Abstract We used a natural experiment to test whether wildfire smoke induced changes in the vertical distribution of zooplankton in Lake Tahoe by decreasing incident ultraviolet radiation (UV). Fires have a variety of effects on aquatic ecosystems, but these impacts are poorly understood and have rarely been observed directly. UV is an important driver of zooplankton vertical migration, and wildfires may alter it over large spatial scales. We measured UV irradiance and the distribution of zooplankton on two successive days. On one day, smoke haze from a nearby wildfire reduced incident UV radiation by up to 9%, but not irradiance in the visible spectrum. Zooplankton responded by positioning themselves, on average, 4.1 m shallower in the lake. While a limited data set such as this requires cautious interpretation, our results suggest that smoke from wildfires can change the UV environment and distribution of zooplankton. This process may be important in drought‐prone regions with increasingly frequent wildfires, and globally due to widespread biomass burning.
Abstract Seabirds have long been thought to exploit social information when searching for their prey, the distribution of which is often patchy and variable. The fact that most seabirds breed colonially has led to speculation that colonies serve as “information centers,” allowing their inhabitants to learn about the distribution of food by observing or following other successful foragers, though this hypothesis is controversial and the evidence for it is mixed. However, several recent studies have documented behaviors that suggest some seabirds do exploit social orientation cues at or near their colonies in order to orient toward food. In this paper, I explore in‐depth one such social orientation behavior, which I call “visual trail following.” I derived a simple model of information transfer and showed that trail following should be favored over other commonly hypothesized foraging behaviors. An individual‐based simulation model was then used to test this theoretical prediction against several other foraging strategies while varying prey patchiness and colony size. The model's results showed that trail following was the optimal strategy across a wide range of conditions. Finally, I used radar data recorded at a tern colony in coastal New York to demonstrate evidence for trail following in the movements of wild seabirds. These results show that trail following and similar behaviors are effective foraging strategies that are likely important for seabirds and other colonial animals.
Abstract Zooplankton are important components of lentic ecosystems, affecting phytoplankton, water clarity, and nutrient cycling, as well as transferring primary production to upper trophic levels. Many of these processes are temporally and spatially heterogeneous, but are difficult to observe at fine scales with traditional sampling methods. High‐resolution sampling has been especially rare in remote and high‐altitude lakes. We measured the vertical distribution of zooplankton and fish in four lakes in the Sierra Nevada Mountains of California, U.S.A. (Independence Lake, Lake Tahoe, Cherry Lake, and Lake Eleanor) using a dual‐frequency echosounder, and estimated lake‐wide biomass in all lakes except Tahoe. For zooplankton, we also quantified trends and patchiness in their horizontal distribution. In two of the lakes, Cherry and Eleanor, surveys were repeated four times at seasonal intervals between autumn 2013 and autumn 2014. Zooplankton were most abundant in these lakes in the spring and summer of 2014, with peak wet‐weight biomasses estimated at 31,000 kg in Lake Eleanor in April and 68,000 kg in Cherry Lake in June. The biomass and vertical distribution of fish also varied, increasing and moving shallower in the water column in June in both Cherry Lake and Lake Eleanor. Zooplankton density was not horizontally homogeneous, displaying gradients at the lake basin scale (5–6 km), and nested patchiness at a range of smaller scales (0–2 km). This small‐scale spatial variability may be generated biologically, not physically. While it is well‐known that the distribution of zooplankton is often patchy, this aspect of their ecology has not been quantified in most lakes, especially in remote montane locations. These results illustrate how acoustic sampling can rapidly and simultaneously measure the biomass and spatial distribution of multiple trophic levels in small lakes. This capability provides unique opportunities to study the processes that generate and maintain gradients and patchiness in these components of the ecosystem.
