A REGIONAL ANALYSIS OF LONG-TERM GRAY AND HARBOR …

1

A REGIONAL ANALYSIS OF LONG-TERM GRAY AND HARBOR SEAL STRANDING EVENTS

Katharine M. L. Jones1 and Michelle D. Staudinger1,2

1 University of Massachusetts Amherst, Amherst, MA 2 Department of the Interior, Northeast Climate Adaptation Science Center, Amherst, MA

INTRODUCTION

Strong indicators of species’ sensitivity, adaptive capacity, and overall vulnerability to

climate change are provided by changes in phenology, the timing of recurring life events

(Parmesan and Yohe, 2003). We possess poor information on climate induced shifts in

phenology of marine organisms, especially top predators. The Gulf of Maine (GOM) Seasonal

Migrants Project is an ongoing effort to determine the phenological changes occurring in the

GOM across marine mammals, sea turtles, and other marine species of conservation concern. As

part of that study, stranding data of injured or dead animals was explored for its utility to serve as

supplemental data to amend more traditional survey data where observations are scarce.

NOAA’s Greater Atlantic Region Marine Mammal Stranding Network Database was

examined for its utility as a potential long-term time series for the evaluation of phenological

patterns and shifts. Although records from stranding events represent sick or injured animals,

these data have been found to be reasonably comparable to survey data and provide useful

information on species’ distribution, abundance, and foraging ecology (Maldini et al., 2005;

Staudinger et al., 2014; Johnston et al., 2012, 2015). In this study, we focused on stranding data

of two marine mammal species, harbor seals (Phoca vitulina) and gray seals (Halichoerus

grypus). It was anticipated these data would also be good indicators of the areas and habitats that

seal populations use on a seasonal and annual basis in the region. Viable stranding data from

Maine to North Carolina included 1,571 gray seals and 4,399 harbor seals from 2001 to 2015.

This paper presents a summary of the spatial and temporal patterns of these data, and suggests

their suitability as supplemental data to other GOM marine species phenological studies, such as

the North Atlantic Right Whale Consortium database modelling efforts.

2

METHODS

Data standardization, preparation for analysis. The NOAA Greater Atlantic Regional Fisheries

Office (GARFO) provided the Greater Atlantic Region Marine Mammal Stranding Network gray

and harbor seal stranding dataset of 8,469 records with approximately 180 fields of Level A data

(these are basic information on stranding events that include morphology measurements, life

history, biology, general health, among other data), collected between 2001 through 2015

(Geraci et al., 2005). Primarily in the early years of data collection, 2001 through 2004, roughly

2,100 records had latitude and longitude formats recorded in variable formats, rather than the

decimal degree format required for conversion to point locations using GIS software (Table 1).

Latitude and longitude descriptive units were also inconsistent.

Through a multiple-step process of sorting, grouping, and calculating the inconsistently-

formatted latitudes and longitudes in Excel, most locational values could be converted to decimal

degrees. Only 45 records (< 1%) had inadequate information to convert latitudes and longitudes

to decimal degree format. Ultimately, a total of 8,424 records were usable for creation of a points

shapefile.

It is worth mentioning the value of remediating the lat/lon format of the records that were

not in decimal degree when initially logged, largely collected between 2001 and 2004.

Resolving their locational data problems extended the time scale of the study, benefitting the

meaning of the spatial and temporal analyses. (See Appendix I for conversion of decimal degree

latitude and longitude Excel data to ESRI point shapefile.)

Data exclusions. Given that our goal was spatial and temporal analyses of seal life events, it was

important to include for analysis only those records in which we had reasonable confidence in

the accuracy of the recorded life stage of animals (adult, subadult, yearling, pup) at first

observation, as well as the approximate date of their stranding. Therefore, 1,914 records were

excluded where the animal’s condition was recorded as ‘Mummified’ or ‘Advanced

Decomposition’ as these stranding states represent animals that have been dead for an extended

period of time, characteristics are often undistinguishable, and life stages are often unknown.

Another 767 stranding records were excluded where life stage was not recorded. Some records

were members of both of these categories. Exclusions reduced the dataset from 8,424 to 5,970

records for a total of N = 1,571 for gray seals, and N = 4,399 for harbor seals.

3

Quality assurance/quality control. Fifty random stranding locations were independently

evaluated for accuracy after conversion from one of the non- decimal degree formats using

several methods. All records contained the state in which the stranding occurred. Color-coding

stranding point symbols by state in ArcGIS verified whether the stranding lat/lon placed each

point within the boundaries of that state. Also, care was taken to validate stranding locations

against information in the ‘Locality_Detail’ field when compared against a base map that

displayed annotated geographic location information. At sea stranding events were examined for

evidence of fisheries association notations or other evidence of an offshore location; similarly,

inland strandings were evaluated for location within a river or lock (Figure 1). (See QA/QC

report for more details of evaluation.)

