Journal of Applied Ecology 2009, 46, 976–982 doi: 10.1111/j.1365-2664.2009.01682.

Fishing impacts on the marine inorganic carbon cycle

Simon Jennings1,2* and Rod W. Wilson3
1
Centre for Environment, Fisheries and Aquaculture Science, Pakefield Road, Lowestoft, NR33 0HT, UK; 2School
of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ, UK; and 3School of Biosciences, Hatherly
Laboratories, University of Exeter, Exeter, Devon EX4 4PS, UK

Summary
1. Teleost fish excrete precipitated carbonate and make significant contributions to the marine inor-
ganic carbon cycle at regional and global scales. As total carbonate production is linked to fish size
and abundance, fishing is predicted to affect carbonate production by modifying fish abundance
and size-structure.
2. We draw on concepts from physiology, metabolic ecology, life history theory, population
dynamics and community ecology to develop, validate and apply analytical tools to assess fishing
impacts on carbonate production. Outputs suggest that population and community carbonate pro-
duction fall rapidly at lower rates of fishing than those used as management targets for sustainable
yield.
3. Theoretical predictions are corroborated by estimated trends in carbonate production by a
herring population and a coral reef fish community subject to fishing. Our analytical results build
on widely applicable relationships between life history parameters and metabolic rates, and can be
generalized to most fished ecosystems.
4. Synthesis and applications. If the maintenance of chemical processes as well as biological process
were adopted as a management objective for fisheries then the methods we have developed can be
applied to assess the effects of fishing on carbonate production and to advise on acceptable rates of
fishing. Maintenance of this ecosystem service would require lower rates of fishing mortality than
those recommended to achieve sustainable yield.
Key-words: community, ecosystem approach, ecosystem services, fish carbonate, fisheries,
management, population

Teleost fish living in salt water precipitate carbonates in the

Introduction
Fisheries managers tend to focus on achieving sustainable and
profitable fisheries while minimizing impacts on non-target
species and habitats (Sinclair & Valdimarsson 2003). However,
fisheries also impact ecosystem services and these impacts need
to be assessed to determine whether they should be managed.
One important ecosystem service provided by teleost fish is car-
bonate production, as a recent (conservative) estimate suggests
they contribute 3–15% of new oceanic carbonate production
globally per year and that this may account for 7Æ7–26Æ2%
of carbonate dissolution in the top 1000 m of the ocean,
with implications for the acid–base balance in the upper ocean
(Wilson et al. 2009). Higher than average rates of fish carbon-
ate production and dissolution are expected in shelf seas and
upwellings, as >50% of global fish biomass occurs in these
regions (Jennings et al. 2008).
*Correspondence author. E-mail: simon.jennings@cefas.co.uk
intestine and subsequently excrete them in mucus-coated tubes
or pellets and in the faeces (Walsh et al. 1991; Wilson et al.
1996; Wilson, Wilson, & Grosell 2002; Grosell 2006). Follow-
ing excretion, the organic parts of the tubes, pellets or faeces
rapidly degrade, leaving inorganic crystals of calcium carbon-
ate (Walsh et al. 1991). Carbonate precipitates are formed
whether or not the fish are feeding (Wilson et al. 1996; Taylor
& Grosell 2006) because the essential process of drinking
seawater results in the supersaturation of calcium and magne-
sium carbonates in the intestine (Wilson et al. 2002; Wilson &
Grosell 2003). Walsh et al. (1991) suggested that carbonate
excretion might make a significant contribution to the
inorganic carbon cycle, and this has since been confirmed by
the global analysis of Wilson et al. (2009). Fish carbonates
have a higher magnesium content and are therefore expected
to have greater solubility than other marine carbonates. This
would result in faster dissolution with depth, providing a novel
explanation for much of the increase in titratable alkalinity
within upper 1000 m of the ocean (Wilson et al. 2009).