Active acoustics are a valuable addition to ocean observatories, allowing the detection of animals throughout the water column with high temporal and spatial resolution. The resulting datasets are large and can be difficult to describe and visualize in their entirety. We assembled a suite of metrics to parsimoniously characterize the vertical distribution of animal densities, including measures of abundance, density, location, dispersion, occupancy, evenness, and aggregation. We also developed and tested an unsupervised algorithm for detecting and counting backscatter layers within an echogram, using a gradient-based classification. These metrics were used to analyze data from the Deep Echo Integrating Marine Observatory System (DEIMOS), a 38 kHz upward-looking acoustic package deployed at the MARS observatory node in Monterey Bay, CA. The metrics successfully captured biological dynamics across multiple time scales, including seasonal, diel, and tidal variability, in addition to episodic events such as transient aggregations. All metric series showed significant positive autocorrelations at lags less than 26 days and as long as 103 days. Their frequency spectra were power-law distributed with exponents between −0.84 and −1.73. This approach could be expanded to two or three dimensions, providing an effective means to quantify the temporal variability of any spatial density distribution.
Identifying biological scatterers is a perennial challenge in fisheries acoustics. Most practitioners classify backscatter based on direct sampling and frequency-difference thresholds, then integrate at a single frequency. However, this approach struggles with species mixtures, and discards multi-frequency information when integrating. Inversion methods do not have these limitations, but are seldom used, because their species identifications are often ambiguous and their algorithms complicated to implement. We address these shortcomings with a probabilistic, Bayesian inversion method. Like other inversion methods, it handles species mixtures, uses all available frequencies, and extends naturally to broadband signals. Unlike prior approaches, it leverages Bayesian priors to rigorously incorporate information from direct sampling and biological knowledge, constraining the inversion and reducing ambiguity in species identification. Because it is probabilistic, it can be trusted to run automatically: it should not produce solutions that are both wrong and confident. Unlike some data-driven machine learning models, it is based on acoustical scattering processes, so its inferences are physically interpretable. Finally, the approach is straightforward to implement using existing Bayesian libraries, and is easily parallelized for large datasets. We present examples using simulations and field data from the Gulf of Alaska, and discuss possible extensions and applications of the method.
Summary Marine surveillance radars are commonly used for radar ornithology, but they are rarely calibrated. This prevents them from measuring the radar cross‐sections (RCS) of the birds under study. Furthermore, if the birds are aggregated too closely for the radar to resolve them individually, the bulk volume reflectivity cannot be translated into a numerical density. We calibrated a commercial off‐the‐shelf marine radar, using a standard spherical target of known RCS. Once calibrated, the radar was used to measure the RCS of common and roseate terns ( Sterna hirundo L. and Sterna dougallii Montagu) tracked from a land‐based installation at their breeding colony on Great Gull Island, NY, USA. We also integrated echoes from flocks of terns, comparing these total flock cross‐sections with visual counts from photos taken at the same time as the radar measurements. The radar's calibration parameters were determined with 1% error. RCS measurements made after calibration were expected to be accurate within ±2 dB. Mean tern RCS was estimated at −28 dB relative to one square meter (dBsm), agreeing in magnitude with a simple theoretical model. RCS was 3–4 dB higher when birds’ aspect angles were broadside to the radar beam compared with head‐ or tail‐on. Integrated flock cross‐section was linearly related to the number of birds. The slope of this line, an independent estimate of RCS, was −32 dBsm, within an order of magnitude of the estimate from individual birds, and near the middle of the frequency distribution of RCS values. These results indicate that a calibrated marine radar can count the birds in an aggregation via echo integration. Field calibration of marine radars is practical, enables useful measurements, and should be done more often.
