Passive Thermal Refugia Provided Warm Water for Florida Manatees During the Severe Winter of 2009−2010

Vol. 462: 287–301, 2012 MARINE ECOLOGY PROGRESS SERIES Published August 21 doi: 10.3354/meps09732 Mar Ecol Prog Ser

Passive thermal refugia provided warm water for Florida manatees during the severe winter of 2009−2010

B. M. Stith1,*, D. H. Slone2, M. de Wit3, H. H. Edwards3, C. A. Langtimm2, E. D. Swain4, L. E. Soderqvist5, J. P. Reid2

1Jacobs Technology, Inc./US Geological Survey, Southeast Ecological Science Center, Gainesville, Florida 32605, USA 2US Geological Survey, Southeast Ecological Science Center, Gainesville, Florida 32605, USA 3Florida Fish and Wildlife Conservation Commission (FWC), Fish and Wildlife Research Institute (FWRI), St. Petersburg, Florida 33701, USA 4US Geological Survey, Florida Water Science Center, Ft. Lauderdale, Florida 33135, USA 5US Geological Survey, Florida Water Science Center, Ft. Myers, Florida 33901, USA

ABSTRACT: Haloclines induced by freshwater inflow over tidal water have been identified as an important mechanism for maintaining warm water in passive thermal refugia (PTR) used by Florida manatees Trichechus manatus latirostris during winter in extreme southwestern Florida. Record- setting cold during winter 2009−2010 resulted in an unprecedented number of manatee deaths, adding to concerns that PTR may provide inadequate thermal protection during severe cold periods. Hydrological data from 2009−2010 indicate that 2 canal systems in the Ten Thousand Islands (TTI) region acted as PTR and maintained warm bottom-water temperatures, even during severe and prolonged cold periods. Aerial survey counts of live and dead manatees in TTI during the winter of 2009−2010 suggest that these PTR were effective at preventing mass mortality from hypothermia, in contrast to the nearby Everglades region, which lacks similar artificial PTR and showed high manatee carcass counts. Hydrological data from winter 2008−2009 confirmed earlier findings that without haloclines these artificial PTR may become ineffective as warm-water sites. Tidal pumping of groundwater appears to provide additional heat to bottom water during low tide cycles, but the associated thermal inversion is not observed unless salinity stratification is present. The finding that halocline-driven PTR can maintain warm water even under extreme winter conditions suggests that they may have significant potential as warm-water sites. However, availability and conflicting uses of freshwater and other management issues may make halocline-driven PTR unreliable or difficult to manage during winter.

KEY WORDS: Halocline · Thermal inversion · Tidal pumping · Groundwater · Ten Thousand Islands · Picayune Strand restoration · Everglades restoration · Salinity stratification · Aquatic species

Resale or republication not permitted without written consent of the publisher

INTRODUCTION odylus acutus, sport fish such as common snook Cen- tropomus undecimalis and tarpon, and exotic species Large numbers of cold-sensitive wildlife species such as fresh water cichlids, pythons, and iguanas died in Florida during the record-setting winter of (FWRI 2010). Florida recorded its coldest January to 2009−2010, including endangered species such as March in the 116 yr period of record (NCDC 2010) manatees, sea turtles, and American crocodiles Croc- during this winter. The month of January was espe-

*Email: [email protected] © Inter-Research 2012 · www.int-res.com 288 Mar Ecol Prog Ser 462: 287–301, 2012