 2009 The Authors. Journal compilation  2009 British Ecological Society


The rate of carbonate production by fish is assumed to be
proportional to the seawater drinking rate and metabolic rate
(Takei & Tsukada 2001). This is because osmoregulatory pro-
cesses such as drinking and active ion transport serve to coun-
terbalance passive ion and water fluxes (primarily at the gills)
and these passive fluxes (water loss and ion gain in marine fish,
and the opposite in freshwater fish) are directly proportional to
gill ventilation and perfusion and therefore proportional to the
oxygen uptake rate and metabolic rate (Nilsson 1986; Gonz-
alez & McDonald 1992). As the metabolic rates of individuals
and species vary with environmental temperature and body
size (Clarke & Johnston 1999; Glazier 2005), the temperature
of the surrounding environment as well as the size composition
and total abundance of a fish population or community, will
determine the total rate of carbonate production.
Fishing takes place in all the global oceans and has substan-
tially modified the structure of fish populations and communi-
ties. Of those factors that influence rates of carbonate
production by fish communities, both total biomass and size
structure are affected by fishing (Quinn & Deriso 1999; Bianchi
et al. 2000; Shin et al. 2005). Comparisons among areas subject
to different fishing intensities and temporal comparison within
areas where fishing effort has increased over time, have both
shown that increased fishing mortality is associated with
decreases in total biomass and a shift in the size distribution
from larger to smaller individuals (Bianchi et al. 2000; Shin
et al. 2005).
Here, we develop, validate and apply methods for describing
relationships between fishing intensity and carbonate produc-
tion by fish populations and communities. These methods can
be used to predict how the size composition and abundance of
fish communities changes in response to fishing mortality and
the consequent impact on rates of carbonate production. Our
new methods provide a quantitative approach for assessing
whether the management of renewable resources should focus
on chemistry as well as biology, an important step in incorpo-
rating concerns about the sustainability of ecosystem services
into environmental management.
Materials and Methods
The analyses comprise four stages: (i) development of a model linking
fish carbonate production to body mass and temperature, (ii) devel-
opment of a model of fishing effects on population carbonate produc-
tion, (iii) development of a model of fishing effects on community
carbonate production, and (iv) validation and application of the
models, based on data that demonstrate fishing-related changes in
the body size composition and abundance of a population and
community.
CARBONATE PRODUCTION
A model that links the rate of carbonate production to fish body size
and temperature was used to estimate rates of carbonate production.
This is based on the observation that rates of drinking by fish are
directly proportional to metabolic rate, and that drinking rates deter-
mine rates of carbonate production (Takei & Tsukada 2001; Wilson
et al. 2002; Taylor & Grosell 2006). Given this indirect link between
 2009 The Authors. Journal compilation  2009 British Ecological Society, Journal of Applied Ecology, 46, 976–982
Fishing impacts on the carbon cycle 977
carbonate production and metabolic rate, changes in relative rates of
carbonate production with temperature can be approximated with
the Arrhenius relationship. This relationship provides a good descrip-
tion, but not a causal explanation, of the effects of temperature on
metabolic rate (e.g. Clarke & Johnston 1999). The Arrhenius relation-
ship
R ¼ AeE=kT eqn 1
links the rate coefficients of a chemical reaction (R) to the absolute
temperature T, where A is a prefactor, E is the activation energy of
the reaction and k is the Boltzmann constant (or the Gas constant
when E is expressed in molar units). Over biologically relevant tem-
perature ranges E is assumed to be independent of temperature and
the minor temperature dependence of A is regarded as negligible com-
pared with the temperature dependence of the e)E ⁄ kTterm (Clarke &
Johnston 1999).
Taking the natural logs of the Arrhenius equation gives:
E 1
loge R ¼ þ loge A eqn 2
k T
thus a plot of logeR vs T)1 is a straight line of slope –E ⁄ k and intercept
logeA. This approach was used to estimate )E ⁄ k from the data com-
pilation of Clarke & Johnston (1999) that listed temperature and pre-
