BSc thesis
INTERACTIVE EFFECTS OF TEMPERATURE
CHANGES AND HARVESTING BOTH
PISCIVORE AND FORAGE FISH ON FISH
COMMUNITY-DYNAMICS
Author: Esm´
ee van der Mark
Primary supervisor: Dhr. prof. dr. A.M. (Andr´
e) de Roos
Second supervisor: Dr. FH (Floor) Soudijn
Date and place: 6th of June 2021, Amsterdam
University of Amsterdam
Future Planet Studies
Abstract
Human global fish consumption has increased at an average annual rate of 3.1 percent from 1961 to 2017. As a result of increased fishing, the percentage of fish stocks that are within biologically sustainable levels has decreased from 90 percent in 1974 to 65.8 percent in 2017. Concomitantly, the mean trophic level of the global catches has decreased, a phenomenon referred to as ‘fishing down marine food webs’. The overfishing of fish at higher trophic levels can disrupt marine ecosystems. Next to overexploitation, rising temperatures threaten fish community dynamics. Global warming is predicted to increase temperature with 1.5°C above the pre-industrial levels between 2030 and 2052. Research has shown that rising temperatures affect metabolic and ingestion rates in fish. Fisheries quota are mostly set using single-species stock assessment models, to control exploitation and to optimize economic performance. However, in many ecosystems fishing occurs at multiple trophic levels simultaneously. Fishing may lead to surprising indirect effects in food chains, such as trophic cascades. Here, I investigate the interactive effects of temperature changes and harvesting at different trophic levels on fish community-dynamics to create robust management plans for fisheries, which will prepare them for the effects of global warming in the near future. I extended a multitrophic, size-structured, bioenergetics model of the fish community in the Baltic Sea with temperature dependent vital rates and parameters in the model. These rates and parameters included metabolism, maximum ingestion rate, resource turnover rate, maximum resource density and background mortality. This research shows that rising temperatures will negatively affect mean biomass levels of the piscivore population. A temperature rise of 2.0°C causes the piscivore population to go extinct, when the fishing mortality imposed on forage fish (Fs) is low. However,
when Fsis high, this does not happen until the piscivore fishing mortality (Fc) is higher than 0.6
year−1. The maximum sustainable yield (MSY) is also negatively affected by temperature rise, as the MSY is lower for higher temperatures. When Fsis low, the MSY is higher for every temperature
than the MSY values for high Fs. However, the MSY occurs at higher Fc values for high Fs, which
demonstrates the importance of investigating the interactive effects of temperature change and harvesting on different trophic levels. When resource productivity increases with temperature, the fish community-dynamics change as well, as this has a positive effect on the forage fish population and thereby also on the piscivore population, especially with the highest temperature (12°C). Finally, this research shows that the temperature dependence of resource productivity, which differs per region, is an important factor of how fish community-dynamics will change and thus important for fisheries management to incorporate in the future.
Contents
1 Introduction 4
2 Methods and Materials 9
3 Results 11 4 Discussion 17 5 Conclusion 21 6 Reference list 22 7 Acknowledgements 24 8 Appendix 1 25 8.1 Appendix 1.1 . . . 25 8.2 Appendix 1.2 . . . 26 8.3 Appendix 1.3 . . . 27 8.4 Appendix 1.4 . . . 28 9 Appendix 2 29 9.1 Appendix 2.1 . . . 29 9.2 Appendix 2.2 . . . 30
1
Introduction
Global fish consumption, which refers to fish destined for human consumption, has increased at an average annual rate of 3.1 percent from 1961 to 2017 (FAO, 2020). This is almost twice the rate of the annual world population growth for the same period. Through the 1970s and 1980s, it became clear that many fish stocks were overfished and depleted to low levels. In the late 1960s and early 1970s, all the northeast Atlantic herring stocks had declined or even collapsed, the Peruvian anchoveta collapsed in 1971 to 1972 and the Newfoundland cod fishery collapsed in the 1990s (Hilborn et al., 2020; Jakobsson, 1985; Myers, Hutchings, & Barrowman, 1996; Pauly et al., 2002). Collapsed fish stocks have a biomass that produces less than 20% of the maximum sustainable yield, which therefore need a longer time to recover than other overfished stocks (Hilborn et al., 2020). According to the FAO (2020), the percentage of fish stocks that are within biologically sustainable levels has decreased from 90 percent in 1974 to 65.8 percent in 2017, which means that 34 percent of the global fish stocks is classified as being overfished (M¨ollman & Diekmann, 2012).
Especially the overfishing of predatory, or piscivorous, fish has been a growing threat to marine ecosystems (M¨ollman & Diekmann, 2012). Overexploitation of large piscivores, which are at the top of the trophic system, can cause population collapses with cascading effects on lower trophic levels in the food web (G˚ardmark et al., 2015). The spawning-stock biomass of the Atlantic cod in the North Sea has declined with approximately 80 percent, from a maximum of 253.000 tonnes in 1971 to 50.000 tonnes in 2008 (Fauchald, 2010). Such a significant decline can cause population collapses, as the mortality has been above the limit that is considered to give a considerable risk of collapse (Fauchald, 2010). The Baltic Sea has also experienced an accelerated cod decline due to unsustainable cod fishing in the late 1980s which changed the fish community dynamics (M¨ollman & Diekmann, 2012).
Not only do fisheries target piscivores, they also target smaller species which serve as a food source for these large piscivores, called forage fish (Soudijn, Denderen, Heino, Dieckmann, & De Roos, 2021). Initially, fisheries tends to remove large, slower-growing fishes first, which reduces the mean trophic level of the remaining fish in the ecosystem (Pauly et al., 2002). A good example is the North Atlantic, where the biomass of the predatory fishes has declined with two-thirds, even though this area was already depleted before this time period (Pauly et al., 2002). Subsequently, the fisheries starts to harvest fish on lower trophic levels, a phenomenon called ‘fishing down marine food webs’ (Pauly et al., 2002). ’Fishing down marine food webs’ can disrupt a whole ecosystem, as
the larger individuals of species are responsible for a large part of the reproduction and predators are important for the structure of the food webs (Pauly et al., 2002).
