Near-term forecasts of stream temperature using deep learning and data assimilation in support of management decisions

Deep learning (DL) models are increasingly used to make accurate hindcasts of management-relevant variables, but they are less commonly used in forecasting applications. Data assimilation (DA) can be used for forecasts to leverage real-time observations, where the difference between model predictions and observations today is used to adjust the model to make better predictions tomorrow. In this use case, we developed a process-guided DL and DA approach to make 7-day probabilistic forecasts of daily maximum water temperature in the Delaware River Basin in support of water management decisions. Our modeling system produced forecasts of daily maximum water temperature with an average root mean squared error (RMSE) from 1.1 to 1.4°C for 1-day-ahead and 1.4 to 1.9°C for 7-day-ahead forecasts across all sites. The DA algorithm marginally improved forecast performance when compared with forecasts produced using the process-guided DL model alone (0%–14% lower RMSE with the DA algorithm). Across all sites and lead times, 65%–82% of observations were within 90% forecast confidence intervals, which allowed managers to anticipate probability of exceedances of ecologically relevant thresholds and aid in decisions about releasing reservoir water downstream. The flexibility of DL models shows promise for forecasting other important environmental variables and aid in decision-making.     

Patterns of co-occurrence and interactions between age classes of the common triplefin,Forsterygion lapillum

Many marine organisms have pelagic larval stages that settle into benthic habitats occupied by older individuals; however, a mechanistic understanding of intercohort interactions remains elusive for most species. Patterns of spatial covariation in the densities of juvenile and adult age classes of a small temperate reef fish, the common triplefin (Forsterygion lapillum), were evaluated during the recruitment season (Feb–Mar, 2011) in Wellington, New Zealand (41°17′S, 174°46′E). The relationship between juvenile and adult density among sites was best approximated by a dome-shaped curve, with a negative correlation between densities of juveniles and adults at higher adult densities. The curve shape was temporally variable, but was unaffected by settlement habitat type (algal species). A laboratory experiment using a “multiple-predator effects” design tested the hypothesis that increased settler mortality in the presence of adults (via enhanced predation risk or cannibalism) contributed to the observed negative relationship between juveniles and adults. Settler mortality did not differ between controls and treatments that contained either one (p = 0.08) or two (p = 0.09) adults. However, post hoc analyses revealed a significant positive correlation between the mean length of juveniles used in experimental trials and survival of juveniles in these treatments, suggesting that smaller juveniles may be vulnerable to cannibalism. There was no evidence for risk enhancement or predator interference when adults were present alongside a heterospecific predator (F. varium). These results highlight the complex nature of intercohort relationships in shaping recruitment patterns and add to the growing body of literature recognizing the importance of age class interactions.     

Biological and environmental influences on the trophic ecology of leatherback turtles in the northwest Atlantic Ocean

Understanding the causes and consequences of variability in trophic status is important for interpreting population dynamics and for identifying important habitats for protected species like marine turtles. In the northwest Atlantic Ocean, many leatherback turtles (Dermochelys coriacea) from distinct breeding stocks throughout the Wider Caribbean region migrate to Canadian waters seasonally to feed, but their trophic status during the migratory and breeding cycle and its implications have not yet been described. In this study, we used stable carbon and nitrogen isotope analyses of bulk skin to characterize the trophic status of leatherbacks in Atlantic Canadian waters by identifying trophic patterns among turtles and the factors influencing those patterns. δ¹⁵N values of adult males and females were significantly higher than those of turtles of unknown gender (i.e., presumed to be subadults), and δ¹⁵N increased significantly with body size. We found no significant differences among average stable isotope values of turtles according to breeding stock origin. Significant inter-annual variation in δ¹⁵N among cohorts probably reflects broad-scale oceanographic variability that drives fluctuations in stable isotope values of nutrient sources transferred through several trophic positions to leatherbacks, variation in baseline isotope values among different overwintering habitats used by leatherbacks, or a combination of both. Our results demonstrate that understanding effects of demographic and physiological factors, as well as oceanographic conditions, on trophic status is key to explaining observed patterns in population dynamics and for identifying important habitats for widely distributed, long-lived species like leatherbacks.     

Reproductive success of parasitized males in a marine reef fish

Parasitism is hypothesized to reduce reproductive success in heavily parasitized males because females may preferentially mate with less parasitized males (parasite-mediated sexual selection) or parasites may compromise male competitiveness. In marine systems, this hypothesis is largely unexplored. This paper provides the first confirmed record of a copepod ectoparasite (Caligus buechlerae Hewitt 1964) on the common triplefin (Forsterygion lapillum) and evaluates the hypothesis that males parasitized with C. buechlerae experience lower reproductive success than unparasitized males (as determined by the presence and area of eggs within male nests). We found that 38 % of males we surveyed were infected with at least one C. buechlerae, with a median of two individuals per infected male. About 32 % of males were defending eggs, with 62.5 % of those males infected with at least one parasite. Males of greater total length (TL) were both more likely to be infected and more likely to be defending eggs. However, when statistically accounting for the effects of TL, parasite infection had no effect on the probability of defending eggs, or the average surface area of eggs when present. Positive covariation in fish length, the presence of eggs and parasite infection observed here potentially suggest that the importance of parasitic infection on reproductive success may depend upon the strength of selection for larger male body size. Our study is one of the few studies to investigate the effects of ectoparasites on reproductive success in reef fish and also provides a quantitative measure of infection for a widespread species within New Zealand.     

