Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Year-round spatiotemporal distribution of harbour porpoises within and around the Maryland wind energy area

  • Jessica E. Wingfield ,

    jwingfie@cbl.umces.edu

    Affiliation Chesapeake Biological Laboratory, University of Maryland Center for Environmental Science, Solomons, Maryland, United States of America

    ORCID http://orcid.org/0000-0003-3679-8184

  • Michael O’Brien,

    Affiliation Chesapeake Biological Laboratory, University of Maryland Center for Environmental Science, Solomons, Maryland, United States of America

  • Vyacheslav Lyubchich,

    Affiliation Chesapeake Biological Laboratory, University of Maryland Center for Environmental Science, Solomons, Maryland, United States of America

  • Jason J. Roberts,

    Affiliation Marine Geospatial Ecology Laboratory, Nicholas School of the Environment, Duke University, Durham, North Carolina, United States of America

  • Patrick N. Halpin,

    Affiliation Marine Geospatial Ecology Laboratory, Nicholas School of the Environment, Duke University, Durham, North Carolina, United States of America

  • Aaron N. Rice,

    Affiliation Biacoustics Research Program, Cornell Lab of Ornithology, Cornell University, Ithaca, New York, United States of America

  • Helen Bailey

    Affiliation Chesapeake Biological Laboratory, University of Maryland Center for Environmental Science, Solomons, Maryland, United States of America

Year-round spatiotemporal distribution of harbour porpoises within and around the Maryland wind energy area

  • Jessica E. Wingfield, 
  • Michael O’Brien, 
  • Vyacheslav Lyubchich, 
  • Jason J. Roberts, 
  • Patrick N. Halpin, 
  • Aaron N. Rice, 
  • Helen Bailey
PLOS
x

Abstract

Offshore windfarms provide renewable energy, but activities during the construction phase can affect marine mammals. To understand how the construction of an offshore windfarm in the Maryland Wind Energy Area (WEA) off Maryland, USA, might impact harbour porpoises (Phocoena phocoena), it is essential to determine their poorly understood year-round distribution. Although habitat-based models can help predict the occurrence of species in areas with limited or no sampling, they require validation to determine the accuracy of the predictions. Incorporating more than 18 months of harbour porpoise detection data from passive acoustic monitoring, generalized auto-regressive moving average and generalized additive models were used to investigate harbour porpoise occurrence within and around the Maryland WEA in relation to temporal and environmental variables. Acoustic detection metrics were compared to habitat-based density estimates derived from aerial and boat-based sightings to validate the model predictions. Harbour porpoises occurred significantly more frequently during January to May, and foraged significantly more often in the evenings to early mornings at sites within and outside the Maryland WEA. Harbour porpoise occurrence peaked at sea surface temperatures of 5°C and chlorophyll a concentrations of 4.5 to 7.4 mg m-3. The acoustic detections were significantly correlated with the predicted densities, except at the most inshore site. This study provides insight into previously unknown fine-scale spatial and temporal patterns in distribution of harbour porpoises offshore of Maryland. The results can be used to help inform future monitoring and mitigate the impacts of windfarm construction and other human activities.

Introduction

With the development of offshore energy infrastructure and increases in ship traffic, the world’s oceans are becoming busier and noisier [1, 2]. Noisier oceans are a concern for marine mammals as they use sound for communication, foraging, and navigation [3, 4]. Increased background noise from ships, dredging, pile-driving, seismic surveys and other anthropogenic sources has caused a variety of behavioural responses in many cetacean species [59]. It is critical to understand the fine-scale spatial and temporal distribution of cetaceans in areas of planned developments, like offshore windfarms, in order to inform regulators and developers on how to most effectively avoid and minimize negative impacts during the construction phase when loud sounds may be emitted.

Over the last decade there has been rapid development of offshore wind energy off the coast of the United Kingdom and elsewhere in Europe [10]. Disturbance to cetaceans may occur during pile-driving of the wind turbine foundation and in response to increased vessel traffic associated with the construction [11]. The harbour porpoise, Phocoena phocoena, is the most common and widely distributed cetacean in European waters [12], and has therefore been the focus of many studies on the effects of offshore wind turbine construction. The responses of porpoises to windfarms have varied depending on the life-cycle phase of the windfarm (construction, early operation, long-term operation). Harbour porpoises decreased their acoustic activity for up to 24 hours at a distance of 18 km from the Danish Horns Rev II windfarm following pile-driving [13]. A study conducted during the operational phase of an offshore windfarm in the Dutch North Sea found that porpoise presence had increased in the area, possibly due to an increase in fish or the absence of fishing vessels [14]. In contrast, Teilmann & Carstensen [15] observed a significant decline in porpoise echolocation activity from 2003 to 2012 relative to baseline levels in 2001 and 2002 inside a windfarm constructed in 2002–2003 in the Danish western Baltic Sea. However, there was a significant increase in encounter rate and echolocation activity in 2011 and 2012 relative to previous years (2003–2009) [15].

The Maryland Wind Energy Area (WEA) is located 20 to 40 km offshore of Maryland in the northwestern Atlantic, and is approximately 324 km2. The Gulf of Maine/Bay of Fundy population of harbour porpoises in the northwestern Atlantic consists of approximately 80,000 individuals [16, 17]. They do not appear to follow a specific migratory route nor do they have a temporally coordinated migration, but they typically occur off New Jersey to North Carolina in winter (January to March), and from the Bay of Fundy to New Jersey in spring, summer, and fall [1821] (Fig 1). Annually, an estimated 709 harbour porpoises from this stock are incidentally bycaught in fisheries in US and Canadian waters [17]. Fisheries bycatch is considered one of the single greatest threats facing marine mammals in the United States [22]. Despite there being a number of both aerial and boat-based visual surveys conducted offshore of Maryland, harbour porpoises have been sighted very few times [23, 24]. Year-round distribution of harbour porpoises off Maryland is therefore not well understood. In an attempt to increase understanding of cetacean distribution off the east coast of the United States, Roberts et al. [25] developed a habitat-based density model using aerial and boat-based sightings data to predict year-round harbour porpoise densities. Despite there being no sightings south of New Jersey included in the model, the predictions extend to Cape Hatteras, North Carolina. However, the few porpoise sightings reported off Maryland and strandings on beaches in North Carolina, justified the model’s extension southwards beyond New Jersey [24, 26, 27].

thumbnail
Fig 1. Map of the northeastern coast of the United States and study location.