A second threat to the marine ecosystems and their fish community-dynamics are rising temperatures. According to Lindmark, Ohlberger, Huss and G˚ardmark (2019), metabolic rates and also other ecological traits such as feeding, mortality and population growth rate, depend strongly on body size and temperature. Research has shown that metabolic rates increase with rising temperatures, while ingestion rates increase more slowly (Uszko, Diehl, Englund, & Amarasekare, 2017). Especially for larger piscivores, the ingestion efficiency (ingestion relative to metabolism) decreases. This causes a bottleneck in the adult stages of the piscivore populations, which can have disruptive effects on marine ecosystems (Soudijn et al., 2021; Uszko et al., 2017).
To protect fish stocks from declining and collapsing and thereby also protect the fisheries depending on the fish stocks, fisheries management is needed. Fisheries quota are used in fisheries management, to control exploitation and to optimize economic performance (Morgan, 1997). In some fisheries, quota arrangements have been in place for more than 60 years. (Morgan, 1997). These quota are mostly set using single-species stock assessment models (Pauly et al., 2002). These stock assessment models predict future recruitment to the stock and thereby establish quota.
Quota are not only imposed for piscivores, but also for forage fish. One of the concerns following the ’fishing down marine food webs’ phenomenon is that fishing on lower trophic levels has a direct, negative effect on forage fish density and can hence be expected to harm population growth of piscivores. However, the recent study of Soudijn et al. (2021) demonstrated that this direct negative effect does not always prevail and that harvesting forage fish can have indirect positive effects on piscivores. This finding is supported by observations from the Baltic Sea (see Fig. 1), where low densities of forage fish (sprat, herring) had indirect positive effects on the piscivore (cod) population, due to reduced competition between forage fish (Van Leeuwen, De Roos, & Persson, 2008). In contrast, following the overexploitation of the cod population, forage fish density increased and high intraspecific competition between forage fish emerged (De Roos & Persson, 2013; Soudijn et al., 2021; Van Leeuwen et al., 2008). This competition resulted in decreased growth and reproduction of sprat and herring, which lead to a decline of the availability of newborn forage fish (Soudijn et al., 2021; Van Leeuwen et al., 2008). This limited the capacity of cod to recover and reach its former high densities and thus remained in its current regime (G˚ardmark et al., 2015). In this situation, intermediate fishing pressure on forage fish decreases competition and increases availability of newborn forage fish and therefore can prevent a fishing-induced collapse on cod,
which illustrates the indirect positive effects of harvesting forage fish on the piscivore population (De Roos & Persson, 2013; Soudijn et al., 2021).
Figure 1: The total fecundity of the forage fish (herring and sprat together are called clupeids) and the total cod biomass. Since the 1970s, the total fecundity of clupeids steadily declined, which is a sigh of high intraspecific
competition between the forage fish. Reprinted from Van Leeuwen et al. (2008).
Soudijn et al. (2021) used a multispecies community-dynamics model for the fish community in the Baltic Sea (see Fig. 2), to investigate the effects of fishing on both forage fish and piscivores. The indirect positive effects of harvesting forage fish on the cod population depends on a phenomenon called ’biomass overcompensation’ (De Roos & Persson, 2013; Soudijn et al., 2021). Biomass overcompensation occurs when the biomass in certain size ranges of fish increases when mortality increases, as a consequence of the relaxation of intraspecific competition. Crucial for biomass overcompensation to occur in the model of Soudijn et al. (2021) is that this model accounts for ontogenetic development and in particular food-dependent growth in body size (De Roos & Persson, 2013; Persson, Van Leeuwen, & De Roos, 2014). Other multispecies models are based on the Lotka-Volterra models, which use population abundance to characterize a population and describe dynamics as the resultant of reproduction and mortality only (De Roos & Persson, 2013). These
models ignore ontogenetic development, and thus assume that the individual body conditions are constant and that the growth processes of an individual are not linked to food availability.
Figure 2: The interactions in the multispecies model of the Baltic Sea fish community. This multispecies model of the Baltic Sea incorporates essential components of fish community dynamics, such as individual-level energy budgets, size-dependent feeding interactions and size-structured fish populations. The Baltic Sea model includes both
exploited forage fish (herring and sprat) and piscivores (cod). Herring and sprat are assumed to have a similar ecological role, thus they are modeled together as a clupeid population (Soudijn et al., 2021). The model makes a distinction between juvenile, small-adult and large-adult stages for both the clupeids and cod, which is an example of
a size-structured component of this model. These different stages of clupeids and cod also have size-dependent feeding interactions, as displayed in the figure with the lines between different stages of clupeids and forage fish.
Reprinted from Soudijn et al. (2021).
Since rising temperatures affect fish community dynamics it is important to include these effects in fisheries models and management. The model analyzed by Lindmark et al. (2019) provides an example of a dynamic stage-structured biomass model with temperature dependent vital rates, that generates predictions about how a tri-trophic fish community responds to rising temperatures. Not only did the study of Lindmark et al. (2019) find that a warmer environment reduced the scope for predator persistence, because energetic costs are higher for larger organisms, but they also found that warming can cause predator populations to collapse due to Allee effects and alternative stable states (Lindmark et al., 2019).
this research is to answer the following research question: What are the interactive effects of temperature changes and harvesting at different trophic levels on fish community-dynamics? This question will be addressed by combining the model of Soudijn et al. (2021) with the temperature-dependent model of Lindmark et al. (2019), to gain even better insight into population responses to climate warming. According to IPCC (2019), human activities have caused approximately 1.0°C of global warming above pre-industrial levels. Global warming is likely to reach 1.5°C between 2030 and 2052, which underlines the importance of investigating the temperature dependence of Lindmark et al. (2019) in the more specific model of Soudijn et al. (2021) (see the ’Method and Materials’ section for the explanation of why this model is more specific). Fish and fishery products are furthermore some of the most internationally traded food commodities in the world (FAO, 2020). In 2018, 38 percent of the total aquaculture and fisheries production, which equals 67 million tonnes, were traded internationally (FAO, 2020). As for consumption in 2017, fish accounted for 17 percent of the global population’s intake of animal proteins and in the coastal developing countries fish provide as much as 50 percent of animal protein consumption (FAO, 2020; Free et al., 2021). Research has also shown that in regions with little fisheries management, stocks are also in poor shape (Hilborn et al., 2020). The same research also shows that in regions where stocks are assessed, the fish populations are increasing and above target levels or rebuilding (Hilborn et al., 2020). This research can thus help improving fisheries management, by creating an even better community-dynamics model including both ontogenetic development and the effects of rising temperatures, which will help to protect the future of marine fisheries (Mullon, Fr´eon & Cury, 2005).
