79717 research outputs found
Sort by
The role of material properties in an ISBM process to manufacture polypropylene bottles from virgin and PCR resins
Get PDFPost-Consumer Recycled (PCR) resin grades have been successfully incorporated into plastic ink supply cartridges, with 80% of cartridge products containing 45% to 70% recycled polyethylene terephthalate (rPET). With this tremendously successful experience, incorporation of recycled polypropylene (rPP) into Injection Stretch Blow Molding (ISBM) bottles has been attempted to increase sustainability across all plastic ink supply containers for better ink seal and protection. Virgin PP comes in three forms – homopolymer (Ho_PP), impact copolymer (Im_PP) and random copolymer (Ra_PP). Preforms and bottles could be made with virgin PP in the Melt Flow Rate (MFR) range from 2 to 20 g/10min. However, although al 100% of rPP material was able to make preforms in the injection molding process with a similar range of MFR, it failed in stretch-blowing molding, and in addition, very few ISBM bottles have been made above 50% rPP. The higher crystallization temperatures upon cooling that are observed with the Ho_PP, Im_PP, and Ra_PP, blends of rPP reduces the blend elasticity and the ability to blow a uniform bottle, as compared to the virgin PP grades.
This study is focused on using Melting Flow Rate (MFR), Differential Scanning Calorimetry (DSC), Thermogravimetric Analysis (TGA), single screw capillary Extrusion, and Dynamic Oscillatory Shear (DOS) rheology to understand the thermal and mechanical properties, and processability in eight grades of homopolymer, random copolymer, impact copolymer and rPP with varying MFRs, including five formulations with varying rPP contents (80%, 50%, 30%, 20%, and 10%).
The results showed that MFR alone might be adequate as a quality control tool to grade plastic materials batch to batch, but it cannot adequately predict ISBM processing performance. While MFR, TGA, DSC heating, Melt Capillary Extrusion, and frequency sweep DOS were useful in delineating the basic thermal and rheological properties of the various materials, they did not predict processability in the ISBM process. The current ISBM process for the polypropylene bottles does not have additional heating after the injection molding step, so the temperature of the preform in the blow molding step cannot be controlled. Therefore, for the current ISBM process, it was found that the crystallization temperature upon cooling (Tc) and the DOS rheological properties in the temperature sweep (hot to cool) experiment provided the best correlation with processability. The preliminary result of this study, which will be further studied in the next phase of testing, is that a new specification for materials to be used in the ISBM process should be the crystallization temperature upon cooling as determined in the DSC slow cooling and DOS temperature sweep experiments
The Geology, Geochronology, and Geochemistry of the Miocene Volcanic Rocks at the Inmaculada Ag-Au Mine, Southern Peru
No full textThe Inmaculada Mine is located in the Miocene belt of epithermal deposits that extends from southern Peru to northern Chile and Bolivia. This belt is known for its silver-rich epithermal veins that have been worked since colonial times. The Inmaculada Mine belongs to a mining district that includes, from north to south, the Selene, Pallancata, and Inmaculada mines. This study presents a new geologic map of a 100 km² area, including the Inmaculada Mine at a scale 1:25,000 and the results of laboratory studies that include petrography, geochronology, whole-rock geochemistry, as well as electron microprobe mineral analysis. Also, it reports adularia ages for samples from the Pallancata Mine.
The oldest dated volcanic rocks in the area, Aniso Rhyolite Volcaniclastics, Domes and Tuffs, are found north of the Inmaculada Mine and consist of volcaniclastic rocks, minor lacustrine limestone, and interbedded rhyolite tuffs which are intruded by rhyolitic dikes and domes. They are interpreted to potentially belong to an intra-caldera environment. The ⁴⁰Ar/³⁹Ar ages of sanidine from tuffs and rhyolite intrusions yielded two population a) ~ 23.7 Ma and b) 22.6 Ma, which are similar to the ages of regional Nazca 1 and Nazca 2 Ignimbrites.
