Post on 30-Dec-2015
Effects of pollution on reproductive behaviourof ®shes
JACKIE C. JONES and JOHN D. REYNOLDS�
School of Biological Sciences, University of East Anglia, Norwich, NR4 7TJ, United Kingdom
Contents
Abstract page 463
Introduction 464
Fish behaviour in pollution research 465
Case study 1: mosquito®sh and paper mill ef¯uent
Case study 2: sun®sh and herbicide
Comparisons between case studies
Review of studies 480
Motivation for research
Selection of study species and pollutant
Laboratory or wild
Emphasis on reproductive behaviour
Effects on reproductive success
Experimental designs
Secondary effects of pollution 483
Prospects for the future: integrating pollution and reproductive behaviour 483
From individual behaviour to populations
Effects on future reproductive success
Population size
Genetic changes
Adaptation
Uses of reproductive behaviour studies
Behavioural assays for protecting other species and habitats
Conservation
Conclusion 488
Acknowledgements 488
References 488
Abstract
This review attempts to integrate pollution research with behavioural ecology by focusingon reproductive behaviour of ®shes. A search of Aquatic Sciences and FisheriesAbstracts and other sources showed that only 0.1% of 19 199 studies of aquatic pollution
Reviews in Fish Biology and Fisheries 7, 463±491 (1997)
0960±3166 # 1997 Chapman & Hall
�Author to whom correspondence should be addressed (e-mail: Reynolds@uea.ac.uk).
and ®shes during the past 20 years have made this link. Effects on parental care andcourtship have been investigated using a variety of pollutants (e.g. acidi®cation,herbicide, thermal ef¯uent) in several ®sh families (e.g. Cichlidae, Poeciliidae,Gasterosteidae, Cyprinidae).
Eleven of the 19 studies found a change in behaviour from the norm. Effects oncourtship included decreases or increases in frequency of displays, increased courtshipduration, or performance of male-like behaviour by masculinized females. Studies ofparental care have found decreased nest-building activity, decreased offspring defence,or changes in division of parental care between the sexes. Few studies have measuredreproductive success or extrapolated their results to effects on populations.
We develop a framework for exploring links between pollution and behavioural ecologywhich suggests potential impacts on life history trade-offs in reproduction, genetic changesin populations, and population sizes. Many reproductive behaviours of ®sh species arereadily quanti®able and behaviours such as courtship by male guppies and other membersof the Poeciliidae show some promise for pollution monitoring and behavioural toxicitytests. Choice of such assays would have to compete with the sensitivity and practicality ofmore traditional methods but may serve as useful complements. There is considerablescope for further research into conservation. A synthesis between behavioural ecologyand toxicology should thus provide useful insights for both ®elds.
Introduction
Evidence of the state of many major British rivers during the reign of George III (1760±1820) was made apparent by a letter reportedly written by a member of Parliament to thePrime Minister. The letter complained about the odour and appearance of the RiverThames. It was written not in ink, but with water from the Thames itself (Strandberg,1971). Historically, the Thames had been a major salmon river but towards the end of the18th century, ®sh species had declined dramatically and large areas of the river becamedevoid of ®sh altogether (Wood, 1982). This was typical of many rivers in heavilypopulated areas of Europe.
Fish populations are vulnerable because the aquatic environment is the recipient ofvirtually every form of human waste (Moyle and Leidy, 1992). The IUCN (InternationalUnion for Conservation of Nature) Red List (1996) records 734 ®sh species asthreatened and 92 species as extinct worldwide. Water pollution is one of severalcontributors to such declines in ®sh populations (Clark, 1992; Moyle and Leidy, 1992;Lawton and May, 1995; Maitland, 1995).
Pollution can be de®ned as `̀ the presence in the environment, or the introduction intoit, of products of human activity which have harmful or objectionable effects'' (OxfordEnglish Dictionary, CD-ROM). Studies of ®shes have ®gured prominently in pollutionresearch, particularly in sublethal physiological effects (review: Kime, 1995). Manystudies focus on reproduction because this is one of the most vulnerable periods in thelife cycle of ®shes (Gerking, 1980; Little et al., 1985).
Recently there has been development towards the use of behaviour in toxicologicalresearch (Little et al., 1985; Dùving, 1991; Smith and Logan, 1997). This represents afusion of the ®elds of behaviour, ecology, toxicology and conservation biology. In thisarticle, we present what might be termed a `prospective review'. The `review' elementconcerns studies that have examined impacts of pollutants on reproductive behaviour of
464 Jones and Reynolds
®shes. The `prospects' that we are interested in concern the value of integratingpollution research into the ®eld of behavioural ecology for gaining insights into waterquality management and conservation of ®shes.
Fish behaviour in pollution research
Before focusing on reproductive behaviour speci®cally, it is instructive to consider theoverall role of ®sh behaviour in pollution research. We searched Aquatic Sciences andFisheries Abstracts (ASFA) on CD-ROM from 1978 to 1997 for references containingvarious combinations of truncated search terms such as `behav�', to include variousspellings and forms of the key words (e.g. `behave', `behaviour', `behavioral'). The result(Fig. 1) shows that only 13.8% of studies involving ®sh and pollution included behaviourand of these, only 1% involved reproductive behaviour as opposed to other behaviourssuch as attraction and avoidance. Thus, of the original 19 199 studies of aquatic pollutionand ®shes, only 0.14% involve reproductive behaviour. This con®rms the impressionfrom previous reviews of the paucity of such studies (Atchison et al., 1987; Beitinger,1990; Dùving, 1991; Scherer, 1992).
The lack of studies of reproductive behaviour in pollution research can be attributedin part to a perception that behaviour is dif®cult to measure and interpret. This makesbehavioural assays less appealing than lethal toxicity tests; death is a rather de®nitiveendpoint! However, behavioural responses may provide more sensitive early warningsthan standard test methods (Smith and Logan, 1997), and ethologists have been makingaccurate measurements and analyses of many forms of animal behaviour for over 50
Reproduc*Behav*
Feed*Behav*
Predat*or Prey
Avoid* orAttract*
Behav*
(No Extras)
0.1 1 10 100% References With Pollut*, Fish* and...
26
152
576
366
2641
19199
Fig. 1. Results of a search of the Aquatic Sciences and Fisheries Abstracts CD-ROM from 1978 to
1997 for references on pollution and various types of ®sh behaviour. Numbers beside each bar show
the number of studies found for that category.
Effects of pollution on reproductive behaviour 465
years. Furthermore, during the past 20±30 years there has been a vigorous developmentof the ®eld of behavioural ecology, which is concerned with how behaviour affects lifehistories, reproductive success of individuals and their interactions with the surroundingenvironment (Krebs and Davies, 1997). In linking the reproductive behaviour ofindividuals to their environment, one can explore how stressful conditions such as thosecaused by pollution will affect offspring production, survival, and population dynamics.
We begin by reviewing the studies that have integrated the ®elds of reproductivebehaviour and pollution research. The behaviours considered include courtship,spawning interactions and parental care. Studies that investigated effects of pollutionon fecundity or reproductive success, or presence±absence of spawning only, withoutmeasuring associated reproductive behaviours, are not included in this review.
Based on the above ASFA search and an additional literature survey, 19 studies werefound that investigated the effects of pollution on reproductive behaviour (Table 1).This total excludes a few studies revealed by the initial ASFA keywords that proved notto be relevant to this review. Courtship was studied most often, followed by parentalcare. The studies varied widely in selection of ®sh species and pollutant, and in therelative contribution that reproductive behaviour made to the research.
Eleven of the 19 studies found a change in reproductive behaviour from the norm. Itshould be borne in mind that there may be a publication bias towards signi®cant effectsof pollutants. Studies investigating parental care have found either no effect, decreasednest-building or egg-directed activity, decreased protection of the young, disruption tocare or changes in division of parental care between the sexes. Effects on courtshipranged from no effect to either decreases or increases in display frequency, increasedduration of courtship and performance of male-like display behaviour by masculinizedfemales. Additional effects reported, but without strong supporting evidence, aredescribed in Table 1. Two contrasting cases from Table 1 illustrate the range ofapproaches used.
CASE STUDY 1: MOSQUITOFISH AND PAPER MILL EFFLUENT
Howell et al. (1980) investigated the impact of paper mill ef¯uent on a local populationof mosquito®sh (Gambusia af®nis holbrooki, Poeciliidae), in Elevenmile Creek,Escambia County, Florida, USA. This research was precipitated by observations thatsome females in the study area exhibited morphological signs of masculinization. Manyfemales had developed a gonopodium, the modi®ed anal ®n used by males for internalfertilization in this family of ®shes (Figs 2 and 3). Fish from above and below theef¯uent out¯ow were compared using histological examinations of gamete abnormalitiesand karyotypes. Comparisons of reproductive behaviour between the two ®sh groupswere also made in the laboratory. Fish were observed in each of the followingcombinations: (1) three pregnant masculinized females and three normal females, (2) twomasculinized females, (3) a precociously masculinized male and a normal female, (4) aprecociously masculinized male, a normal male and three normal females, and (5) anormal male and a normal female. No quantitative data were recorded, but descriptionsof morphology and reproductive behaviour were provided.
