Assessing Risks from Harbor Dredging to the Northernmost Population of Diamondback Terrapins Using Acoustic Telemetry T. Castro-Santos1,2 & M. Bolus2 & A. J. Danylchuk2 Received: 2 August 2017 /Revised: 28 September 2018 /Accepted: 1 November 2018 # This is a U.S. government work and its text is not subject to copyright protection in the United States; however, its text may be subject to foreign copyright protection 2018 Abstract The northern diamondback terrapin (Malaclemys terrapin terrapin) is a saltmarsh-dependent turtle that occupies coastal habitats throughout much of the Atlantic coast of North America. We used a novel application of acoustic telemetry to quantify both mobility and occupancy of terrapins within a dredged harbor and surrounding habitats, and used these metrics to quantify relative risk to individuals posed by harbor dredging. Terrapins showed strong fidelity to brumating areas within subdrainages, but extensive movements between these zones during the active period. Activity was greatest in late spring and early summer, declining to near zero by December. Occupancy of the dredge zone was also greatest during spring and summer and declined throughout the autumn months to an annual minimum during winter. Taken together, these data indicate that risks from harbor dredging are minimized during the autumn and early winter months. Keywords Terrapin . Telemetry . Dredging . Brumation . Hibernation .Movement . Modeling . Risk . Assessment Introduction Coastal and estuarine environments accommodate a range of human activities, including fisheries, aquaculture, recreation, and transportation. These activities are supported by infrastruc- ture, the maintenance of which can affect sensitive species that inhabit these zones (Culloch et al. 2016;MoserandRoss 1995). One such activity is harbor dredging. Harbors are typically located in sheltered waterways, often with additional armoring to protect anchorages and ports. This sheltered quality creates an opportunity for sediments to accumulate, and harbors must be dredged periodically to maintain navigability. Dredging requires heavy equipment and the removal of large volumes of sediment. It is highly disruptive to the environment and esthetically unpleasant, interfering with economically and cul- turally important activities such as boating, aquaculture, and tourism. To minimize these effects, dredging is often undertak- en during winter months, when human activities are reduced. Dredging also poses several potential threats to wildlife, including suspension of sediments (which may contain toxic chemicals) and direct mechanical contact with the dredge equipment (O’Donnell et al. 2007; Wilkens et al. 2015). One species that is vulnerable to harbor dredging is the northern diamondback terrapin (Malaclemys terrapin terrapin, hereaf- ter referred to as Bterrapin^), a saltmarsh-adapted species that occupies nearshore habitats along the East Coast of North America (Brennessel 2006). Historic overharvest, combined with fishery bycatch and habitat destruction, has caused many populations to be extirpated, and the species is listed for pro- tection in many states throughout its range (Butler et al. 2006; Hart and Lee 2006). During winter months, terrapins enter a period of dormancy known as brumation, during which they aggregate and remain essentially immobile, buried in the ben- thos (Brennessel 2006; Haramis et al. 2011; Yearicks et al. 1981). Given that coastal dredging also often occurs during this period, there is the potential for considerable mortality of terrapins that brumate within the dredge zone, prompting management agencies to restrict dredging to summer months, Communicated by Nathan Waltham Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12237-018-0481-9) contains supplementary material, which is available to authorized users. * T. Castro-Santos tcastrosantos@usgs.gov 1 USGS, Leetown Science Center–S.O. Conte Anadromous Fish Research Center, One Migratory Way, Turners Falls, MA 01376, USA 2 Department of Environmental Conservation, University of Massachusetts Amherst, Amherst, MA 01003, USA Estuaries and Coasts https://doi.org/10.1007/s12237-018-0481-9 when the high level of activity presumably confers ability to volitionally avoid dredging equipment. The most northerly terrapin population resides in Wellfleet Harbor, Massachusetts (USA), where the species is listed as threatened under the Massachusetts Endangered Species Act (MESA, MGL c131A). The harbor is a federal waterway and municipal anchorage, both subject to periodic dredging, and terrapins are regularly observed throughout the dredge zone. Large breeding aggregations occur each spring in a cove ad- jacent to the harbor, indicating that this is vital habitat for this population, and raising concerns that brumation may be oc- curring within the dredge zone. While any mortality holds the potential to affect the popu- lation, the magnitude of the threat posed by dredging depends in part on which portion of the population is at risk. If dredg- ing is performed during brumation, and if the dredge zone comprises preferred brumation habitat, then dredging will likely pose serious risks, both to individuals and to the local population. The objective of this study was to use acoustic telemetry to assess seasonality of risk from dredging by quan- tifying rates of movement and occupancy of terrapins within the dredge zone of Wellfleet Harbor. Our observations can inform decisions on timing of dredging activity to minimize risk to this threatened coastal species. Methods Study Area Wellfleet Harbor is a protected harbor in Cape Cod Bay locat- ed in the Town ofWellfleet, MA, USA (41° 55′ 48″N, 70° 01′ 30″ W; Fig. 1). It is a Spartina spp. grass-dominated marsh system, comprising several subdrainages with a mix of inlets, rivers, and creeks. It has an extensive intertidal zone, with regular tidal fluctuations of 3–4 m. The area of primary con- cern was the harbor proper, which consists of the federal and town anchorages (hereafter called the anchorage) and the main navigation channel, all of which are dredged (Fig. 1). In addi- tion to the anchorage, the study area also included each of the primary subdrainages within the Wellfleet Harbor watershed. Capture and Handling Mature terrapins were collected during three distinct time pe- riods to ensure that we