Displayed is the Maryland Wind Energy Area (yellow) and the four C-POD sites (inset). Bar plot shows the mean (± SE) number of hours per day that porpoises were detected throughout the study period.

https://doi.org/10.1371/journal.pone.0176653.g001

Here, we aimed to characterize year-round patterns in harbour porpoise occurrence and foraging behaviour in relation to temporal and environmental variables within and around the Maryland WEA. Data were collected using passive acoustic devices called C-PODs, which detect and log cetacean echolocation clicks (Chelonia Ltd., UK). C-PODs, and their predecessor, T-PODs, have been widely used to detect both dolphins and porpoises [11, 13, 2831]. Harbour porpoises can be difficult to observe during boat-based or aerial surveys because of their small size and elusive behaviour [32]. Visual detectability of harbour porpoises significantly declines in sea states of Beaufort 2 or higher, limiting the number of reliable sightings available for abundance estimates [33, 34]. Therefore, passive acoustic monitoring is a cost-effective alternative for detecting porpoises in any sea state and at any time of day. Passive acoustic devices also allow for the collection of continuous and long-term occurrence data, and therefore provide a useful tool for validating patterns of relative abundance from habitat-based predictive models [35, 36]. The acoustic data collected in this study were also used to evaluate the accuracy of Roberts et al.’s [25] habitat-based density predictions for harbour porpoises offshore of Maryland.

Materials and methods

Data collection and processing

The Maryland WEA is located approximately 20–40 km offshore of Ocean City, Maryland, USA (Fig 1). The substrate within and around the WEA is predominantly sand [37]. The eastern edge of the WEA has high ship traffic, where ships pass as they approach or exit the Delaware Bay [38]. Passive acoustic monitoring of marine mammals in the area began in November 2014 to obtain baseline data prior to windfarm construction. C-PODs (Version 1, Chelonia Ltd., UK) were deployed at four sites up to 63 km offshore, including within and up to 35 km outside of the WEA to detect harbour porpoises (Phocoena phocoena) (Fig 1). Moorings were bottom-anchored, with the C-POD positioned approximately 5 m from the sea floor, in approximately 20–45 m water depth. C-PODs were recovered and re-deployed approximately every three months. Data in this study extend to May 2016. C-PODs continuously monitor the 20–160 kHz frequency range, logging the center frequency, frequency trend, duration, intensity, and bandwidth of tonal clicks. High-frequency harbour porpoise clicks, which have a peak frequency of approximately 131 kHz and range from 110 to 180 kHz [39, 40], can be detected by a C-POD from several hundred meters away [41]. The KERNO classifier within the CPOD.exe software (Chelonia Ltd., v. 2.044) then identifies click trains (sequences of at least 5 clicks) and assesses the likelihood of each click-train belonging to a dolphin or porpoise as either high (CetHi), medium (CetMed), or low (CetLow) [42]. A conservative approach was taken and only CetHi and CetMed harbour porpoise click trains were included in the analyses [31, 35]. A study combining T-POD detections (the predecessor of the C-POD) with visual observations determined the false detection rate to be very low indicating that the click train algorithm is efficient and conservative [43]. The data were exported and formatted to an hourly resolution with the number of minutes in each hour that a harbour porpoise click train was detected. Only hours with a complete 60 minutes of recording were used in the analyses.

Temporal occurrence of porpoises

Temporal patterns in harbour porpoise detections were investigated for each site using generalized autoregressive moving average (GARMA) models [44]. This type of model accommodates non-Gaussian time-series data (e.g. auto-correlated count series), with potentially time-dependent covariates. The response variable in the model was the number of minutes per hour that harbour porpoises were detected and the explanatory variables were hour of the day (Eastern Standard Time, EST) and Julian day. To model cyclical annual and daily patterns, we applied two pairs of sinusoidal functions: sin(2πt/d) and cos(2πt/d), where period d is one day or one year, and t is the hour of the day or Julian day, respectively. Models were fit in the statistical software R [45] using the package gamlss.util [46]. Model selection was performed using the Akaike Information Criterion (AIC). Autocorrelation and partial autocorrelation plots were used to examine the remaining serial dependence (if any) in the final models’ residuals. Model fit was assessed by examining the residual plots [46].

Foraging behaviour

A subset of the data was created that consisted of only hours during which harbour porpoises were detected. The C-POD custom software was used to calculate and export inter-click intervals (ICI), as the number of micro-seconds between clicks, for each click train detected. The inter-click intervals of harbour porpoise click trains have been found to vary in duration depending on the behaviour of the porpoise [4749]. Click trains associated with foraging have lower ICIs and faster repetition rates than those associated with travelling [49, 50]. An ICI of 10 ms or less was used as the threshold to infer foraging [5055]. Therefore, hours during which at least one of the ICIs was 10 ms or less were considered “foraging positive”, and hours with ICIs greater than 10 ms were deemed “foraging negative”. The presence/absence of foraging behaviour in each hour from the subset data was modeled for each site using generalized additive models (GAMs) with a binomial error distribution and logit link function [56]. The explanatory variables were hour of the day (EST) and Julian day. Due to the cyclical nature of the explanatory variables, a circular spline was used. Models were fit using the R package mgcv [57]. AIC was used to select the best model, and goodness of fit was evaluated using confusion matrices [58] and area under the receiver operating characteristic (ROC) curve [59]. Confusion matrices compare the binary predictions from the model to the observed presence/absence values [58], in this case the presence or absence of foraging. The closer the area under the ROC curve is to 1, the better the model fit [59]. Confusion matrices were calculated using the R package PresenceAbsence [60] and the area under the ROC curves was calculated using the package ROCR [61].

Environmental data analysis

Using the full data set, the proportion of hours per week during which harbour porpoises were detected was compared to weekly median sea surface temperature (°C, SST), the natural log of chlorophyll a concentration (mg m-3), and fraction of the moon illuminated. Due to their small size and that they consume small forage-fish, it is difficult for harbour porpoises to retain large energy stores, and therefore they forage almost continuously with generally high capture success rates [62, 63]. Fine-scale distribution of forage-fish species is difficult to assess, but due to the attraction of these fish to areas of high primary productivity, chlorophyll a concentration can be used as a proxy for prey abundance [64, 65]. Although the degree of lunar illumination has been shown to affect the foraging behaviour of dolphins, its effect on porpoise foraging is not well studied [66]. Week numbers were assigned using the ISO week date standard (ISO-8601). Eight-day composites of SST (GOES Imager) and chlorophyll a concentration (Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Aqua satellite) were extracted for each day during the study period at each site using the NOAA Coastwatch tool Xtractomatic tool (http://coastwatch.pfel.noaa.gov/xtracto/) in R. Data on the fraction of the moon illuminated for each night were available from the Astronomical Applications Department of the US Naval Observatory (http://aa.usno.navy.mil/index.php). Weekly medians for each environmental variable were then calculated and compared to the proportion of hours per week with a harbour porpoise detection using a GAM with a Gaussian error distribution for each site (R package mgcv [57]). AIC was used for model selection, and the function gam.check within the mgcv package was used to assess goodness of fit by visualizing the model residuals [57]. Residual autocorrelation and partial autocorrelation plots were used to assess if any serial dependence remained uncaptured by the models.