dicted resting (standard) metabolic rates for a range of fish species at
body mass 50 g. The relationship was highly significant F = 81Æ451,88
(P < 0Æ0001), slope ()E ⁄ k) was )4727Æ36 (95% C.I. )3686Æ4 to
)5768Æ3) and the intercept was 14Æ27 (95% C.I. 10Æ59–17Æ95).
We assumed that the scaling of metabolism with body mass (W),
both within and among species, could be approximated as W0Æ75. In
reality, the value of the exponent can vary within and among species
(Clarke & Johnston 1999; Glazier 2005) but we consider W0Æ75 an ade-
quate approximation for developing a generically applicable
approach, and the exponent could easily be modified in the subse-
quent equations if species-specific data were available. We combined
the relationship between body mass and metabolism with the Arrhe-
nius equation describing temperature effects following the approach
of Gillooly et al. (2002). Assuming that the rate of metabolism is pro-
portional to the rate of carbonate production C (given the effects of
metabolism on drinking rate; Takei & Tsukada 2001)
C ¼ aqaW0:75 AeE=kT eqn 3
where a is a constant. Constants a and q were added to correct experi-
mentally measured mass specific rates of carbonate production (Wil-
son et al. 2009) for the ratio between carbonate production in active
and resting fish (a) and the relatively higher resting metabolism and
drinking rates of fish species living in the water column (q) (Clarke &
Johnston 1999; Takei & Tsukada 2001). Alpha exceeds one in wild
fish because metabolic rate, and hence drinking rate and carbonate
production, rise above resting (experimental) levels during normal
activity (Kerr 1982). Carbonate production per unit mass can thus be
expressed as
C=W ¼ aqaW0:25 AeE=kT eqn 4
To fit equation (4) to data for resting unfed benthic fish, the values of
a and q were both set to one. This equation was fitted to data for car-
bonate production per unit mass by Gulf toadfish Opsanus beta and
European flounder Platichthys flesus, as recorded experimentally in
978 S. Jennings & R. W. Wilson
resting unfed fish (Walsh et al. 1991; Wilson et al. 2002; Taylor &
Grosell 2006; Taylor et al. 2007), to determine constants a and A, giv-
ing the equation
C=W ¼ aq9:81  108 W0:25 e4727ð1=TÞ eqn 5
where carbonate production per unit mass is expressed as l mol C
kg)1 h)1 (molar C = g C ⁄ 12), W is body mass in g and T is tempera-
ture in Kelvin (C+273) (Wilson et al. 2009). We set the value of a to
2Æ5 to q 2Æ4 based on the differences between resting and activity
metabolism and the relative activity levels of bottom living and pela-
gic fishes reported in Wilson et al. (2009). For simplicity, and given
the very limited data currently available to parameterise the model,
we assume the same model applies within and among species. How-
ever, the general form of the model will allow it to be re-parameterised
if additional data on rates of carbonate production are collected.
FISHING EFFECTS ON POPULATIONS
Total carbonate production by a population at a given temperature
depends on size composition and abundance. Changes in carbonate
production by a cohort (year class) with time are a function of the
changes in the number of individuals owing to mortality and the
changes in the size of individuals owing to growth. The number of
individuals in a cohort at time t can be estimated using (e.g. Quinn &
Deriso 1999)
Nt ¼ N0 eðMþFÞt eqn 6
where N0 is the number of individuals present at t = 0, F is fishing
mortality and M is natural mortality. The von Bertalanffy Growth
Equation can be used to describe W at time t as a function of the
asymptotic mass W¥(e.g. Quinn & Deriso 1999)
Wt ¼ W1 ð1  eKðtt0Þ Þ3 eqn 7
where t0 is the time when W is theoretically zero and K is the Brody
growth coefficient.
Following equation 3, carbonate production at time t will be
Ct ¼ Nt aqaW0:75
t AeE=kT eqn 8
where Nt is the number of individuals present at time t as determined
from equation (6) and Wt is determined from equation (7). Assuming
temperature is constant through the cohort lifespan, the time when a
cohort is producing the maximum amount of carbonate tCmax can
thus be determined by substituting (6) and (7) into (8), differentiating
with respect to t and solving for tCmax when the first derivative is set to
zero.
  1=K
9 K
tC max ¼ t0 þ loge 1 þ eqn 9
4 MþF
Equation (9) can be substituted into (7) to give the weight of fish in a
cohort when they are producing the maximum amount of carbonate
WCmax and the equation for WCmax reduces to
  