characterized movement behaviors of a representative sample of the population. We focused on mature individuals, both to ensure that they were of sufficient size to bear the mass of the tag, and also because impacts on mature individuals are more likely to affect the population (Heppell et al. 2000). The first and third collections were performed during the mating aggregation in Chipman’s Cove (Duck Creek, adjacent to the dredge zone) shortly after terrapins emerged from brumation (April–May, 2011 and 2012). These individuals were intended to provide information on whether and to what extent terrapins brumated nearby, and to maximize the likelihood of identifying individuals at risk if the dredge zone proved to include important brumation habitat. The second collection was performed during July, 2011, and was intended to be representative of the entire population. To that end, terrapins were captured in the tributaries (South of Lt. Island (hereafter termed BWBWS^ or Bsanctuary^), Blackfish Creek, Herring River, and Duck Creek). The July collection occurred after their reproductive period, when ter- rapins were dispersed throughout the watershed. The purpose of this group was to help us to identify other brumation sites, in the event that dispersal rates were low. Transmitters and Tag Lots Terrapins were tagged with acoustic transmitters (V9 coded tags, 69 kHz; VEMCO Division, AMIRIX Systems Inc., Halifax, Nova Scotia, Canada). Tags were purchased in two separate groups (Btag lots^). Each group had a distinct pro- gramming scheme, intended to ensure maximum longevity, and in particular to allow for documentation of onset of brumation in each of 2 years for each tag. To achieve this, they were programmed to transmit every 50–180 s (randomly dis- tributed), and to have a dormant period during the first winter following tagging. These were originally intended to be applied during 2010 and 2011. However, owing to permitting restric- tions, all tags were deployed during 2011 and 2012. This meant that the first tag lot became dormant and re-activated later than the initial study design intended, and only one brumation onset event was observed for this tag lot (Tables 2 and 3). Collection Methods and Transmitter Attachment Terrapins were collected manually using dip nets. Transmitters were attached to the first and second marginal scutes to the left of the supracaudal scute (Fig. S1). Two small holes were drilled through the scutes and the supporting der- mal plate, through which were inserted small cable ties. Each transmitter was secured with the cable ties and embedded in epoxy (ACE Quick Set epoxy, Henkel AG & Co. KGaA, 40191 Düsseldorf, Germany). The epoxy was thickened with colloidal silica (West Systems 406 Colloidal Silica Adhesive Filler, West Systems, Bay City, MI, USA) and mixed with black pigment (Evercoat Coloring Agent, Evercoat, Cincinnati, OH, USA). This produced a smooth finished prod- uct with no sharp edges to snag on vegetation and fishing gear, and cryptic coloration which we hoped would reduce risk of predation. To avoid biofouling on the transmitters and poten- tial adverse effects on the animals, we coated the entire trans- mitter package with anti-fouling paint (Interlux Micron CSC, International Paint LLC, Union, NJ, USA). The tagging Estuaries and Coasts procedure was approved by the University of Massachusetts Amherst IACUC (protocol 2011-0009). All captured terrapins were sexed (Brennessel 2006), measured, weighed, and re- leased at their point of capture. Fixed-Receiver Array We used an array of fixed-station receivers to monitor broad- scale movements and distribution of terrapins. We deployed 20 submersible receivers (VR2W, VEMCO Division, AMIRIX Systems Inc., Halifax, Nova Scotia Canada) throughout Wellfleet Harbor during 2011, and 21 during 2012 and 2013 (Fig. 1; Tables 1 and 2). Receivers were placed to optimize our ability to test key hypotheses regarding movement of terrapins inWellfleet Harbor, and placement was consistent across years. Themost important metrics concerned rates of occupancywith- in the anchorage throughout the year, seasonal changes in mo- bility (including onset of brumation), and fidelity to brumation sites among years. The array was deployed in March and re- trieved in January, which ensured complete coverage during the active phase of the transmitters (Table 2). Each receiver was attached to a 40-kg concrete mooring using a buoy system that optimized its ability to detect passing terrapins at any phase of the tidal cycle (Fig. S2). Receivers were deployed in a series of Bgates^ (channel constrictions where a swimming terrapin would have to pass through the detection zone in order to pass the receiver) and Bnodes^ (distributed sites within an area of interest) so that movement among detection zones aswell as residency at key locations (e.g., brumation sites) was documented. Maximum receiver detection ranges were measured by deploying a range testing transmitter (V9, coded tag range testing transmitter emitting at a 69 kHz signal, battery life = 14 days, VEMCO Division, AMIRIX Systems Inc., Halifax, Nova Scotia, Canada) concentrically in all four cardinal directions in 50-m increments (max = 600m) from each receiver location (Selby et al. 2016). The maximum range at which a tag was detected was assumed to represent the maximum detection range for that receiver. Receivers were downloaded biweekly, and on each download, clocks were synchronized with official US Naval time, and any clock drift was corrected assuming a linear rate of drift between downloads. Temperature loggers (HOBO Pendant Temperature/Light Data Logger 64 bit—UA-002-64, Onset, Pocasset, MA, USA) were also deployed on four receivers (one each at the mouths of Blackfish Creek (receiver 14), Herring River (re- ceiver 3), and Duck Creek (receiver 9) as well as in the Main Channel (receiver 11; Fig. 1). Loggers were configured to record water temperature every 15 min. Manual Tracking We performed manual tracking from a variety of vessel types to verify fixed-station observations and to identify exact loca- tions of brumation sites of those terrapins we were able to detect. Terrapins that remained stationary during multiple scanning events were assumed to be brumating. Surveys Fig. 1 Receiver deployment locations in Wellfleet Harbor, Wellfleet, MA (white area indicates