Analyses overview. Several independent approaches to spatial and temporal exploration of the

dataset were taken. These included: 1) charts, graphs, tables, and pivot tables, 2) kernel density

analysis by age class with seasonally and annually grouped data, and 3) mean annual location of

aggregated life stages by species. (See Appendix II.)

Kernel density analysis. Kernel density analyses were applied to gain an overview of stranding

density within different time-frames and across the geographic extent of the region. The kernel

density algorithm shows the density of point features around each output raster cell. An output

cell size of 500 meters on a side, and a search radius of 50,000 meters was applied. For each seal

species we performed kernel density analyses:

• annually across the 15-year dataset (all locations within year),

• seasonally (winter, spring, summer, fall) for adult, subadult, yearling, and pup life stages

(See Appendix II for established seasonality),

• monthly adult, subadult, yearling, and pup life stages.

Mean annual location analysis. For each species, all locations within each year were averaged to

calculate a single location. This was accomplished by averaging the values of all latitudes and

likewise for the longitudes.

4

RESULTS AND DISCUSSION

Monthly stranding patterns. Exploratory statistical and GIS analyses revealed distinct patterns

unique to species, age class, and time of year. Figure 2A displays the number of gray seal

strandings by month for the 15-year study period. Peak stranding events across the region for

gray seal pups and yearlings occurred in April, while harbor seal pups peaked in July, and

yearlings in November (Figure 2B). Peak stranding events of adults for both species occurred

approximately one month after the respective pup peaks, with gray seals in May and harbor seals

in August.

Annual stranding patterns. Notable annual stranding events for gray seal

yearlings occurred in 2011, for pups in 2007 and 2013 (Figure 3A). Gray seal

stranding events display a general increase from 2001 to 2015, possibly

paralleling similar recent trends in population growth (Hayes et al., 2016). Harbor seal pups

peaked in 2004, 2005, 2006, and 2011 (Figure 3B). Some of these events, which include die-

offs, have been confirmed elsewhere (Hayes et al., 2016).

Kernel density analysis. Kernel density analyses in Figures 4-6 give us a picture of adult and

pup stranding densities through the seasons with pooled annual data. (See Appendix II for

seasonal and annual stranding tables.) Stranding point locations are laid on top of the kernel

density displays.

Gray seals. Gray seal pups strand in winter from Maine to Maryland, with a concentration on

Nantucket (Hayes et al., 2016). By spring the stranding frequency quadruples from 135 to 509

with high densities on Long Island and the New Jersey coast. In summer and fall, stranding

occurrences drop precipitously from 32 to 11, with small concentrations on the Massachusetts

coast and Nantucket (Figure 4A).

Gray seal adult stranding occurrences were considerably fewer across the region in

comparison to pups, with a concentration on Cape Cod and Nantucket in general (Figure 4B).

By fall, stranding occurrences exhibit a geographical shift northward, up the Maine coast.

5

Harbor seals. Winter harbor seal pup stranding occurrences were low and generally loosely

dispersed across the region from Maine to Maryland with slightly higher numbers of events of

Cape Cod and southern Maine (Figure 5A). In spring the number of stranding events increased

dramatically from 58 to 504, with the densest occurrences along the entire Maine coast.

Stranding events triple by summer and show the highest densities along the mid-coast of Maine

to Cape Cod. In fall, stranding events decrease in frequency from 1,568 to 671 and were centered

along the coast of southern Maine to Massachusetts.

The range of harbor seal adult stranding events exhibited a similar seasonal distribution

in winter and spring from Maine to Maryland, and occurrences only extend as far south as New

Jersey in summer and fall (Figure 5B). The frequency of stranding occurrences was relatively

stable during spring, summer, and fall; however, concentrations shifted northwards from the

Massachusetts coast during winter and spring, to the southern Maine during summer and fall.

Mean annual centroids. Mean annual centroids of stranding events were calculated by averaging

all the latitude values within each year, and all longitudes likewise, to calculate one annual

location for each seal species (Figure 6). Gray seal stranding events for all age classes combined

were clustered from the east end of Long Island to Buzzard’s Bay, whereas harbor seals showed

a more northern pattern stretching from Boston to southernmost Maine.

Kernel density plots and line graphs demonstrate very clear species, temporal, and spatial

patterns that conform to known life history patterns for both seal species. Harbor seal pup

stranding events peak in July, centered on the New Hampshire and southern Maine coasts (See

Appendix II). A very high percentage of all harbor seal stranding events are pups in summer.