MþF 3
WC max ¼ W1 1= 1 þ eqn 10
2:25K
The advantage of equations (9) and (10) is that for F = 0, the
M ⁄ K ratio, which is relatively constant among many fish populations
(Beverton 1992), can be used to predict the time and body mass when
 2009 The Authors. Journal compilation  2009 British Ecological Society, Journal of Applied Ecology, 46, 976–982
an unexploited cohort is producing most carbonate. This allows the
application of the method when M and K are not known separately.
Observed values of tCmax and WCmax in fished populations can be
compared with theoretical values for unfished populations, providing
an indicator of the relative impacts of fishing on carbonate produc-
tion.
As most fish population assessments are age based, a summation of
Ct across age classes up to the maximum age tmax provides an ade-
quate assessment of total carbonate production throughout the life-
span of a cohort Ctotp (and hence the carbonate production of a
population at steady state). This is given by
X
tmax
Ctotp ¼ ðF þ MÞt Nt aqaWt075 AeE=kT eqn 11
t¼0
The methods of population-based analysis were applied to the her-
ring population in the North Sea, for which there are long-term age-
structured data and very large fluctuations in abundance and mortal-
ity over time (ICES 2007). We estimated carbonate production based
on a full age-structured population assessment for the fished popula-
tion and for the population in the absence of fishing. The life history
parameters of the herring population were W¥ = 332g, t0 = )1Æ1,
K = 0Æ4 and mean M = 0Æ31, with age-specific M based on ICES
(2007) in the age-structured analysis. Mean sea temperature in the
North Sea was taken as 10Æ5 C (ICES, unpublished data).
FISHING EFFECTS ON COMMUNITIES
To assess the potential effects of fishing on carbonate production by
fish communities we modified a model that captures the direct and
indirect effects of fishing on community abundance and size structure
(Pope et al. 2006). The model predicts interrelationships between fish-
ing, population and community dynamics that are supported by
empirical analysis and uses 15 parameters to describe a 13 ‘species’
fish community, where species are defined by their maximum body
size (asymptotic length L¥) and size-related life history parameters.
An overall F acts on all species and can be modified by defining spe-
cies and size selectivity. The parameter values followed the ‘key run’
of Pope et al. (2006) but the exploitation pattern was modified so that
all ‘species’ were fished at the same F. This pattern is indicative of
exploitation in many multispecies fisheries where small and large
fishes are targeted. The model is intended to mimic the effects of fish-
ing in a shallow (typically <200 m depth) shelf sea ecosystem, and
was extended to output relative carbonate production.
As the W and length (L) in fish are typically related as W  L3, the
scaling of C ⁄ W and L will be  L)0Æ75 if the scaling of C ⁄ W and W is
 W)0.25. The total relative carbonate production of the fish commu-
nity (Ctotc) at different levels of fishing mortality F was thus defined as
X
Lmax
Ctotc ¼ WL NL ðC=WÞL eqn 12
Lmin
where Lmin and Lmax were taken as 5 cm and 130 cm and values of
weight at length (WL) and numbers at length (NL) for 10 cm size clas-
ses were output from the model.
The theoretical analysis of fishing effects on carbonate production
suggested that the largest changes in carbonate production would
occur at relatively low levels of fishing mortality. Therefore, to cor-
roborate or reject this pattern we needed to analyse empirical data
from areas with very low levels of fishing mortality. Most fisheries
time series data were collected well after exploitation rates exceeded
F = 0Æ3 or 0Æ4 and thus we analysed data from spatial comparisons
 Fishing impacts on the carbon cycle 979