water). The map shows the proposed dredging project (stippled and striped area near the breakwater), subdrainages (labeled according to either the legend or body of Table 1), and receiver locations (numerals). Receiver numbers are sized to approximate the typical detection range at each site Estuaries and Coasts began in October and were performed weekly as weather and tides permitted. We used two receiver types for this: a Vemco VR-100 (VEMCO Division, AMIRIX Systems Inc., Halifax, Nova Scotia, Canada) and a Sonotronics USR-08 (Sonotronics, Tucson, AZ, USA). Both receiver types were outfitted with directional and omni-directional hydrophones and were able to decode all tags. We established density of scanning transects based on de- tection ranges of known brumating terrapins. Wind and tide made it impossible to follow planned transects exactly, but this detection range was used to establish a grid that was covered throughout the dredge zone during each manual tracking ses- sion. Thus, the individual transects scanned varied by session, but were always designed to ensure a strong likelihood of detecting tags that were present anywhere within the surveyed zone. Similar methods were used outside the dredge zone, but because of the importance of identifying any brumating terra- pins within that zone, we concentrated our efforts there. Data Management and Analysis All data for this study were compiled in a relational Microsoft Access database and analyzed using R statistical software (R 3.2.3; R Core Team 2014). Occupancy and Movement Vulnerability of terrapins was assessed by measuring both occupancy of the dredge zone and mobility. Occupancy ad- dresses when animals were present within the dredge zone, Table 1 Locations of the 21 acoustic telemetry receivers (latitude and longitude in decimal degrees). Map ID refers to numbers shown on Fig. 1; station names are referred to throughout the text. Subdrainages refer to the four primary drainage areas in the system (Herring River (HR), Duck Creek (DC), Blackfish Creek (BFC), Wellfleet Bay Wildlife Sanctuary (WBWS)) or various locations outside of these drainages within Wellfleet Harbor (WH). Receivers marked with an asterisk (*) were affixed with a temperature logger Map ID Station name Subdrainage Intertidal Latitude Longitude 1 Herring River 3 HR Yes 41.931394 − 70.063555 2 Herring River 2 HR Yes 41.930683 − 70.066383 3* Herring River 1 HR Yes 41.92375 − 70.056983 4 Great Island WH No 41.912694 − 70.060944 5 Jeremy Point WH No 41.89137 − 70.06682 6 Mooring Basin DC No 41.9284 − 70.033016 7 Anchorage DC No 41.926194 − 70.028194 8 The Cove DC Yes 41.922138 − 70.027277 9* Duck Creek DC Yes 41.930133 − 70.024166 10 Railroad Bridge DC Yes 41.933861 − 70.027194 11 Channel 1 WH No 41.922166 − 70.036388 12 Indian Neck 1 WH No 41.91295 − 70.030616 13 Indian Neck 2 WH No 41.912950 − 70.037236 14* Blackfish 1 BFC Yes 41.9029 − 70.012916 15 Fox Island BFC Yes 41.910277 − 70.014416 16 Pleasant Point BFC Yes 41.910116 − 69.995183 17 Loagy Bay BFC Yes 41.900111 − 70.005527 18 Lt. Island WH No 41.894333 − 70.02780 19 Sanctuary 2 WBWS Yes 41.890055 − 70.00850 20 Sanctuary WBWS Yes 41.8876 − 69.99936 21 Eastham WH No 41.877983 − 70.020041 Table 2 Period of coverage for fixed receivers (Rx coverage) and tags. Date on and Date off indicate expected on-off cycles, based on manufac- turer’s programming. First detection and Last detection indicate actual observations, and are presented as median (range). Note that some tags were detected before their programmed activation date—evidently arising from a manufacturing defect in programmed on/off times Tag lot Date on Date off First detection Last detection Rx coverage 1 02 Jun 2011 3 Jan 2012 25 May 2011 (24 May–15 Jun) 18 Oct 2011 (13 Jul–25 Dec) 18 Mar 2011–10 Jan 2012 2 04 Jul 2011 18 Dec 2012 19 Jul 2011 (5 July–14 Aug) 05 Oct 2011 (12 Sep–15 Nov) 1 01 Jun 2012 06 Aug 2012 (expired) 23 May 2012 (16 Apr–24 May) 05 Aug 2012 (15 Jun–27 Aug) 20 Mar 2012–07 Jan 2013 2 18 Apr 2012 18 Dec 2012 25 Apr 2012 (18 Apr–13 May) 09 Oct 2012 (01 Jun–18 Dec) 2 19 Apr 2013 13 Jul 2013 (expired) 29 Apr 2013 (9 Apr–25 Jul) 01 Jul 2013 (28 Apr–02 Aug) 19 Mar 2013–02 Aug 2013 Estuaries and Coasts and so vulnerable to equipment; mobility addresses the ability of terrapins to actively avoid the dredge, given that they be present during operations. Occupancy of the anchorage and dredge zone was mea- sured as the proportion of available active tags detected within the Duck Creek subarray. This included all receivers deployed from the mouth of Duck Creek (receiver 10) out to the break- water, including the entire dredge zone (Fig. 1) on each day of the study. Movement was considered an indicator of each animal’s Bmobility^ (and by association its ability to avoid dredge equipment), and was measured by identifying transitions in detections between receivers within the Duck Creek system. Transitions were determined based on occupancy events with- in the detection zone of each receiver. We applied time-to- event techniques, whereby the log density function of interval durations between detections at each receiver were used to identify times when terrapins entered and left each detection zone (Castro-Santos and Perry 2012). The density of receivers within the Duck Creek subarray meant that it was possible to discern movements throughout the subdrainage. Data from all receivers were combined, and any detections occurring within 20 s of each other on two or more receivers were identified, retaining only the first of these detections and assigning the location to that receiver. These simultaneous detections were very rare, comprising < 0.1% of the total detections. More commonly, there was a gap between detections as terrapins moved around the har- bor, passing among the detection zones of the array; > 95% of all new detections were separated by > 1 transmission interval. Taken together, the low incidence of simultaneous detections combined with successive receivers typically be- ing separated by more than one transmission interval sug- gests that any error associated with animals occupying over- lapping zones must be trivial, and that successive detections at distinct receivers represented actual movements of ani- mals among the receivers in the array. Distance between receivers was known, and at each transition, an individual animal was