Automatic Detection of Marine Mammal from Aerial Imagery

The following technical report describes the approach and algorithm used to detect marine mammals from aerial imagery taken from manned/unmanned platform. The aim is to automate the process of counting the population of dugongs and other mammals. We have developed and algorithm that automatically presents to a user a number of possible candidates of these mammals. We tested the algorithm in two distinct datasets taken from different altitudes. Analysis and discussion is presented in regards with the complexity of the input datasets, the detection performance.

Automated marine mammal detection from aerial imagery

This paper presents two algorithms to automate the detection of marine species in aerial imagery. An algorithm from an initial pilot study is presented in which morphology operations and colour analysis formed the basis of its working principle. A second approach is presented in which saturation channel and histogram-based shape profiling were used. We report on performance for both algorithms using datasets collected from an unmanned aerial system at an altitude of 1000 ft. Early results have demonstrated recall values of 48.57% and 51.4%, and precision values of 4.01% and 4.97%.

Automating marine mammal detection in aerial images captured during wildlife surveys: A deep learning approach

Aerial surveys conducted using manned or unmanned aircraft with customized camera payloads can generate a large number of images. Manual review of these images to extract data is prohibitive in terms of time and financial resources, thus providing strong incentive to automate this process using computer vision systems. There are potential applications for these automated systems in areas such as surveillance and monitoring, precision agriculture, law enforcement, asset inspection, and wildlife assessment. In this paper, we present an efficient machine learning system for automating the detection of marine species in aerial imagery. The effectiveness of our approach can be credited to the combination of a well-suited region proposal method and the use of Deep Convolutional Neural Networks (DCNNs). In comparison to previous algorithms designed for the same purpose, we have been able to dramatically improve recall to more than 80% and improve precision to 27% by using DCNNs as the core approach.

A spatial assessment of the risk to a mobile marine mammal from bycatch

Several species of marine mammals are at risk of extinction from being captured as bycatch in commercial fisheries. Various approaches have been developed and implemented to address this bycatch problem, including devices and gear changes, time and area closures and fisheries moratoria. Most of these solutions are difficult to implement effectively, especially for artisanal fisheries in developing countries and remote regions. Re-zoning of the Great Barrier Reef World Heritage Area (GBRWHA) in 2004 closed 33% of the region to extractive activities, including commercial fishing. However, the impact of re-zoning and the associated industry restructuring on a threatened marine mammal, the dugong (Dugong dugon), is difficult to quantify. Accurate information on dugong bycatch in commercial nets is unavailable because of the large geographic extent of the GBRWHA, the remoteness of the region adjacent to the Cape York Peninsula where most dugongs occur and the artisanal nature of the fishery. In the face of this uncertainty, a spatial risk-assessment approach was used to evaluate the re-zoning and associated industry restructuring for their ability to reduce the risk of dugong bycatch from commercial fisheries netting. The new zoning arrangements appreciably reduced the risk of dugong bycatch by reducing the total area where commercial netting is permitted. Netting is currently not permitted in 67% of dugong habitats of high conservation value, a 56% improvement over the former arrangements. Re-zoning and industry restructuring also contributed to a 22% decline in the spatial extent of conducted netting. Spatial risk assessment approaches that evaluate the risk of mobile marine mammals from bycatch are applicable to other situations where there is limited information on the location and intensity of bycatch, including remote regions and developing countries where resources are limited.

Report on Marine Mammal Unusual Mortality Event UMESE0501Sp: Multispecies Mass Stranding of Pilot Whales (Globicephala macrorhynchus), Minke Whale (Balaenoptera acutorostrata), and Dwarf Sperm Whales (Kogia sima) in North Carolina on 15-16 January 2005