2
Methods and Materials
The community-dynamics model of the Baltic Sea of Soudijn et al. (2021) (from here on referred to as the Soudijn model), has been extended with the temperature-dependent model of Lindmark et al. (2019), (from here on the Lindmark model). The modelling has been done in Python (version 3.8.2288.0) , using Visual Studio Code (version 1.54.3) as code editor. The Soudijn model has been used as a base, for multiple reasons. First of all, the food web of the Baltic Sea is relatively simple, but nonetheless includes the key ecological interactions between piscivore (cod) and forage fish (herring and sprat, together called clupeids) and their resource (Soudijn et al., 2021). Secondly, this tri-trophic fish community model has size-based, stage-specific parameterization, and makes a distinction between juvenile, small adults and large adult piscivores and clupeids. Research has shown that, when distinguishing between juvenile and adult predators, new and more realistic insights are developed (Soudijn et al., 2021). The tri-trophic temperature-dependent Lindmark model is also stage-structured for consumers and accounts for a stage-selective predator, but the predator population is unstructured (Lindmark et al., 2019). This means that the dynamics of the predator population are only determined by its net biomass production and background mortality (Lindmark et al., 2019). Furthermore, the Soudijn model implements pulsed reproduction, which makes it a more realistic starting point for extending. This, and the implementation of a structured piscivore population, are the reasons that the Soudijn model forms the base and the Lindmark model has been used as an extension.
The article of Soudijn et al. (2021) used a C-based simulation program to carry out the analysis. In this research I recreated the Soudijn model in Python, using the supporting information of Soudijn et al. (2021). The model is defined in terms of biomasses per volume and consists of 13 ordinary differential equations (ODEs), which describe the continuous-time dynamics over time during the growing season. The Soudijn model is based on the bioenergetics approach of Yodzis and Innes (1992), which means that assimilated energy is first used to cover maintenance costs (Soudijn et al., 2021). If there is energy left over after covering maintenance costs, the net-energy production is used for either somatic growth and/or reproduction, depending on the life stage. To account for these differences between life stages, the parameter κ is included in the clupeid and cod ODEs, which is the fraction of energy invested in somatic growth (see Appendix 1.1 for a more detailed description of the model, including parameter κ). During the growing season, which lasts 250 days, the energy investments into reproduction are stored in the reproduction buffer. After these 250 days, the reproduction season starts and all the biomass stored in the reproductive buffer
is converted to juvenile biomass. See Appendix 1.2 for the validation steps which have been taken to check this model.
After the Soudijn model was implemented and its correctness validated, the temperature dependency of the Lindmark model was added to the implementation. In the Lindmark model, temperature dependency is represented using an Arrhenius term, see equation 1. The Lindmark model models the following vital rates and parameters as temperature dependent: metabolism (M), maximum ingestion rate (Imax), background mortality (µ), resource turnover rate (δ) and
the maximum density of the basal resource (Rmax). To implement this temperature dependency
in the Soudijn model, the vital rates and parameters are multiplied with a factor such as shown in equation 1. Also, to account for the differences of temperature dependency between life stages (temperature affects larger individuals more than smaller individuals), an interactive effect between temperature and body size was implemented parameterised by the parameter c (see Appendix 1.3 on further explanation on this parameter c and of equation 1).
rY(T ) = e
EY (T −T0)
kT T0 (1)
The model was analyzed by creating multiple bifurcation graphs, for different bifurcation parameters. In terms of my model, which is a community-dynamics model by means of numerical integration of the ODEs, a bifurcation graph shows how the outcomes will change while varying one particular parameter (bifurcation parameter) in a step-wise manner. This will be done by both increasing and decreasing values of the bifurcation parameter, which allows to locate alternative stable states. For more details about bifurcation graphs I refer to Box 3.5 in De Roos & Persson (2013). In this research, temperature is used as bifurcation parameter (from 6◦C to 20◦C) for both low and high clupeid mortality (Fs), to see whether the results of the Soudijn model still apply with
higher temperatures. Secondly, cod fishing mortality (Fc) has been used as bifurcation parameter as
well, for low and high Fsand for 4 different temperatures, to see how the fish community-dynamics
react to changing Fc values with different temperatures. See Appendix 1.4 for the parameter values
used for these bifurcation graphs. Lastly, bifurcation plots have been made to show the mean total annual yield of cod (all stages except for juveniles) and clupeids (all stages including juveniles), with Fc as bifurcation parameter as well.
3
Results
For high temperatures and low clupeid fishing mortality Fs, cod goes extinct even at low values of
the fishing mortality Fc. On the other hand, for low temperatures, cod can sustain much higher Fc
values. Temperatures lower than the reference temperature of 10◦C (thus 6.5 and 7.7◦C) allow cod to persist for a longer period and not reach extinction even at the highest Fc value. Temperature
affects the cod population in different ways, depending on the value of Fc (see Fig. 3). When the
Fc values are low (< 0.2 year−1 approximately) the total biomass of adult cod reaches higher levels
with higher temperatures than for lower temperatures. However, when Fc values are higher (> 0.2
year−1), this changes and temperature increases have a negative effect on the adult cod biomass. With high clupeid fishing mortality, cod can persist for higher Fc values than with low
clupeid fishing mortality as was also reported by Soudijn et al. (2021) (see Fig. 3). In addition, with high clupeid fishing mortality, cod adult biomass levels are higher at high Fc values than with
low clupeid fishing mortality. These effects occur for all temperatures. The positive effect of Fs on
cod are particularly showing in the plots of the highest temperature (12◦C). When there is low-level clupeid fishing, the cod population goes extinct for all Fc values. However when there is high-level
clupeid fishing, the cod population goes extinct only at Fc values higher than approximately 0.6
year−1. Before that the cod population persists and even reaches high cod adult biomass values when the Fc levels are lower than 0.2 year−1 (see Fig. 3).