Comparison of acoustic data with habitat-based density predictions

Roberts et al. [25] developed habitat-based density models for several species of cetaceans, including harbour porpoises, off the US east coast using aerial and boat-based sightings data. A porpoise positive hour (PPH) is an hour during which the C-POD software identified at least one porpoise click train. Roberts et al.’s monthly density estimates [25] were compared with the median number of PPHs per day, total PPHs per month, maximum PPHs per day, and proportion of days per month harbour porpoises were present in the study area offshore of Maryland based on our acoustic detection data using Spearman’s rank correlation tests for each site [35, 36].

Results

C-PODs were deployed and recording for a median 521 days from 4th November 2014 to 18th May 2016 (Table 1). Instrument loss and malfunction resulted in some data gaps at sites 2 and 4 (Fig 2). Harbour porpoises were detected during the greatest proportion of days at the most inshore site, site 1, but were detected for the most hours at the farther offshore site, site 3 (Table 1).

thumbnail
Table 1. Summary of the harbour porpoise acoustic data collected at each of the four sites offshore of Maryland.

https://doi.org/10.1371/journal.pone.0176653.t001

thumbnail
Fig 2. The C-POD deployment periods.

Green indicates a complete, uninterrupted dataset and a blank space indicates there were no data during the corresponding deployment period, either due to instrument loss or malfunction.

https://doi.org/10.1371/journal.pone.0176653.g002

Temporal occurrence of harbour porpoises

A Poisson inverse-Gaussian distribution yielded the lowest AIC scores for GARMA models of the temporal patterns in harbour porpoise presence at all sites except site 4. The Poisson inverse-Gaussian distribution is well suited to handle extra-Poisson variation and has been used in a variety of disciplines [67]. A zero-inflated Poisson distribution yielded the lowest AIC score for the GARMA model of the data at site 4. Julian day was retained in all final models as a significant predictor for the number of minutes harbour porpoises were detected in an hour (S1 Table). Harbour porpoises were present significantly more often during the winter and spring months (January to May), and rarely in the summer and fall (Fig 3). There was a high degree of inter-annual variability in the number of minutes harbour porpoises were detected per day (Fig 3). There were more detections at site 1 in 2016 than in 2015, but more detections at sites 2 and 3 in 2015 compared to 2016 (Fig 3). The hour of the day was retained as a significant factor in the GARMA models for sites 1 and 2 only (S1 Table). A particularly strong diel pattern was seen at site 2 with peaks in occurrence at 01:00 and 20:00 and lowest occurrence at noon (Fig 4).

thumbnail
Fig 3. The mean (± SE) number of minutes per day during which harbour porpoises were acoustically detected at each site offshore of Maryland.

There were no data in March at site 4 due to instrument loss.

https://doi.org/10.1371/journal.pone.0176653.g003

thumbnail
Fig 4. The mean number of minutes in each hour that harbour porpoises were detected at sites 1 and 2, where the hour of the day was a significant factor in the generalized auto-regressive moving average (GARMA) models of hourly porpoise presence.

The shaded polygons represent the standard error.

https://doi.org/10.1371/journal.pone.0176653.g004

Foraging behaviour

Although we deemed hours with at least one ICI equal to or less than 10 ms as foraging positive hours, an average of 94% of foraging hours analyzed for sites 1 to 3 contained at least 5 ICIs less than 10 ms. The occurrence of foraging at site 4 was not modeled because harbour porpoises were present for only 12 hours in total and were identified as foraging for 6 of those hours at that site in April, May and October 2015, and January 2016. In contrast, harbour porpoises foraged during 61% of the hours they were present at site 3 from November to May (Fig 5). Julian day and hour of the day were both retained in the final models for sites 1, 2, and 3 (Table 2). Both of the variables had a significant relationship with the foraging behaviour of harbour porpoises at sites 2 and 3 (Table 2). At site 2, the proportion of hours during which foraging activity occurred decreased from January (0.51) to April (0.32), before rising again in May (0.50) (Fig 5). At site 3, foraging activity decreased from January (0.57) to March (0.44), and began to increase again in April (0.48) and May (0.80) (Fig 5). The high proportion of foraging activity in November at site 3 (1.00) is based on only one hour of data when porpoises were detected. Diel patterns in foraging varied between sites (S1 Fig). A decline in foraging during daytime hours was most pronounced at site 2, where the occurrence of foraging was lowest from 08:00 to 11:00 and highest in the evening to early morning (Fig 5). All models correctly predicted the presence of foraging behaviour greater than 50% of the time, and areas under the ROC curves were all greater than 0.60.

thumbnail
Fig 5. Summary of harbour porpoise foraging behaviour.

The proportion of hours harbour porpoise foraging behaviour was detected in each month (a) and the proportion of days that harbour porpoise foraging was detected in each hour (b).

https://doi.org/10.1371/journal.pone.0176653.g005

thumbnail
Table 2. The results of the binomial generalized additive models (GAM) used to relate presence/absence of foraging to hour of the day (EST) and Julian day at sites 1, 2, and 3.

https://doi.org/10.1371/journal.pone.0176653.t002

Association with environmental variables

The weekly proportion of hours harbour porpoises were present was significantly affected by SST at sites 1, 2, and 3, and by chlorophyll a concentration at sites 2 and 3 (Table 3). Harbour porpoises were present more often when SST was low at all three sites, with a peak in occurrence at approximately 5°C (Fig 6). At sites 2 and 3, harbour porpoises were present most often when the chlorophyll a concentration was approximately 4.5 to 7.4 mg m-3 (Fig 6). Data from site 4 were not modeled due to a high number of missing weeks and low number of detections.

thumbnail
Table 3. The results of the generalized additive models (GAM) used to relate the weekly occurrence of harbour porpoises to sea surface temperature and the natural logarithm of chlorophyll a concentration at sites 1, 2, and 3.

https://doi.org/10.1371/journal.pone.0176653.t003

thumbnail
Fig 6. Smoothers from the generalized additive model (GAM) for site 2.