On 15-16 January 2005, three offshore species of cetaceans (33 short-finned pilot whales, Globicephala macrorhynchus, one minke whale, Balaenoptera acutorostrata, and two dwarf sperm whales, Kogia sima) stranded alive on the beaches of North Carolina. The pilot whales stranded near Oregon Inlet, the minke whale in northern North Carolina, and the dwarf sperm whales near Cape Hatteras. Live strandings of three species in one weekend was unique in North Carolina and qualified as an Unusual Mortality Event. Gross necropsies were conducted on 16-17 January 2005 on 27 pilot whales, two dwarf sperm whales, and the minke whale. Samples were collected for clinical pathology, parasitology, gross pathology, histopathology, microbiology and serology. There was variation in the number of animals sampled for each collection type, however, due to carcasses washing off the beach or degradation in carcass condition during the course of the response. Comprehensive histologic examination was conducted on 16 pilot whales, both dwarf sperm whales, and the minke whale. Limited organ or only head tissue suites were obtained from nine pilot whales. Histologic examination of tissues began in February 2005 and concluded in December 2005 when final sampling was concluded. Neither the pilot whales nor dwarf sperm whales were emaciated although none had recently ingested prey in their stomachs. The minke whale was emaciated; it was likely a dependent calf that became separated from the female. Most serum biochemistry abnormalities appear to have resulted from the stranding and indicated deteriorating condition from being on land for an extended period. Three pilot whales had clinical evidence of pre-existing systemic inflammation, which was supported by histopathologic findings. Although gross and histologic lesions involving all organ systems were noted, consistent lesions were not observed across species. Verminous pterygoid sinusitis and healed fishery interactions were seen in pilot whales but neither of these changes were causes of debilitation or death. In three pilot whales and one dwarf sperm whale there was evidence of clinically significant disease in postcranial tissues which led to chronic debilitation. Cardiovascular disease was present in one pilot whale and one dwarf sperm whale; musculoskeletal disease and intra-abdominal granulomas were present in two pilot whales. These lesions were possible, but not definitive, causal factors in the stranding. Remaining lesions were incidental or post-stranding. The minke whale and three of five tested pilot whales had positive morbillivirus titers (≥1:8 with one at >1:256), but there was no histologic evidence of active viral infection. Parasites (nematodes, cestodes, and trematodes) were collected from 26 pilot whales and two dwarf sperm whales. Sites of collection included stomach, nasal/pterygoid, peribullar sinuses, blubber, and abdominal cavity. Parasite species, locations and loads were within normal limits for free-ranging cetaceans and were not considered causative for the stranding event. Gas emboli lesions which were considered consistent with or diagnostic of sonarassociated strandings of beaked whales or small cetaceans were not found in the whales stranded as part of UMESE0501Sp. Twenty-five heads were examined with nine specific anatomic locations of interest: extramandibular fat, intramandibular fat, auditory meatus, peribullar acoustic fat, peribullar soft tissue, peribullar sinus, pterygoid sinus, melon, and brain. The common finding in all examined heads was verminous pterygoid sinusitis. Intramandibular adipose tissue reddening, typically adjacent to the vascular plexus, was observed in some individuals and could represent localized hemorrhage resulting from vascular rete rupture, hypostatic congestion, or erythrocyte rupture during the freeze/thaw cycle. One cetacean had peracute to acute subdural hemorrhage that likely occurred from thrashing on the beach post-stranding, although its occurrence prior to stranding cannot be excluded. Information provided to NMFS by the U.S. Navy indicated routine tactical mid-frequency sonar operations from individual surface vessels over relatively short durations and small spatial scales within the area and time period investigated. No marine mammals were detected by marine mammal observers on operational vessels; standard operating procedure for surface naval vessels operating mid-frequency sonar is the use of trained visual lookouts using high-powered binoculars. Sound propagation modeling using information provided to NMFS indicated that acoustic conditions in the vicinity likely depended heavily on position of the receivers (e.g., range, bearing, depth) relative to that of the sources. Absent explicit information on the location of animals meant that it was not possible to estimate received acoustic exposures from active sonar transmissions. Nonetheless, the event was associated in time and space with naval activity using mid-frequency active sonar. It also had a number of features in common (e.g., the “atypical” distribution of strandings involving multiple offshore species, all stranding alive, and without evidence of common infectious or other disease process) with other sonar-related cetacean mass stranding events. Given that this event was the only stranding of offshore species to occur within a 2-3 day period in the region on record (i.e., a very rare event), and given the occurrence of the event simultaneously in time and space with a naval exercise using active sonar, the association between the naval sonar activity and the location and timing of the event could be a causal rather than a coincidental relationship. However, evidence supporting a definitive association is lacking, and, in particular, there are differences in operational/environmental characteristics between this event and previous events where sonar has apparently played a role in marine mammal strandings. This does not preclude behavorial avoidance of noise exposure. No harmful algal blooms were present along the Atlantic coast south of the Chesapeake Bay during the months prior to the event. Environmental conditions, including strong winds, changes in upwelling- to downwelling-favorable conditions, and gently sloping bathymetry, were consistent with conditions which have been correlated with other mass strandings. In summary, we did not find commonality in gross and histologic lesions that would indicate a single cause for this stranding event. Three pilot whales and one dwarf sperm whale had debilitating conditions identified that could have contributed to stranding, one pilot whale had a debilitating condition (subdural hemorrhage) that could have been present prior to or resulting from stranding. While the pilot and dwarf sperm whale strandings may have had a common cause, the minke whale stranding was probably just coincidental. On the basis of examination of physical evidence in the affected whales, however, we cannot definitively conclude that there was or was not a causal link between anthropogenic sonar activity or environmental conditions (or a combination of these factors) and the strandings. Overall, the cause of UMESE0501Sp in North Carolina is not and likely will not be definitively known. (PDF contains 240 pages)