cially severe, with 175 new lowest daily minimum temperature records and 158 new lowest maximum temperature records set across the state (NCDC 2010). The number of manatee deaths was unprecedented, even in south Florida, with 480 carcasses recorded statewide during 3 mo from January 11 to April 9, 2010, a record number even for any previous 12 mo period (Barlas et al. 2011). This prompted the Working Group on Marine Mammal Unusual Mortal- ity Events to declare an unusual mortality event (UME; Title IV, Marine Mammal Protection Act). Florida manatees Trichechus manatus latirostris, an endangered sub-species of the West Indian manatee T. manatus manatus, inhabit fresh- and brackish- water rivers, bays, and estuaries in the subtropical regions of Florida and the southeastern United States Fig. 1. Halocline conditions at Port of the Islands canal and (northern most limit for this species). Manatees in Faka Union weir showing (large arrow) tidal wedge and Florida have a limited ability to thermoregulate in (small arrows) freshwater surface layer cold water and regularly suffer from cold-related mortality (Irvine 1983, Bossart et al. 2003). Manatees the PTR in winter 2006−2007 and 2007−2008 and are increasingly seen to retreat to warm-water refu- confirmed by simulations from a 3-dimensional gia as water temperatures fall below 20°C to avoid physics-based hydrology model for one of the PTR cold stress (Deutsch et al. 2003). Most manatees take (Decker et al. 2012). shelter in power plant effluents or artesian springs An understanding of the mechanism that maintains during cold periods, but a substantial proportion of the thermal properties raised the possibility of man- the population (typically several hundred manatees) aging or creating halocline-driven PTR (Stith et al. are regularly found at passive thermal refugia (PTR) 2011). However, lack of hydrological data for winters especially in areas that lack springs or industrial with severe cold periods left open the possibility that warm water, such as extreme southern Florida factors such as tidal mixing, wind, varying freshwater (USFWS 2001, Laist & Reynolds 2005a). These PTR inputs, or conduction could disrupt the temperature differ from springs or power plants in not having con- inversion during severe cold. The unprecedented spicuous, active point sources of warm water inflow. cold of winter 2009−2010 provided data to evaluate As such, the ability of PTR to adequately support the hydrodynamics under extreme conditions. Stith large numbers of manatees during cold winters, even et al. (2011) also noted evidence for a source of bot- in south Florida, has remained unclear (Laist & tom heat, indicated by multi-day periods when the Reynolds 2005a,b). bottom layer was warming while the surface layer Recently Stith et al. (2011) studied 2 PTR used by was cooling, but the source of heat was not deter- manatees in extreme southwestern Florida and iden- mined. We examined additional hydrological data, tified the hydrological mechanism that maintained including canal tidal stage and groundwater level, to warm water temperatures. The PTR were part of look for potential hydraulic head differences and evi- canal systems that received freshwater from up - dence that tidal pumping of groundwater could be a stream and interaction with tidal waters downstream source of heat. (Fig. 1). The lighter freshwater flows over heavier Additionally, the large number of wildlife deaths saltwater, forming a halocline, or sharp, vertical prompted extensive aerial and boat surveys that sup- salinity gradient between stratified layers. When suf- plemented year-round manatee carcass recovery ficient fresh water flow was present, strong vertical efforts to find and retrieve all carcasses and to deter- salinity gradients provided barriers to convective mine cause of death. We used these data to assess mixing, which in winter allowed warm bottom water whether manatee mortality from cold was reduced to persist below colder surface water. These benefi- for manatees in the Ten Thousand Islands (TTI) cial temperature inversions would disappear if no region, where 2 halocline-driven PTR have been doc- halocline was present due to convective mixing of umented, compared to areas to the south in the west- the water column. The mechanism was identified by ern coastal Everglades (WCE), which lack similar analysis of empirical hydrological data collected at artificial PTR. This 2 pronged approach allowed us to Stith et al.: Passive thermal refugia for manatees 289

examine whether halocline-driven PTR can function Mexico and inner bays/rivers, whereas in TTI there as adequate warm-water refugia for manatees dur- are hundreds of pathways. Telemetry data suggest ing severe winters in this region. that manatee movements in the 2 areas are very different (USGS unpubl.). Moreover, the availability of artificial warm-water refugia differs between the 2 MATERIALS AND METHODS regions. The TTI region included the 2 PTR studied by Stith Study area et al. (2011), both associated with dredged canal systems southeast of Naples; one at Port of the Islands Our study area encompassed extreme southwest- (POI), and the second at Big Cypress National Pre- ern Florida, including the southern coastal region of serve Headquarters (BCNP) (Fig. 2), Ochopee, Flo- Collier County, centering on the TTI region and rida. POI regularly harbors more than 100 manatees extending south to Cape Sable in the Everglades during typical winter aerial surveys (USFWS 2001). (Fig. 2). We divided the study area into 2 large Manatees aggregate in the residential canal system regions that correspond roughly to the TTI region at the tidal head of the Faka Union (FU) Canal, which (west of Turner River to Marco Island) and the west- is directly downstream from the Picayune Strand ern coastal Everglades (Turner River south to Cape Restoration Project. The restoration project will Sable) (Fig. 2). Although the 2 regions are very simi- remove or plug most of the 77 km canal system that lar ecologically, the underlying geology and historic was completed in 1971 to drain a failed real estate hydrology has produced pronounced physiographic development that is now under public ownership (US differences in the 2 regions. The TTI coast is highly Army Corps of Engineers and South Florida Water dissected with innumerable channels, cuts, and small Management District 2004). The PTR at BCNP con- islands, whereas the Everglades National Park (ENP) sists of dredged canals that receive freshwater from is much less dissected with only a few large rivers the Big Cypress Basin, with a shallow, circuitous, and and large islands (Parkinson 1998). Manatees in the lengthy tidal connection to Chokoloskee Bay via ENP have very few corridors connecting the Gulf of Halfway Creek. The western coastal Everglades region has a nearly continuous coastline with 5 large rivers connecting the Gulf of Mexico to an extensive set of inner bays, the largest being Whitewater Bay at the southern end. Several small winter ag - gregation sites for manatees, the most noteworthy of which being Mud Bay, have been identified based on winter aerial surveys (Stith et al. 2006) and a USGS telemetry study (USGS unpubl. data). Conversely, the WCE region lacks canal systems, like POI or BCNP, that are de signed to drain wetlands and are accessible to manatees. The few canals that exist, such as Buttonwood Canal and boat basin complex at Flamingo, were designed as transportation short- cuts for boats, rather than as parts of a drainage project that receives freshwater discharge. Water depths in the canals at POI ave - Fig. 2. Study area showing the Ten Thousand Islands (TTI) and rage around 2.5 to 3.0 m with a few western coastal Everglades regions, separated by a bold dashed line. deeper holes reaching 3.5 to 4.5 m. The The 2 artificial passive thermal refugia (PTR) in TTI at Port of the Is- canals at BCNP are considerably deeper, lands (POI) and Big Cypress National Park are shown, along with a natural PTR, Mud Bay, in the Everglades. SGT5W3 is a ground water averaging around 5.0 to 6.0 m with some monitoring well approximately 3 km from POI deeper sections reaching 7.0 to 8.5 m. Wa- 290 Mar Ecol Prog Ser 462: 287–301, 2012