Howell et al. (1980) showed that ®sh from above the ef¯uent discharge exhibitednormal sexual expression. However, all females 13±45 mm standard length, collectedfrom below the ef¯uent out¯ow, showed some degree of gonopodial development. Therewas no evidence of testicular tissue, and heteromorphic sex chromosomes were absent.
466 Jones and Reynolds
Table 1. Summary of effects of pollution on ®sh reproductive behaviour
Behavioural
category
Species Pollutant and
sublethal level
used
Effect on reproductive
behaviour
Effect on reproductive output Comments Reference
Shown Extrapolated
Courtship Guppy
Poecilia
reticulata
Phenol.
10 mg lÿ1 for 10
days.
Decreased male court-
ship, e.g. approaches
to females and sigmoid
displays.
Not
determined.
Probably
disruptive
because sexual
selection and
elaborate
courtship are
prominent in
this species.
Unclear if phenol expo-
sure affects females and
their reproductive out-
put.
Colgan et
al. (1982)
Courtship Guppy
Poecilia
reticulata
Wastewater,
lindane.
10% sewage
plant water at
stage of release
to river, or
1 ìg lÿ1 lindane.
Both exposures
over observation
period only (2±
3 days).
Authors state that a
difference between two
laboratory strains in
male courtship displays
decreases. A difference
in gonopodial thrusting
increases under both
pollutants.
Not
determined.
A reduction of
®tness by one
laboratory
strain relative
to the other
due to the
pollutants
would be
expected.
Only 3 pairs of males
were tested with lindane,
and 9 pairs with waste-
water. Statistics erro-
neously used multiple
observations of same
®sh as independent data
points (hence sample
sizes of 49 and 153
respectively). Time was
confounded with treat-
ments. Interesting results
inconclusive.
SchroÈder
and Peters
(1988a)
continued overleaf
Effects
of
pollu
tion
on
repro
ductive
beh
avio
ur
467
Table 1. (continued)
Behavioural
category
Species Pollutant and
sublethal level
used
Effect on reproductive
behaviour
Effect on reproductive output Comments Reference
Shown Extrapolated
Courtship Guppy
Poecilia
reticulata
Wastewater.
10% sewage
plant water at
stage of release
to river. Expo-
sure was for
observation
period only
(3 days).
Authors state that
males of the `Macula-
tus' strain increased
gonopodial thrusting
and `closed' courtship
display, but had
decreased `open'
displays. Males of
`Blue Iridescence'
strain increased
thrusting.
`Maculatus'
males sired a
larger propor-
tion of young
when in com-
petition with
`Blue Irides-
cence' males
in wastewater
than in tap-
water.
None
suggested.
Only 2 pairs of males
used (one of each strain
in a pair) in each treat-
ment (wastewater or
tapwater). Multiple ob-
servations of the same
®sh arti®cially in¯ated
sample sizes on beha-
viour to 16 for each
treatment. Findings
inconclusive.
SchroÈder
and Peters
(1988b)
Courtship Fathead
minnow
Pimephales
promelas
Selenium.
20 mg lÿ1 and
30 mg lÿ1 for 24
hours.
No signi®cant differ-
ence in frequencies of
behaviours (approach,
leading, lateral display,
tail beating, vibrating
and butting) between
exposed males and
control males.
None due to
behaviour, but
many larvae
were born
deformed and
died.
None
suggested.
Fish died at higher con-
centrations (60 mg lÿ1).
Sample sizes unclear,
apparently 5 pairs of
exposed ®sh per treat-
ment. Uncontaminated
water used for breeding
tests.
Pyron and
Beitinger
(1989)
468
Jones
and
Reyn
old
s
Courtship Orange
chromide
Etroplus
maculatus
Monochloramine
Either 0.025 or
0.05 mg lÿ1 from
within 24 hours
of pair being
placed together
until spawning.
Little courtship and no
spawning occurred at
0.05 mg lÿ1, with ®sh
dying or becoming ill.
Comparisons between
0.025 mg lÿ1 and con-
trol showed: 64% in-
crease in length of
courtship period, 34%
reduction in clutch
size, increased ®n-
¯icker by males,
reduced quivering
and ®n-¯icker by
females.
No spawning
at higher
concentration.
Hatching data
not given.
Increased
courtship
length may
increase preda-
tion risk. An
additional be-
haviour which
increased was
gill-purging.
This might be
misinterpreted
by conspeci®es
as aggression
and hamper
pair formation.
A comprehensive study,
primarily aimed at
determining suitability
of behavioural toxicity
for pollution monitoring.
Stafford
and Ward
(1983)
Courtship Mosquito®sh
Gambusia
af®nis
holbrooki
Paper mill
ef¯uent.
Fish collected
6.5 km down-
stream of the
ef¯uent output.
Exposed females
became masculinized.
Masculinized pregnant
females displayed only
typical male reproduc-
tive behaviour (chasing
normal and masculi-
nized females with
gonopodial swings and
thrust attempts). Ex-
posed males matured
precociously and
exhibited typical but
more aggressive court-
ship. A masculinized
male was dominant
over a normal male.
Not
determined.
None
suggested.
No data or analyses
presented. One replicate
for each combination
involving 1±3 masculi-
nized or normal ®sh.
Interesting ®ndings from
wild-exposed ®sh merit
further investigation.
Howell
et al. (1980)
continued overleaf
Effects
of
pollu
tion
on
repro
ductive
beh
avio
ur
469
Table 1. (continued)
Behavioural
category
Species Pollutant and
sublethal level
used
Effect on reproductive
behaviour
Effect on reproductive output Comments Reference
Shown Extrapolated
Courtship
and
aggression
Mosquito®sh
Gambusia
af®nis
Kraft mill
ef¯uent.
Exposure in
wild 1±5 km
downstream of
discharge.
No signi®cant differ-
ence for either sex in
any behaviour (ap-
proach, chase, display,
thrust and penetrate)
between exposed and
control ®sh. Exposed
females had become
masculinized morpho-
logically.
Not
determined.
None
suggested.
Further work
was recom-
mended.
Fish had been in clean
water in captivity for 2±
4 weeks before testing.
Results might have been
different if ®sh had been
held in water from con-
taminated site. Good
number of replicates
used (20 per ®sh com-
bination).
Bortone et
al. (1989)
Courtship
and
aggression
Mosquito-
®sh
Gambusia
af®nis
af®nis
Simulated mas-
culinizing effect
of kraft mill
ef¯uent.
Degraded pro-
ducts of 65%
stigmastanol and
30% B-sitosterol
for one month.
Exposed females
became masculinized
and showed more
male-like courtship
than control females in
the presence of a
female. These beha-
viours were less fre-
quent than in normal
males. When two
masculinized females
were placed together,
the smaller one acted
as a male, and the
larger as a female.
Masculinized females
did not have increased
aggression.
Not
determined.
Detrimental
effects prob-
ably minimal
in an exposed
population be-
cause masculi-
nized females
reverted to
normal female
behaviour in
the presence of
a normal male,
and could still
reproduce.
It would be interesting
to examine longer-term
effects on both sexes.
Krotzer
(1990)
470
Jones
and
Reyn
old
s
Male
courtship
Cichlid
Sarothero-
don
mossambi-
cus
Endosulfan
insecticide.
Three treat-
ments: (A)
adults exposed
to 0.5 ìg lÿ1 for
4 weeks;
(B) juveniles
exposed to
0.6 ìg lÿ1 for 9
weeks (with
1.5 ìg lÿ1 for
®rst 3 days);
(C) juveniles
exposed to
either 0.6 or
0.2 ìg lÿ1 for 6
weeks.
(A) No effect on adult
male sexual behaviour
of female clutch
production or mouth-
brooding time.
(B) Onset of juvenile
male sexual behaviour
and female clutch
production delayed.
Females had higher
clutch production rates
but lower retention
times than controls,
suggesting that clutch
abortion was
occurring.
(C) More control
males showed normal
breeding behaviour
between days 36 and
42 than exposed males.
Many fry died
at low expo-
sure concentra-
tion (24 hour
LC50 was
0.5 ìg lÿ1).
Delayed breed-
ing displays in
newly matur-
ing males may
explain a
reduction of
approximately
75% in nests
of related
Tilapia ren-
dalli in the
Okavango
Delta,
Botswana.
Concentrations were
similar to those found in
the wild soon after
spraying. Additional
direct physiological
effects noted on ®sh.