assigned this distance minus 300 m (the average radius of one receiver’s detection range). This provided a conservative estimate of movement distance, balancing con- siderations of detection range, detection probability, trans- mission rate, and rate of movement of the turtles, and as- suming a straight-line path between receivers. In reality, movement paths are rarely truly straight, and so, actual movement was probably greater that what we estimate using this method. The transition distances were then summed for each detected terrapin on each day, and median daily values were then calculated based on all detected animals. Note that by using only receivers within the Duck Creek drainage, this also provides a conservative measure of mobility. Since most movements occurred within drainages, however, and because we used the median value, influence of extreme values was minimized, providing a conservative but realistic index of movement. Loess smooth functions were fitted to both the occupancy and movement data, providing a continuous estimate of aver- age occupancy and movement by terrapins throughout the year. Means and 95% confidence intervals were calculated for both metrics. Combined Risk The exposure risk was assumed to be directly proportional to the probability of being present within the dredge zone on a given date. We therefore used the untransformed occupancy metric described above as a direct index of risk (Ro): Ro ¼ P Occupancy ð Þ ð1Þ where P (Occupancy) ∈[0, 1] is the proportion of the popula- tion present in the dredge zone on a given day of the year as estimated by the loess smooth. Next, we produced an index of relative mobility risk (Rm): Rm ¼ 1− Md Mmax ð2Þ whereMd is the median observedmovement on a given day and Mmax is the annual maximum of these values. Values for Rm range from 0 to 1, with 0 being associated with the day of greatest mobility and 1 associated with the day of least mobil- ity within a given year. Thus, brumating terrapins experience the greatest mobility-related risk, and this risk is minimized on the day of greatest activity. Values used for calculating both Ro and Rm were taken from the among-years loess smooth. Combined risk exposure (RT)was then calculated as the product of occupancy and mobility risks: RT ¼ Ro  Rm ð3Þ Results Capture and Handling Seventy-five terrapins (56 females and 19 males) were tagged during the 2011 field season (Table 3). Of these, 30 females and 19 males comprised the spring collection from the pro- posed dredging area (Table 2; Fig. 1). Twenty-six terrapins were also tagged, distributed throughoutWellfleet Harbor dur- ing July (Table 3). During this second period, no males were captured that were large enough to tag. Instead, all transmitters were attached to mature females. An additional 25 terrapins (13 females and 12 males) were tagged during the 2012 mating aggregation (Table 3). Thus, 74 of 100 terrapins were captured within the anchorage area. Estuaries and Coasts By concentrating our collections near the dredge zone, we expected the observed proportion of brumating terrapins with- in this zone to overestimate their representation in the larger population. In this way, we ensured that assessments of occu- pancy risk were conservative, i.e., biased in favor of detecting occupancy when it occurred. Females were larger and more variable in size than males. Mean and standard deviations of straight carapace length and in-air mass of females were 18.4 ± 1.3 cm and 1071 ± 224 g; of males, they were 12.3 ± 0.4 cm and 280 ± 27 g. There was no difference in size or mass between years within sexes (p- > t > 0.3 in all cases). All terrapins were held overnight and released within 18–30 h of capture. They all swam away once released; no abnormal or disoriented behavior was observed. The total transmitter package weighed 9 g in air, or 0.84% ± 18% of the average body mass of females and 3.21% ± 0.31% of the average body mass of males; on no individual did it exceed 3.7% of the body mass. Most tags activated and deactivated within a day of their programmed dates. Some drift in these settings did occur, however with at least one tag activating at least 10 days before its programmed date (Table 2; Fig. S3). Receiver Array Receivers performed well throughout the deployment period (Table 2). Tested detection ranges ranged from 117 to 602 m (mean = 421 m), and varied with tide and bathymetry. Range testing was typically performed within 2 h of high tide, and so, these represent maximum values. At low tide and shallow conditions, these ranges were reduced, with periods (< 2 h) of zero efficiency for receivers in the intertidal zone (Table 1). Thus, the 300-m detection ranges used for calculat- ing relative movement over the course of the study were ap- propriate, given the available information. Mobility Terrapins were active during the entire period of telemetry coverage (April–December: Fig. 2). Total movement had little relationship to water temperature (Figs. 2 and S4; Kendall’s tau = 0.056). Instead, activity was greatest in mid-May, with median movement values of about 2 km/day. This corresponded with the breeding period of this population. Table 3 Numbers and location of terrapin captures. Capture locations are abbreviated as CC (Chipman’s Cove), BFC (Blackfish Creek), DC (Duck Creek), WBWS (Wellfleet Bay Wildlife Sanctuary—actually in- cludes marshes and beaches south of Lt. Island), and HR (Herring River). Anchorage refers to the federal and town dredge zones, adjacent to Chipman’s Cove Date Tag lot Capture location Females Males Total May 2011 1 CC, anchorage 30 19 49 July 2011 2 WBWS 8 0 8 2 HR 8 0 8 2 BFC 8 0 8 2 CC, anchorage 2 0 2 May 2012 2 CC, anchorage 13 12 25 Total 69 31 100 0 2000 4000 6000 8000 Apr Jul Oct Jan Date Median Distance (m)Year 2011 2012 2013 Fig. 2 Median known distance traveled per day by detected terrapins. Points represent median for a given day on a given year. The loess smooth averages across years Estuaries and Coasts There was a slight decrease in movement in mid-summer, which corresponded with the female nesting season, followed by another broad peak from August to October, where the median movement was about 1.8 km/day. Movement then began to decrease until reaching a minimum in December, shortly before the tags shut down for the winter. By December 1, it became common for entire