Clupeid adult biomass is lower for high Fs, while clupeid juvenile biomass is equal or higher
for high Fs. This effect is most pronounced for high temperatures (12◦C) (see Fig. 4). Secondly,
considering the clupeids (see Fig. 3, second and third row), it shows that clupeids react for the most part the same as cod, where both juvenile an adult clupeid biomass are lower with higher temperatures. Also, the density of the clupeid resource decreases with temperature (see Fig. 3, fourth row and Fig. 4). Figure 4 shows the bifurcation graph with temperature as the bifurcation parameter, where Fc is 0.75 year−1. For this Fc value, cod goes extinct around 9◦C. When this
happens, adult clupeid biomass increases, but juvenile clupeid biomass decreases. This decrease shows the biomass overcompensation, which I mentioned in the introduction. In both regimes (low and high Fs), however (with and without cod) juvenile and adult clupeid biomass decreases with
temperature. In figure 4 the bistability is shown as well, as you can see that the plots where the bifurcation parameter (temperature) is scanned from low to high temperatures (solid blue lines), is different from the plots where it is scanned from high to low temperatures (orange dashed lines).
0.00
0.01
0.02
Cod adult
bi
om
as
s (
kg
/m
3
)
Low-level clupeid fishing
High-level clupeid fishing
6.5 °C
7.7 °C
10 °C
12 °C
0.2
0.3
0.4
Clupeid adult bio
ma
ss
(kg
/m
3
)
0.02
0.03
0.04
Clupeid juvenile bi
om
as
s (
kg
/m
3
)
0.00 0.25 0.50 0.75 1.00 1.25
0.05
0.10
0.15
0.20
Resource for clupeids
bi
om
as
s (
kg
/m
3
)
0.00 0.25 0.50 0.75 1.00 1.25
Cod fishing mortality F
c
(year
1
)
Figure 3: Mean biomass of adult cod, adult clupeid, juvenile clupeid and resource for clupeids, with ERmax= −0.43.
Included in adult cod biomass are the the small adults (Ca), the large adults (Cb) and the reproductive buffers of the small adults (Cga) and large adults (Cgb). Same applies for the clupeid adult biomass. Low Fs= 0.2 year−1and
0.000
0.005
0.010
0.015
Cod adult
bi
om
ass
(k
g/m
3
)
Low-level clupeid fishing
High-level clupeid fishing
low to high T
high to low T
0.2
0.3
0.4
Clupeid adult bio
ma
ss
(kg
/m
3
)
0.01
0.02
0.03
0.04
Clupeid juvenile bio
ma
ss
(kg
/m
3
)
7.5
10.0 12.5 15.0 17.5 20.0
0.05
0.10
Resource for clupeids
bi
om
ass
(k
g/m
3
)
7.5
10.0 12.5 15.0 17.5 20.0
Temperature (in °C)
Figure 4: Mean biomass of adult cod, adult clupeids, juvenile clupeids and resource for clupeids, for temperature ranging from 6◦C to 20◦C. Cod fishing mortality is high, Fcis 0.75 year−1. The blue solid lines represent the bifurcation plots in which temperature was gradually increased in a step-wise manner (low to high T), while the orange dashed lines represent the bifurcation plots in which temperature was gradually decreased (high to low T). Included in adult cod biomass are the the small adults (Ca), the large adults (Cb) and the reproductive buffers of the
Adult and juvenile clupeids react differently to higher Fc levels. As long as cod persists,
adult clupeid biomass increases with Fc, due to less predation by cod (see Fig. 3, second row).
However, juvenile clupeid biomass shows a hump-shaped pattern, first increasing with Fc and after
that decreasing until cod goes extinct (see Fig. 3, third row).
Furthermore, higher temperatures cause the annual mean cod yield to be lower (Fig. 5) and the yield reaches 0 sooner than for lower temperatures. The maximum sustainable yield (MSY) lies at higher Fc values for lower temperatures. For high Fs values the MSY is lower and located at
higher Fc values than for low Fsvalues. This effect is independent of temperature, except for the
scenario at 12◦C, as the cod population is here extinct with low Fs and the yield is therefore at
this temperature always higher for high Fs. See Appendix 2.1 for the annual mean yield of clupeids.
Figure 6 shows the effect of ERmax on the fish community dynamics. The default value of ERmax = −0.43, which implies that with increasing temperatures the total productivity of the resources remains unchanged. When ERmax = 0, the total productivity of the resources increases with temperature and hence energy supply to the fish community increases at the same time that energetic costs (maintenance and mortality) increase. At the reference temperature (10◦C), changing the ERmax so resource productivity increases with temperatures, does not have any effect on resource productivity, because resource productivity stays the same for both ERmax values. For lower and higher temperatures, however, resource productivity is lower and higher, respectively when ERmax = 0.0. For ERmax = 0.0 , temperature does no longer have a strong effect on the persistence of cod (compare Fig. 6 and Fig. 3). Yet, there is still some effect of temperature. Cod is positively affected by higher temperatures, as the mean biomass is higher than the colder temperatures for low up until medium Fc levels. Higher temperatures are also more beneficial for
adult and juvenile clupeids in this situation, as the biomass levels are higher compared to when ERmax= −0.43. In the situation of high clupeid fishing, the adult clupeid population even has the highest biomass of all the temperatures when Fc is high (> 0.75 year−1). The effect of Fs on the
persistence of cod is independent of the value of ERmax, as cod still persists for higher values of Fc with higher Fs. Moreover, the effect of Fson the persistence of cod has become stronger for
ERmax= 0 with 12
◦C. While cod goes extinct at a cod fishing mortality value of 0.6 year−1 for
ERmax= −0.43, cod does not go extinct at all for ERmax= 0.0 with high clupeid fishing mortality. For the two lowest temperatures (6.5◦C and 7.7◦C) however, cod does go extinct at Fc values of
0.9 year−1 and 0.75 year−1 respectively (for low Fs), which did not happen for ERmax= −0.43. Lastly, whether or not an interactive effect between temperature and body size is present
does not have major effects on the community-dynamics (see Appendix 2.2).
