Linking statistical and mechanistic modelling to understand the dynamics of the free-floating eggs and larvae of a marine fish

Many marine fishes experience tremendous mortality during their first months of life. Understanding the causes of this mortality and why it varies from year to year has challenged fisheries ecologists for more than a century. Part of the difficulty comes from the fact that many fishes have free-floating larvae. It is therefore difficult to follow a group of fish larvae over time in the field and investigate which factors cause mortality.

We here combine two approaches to study what causes spatial and temporal differences in mortality of fish larvae. We first focus on the spatial differences, asking which factors influence survival locally. Using mechanistic modelling, we tried to realistically simulate drift, growth and survival of eggs and larvae of Barents Sea cod through their first months of life. Using statistical modelling, we analysed observation data from >30 years of field surveys conducted by PINRO, Murmansk. From these surveys we have the distribution of cod eggs and larvae in April-May as well as the distribution of the larvae in June-July each year. The mechanistic model (a coupled physical-biological model) allows us to link the observations in April-May to those in June-July. We can then look at which variables explain the changes in abundance from spring to summer at the local scale.

We find that water masses with high occurrence of cod larvae in summer are characterized by high concentrations of copepod nauplii in spring.  Copepod nauplii are the main prey of the young cod larvae, and it has long been thought that food availability is critical for their survival during this period. Our findings support this idea, by suggesting that cod eggs and larvae located in patches with high food concentrations in spring have higher chance of surviving to summer.

In addition, our findings suggest that being located in relatively warm water masses in spring also leads to higher survival to summer. Note that this is one of the northernmost cod populations, and “warm” here means around 4−6 °C. High temperature leads to faster development. This may indirectly lead to higher survival, as the eggs and larvae by growing faster sooner acquire a size when they can escape predators. The egg stage is the most vulnerable, and in an earlier study we have found that about 16% of the cod eggs die each day. As larvae, “only” about 8% die per day.

In addition to this analysis of which factors explain spatial differences in survival, we analysed spatially averaged data to explore if the same factors explain year-to-year differences in survival at the cohort scale. The results were at first disappointing. Annual mean abundance of cod larvae in summer was strongly correlated with the abundance of eggs and larvae in spring, but not with copepod nauplii concentration or temperature in spring. Does this mean that the correlations found in the spatial analysis were spurious, that is, arisen by chance or through correlations with other, unknown factors?

Perhaps. This possibility can never be excluded in correlational studies such as ours. On the other hand, it turned out that annual means of copepod nauplii concentration and temperature in spring were both correlated with the mean length of the cod larvae in summer. High food abundance and high temperature thus meant that the cod larvae on average were larger in summer. This finding could be a result of food and temperature effects on growth as well as food and temperature effects on survival. Higher survival leads to higher mean body size by increasing the frequency of old and hence large larvae in the samples. The results of the local scale and the cohort scale analyses were therefore not as conflicting as first thought.

Finally, it turned out that survival to age 3 years was higher in years when the mean body size of the larvae in summer was larger. This result highlighted the biological significance of the findings, by showing that the investigated dynamics of these early life stages are also likely to influence the dynamics of the population as a whole.



Stige, L., Langangen, Ø., Yaragina, N., Vikebø, F., Bogstad, B., Ottersen, G., Stenseth, N.C., & Hjermann, D.Ø. (2015). Combined statistical and mechanistic modelling suggests food and temperature effects on survival of early life stages of Northeast Arctic cod (Gadus morhua) Progress in Oceanography DOI: 10.1016/j.pocean.2015.01.009


Langangen, Ø., Stige, L., Yaragina, N., Vikebo, F., Bogstad, B., & Gusdal, Y. (2013). Egg mortality of northeast Arctic cod (Gadus morhua) and haddock (Melanogrammus aeglefinus) ICES Journal of Marine Science, 71 (5), 1129-1136 DOI: 10.1093/icesjms/fst007

Tags: recruitment, fish larvae, statistical models, coupled physical-biological models, environmental effects, cod, Northeast Atlantic, Barents Sea By Leif Christian Stige
Published Mar. 27, 2015 1:01 AM
Add comment

Log in to comment

Not UiO or Feide account?
Create a WebID account to comment