days to pass with- out any observed movement between receivers (Fig. 2). Occupancy and Brumation Because the tags became active in late April or later (Table 2), it is not possible to estimate actual dates of emergence and onset of activity. The fixed array, however, provides some insights into this. In both 2012 and 2013, many terrapins were already active on the date that the tags turned on (Fig. S3). Dates of last detection in the fall do not correspond with re- duced activity. Median dates of last detection were 30 September (2011) and 3 October (2012), with no difference between sexes (Kruskal-Wallis, p > 0.57 in both years). Although it is tempting to infer from this that brumation had begun, those terrapins that were detected after these dates continued to be highly active, and it is likely that the absence of detections instead indicates that the terrapins had moved elsewhere in the system, away from the anchorage and the rest of the fixed-receiver array. This Bdeparture hypothesis^ is supported by the occupancy data (Fig. 3; Table 4). Occupancy in the anchorage was greatest during the May breeding aggregation (58% of tagged terrapins being present on any given day (%/day)), with another strong plateau inmidsummer, with about 45%/day. Occupancy declined rapidly from late August– to September. By October 1, only about 20% of the population occupied the dredge zone, and this proportion continued to decline until most movement ceased in December. It is possible that as movement declined, some terra- pins might have been present but undetected because they occu- pied zones outside the detection radius of the receivers. Any associated error in the occupancy estimate appears to be negligi- ble, however: in each year, eight individuals were detected in the anchorage using manual tracking, corresponding to 11% of tagged terrapins in 2011 and 8% in 2012. This is consistent with the smoothed occupancy estimate (Fig. 3). Data from the larger receiver array supports the interpreta- tion that most terrapins left the anchorage for the winter. Of 128 last fixed-receiver detections in 2011 and 2012, 97%were within creeks, with only four at open water receivers (Table 4, Fig. S5; see also Table 1 and Fig. 1). The distribution of the last date of detection was broadly distributed among the four major drainages within the Wellfleet Harbor estuary (mean ± SD number of individuals per drainage were 18 ± 3 in 2011 and 12 ± 2 in 2012), suggesting a pattern of dispersal, rather than aggregation. 0.0 0.3 0.6 0.9 Apr Jul Oct Jan Date ProportionYear 2011 2012 2013 Fig. 3 Mean proportion of tagged terrapins detected within the dredge zone. Data from each year are indicated by dots. The loess smooth represents the mean across years. The shaded area is the 95% confidence interval for this mean. Note the consistent aggregation that occurs in May of each year. This appears to correspond with the known mating aggregation that occurs at that time. This is followed in each year by a departure (perhaps reflecting nesting activity) but then by a subsequent return to the dredge zone in late June, which persists until mid-August. From mid-August through September, there is a dramatic exodus, with about half of the terrapins leaving the Duck Creek subdrainage. The remaining terrapins stay within the subdrainage until early November, when they also begin to leave. By December, nearly all of the terrapins have left the subdrainage, with only about 5% detected on any given day Estuaries and Coasts We also observed evidence of interannual, creek-specific fidelity in selection of fall and winter habitat. Of the 26 terra- pins that carried active tags through December of both 2011 and 2012 (Table 2), 23 individuals (88%) were last detected by fixed receivers after September 1 in both years.Of these, all but one were last detected in the same subdrainage in both years, most of them being detected on the same receiver as on the previous year (Fig. 4). We observed a similar pattern with emergence. During the spring emergence events of 2012 and 2013, terrapins were typically detected in the same drain- age as their last detection (Fig. S6). Notably, however, there were several first detections outside of the drainages where they had been last detected the previous fall, and nearly all of these occurred within the zone between the breakwater and Duck Creek. Because tags from tag lot 1 were dormant from January to May, and from December to April for tag lot 2 (Tables 2 and 3), it was not possible to determine with certain- ty when this movement occurred. By comparing the distribu- tions of last vs. first detections, however, it is possible to infer that activity had already begun by mid-April, making it more likely that those movements that were observed occurred in the early spring, before tag re-activation (Fig. S3). Most terrapins occupied habitat outside the anchorage dur- ing the fall and winter periods. Combined risk from occupan- cy and mobility restrictions was calculated by multiplying occupancy (Fig. 3; Eq. 1) by the transformed mobility risk (Fig. 2, Eqs. 2 and 3). The results indicate that a tradeoff exists between the two risk metrics: Occupancy is greatest at the same time that mobility is near its peak, resulting in a mini- mum combined risk occurring both during the spring period and again during the fall (Fig. 5). The low risk in spring is Table 4 Relationship between capture and brumation sites in 2011 and 2012. Data are presented as n (proportion of population captured within that subdrainage). Subdrainage labels are defined in Table 1. Note that a substantial proportion of terrapins captured within the Duck Creek watershed distribute throughout the harbor, while those caught outside of Duck Creek tend to brumate within the subdrainage where they were captured Subdrainage Year Capture Brumation 2011 2012 BFC BFC 5 (0.63) 4 (0.67) DC 0 (0.00) 0 (0.00) HR 0 (0.00) 1 (0.17) WBWS 3 (0.38) 1 (0.17) WH 0 (0.00) 0 (0.00) DC BFC 11 (0.22) 6 (0.23) DC 23 (0.45) 14 (0.54) HR 8 (0.16) 2 (0.08) WBWS 6 (0.12) 1 (0.04) WH 3 (0.06) 3 (0.12) HR BFC 0 (0.00) 0 (0.00) DC 0 (0.00) 0 (0.00) HR 8 (1.00) 7 (1.00) WBWS 0 (0.00) 0 (0.00) WH 0 (0.00) 0 (0.00) WBWS BFC 1 (0.13) 1 (0.13) DC 0 (0.00) 0 (0.00) HR 0 (0.00) 0 (0.00) WBWS 7 (0.88) 7 (0.88) WH 0 (0.00) 0 (0.00) 2011 H e rrin g R ive r 3 H e rrin g R ive r 2 H e rrin