Matthiessen
and Logan
(1984)
continued overleaf
Effects
of
pollu
tion
on
repro
ductive
beh
avio
ur
471
Table 1. (continued)
Behavioural
category
Species Pollutant and
sublethal level
used
Effect on reproductive
behaviour
Effect on reproductive output Comments Reference
Shown Extrapolated
Nest
building
Cichlid
Tilapia
rendalli
Endosulfan.
Application of
9.5 g haÿ1
sprayed onto the
river by aircraft.
The density of Tilapia
nests was about 75%
less in sprayed areas
than in unsprayed
areas.
Juveniles were
less numerous
in a sprayed
lagoon com-
pared with
control.
The 1977 year-
class ®sh were
less abundant
in the areas
sprayed in that
year than in
those areas not
sprayed,
implying fewer
®sh were pro-
duced. As long
as the same
sites are not
sprayed an-
nually, then
®sh population
should not be
reduced per-
manently.
Although effects on the
reproductive success
were only inferred from
the ®ndings, this was a
thorough study investi-
gating all aspects of
effects on ®sh, from
mortality to behaviour
and populations.
Douthwaite
et al.
(1981)
Parental
care
Threespine
stickleback
Gasterosteus
aculeatus
Thermal ef¯uent.
28±30 8C.
Disruption of parental
behaviour occurred.
May cause males to
eat own eggs or aban-
don nest.
Eggs died,
apparently due
to disruption
of parental
behaviour.
None
suggested.
This paper is an abstract
from a meeting. Insuf®-
cient detail to evaluate
it.
Ryabov
(1985)
472
Jones
and
Reyn
old
s
Parental
care
Naked goby
Gobiosoma
bosc
Episodic
hypoxia.
Field; down to
,0.5 mg lÿ1
Laboratory;
down to
0.15 mg lÿ1
Males guarded their
eggs and nests until
the dissolved oxygen
level reached almost
lethal levels (0.15±
0.6 mg lÿ1), when they
abandoned.
The eggs died
if males left
the nest. If the
male does not
leave the nest
when lethal
limits are
being reached
he will die.
Even if there
are no eggs in
the nest, the
male will lose
mating oppor-
tunities if he
abandons
unnecessarily.
Thorough study, unusual
in incorporating both
®eld observations and
laboratory experiments.
Breitburg
(1992)
Parental
care
Bluegill
sun®sh
Lepomis
macrochirus
Redear
sun®sh
Lepomis
microlophus
Herbicides
2,4-dichlorophe-
noxyacetic acid
and Aquathol-K
sprayed on sur-
face of lake to
obtain a con-
centration of
4 mg lÿ1.
Sprayed water
was used as a
control.
Although some males
abandoned nests tem-
porarily in each treat-
ment, there was no
signi®cant difference
between them. Conge-
ners usually intruded
during male absence
and some egg preda-
tion was observed in
all treatments. There
was no signi®cant ef-
fect on male parental
care (e.g. displays to-
wards intruding males,
egg fanning).
Not
determined.
None expected
from herbicide
exposure.
Sample sizes were low
(5±9 males per treat-
ment). A possible trend
towards higher rates of
temporary desertion in
both species and conse-
quently effects on egg
predation might prove
signi®cant with larger
sample sizes.
Bettoli and
Clark
(1992)
Parental
care
Fathead
minnow
Pimephales
promelas
Waterborne lead.
0.5 mg lÿ1 lead
for 30 days
while ®sh
spawned.
Males exposed to lead
spent signi®cantly less
time in activities
directed towards ceil-
ing of nest, including
preparation of sub-
strate for eggs, and
touching ceiling with
dorsal pad.
Drastic
decrease in
egg production
by exposed fe-
males. Role of
reduced paren-
tal behaviour
in reproductive
output unclear.
Reproductive
success may
be reduced
due to less
cleaning of
nest ceiling
and egg mass
surface.
Dif®cult to interpret the
cause of reduced nest
maintenance behaviours,
i.e. an effect of lead
exposure versus fewer
eggs received.
Weber
(1993)
continued overleaf
Effects
of
pollu
tion
on
repro
ductive
beh
avio
ur
473
Table 1. (continued)
Behavioural
category
Species Pollutant and
sublethal level
used
Effect on reproductive
behaviour
Effect on reproductive output Comments Reference
Shown Extrapolated
Parental
care and
aggression
Convict
cichlid
Cichlasoma
nigrofascia-
tum
Acid stress
pH 5±5.5 versus
control pH 7.
Administered to
pair-bonded ®sh
24 hours after
introduction to
breeding tank.
Continued expo-
sure.
Experiments were car-
ried out in the pre-
sence of conspeci®cs.
Egg fanning time was
reduced by 25% due to
a decrease in female
fanning activity, which
was only partly
compensated by an
increase in male fan-
ning. Females
increased their guard-
ing time, while this
decreased in males.
Males increased their
aggressive behaviours.
Not
determined.
Acidication in
the natural
environment
could reduce
survival of fry
due to changes
in parental
behaviour.
Experiments capitalize
effectively on the rich
repertoire of behaviours
by both sexes in this
species.
Lorenz and
Taylor
(1992a)
Parental
care
Convict
cichlid
Cichlasoma
nigrofascia-
tum
Acid stress.
pH 5±5.5 versus
control pH 7.
Administered to
pair-bonded ®sh
24 hours after
introduction to
breeding tank.
Continued expo-
sure until fry 5
days old.
Survival of fry unaf-
fected by pH alone.
However, conspeci®cs
preyed upon more fry
at low pH compared
with control pH. Total
number of displays by
both parents towards
predators increased at
low pH, and there was
a trend towards fewer
attacks.
Decreased per
cent fry survi-
val after day 5
of free-swim-
ming stage
under acid
stress when in
the presence of
predators.
Switching
from attacks to
displays may
conserve
energy of
parents, but
this is not
effective
against
predators.
Shows the importance of
ecological context (i.e.
predation risk) for
understanding sublethal
effects of pollution on
®tness.
Lorenz and
Taylor
(1992b)
474
Jones
and
Reyn
old
s
Parental
care
Bluegill
sun®sh
Lepomis
macro-
chirus
Di¯ubenzuron; a
pesticide
2.5 ìg lÿ1 and
30 ìg lÿ1
applied twice,
®rst application
into the desig-
nated ®sh enclo-
sures 2 months
after ®sh intro-
duced and sec-
ond application
1 month later.
No effects on male
reproductive behaviour
observed.
None due to
behaviour.
Only one
spawning event
was observed
in all exposure
treatments
(30 ìg lÿ1) but
most spawning
events (86%)
took place
prior to expo-
sure. Spawning
appeared to be
more in¯u-
enced by water
temperature
than di¯uben-
zuron expo-
sure. Reduced
juvenile
growth rates.
Reduced juve-
nile growth
could have
major impoli-
cations for ®sh
survival during
winter through
starvation and
increased pre-
dation risk.
No statistical analysis
was performed and only
2 replicates per treat-
ment used. Reproductive
behaviour effects not
quanti®ed and hatching
success based on one
spawning event at
30 ìg lÿ1. Most spawn-
ing events took place
before exposure. How-
ever, focused on ecolo-
gical context of effect of
pesticide exposure on
natural food supply, and
effects on reproductive
behaviour were not in-
tended as a main focus.
Tanner and
Hoffett
(1995)
continued overleaf
Effects
of
pollu
tion
on
repro
ductive
beh
avio
ur
475
Table 1. (continued)
Behavioural
category
Species Pollutant and
sublethal level
used
Effect on reproductive
behaviour
Effect on reproductive output Comments Reference
Shown Extrapolated
Parental
care
Bluegill
sun®sh
Lepomis
macro-
chirus
Azinphos-
methyl: a
pesticide.
1 ìg lÿ1 and
4 ìg lÿ1 7 weeks
after ®sh
introduced.
Effect on reproductive
behaviour unclear be-
cause most spawning
took place prior to
exposure.
None due to
behaviour.
Although
spawning
caused in three
of the four
ìg lÿ1 repli-
cates after
exposure, it
also ceased in
three of the
four control
replicates.
None
suggested.
Statistical analyses used
for larval growth and
biomass data. Most
spawning events took
place prior to exposure.
Reproductive behaviour
effects not quanti®ed but
played a minor part.
Data from before and
after pesticide applica-
tion were used to calcu-
late each average
hatching success. The
study took place in lit-
toral enclosures and in-
vestigated effects on
reproductive success
from spawning to juve-
nile production.
Tanner and
Knuth
(1995)
476
Jones
and
Reyn
old
s
Parental
care
Bluegill
sun®sh
Lepomis
macro-
chirus
Esfenvalerate: a
pesticide
0.01, 0.08, 0.2,
1.0 and 5 ìg lÿ1
applied twice,
®rst application
1 month after
®sh introduced
and second ap-
plication another
month later.