The next youngest unit in the area is the Huallhua Rhyolite Ignimbrite, which is as thick as ~850 m was erupted at ~13.2 Ma. It crops out to the south of the Inmaculada Mine and consists of poorly sorted, massive, lithic-rich, poorly welded, and crystal-poor rhyolite tuff containing previously altered accidental fragments. The rock shows little variation in welding and varies from 62-74 wt% SiO₂ in composition. It is proposed that the contact between the 23-24 Ma Aniso Rhyolite Volcaniclastics Domes and Tuffs and the ~ 13.2 Ma Huallhua Rhyolite Ignimbrite is the San Salvador fault. This fault is the border of a proposed caldera named the Alpabamba caldera located southwest of the Inmaculada Mine. Immediately after the caldera collapse, the ~13.2 Ma Inmaculada Volcanics erupted, which range in composition from andesite to dacite and probably to rhyolite. Two pyroxene barometry suggests that the depth of pyroxene crystallization of the Inmaculada Volcanics ranges from 10 to 22 km. Therefore, it is proposed that after the rhyolite pyroclastic eruption and the caldera collapse, intermediate composition magma rose from a source at 10 to 22 km and erupted as the Inmaculada Volcanics.
The Inmaculada Mine is hosted in veins cutting the Inmaculada Volcanics and has two epithermal alteration styles, quartz-alunite, and quartz-adularia-illite, which may be genetically related. Only the quartz-adularia-illite alteration is associated with economic Ag-Au mineralization. Two Ar-Ar ages of adularia from Inmaculada yielded 12.42 ± 0.03 Ma and 12.42 ± 0.05 Ma; consequently, the quartz-adularia veins formed ~ 0.8 Ma after the eruption of the dated Inmaculada Volcanic host rocks. In the context of Andean uplift, the ~0.8 Ma time-period may have allowed the formation of a hydraulic gradient, which drove a lateral flow of hydrothermal fluids along the km-long northeast-southwest oriented vein system. The sphalerite-galena-pyrite-chalcopyrite-acanthite mineral assemblage indicates that the Inmaculada Mine is an intermediate sulfidation state epithermal deposit.
After the mineralization event, a ~10 Ma hornblende andesite was emplaced and is exposed to the northeast of the Inmaculada Mine. It is likely a subvolcanic intrusion. Then, the ~ 9.2 Ma Huancarama Rhyolite Ignimbrite covered the area of study. Based on its mineral assemblage plagioclase-quartz-sanidine-biotite-hornblende, it is thought to have erupted from an upper crust magma chamber. The igneous activity finalized at ~ 6.9 Ma when both the Chibchi Rhyolite Tuff and the Coñaccahua Andesite Lavas erupted. They crop out to northwest of the Inmaculada Mine.
Previous studies from the Selene Mine (Explorador vein) reported 14.2 ± 0.2 Ma (⁴⁰Ar/³⁹Ar of whole rock) as the possible age of Ag-Au mineralization. Adularia samples from the Pallancata Mine yielded ages of ~13.4 Ma. Therefore, there was a north to south migration of hydrothermal, and presumably causative magmatic activity in the district, from Selene to Pallancata to the Inmaculada Mine (~12.4 Ma) over a period of about 2 Ma
Modeling Human Learning of Data Quality Rules
Get PDFThe process of cleaning data is quite laborious and can be complicated by the presence of inconsistencies and other issues in the data, leading many data scientists to utilize interactive data cleaning software. However, many current state-of-the-art cleaning solutions fail to account for the reality that the human needs to explore the data to learn and as a result may periodically provide noisy feedback. Therefore, we need to understand how humans fundamentally learn data quality rules over the data and how that translates to identifying violations of these rules in the data they work with. In this thesis, we perform an empirical analysis of how humans understand and iteratively learn data quality rules, working to both quantify and model this learning process. We find that a Bayesian learning model replicates user behavior well when the model success definition is strict, while a hypothesis testing model also performs well when the success definition is loosened
Evaluation of the Efficiency of Supplementary Cementitious Materials to Prevent Alkali-Silica Reaction in Concrete with Highly and Very Highly Reactive Aggregates
Get PDFThe most common test methods to evaluate alkali-silica reaction (ASR), the concrete prism test (CPT) and the accelerated mortar bar tests (AMBT), have limitations and are not entirely reliable in terms of predicting the performance of concrete with prevention measures under field conditions. In this study, the miniature concrete prism test (MCPT), which is a relatively new test method, was benchmarked against the field exposed blocks to determine its reliability in predicting the efficiency of ASR preventive measures. The MCPT method was compared to two other recently developed laboratory test methods - the concrete cylinder test (CCT) and the University of New Brunswick concrete cylinder test (UNBCCT). The MCPT was shown to be a rapid and easy to perform method among the three methods. For the mixtures evaluated, the MCPT showed higher reliability than the CPT in terms of predicting the performance of mixtures with supplementary cementitious materials (SCMs) when compared with the long-term results obtained from the field exposed blocks.