Cars on crutches: How much abatement do smog check repairs actually provide?

Not as much abatement as has been presumed. Smog check programs aim to curb tailpipe emissions from in-use vehicles by requiring repairs whenever emissions, measured at regular time intervals, exceed a certain threshold. Using data from California, we estimate that on average 41% of the initial emissions abatement from repairs is lost by the time of the subsequent inspection, normally two years later. Our estimates imply that the cost per pound of pollution avoided is an order of magnitude greater for smog check repairs than alternative policies such as new-vehicle standards or emissions trading among industrial point sources.     

Invasive ecosystem engineer selects for different phenotypes of an associated native species

Invasive habitat-forming ecosystem engineers modify the abiotic environment and thus represent a major perturbation to many ecosystems. Because native species often persist in these invaded habitats but have no shared history with the ecosystem engineer, the engineer may impose novel selective pressure on native species. In this study, we used a phenotypic selection framework to determine whether an invasive habitat-forming ecosystem engineer (the seaweed Caulerpa taxifolia) selects for different phenotypes of a common co-occurring native species (the bivalve Anadara trapezia). Compared to unvegetated habitat, Caulerpa habitat has lower water flow, lower dissolved oxygen, and sediments are more silty and anoxic. We determined the performance consequences of variation in key functional traits that may be affected by these abiotic changes (shell morphology, gill mass, and palp mass) for Anadara transplanted into Caulerpa and unvegetated habitat. Both linear and nonlinear performance gradients in Anadara differed between habitats, and these gradients were stronger in Caulerpa compared to unvegetated sediment. Moreover, in Caulerpa alternate phenotypes performed well, and these phenotypes were different from the dominant phenotype in unvegetated sediment. By demonstrating that phenotype-performance gradients differ between habitats, we have highlighted a role for Caulerpa as an agent of selection on native species.     

Shoaling internal waves may reduce gravity current transport

Gravity currents descending along slopes have typically been studied in quiescent environments, despite the fact that in many geophysical settings there is significant externally driven motion. Here we investigate how the head of a gravity current is influenced by interfacial internal waves at the pycnocline of a two-layer ambient water column. Our experimental measurements show that larger amplitude internal waves, interacting with the gravity current, reduce both the mass transport by the gravity current and its thickness. These results suggest that the ambient internal wave field should be considered when estimating transport by gravity currents in geophysical settings with strong internal waves, such as lakes and the coastal ocean.     

An ecosystem model for estimating potential shellfish culture production in sheltered coastal waters

A generic ecosystem model has been developed for estimating the potential production of shellfish culture and the effect of that cultivation on the pelagic ecosystem in sheltered coastal waters. The model describes the dynamics of a simple food web, nutrient cycling and growth of shellfish. The design of the model is closely tied to the temporal and spatial scales that are important in determining the sustainable production level for a particular embayment. The pelagic ecosystem, mussel energetics, population dynamics and hydrodynamics are coupled to allow fully dynamic predictions of the effect of the shellfish density. When applied to Beatrix Bay, an intensive culture embayment in the Pelorus Sound of New Zealand, the model successfully captured main features of the observed system behaviour. The hydrodynamic regime of the bay controls mussel growth and production. Although high fluxes of water into the bay suppress nutrient and carbon cycling signals in the system, the model simulations demonstrated that the mussel cultivation can have considerable effects on the ecosystem of the bay including food depletion and nutrient cycling. One of the most obvious effects is nutrient enhancement through mussel excretion at low cultivation densities, which promotes primary production particularly during the N-limitation period in summer. The sensitivity analysis identified uncertainty in some parameters and indicated areas for which experimental studies could lead to model improvement. The modelling exercise has established a primary predictive tool for managing mussel aquaculture of a coastal embayment to estimate relationships between the stock level and the growth rate of mussels, and the potentially achievable harvest and stocking density.     

Toward Consensus-Based Actions that Balance Invasive Plant Management and Conservation of At-Risk Fauna

Limiting the spread of invasive plants has become a high priority among natural resource managers. Yet in some regions, invasive plants are providing important habitat components to native animals that are at risk of local or regional extirpation. In these situations, removing invasive plants may decrease short-term survival of the at-risk taxa. At the same time, there may be a reluctance to expand invaded habitats to benefit at-risk species because such actions may increase the distribution of invasive plants. Such a dilemma can result in “management paralysis,” where no action is taken either to reduce invasive plants or to expand habitats for at-risk species. A pragmatic solution to this dilemma may be to develop an approach that considers site-specific circumstances. We constructed a “discussion tree” as a means of initiating conversations among various stakeholders involved with managing habitats in the northeastern USA to benefit several at-risk taxa, including New England cottontails (Sylvilagus transitionalis). Major components of this approach include recognition that expanding some invaded habitats may be essential to prevent extirpation of at-risk species, and the effective control of invasive plants is dependent on knowledge of the status of invasives on managed lands and within the surrounding landscape. By acknowledging that management of invasive plants is a complex issue without a single solution, we may be successful in limiting their spread while still addressing critical habitat needs.