g R ive r 1 Gre a t Isla nd Je re m y Po in t M o o rin g Ba sin An ch o ra g e Th e C o ve D u ck C re e kR a ilro a d Brid ge C h a n n e l 1 In d ia n N e ck 1 In d ia n N e ck 2 Bla ckfish 1 Fo x I sla n d Ple a sa n t Po in t L o a g y BayL t. I sla nd Sa n ctu a ry 2 Sa n ctu a ryE a sth a m 2012Herring River 3 Herring River 2 Herring River 1 Great Island Jeremy Point Mooring Basin Anchorage The Cove Duck Creek Railroad Bridge Channel 1 Indian Neck 1 Indian Neck 2 Blackfish 1 Fox Island Pleasant Point Loagy Bay Lt. Island Sanctuary 2 Sanctuary Eastham # Observations 1 2 3 4 Fig. 4 Correlation between years of last receiver detection locations. Dashed rectangles delineate the four primary subdrainages (Fig. 1) Estuaries and Coasts driven by mobility, and in the fall by reduced occupancy. Any terrapins that remain in the anchorage after late November are assumed to be at high mortality risk, owing to their limited mobility. Although a small proportion of the population did remain near the anchorage at this time in both years, their locations fell primarily outside of the dredge zone (Fig. S7). Nevertheless, across the 2 years, three individuals were last detected within the dredge zone. The combined data thus in- dicate that a prolonged period of reduced risk exists from September to December, and probably through much of the winter. Nevertheless, there is a small but significant risk to individuals during the winter months. Discussion Data from this study provide a useful gauge to infer risk from harbor dredging, but the implications go beyond this. Risks from other activities can also be inferred, along with associat- ed seasonality, etc. The study has also shown some interesting patterns that shed light more generally on the movement ecol- ogy of this species. The tagged terrapins were highly mobile, in some cases traveling several kilometers in a single day. They actively moved between creeks, although this move- ment appeared to be primarily restricted to spring and summer months, and they exhibited strong fidelity to wintering habitat. This fidelity puts them at some risk from catastrophic events (e.g., severe weather, chemical spills, etc.)—a brumating pop- ulation that is locally extirpated may take considerable time to recover (Tucker et al. 2001). The evidence suggests, however, that the Wellfleet population is widely distributed throughout the available habitat, and there does not appear to be a single site that would render the population vulnerable to such an event. These observations differ from previous studies in that although other authors have described strong within-creek site fidelity, movements among creeks have been thought to be rare (Muehlbauer 1987; Gibbons et al. 2001; Tucker et al. 2001; Harden et al. 2007). Some of this difference, however, may be an artifact of the techniques used. By continuously monitoring movement with a fixed-receiver array, we were able to identify movements that can be missed using other mark-recapture techniques, and our observations may just be the result of improved resolution offered by acoustic teleme- try. Regardless, we did observe strong between-year fidelity to 0.00 0.25 0.50 0.75 1.00 Apr Jul Oct Jan Date Relative Daily RiskFig. 5 Three risk components, showing total risk scale. Occupancy risk (Ro. Eq. 1; dashed curve) is the proportion of known terrapins in the dredge zone and is on the same scale as in Fig. 3. Mobility risk (Rm Eq. 2; dash-dot) was calculated from the loess smooths for each year of data shown in Fig. 2 (Mmax = 2416m/day; Eq. 2). Combined risk (RT Eq. 3) is the product of occupancy andmobility components, and is shown by a solid black curve Estuaries and Coasts creeks during the fall months, which is consistent with other studies. In other ways, however, our observations were similar to previous studies. For example, we were unable to find and capture males during the summer months. Roosenburg et al. (1999) described within-creek habitat partitioning by sex and size, with larger mature females using more open, deep water habitats in contrast to the males and juvenile females using upper marsh habitats. Such sex-based partitioning might ex- plain our inability to capture males in summer. The fixed- station data did not show clear evidence of this, however; in our study, both sexes occupied open water areas, moving to the upper intertidal zones in late summer and early fall. Other authors have described age-dependent variability in location of brumacula (brumation sites), with adults remaining in the intertidal zone, either on top of the substrate, under scarred-out banks, or buried in mud, frequently in aggregations of several individuals (Haramis et al. 2011; Yearicks et al. 1981). Hatchlings, juveniles, and subadult terrapins often overwinter terrestrially, buried under the soil or dense vegetation above mean high water line (Lawler andMusick 1972; Muldoon and Burke 2012; Pitler 1985). We did not observe these patterns; however, we specifically targeted adult individuals in this study. If demographic segregation is occurring, it is likely to be happening primarily within the upper intertidal zone or above, in locations where we were unable to monitor. High levels of gene flow are common among terrapin pop- ulations. Published data suggest that there is a tendency towards male-biased dispersal, with limited genetic separation by dis- tance (Hauswaldt and Glenn 2005; Sheridan et al. 2010). Our collection data suggest that some sex-based habitat partitioning did occur. More work is needed to determine the extent of this partitioning, and whether this is typical of the species or if it reflects a unique characteristic of this population. It is unclear why the terrapins congregate in the harbor during summer months. Interestingly, this is the period of greatest human activity within the harbor. Also during this period, ground tackle (moorings, floats, etc.) is deployed throughout the anchorage. This gear creates a reef effect, with abundant invertebrate communities that may serve as forage for terrapins (Tucker et al. 1995). This equipment is removed in the winter to prevent ice damage—if terrapins are attracted to this structure in summer, its absence in winter might help to explain the timing of the fall exodus. Comparable studies per- formed at lower latitudes where ground tackle remains in