The MCPT mixtures were further evaluated by analyzing pore solution, measuring bulk electrical resistivity, and doing petrographic analysis. Petrographic analysis showed that the extent of cracking observed in the MCPT prisms was on the same order as the observed expansions. The pore solution alkalinity and electrical resistivity of the MCPT specimens showed a good correlation (R² > 0.90) to expansions; suggesting these could be used as parameters to monitor the progress of ASR in accelerated laboratory evaluation. The correlations between MCPT expansions and hardened paste properties and binder oxide composition were also examined. The bulk chemical index - Na₂Oₑ.CaO/SiO₂² correlated the best (R² > 0.99) to the MCPT expansions. The possibility of recommending threshold chemical indices for different aggregate reactivities to predict an SCM combination’s ASR prevention efficiency was shown. The paste pore solution alkalinity and paste bulk electrical resistivity correlated well (R² > 0.90) with the MCPT expansions; the SCM combinations that resulted in lower alkalis in the pore solution and reduced transport properties showed higher efficiency to prevent ASR expansion. The correlation analysis showed a possibility of predicting MCPT expansion from the hardened paste properties and could potentially be used as an approach to determine preventive measures for ASR.
The efficiency of SCMs in combination with portland-limestone cements (PLCs) to prevent ASR has not been well studied previously in the literature. Therefore, in this study, several accelerated test methods – the Pyrex mortar bar test, the AMBT, and the MCPT were used to compare the performance of OPC and PLC systems with SCMs. The PLCs studied include both interground (10-15% limestone by mass) and interblended (10% limestone by mass) systems. Two very highly reactive fine aggregates were considered. The expansion results from the test methods revealed that the mixtures with PLCs and SCMs had similar or better overall performance compared to the mixtures with their corresponding OPCs and SCMs. Pore solution alkalinity and bulk electrical resistivity determined on concrete cylinders exposed to the MCPT conditions revealed that the lower pore solution alkalinity and higher resistance to mass transport were the main contributing factors towards PLCs’ improved performance for ASR prevention in the presence of SCMs. Further, the findings showed that the concrete pore solution alkalinity and bulk electrical resistivity correlated well (R² > 0.71 and 0.85, respectively) to the MCPT expansion. Therefore, a new approach that has a potential to reduce the amount of long-term performance testing needed to determine the efficiency of SCM combinations to prevent ASR was proposed by combined use of the pore solution alkalinity and bulk electrical resistivity of concrete
Passive acoustic monitoring of owls: two studies in forested landscapes of the Pacific Northwest
Get PDFNearly all birds communicate through sound, and there has been much study of avian populations and communities using song and other vocalizations. Owls are no exception as they defend territories, advertise for mates, and defend against threats using various vocalizations. However, due to their generally nocturnal habits, some owl species have not been well-studied. With the advent of passive acoustic monitoring using autonomous recording units (ARUs), we can now collect large amounts of nocturnal acoustic data without disturbing target species and with relatively low field effort. In this thesis I used ARUs to evaluate their ability to detect calls of northern spotted owls and barred owls in three study areas in the Pacific Northwest (Chapter 2), and to quantify landscape use by five species of owls of an area in Southwest Oregon burned two years prior by a mixed-severity wildfire (Chapter 3).