The relationship between the proportion of hours per week that harbour porpoises were detected and (a) sea surface temperature (SST, °C) and (b) the natural logarithm of chlorophyll a concentration (mg m-3). The predictor is on each x-axis, the centered fitted values are on each y-axis, the dashed lines are error bands. Tick marks on the x-axes—rug plot—show the distribution of the underlying data. Similar smoother patterns occurred for sites 1 and 3.

https://doi.org/10.1371/journal.pone.0176653.g006

Comparison of acoustic data with habitat-based density predictions

Acoustic detections of harbour porpoises (median PPHs per day, total PPHs, maximum number of PPHs per day, and proportion of days per month detected) were significantly correlated with monthly habitat-based density estimates [25] at sites 2, 3, and 4 (Table 4). The median number of PPHs per day at site 4 was 0 in all months, and therefore there is no correlation value for this metric. The strongest correlations were between predicted densities and the total number of PPHs at site 2 (Fig 7), and the median number of PPHs per day at site 3 (Table 4). None of the acoustic detection metrics from site 1 were significantly correlated with the density estimates (Table 4). The highest predicted density of harbour porpoises at this site occurred in October [25], a month during which there were very few acoustic detections (Fig 7).

thumbnail
Table 4. Spearman’s rank correlation coefficients (p-values are in parentheses) for the median porpoise positive hours (PPHs) per day, total PPHs per month, maximum number of PPHs per day and proportion of days harbour porpoises were detected acoustically in each month compared to Roberts et al.’s [25] monthly predictions of porpoise density at each site.

https://doi.org/10.1371/journal.pone.0176653.t004

thumbnail
Fig 7. The predicted densities of harbour porpoises per month (red) and the total number of acoustically detected porpoise positive hours (PPHs) per month offshore of Maryland (black).

Predictions (in individuals per 100 km2) are from Roberts et al.’s [25] model and acoustic data were collected from November 2014 to May 2016. There were no acoustic data for March at site 4.

https://doi.org/10.1371/journal.pone.0176653.g007

Discussion

Harbour porpoises were regularly detected offshore of Maryland during the winter and spring, particularly from January to May. This is in contrast to low sighting rates during many boat-based and aerial surveys conducted over many years [23, 24]. Our study has shown that harbour porpoise occurrence is greatest during the winter and spring, and during hours of the day with reduced light or darkness. These are periods of time during which conditions for sighting this small species are generally poor, and visual surveys are expected to underestimate harbour porpoise occurrence.

The observed seasonal pattern in harbour porpoise occurrence is consistent with prior information on their general distribution [1821]. Harbour porpoises move between their summer habitat in the Bay of Fundy and Gulf of Maine to as far south as North Carolina in the winter [1821]. Harbour porpoises in this population have been found to travel a range of distances between productive habitats, where aggregations of prey may occur [68]. Our analysis of the surface chlorophyll a concentration suggested March to May is a period of high primary productivity offshore of Maryland, as it is during the winter-spring phytoplankton bloom[69, 70].

There was a high degree of inter-annual variation in the number of minutes per day that porpoises were detected. The maximum periods of time between the clicks of three free-ranging, tagged harbour porpoises in Danish waters were brief (1.6, 4, and 22 minutes), demonstrating that porpoises click regularly [53]. Because of this regularity in click production, patterns in the C-POD detection rates of clicks were assumed to reflect occurrence of harbour porpoises [71]. Inter-annual variability in occurrence is also reflected in the stranding record, as 22 strandings were recorded in 2005, and only two in 2011 and 2012 on the shorelines of Virginia [72]. The change in porpoise occurrence between years could be due to a number of biological and oceanographic factors affecting the environment offshore of Maryland and in more northern foraging grounds. For example, favourable conditions in more northern foraging grounds could delay porpoise movement southwards, leading to decreased or delayed occurrence offshore of Maryland. Chlorophyll a concentration at site 3 was greater in 2015 compared to 2016, which is likely to have led to increased prey abundance and in turn higher porpoise occurrence at this site in 2015. Further investigation into the environmental conditions in areas beyond our study area would provide insight into which factors affect broader porpoise movement up and down the coastline from year to year. Anthropogenic noise may also influence harbour porpoise occurrence and behaviour in the area, although we were unable to measure this with the C-POD.

In addition to seasonal variation in occurrence, a particularly strong diel pattern was observed at the site within the Maryland WEA (site 2), where porpoises occurred most frequently in the evening to early morning hours. This is consistent with previous studies, in which diel patterns in porpoise echolocation rates were hypothesized to be linked to prey availability [51, 55]. As visual surveys are not conducted during these hours because of reduced visibility, it is probable that porpoise occurrence at this site will be underestimated by visual surveys. It is thus recommended that future monitoring of harbour porpoise distribution in this area be conducted using passive acoustic monitoring with moored or towed hydrophones.

Foraging behaviour was analyzed using only the hours during which harbour porpoises were detected, and therefore the dataset was unevenly spaced. Temporal autocorrelation in unevenly spaced datasets cannot be correctly assessed using standard methods, and requires non-trivial estimation techniques [73, 74]. However, as foraging often occurred in non-consecutive hours and there were sometimes long gaps in foraging occurrence, we assumed that the occurrence of foraging in an hour was independent from prior and subsequent hours with foraging and did not explicitly model the autocorrelation structure. The increase in foraging activity during nighttime hours at sites 2 and 3 is consistent with patterns observed in harbour porpoise populations around the world [50, 51, 55, 75]. The diel pattern in foraging may reflect nighttime diving behaviour or prey distribution. Porpoises occurring offshore of Maryland may increase their mean dive depth during nighttime hours, as was seen in the Bay of Fundy [76], and are therefore more likely to have been detected by the bottom-moored C-POD at night. However, there was no diel pattern in dive-depth observed in Japanese waters [77]. Herring (Clupea harengus), one of the main prey species for harbour porpoises in the Northwestern Atlantic [78, 79], migrate vertically in the water column at night [80, 81]. This behaviour may make herring easier to prey upon at night, leading to an increase in porpoise foraging.

The deviance explained by each of the foraging models was low (<10%), and would likely increase in subsequent models with the inclusion of information on environmental conditions and the distribution and abundance of prey species. The fine-scale spatial and temporal distributions of harbour porpoise prey, such as herring, silver hake (Merluccius bilinearis), and pearlsides (Maurolicus weitzmani) [79], are not well known as their availability to trawl surveys is low [82]. Even if trawls effectively captured forage fish, the surveys cover large areas and data are often aggregated on a seasonal scale. Sediment type has been used in harbour porpoise habitat association models as a proxy for sandeels (also known as sand lance, Ammodytes species) [35], a key prey species for harbour porpoises in European waters [83], which prefer fine and coarse sand sediments. Sand lance also occur in the mid-Atlantic Bight [84], and the dominant sediment type in our study area is sand. Fine-scale data on prey abundance, for example using sonar-imaging technology [85], is another way to improve our understanding of factors driving porpoise foraging behaviour.

