figure 1)).
Modelling recruitment and parent stock may be improved by carefully selecting and including environmental variables in the S‐R models, an approach that can be particularly appealing for many estuarine species, as noted above for effects of winter temperature on the sciaenid Micropogonias undulatus (Hare et al. 2010) (Figure 3.17b). For the clupeid Brevoortia patronus in the Gulf of Mexico, Mississippi River discharge is inversely related to recruitment (Govoni 1997). Inclusion of a river flow variable in either a Ricker of Beverton–Holt S‐R model improves the model fit for recruitment of B. patronus sufficiently to provide predictive power (Vaughan et al. 2011). In another example, for the anadromous moronid Morone saxatilis from Chesapeake Bay, including freshwater discharge in spring months as a variable in a Ricker S‐R model increased the model's predictive power (r 2 increased from 0.03 to 0.44) (North & Houde 2003). Furthermore, a modified Ricker S‐R model with both an added environmental variable (salinity) and a stock variable (age diversity of adult spawners) (Figure 3.18) explained a high proportion of the variability in observed recruitments of age‐0+ juvenile M. saxatilis (Houde 2008).
For the anadromous alosine Alosa sapidissima in the Connecticut River (USA), a Ricker model fitted to spawner abundance explained little of the observed recruitment variability (Crecco & Savoy 1984, Crecco et al. 1986). Including a freshwater‐discharge variable in the S‐R model improved the modelled result (r 2 increased from 0.02 to 0.26). This relationship, while still far from a reliable predictor of recruitment, convincingly indicated that freshwater discharge (in this case a negative factor) may control recruitment and, in many years, may be more important than abundance of adult spawners.
Figure 3.18 Modified Ricker stock–recruitment model for Morone saxatilis, Chesapeake Bay. Juvenile (age‐0+) abundance index, i.e. Recruitment (R), Adult stock biomass (P), Salinity (Sal), and spawner age diversity (H′). H′ is the Shannon–Wiener Diversity Index applied to the abundances‐at‐age of adult spawners in each year
(from Houde (2008, his figure 6)).
Recruitment levels of an unfished species, the short‐lived engraulid Anchoa mitchilli, in Chesapeake Bay were described by a modified Ricker S‐R model with ΔL (a latitudinal variable that defines the centroid of adult spawner biomass, itself determined by dissolved oxygen level) included as a delineator of the adult stock's centre of abundance (Jung & Houde 2004a):
where R y = recruitment level, S = estimated baywide A. mitchilli spawning stock biomass, ΔL = centre of latitudinal location of the A. mitchilli spawning stock. A high percentage of the annual variability in recruitment was explained by the modified Ricker model.
Including water temperature as an environmental variable in a Ricker S‐R model effectively described age‐1 recruitment of a Baltic Sea percid Sander lucioperca in the Archipelago Sea near Finland (Figure 3.19) (Heikinheimo et al. 2014). Although recruitment of S. lucioperca exhibited a moderate degree of density dependence related to adult stock abundance, temperature was the major driving force determining recruitment level.
3.4.2.3 Predicting and forecasting recruitment
Predicting and forecasting recruitment has obvious value for stock assessments and management of estuary‐dependent and ‐associated fishes. Beyond the importance for fishery management, forecasting recruitment has ecological value to predict short‐ (annual) and long‐term trends in stock abundances. The search for environmental predictors of recruitment is at least a century old (Cushing 1982), and despite substantial efforts by scientists to include environment–recruitment correlations in stock assessments and forecasts, success has remained rather poor in practice (Walters & Collie 1988, Myers 1998, Houde 2008, 2016, Sharma et al. 2019). Nevertheless, forecasts that explore and analyse relationships between the environment (broadly), adult stock and recruitment, i.e. ecological forecasting, have value even if failing to meet the rigour required by fisheries managers for stock assessments.
Figure 3.19 Recruitment time series for Sander lucioperca in the Archipelago Sea, Finland. Recruitment expressed as loge number of recruits (R) per unit spawning biomass (S). Modified Ricker model, with July–August water temperature included in the stock–recruitment model, is fitted to the data
(modified from Heikinheimo et al. (2014, their figure 7)).
The complexity of estuaries and the life cycles of fishes that reproduce in and recruit to estuaries add to the challenge of successful forecasting. The complexity itself suggests that forecasting based on adult abundance alone is insufficient and that environmental variables must be incorporated into analyses and models to account for factors driving recruitment variability. For many estuary‐dependent and ‐associated species in which egg, larval and juvenile stages occupy different habitats, identifying variables and life stages most linked to recruitment variability is particularly challenging. Incorporating freshwater flow and temperature variables into stock‐recruitment models for estuarine‐associated fishes has gained considerable success in predicting recruitments in the past three decades. Two notable examples include the clupeid Brevoortia patronus in the Gulf of Mexico, in which freshwater discharge from the Mississippi River has predictive power (Vaughan et al. 2011) and the sciaenid Micropogonias undulatus along the coast of the northeast USA in which winter temperatures are directly related to recruitment success (Figure 3.17a) (Hare et al. 2010). Predictive models presented in Figures 3.16 and 3.17 for two estuary‐dependent species are good examples of how S‐R models that include adult biomass and environmental variables can be effective for hindcasting but are not necessarily used or effective for forecasting (Haltuch et al. 2019).
Climate variability in marine ecosystems often is expressed in patterns that dominate for periods of a few years to decades and thus has the potential for predicting and forecasting trends in fish recruitment. Both oscillating climate regimes and directional climate shifts may affect reproduction and recruitment in marine and estuarine fishes (Nye et al. 2014, Gillanders et al. 2022). Using a synoptic climatology approach, inter‐annual recruitment variability in several offshore‐spawning, estuary‐dependent and anadromous fishes was investigated (Wood 2000, Wood & Austin 2009). These analyses indicated considerable potential and ability to hindcast recruitment and to predict probable trends for these species.
In many fishery management