Northern spotted owl (Strix occidentalis caurina) populations have been monitored since the mid-1980s using mark-recapture survey methods. To evaluate an alternative survey method, I used ARUs to detect calls of northern spotted owls and barred owls (S. varia) (Chapter 2), a congener that has expanded its historic range into the Pacific Northwest and now threatens northern spotted owl persistence primarily through interference competition. From March-July 2017, I set ARUs to record continuously each night at 30, 500ha hexagons (150 ARU stations) with recent northern spotted owl activity, and high barred owl use in Oregon and Washington. I reviewed spectrograms (visual representations of sound) and manually tagged target vocalizations to extract calls from ~160,000 hours of recordings. Even in a study area with low occupancy rates on historic territories (Washington’s Olympic Peninsula), the probability of detecting a northern spotted owl when it was present in a hexagon exceeded 0.95 after 3 weeks of ARU deployment. Background noise, mainly from streams, rain, and wind negatively affected detection probability for both species over all study areas. Using known demographic information about northern spotted owl pair locations I quantified patterns in northern spotted owl vocalization intensity that could help distinguish paired vs. single birds using only passive acoustic data. I found that weekly detection probabilities of northern spotted owls were higher for ARUs placed closer to northern spotted owl nests and activity centers. Examination of vocal activity patterns of both northern spotted and barred owls revealed strong synchrony between the two species through the lunar cycle with calling peaks near the full moon, but asynchrony over the diel period. Northern spotted owls called more frequently during the crepuscular period, while barred owl calling was highest in the nocturnal period, suggesting there may be fine-scale temporal partitioning of calling activity. These results demonstrate that ARUs can be used to effectively detect northern spotted owls when they are present, even in a landscape with high barred owl density, thereby facilitating the use of passive, occupancy-based study designs to monitor northern spotted owl populations.
Mixed-severity fire is the dominant fire type in southwest Oregon, yet its effects on wildlife communities is complex and poorly understood. In particular, little attention has been paid to the use of burned landscapes by owls. I performed passive acoustic surveys using ARUs in and around the area of a ~10,500 hectare mixed-severity wildfire two years post-fire (Chapter 3). I detected the vocalizations of six owl species including northern spotted owls, barred owls, great horned owls (Bubo virginianus), western screech-owls (Megascops kennicottii), northern pygmy-owls (Glaucidium gnoma), and northern saw-whet owls (Aegolius acadicus). I did not detect enough northern spotted owl vocalizations, so they were excluded from analyses. I evaluated landscape use (barred owls and great horned owls; large home ranges prevent estimation of true occupancy) or occupancy (three small owl species), with a single-species, single-season occupancy model using detection data gathered on the remaining five owl species. Western screech-owls had higher probability of occupancy at more severely burned sites, northern pygmy-owls were widely distributed but increased their occupancy as burn severity increases, and northern saw-whet owls appeared to avoid nearly all burned areas, no matter the severity. Barred owls were somewhat less likely to use areas deeper within the fire’s perimeter than unburned areas outside the fire, but ARUs did detect them in burned areas. Great horned owls showed patterns of landscape use unrelated to fire severity. Four out of five species of owls occupied or used recently burned areas without salvage logging. Thus, the high level of landscape heterogeneity and unique forest conditions that result from mixed-severity fire appear to be suitable for use by many owl species. These two studies evaluated the use of passive acoustic detection of forest owl species both as a potential long-term monitoring method, and to gain understanding of spatial patterns within a single survey season. I established that northern spotted owls and barred owls are highly detectable with ARUs using a landscape-level sampling scheme, and identified environmental factors that affect the ability to detect both species. Even with relatively few individuals in an area, ARUs can detect owls due to large spatial coverage and long duration of recording. This is particularly advantageous and may have management implications when it comes to monitoring threatened and endangered species. Additionally, I identified previously undocumented patterns of use in a post-fire landscape by five species of forest owls, adding to the growing body of literature regarding wildfire’s effects on wildlife communities
Assessing Renesting Rate and Survey Methods of Red Phalarope (Phalaropus fulicarius), a Polyandrous Arctic-breeding Shorebird