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Low-level clupeid
fishing
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
High-level clupeid
fishing
6.5 °C
7.7 °C
10 °C
12 °C
Cod fishing mortality F
c
(year
1
)
Co
d
m
ea
n
an
nu
al
y
ie
ld
(k
g/
m
3
)
Figure 5: Cod annual mean yield for low and high clupeid fishing mortality’s. Low Fs = 0.2 year−1and high Fs = 0.5 year−1.
0.00
0.01
0.02
Cod adult
bi
om
as
s (
kg
/m
3
)
Low-level clupeid fishing
High-level clupeid fishing
6.5 °C
7.7 °C
10 °C
12 °C
0.2
0.3
Clupeid adult bio
ma
ss
(kg
/m
3
)
0.020
0.025
0.030
0.035
0.040
Clupeid juvenile bi
om
as
s (
kg
/m
3
)
0.00 0.25 0.50 0.75 1.00 1.25
0.05
0.10
0.15
Resource for clupeids
bi
om
as
s (
kg
/m
3
)
0.00 0.25 0.50 0.75 1.00 1.25
Cod fishing mortality F
c
(year
1
)
Figure 6: Mean biomass of adult cod, adult clupeid, juvenile clupeid and resource for clupeid, when resource productivity increases with temperature (ERmax= 0.0). Included in adult cod biomass are the the small adults (Ca),
the large adults (Cb) and the reproductive buffers of the small adults (Cga) and large adults (Cgb). Same applies for the clupeid adult biomass. Low Fs= 0.2 year−1 and high Fs= 0.5 year−1
4
Discussion
The results presented above show that there are several interactive effects of temperature changes and harvesting at different trophic levels on the fish community. First of all, warming in general negatively affects biomass levels, and most importantly cod goes extinct at lower Fc values for
higher temperatures. With high clupeid fishing mortality, cod can persist for higher Fc values than
with low Fs. This is particularly showing at 12◦C, where cod goes extinct for all Fc values when
there is low-level clupeid fishing, but for high-level clupeid fishing only goes extinct for Fc values
higher than 0.6 year−1. When Fs is high, cod adult biomass is even the highest for 12◦C for low
Fc values (< 0.2). Furthermore, clupeid adult biomass is lower for high Fs, while clupeid juvenile
biomass is equal or higher for high Fs, which is most pronounced for high temperatures (12◦C).
The annual mean cod yield is lower with higher temperatures and the MSY is also higher for low Fs. For high Fs the MSY is lower, however it is located at higher Fc levels than for low Fs. When
resource productivity increases with temperature as well (ERmax = 0.0), adult cod is positively affected by higher temperatures (for low up until medium Fc levels). High clupeid fishing mortality
causes the biomass of adult clupeids to be higher than the other temperatures with high Fc levels.
Other studies support the finding that in general warming has a negative effect on biomass levels and thus decreases the persistence of cod at higher temperatures. Cod extinction at lower Fc values with higher temperatures was already the case in the original Lindmark model, which
thus did not change with the Soudijn model as a basis. This is also supported by the finding of Uzsko et al. (2017), where with lower temperatures the predator’s ingestion and growth rates first increased with warming more strongly than the metabolic losses and mortalities, however this eventually decreased with higher temperatures. Not only Lindmark et al. (2019), but also other articles found negative effects of rising temperature on biomass levels, such as Barneche et al (2021). According to Barneche et al. (2021), the mechanism behind this effect is that warming decreases the efficiency of energy transfer between trophic levels. The efficiency of energy transfer from resources to consumers determines the biomass structure of food webs in ecosystems (Barneche et al., 2021). Due to warming, this efficiency decreases by up to 56 % with a 4◦C rise, which means that less energy can be transferred to higher trophic levels. Consequently, this leads to relatively low biomass of upper trophic levels (Barneche et al., 2021). This lower energy transfer efficiency can be further explained by the higher costs of growth with higher temperatures. According to Barneche, Jahn and Seebacher (2019), the cost of growth (Em), which is the amount of energy
temperatures and higher trophic levels (Barneche & Allen, 2018). This means that an increasing proportion of the carbon fixed by photosynthesis will be lost to the atmosphere as temperatures rise, and thus less carbon will be retained in the food chain and ecosystem (Barneche et al., 2021).
Another phenomenon behind lower biomass with higher temperatures, is ‘the temperature-size rule’ (TSR). The TSR refers to the observations in a wide range of organisms that increasing temperatures often display smaller size at maturity and reduction in maximal body size (Van Rijn, Buba, DeLong, Kiflawi, & Belmaker, 2017). The research of the original Lindmark model, indeed showed that with warming there is a decline in body size (Lindmark et al., 2019). According to Van Rijn et al. (2017), oxygen supply is an important mechanism behind this, as oxygen supply has been shown to limit maximal body size of aquatic organisms. When individuals (fish) grow, their gill surface area to body mass ratio decreases, which means that at a certain size they can no longer acquire the oxygen needed for maintaining their metabolic demands (Van Rijn et al, 2017). This oxygen limitation can happen at smaller body sizes with higher temperatures, as this increases metabolism and thus oxygen demand increases as well, which means that metabolic demands are higher than oxygen supply sooner than with lower temperatures. See figure 7 for the relationship between body mass and oxygen consumption (left), and the relationship between temperature and oxygen consumption (right). Furthermore, Van Rijn et al. (2017) show that especially active larger species experience the most reduction in body size in response to increasing temperatures. This can be explained with the fact that active species have in general a higher metabolic rate than resting (less active) species, and thus the oxygen demand is more temperature dependent in these active
species.
Our results still have the same effect reported by Soudijn et al. (2021), where high Fs has a
positive effect on the persistence of cod to higher Fc levels and this effect is even stronger with
a 2◦C temperature rise. The effect of biomass overcompensation, which has been explained in the introduction, is thus still showing in the extended model analysed here. Cod biomass is even higher than all the other temperatures, with high Fsand Fc value lower than 0.2 year−1. This can
be explained by the fact that metabolism increases with temperature, and this effect is stronger than low Fc values, so biomass is higher than other temperatures at that point. Finding the right
balance between Fsand Fc will thus be very important for future fisheries and management plans,
as temperature increases even strengthen the effect.