place throughout the year might yield different results. In any event, it is evident that activity within the harbor does not repel terrapins, which calls into question the assumption that they would actively avoid a dredge should that activity occur during summer (Brennessel 2006; Cecala et al. 2009). Further, the summer occupancy of this zone suggests that any activity that did repel them might constitute harm to the pop- ulation through deprivation of access to vital resources. Adequacy of the Approach We applied a movement-theoretic approach to quantifying movement and occupancy. By analyzing the log-linear rela- tionship between the density function of the intervals between detections, we were able to differentiate between departures from receivers and the missed detections associated with im- perfect detection efficiency and movements near the periphery of receiver detection zones (Castro-Santos and Perry 2012). This approach recognizes the mixed distribution nature of movement data (Langton et al. 1995), and is an improvement over techniques that apply arbitrary thresholds to these inter- vals (e.g., Andrews and Quinn 2012; Chamberlin et al. 2011; Rohde et al. 2013). This approach, combined with a dense array of receivers, provided near-continuous monitoring of movements within the system of interest. By combining occupancy and mobility, we were able to produce a reasoned estimate of how these two factors combine to produce total risk. The occupancy component of the model is robust, biased only by the deliberately disproportionate number of terrapins collected within the zone of interest. The interannual fidelity raises concerns that portions of the population that are impacted might be slow to re-colonize (Tucker et al. 2001). These data were for last observations on fixed receivers, however, which we have shown did not correspond with cessation of movement or actual locations of brumation. Instead, brumation appears to occur further up the intertidal zone within the creeks. Furthermore, the anchorage had the lowest level of interannual fidelity, suggesting that any disturbance in this area would represent low risk to the popu- lation. A less biased study would have selected terrapins from throughout the harbor. However, given that the aggregation appears to draw from the entire population, the actual bias may be small, and the results are conservative. The other source of bias in our risk assessment comes from the assumption that there is zero risk when terrapins are max- imally mobile and 100% risk when they are dormant. While it is likely true that a dormant terrapin would not be able to avoid a dredge, the dredge would have to be at the same site, and if the brumation location does not co-occur with the dredge, then the risk to the dormant animal is overestimated using our approach. Conversely, the risk may be underestimated to the extent that terrapins are unable to avoid the dredge when max- imally active. Furthermore, we assume that the greatest dis- tance traveled corresponds with greatest ability to avoid the dredge. At the time of greatest movement in late May, the water temperature ranged from 15 to 20 °C. Mean daily tem- peratures reached as high as 25 °C in the summer, however. Terrapins are ectotherms, and performance tends to increase with temperature. This relationship is not linear though, and for many species, the greatest scope for activity occurs at temperatures below the maximum environmental temperature that they are likely to encounter (Brett and Glass 1973; Estuaries and Coasts Peterson et al. 1990). More work is needed to better under- stand the relationship between temperature and avoidance per- formance of terrapins. Still, it is important to recognize that the increased metabolic costs associated with activity during ele- vated temperatures often induce a state of reduced activity, meaning that it is possible for terrapins to be at greater risk from both exposure and mobility during the summer months. Given the data, it is possible to gauge the effect of any error that might exist in our approach. First, if some vul- nerability exists even during maximum activity, then the combined outcome would increase the expected risk dur- ing spring and summer. The fall/winter predictions would be relatively unaffected, indicating that the same period would still constitute the optimum period for risk minimi- zation. This same logic applies to any reduction in protec- tion from avoidance performance: If we are willing to as- sume that a dormant terrapin has zero avoidance ability, the general shape of the relationship will remain unchanged, with increased benefits accruing to performing dredging during fall and winter. In conclusion, the data compiled in this study provide strong evidence that risk to the Wellfleet Harbor population of diamondback terrapins is minimized during late fall through early winter. The approach described here could read- ily be applied to other species and environments, and holds promise as a conservation tool for sensitive populations. Acknowledgements Many people and organizations made important contributions to this project. Special thanks to Bob Prescott and the staff of the Massachusetts Audubon’s Wellfleet Bay Wildlife Sanctuary; Dr. Barbara Brennessel of Wheaton College; and Michael Flanagan and the staff of the Town of Wellfleet Harbormaster’s Office. These individuals and groups provided key resources and guidance, without which the project would have been impossible. We also thank agency staff for pro- viding guidance and insights into the management context for the study, including Eve Schluter, KevinMooney, and the staff of theMassachusetts Department of Conservation and Recreation and the Natural Heritage and Endangered Species Program, as well as Todd Randall and Craig Martin (US Army Corps of Engineers). This project was funded by a grant from the Massachusetts Department of Conservation and Recreation Waterways Grant (Project # P11-2660-G01 (3803-G)), the University of Massachusetts Amherst, Department of Environmental Conservation, the Diamondback Terrapin Working Group, and Zoar Outdoor (Charlemont, MA). A.J. Danylchuk is supported by the National Institute of Food & Agriculture, U.S. Department of Agriculture, the Massachusetts Agricultural Experiment Station and Department of Environmental Conservation. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government. References Andrews, K.S., and T.P. Quinn. 2012. Combining fishing and acoustic monitoring data to evaluate the distribution and movements of spot- ted ratfish Hydrolagus colliei. Marine Biology 159 (4): 769–782. Brennessel, B. 2006. Diamonds in the marsh: A natural history of the diamondback terrapin. Hanover: University Press of New England. Brett, J.R., and N.R. Glass. 1973. Metabolic rates and critical swimming speeds of sockeye salmon (Oncorhynchus nerka) in relation to size and temperature. Journal of the Fisheries Research Board of Canada 30 (3): 379–387. Butler, J.A., G.L. Heinrich, and R.A. Seigel. 2006. Third workshop on the ecology, status and conservation of diamondback terrapins (Malaclemys terrapin): results and recommendations. Chelonian Conservation and Biology 5 (2): 331–334. Castro-Santos, T., and R.W. Perry. 2012. Time-to-event analysis as a framework for quantifying fish passage performance. In Telemetry Techniques, ed. N.S. Adams, J.W. Beeman, and J. Eiler, 427–452. Bethesda: American Fisheries Society. Cecala, K.K., J.W. Gibbons, andM.E. Dorcas. 2009. Ecological effects of major injuries in diamondback terrapins: implications for conserva- tion and management. Aquatic Conservation-Marine and Freshwater Ecosystems 19 (4): 421–427. Chamberlin, J.W., A.N. Kagley, K.L. Fresh, and T.P. Quinn. 2011. Movements of yearling Chinook salmon during the first summer in marine waters of Hood Canal, Washington. Transactions of the American Fisheries Society 140 (2): 429–439. Culloch, R.M., P. Anderwald, A. Brandecker, D. Haberlin, B.McGovern, R. Pinfield, F. Visser, M. Jessopp, and M. Cronin. 2016. Effect of construction-related activities and vessel traffic on marine mam- mals. Marine Ecology Progress Series 549: 231–242. Gibbons, J.W., J.E. Lovich, A.D. Tucker, N.N. Fizsimmons, and J.L. Greene. 2001. Demographic and ecological factors affecting conser- vation and management of diamondback terrapins (Malaclemys terrapin) in South Carolina. Chelonian Conservation and Biology 4: 66–74. Haramis, G.M., P.P.F. Henry, and D.D. Day. 2011. Using scrape fishing to document terrapins in hibernacula in Chesapeake Bay. Herpetological Review 42: 170. Harden, L.A., N.A. Diluzuzio, J.W. Gibbons, and M.E. Dorcas. 2007. Spatial and thermal ecology of the diamondback terrapin (Malacelmys terrapin) in a South Carolina salt marsh. Journal of the North Carolina Academy of Sciences. 123: 154–162. Hart, K.M., and D.S. Lee. 2006. The diamondback terrapin: the biology, ecology, cultural history, and conservation status of an obligate es- tuarine turtle. Studies in Avian Biology 32: 206–213. Hauswaldt, J.S., and T.C. Glenn. 2005. Population genetics of the dia- mondback terrapin (Malaclemys terrapin). Molecular Ecology 14 (3): 723–732. Heppell, S.S., H. Caswell, and L.B. Crowder. 2000. Life histories and elasticity patterns: perturbation analysis for species with minimal demographic data. Ecology 81 (3): 654–665. Langton, S.D., D. Collett, and R.M. Sibly. 1995. Splitting behavior into bouts—a maximum-likelihood approach. Behaviour 132 (9): 781– 799. Lawler, A.R., and J.A. Musick. 1972. Sand beach hibernation by a north- ern diamondback terrapin, Malaclemys terrapin terrapin (Schoepff). Copeia 1972 (2): 389–390. Moser, M.L., and S.W. Ross. 1995. Habitat use and movements of shortnose and Atlantic sturgeons in the Lower Cape-Fear River, North-Carolina. Trans. Am.Fish.Soc. 124 (2): 225–234. Muehlbauer, E. (1987). Field and laboratory studies of tidal activity in the turtle Malaclemys terrapin terrapin. Thesis New York University. New York, New York. USA. PhD thesis, New York University. Muldoon, K.A., and R.L. Burke. 2012. Movements, overwintering, and mortality of hatchling diamond-backed terrapins (Malaclemys terra- pin) at Jamaica Bay, New York. Canadian Journal of Zoology- Revue Canadienne de Zoologie 90 (5): 651–662. O’Donnell, K.P., R.A. Wahle, M. Bell, and M. Dunnington. 2007. Spatially referenced trap arrays detect sediment disposal impacts on lobsters and crabs in a New England estuary. Marine Ecology Progress Series 348: 249–260. Estuaries and Coasts Peterson, C.C., K.A. Nagy, and J. Diamond. 1990. Sustained metabolic scope. Proceedings of the National Academy of Sciences of the United States of America 87 (6): 2324–2328. Pitler, R. 1985. Malaclemys terrapin terrapin (northern diamondback ter- rapin) behavior. Herpetological Review 16: 82. R Core Team. 2014. R: A language and environment for stattistical com- puting. (3.1.1). Vienna: R Foundation for Statistical Computing. Rohde, J., A.N. Kagley, K.L. Fresh, F.A. Goetz, and T.P. Quinn. 2013. Partial migration and diel movement patterns in Puget Sound Coho Salmon. Transactions of the American Fisheries Society 142 (6): 1615–1628. Roosenburg, W.M., K.L. Haley, and S. McGuire. 1999. Habitat selection and movements of diamondback terrapins, Malaclemys terrapin, in a Maryland estuary. Chelonian Conservation and Biology 3: 425– 429. Selby, T.H., K.M. Hart, I. Fujisaki, B.J. Smith, C.J. Pollock, Z. Hillis- Starr, I. Lundgren, and M.K. Oli. 2016. Can you hear me now? Range-testing a submerged passive acoustic receiver array in a Caribbean coral reef habitat. Ecology and Evolution 6 (14): 4823– 4835. Sheridan, C.M., J.R. Spotila, W.F. Bien, and H.W. Avery. 2010. Sex- biased dispersal and natal philopatry in the diamondback terrapin, Malaclemys terrapin. Molecular Ecology 19 (24): 5497–5510. Tucker, A.D., N.N. Fitzsimmons, and J.W. Gibbons. 1995. Resource partitioning by the estuarine turtle Malaclemys terrapin—trophic, spatial, and temporal foraging constraints. Herpetologica 51: 167– 181. Tucker, A.D., J.W. Gibbons, and J.L. Greene. 2001. Estimates of adult survival and migration for diamondback terrapins: conservation in- sight from local extirpation within a metapopulation. Canadian Journal of Zoology 79 (12): 2199–2209. Wilkens, J.L., A.W. Katzenmeyer, N.M. Hahn, J.J. Hoover, and B.C. Suedel. 2015. Laboratory test of suspended sediment effects on short-term survival and swimming performance of juvenile Atlantic sturgeon (Acipenser oxyrinchus oxyrinchus, Mitchill, 1815). Journal of Applied Ichthyology 31 (6): 984–990. Yearicks, E., R.Wood, andW. Johnson. 1981. Hibernation of the northern diamondback terrapin, Malaclemys terrapin terrapin. Estuaries 4 (1): 78–80. Estuaries and Coasts