Get PDFThe life history traits of polygamous, non-territorial shorebirds like the Red Phalarope (Phalaropus fulicarius) can make it challenging to assess their population status. There are two general approaches to statistical inference used to understand shorebird (Order: Charadriiformes) population change: design-based inference, which implements an a-priori sampling scheme with known selection probabilities, and model-based inference, which seeks to identify a stochastic model that explains observed variation in a fixed sample. Model-based approaches rely on good parameter estimates, and renesting rate is one such population parameter that is poorly understood in Red Phalaropes. I conducted a clutch-removal experiment to estimate renesting rate in Red Phalaropes breeding near Utqiaģvik, Alaska, and found that 19 of 24 males whose nests we experimentally removed (n=19) or had been abandoned/depredated (n=5) left the study area prior to the end of the laying period (Chapter 2). Three of the 19 males were observed to have paired again after clutch removal, suggesting they may have renested outside the range of my telemetry system; however, I was unable to assess renesting rate for my study population. Design-based approaches tend to minimize sampling error, but unaddressed measurement errors can affect the accuracy of their estimates. I investigated four untested assumptions of the Arctic PRISM monitoring program, which uses a double-sampling scheme to estimate shorebird population size and trends, for Red Phalaropes breeding near Utqiaģvik, Alaska. I found a weak relationship between rapid and intensive survey variables in my dataset (marginal R2 values <0.12, Chapter 3), indicating low precision of population estimates and a limited ability to improve the precision by re-summarizing existing survey data. Imperfect nest detection on intensive surveys biased population estimates low, while occurrences of polyandry and renesting biased population estimates slightly high, which taken together resulted in a 33% population underestimate for my study area in 2019 (Chapter 3). A combined approach to making Red Phalarope population estimates, which takes advantage of the strengths of a design-based sampling scheme while using model-based approaches to minimize measurement and detection errors, may optimize the accuracy of population estimates for this species
Balancing Timber Production and Conservation in the Oregon Coast Range: an Examination of the Response by Birds and Bird-induced Ecosystem Services to Forest Management at Multiple Scales
Get PDFSociety derives many critical and irreplaceable values from forests. With a growing global human population and rates of consumption, forests are under increasing pressure to provide all these values simultaneously. To meet societal demands for wood products, tree plantations are becoming increasingly common and are replacing native forests. Yet, forests house the majority of terrestrial biodiversity. While there is strong consensus that tree plantations, or intensively managed forests, have negative consequences for biodiversity, plantations may still provide some benefits. The degree to which plantations can benefit biodiversity, and at what scales, is actively debated and forms the foundation of my research.
To inform this question and better inform sustainable management, I examined the effects of intensive forest management on an ecosystem service (Ch. 2) and the provision of early seral forest habitat at forest stand scales (Ch. 3), and finally, examined how the allocation of management treatments across a managed forest landscape will affect the provision of multiple forest values (Ch. 4). My study area is the Oregon Coast Range Physiographic Province, a temperate forest region in the Pacific Northwest of the USA. In this dissertation, I provide a
general overview of the problem in Chapter 1 and then describe individual research studies in Chapters 2, 3, and 4. By examining the effects of intensive forest management at multiple scales, my goal is for my research to inform both stand management practices and strategic forest planning.
In Chapter 2, I hypothesized that, in plantations, birds help trees grow by consuming herbivorous arthropods, and that the degree to which birds help trees grow is influenced by management intensity. This study of a trophic cascade utilized data from a broad-scale, 4-year experimental study in intensively managed tree plantations. This experiment manipulated management intensity by adjusting the intensity of herbicide application (following clearcut harvest) and used exclosures to manipulate bird access. I found that the presence of birds reduces the number of arthropods known to be pests upon the Douglas-fir (Pseudotsuga menziesii) crop trees, and reduces damage to trees, but I detected no change in tree growth due to the presence of birds. Furthermore, the effect of birds did not change with management intensity. My findings suggest that management intensity does not alter the ecosystem services provided by trophic cascades, but that under changing environmental or management conditions (e.g., climate change and increased pest densities), birds could provide important services.
In Chapter 3, using birds as an indicator I examined the degree to which intensive forest management affects biodiversity by 1) quantifying songbird abundance across a gradient in forest stand age, 2) estimating the duration of early seral habitat, and 3) testing if forest structural and compositional elements prolong habitat availability. Using a chronosequence approach, I surveyed species’ densities in intensively managed forests aged 0 to 30 years, which encompassed the critical stage of canopy closure, where biodiversity values are assumed to dramatically decline. I found that the densities of individual species are highly dynamic during the short duration of early seral habitat availability, with most species declining in abundance by the end of the chronosequence. Early seral habitat is available for less than half the time when compared to early seral habitat in less intensively managed forests. This result suggests that for birds to benefit from the ephemeral and patchy nature of early seral in managed forests, they would need to have efficient dispersal and habitat prospecting capabilities.