Gray seal pup stranding events peak earlier than harbor seals, in April, and are densest on Long

Island and New Jersey coasts, spanning from Maine to Delaware. Seasonally, most gray pup and

yearling strandings were in spring, while gray adult and subadult stranding events were in

summer.

CONCLUSIONS AND NEXT STEPS

Preliminary spatial and temporal evaluation of seal stranding data has demonstrated that

the data are suitable to evaluate phenological patterns and shifts in these two pinniped species.

The approach outlined in this report also shows these stranding data have strong potential to

6

answer questions beyond their primary focus (e.g., of serving as records of injury and illness),

and as has been shown in previous studies, to provide data that may complement and supplement

more traditional survey data in support of modelling efforts (Maldini et al., 2005; Johnston et al.,

2012, 2015). The data strongly display species, life stage, and season-specific spatial and

temporal signals. Because observed patterns in time and location match known life history

patterns, the data set is appropriate for evaluating seal phenology and has additional potential for

other regional applications in the GOM across marine mammals, sea turtles, and other marine

species of conservation concern.

This project only scratched the surface of how this type of dataset may be utilized to

investigate long-term trends in occurrence relative to climate drivers and other factors

influencing population dynamics (e.g., Johnston et al., 2012, 2015). Next investigatory steps

include more in-depth examination of changing spatial patterns over time by age class and as a

function of stranding causation (e.g., disease, human interactions), as well as in relationship to

environmental stressors such as temperature and coastal storm events. Opportunities to integrate

additional data series (e.g., bycatch data) and apply this approach to other species can also

provide opportunities for additional analyses.

Development of an automated protocol for the correction of inconsistent latitude and

longitude formats (i.e. non-decimal degree formats) is possible and future studies could use the

correction process developed here as a validation step to determine which approach yields the

largest return of corrected values for the least investment of time and other resources. The results

of such a protocol could facilitate rapid conversion of other stranding datasets for additional

species of interest. Evaluating monthly and annual stranding trends in relationship to

environmental variables, would allow the exploration of the drivers of anomalous years and

events, and help quantify monthly and spatial shifts over years (Johnston et al., 2012, 2015). Co-

variates to be considered in future analyses include the timing of seasonal transitions, regional

sea and air temperature data, frequency and degree of extreme heat days and storm events.

The presence in late fall of ‘pup’ stranding events is a potential anomaly given that both

species in the GOM generally pup by late spring (May-June). An evaluation of body size (i.e.,

length), age and reproductive state for these individuals would provide additional information

and potential insights into the accuracy of life stage designations of individual animals.

7

ACKNOWLEDGEMENTS

We would like to express our gratitude to NOAA Fisheries Service, Greater Atlantic Regional

Fisheries Office and the Greater Atlantic Region Marine Mammal Stranding Network for

providing the stranding dataset for this study, especially Mendy Garron for answering a plethora

of questions. Thank you to Daniel Pendleton, Forrest Bowlick and the Staudinger University of

Massachusetts Phenology Lab Research Group for feedback during this effort. Finally, we thank

Brigid Ryan, University of Massachusetts, for her quality assurance and quality control efforts in

evaluating repaired latitude and longitude data. Data on marine mammal strandings were

collected by organizations authorized under Marine Mammal Protection Act Stranding

Agreements issued by NOAA Fisheries. Data are reported on NOAA Form 89-864 (Level A

data form) and are as complete and accurate as possible at time of data entry. This research was

funded through a student contract (#G17PX00971) by the Department of the Interior Northeast

Climate Science Center as part of the project How and why is the timing and occurrence of

seasonal migrants in the Gulf of Maine changing due to climate?.

REFERENCING THIS REPORT

Jones, K. L., and M. D. Staudinger. 2018. A regional analysis of long-term gray and harbor seal stranding events. DOI Northeast Climate Adaptation Science Center Report. Amherst, MA.

LITERATURE CITED

Geraci JR, Lounsbury VJ. 2005. Marine Mammals Ashore: A Field Guide for Strandings. Texas A&M University Sea Grant College Program. 344 p. http://www.dtic.mil/dtic/tr/fulltext/u2/a456126.pdf

Hayes SA, Josephson E, Maze-Foley K, Rosel, PE, editors. 2016. US Atlantic and Gulf of Mexico Marine Mammal Stock Assessments -- 2016. NOAA Tech Memo NMFS NE 241; 274 p. Available from: National Marine Fisheries Service, 166 Water Street, Woods Hole, MA 02543-1026, or online at http://www.nefsc.noaa.gov/publications/ or doi:10.7289/V5/TM-NEFSC-241

