Modeling the spatial distribution of larval fish abundance provides essential information for management

Abstract

Productive fisheries are strongly linked to the ecological state of the essential habitats. In this study, we developed a methodology to assess the most important reproduction habitats of fish by using larval survey data and Bayesian species distribution models that predict the spatial distribution and abundance of fish larvae. Our case study with four commercially and ecologically important fish species in the coastal zone of the northern Baltic Sea demonstrated that the production of fish stocks can be concentrated to an extremely limited area compared with the entire suitable production area. The area suitable for larval production varied from 3.7% to 99.8% among species, but the smallest area responsible for 80% of the cumulative larval production was two to five times more limited, varying from 1.4% to 52.9% among species. Hence, instead of the traditional approach of modeling only habitat suitability for fish production, marine spatial planning and management should take into account the areal production potential. Moreover, the developed methodology enables linking of the total production potential across the whole distribution area to fisheries stock assessment and management.

Additive Multivariate Gaussian Processes for Joint Species Distribution Modeling with Heterogeneous Data

“…Such limits, for example, along temperature and salinity are typical for large variety of taxa in aquatic domains (MacKenzie et al, 2007;Kotta et al, 2019). With SSDMs Vanhatalo et al (2012), Shelton et al (2014), Golding and Purse (2016) and Kallasvuo et al (2017) have demonstrated the benefits of semiparametric models in such situations. The GP approach for modeling environmental responses is similar to generalized additive models (Guisan et al, 2002) but the latter have not been implemented as JSDMs.…”

“…However, predictions using these covariance functions revert to prior predictive distributions when predicting beyond covariate range covered by data. Hence, in extrapolation tasks stationary covariance functions may not be the optimal choice (Vanhatalo et al, 2012;Kallasvuo et al, 2017) and combining the multivariate GP models with functional constraints (e.g. Kotta et al, 2019) could improve their predictive performance further.…”

“…The Gulf of Bothnia is a brackish water basin between Sweden and Finland covering an area of approximately 600 × 120 km. Its coastal areas play a central role in the ecosystem and many Baltic fish stocks are dependent on the coastal regions for their reproduction (Veneranta et al, 2013;Kallasvuo et al, 2017). Coastal zones are also the most sensitive regions of the Baltic sea to both natural variation and anthropogenic pressures (Reusch et al, 2018).…”

Long‐term demographic surveys reveal a consistent relationship between average occupancy and abundance within local populations of a butterfly metapopulation

“…Species distributions and occupancy models usually use only occupancy data and model patterns instead of demographic processes (Pearce and Ferrier 2001, Sileshi 2007, Keith et al 2008, Duff et al 2012. Although species distribution models built on abundance data give a more detailed view of the relationship between population density and the environment (Kallasvuo et al 2017), they still implicitly assume a simple relationship between patterns of occupancy and abundance (Sileshi et al 2009, Dallas andHastings 2018). Models of abundance conditioned on covariates that use for example a Poisson or negative binomial distribution yield occupancy probabilities as a simple function of the distributional form (Holt et al 2002).…”

Mosaic environments shape the distribution of Neotropical freshwater ichthyoplankton

“…Rivers are complex aquatic ecosystems that exhibit zonal variation in physical, chemical and biological characteristics, resulting in considerable habitat heterogeneity (Nannini, Goodrich, Dettmers, Soluk, & Wahl, ; Ward & Tockner, ). Fish exploit these diverse fluvial biotopes during different life cycle stages, particularly spawning and early development (Gogola, Daga, Gubiani, Da Silva, & Sanches, ; Kallasvuo, Vanhatalo, & Veneranta, ), without which recruitment is seriously hampered (Humphries & Lake, ). Because population maintenance depends on both reproductive success and recruitment, habitat loss will alter assemblage structure over time, reducing fish stocks and causing local extinction (Agostinho, Gomes, Veríssimo, & Okada, ; Humphries, King, & Koehn, ).…”