, R.; riparius (Diptera: Chironomidae) under the impact of copper ions.
Bicker, R.; Moffitt, D. Structural changes in freshwater fish and Hydrobiol. J. 2012, 48 (6), 86−94.
chironomids exposed to bacterial exotoxins. Ecotoxicol. Environ. Saf. (63) Wise, R. R.; Pierstorff, C. A.; Nelson, S. L.; Bursek, R. M.; Plude, J.
2012, 80, 37−44. L.; McNello, M.; Hein, J. Morphological deformities in Chironomus
(46) Martinez, E. A.; Moore, B. C.; Schaumloffel, J.; Dasgupta, N. (Chironomidae: Diptera) larvae as indicators of pollution in Lake
Winnebago, Wisconsin. J. Great Lakes Res. 2001, 27 (4), 503−509.
Effects of exposure to a combination of zinc- and lead-spiked sediments
on mouthpart development and growth in Chironomus tentans. Environ.
Toxicol. Chem. 2004, 23 (3), 662−667.
(47) Janssens de Bisthoven, L. J., Biomonitoring with morphological
deformities in aquatic organisms. In Biomonitoring of Polluted Water -
Reviews on Actual Topics, Gerhardt, A., Ed.; Trans Tech Publications:
Zurich, 2000; Vol. 9, pp 65−93.
(48) Calabrese, E. J. Paradigm lost, paradigm found: the re-emergence
of hormesis as a fundamental dose response model in the toxicological
sciences. Environ. Pollut. 2005, 138 (3), 378−411.
(49) Janssens de Bisthoven, L.; Huysmans, C.; Ollevier, F. The in situ
relationships between sediment concentrations of micropollutants and
morphological deformities in Chironomus gr. thummi larvae (Diptera,
Chironomidae) from lowland rivers (Belgium): a spatial comparison. In
Chironomids: from genes to ecosystems; Cranston, P. S., Ed.; CSIRO: East
Melbourne, 1995; pp 63−80.

I DOI: 10.1021/acs.est.6b04020
Environ. Sci. Technol. XXXX, XXX, XXX−XXX