2016. Gray Seal (Halichoerus grypus grypus): Western North Atlantic Stock. February 2016, p 147.

2014. Harbor Seal (Phoca vitulina): Western North Atlantic Stock. May 2015, p 165.

8

Johnston DW, Bowers MT, Friedlaender AS, Lavigne DM. 2012. The Effects of Climate Change on Harp Seals (Pagophilus groenlandicus). PLoS ONE 7(1): e29158. https://doi.org/10.1371/journal.pone.0029158

Johnston DW, Frungillo J, Smith A, Moore K, Sharp B, Schuh J, et al. 2015. Trends in Stranding and By-Catch Rates of Gray and Harbor Seals along the Northeastern Coast of the United States: Evidence of Divergence in the Abundance of Two Sympatric Phocid Species? PLoS ONE 10(7): e0131660. https://doi.org/10.1371/journal.pone.0131660

Maldini D, Mazzuca L, Atkinson S. 2005. Odontocete Stranding Patterns in the Main Hawaiian Islands (1937–2002): How Do They Compare with Live Animal Surveys? Pacific Science 59(1):55-67. https://doi.org/10.1353/psc.2005.0009

Parmesan C, Yohe G. 2003. A Globally Coherent Fingerprint of Climate Change Impacts Across Natural Systems. Nature: 421:37. http://dx.doi.org/10.1038/nature01286

Staudinger MD, McAlarney R, Pabst A, McLellan W. 2014. Foraging Ecology and Niche Overlap in Pygmy (Kogia breviceps) and Dwarf (Kogia sima) Sperm Whales from Waters of the U.S. Mid-Atlantic Coast. Marine Mammal Science: 30(2): 626-655. https://doi.org/10.1111/mms.12064

9

TABLES

Table 1: Examples of problematic location formats from stranding records recorded in Level A datasheets between 2001-2015.

10

Figure 1: Examples of how location information was validated by mapping points using geographical (e.g., state) and ‘Locality_Detail’ fields to resolve at sea and inland points.

11

Figure 2: A) Gray seal (N = 1,571) and B) harbor seal (N = 4,399) stranding events by month and life stage from Virginia to Maine between 2001-2015.

0

50

100

150

200

250

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

# of

stra

ndin

gs

Month

A)

0

200

400

600

800

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

# of

stra

ndin

gs

Month

B)

Adult Subadult Yearling Pup

12

Figure 3: A) Annual gray seal (N = 1,571) and B) harbor seal (N = 4,399) stranding events by life stage from Virginia to Maine between 2001-2015.

0

20

40

60

80

10020

01

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

# of

stra

ndin

gs

Year

A)

0

100

200

300

400

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

# of

stra

ndin

gs

Year

B)

Adult Subadult Yearling Pup

13

Figure 4: Seasonal kernel density analysis of gray seal A) pups and B) adults for all years (pooled) from 2001 - 2015. Occurrences of stranding events are divided into density groupings (0-1%, 1-5%, 5-50%, 50-100%) with the darkest red color indicating the most dense occurrences and the lightest shade of red indicating the fewest.

A

B

14

Figure 5: Seasonal kernel density analysis of harbor seal A) pups and B) adults all years (pooled) from 2001 - 2015. Occurrences of stranding events are divided into density groupings (0-1%, 1-5%, 5-50%, 50-100%) with the darkest red color indicating the most dense occurrences and the lightest shade of red indicating the fewest.

A

B

15

Figure 6: Gray and harbor seal mean annual centroids of stranding events.

16

APPENDIX I Conversion of decimal degree latitude and longitude Excel data to ESRI point shapefile. 1. ArcGIS.

A. ADD (+) table to Table of Contents (use the Excel Level A lat/lon-corrected file in .xls format). This action will create an Event. Output = 8424_pts_uniqid$.

B. Select the new Event, right click, Display XY Data. Select fields ‘latitude’ and ‘longitude’ and the unique record identifier, National_Database_Number. A points shapefile containing these three fields is created. Output = data_8424_geog.shp.

C. Right click new shapefile, Data, Export to preserve lat/lon . D. Use the PROJECT tool to project the data from Geographic coordinate system to UTM

19 projection. Output = data_8424_u19.shp.

2. Excel. Create a ‘stripped down’ Excel file containing the meaningful fields for future spatial and temporal analyses.

3. ArcGIS. A. Perform JOIN operation on the new shapefile that has the lat/lon locations

(seal_pts_8424). This action will connect the Level A data fields to the points shapefile permanently throughout all subsequent operations. Use the unique identifier, National_Database_Number. Output = data_8424_u19_join.shp.

4. From this point forward, all processing was done in ArcGIS.

17

APPENDIX II Monthly, seasonal, and annual occurrence by life stage Monthly

Seasonal

18

Annual