Many pelagic animals, such as krill, lanternfish, and cephalopods, migrate to deep water at dawn to avoid visual predators during daylight hours and move up toward the sea surface at dusk to search for food. This behavior is termed "diel vertical migration." Migrating animals graze on phytoplankton or zooplankton and in turn serve as food for higher trophic levels, hence providing a key mechanism for carbon export via this migration. These animals are often observed as sound-scattering layers by echosounders, but the animals causing the acoustic scattering are difficult to identify using acoustics alone. In a spring 2019 experiment in Monterey Bay, we deployed autonomous underwater and surface vehicles over a seabed-mounted upward-looking echosounder to collect environmental DNA (eDNA) with the goal of identifying the vertically migrating animals. The echosounder was installed at 890-m depth on the Monterey Accelerated Research System (MARS) seabed cabled ocean observatory, providing real-time data of acoustic backscatter from the full water column. One long-range autonomous underwater vehicle (LRAUV) carrying a Third-Generation Environmental Sample Processor (3G-ESP) acquired water samples from a sequence of layers from near surface down to ~ 290 m as directed by the distribution of animals observed by the echosounder. During the sampling of each layer, the LRAUV ran on a tight circular yo-yo trajectory directly above the echosounder, remaining in its beam by acoustically tracking a station-keeping Wave Glider on the sea surface marking the echosounder's latitude and longitude. The persistent and simultaneous acoustic observation and eDNA acquisition enables identification of animals at precise locations to better understand their vertical migration behaviors. We present the methods and the system performance in the experiment.
Abstract A 38 kHz upward-facing echosounder was deployed on the seafloor at a depth of 875 m in Monterey Bay, CA, USA (36° 42.748’ N, 122° 11.214’ W) from 27 February 2009 to 18 August 2010. This 18-month record of acoustic backscatter was compared to oceanographic time series from a nearby data buoy to investigate the responses of animals in sound-scattering layers to oceanic variability at seasonal and sub-seasonal time scales. Pelagic animals, as measured by acoustic backscatter, moved higher in the water column and decreased in abundance during spring upwelling, attributed to avoidance of a shoaling oxycline and advection offshore. Seasonal changes were most evi-dent in a non-migrating scattering layer near 500 m depth that disappeared in spring and reappeared in summer, building to a seasonal maximum in fall. At sub-seasonal time scales, similar responses were observed after individual upwelling events, though they were much weaker than the seasonal relationship. Correlations of acoustic backscatter with oceanographic variability also differed with depth. Backscatter in the upper water column decreased immediately following upwelling, then increased approximately 20 days later. Similar correlations existed deeper in the water column, but at increasing lags, suggesting that near-surface productivity propagated down the water column at 10-15 m d ‒1 , consistent with sinking speeds of marine snow measured in Monterey Bay. Sub-seasonal variability in backscatter was best correlated with sea-surface height, suggesting that passive physical transport was most important at these time scales.
Marine area-based conservation measures including no-take zones (areas with no fishing allowed) are often designed through lengthy processes that aim to optimize for ecological and social objectives. Their (semi) permanence generates high stakes in what seems like a one-shot game. In this paper, we theoretically and empirically explore a model of short-term area-based conservation that prioritizes adaptive co-management: temporary areas closed to fishing, designed by the fishers they affect, approved by the government, and adapted every 5 years. In this model, no-take zones are adapted through learning and trust-building between fishers and government fisheries scientists. We use integrated social-ecological theory and a case study of a network of such fisheries closures (“fishing refugia”) in northwest Mexico to hypothesize a feedback loop between trust, design, and ecological outcomes. We argue that, with temporary and adaptive area-based management, social and ecological outcomes can be mutually reinforcing as long as initial designs are ecologically “good enough” and supported in the social-ecological context. This type of adaptive management also has the potential to adapt to climate change and other social-ecological changes. This feedback loop also predicts the dangerous possibility that low trust among stakeholders may lead to poor design, lack of ecological benefits, eroding confidence in the tool’s capacity, shrinking size, and even lower likelihood of social-ecological benefits. In our case, however, this did not occur, despite poor ecological design of some areas, likely due to buffering by social network effects and alternative benefits. We discuss both the potential and the danger of temporary area-based conservation measures as a learning tool for adaptive co-management and commoning.