Whether the maximum density of resource declines with temperature or is temperature-independent, and resource productivity hence remains constant or increases with temperature,
Figure 7: Zebrafish resting rates of oxygen consumption with respect to body mass (left) and temperature treatment (right). Reprinted from Barneche et al. (2019).
influences the cod persistence to such an extent that this should be taken into account for future research. Changing the activation energy of ERmax from -0.43 to 0.0, means that the resource productivity increases with temperature. This immediately changes the fish community-dynamics, as cod persists for much higher Fc levels and also the clupeids (both juvenile and adult) benefit
from it. This effect is most pronounced for the highest temperature (12◦C), where biomass levels are much higher than before. The resource for clupeids does not reach low biomass levels up until high Fc levels, which has a direct positive effects on the clupeid population, and especially on
the adult clupeids. The importance of resource availability to fish community-dynamics is also supported by other articles. According to O’Connor, Piehler, Leech, Anton and Bruno (2009), warming temperatures strengthen consumer control of primary production when resource biomass scales with temperature as well. This supports our finding of the higher adult clupeid biomass for the highest temperature. Whether resource biomass scales with temperature and the trophic structure is affected, depends on the nutrient levels of the region (O’Connor et al., 2009). If a region is nutrient-poor, the research of O’Connor et al. (2009) suggest that food webs may be more resilient to warming because consumer production is limited by resource availability. In contrast, they suggest that nutrient-rich regions small amounts of warming may have dramatic
effects on primary productivity, standing biomass and trophic structure (O’Connor et al. ,2009). In these nutrient-rich regions, warming should increase productivity and biomass in higher trophic levels. This causes shifts in food web structures and there will be a stronger consumer control of phytoplankton standing stock (O’Connor et al., 2009). Following our results about the productivity of resource, and also following the research of O’Connor et al. (2009), it will be important for fisheries in the future to investigate more extensively the effect of temperature rise on resource productivity, as there will be differences between resource productivity in different regions.
Thus, biomass levels decline with rising temperatures and the persistence of cod to cod fishing mortalities as well. This study emphasizes that fisheries have to create a good balance between forage fish fishing mortality and piscivore fishing mortality, as this research shows that a good balance between Fsand Fc can save cod from going extinct for all Fc levels with higher
temperatures. This tri-trophic, size-bases, bioenergetics model of the Baltic Sea fish community, extended with temperature dependency, can thus be used for future research about interactive effects of temperature rise and harvesting on different trophic levels on fish community-dynamics. This model can be useful for fisheries management and thus for species conservation. Suggestions to improve our model are specifying the temperature dependence parameters more than they are specified now. For example, the activation energy for the maximum resource density is the same for all three resources, so this could be more specific in the future, as these resources differ from each other. Also, fisheries should take into account that active large predatory species are most vulnerable to rising temperatures, which means that more future research should be done about which species are active and non-active. Thereby, fisheries can decide better whether they should adapt their fishing strategy in the future (Van Rijn et al., 2017). Also for future research, incorporating future oxygen levels may be important for predicting size-changes, as Van Rijn et al. (2017) showed that oxygen limitation is an important mechanism behind size changes.
5
Conclusion
Biomass levels and the persistence of cod to Fclevels are negatively affected by rising temperatures.
On the other hand, for low temperatures cod can sustain much higher Fc values. Cod goes extinct
for all Fc values with low Fs, but cod can persist for higher Fc values with high Fs. Adult clupeid
biomass is negatively affected by high Fs, while juvenile clupeid biomass is equal or higher for
high Fs than for low Fs. Also the resource for clupeids experiences lower biomass levels with
higher temperatures. The MSY of cod is higher for low Fs and although the MSY is lower for
high Fs, the MSY is located at higher Fc values. Furthermore, the temperature dependence of
resource productivity is important for the fish community-dynamics. When the activation energy of maximum resource density is equal to zero, which means that resource productivity scales with temperature, cod will persist for much higher Fc values, and also the clupeids biomass benefits from
it. This effect is most pronounced for 12◦C, where biomass levels are much higher than before, and this is especially the case for clupeid adult biomass which is higher than all temperatures for high Fc and high Fs.
Our finding of lower biomass levels with warming is supported by other articles, in which they refer to different mechanisms behind this effect of rising temperatures. According to Barneche et al. (2021), warming causes a decrease in energy transfer efficiency between trophic levels. This is caused by a higher cost of growth with higher temperatures, especially for large piscivores. Another article supporting our findings refers to the TSR as the underlying mechanism of lower biomass levels with higher temperatures (Van Rijn et al., 2017). An important factor behind this is oxygen limitation, which occurs at lower maximum body sizes with warming. This effect is stronger for large, active species which show the most reduction in body size in response to increasing temperatures. Finally, this research shows that temperature dependency of the maximum resource density influences the fish community-dynamics in such a way that it should certainly be taken into account for future fisheries management plans. According to O’Connor et al. (2009), whether resource productivity scales with temperatures depends on the nutrients availability of the region. For fisheries it will thus be important to research the nutrient availability of the region, to know how the fish community-dynamics will change in the future.
6
Reference list
Barneche, D. R., & Allen, A. P. (2018). The energetics of fish growth and how it constrains food-web trophic structure. Ecology letters, 21(6), 836-844.
Barneche, D. R., Jahn, M., & Seebacher, F. (2019). Warming increases the cost of growth in a model vertebrate. Functional Ecology, 33(7), 1256-1266.
Barneche, D. R., Hulatt, C. J., Dossena, M., Padfield, D., Woodward, G., Trimmer, M., & Yvon-Durocher, G. (2021). Warming impairs trophic transfer efficiency in a long-term field experiment. Nature, 592(7852), 76-79.
De Roos, A. M., & Persson, L. (2013). Population and community ecology of ontogenetic development. Princeton University Press.