of areas subject to different levels of fishing effort that included lightly (a) 0·25
or unfished areas (Jennings & Polunin 1997).
Fish abundance was determined by underwater visual census in ten 0·20

WCmax/W∞
reef fishing grounds on the western coast of Kadavu Island, Fiji. The
boundaries of each fishing ground enclose areas of reef where people 0·15
from specific villages have exclusive rights to fish. As a result, varia-
tions in human population density and reef area among grounds 0·10
mean that they are subject to a range of fishing intensities. Reef fishes
0·05
in the families studied do not move extensively among grounds and
thus their abundance is determined by the recruitment of larvae from
0·00
the plankton, natural mortality and local fishing intensity. Further 0·0 0·2 0·4 0·6 0·8 1·0
details of the study areas, associated fish communities, data collection Fishing mortality
and processing are provided in Jennings & Polunin (1997).
All fish census work was conducted in 1995 and 1996 and 144 spe- (b) 1·0
cies were censused. Abundances were determined at seven randomly

Relative C/R or Y/R

0·8
selected replicate sites in each of the fishing grounds. At each site, the
abundance and size of census species ‡8 cm length was estimated
0·6
within 12 adjacent census areas of 7 m radius by counting each fish
and estimating its length to ±1 cm. Species in each census area were 0·4
recorded sequentially, with the most active species being first. When
a count for one species was complete, all further movements of that 0·2
species were disregarded. Fish lengths were converted to mass from
published length–mass relationships. Carbonate production was cal- 0·0
culated from mass at a temperature of 27Æ5 C, the annual mean 0·0 0·2 0·4 0·6 0·8 1·0
Fishing mortality
water temperature (NOAA 2007).
An index of fishing intensity in each fishing ground was calculated (c) 1·0
by dividing the number of people in the villages that have fishing
rights in the fishing grounds by the length of reef front. In Fijian vil- 0·8
lages, all villagers have fishing rights and so the population approxi-
Relative Ctotp

mates the number of fishers and consumers (Jennings & Polunin 0·6
1997). Human population data were obtained from the most recent
census. The length of reef front was measured on aerial photographs 0·4
or navigational charts.
0·2