Workshop on marine mammal research & monitoring in the National Marine Sanctuaries, Wailea, Maui, Hawaii, 28 November 1999

The Second National Workshop on Marine Mammal Research and Monitoring in the National Marine Sanctuaries was held on 28 November 1999 in Maui, Hawaii. The workshop preceded the Thirteenth Biennial Conference on the Biology of Marine Mammals, and provided an opportunity to review and promote marine mammal research and monitoring in the National Marine Sanctuaries (NMS). The purpose of the workshop was to bring together researchers and sanctuary staff and to improve marine mammal research and monitoring throughout the sanctuaries. Discussion topics included: potential multi-sanctuary projects, sources of funding for multi-sanctuary projects, services and equipment for researchers through the sanctuaries, consolidating small levels of funding, help in funding and support for writing up data, publishing documents in Technical Memoranda, and letters of support. Representatives from the NMS national office and nine sanctuaries provided participants with overviews of marine mammal research within the sanctuaries. Presentations were also given by representatives from the National Marine Fisheries Service’s Permits and Health and Stranding programs. During the breakout working groups, there were several comments and suggestions consistent among each of the groups to improve marine mammal research. Each group emphasized the need to improve communication among researchers and to better share data. These suggestions included web-based information networks, advisory panels, and workshops. Regionally based research projects were also emphasized. In order to best study marine mammal populations, collaborative studies must take place throughout multiple sanctuaries. In order to achieve these large scale studies, funding and staffing must be directed towards these studies and distributed among each of the sanctuaries so that they may all be able to have the staffing, equipment, and vessels necessary to achieve a collaborative, ecosystem-based, regional marine mammal monitoring program. It will take several years to achieve all of the suggestions from the workshop, but thanks to the workshop participants, the National Marine Sanctuary Program has begun to direct marine mammal research and monitoring in order to achieve the goals of the workshop. This document provides a summary of the workshop with a focus on key points/main issues. We have included contact information intended to encourage continued collaboration among the individuals and organizations represented at the 1999 Marine Mammal Research and Monitoring in the National Marine Sanctuaries Workshop. (PDF contains 71 pages.)

Marine Mammal Strandings in the United States: Proceedings of the Second Marine Mammal Stranding Workshop, Miami, Florida, December 3-5, 1987

Over a decade ago, in August 1977, the First Marine Mammal Stranding Workshop was convened in Athens, Georgia. That workshop, organized by j.R. Geraci and D.J. St. Aubin, not only considered biology and pathology of stranded marine mammals, but it also served as a springboard for the formation of regional marine mammal stranding networks in the United States. The ramifications have been extremely important to the field of marine mammalogy since, for some species, examination or rehabilitation of stranded specimens serves as virtually the only source of information on distribution, anatomy, physiology, reproduction, and pathology. The First Marine Mammal Stranding Workshop led to increased awareness of the marine mammals themselves, as well as the logistic and legal factors associated with effective handling of the animals. A number of individuals indicated that they felt that a Second Marine Mammal Stranding Workshop held prior to the Seventh Biennial Conference on the Biology of Marine Mammals (Miami, Florida; December 1987) would be both timely and productive. Accordingly, we organized the workshop and scheduled it to occur on 3-5 December. Our goals for the workshop were several, including 1) providing descriptions of some research, especially new techniques, regarding stranded marine mammals; 2) providing a forum where scientists could interact and possibly initiate cooperative research activities; 3) presenting information regarding procedures used effectively to handle stranded animals; 4) assessing ways to standardize data and specimen collection, archiving, and retrieval; and 5) providing a forum for assessing accomplishments and status of regional stranding networks to date, as well as for making recommendations regarding future activities of the networks. Nearly 100 individuals representing Federal and State governments, academic institutions, the oceanarium industry, consulting groups, conservation organizations, and the private sector attended the workshop (see Workshop Participants, this volume). (PDF file contains 166 pages.)