FAO, (2020). The State of World Fisheries and Aquaculture 2020, Sustainability in Action, Food and Agriculture Organization of the United Nations, Rome Free, C. M., Thorson, J. T., Pinsky, M. L., Oken, K. L., Wiedenmann, J., & Jensen, O. P. (2019). Impacts of historical warming on marine fisheries production. Science, 363(6430), 979-983.
G˚ardmark, A., Casini, M., Huss, M., Van Leeuwen, A., Hjelm, J., Persson, L., & De Roos, A. M. (2015). Regime shifts in exploited marine food webs: detecting mechanisms underlying alternative stable states using size-structured community dynamics theory. Philosophical Transactions of the Royal Society B: Biological Sciences, 370(1659), 20130262.
Hilborn, R., Amoroso, R. O., Anderson, C. M., Baum, J. K., Branch, T. A., Costello, C., ... & Ye, Y. (2020). Effective fisheries management instrumental in improving fish stock status. Proceedings of the National Academy of Sciences, 117(4), 2218-2224.
IPCC, 2019: Global Warming of 1.5°C. An IPCC Special Report on the impacts of global warming of 1.5°C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty [Masson-Delmotte, V., P. Zhai, H.-O. P¨ortner, D. Roberts, J. Skea, P.R. Shukla, A. Pirani, W. Moufouma-Okia, C. P´ean, R. Pidcock, S. Connors, J.B.R. Matthews, Y. Chen, X. Zhou, M.I. Gomis, E. Lonnoy, T. Maycock, M. Tignor, and T. Waterfield (eds.)]
Jakobsson, J. (1985). Monitoring and management of the northeast Atlantic herring stocks. Canadian Journal of Fisheries and Aquatic Sciences, 42(S1), s207-s221.
Lindmark, M., Ohlberger, J., Huss, M., & G˚ardmark, A. (2019). Size-based ecological interactions drive food web responses to climate warming. Ecology letters, 22(5), 778-786.
M¨ollmann, C., & Diekmann, R. (2012). Marine ecosystem regime shifts induced by climate and overfishing: a review for the Northern Hemisphere. Advances in ecological research, 47, 303-347.
Morgan, G. R. (1997). Individual quota management in fisheries: methodologies for deter-mining catch quotas and initial quota allocation (No. 371). Food & Agriculture Org..
Mullon, C., Fr´eon, P., & Cury, P. (2005). The dynamics of collapse in world fisheries. Fish and fisheries, 6(2), 111-120.
Myers, R. A., Hutchings, J. A., & Barrowman, N. J. (1996). Hypotheses for the decline of cod in the North Atlantic. Marine ecology progress series, 138, 293-308.
O’Connor, M. I., Piehler, M. F., Leech, D. M., Anton, A., & Bruno, J. F. (2009). Warming and resource availability shift food web structure and metabolism. PLoS Biol, 7(8), e1000178.
Pauly, D., Christensen, V., Gu´enette, S., Pitcher, T. J., Sumaila, U. R., Walters, C. J., ... & Zeller, D. (2002). Towards sustainability in world fisheries. Nature, 418(6898), 689-695.
Persson, L., Van Leeuwen, A., & De Roos, A. M. (2014). The ecological foundation for ecosystem-based management of fisheries: mechanistic linkages between the individual-, population-, and community-level dynamics. ICES Journal of Marine Science, 71(8), 2268-2280.
Soudijn, F. H., van Denderen, P. D., Heino, M., Dieckmann, U., & De Roos, A. M. (2021). Harvesting forage fish can prevent fishing-induced population collapses of large piscivorous fish. Proceedings of the National Academy of Sciences, 118(6).
Uszko, W., Diehl, S., Englund, G., & Amarasekare, P. (2017). Effects of warming on predator–prey interactions–a resource-based approach and a theoretical synthesis. Ecology Letters, 20(4), 513-523.
Van Leeuwen, A., De Roos, A. M., & Persson, L. (2008). How cod shapes its world. Journal of Sea Research, 60(1-2), 89-104.
Van Rijn, I., Buba, Y., DeLong, J., Kiflawi, M., & Belmaker, J. (2017). Large but uneven reduction in fish size across species in relation to changing sea temperatures. Global Change Biology, 23(9), 3667-3674.
Yodzis, P., & Innes, S. (1992). Body size and consumer-resource dynamics. The American Naturalist, 139(6), 1151-1175.
7
Acknowledgements
I would like to thank my first supervisor, dr. prof. Andr´e de Roos, for helping me with modelling this fish community-dynamics model extended with temperature and also with writing my BSc thesis. Your insightful feedback pushed me to sharpen my thinking and brought my work to a higher level. I also really appreciated the good and fast communication. I have really developed my modelling skills and I now also know that I really want to proceed my studies in the ecology direction.
Secondly, I would like to thank my second supervisor, dr. Floor Soudijn, for helping me analyze my findings and for helping me write my abstract, results and discussion section. I also really appreciated the good and fast communication and also for checking in with me, which helped me to keep on track with my planning.
Third and last, I would like to thank Anne Uilhoorn, for answering my practical questions and also for giving feedback on my writing, which helped me a lot as well.
8
Appendix 1
8.1
Appendix 1.1
The Soudijn model consists of three resource ODEs: a resource for clupeids, a resource for cod juveniles and a resource for cod adults (Soudijn et al., 2021). These three unstructured resources are assumed to have a constant productivity and turnover rate and the growth thus depends on the grazing rate of their predators (Soudijn et al., 2021). Furthermore, the model has size-structured clupeid and cod populations, which means that both the clupeid and cod population is divided in five separate stages: one juvenile stage, two adult stages and their reproductive storages. For a complete description of the model I refer to the original paper by Soudijn et al. (2021).
As stated in the main text, assimilated energy is first used to cover maintenance costs. If the maintenance costs are higher than the assimilated energy, biomass is lost due to starvation mortality and no growth or reproduction occurs (Soudijn et al., 2021). If there is energy left after covering maintenance costs, the energy is invested in growth and/or reproduction, depending on the life stage. For both juvenile clupeid and cod, all net-energy production is used for somatic growth. Small adults use a part of their net-energy production for somatic growth and the remainder for reproduction. The large adult stage of both clupeid and cod allocate all energy to reproduction after covering maintenance costs (Soudijn et al., 2021). See table S1 for the exact values for parameter κ, which is the fraction of energy invested in somatic growth. If a life stage has a κ-parameter value of 0 this means that all the net-energy production will be invested in reproduction, which is the case for the large adult life stages of clupeid and cod.