As in previous studies (e.g. [65]), we used environmental variables as proxies for prey abundance because fine-scale data on prey were not available. Chlorophyll a concentration, SST and fraction of the moon illuminated were readily available data sets. Despite being a significant factor influencing the echolocation of some dolphin species [66], lunar illumination did not significantly affect harbour porpoise echolocation offshore of Maryland. SST significantly affected harbour porpoise occurrence at all three sites. This result is consistent with Roberts et al.’s [25] model, which predicted greater harbour porpoise presence at lower SSTs. Harbour porpoises were expected to be present at colder temperatures given their seasonal distribution pattern. The peak in harbour porpoise detection rate at 5°C at all sites may also relate to the presence of herring, as catches were greatest in waters of 7–8°C in winter and 5°C in spring [86]. Summertime (June to October) concentrations of chlorophyll a in the mid-Atlantic Bight are typically below 1 mg m-3 [69], compared to values exceeding 3 mg m-3 in coastal areas during the winter-spring bloom, which begins as early as January and continues until March or April [69, 70]. It is during this winter-spring bloom that porpoise presence peaked at sites 2 and 3, at chlorophyll a concentrations of 4.5 to 7.4 mg m-3. These values are particularly high, even for this productive period in the mid-Atlantic coastal waters. Peaks in porpoise occurrence at higher chlorophyll a concentrations may be linked to prey, as areas of higher primary productivity are likely to have greater numbers of forage fish [64]. Roberts et al.’s [25] final models of summer and winter harbour porpoise density also retained productivity parameters, which had positive effects on porpoise density. The models at sites 2 and 3 relating our acoustic detections to environmental variables explained a high percentage of the deviance in weekly porpoise occurrence (78.0 and 82.1% respectively), indicating SST and chlorophyll a concentration are appropriate indicators for porpoise occurrence offshore of Maryland. The inclusion of tidal parameters may help to improve model fit for site 1 occurrence, where the deviance explained was low [32, 87].

All of the acoustic metrics for sites 2–4 were significantly correlated with monthly habitat-based predictions of harbour porpoises from sightings data recorded during aerial and boat-based surveys [25]. However, the monthly density predictions for site 1 did not correlate well with the acoustic data. Roberts et al. [25] fit two separate models, one for the winter (November to May) and another for the summer (June to October) data, as it was assumed porpoises switch environmental preferences during different phases of their annual migratory cycle. Although this strategy worked well when modelling baleen whale occurrence, it resulted in a rise in porpoise density at the May to June transition and discontinuity at the October to November transition, which was most evident at site 1. The results from this study can be used to inform how to refine and improve the density models. Although it is difficult to determine absolute densities of cetacean species using passive acoustic data [43, 88], this type of data can be a useful, independent data source to validate relative patterns and improve habitat-based models.

This study provides insight into the previously poorly understood occurrence of harbour porpoises offshore of Maryland and indicates that it is underestimated when using boat-based and aerial survey methods. The diel pattern in detections can be used to improve estimates of the detection probability for harbour porpoises during line transect surveys. Harbour porpoises occurred frequently offshore of Maryland from January to May. Consistent with our findings on their seasonal occurrence in the southern part of their range, strandings of porpoises after entanglement in fishing nets occurred primarily from January to May along the shores of Maryland, Virginia, and North Carolina [72, 89]. Scheduling wind farm construction activities in the Maryland WEA to take place during the summer months (June to September) would reduce the likelihood of disturbance to harbour porpoises. However, there are many other protected species that occur in the area, including the endangered North Atlantic right whale (Eubalaena glacialis) and endangered Atlantic sturgeon (Acipenser oxyrinchus), which should also be considered.

Supporting information

S1 Fig. The estimated relationships between the presence/absence of foraging and hour of the day (Eastern standard time, EST) at sites 1, 2, and 3.

https://doi.org/10.1371/journal.pone.0176653.s001

(TIF)

S1 Table. Estimated parameters (standard errors in parentheses) from the generalized auto-regressive moving average (GARMA) models used to relate the number of minutes harbour porpoises were present in an hour to the hour (EST) and Julian day.

Explanatory variables were sine and cosine transformed to capture the daily and seasonal cycles. PIG = Poisson inverse-Gaussian, ZIP = Zero-inflated Poisson. Asterisks represent significance level: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ‘ 1. The general formula for a GARMA model is: (1) where g(·) is the link function, μt is a conditional mean of the dependent variable, β is the regression coefficients, φj and θj are the autoregressive and moving average parameters, and p and q are the orders, respectively [1, 2].

https://doi.org/10.1371/journal.pone.0176653.s002

(DOCX)

Acknowledgments

Thank you to the many volunteers from the Chesapeake Biological Laboratory and to Fred Channel and Jason Michalec from Cornell University for their assistance in the field, and Dong Liang from the Chesapeake Biological Laboratory for his advice regarding statistical analyses. Disclaimers: The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the opinions or policies of the U.S. Government, the Maryland Department of Natural Resources, or the Maryland Energy Administration. Mention of trade names or commercial products does not constitute their endorsement by the U.S. Government or the state.

Author Contributions

  1. Conceptualization: HB ANR.
  2. Data curation: JEW.
  3. Formal analysis: JEW VL.
  4. Funding acquisition: HB ANR.
  5. Investigation: JEW MO'B JJR PNH.
  6. Methodology: JEW HB VL.
  7. Project administration: HB.
  8. Resources: HB JEW.
  9. Supervision: HB MO'B JEW ANR.
  10. Visualization: JEW.
  11. Writing – original draft: JEW.
  12. Writing – review & editing: JEW VL JJR PNH ANR HB.