At 1 ìg lÿ1, one male
guarding a nest with
embryos rim-circled
rapidly. In all enclo-
sures except 5 ìg lÿ1,
where the adults died,
rim-circling occurred
both before and after
exposure.
None due to
behaviour.
Adults died at
5 ìg lÿ1.
Spawning
ceased for 15
days in
1 ìg lÿ1 enclo-
sure following
both applica-
tions and lar-
val mortality
was observed
following the
second appli-
cation. A
reduction in
juvenile
growth rate
was measured
at 1.0, 0.2 and
0.08 ìg lÿ1.
If esfenvalerate
were to be ap-
plied repeat-
edly biweekly
at concentra-
tions greater
than
0.44 ìg lÿ1,
spawning and
hence juvenile
production
would cease.
Reduced juve-
nile growth
could have
major implica-
tions for ®sh
survival over
winter through
starvation and
increased
predation risk.
No statistical analysis
was performed and 1
replicate per treatment
was used. Reproductive
behaviour effects not
quanti®ed but played a
minor role. The rapid
rim-circling that oc-
curred in the 1 ìg lÿ1
treatment was observed
in only one male. Com-
bined data from before
and after pesticide apli-
cation were used to
calculate average hatch-
ing success.
Tanner and
Knuth
(1996)
Effects
of
pollu
tion
on
repro
ductive
beh
avio
ur
477
Masculinized females exhibited some male reproductive behaviour and males exhibitedprecocious development of secondary sex characters and reproductive behaviour. Aphysiologically normal male and female when placed together exhibited normalbehaviour (see Table 1 for detailed behavioural ®ndings).
CASE STUDY 2: SUNFISH AND HERBICIDE
Bettoli and Clark (1992) investigated the effects of herbicide exposure on nestingbehaviour in bluegill sun®sh, Lepomis macrochirus and redear sun®sh, L. microlophus(Centrarchidae) in Old Monterey Lake, Monterey, Tennessee, USA. Underwater videocameras were set up near nests in the wild and ®sh were ®lmed prior to, during and afterherbicide application (Aquathol-K or 2,4-dichlorophenoxyacetic acid). Some of theredear sun®sh observations were also made from the lakeside using binoculars. Speci®cbehaviours such as display, spawning and egg fanning were recorded, along withincidence and timing of nest abandonment. Statistical comparisons were made betweenmales with nests in test plots sprayed with Aquathol-K, 2,4-D or with water (control).
There was no difference in the number of nests abandoned nor in the actual time ofabandonment between treatments (herbicide or water sprayed). When abandonment didoccur, congeners usually entered the nest to feed on eggs or fry until the male returned.Finally, frequencies of rim circling, egg fanning and agonistic behaviours of theguarding males did not differ between treatments. Bettoli and Clark (1992) concludedthat these herbicides, if applied properly, cause no pronounced shift in sun®shreproductive behaviour.
Fig. 2. Normal specimens of Gambusia af®nis holbrooki collected above paper-mill ef¯uent in
Elevenmile Creek, Escambia County, Florida (Howell et al., 1980; reproduced by permission of
Copeia). The top ®sh is a pregnant female (31 mm standard length). The bottom ®sh is a male
(25 mm standard length) showing gonopodial differentiation of the anal ®n.
478 Jones and Reynolds
COMPARISONS BETWEEN CASE STUDIES
These studies show the diversity of approaches to the study of effects of pollution on ®shreproductive behaviour. Whereas the mosquito®sh study focused primarily on morph-ology, with some unquanti®ed behavioural observations, the sun®sh study was restrictedentirely to reproductive behaviour. Many mosquito®sh were put into each observationtank and only one replicate of each ®sh combination was used. The sun®sh study usedbetween ®ve and nine individually sprayed areas per treatment to study effects ofherbicide exposure on ®sh in the wild. Owing to the nature of the data collected in the®rst study, no analysis could be undertaken, while in the second study a rigorous analysisof the data was performed. However, the mosquito®sh study was concerned primarilywith morphological effects, with behavioural observations comprising a small part of theresearch. Although the behavioural data were non-quantitative, they provide a startingpoint for further work.
Fig. 3. Masculinized specimens of Gambusia af®nis holbrooki collected below a paper-mill ef¯uent in
Elevenmile Creek, Escambia County, Florida (Howell et al., 1980; reproduced by permission of
Copeia). (A) Immature male (15 mm SL) showing precocious anal ray elongation. (B) Mature male
(17 mm SL) showing precocious gonopodium. (C) Pregnant female (23 mm SL) with a gonopodium.
(D) Pregnant female (28 mm SL) with a gonopodium. (E) Pregnant female (34 mm SL) with a
gonopodium. (F) pregnant female (36 mm SL) with a gonopodium.
Effects of pollution on reproductive behaviour 479
To obtain a clearer picture of effects of pollution on reproductive behaviour, weassess various aspects of the studies carried out so far.
Review of studies
MOTIVATION FOR RESEARCH
Some studies were aimed towards development of a behavioural assay for toxicity testing,based on changes in behaviour such as courtship (Stafford and Ward, 1983; SchroÈder andPeters, 1988a; Bortone et al., 1989). This information could be used in a similar vein tolethal toxicity tests, but with behavioural responses as an end point (Stafford and Ward,1983; Little et al., 1985; Bortone et al., 1989). SchroÈder and Peters (1988a) contendedthat as a behavioural assay, courtship by guppies (Poecilia reticulata, Poeciliidae) isextremely sensitive to very low concentrations of aquatic contaminants. Stafford andWard (1983) found orange chromides (Etroplus maculatus, Cichlidae) also to be asuitable test organism and proposed that behavioural assays are liable to be moresensitive measures of effects of toxicants than previous methods.
The aims of the other studies were less clear, but many were concerned with effectson reproductive behaviour and reproductive success (Howell et al., 1980; Matthiessenand Logan, 1984; Ryabov, 1985; SchroÈder and Peters, 1988b; Pyron and Beitinger,1989; Bettoli and Clark, 1992; Lorenz and Taylor, 1992a,b; Weber, 1993; Tanner andKnuth 1995, 1996; Tanner and Moffett, 1995). For example, Lorenz and Taylor (1992b)tested the hypothesis that if offspring survival is reduced in an unfavourableenvironment (in this case due to acid stress), then parents may reduce their care toconserve energy for future reproduction (Carlisle, 1982). This was supported by testswith convict cichlids (Cichlasoma nigrofasciatum, Cichlidae), which suffered a decreasein reproductive success. Some studies went further in hypothesizing that reducedreproductive output owing to changes in behaviour may have a detrimental effect onpopulation numbers (Douthwaite et al., 1981; Colgan et al., 1982; Krotzer, 1990;Breitburg, 1992). Krotzer (1990) studied the effect of masculinization on the behaviourof female mosquito®sh (Gambusia af®nis af®nis, Poeciliidae). It was suggested that iffemales became more aggressive and less receptive to males, exposed populations inthe wild could suffer. The primary aim of three studies of bluegill sun®sh by oneresearch group was to measure effects of pesticides on reproductive success fromspawning to juvenile production, with reproductive behaviour playing a very minor role(Tanner and Knuth, 1995, 1996; Tanner and Moffett, 1995). Although some studieswent into greater depth on the consequences of changes in behaviour than others, theinformation provided by most studies could be used as a starting point towards furtherresearch.
SELECTION OF STUDY SPECIES AND POLLUTANT
Studies varied widely in species and pollutant investigated. Most of those that focused oneffects on the reproductive behaviour of the ®sh chose a ®sh±pollutant combinationoccurring in the wild at a site of speci®c concern. Examples include ®sh±pollutantcombinations such as mosquito®sh exposed to paper mill or kraft mill ef¯uent (Howell etal., 1980; Bortone et al., 1989), cichlids (Tilapia rendalli, Cichlidae) exposed toendosulfan insecticide (Douthwaite et al., 1981), threespine sticklebacks (Gasterosteus
480 Jones and Reynolds
aculeatus, Gasterosteidae) exposed to power station thermal ef¯uent (Ryabov, 1985),naked gobies (Gobiosoma bosc, Gobiidae) exposed to hypoxic water conditions(Breitburg, 1992) and bluegill sun®sh exposed to herbicide contamination (Bettoli andClark, 1992). Although not all of these studies were carried out in the ®eld, suchcombinations were known to occur in the wild. This gives the ®ndings direct relevance tosuch affected areas.
Other criteria may affect selection of ®sh species. For example, Weber (1993) usedfathead minnows (Pimephales promelas, Cyprinidae) because this species is a standardtest species for use in aquatic toxicity studies (Denny, 1987). Fathead minnows werealso used by Pyron and Beitinger (1989) to investigate the in¯uence of seleniumexposure.