Table S1: κ parameter values for the different life stages of clupeid and cod Clupeid Cod
Juvenile 1.0 1.0 Small adult 0.8 0.8 Large adult 0.0 0.0
8.2
Appendix 1.2
I took validation steps to check whether my model completely matches with the original model of Soudijn et al. (2021). I obtained a dataset from my primary supervisor, with biomass results from the original Soudijn model for specific parameter and initial values. I adjusted my model to these parameter values and initial values, and graphically compared the results from my model together with the results from the original model. See figure S1 for the validation graph of adult clupeid biomass, showing complete correspondence between the results of my model and the original model of Soudijn et al. (2021).
0
1000
2000
3000
4000
5000
Time (in days)
60
80
100
120
140
160
Biomass (g/Vol)
Adult clupeids biomass
My model
Soudijn model
Figure S1: Biomass of adult clupeids over time. The small adult (Sa), large adult (Sb) and both the reproductive storages (Sga and Sgb) are included in this graph. Low Fs= 0.2 year−1and high Fs= 0.5 year−1
8.3
Appendix 1.3
Equation 1 in the main text is the Boltzmann-Arrhenius function, where T is the temperature (in Kelvin), T0 is the reference temperature (in Kelvin), k is Boltzmann’s constant and EY is the
activation energy of parameter or rate Y (in eV K1). As reference temperature I adopted 10 ◦C
following Soudijn et al. (2021), which is 283 K. The Boltzmann’s constant is equal to 8.617332·E−5 eV K−1. As stated in the main text, these parameters/rates include metabolism (M ), maximum ingestion rate (Imax), background mortality (µ), resource turnover rate (δ) and the maximum
density of the basal resource (Rmax). These parameters/rates have different default activation
energies, as shown in table S2. Activation energy is the amount of energy needed for a certain reaction, for example metabolism. If the activation energy is 0 (which could be the case of ERmax),
this means that there is no effect of temperature on Rmax. If ERmaxis equal to -0.43, this means
that Rmax will be declining with temperature at the same rate as turnover rate δ increases (0.43).
The joint effect of a decrease in Rmax and an equal increase in turnover rate δ is that the total
resource productivity, which is the product of δ and Rmax, remains constant.
Table S2: Activation energy values EY for the different rate parameters in the model (in eV K1)
EM EI Eµ ERmax Eδ
0.594 0.594 0.45 -0.43/0 0.43
To account for different temperature dependencies for different life stages, an interaction effect between temperature and the body size of individuals in a particular life stage has been implemented, which is parameterised by the parameter c (see equation 2 and 3 below). A larger individual experiences more pronounced effects from temperature than smaller individuals. For juvenile clupeid or cod, nothing will change and thus a parameter such as metabolism will only be multiplied with the Arrhenius term (equation 1). However, for small adults and large adults (clupeid and cod), equation has been extended to include a body size dependent term. See equation 2 and 3 for the example of small and large clupeid adults (which is the same for cod). For small adults, the ratio between juvenile and small adult average body size is used, and for large adults, the ratio between juvenile and large adult average body size is used (these average size values can be found in the supporting information of Soudijn et al. (2021)). These ratios are raised to the power c(T − T0) with the parameter c set either to 0 or 0.005. When the parameter c is set equal
individuals. However, when the parameter c is set equal to 0.005, this means that larger individuals do experience stronger effects due to temperature change.
rY(T ) = e
EY (T −T0) kT T0 ·
Average Juvenile Clupeid size Average Small Adult Clupeid Size
c(T −T0)
(2)
rY(T ) = e
EY (T −T0) kT T0 ·
Average Juvenile Clupeid size Average Large Adult Clupeid Size
c(T −T0)
(3)
8.4
Appendix 1.4
Bifurcation graphs have been constructed with temperature as the bifurcation parameter and with cod fishing mortality (Fc) as the bifurcation parameter. For both bifurcation graphs, the model
has been run for both low and high clupeid fishing mortality (Fs). See table S3 for the exact Fs
values. In my model I have used the Fsvalues per day.
Table S3: Fs values
low Fs high Fs
per year 0.2 0.5 per day 0.0008 0.002
9
Appendix 2
9.1
Appendix 2.1
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
20
30
40
50
Low-level clupeid
fishing
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
High-level clupeid
fishing
6.5 °C
7.7 °C
10 °C
12 °C
Cod fishing mortality F
c
(year
1
)
Cl
up
ei
d
m
ea
n
an
nu
al
y
ie
ld
(k
g/
m
3
)
Figure S2: Clupeid mean annual yield for low and high clupeid fishing mortality’s. Low Fs = 0.2 year−1and high Fs = 0.5 year−1.
9.2
Appendix 2.2
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
0.000
0.005
0.010
0.015
0.020
0.025
Cod adult
b
io
m
as
s (
kg
/m
3)
Low-level clupeid fishing
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
High-level clupeid fishing
6.5 °C , c = 0.005
7.7 °C, c = 0.005
10 °C, c = 0.005
12 °C, c = 0.005
6.5 °C, c = 0.0
7.7 °C, c = 0.0
10 °C, c = 0.0
12 °C, c = 0.0
Cod fishing mortality F
c
(year
1
)
Figure S3: Cod adult mean biomass for two different situations (and for low and high Fs values): c = 0.005 (solid lines) and c = 0.0 (dashed lines). Low Fs = 0.2 year−1 and high Fs = 0.5 year−1.
9.3
Appendix 2.3 (Data repository)
Python code used for implementation of the fish community-dynamics model extended with temperature dependency and also the Python code used for data analysis are available on this Github repository: https://github.com/esmeevdmark/BSc-thesis. The figures produced and used in my BSc thesis are also available in this repository, and also the CSV files which are needed to produce these figures are available. More information can be found in the README.md file in this repository.