References

  1. 1. Andrew R, Howe B, Mercer J. Long-time trends in ship traffic noise for four sites off the North American West Coast. Acoustical Scoiety of America. 2011;129(2):642–51.
  2. 2. Hildebrand JA. Anthropogenic and natural sources of ambient noise in the ocean. Mar Ecol Prog Ser. 2009;395:5–20.
  3. 3. Tyack PL, Miller EH. Vocal anatomy, acoustic communication and echolocation. In: Hoelzel AR, editor. Marine Mammal Biology: An Evolutionary Approach. Oxford, UK: Blackwell Science Ltd; 2002. p. 142–84.
  4. 4. Weilgart LS. The impacts of anthropogenic ocean noise on cetaceans and implications for management. Can J Zool. 2007;85(11):1091–116.
  5. 5. Anderwald P, Brandecker A, Coleman M, Collins C, Denniston H, Haberlin MD, et al. Displacement responses of a mysticete, an adontocete, and a phacid seal to construction-related vessel traffic. Endang Spec Res. 2013;21(3):231–40.
  6. 6. Di Iorio L, Clark CW. Exposure to seismic survey alters blue whale acoustic communication. Biol Lett. 2010;6(1):51–4. pmid:19776059
  7. 7. Finley KJ, Miller GW, Davis RA, Greene CR. Reactions of belugas, Delphinapterus leucas and narwhals, Monodon monoceros, to ice-breaking ships in the Canadian high Arctic. Can Bull Fish Aquat Sci. 1990;224:97–117.
  8. 8. Holt MM, Noren DP, Veirs V, Emmons CK, Veirs S. Speaking up: Killer whales (Orcinus orca) increase their call amplitude in response to vessel noise. J Acoust Soc Am. 2009;125(1):EL27–32. pmid:19173379
  9. 9. Pirotta E, Merchant ND, Thompson PM, Barton TR, Lusseau D. Quantifying the effect of boat disturbance on bottlenose dolphin foraging activity. Biol Conserv. 2015;181:82–9.
  10. 10. Bailey H, Brookes KL, Thompson PM. Assessing environmental impacts of offshore wind farms: lessons learned and recommendations for the future. Aquat Biosyst. 2014;10:8. pmid:25250175
  11. 11. Carstensen J, Henriksen OD, Teilmann J. Impacts of offshore wind farm construction on harbour porpoises: acoustic monitoring of echolocation activity using porpoise detectors (T-PODs). Mar Ecol Prog Ser. 2006;321:295–308.
  12. 12. Reid JB, Evans PGH, Northridge SP. Atlas of cetacean distribution and north-west European waters. Peterborough, UK: Joint Nature Conservation Committee; 2003.
  13. 13. Brandt MJ, Diederichs A, Betke K, Nehls G. Responses of harbour porpoises to pile driving at the Horns Rev II offshore wind farm in the Danish North Sea. Mar Ecol Prog Ser. 2011;421:205–16.
  14. 14. Scheidat M, Tougaard J, Brasseur S, Carstensen J, van Polanen Petel T, Teilmann J, et al. Harbour porpoises (Phocoena phocoena) and wind farms: a case study in the Dutch North Sea. Environ Res Lett. 2011;6(2):025102.
  15. 15. Teilmann J, Carstensen J. Negative long term effects on harbour porpoises from a large scale offshore wind farm in the Baltic—evidence of slow recovery. Environ Res Lett. 2012;7(4):045101.
  16. 16. Palka DL. Cetacean abundance estimates in US Northwestern Atlantic Ocean waters from summer 2011 line transect survey. Northeast Fisheries Science Center Reference Document 12–29. Woods Hole, MA: National Marine Fisheries Service; 2012.
  17. 17. Waring GT, Josephson E, Maze-Foley K, Rosel PE, editors. U.S. Atlantic and Gulf of Mexico Marine Mammal Stock Assessments– 2014. NOAA Tech Memo NMFS NE 231. Woods Hole, MA: National Marine Fisheries Service, National Oceanic and Atmospheric Administration; 2015.
  18. 18. Gaskin D. Harbour porpoise, Phocoena phocoena (L.), in the western approaches to the Bay of Fundy 1969–75. Rep Int Whal Comm. 1977;27:487–92.
  19. 19. Kraus SD, Prescott JH, Stone GS. Harbor porpoise, Phocoena phocoena, in the US coastal waters off the Gulf of Maine: A survey to determine seasonal distribution and abundance. Technical Report Submitted to Naitonal Marine Fisheries Service. Boston, MA: New England Aquarium; 1983.
  20. 20. Palka D. Influences on spatial patterns of Gulf of Maine harbor porpoises. In: Blix AS, Walloe L, Ultang O, editors. Whales, Seals, Fish and Man. Developments in Marine Biology. 4. Amsterdam: Elsevier Science; 1995. p. 69–75.
  21. 21. Palka DL. Abundance estimate of Gulf of Maine harbor porpoise. Rep Int Whal Comm. 1995;16:27–50.
  22. 22. Read AJ, Drinker P, Northridge S. Bycatch of marine mammals in U.S. and global fisheries. Conserv Biol. 2006;20(1):163–9. pmid:16909669
  23. 23. Garrison LP, Barry KP. Appendix A: Aerial abundance survey data during February-March 2013: Southeast Fisheries Science Center. Annual Report of a Comprehensive Assessment of Marine Mammal, Marine Turtle, and Seabird Abundance and Spatial Distribution in US Waters of the Western North Atlantic Ocean Woods Hole, MA: Northeast Fisheries Science Center and Southeast Fisheries Science Center, National Oceanic and Atmospheric Administration; 2013. p. 17–29.
  24. 24. Connelly EE, Duron M, Williams KA, Stenhouse JJ. Summary of high resolution digital video aerial survey data. In: Williams KA, Connelly EE, Johnson SM, Stenhouse JJ, editors. Wildlife densities and habitat use across temporal and spatial scales on the mid-Atlantic outer continental shelf: Final report to the Department of Energy EERE Wind & Water Power Technologies Office. Portland, Maine: Biodiversity Research Institute; 2015. p. 34.
  25. 25. Roberts JJ, Best BD, Mannocci L, Fujioka E, Halpin PN, Palka DL, et al. Habitat-based cetacean density models for the U.S. Atlantic and Gulf of Mexico. Sci Rep. 2016;6:22615. pmid:26936335
  26. 26. Cox TM, Read AJ, Barco S, Evans J, Gannon DR, Koopman HN, et al. Documenting the bycatch of harbor porpoises, Phocoena phocoena, in coastal gillnet fisheries from stranded carcasses. Fish Bull. 1998;96:727–34.
  27. 27. Hohn AA, Rotstein DS, Byrd BL. Unusual Mortality Events of Harbor Porpoise Strandings in North Carolina, 1997–2009. J Mar Biol. 2013;2013:1–13.