In Chapter 4, I expanded the spatial and temporal scope of my research by examining the potential of Triad-based scenario allocations to balance multiple forest management objectives. I
developed correlative relationships between multiple forest value indicators (wood supply, the densities of three species of songbirds, habitat suitability for marbled murrelet, carbon storage, and old growth forest structure) and used Woodstock, a widely-used strategic forest planning tool, to calculate the yields of these indicators over a 125-year planning horizon, in an 82,520 acre forest, and under 43 Triad scenarios. Each Triad scenario allocates proportions of the forest to reserve management (where conservation is the primary objective), intensive management (wood-production objective), or extensive management (combined objectives). I found that identifying a single landscape-planning scenario to maximize all indicators is not possible. The ideal scenario to maximize each forest value varied greatly by both the indicator and the level of wood supply. Even though identifying one scenario to balance all values was elusive, multiple scenario options exist for each wood supply level. These options provide ample decision space for forest managers, policy makers, and the pubic to deliberate tradeoffs among indicators.
Collectively, my research shows that tree plantations are not biological deserts – they have intact ecosystem processes and continue to provide early seral habitat. However, tree plantations are novel types of ecosystems that are likely sensitive to additional stressors, such as climate change, and where species need to adapt to take advantage of the patchy and ephemeral nature of habitat conditions in managed forests. The long-term persistence of these species may therefore depend on their adaptive capacity (which is currently poorly understood) and the benefits derived from the surrounding forest matrix. Fortunately, my research has also shown that many options exist for strategically managing forests for multiple objectives by allocating treatments across the landscape. Ultimately, society will need to explore these options and prioritize and establish minimum thresholds for each forest value, in order to inform forest management decisions
Decentralized Multi-Robot Exploration with Sparse Communication through Deep Imitation Learning
Get PDFWe present a method for decentralized, multi-robot exploration in adverse environments where communication is minimal. A key conceptual feature of our method is enabling implicit coordination between robots by training a Convolutional Neural Network (CNN) as a heuristic for planning using Monte Carlo Tree Search (MCTS). Our method consists of three interacting components: (i) a centralized oracle planner that uses ground truth information to generate expert demonstrations, (ii) a CNN that learns to assign rewards to action sequences by imitating a team of expert robots, and (iii) MCTS for online, decentralized planning that uses the trained CNN as the reward function for the action sequences produced by the simulation phase of MCTS.
The CNN also leverages structural information to adapt the planner to different environments. Our experiments in simulation, which include a real-world environment, show that performance gracefully degrades with increasing communication loss for a team of four robots. The proposed method enables a random rollout policy to be used in the simulation phase of MCTS, which saves a significant amount of computation time compared to other heuristics. Our tests further show that the proposed method is scalable to a team of sixteen robots
Decadal changes in lichen and vegetation communities in relation to metal deposition and climate in northern Alaska
Get PDFThe Red Dog Mine transports concentrated zinc and lead ore in northwestern Alaska through Cape Krusenstern National monument along a haul road. High metal levels in the moss Hylocomium splendens along the haul road have been attributed to ore dust released during transport. The mine has implemented several pollution abatement measures to reduce fugitive ore dust since the original study in 2001, and a subsequent remeasure in 2006.
In 2017, we remeasured moss metal concentrations and lichen and vegetation communities with the purpose of documenting potential effects of metal deposition, and climate change to the local ecosystem from 2006 to 2017. Our results indicate significant changes in lichen and vegetation communities as seen through a loss in diversity and a gain in mean total cover. In both years lichen diversity and cover had negative relationships to metal deposition and distance from the haul road. We observed a decrease in lichen diversity and a doubling in cover over time. Vegetation diversity was nearly constant but similarly doubled in cover over time. Changes in metal content of mosses during the study period were not significant as evident from the subset of data from 60 plots available. The lichen community was correlated to metal content, which suggests a degree of sensitivity and their appropriate use as biomonitors in this system. We found no significant differences in communities between the north and south sides of the haul road. Changes observed through the study period suggest a relationship to climate change. This was observed through increases in total lichen and vegetation cover as likely mediated by regional climate, biome, temperature and drying in wetlands. Overall, pronounced changes were documented within a mere decade, indicating that continued monitoring into the future is necessary and will be an invaluable resource for future generations
Larval Biogeography and Varying Spawning Behavior of Bering Sea Groundfishes: Implications for Management in a Changing Seascape
Get PDFEarly life stages of marine fishes exert substantial control on the productivity and resilience of marine fisheries, as these life stages represent periods of extreme mortality. Additionally, eggs and larvae of marine fishes respond rapidly to environmental change and can thus be used to infer ecosystem-wide responses to climate change or environmental regime shifts. In the Bering Sea (BS), considerable populations of fishes and shellfishes support huge fisheries of economic importance for Alaska and the United States. Groundfishes represent a considerable sector of this economy; in 2019, 83% of all harvest in Alaskan waters was for groundfishes and the ex-vessel value of groundfish fisheries comprised 50% of total Alaskan ex-vessel value. The BS is also a physically unique large marine ecosystem, with three hydrographic domains that represent annually consistent vertical patterns in temperature and salinity. Tidal processes, atmospheric forcing, coastal currents and the annual cycle of sea ice formation and breakup impart nutrients into the region, thus supporting the considerable populations of marine micro- and macro-organisms.