Estimates of Marine Mammal, Sea Turtle, and Seabird Mortality in the California Drift Gillnet Fishery for Swordfish and Thresher Shark, 1996–2002

Estimates of incidental marine mammal, sea turtle, and seabird mortality in the California drift gillnet fishery for broadbill swordfish, Xiphias gladius, and common thresher shark, Alopias vulpinus, are summarized for the 7-year period, 1996 to 2002. Fishery observer coverage was 19% over the period (3,369 days observed/17,649 days fished). An experiment to test the effectiveness of acoustic pingers on reducing marine mammal entanglements in this fishery began in 1996 and resulted in statistically significant reductions in marine mammal bycatch. The most commonly entangled marine mammal species were the short-beaked common dolphin, Delphinus delphis; California sea lion, Zalophus californianus; and northern right whale dolphin, Lissodelphis borealis. Estimated mortality by species (CV and observed mortality in parentheses) from 1996 to 2002 is 861 (0.11, 133) short-beaked common dolphins; 553 (0.16, 103) California sea lions; 151 (0.25, 31) northern right whale dolphins; 150 (0.21, 27) northern elephant seals, Mirounga angustirostris; 54 (0.41, 10) long-beaked common dolphins, Delphinus capensis; 44 (0.53, 6) Dall’s porpoise, Phocoenoides dalli; 19 (0.60, 5) Risso’s dolphins, Grampus griseus; 11 (0.71, 2) gray whales, Eschrichtius robustus; 7 (0.83, 2) sperm whales, Physeter macrocephalus; 7 (0.96, 1) short-finned pilot whales, Globicephala macrorhychus; 12 (1.06, 1) minke whales, Balaenoptera acutorostrata; 5 (1.05, 1) fin whales, Balaenoptera physalus; 11 (0.68, 2) unidentified pinnipeds; 33 (0.52, 4) leatherback turtles, Dermochelys coriacea; 18 (0.57, 3) loggerhead turtles, Caretta caretta; 13 (0.73, 3) northern fulmars, Fulmarus glacialis; and 6 (0.86, 2) unidentified birds.

Strandings as indicators of marine mammal biodiversity and human interactions off the coast of North Carolina

The adjacency of 2 marine biogeographic regions off Cape Hatteras, North Carolina (NC), and the proximity of the Gulf Stream result in a high biodiversity of species from northern and southern provinces and from coastal and pelagic habitats. We examined spatiotemporal patterns of marine mammal strandings and evidence of human interaction for these strandings along NC shorelines and evaluated whether the spatiotemporal patterns and species diversity of the stranded animals reflected published records of populations in NC waters. During the period of 1997–2008, 1847 stranded animals were documented from 1777 reported events. These animals represented 9 families and 34 species that ranged from tropical delphinids to pagophilic seals. This biodiversity is higher than levels observed in other regions. Most strandings were of coastal bottlenose dolphins (Tursiops truncatus) (56%), harbor porpoises (Phocoena phocoena) (14%), and harbor seals (Phoca vitulina) (4%). Overall, strandings of northern species peaked in spring. Bottlenose dolphin strandings peaked in spring and fall. Almost half of the strandings, including southern delphinids, occurred north of Cape Hatteras, on only 30% of NC’s coastline. Most stranded animals that were positive for human interaction showed evidence of having been entangled in fishing gear, particularly bottlenose dolphins, harbor porpoises, short-finned pilot whales (Globicephala macrorhynchus), harbor seals, and humpback whales (Megaptera novaeangliae). Spatiotemporal patterns of bottlenose dolphin strandings were similar to ocean gillnet fishing effort. Biodiversity of the animals stranded on the beaches reflected biodiversity in the waters off NC, albeit not always proportional to the relative abundance of species (e.g., Kogia species). Changes in the spatiotemporal patterns of strandings can serve as indicators of underlying changes due to anthropogenic or naturally occurring events in the source populations.