  28. 28. Tougaard J, Carstensen J, Teilmann J, Skov H, Rasmussen P. Pile driving zone of responsiveness extends beyond 20 km for harbor porpoises (Phocoena phocoena (L.)). J Acoust Soc Am. 2009;126(1):11–4. pmid:19603857
  29. 29. Bailey H, Clay G, Coates EA, Lusseau D, Senior B, Thompson PM. Using T-PODs to assess variations in the occurrence of coastal bottlenose dolphins and harbour porpoises. Aquat Conserv Mar Freshw Ecosyst. 2010;20(2):150–8.
  30. 30. Thompson PM, Lusseau D, Barton T, Simmons D, Rusin J, Bailey H. Assessing the responses of coastal cetaceans to the construction of offshore wind turbines. Mar Pollut Bull. 2010;60(8):1200–8. pmid:20413133
  31. 31. Pirotta E, Thompson PM, Miller PI, Brookes KL, Cheney B, Barton TR, et al. Scale-dependent foraging ecology of a marine top predator modelled using passive acoustic data. Funct Ecol. 2014;28(1):206–17.
  32. 32. Embling CB, Gillibrand PA, Gordon J, Shrimpton J, Stevick PT, Hammond PS. Using habitat models to identify suitable sites for marine protected areas for harbour porpoises (Phocoena phocoena). Biol Conserv. 2010;143(2):267–79.
  33. 33. Hammond PS, Berggren P, Benke H, Borchers DL, Collet A, Heide-Jorgensen MP, et al. Abundance of harbour porpoise and other cetaceans in the North sea and adjacent waters. J Appl Ecol. 2002;39:361–76.
  34. 34. Teilmann J. Influence of sea state on density estimates of harbour porpoises (Phocoena phocoena). J Cet Res Manage. 2003;5(1):85–92.
  35. 35. Brookes KL, Bailey H, Thompson PM. Predictions from harbor porpoise habitat association models are confirmed by long-term passive acoustic monitoring. J Acoust Soc Am. 2013;134(3):2523–33. pmid:23968050
  36. 36. Williamson LD, Brookes KL, Scott BE, Graham IM, Bradbury G, Hammond PS, et al. Echolocation detections and digital video surveys provide reliable estimates of the relative density of harbour porpoises. Methods Ecol Evol. 2016;7(7):762–9.
  37. 37. Guida V, Drohan A, Johnson D, Pessutti J, Fromm S, McHenry J. Report on Benthic Habitats in the Maryland Wind Energy Area. Report to the U.S. Department of the Interior, Bureau of Ocean Energy Management, Interagency Agreement M13PG00019/02. Sandy Hook, NJ: Northeast Fisheries Science Center; 2015.
  38. 38. Bureau of Ocean Energy Management (BOEM) and National Oceanic and Atmospheric Administration (NOAA). MarineCadastre.gov. 2013 Vessel Density. Retrieved March 22nd, 2017 from marinecadastre.gov/data.
  39. 39. Teilmann J, Miller LA, Kirketerp T, Kastelein RA, Madsen PT, Nielsen BK, et al. Characteristics of echolocation signals used by a harbour porpoise (Phocoena phocoena) in a target detection experiment. Aquat Mamm. 2002;28:275–84.
  40. 40. Villadsgaard A, Wahlberg M, Tougaard J. Echolocation signals of wild harbour porpoises, Phocoena phocoena. J Exp Biol. 2007;210:56–64. pmid:17170148
  41. 41. Dähne M, Gilles A, Lucke K, Peschko V, Adler S, Krugel K, et al. Effects of pile-driving on harbour porpoises (Phocoena phocoena) at the first offshore wind farm in Germany. Environ Res Lett. 2013;8:16.
  42. 42. Sarnocinska J, Tougaard J, Johnson M, Madsen PT, Wahlberg M. Comparing the performance of C-PODs and SoundTrap/PAMGUARD in detecting the acoustic activity of harbor porpoises (Phocoena phocoena). Proc of Meet Acoust. 2016;27.
  43. 43. Kyhn LA, Tougaard J, Thomas L, Duve LR, Stenback J, Amundin M, et al. From echolocation clicks to animal density-Acoustic sampling of harbor porpoises with static dataloggers. J Acoust Soc Am. 2012;131(1):550–60. pmid:22280616
  44. 44. Benjamin MA, Rigby RA, Stasinopoulos DM. Generalized autoregressive moving average models. J Am Stat Assoc. 2003;98(461):214–23.
  45. 45. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2015.
  46. 46. Stasinopoulos M, Rigby B, Eilers P. gamlss.util: GAMLSS Utilities. R package version 4.3–4. https://CRAN.R-project.org/package=gamlss.util2016.
  47. 47. Richardson WJ, Greene CR, Malme CI, Thomson DH. Marine Mammals and Noise, 1st ed. San Diego: Academic Press; 1995.
  48. 48. Koschinski S, Diederichs A, Amundin M. Click train patterns of free-ranging harbour porpoises acquired using T-PODs may be useful as indicators of their behaviour. J Cet Res Manage. 2008;10(2):147–55.
  49. 49. Nuuttila H, Meier R, Evans P, Turner J, Bennell J, Hiddink J. Identifying foraging behaviour of wild bottlenose dolphins (Tursiops truncatus) and harbour porpoises (Phocoena phocoena) with static dataloggers Aquat Mamm. 2013;39(2):147–61.
  50. 50. Carlström J. Diel variation in echolocation behavior of wild harbor porpoises. Mar Mamm Sci. 2005;21(1):1–12.
  51. 51. Todd VLG, Pearse WD, Tregenza NC, Lepper PA, Todd IB. Diel echolocation activity of harbour porpoises (Phocoena phocoena) around North Sea offshore gas installations. ICES J Mar Sci. 2009;66(4):734–45.
  52. 52. Verfuß UK, Miller LA, Pilz PK, Schnitzler HU. Echolocation by two foraging harbour porpoises (Phocoena phocoena). J Exp Biol. 2009;212(Pt 6):823–34. pmid:19251999
  53. 53. Linnenschmidt M, Teilmann J, Akamatsu T, Dietz R, Miller LA. Biosonar, dive, and foraging activity of satellite tracked harbor porpoises (Phocoena phocoena). Mar Mamm Sci. 2013;29(2):E77–E97.
  54. 54. Nuuttila H. Identifying foraging behaviour of wild bottlenose dolphins (Tursiops truncatus) and harbour porpoises (Phocoena phocoena) with static acoustic dataloggers. Aquat Mamm. 2013;39(2):147–61.
  55. 55. Schaffeld T, Bräger S, Gallus A, Dähne M, Krügel K, Herrmann A, et al. Diel and seasonal patterns in acoustic presence and foraging behaviour of free-ranging harbour porpoises. Mar Ecol Prog Ser. 2016;547:257–72.
  56. 56. Hastie T, Tibshirani R. Generalized additive models. 1st ed. London: Chapman and Hall; 1990.
  57. 57. Wood SN. Generalized additive models: An introduction with R. Boca Raton, FL: Chapman & Hall/CRC Press; 2006.