Results
The theoretical analysis of relationships between the body
weight at which a cohort has maximum carbonate production
WCmax, asymptotic weight W¥ and fishing mortality F
showed that WCmax was 21Æ6% of W¥ in the absence of fish-
ing and decreased to <5% of W¥ as F increased to 1Æ0
(Fig. 1a). Carbonate production per recruit (C ⁄ R) decreased
by 90% over this range of fishing mortality, with a faster rate
of decrease per unit increase in F at lower F (Fig. 1b). Indeed,
when yield per recruit (Y ⁄ R) attains the maximum value
(F0Æ3), C ⁄ R has decreased by >60% from the value when
F = 0. Total carbonate production by the population (Ctotp)
as a proportion of Ctotp when F = 0 is almost linearly related
to WCmax ⁄ W¥ (Fig. 1c).
Estimated carbonate production per recruit by the North
Sea herring population fluctuated from 1960 to 1996 (Fig. 2a)
and, when C ⁄ R was expressed as a proportion of C ⁄ RF = 0, it
was negatively correlated with fishing mortality (Fig. 2b).
Total carbonate production varied substantially among
cohorts and among years, with both trends tending to precede
decreases in total population biomass (Fig. 3).
The relationship between the age at maximum carbonate
production tCmax or WCmax and F for herring (Fig. 4) showed
that tCmax or WCmax would be expected to occur early in life
0·0
0·05 0·10 0·15 0·20 0·25
WCmax/W∞
Fig. 1. Relationships between (a) body mass at maximum carbonate
production as a proportion of asymptotic mass (WCmax ⁄ W¥) and
fishing mortality, (b) relative carbonate production per recruit (con-
tinuous line) or yield per recruit (solid line) and fishing mortality and
(c) relative carbonate production by the population (Ctotp ⁄ CtotpF=0)
and body mass at maximum carbonate production as a proportion of
asymptotic mass (WCmax ⁄ W¥). We assumed W¥ = 1000, k = 0Æ3
and M ⁄ K = 1Æ5.
and at low body mass if the population were fished at
F = 0Æ68; the mean F in the period 1960–1996. At the target
F of 0Æ25 (ages 2–6) which applies when spawning population
biomass is >1Æ3 · 106 tonnes (ICES 2007), carbonate produc-
tion at tCmax or WCmax would be more than double the value at
F = 0Æ68. The predicted tCmax or WCmax for the population
are broadly consistent with the values calculated from the age-
structured assessment, with the age 0 group always having
highest estimated carbonate production from 1960 to 1996 and
average mass of this group being 13 g.
The model of the effects of fishing on carbonate production
suggested that the greatest decreases in community carbonate
production would occur at relatively low rates of mortality

 2009 The Authors. Journal compilation  2009 British Ecological Society, Journal of Applied Ecology, 46, 976–982
980 S. Jennings & R. W. Wilson

(a) 1·1 200 2·5

1·0 2·0
150

0·9 1·5

WCmax
C/R (g)

tCmax
100

0·8 1·0

50
0·5
0·7

0 0·0
0·6 0·0 0·2 0·4 0·6 0·8 1·0
1960 1970 1980 1990 Fishing mortality F
Year class
(b) 0·35 Fig. 4. Relationship between WCmax (solid line) and tCmax (broken
line) and fishing mortality for the North Sea herring population.
C/R as proportion C/RF = 0

0·30
tion with fishing was slightly lower than the rate of decrease in
total biomass (Fig. 5a). The mean mass of individuals in the
0·25
modelled community broadly declined with increasing fishing
mortality while the mean rate of carbonate production per unit
mass showed a corresponding increase (Fig. 5b). In these simu-
0·20 lations, F = 0Æ25 corresponded to the multispecies F at which
the maximum sustainable yield could be taken from the most
vulnerable species, while maximum multispecies yield would
0·15 be taken at F > 1Æ0.
0·0 0·2 0·4 0·6 0·8 1·0 1·2 1·4 1·6 1·8
Fishing intensity in the Fijian reef fisheries was expressed in
Fishing mortality F
terms of effort rather than F, owing to the absence of data
Fig. 2. Predicted carbonate production per recruit for the North Sea describing population-specific exploitation rates. The range of
herring stock for the year classes from 1960 to 1996 (a) and the rela- fishing effort spanned more than an order of magnitude with
tionship between C ⁄ R as a proportion C ⁄ RF = 0 and the fishing mor- one ground infrequently fished. Community biomass
tality (F for ages 3–6) in each of these cohorts (b).
decreased rapidly with low and increasing levels of fishing
effort but stabilized at higher effort (Fig. 6a). Estimated car-
bonate production was highest in the least frequently fished
120 5
ground, on average 20% higher than at other grounds, where
Population biomass (106 tonnes)