C-values of seven marine mammal species determined by flow cytometry

C-values, which estimate genome size, have puzzled geneticists for years because they bear no relationship to organismal complexity. Though C-values have been estimated for thousands of species, considerably more data are required in order to better understanding genome evolution. This is particularly true for mammals, in which C-values are known for less than 8% of the total number of mammalian species. Among marine mammals, a C-value has been estimated only for the bottlenose dolphin (Tursiops truncatus). Thus examination of additional species of marine mammals is necessary for comparative purposes. It will enable a better understanding of marine mammal genome evolution, and it is also relevant to conservation, because larger genome size has been linked to increased likelihood of extinction in some plant and animal groups. Our study presents C-values of seven marine mammal species, including five cetacean species that are endangered to varying degrees. Similarly to the results for other groups, our results suggest that larger genome size in cetaceans is related to an increased likelihood of extinction.

Genetic status of an endemic marine mammal, the Australian fur seal, following historical harvesting

Genetic variation, and the way in which it is partitioned among populations, has implications for a species’ survival and evolutionary potential. Such information is particularly important for the successful conservation and management of species that have experienced past human impacts and potential losses of genetic diversity. Overharvesting of the Australian fur seal Arctocephalus pusillus doriferus in the 18th and 19th centuries resulted in severe population reductions and elimination of an estimated 17 of 26 colonies. Currently, the subspecies is recovering and c. 20 000 pups are produced annually at 13 colony sites, most of which are situated in Bass Strait in south-eastern Australia. Genetic analysis of samples collected from pups captured at nine colonies revealed no difference in allelic diversity or heterozygosity at five microsatellite loci and no differences in haplotype diversity within a 344 bp region of the mitochondrial DNA control region. There was some evidence for isolation by distance but the program STRUCTURE predicted a single cluster of individuals. Gene flow among colonies appears to be substantial at present, indicating that the Australian fur seal is currently a single, panmictic unit.

Seabird and marine mammal management options in the face of climate change

Climate change is already impacting Australia’s oceans. Responses by marine life to both climate variability and change have been documented for low trophic levels, however, responses for Australia’s iconic higher trophic level marine taxa are poorly understood, including for many conservation-dependent seabirds and marine mammals. We report initial results from a national study evaluating impacts an adaptation options. Individual time series and combined analyses show consistent responses to historical climate signals, however, improved monitoring protocols are needed to maximize detection of any climate-related demographic signals. Despite difference in sampling , the development of regional multi-species-indices of environmental change provides robust climate indicators over large regions.

Determining trends and environmental drivers from long-term marine mammal and seabird data : examples from Southern Australia

Climate change is acknowledged as an emerging threat for top-order marine predators, yet obtaining evidence of impacts is often difficult. In south-eastern Australia, a marine global warming hotspot, evidence suggests that climate change will profoundly affect pinnipeds and seabirds. Long-term data series are available to assess some species' responses to climate. Researchers have measured a variety of chronological and population variables, such as laying dates, chick or pup production, colony-specific abundance and breeding success. Here, we consider the challenges in accurately assessing trends in marine predator data, using long-term data series that were originally collected for other purposes, and how these may be driven by environmental change and variability. In the past, many studies of temporal changes and environmental drivers used linear analyses and we demonstrate the (theoretical) relationship between the magnitude of a trend, its variability, and the duration of a data series required to detect a linear trend. However, species may respond to environmental change in a nonlinear manner and, based on analysis of time-series from south-eastern Australia, it appears that the assumptions of a linear model are often violated, particularly for measures of population size. The commonly measured demographic variables exhibit different degrees of variation, which influences the ability to detect climate signals. Due to their generally lower year-to-year variability, we illustrate that monitoring of variables such as mass and breeding chronology should allow detection of temporal trends earlier in a monitoring programme than observations of breeding success and population size. Thus, establishing temporal changes with respect to climate change from a monitoring programme over a relatively short time period requires careful a priori choice of biological variables. © 2014 Springer-Verlag Berlin Heidelberg.