  58. 58. Fielding AH, Bell JF. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ Conserv. 1997;24(1):38–49.
  59. 59. Boyce MS, Vernier PR, Nielsen SE, Schmiegelow FKA. Evaluating resource selection functions. Ecol Model. 2002;157(2–3):281–300.
  60. 60. Freeman EA, Moisen G. PresenceAbsence: An R package for presence-absence model analysis. J Stat Softw. 2008;23(11):1–31.
  61. 61. Sing T, Sander O, Beerenwinkel N, Lengauer T. ROCR: visualizing classifier performance in R. Bioinformatics. 2005;21(20):3940–1. pmid:16096348
  62. 62. Koopman HN. Topographical distribution of the blubber of harbor porpoises (Phocoena phocoena). J Mammal. 1998;79:260–70.
  63. 63. Wisniewska D, Johnson M, Teilmann J, Rojano-Donate L, Shearer J, Sveegaard S, et al. Ultra-high foraging rates of harbor porpoises make them vulberable to anthropogenic disturbance. Curr Biol. 2016;26:1441–6. pmid:27238281
  64. 64. Johnston DW, Westgate A.J., Read A.J. Effects of fine-scale oceanographic features on the distribution and movements of harbour porpoises Phocoena phocoena in the Bay of Fundy. Mar Ecol Prog Ser. 2005;295:279–93.
  65. 65. Soldevilla MS, Wiggins SM, Hildebrand JA, Oleson EM, Ferguson MC. Risso’s and Pacific white-sided dolphin habitat modeling from passive acoustic monitoring. Mar Ecol Prog Ser. 2011;423:247–60.
  66. 66. Benoit-Bird KJ, Dahood AD, Würsig B. Using active acoustics to compare lunar effects on predator–prey behavior in two marine mammal species. Mar Ecol Prog Ser. 2009;395:119–35.
  67. 67. Dean C, Lawless J, Willmot G. A mixed poisson–inverse-gaussian regression model. Can J Stat. 1989;17(2):171–81.
  68. 68. Read AJ, Westgate AJ. Monitoring the movements of harbour porpoises (Phocoena phocoena) with satellite telemetry. Mar Biol. 1997;130(2):315–22.
  69. 69. Xu Y, Chant R, Gong D, Castelao R, Glenn S, Schofield O. Seasonal variability of chlorophyll a in the Mid-Atlantic Bight. Cont Shelf Res. 2011.
  70. 70. O'Reilly JE, Zetlin C. Seasonal, horizontal, and vertical distribution of phytoplankton chlorophyll a in the northeast U.S. continental shelf ecosystem. NOAA Tech. Rep. NMFS 139. Seattle, WA: U.S. Department of Commerce; 1998.
  71. 71. Thompson PM, Brookes KL, Graham IM, Barton TR, Needham K, Bradbury G, et al. Short-term disturbance by a commercial two-dimensional seismic survey does not lead to long-term displacement of harbour porpoises. Proc R Soc Lond [Biol]. 2013;280(1771):20132001.
  72. 72. Swingle WM, Barco SG, Bates EB, Lockhart GG, Phillips KM, Rodrique KR, et al. Virginia Sea Turtle and Marine Mammal Stranding Network 2015 Grant Report. VAQF Scientific Report 2016–01. Virginia Beach, VA: Virginia Aquarium Foundation Stranding Response Program; 2016.
  73. 73. Jones R. Time series resgression with unequally spaced data. J Appl Probab. 1986;23:89–98.
  74. 74. Bos R, de Waele S, Broersen P. Autoregressive spectral estimation by application of the Burg Algorithm to irregularly sampled data. IEEE Trans Instrum Meas. 2002;51(6):1289–94.
  75. 75. Cox TM, Read AJ, Solow A, Tregenza N. Will harbour porpoises (Phocoena phocoena) habituate to pingers? J Cet Res Manage. 2001;3(1):81–6.
  76. 76. Westgate AJ, Read AJ, Berggren P, Koopman HN, Gaskin DE. Diving behavior of harbor porpoises, Phocoena phocoena. Can J Fish Aquat Sci. 1995;52(5):1064–73.
  77. 77. Otani S, Naito Y, Kawamura A, Kawasaki M, Nishiwaki S, Kato A. Diving behavior and performance of harbor porpoises, Phocoena phocoena, in Funka Bay, Hokkaido, Japan. Mar Mamm Sci. 1998;14(2):209–20.
  78. 78. Recchia CA, Read AJ. Stomach contents of harbor porpoises, Phocoena phocoena (L), from the Bay of Fundy. Can J Zool. 1989;67(9):2140–6.
  79. 79. Gannon DP, Craddock JE, Read AJ. Autumn food habits of harbor porpoises, Phocoena phocoena, in the Gulf of Maine. Fish Bull. 1998;96(3):428–37.
  80. 80. Shotton R, Randall RG. Results of acoustic surveys of the southwest Nova Scotia (NAFO Division 4WX) herring stock during February and July 1981. Can Atl Fish Sci Adv Comm Res Doc 82/441982.
  81. 81. Scott WB, Scott MG. Atlantic Fishes of Canada. Toronto, Ontario: University of Toronoto Press; 1988. 730 p.
  82. 82. McQuinn IH. Pelagic fish outburst or suprabenthic habitat occupation: legacy of the Atlantic cod (Gadus morhua) collapse in eastern Canada. Can J Fish Aquat Sci. 2009;66(12):2256–62.
  83. 83. Santos MB, Pierce GJ. The diet of harbour porpoise (Phocoena phocoena) in the northeast Atlantic. Oceanogr Mar Biol. 2003;41:355–90.
  84. 84. Nelson GA, Ross MR. Biology and population changes of northern sand lance (Ammodytes dubius) from the Gulf of Maine to the Middle Atlantic Bight. J Northw Atl Fish Sci. 1991;11:11–27.
  85. 85. Boswell KM, Wilson MP, Cowan JH Jr. A semiautomated approach to estimating fish size, abundance, and behavior from dual-frequency identification sonar (DIDSON) data. N Am J Fish Manage. 2008;28(3):799–807.
  86. 86. Reid RN, Cargnelli LM, Griesbach SJ, Packer DB, Johnson DL, Zetlin CA, et al. Atlantic herring, Clupea harengus, life history and habitat characteristics. NOAA Technical Memorandum NMFS-NE-126. Woods Hole, MA: National Marine Fisheries Service; 1999.
  87. 87. Marubini F, Gimona A, Evans PGH, Wright PJ, Pierce GJ. Habitat preferences and interannual variability in occurrence of the harbour porpoise Phocoena phocoena off northwest Scotland. Mar Ecol Prog Ser. 2009;381:297–310.
  88. 88. Marques TA, Thomas L, Martin SW, Mellinger DK, Ward JA, Moretti DJ, et al. Estimating animal population density using passive acoustics. Biol Rev. 2013;88(2):287–309. pmid:23190144
  89. 89. Byrd BL, Hohn AA, Lovewell GN, Altman KM, Barco SG, Friedlaender A, et al. Strandings as indicators of marine mammal biodiversity and human interactions off the coast of North Carolina. Fish Bull. 2014;112(1):1–23.