CaCO3 by cohorts
CaCO3 (103 tonnes years–1)

100 CaCO3 by years production was lower and more variable, and did not show a
population biomass 4
clear relationship with fishing intensity (Fig. 6b). Carbonate
80 production per unit biomass increased with fishing effort,
3 probably reflecting the dominance of smaller fishes that pro-
60 duce more carbonate per unit mass in the more heavily
2 exploited grounds (Fig. 6c).
40

20
0 0
1960 1970 1980
Year class or year
Fig. 3. Trends in estimated CaCO3 production by the North Sea
herring population by cohorts (closed circles, solid line) and by years
(open circles, broken line) from 1960 to 1996 Total population
biomass over the same time period is shown with the dotted line and
small circles.
(Fig. 5a). At F = 0Æ5, relative carbonate production had fallen
to about 60% of that in the unfished community. At higher
rates of F, there was relatively little change in the rate of car-
bonate production. The rate of decrease in carbonate produc-
1
Discussion
For populations and communities, lower rates of fishing mor-
1990
tality than those associated with obtaining high and sustain-
able yields lead to substantial reductions in population
carbonate production. Relatively small changes in rates of
carbonate production at higher fishing mortalities imply that
current management interventions intended to achieve high
and sustainable yield will have limited effects on carbonate
production. The analytical methods and hence the results
depend on widely applicable relationships between mortality,
population and community size structure and metabolism, and
we therefore expect they can be generalized to populations and
communities in most fished ecosystems. In general, fishing
mortality reduces total carbonate production and C ⁄ R in the

 2009 The Authors. Journal compilation  2009 British Ecological Society, Journal of Applied Ecology, 46, 976–982
 Fishing impacts on the carbon cycle 981

(a) 80
Carbonate production (C) as proportion CF = 0

(a)

1·0 1·0 70

Biomass (B) as proportion BF = 0

Biomass (g m–2)
0·8 0·8 60

50
0·6 0·6
40
0·4 0·4
30

0·2 0·2
20
(b) 8·5
0·0 0·0
0·0 0·2 0·4 0·6 0·8 1·0 1·2 1·4
8·0

C production (g m–2 years–1)

(b) 1·3 7·5
4·4
7·0
Corbonate production per unit

4·2 1·2
Mean individual mass (g)

6·5
4·0
biomass (relative)

6·0
3·8 1·1
5·5
3·6
5·0
3·4 1·0
3·2 4·5
(c)
3·0 0·9 0·20
C production per unit biomass (g g–1)

2·8
0·8 0·18
2·6
0·0 0·2 0·4 0·6 0·8 1·0 1·2 1·4
Fishing mortality F
0·16

Fig. 5. (a) Changes in total carbonate production (continuous line)

and relative biomass (broken line) of a fish community as a function 0·14
of changes in the rate of fishing mortality The rate of carbonate pro-
duction (Ctotc) and biomass (B) are expressed as a proportion of their 0·12
values with no fishing (F = 0). (b) Relationship between mean indi-
vidual body mass (continuous line) or carbonate production per unit
biomass (broken line) and fishing mortality in the modelled fish com- 0·10
munity. 0 50 100 150 200 250 300
Fishing effort (persons km–2 reef front)

population because the abundance of fished cohorts will fall Fig. 6. Relationships between (a) biomass, (b) carbonate production
more rapidly with age. However, carbonate production per and (c) carbonate production per unit mass and fishing effort on
unit mass increases with fishing mortality in fished populations Fijian reef fishing grounds. Vertical bars are 95% confidence inter-
and communities because these are dominated by smaller indi- vals.
viduals.
The model that links size and temperature to carbonate relative increases in the rate of carbonate production per unit
production assumed that the intercepts of relationships biomass at different fishing intensities. The predicted trend in
between R and C and W were the same for all species. This carbonate production with fishing will also be influenced by
would not be the case in reality, as active species of pelagic fish the proportion of teleosts (carbonate producing) and elasmo-
(e.g. tunas) have higher metabolic rates than less active bottom- branchs (not carbonate producing) in the fish community.
dwelling species (e.g. groupers, flatfishes) at a given body size Elasmobranchs tend to have relatively large body sizes and to
and temperature (Clarke & Johnston 1999) and this would be more vulnerable to fishing owing to their low intrinsic rates
influence their drinking rates. However, the model structure is of increase (Stevens et al. 2000), so it might be expected that
sufficiently general that it could easily be parameterised with they will form a smaller proportion of total biomass at high
species-specific data as they become available. In the case of fishing mortality. This would exaggerate any predicted
the community analysis, larger bottom-dwelling species do decrease in carbonate production.
tend to be more vulnerable to fishing than smaller pelagic The relationship between WCmax and W¥ provides a linear
species and changes in their relative abundance could lead to indicator of the extent to which relative carbonate production