Other ®sh may be readily available, have well-documented reproductive behaviour,survive well in the laboratory, or breed readily. These were used speci®cally in thebehavioural toxicity tests. For example, the suitability of orange chromide cichlids fortoxicity tests was investigated under exposure to monochloramine (Stafford and Ward,1983) and guppies were exposed to treatment-plant wastewater (SchroÈder and Peters,1988a). However, these ®sh species may not encounter such pollutants in the wild.Researchers have thus sacri®ced direct relevance of the environmental situationnormally faced by the ®sh for the sake of practicality and standardization.
LABORATORY OR WILD
Studies have differed greatly in the origin of the ®sh used and the type of exposurereceived. Nine of the 19 studies in Table 1 used laboratory-bred and laboratory-exposed®sh (Colgan et al., 1982; Stafford and Ward, 1983; Matthiessen and Logan, 1984;SchroÈder and Peters, 1988a,b; Pyron and Beitinger, 1989; Lorenz and Taylor, 1992a,b;Weber, 1993). One study used wild mosquito®sh, exposed and observed in the laboratory(Krotzer, 1990) while three used hatchery-reared bluegill sun®sh and exposed them inoutdoor littoral enclosures (Tanner and Knuth, 1995, 1996; Tanner and Moffett, 1995).Five studies used wild ®sh that were exposed to the pollutant in the ®eld. Of these, threeobserved the ®sh in the wild: sun®sh, Lepomis macrochirus and L. microlophus (Bettoliand Clark, 1992), Tilapia rendalli (Douthwaite et al., 1981), and threespine stickleback(Ryabov, 1985). The other two observed mosquito®sh under non-polluted conditions inthe laboratory (Howell et al., 1980; Bortone et al., 1989) after exposure in the ®eld. Onlyone study used wild ®sh (naked gobies) in both wild and laboratory exposureobservations (Breitburg, 1992). Differences in the results of these studies could, in part,have arisen because of the varied background and exposure conditions of experimental®sh.
Although results may be less directly relevant to the wild, there is a bene®t in usinglaboratory ®sh strains compared with wild ®sh because one can eliminate the potentialeffects of tolerance through genetic adaptation or physiological acclimation. Toleranceto pollutants has been shown in several ®sh species, including mosquito®sh, Gambusiaholbrooki (Poeciliidae) (Benton et al., 1994), rainbow trout, Oncorhynchus mykiss(Salmonidae) (Pascoe and Beattie, 1979) and central mudminnow, Umbra limi(Umbridae) (Kopp et al., 1992), as well as in other taxa (Bryan and Langston, 1992;Gustavson and Wangberg, 1995). There is considerable scope for further research intodifferences among wild populations in pollution tolerance.
Effects of pollution on reproductive behaviour 481
EMPHASIS ON REPRODUCTIVE BEHAVIOUR
Some studies were purely behavioural (e.g. SchroÈder and Peters, 1988a; Pyron andBeitinger, 1989; Lorenz and Taylor, 1992a). Others contained combinations ofreproductive physiology, morphology and behaviour (e.g. Howell et al., 1980; Weber,1993) while others dedicated little to effects on reproductive behaviour (Tanner andKnuth, 1995, 1996; Tanner and Moffett, 1995). Usually, only a small part of thereproductive behaviour repertoire was studied, such as courtship, spawning behaviour,nest building, parental care of eggs, or defence of newly hatched fry. Ideally, the entirereproductive behaviour repertoire should be investigated (from mate choice and courtshipto egg or larval independence) to see at what stages pollution exerts an effect. Time andexpense will hinder such an approach, but if the purpose is to develop a toxicity test forwater quality management, full reproductive cycle studies could serve as an important®rst step, to be re®ned by simpler, focused behavioural assays.
EFFECTS ON REPRODUCTIVE SUCCESS
Eleven of the studies in Table 1 considered the possible effects of a change in reproductivebehaviour on reproductive success. Only three studies actually measured this directly: twofound a negative effect (Breitburg, 1992; Lorenz and Taylor, 1992b) and one reported anincrease in reproductive success (SchroÈder and Peters, 1988b). The latter result should betreated with caution owing to a small sample size and inappropriate experimental design(see Table 1 and general comments under `Experimental designs', below). A possibledecrease in reproductive success was inferred by Douthwaite et al. (1981), who measuredthe age distribution of juvenile Tilapia rendalli in the Okavango Delta, Botswana,comparing sites that had been sprayed with insecticide 2 years previously with controlsites. Although few studies have measured impacts on reproductive output via behaviouralchanges, in some cases it was likely that any such effects would have been overshadowedby physiological problems. For example, Weber (1993) found reduced egg production infemale fathead minnows upon exposure to lead. Stafford and Ward (1983) investigated theeffect of monochloramine on orange chromide. At levels of 0.05 mg lÿ1, some ®sh becameill and courtship and spawning stopped, while at levels of 0.025 mg lÿ1, female eggproduction was reduced. Finally, some studies estimated effects on reproductive successwithout trying to establish a causal link to changes in reproductive behaviour (Tanner andKnuth, 1995, 1996; Tanner and Moffett, 1995).
EXPERIMENTAL DESIGNS
In Table 1 most of our comments concerning experimental designs are rather critical. Infairness to the authors, we wish to emphasize two points strongly. First, many of thestudies were done before biologists were fully aware of the problems of non-independence of samples within treatments (`pseudoreplication', Hurlbert, 1984). Second,as noted earlier, a combination of physiological and behavioural effects of pollutants wasinvestigated in some studies, sometimes with effects on reproductive behaviour appearingas a secondary objective (e.g. Howell et al., 1980).
Nevertheless, to evaluate our current state of knowledge in this ®eld, it must be bornein mind that many of the studies lacked adequate controls, replication, or statisticalanalyses. Multiple observations of the same ®sh under the same treatment cannot becounted as separate statistical data points. One ®sh may be watched 10 times whenexposed to a pollutant and another ®sh may be watched 10 times in a control situation,
482 Jones and Reynolds
but effectively this still leaves one data point (e.g. mean behaviour per ®sh) under eachtreatment (Hurlbert, 1984). Furthermore, 10 ®sh in one aquarium exposed to a pollutantshould not be compared as 10 independent samples to ®sh in a control tank. Fishwithin aquaria are not independent because individuals may affect each other bothbehaviourally and physiologically. For example, there may be in¯uences from eachother's activity levels, dominance, pheromones, or metabolic wastes. These problemsapply to more traditional non-behavioural tests as well. Furthermore, aquaria are boundto differ by more than the presence or absence of the pollutant concerned. For example,levels of nutrients, light intensities, or other uncontrolled and unknown factors mayconfound treatments. This necessitates the use of multiple aquaria per treatment inrandomized or alternating spatial patterns. If multiple ®sh or observations are used peraquarium, each aquarium should yield a single mean data point in a t-test or standardANOVA design, or observations from individual ®sh within tanks can be used in a nestedANOVA-type design.
Our general impression from the information summarized in Table 1 is that many ofthese early studies report tantalizing results, but the use of ®sh reproductive behaviourin pollution research is still very much in its infancy.
Secondary effects of pollution
Additional studies are available in the literature which measure secondary effects ofpollutants on reproductive behaviour. For example, silting arises from high sedimentloading, and lowered oxygen levels often result from bacterial degradation under highorganic pollution (Clark, 1992; Mason, 1996). Although such studies were not includedin Table 1 because they did not use pollutants directly, their results could still be usefulin water quality management and conservation.
Silting was found to affect parental care in the ®fteenspine stickleback (Spinachiaspinachia, Gasterosteidae) (Potts et al., 1988). During parental care, males use their ®nsto fan water over their egg mass. Silting increased the number of times males fannedtheir eggs. Nest inspection and nest pushing also became more frequent. Loweredoxygen has also been found to affect paternal care in the threespine stickleback (Reebset al., 1984), with an increase in the length of egg-fanning bouts. When oxygen levelsfell below 2.8 mg lÿ1, males spent more time swimming outside the nest but still fannedat intervals. In the guppy, lowered oxygen levels increased the frequency of breathing atthe surface by males at a cost of decreased courtship (Kramer and Mehegan, 1981). Todate, such studies have not examined consequences for reproductive success.
Prospects for the future: integrating pollution and reproductive behaviour
There is now a growing interest in understanding sublethal effects of pollutants onaquatic organisms (Alabaster and Lloyd, 1982; MuÈller and Lloyd, 1994). To evaluate theusefulness of developing this ®eld further, it is helpful to integrate general theory frombehavioural ecology to clarify the biological links between behavioural responses ofindividuals and population changes. We can then use this information to help decide theprospects for using behavioural studies to help protect populations of the ®shesthemselves, as well as developing assays for protecting other species in the habitat.