Currently, fisheries in the BS are managed sustainably; Alaska leads the country in ecosystem-based fisheries management and has done so for decades. However, climate change is exerting sweeping impacts on the BS, transitioning typical interannual variability in sea surface temperature and other physical dynamics to multi-year periods of warm or cold regimes. Such warm regimes are expected to modify the distribution and phenology of reproduction in marine fishes. These two dimensions, though, have not been analyzed in conjunction with one another which has led to a lack of clarity on whether certain species are more likely to modify their reproductive geography or phenology. Furthermore, early life stage preferences of water masses, defined by temperature and salinity, are not mapped for many larval groundfishes at the species level. As early life stages are quick responders to environmental change and dynamics of their survivability can significantly impact marine fisheries, the focus of this thesis is on these life stages and the spawning behavior of BS groundfishes.
The objectives of this thesis were to first map the larval biogeography of six species of groundfishes, Alaska plaice, flathead sole, northern rock sole, Pacific cod, walleye pollock, and yellowfin sole in relation to in situ measurements of sea surface temperature (SST) and sea surface salinity (SSS) in the BS. Secondly, this thesis examined the spawning behavior of four species of groundfishes, Alaska plaice, flathead sole, walleye pollock, and rex sole, in relation to regional SST indices to quantify whether reproductive flexibility occurred primarily in space or time. These species are all ecologically and economically important in the region and represent a diversity of life history strategies and habitat preferences. I used generalized additive models (GAMs) and decades-worth of ichthyoplankton data collected by the Alaska Fisheries Science Center Ecosystem and Fisheries Oceanography Coordinated Investigations team to address these objectives. Spawning flexibility was inferred with egg distribution data.
I found that all larval species exhibited specific SST and SSS associations; Alaska plaice, northern rock sole, and walleye pollock demonstrated catch associations across a smaller breadth in SST than SSS. Alternatively, flathead sole and Pacific cod exhibited catch associations across a wider breadth in SST than SSS. Yellowfin sole demonstrated relatively even breadths in catch associations with SST and SSS. There were spatial variations among all larval species; however, flathead sole and Pacific cod showed similar patterns of SST and SSS catch associations, and both occurred primarily over the outer BS shelf and along the Aleutian Islands. The subset of species analyzed for spawning behavior flexibility were similar in the dimension of flexibility (e.g., either space or time). Alaska plaice, flathead sole, walleye pollock, and rex sole exhibited geographic flexibility, and egg catches (standardized in the same way as larval data) were best modelled by a GAM that included a threshold effect of regional SST whereby geographic egg distributions varied above and below this threshold.
All three hydrographic domains in the BS appear to exhibit markedly different SST variations in warm versus cold regimes. SSS patterns were similar across all three domains in warm versus cold regimes, though it is likely that the inner shelf will experience decreases in SSS in the coming century. Spatial shifts in reproductive aggregations could impact the ability of larvae to reach ideal nursery grounds, and conditions in coastal areas (where many nursery habitats for these species occur) are likely to undergo increases in seasonality of SST and SSS parameters. This thesis therefore elucidates important information regarding the larval biogeography of groundfishes in the BS and how spawning behavior may affect larval biogeography in the future. Current federal fisheries policy has outlined procedures to include findings like these in management strategies. However, the productivity of fisheries into the future depends on our continued ability to link management practices to multi-life stage species data and climate change science