 2009 The Authors. Journal compilation  2009 British Ecological Society, Journal of Applied Ecology, 46, 976–982
982 S. Jennings & R. W. Wilson
per recruit and total carbonate production of a population at
steady state is influenced by fishing. WCmax can be calculated
without a natural mortality for the species concerned, by tak-
ing advantage of the M ⁄ K ratio, and thus provides a simple
method for assessing the relative effects of fishing on carbonate
production. The disadvantage of this simple technique over
age-structured assessment of C ⁄ R is the assumption of a single
value of M in all age classes, when fish typically exhibit higher
M when younger. If size or age-related natural mortality data
are available for some well-studied populations, our analyses
show that it would be straightforward to modify existing popu-
lation assessment methods to predict the effect of various rates
of fishing mortality on Ctotp and C ⁄ R.
This analysis suggests that fishing will alter the rates of car-
bonate production by fish populations and communities, but
the analysis would be refined by significant additional research,
to include (i) obtaining carbonate production data for a wider
range of species, body sizes and temperatures to better para-
meterise the model linking these variables, (ii) accounting for
differences in the relative activity levels of fishes in the carbon-
ate production model and in the predictions of changes in com-
munity structure, (iii) accounting for the effects of fishing on
carbonate production by the smaller size-classes of fish, and
(iv) accounting for changes in relative abundance of teleosts
and elasmobranchs in the model of fishing impacts. In addi-
tion, while the analyses provide a method for giving manage-
ment advice on the effects of fishing on an ecosystem service
other than food production, considerable work would still be
needed to identify realistic management objectives for this
service. Such objectives would ultimately be a matter of choice
for society, albeit informed by science (Jennings 2007). Our
understanding of the wider consequences of changes in the
rates of fish carbonate production on ocean chemistry and the
consequences for biota will need to be improved to inform any
debate on objectives.
Acknowledgements
We thank Andrew Clarke for providing the compilation of teleost oxygen con-
sumption data from Clarke & Johnston (1999) and John Pope for allowing us
to modify the size-based model of Pope et al. (2006) for this analysis. S.J. thanks
UK DFID (formerly ODA) and NERC for funding the collection of the fish
community data used in this analysis, and the EC and Defra for funding this
research. R.W. thanks BBSRC and The Royal Society for funding fundamental
studies on intestinal carbonate production in marine fish.
References
Beverton, R.J.H. (1992) Patterns of reproductive strategy parameters in some
marine teleost fishes. Journal of Fish Biology, 41(Suppl. B), 137–160.
Bianchi, G., Gislason, H., Graham, K., Hill, L., Jin, X., Koranteng, K.,
Manickchand-Heileman, S., Paya, I., Sainsbury, K., Sanchez, F. & Zwanen-
burg, K. (2000) Impact of fishing on size composition and diversity of
demersal fish communities. ICES Journal of Marine Science, 57, 558–571.
Clarke, A. & Johnston, N.M. (1999) Scaling of metabolic rate with body
mass and temperature in teleost fish. Journal of Animal Ecology, 68,
893–905.
 2009 The Authors. Journal compilation  2009 British Ecological Society, Journal of Applied Ecology, 46, 976–982
Gillooly, J.F., Charnov, E.L., West, G.B., Savage, V.M. & Brown, J.H.
(2002) Effects of size and temperature on developmental time. Nature,
417, 70–73.
Glazier, D.S. (2005) Beyond the ‘3 ⁄ 4-power law’: variation in the intra-
and interspecific scaling of metabolic rate in animals. Biological Reviews, 80,
611–662.
Gonzalez, R. J. & McDonald, D.G. (1992) The relationship between oxygen
consumption and ion loss in a freshwater fish. Journal of Experimental Biol-
ogy, 163, 317–322.
Grosell, M. (2006) Intestinal anion exchange in marine fish osmoregulation.
Journal of Experimental Biology, 209, 2813–2827.
ICES (2007) Report of the Herring Assessment Working Group for the Area
South of 62 N. International Council for the Exploration of the Seas,
Copenhagen, CM 2007 ⁄ ACFM: 11.
Jennings, S. (2007) Reporting and advising on the effects of fishing. Fish and
Fisheries, 8, 269–276.
Jennings, S. & Polunin, N.V.C. (1997) Impacts of predator depletion by fishing
on the biomass and diversity of non-target reef fish communities. Coral
Reefs, 16, 71–82.
Jennings, S., Mélin, F., Blanchard, J.L., Forster, R.M., Dulvy, N.K. & Wilson,
R.W. (2008) Global-scale predictions of community and ecosystem proper-
ties from simple ecological theory. Proceedings of the Royal Society B, 275,
1375–1383.
Kerr, S.R. (1982) Estimating the energy budgets of actively predatory fishes.
Canadian Journal of Fisheries and Aquatic Science, 39, 371–379.
Nilsson, S. (1986) Control of gill blood flow. Fish Physiology: Recent Advances
(eds S. Nilsson & S. Holmgren), pp. 87–101. Croom Helm, London.
NOAA (2007) Coral Reef Watch Program. NOAA National Environmental
Satellite Data and Information Service, Fiji-Beqa Sea Surface Temperature
Time Series Charts Coral Reef Watch Program. Camp Springs, MD, USA.
Pope, J.G., Rice, J.C., Daan, N., Jennings, S. & Gislason, H. (2006) Modelling
an exploited marine fish community with 15 parameters: results from a sim-
ple size-based model. ICES Journal of Marine Science, 63, 1029–1044.
Quinn, T.J. & Deriso, R.B. (1999) Quantitative Fish Dynamics. Oxford Univer-
sity Press, New York.
Shin, Y-J., Rochet, M-J., Jennings, S., Field, J. & Gislason, H. (2005) Using
size-based indicators to evaluate the ecosystem effects of fishing. ICES Jour-
nal of Marine Science, 62, 384–396.
Sinclair, M. & Valdimarsson, G. (eds) (2003) Responsible Fisheries in the Mar-
ine Ecosystem, FAO, Rome.
Stevens, J.D., Bonfil, R., Dulvy, N.K. & Walker, P.A. (2000) The effects of fish-
ing on sharks, rays and chimaeras (chrondrichthyans), and the implications
for marine ecosystems. ICES Journal of Marine Science, 57, 476–494.
Takei, Y. & Tsukada, T. (2001) Ambient temperature regulates drinking and
arterial pressure in eels. Zoological Science, 18, 963–967.
Taylor, J.R. & Grosell, M. (2006) Feeding and osmoregulation: dual function
of the marine teleost intestine. Journal of Experimental Biology, 209, 2939–
2951.
Taylor, J.R., Whittamore, J.M., Wilson, R.W. & Grosell, M. (2007) Postpran-
dial acid-base balance in freshwater and seawater-acclimated European
flounder. Journal of Comparative Physiology B, 177, 597–608.
Walsh, P.J., Blackwelder, P., Gill, K.A., Danulat, E. & Mommsen, T.P. (1991)
Carbonate deposits in marine fish intestines: a new source of biomineraliza-
tion. Limnology and Oceanography, 36, 1227–1232.
Wilson, R.W. & Grosell, M. (2003) Intestinal bicarbonate secretion in marine
teleost fish - source of HCO3-, pH sensitivity, and consequences for whole
animal acid-base and Ca2+ homeostasis. Biochimica et Biophysica Acta,
1618, 163–174.
Wilson, R.W., Gilmour, K.M., Henry, R.P. & Wood, C.M. (1996) Intestinal
base excretion in the seawater-adapted rainbow trout: a role in acid-base bal-
ance? Journal of Experimental Biology, 199, 2331–2343.
Wilson, R.W., Wilson, J.M. & Grosell, M. (2002) Intestinal bicarbonate secre-
tion by marine teleost fish - why and how? Biochimica et Biophysica Acta,
1566, 182–193.
Wilson, R.W., Millero, F.J., Taylor, J.R., Walsh, P.J., Christensen, V., Jen-
nings, S. & Grosell, M. (2009) Contribution of fish to the marine inorganic
carbon cycle. Science, 323, 359–362.
Received 7 March 2009; accepted 9 June 2009
Handling Editor: Nick Dulvy