Effects of pollution on reproductive behaviour 483
FROM INDIVIDUAL BEHAVIOUR TO POPULATIONS
Figure 4 is a schematic diagram of behavioural testing procedures, responses of theanimals, and potential uses for this information. Nearly all of the studies reviewed areconcerned with step 1, which links reproductive behaviour of ®shes and their exposure tosublethal doses of the contaminant in question. As we have seen, few of the studies havegone on to step 2, from reproductive behaviour to reproductive success, and these havebeen concerned only with current broods rather than impacts on subsequent reproductivebouts. We have not found any studies that take the `®nal' step (step 3) of testing forpopulation responses, although there was circumstantial evidence of a decrease in thenumbers of Tilapia rendalli in the Okavango Delta in two areas sprayed with endosulfaninsecticide (Douthwaite et al., 1981). Below, we discuss the logic and potential of theseunder-developed steps.
Effects on future reproductive success
Life history theory, supported by a large body of empirical research on a variety of taxa,suggests that commitment of resources to a given reproductive bout should have costs forsurvival, growth, or reproduction in subsequent bouts (Williams, 1966; Roff, 1992;Stearns, 1992; Sargent and Gross, 1993). If this were not true, animals could make hugeinvestments in reproduction forever (Partridge and Harvey, 1985). An example from®shes concerns male bluegill sun®sh, which have been induced to ventilate their eggs byfanning water across them more frequently when given larger broods experimentally(Coleman et al., 1985). Increased fanning reduces the time available to males forcourting new females, and it reduces energy reserves through the loss of non-polar lipids(Coleman and Fischer, 1991). This may impair survival and future reproduction. Suchreproductive trade-offs may be most evident under conditions of stress (Roff, 1992). Forexample, in the sand goby (Pomatoschistus minutus, Gobiidae), females that initiallyproduced large clutches before receiving a low food ration produced smaller clutches inthe next bout (Kvarnemo, 1997). This trade-off did not occur with high food. Thus,environmental stressors such as pollutants could have delayed effects on reproductionwhich would be overlooked if only a single reproductive cycle were studied.
Population size
Links between individual behaviour and population biology have been much sought-afterby ecologists (Sibly and Smith, 1985; Sutherland, 1996). In theory, any external force thatalters behaviour of individuals from the optimum under natural and sexual selectioncould lead to reduced population sizes. But this effect may be far from straightforward,depending on the nature of density dependence in the population. For example, Lorenzand Taylor (1992b) showed that under conditions of low pH, convict cichlids were lessable to rear their young in the presence of conspeci®cs. But if population densities ofthese ®sh were restricted by food or predation at a different stage of life, reduced survivalof young being guarded by parents may have no impact on population size. The dif®cultyof showing relationships between reproduction and population size is well known to®sheries biologists who rarely ®nd clear patterns between stock size and recruitment(Hilborn and Walters, 1992; King, 1995). Thus, if one is concerned with conservation, amajor challenge is to demonstrate clear effects of sublethal doses of pollutants onpopulation size.
484 Jones and Reynolds
EXPOSURE
ORGANISMRESPONSE
POPULATIONRESPONSE
Fish exposed to sublethal dose
Effect on reproductive behaviour?
Effect on current and futurereproductive success?
Effect on population?
Number of individualsGenetic changes
Adaptation
1
2
3
TESTS
Monitoring water quality
USES
4
5
6
7
Protection ofother speciesand habitat
Conservation ofthe test fish
species
Fig. 4. Schematic diagram showing hierarchy of pollution effects on reproductive behaviour and the potential uses for such information. Circled
numbers are reference points for discussion in the text.
Effects
of
pollu
tion
on
repro
ductive
beh
avio
ur
485
Genetic changes
Environmental impacts that change the reproductive behaviour of individuals (e.g. choiceof mate or oviposition site) could alter the genetic composition of populations. Anexample from a natural environmental parameter concerns the effects of light levels onbehaviour of Trinidadian guppies (Reynolds, 1993). Under low light levels, large-bodiedmales have higher mating success than smaller ones, because of female choice. Thispattern is reversed under higher light levels, when males may be hampered by greaterrisk of predation. Because male body size has a genetic basis (Reynolds and Gross,1992), such changes in mating behaviour in response to an environmental change mayhave genetic consequences for the population. Genetic changes such as these have yet tobe shown in the context of pollution, but given the numerous demonstrations ofheritability of traits under natural selection (Falconer, 1989) and sexual selection(Pomiankowski and Mùller, 1995; cf. Alatalo et al., 1997), researchers would be welladvised to bear these possibilities in mind.
Adaptation
Long-term selection may lead to the evolution of resistance. This depends on theintensity of selection, additive genetic variance in resistance and the population size.Benton et al. (1994) provided an example in a comparison of populations of freshwatersnails (Helisoma trivolvis, Planorbidae) and mosquito®sh from a relatively clean site anda site contaminated with a variety of heavy metals and other elements. They found that inthe populations exposed to the pollutants, both species may have evolved increasedtolerance. In mosquito®sh, this tolerant genotype seems to be linked to small body size.
Particularly relevant here is the possibility that altered mate choice, courtship orparental care could be adaptive, leading to a buffering against environmental change.This could occur either through short-term, facultative changes in behaviour or throughgenetic responses to selection. This is a risky proposition, because changes in behaviourcould be non-adaptive, and genetic adaptations may not evolve quickly enough.Furthermore, genetic bottlenecks and inbreeding are risks when the effective populationsize is reduced owing to fewer individuals reproducing (Meffe, 1986).
USES OF REPRODUCTIVE BEHAVIOUR STUDIES
Aside from gaining a better understanding of pollution in an ecological context, whatdirect practical uses can be made from the framework in Fig. 4? We believe the answersdepend on whether one is concerned with development of general assays for protectingwater quality for other species and the habitat, or with conservation of the particularspecies under study.
Behavioural assays for protecting other species and habitats
Toxicity tests may be used to screen new chemicals, and to formulate water qualitycriteria and standards (Council of the European Communities, 1978; Alabaster and Lloyd,1982; Abel, 1989; Lloyd, 1992; Howells, 1994). Standardized protocols for toxicity tests,for example LC50 tests using guppies or Daphnia magna (Daphniidae), or life-cycle testsusing salmonids, are often used (Alabaster and Lloyd, 1982; Lloyd, 1992; Howells, 1994;Mason, 1996). Toxicity tests are also used to assess hazards presented by ef¯uents, forexample from industry or sewage treatment works. Such monitoring of ef¯uent is
486 Jones and Reynolds
especially useful when substances are dif®cult to analyse or where a concoction ofsubstances is present (Alabaster and Lloyd, 1982).
Whether the objective is to monitor ef¯uent water quality (arrow 4 in Fig. 4) orformulate water quality criteria or standards (arrow 5), the ideal protocol involves aspecies that can be reared quickly and easily, with a minimum of equipment and space,and which responds quickly to the testing procedure, in a highly repeatable manner(Alabaster and Lloyd, 1982; Smith and Logan, 1997). Given these requirements, itseems unlikely that many of the ®sh species and reproductive behaviours reviewed inTable 1 will be practical for general use. We suggest that the species that come closestmight be poeciliids such as the guppy (already a standard test species for traditionaltoxicity tests) and mosquito®sh (Gambusia spp.). They are small and breed readily incaptivity, and male courtship is easily quanti®able and occurs under a wide variety ofcircumstances (Farr, 1980; SchroÈder and Peters, 1988a; Reynolds et al., 1993; Houde,1997). Rice®sh (Oryzias latipes, Cyprnodonitidae) and killi®sh (Fundulus, Cyprnodo-nitidae) are also worth exploring, although their courtship behaviour is less elaboratethan in guppies. Female poeciliids have already been shown to undergo physical, and toa less extent behavioural, masculinization as a result of exposure to paper mill ef¯uent(Howell et al., 1980). It would be interesting to see whether the opposite effect ±feminization ± occurs during exposure of males to oestrogenic compounds, and whetherchanges in courtship could yield an early warning system for such pollutants (Purdomet al., 1994; Sumpter et al., 1996).
Overall, it remains to be seen whether courtship by ®shes yields an improvement insensitivity, practicality, and cost over existing methods. This is well worth exploring,particularly as automated behavioural monitoring techniques using videos andcomputers become increasingly advanced.
Conservation
Studies of impacts of pollutants on reproductive behaviour may be more important forunderstanding conservation of the test ®sh themselves (arrows 6 and 7 in Fig. 4) than asgeneral assays for monitoring water quality and recommending criteria on behalf of otherspecies. Such studies could complement, rather than replace traditional studies of effectson physiology and survival.
The ecological framework presented (Fig. 4) suggests several reasons for focusing onreproductive behaviour. For example, behavioural studies of the impacts of oestrogeniccompounds might yield insights into reproductive output in species where males provideparental care. Indeed, parental care itself, including egg guarding and ventilation, placeshigh energetic demands on ®shes, which often lose weight during parental cycles(Sargent and Gross, 1993; Smith and Wootton, 1995). Pollutants may exacerbate thesecosts, as well as costs from sexual selection, including predation, courtship,territoriality, mating competition, and mate choice (Magnhagen, 1991; Andersson,1994; Reynolds and CoÃteÂ, 1995). One or more forms of such mating competition arewidespread in ®shes. Although there are numerous routes towards reduced offspringproduction in current and future breeding attempts (Fig. 4), we need more research onthese to explore properly their links to conservation.
Empirical justi®cation for focusing on reproductive behaviour comes from a recentsurvey of conservation of ®shes (Bruton, 1995). Cross-species comparisons suggestedthat ®sh with more complex reproductive behaviours are particularly at risk. Parental
Effects of pollution on reproductive behaviour 487
care is found in 21% of teleost ®sh families (Gross and Sargent, 1985), and isparticularly common in species breeding in fresh water, estuaries, and coastal marineenvironments. These are also the habitats that receive the greatest inputs of pollutants(Clark, 1992; Mason, 1996).
Conclusion
The study of effects of pollution on reproductive behaviour of ®shes has yielded someclear effects on courtship and partental care for several species, but few studies havegone far enough to draw clear inferences for effects on populations. As a tool forbiomonitoring or use in water quality criteria, such studies are unlikely to be able tocompete with standard toxicity tests, although they may complement them effectively,especially since there is evidence of strong sensitivity to pollutants. From the standpointof conservation of populations of the ®shes themselves, it would be useful to explorefurther the links between alterations in reproductive behaviour, reproductive output, andthe mechanisms that govern population dynamics. Such an integration of behaviouralecology with pollution research may provide useful insights for both ®elds.
Acknowledgements
We wish to thank W. Mike Howell for the photographs used for Figs 2 and 3 and forcomments on the manuscript. We also thank Mike Elliott, Alastair Grant and PeterMatthiessen for helpful discussions and Nick Dulvy, Paul Hart, David Kime, Carl Smithand two anonymous referees for comments on the manuscript.
References
Abel, P.D. (1989) Water Pollution Biology. Chichester: Ellis Horwood Ltd. 231 pp.
Alabaster, J.S. and Lloyd, R. (1982) Water Quality Criteria for Freshwater Fish, 2nd edn. London:
Butterworths.
Alatalo, R., Mappes, J. and Elgar, M.A. (1997) Heritabilities and paradigm shifts. Nature 385,
402±403.
Andersson, M. (1994) Sexual Selection. Princeton: Princeton University Press. 599 pp.
Aquatic Sciences and Fisheries Abstracts. London: SilverPlatter International.
Atchison, G.J., Henry, M.G. and Sandheinrich, M.B. (1987) Effects of metals on ®sh behavior: a
review. Env. Biol. Fishes 18, 11±25.
Beitinger, T.L. (1990) Behavioral reactions for the assessment of stress in ®shes. J. Great Lakes Res.
16, 495±528.
Benton, M.J., Diamond, S.A. and Guttman, S.I. (1994) A genetic and morphological comparison of
Helisoma trivolvis and Gambusia holbrooki from clean and contaminated habitats. Ecotoxicol.
Env. Safety 29, 20±37.
Bettoli, P.W. and Clark, P.W. (1992) Behavior of sun®sh exposed to herbicides: a ®eld study. Env.
Toxicol. Chem. 11, 1461±1467.
Bortone, S.A., Davis, W.P. and Bundrick, C.M. (1989) Morphological and behavioral characters in
mosquito®sh as potential bioindication of exposure to Kraft mill ef¯uent. Bull. Env. Contam.
Toxicol. 43, 370±377.
Breitburg, D.L. (1992) Episodic hypoxia in Chesapeake Bay: interacting effects of recruitment,
behavior, and physical disturbance. Ecol. Monogr. 62, 525±546.
488 Jones and Reynolds
Bruton, M.N. (1995) Have ®shes had their chips? The dilemma of threatened ®shes. Env. Biol. Fishes
43, 1±27.
Bryan, G.W. and Langston, W.J. (1992) Bioavailability, accumulation and effects of heavy metals in
sediments with special reference to United Kingdom estuaries: a review. Env. Pollut. 76, 89±131.
Carlisle, T.R. (1982) Brood success in variable environments: implications for parental care allocation.
Anim. Behav. 30, 824±836.
Clark, R.B. (1992) Marine Pollution, 3rd edn. Oxford: Clarendon Press. 172 pp.
Coleman, R.M. and Fischer, R.U. (1991) Brood size, male fanning effort and the bioenergetics of a
nonshareable parental investment in bluegill sun®sh, Lepomis macrochirus (Teleostei: Centrarch-
idae). Ethology 87, 177±188.
Coleman, R.M., Gross, M.R. and Sargent, R.C. (1985) Parental investment decision rules: a test in
bluegill sun®sh. Behav. Ecol. Sociobiol. 18, 59±66.
Colgan, P.W., Cross, J.A. and Johansen, P.H. (1982) Guppy behaviour during exposure to a sub-lethal
concentration of phenol. Bull. Env. Contam. Toxicol. 28, 20±27.
Council of the European Communities (1978) Directive on the quality of freshwaters needing
protection or improvement in order to support ®sh life (78=659=EEC). Of®cial Journal of the
European Communities. L 222, 14 August 1978.
Denny, J.S. (1987) Guidelines for the culture of fathead minnows Pimephales promelas for use in
toxicity tests. EPA=600=3-87=001. Environmental Protection Agency, Duluth, MN, 42 pp.
Douthwaite, R.J., Fox, P.J., Matthiessen, P. and Russell-Smith, A. (1981) Environmental impact of
aerosols of endosulfan, applied for tsetse ¯y control in the Okavango Delta, Botswana. The Final
Report of the Endosulfan Monitoring Project. London: Overseas Development Administration.
Dùving, K.B. (1991) Assessment of animal behaviour as a method to indicate environmental toxicity.
Comp. Biochem. Physiol. 100C, 247±252.
Falconer, D.S. (1989) Introduction to Quantitative Genetics, 3rd edn. Harlow: Longman Scienti®c and
Technical. 438 pp.
Farr, J.A. (1980) Social behavior patterns as determinants of reproductive success in the guppy,
Poecilia reticulata Peters (Pisces: Poeciliidae). Anim. Behav. 28, 1195±1201.
Gerking, S.D. (1980) Fish reproduction and stress. In Ali, M.A., ed. Environmental Physiology of
Fishes. New York: Plenum Press, pp. 569±587.
Gross, M.R. and Sargent, R.C. (1985) The evolution of male and female parental care in ®shes. Am.
Zool. 25, 807±822.
Gustavson, K. and Wangberg, S.A. (1995) Tolerance induction and succession in microalgae
communities exposed to copper and atrazine. Aquat. Toxicol. 32, 283±302.
Hilborn, R. and Walters, C.J. (1992) Quantitative Stock Assessment: Choice, Dynamics and
Uncertainty. London: Chapman and Hall.
Houde, A.E. (1997) Sex Color and Mate Choice in Guppies. Princeton: Princeton University Press,
210pp.
Howell, W.M., Black, D.A. and Bortone, S.A. (1980) Abnormal expression of secondary sex characters
in a population of mosquito®sh, Gambusia af®nis holbrooki: evidence for environmentally-
induced masculinization. Copeia 1980, 676±681.
Howells, G. (ed.) (1994) Water Quality for Freshwater Fish. Switzerland: Gordon and Breach.
Hurlbert, S.H. (1984) Pseudoreplication and the design of ecological ®eld experiments. Ecol. Monogr.
54, 187±211.
IUCN (1996) 1996 IUCN Red List of Threatened Animals. Gland, Switzerland: IUCN, 368pp.
Kime, D.E. (1995) The effects of pollution on reproduction in ®sh Rev. Fish Biol. Fisheries 5, 52±96.
King, M. (1995) Fisheries Biology, Assessment and Management. Oxford: Blackwell Science Ltd.
341 pp.
Kopp, R.L., Guttman, S.I. and Wissing, T.E. (1992) Genetic indicators of environmental stress in
central mudminnow (Umbria limi) populations exposed to acid deposition in the Adirondack
Mountains. Env. Toxicol. Chem. 11, 665±676.
Effects of pollution on reproductive behaviour 489
Kramer, D.L. and Mehegan, J.P. (1981) Aquatic surface respiration, an adaptive response to hypoxia in
the guppy, Poecilia reticulata (Pisces, Poeciliidae). Env. Biol. Fishes 6, 299±313.
Krebs, J.R. and Davies, N.B. (1997) Behavioural Ecology, 4th edn. Oxford: Blackwell Scienti®c
Publications. 481 pp.
Krotzer, M.J. (1990) The effects of induced masculinization on reproductive and aggressive behaviors
of the female mosquito®sh, Gambusia af®nis af®nis. Env. Biol. Fishes 29, 127±134.
Kvarnemo, C. (1997) Food affects the potential reproductive rates of sand goby females, but not of
males. Behav. Ecol. (in press).
Lawton, J.H. and May, R.M. (1995) Extinction Rates. Oxford: Oxford University Press. 233 pp.
Little, E.E., Flerov, B.A. and Ruzhinskaya, N.N. (1985) Behavioral approaches in aquatic toxicity
investigations: a review. In Mehrle, P.M. Jr, Gray, R.H. and Kendall, R.L., eds. Toxic Substances
in the Aquatic Environment: An International Aspect. Bethesda, MD: American Fisheries Society,
pp. 72±98.
Lloyd, R. (1992) Pollution and Freshwater Fish. Oxford: Fishing News Books. 176 pp.
Lorenz, J.J. and Taylor, D.H. (1992a) Effects of a chemical stressor on the parental behavior of convict
cichlids with offspring in early stages of development. Trans. Am. Fish. Soc. 121, 315±321.
Lorenz, J.J. and Taylor, D.H. (1992b) The effects of low pH as a chemical stressor on the ability of
convict cichlids (Cichlasoma nigrofasciatum) to successfully brood their young. Copeia 1992,
832±839.
Magnhagen, C. (1991) Predation risk as a cost of reproduction. Trends Ecol. Evol. 6, 183±185.
Maitland, P.S. (1995) The conservation of freshwater ®sh: past and present experience. Biol. Conserv.
72, 259±270.
Mason, C.F. (1996) Biology of Freshwater Pollution. Singapore: Longman. 356 pp.
Matthiessen, P. and Logan, J.W.M. (1984) Low concentration effects of endosulfan insecticide on
reproductive behavior in the tropical cichlid ®sh Sarotherodon mossambicus. Bull. Env. Contam.
Toxicol. 33, 575±583.
Meffe, G.K. (1986) Conservation genetics and the management of endangered ®shes. Fisheries 11,
14±23.
Moyle, P.B. and Leidy, R.A. (1992) Loss of biodiversity in aquatic ecosystems: evidence from ®sh
faunas. In Fiedler, P.L. and Jain, S.K., eds. Conservation Biology. New York: Chapman and Hall,
pp. 129±169.
MuÈller, R. and Lloyd, R. (1994) Sublethal and Chronic Effects of Pollution on Freshwater Fish.
Oxford: Blackwell Science Ltd. 371 pp.
Partridge, L. and Harvey, P.H. (1985) Costs of reproduction. Nature 316, 20.
Pascoe, D. and Beattie, J.H. (1979) Resistance to cadmium by pretreated rainbow trout alevins. J. Fish
Biol. 14, 303±308.
Pomiankowski, A. and Mùller, A.P. (1995) A resolution of the lek paradox. Proc. R. Soc. Lond. 260B,
21±29.
Potts, G.W., Keenleyside, M.H.A. and Edwards, J.M. (1988) The effect of silt on the parental
behaviour of the sea stickleback, Spinachia spinachia. J. Mar. Biol. Ass. U.K. 68, 277±286.
Purdom, C.E., Hardiman, P.A., Bye, V.J., Eno, N.C., Tyler, C.R. and Sumpter, J.P. (1994) Estrogenic
effects of ef¯uents from sewage treatment works. Chem. Ecol. 8, 275±285.
Pyron, M. and Beitinger, T.L. (1989) Effects of selenium on reproductive behaviour and fry of fathead
minnows. Bull. Env. Contam. Toxicol. 42, 609±613.
Reebs, S.G., Whoriskey, F.G. and FitzGerald, G.J. (1984) Diel patterns of fanning activity, egg
respiration, and the nocturnal behavior of male three-spined sicklebacks, Gasterosteus aculeatus
L. ( f. trachurus). Can. J. Zool. 62, 329±334.
Reynolds, J.D. (1993) Should attractive individuals court more? Theory and a test. Am. Nat. 141,
914±927.
Reynolds, J.D. and CoÃteÂ, I.M. (1995) Direct selection on mate choice: female redlip blennies pay more
for better mates. Behav. Ecol. 6, 175±181.
490 Jones and Reynolds
Reynolds, J.D. and Gross, M.R. (1992) Female mate preference enhances offspring growth and
reproduction in a ®sh, Poecilia reticulata. Proc. R. Soc. Lond. 250B, 57±62.
Reynolds, J.D., Gross, M.R. and Coombs, M.J. (1993) Environmental conditions and male morphology
determine alternative mating behaviour in Trinidadian guppies. Anim. Behav. 45, 145±152.
Roff, D.A. (1992) The Evolution of Life Histories. New York: Chapman and Hall. 535 pp.
Ryabov, I.N. (1985) The behaviour of Gasterosteus aculeatus L. in the zone of action of warm waste
water. Behaviour 93, 56.
Sargent, R.C. and Gross, M.R. (1993) Williams' principle: an explanation of parental care in teleost
®shes. In Pitcher, T.J., ed. Behaviour of Teleost Fishes, 2nd edn. London: Chapman and Hall,
pp. 333±361.
Scherer, E. (1992) Behavioural responses as indicators of environmental alterations: approaches,
results, developments. J. Appl. Ichthyol. 8, 22±131.
SchroÈder, J.H. and Peters, K. (1988a) Differential courtship activity of competing guppy males
(Poecilia reticulata Peters; Pisces: Poeciliidae) as an indicator for low concentrations of aquatic
pollutants. Bull. Env. Contam. Toxicol. 40, 396±404.
SchroÈder, J.H. and Peters, K. (1988b) Differential courtship acitivity and alterations of reproductive
success of competing guppy males (Poecilia reticulata Peters: Pisces: Poeciliidae) as an indicator
for low concentrations of aquatic pollutants. Bull. Env. Contam. Toxicol. 41, 385±390.
Sibly, R.M. and Smith, R.H. (1985) Behavioural Ecology: Ecological Consequences of Adaptive
Behaviour. Oxford: Blackwell Scienti®c Publications. 620 pp.
Smith, C. and Wootton, R.J. (1995) The costs of parental care in teleost ®shes. Rev. Fish Biol.
Fisheries 5, 7±22.
Smith, E.H. and Logan, D.T. (1997) Linking environmental toxicology, ethology, and conservation. In
Clemmons, J.R. and Buchholz, R., eds. Behavioral Approaches to Conservation in the Wild.
Cambridge: Cambridge University Press, pp. 277±302.
Stafford, C.L. and Ward, J.A. (1983) Effects of monochloramine on courtship and spawning in the
cichlid ®sh Etroplus maculatus. In Noakes, D.L.G., Linquist, D.G., Helfman, G.S. and Ward, J.A.,
eds. Predators and Prey in Fishes. The Hague: Dr W. Junk, pp. 213±220.
Stearns, S.C. (1992) The Evolution of Life Histories. Oxford: Oxford University Press. 249 pp.
Strandberg, C.H. (1971) Water pollution. In Smith, G.H., ed. Conservation of Natural Resources. New
York: John Wiley & Sons, pp. 189±219.
Sumpter, J.P., Jobling, S. and Tyler, C.R. (1996) Oestrogenic substances in the aquatic environment
and their potential impact on animals, particularly ®sh. In Taylor, E.W., ed. Toxicology of Aquatic
Pollution ± Physiological, Cellular and Molecular Approaches. Cambridge: Cambridge University
Press, pp. 205±224.
Sutherland, W.J. (1996) From Individual Behaviour to Population Ecology. Oxford University Press.
213 pp.
Tanner, D.K. and Knuth, M.L. (1995) Effects of azinphos-methyl on the reproductive success of the
bluegill sun®sh, Lepomis macrochirus, in littoral enclosures. Ecotoxicol. Env. Saf. 32, 184±193.
Tanner, D.K. and Knuth, M.L. (1996) Effects of esfenvalerate on the reproductive success of the
bluegill sun®sh, Lepomis macrochirus in littoral enclosures. Arch. Env. Contam. Toxicol. 31,
244±251.
Tanner, D.K. and Moffett, M.F. (1995) Effects of di¯ubenzuron on the reproductive success of the
bluegill sun®sh, Lepomis macrochirus. Env. Toxicol. Chem. 14, 1345±1355.
Weber, D.N. (1993) Exposure to sublethal levels of waterborne lead alters reproductive behavior
patterns in fathead minnows (Pimephales promelas). NeuroToxicology 14, 347±358.
Williams, G.C. (1966) Natural selection, the cost of reproduction and a re®nement of Lack's principle.
Am. Nat. 100, 687±690.
Wood, L.B. (1982) The Restoration of the Tidal Thames. Bristol: Adam Hilger Ltd. 202 pp.
Accepted 3 July 1997
Effects of pollution on reproductive behaviour 491