ter depths at Mud Bay average around 1.2 to 1.6 m, Hydrological data and analysis at the PTR with discrete deeper holes reaching 2.5 to 3.0 m. As a continuation of prior research (Stith et al. 2011), we monitored the temperature and salinity at Manatee data POI and BCNP for 2 additional winters (2008−2009 and 2009−2010) to characterize the vertical tempera- Special aerial and boat surveys were conducted to ture and salinity regime. At each site we deployed search systematically for manatee carcasses, as well 2 YSI 600 XLM water quality sondes within 15 cm of as American crocodile and other cold-sensitive spe- the surface using a floating mount, and the other cies, throughout WCE and TTI during mid to late within 15 cm of the bottom. Date and time, tempera- January 2010. Locations of carcasses were recorded ture, conductance, and salinity (calculated using on maps and via GPS. Reports of carcasses were also temperature and conductance) were logged every obtained from other sources such as boaters and fish- 15 min. ermen. In 2009−2010 we added a vertical string of ther- Carcasses were verified and recovered whenever mistors to the monitoring station at POI to record possible. Necropsies were conducted either on site or temperatures at 30 cm vertical increments, with the at the Fish and Wildlife Research Institute (FWRI) deepest thermistor buried 30 cm into the mud Marine Mammal Pathology Laboratory in St. Peters- layer, and the shallowest 168 cm above the mud. burg, Florida to identify cause of death. The mud temperature was of special interest We restricted our analysis of carcass data to a 2 wk because bottom-resting manatees were often seen period (January 13 to 28, 2010) because the most partially coated in mud as they rose to the surface intensive sampling effort occurred during this period to breathe, suggesting that mud temperatures and was relatively uniform across the study area. might be warmer than bottom water ones, as noted This period also overlapped the first and most severe by Stith et al. (2011). cold weather of the 2010 winter, when immediate Temperature, salinity, and stage data from a death from acute cold shock rather than prolonged groundwater monitoring well near POI (SGT5W3; decline from chronic cold stress was most likely location on Fig. 2; approx. 3 km from POI) were ob - (Bossart et al. 2003, Barlas et al. 2011). The short time tained from the South Florida Water Management frame between extreme cold and carcass recovery District (SFWMD) DBHYDRO data website (www. made it more probable that an individual would have sfwmd.gov) for 2 winters (2008−2009 and 2009− died in the area where it experienced severe cold. 2010). We investigated data from another nearby On January 14, 2010 the Florida Wildlife Commis- well (SGT5W2) but did not use these data because sion (FWC) flew its annual winter manatee synoptic the groundwater was much less saline than both aerial survey during the first extended cold spell, SGT5W3 and the POI bottom water, and SGT5W2 providing an initial snapshot of the distribution of stage data showed large, non-tidal fluctuations that live animals as the mortality event unfolded. These were absent at SGT5W3. We also obtained stage data statewide (synoptic) aerial surveys cover manatee (3 to 4 min intervals and daily averages) from just habitat in the major river systems, inland bays, shal- downstream of the FU-1 weir to investigate the role low Gulf shoals, and all known primary and second- of tidal fluctuations in the canal. The uneven- ary manatee winter aggregation sites in the study time-interval tidal data were interpolated to match area. Synoptic surveys (Ackerman 1995) follow a the 15 min interval groundwater stage. standardized protocol and flightlines; however both We plotted surface and near-bottom temperature have been modified slightly over the past 20 yr as and salinity data in vertical panels for the 2 winters more information about survey timing and manatee for both sites. We plotted groundwater temperature distribution has been ascertained. Because of factors data for SGT5W3 with the POI mud and water tem- such as survey design, availability and hetero - perature data to see if groundwater was a possible geneous detection of manatees, counts from these source of bottom heat. We also plotted groundwater surveys are usually considered underestimates (min- levels and tidal stage in the canal as an indicator of imum counts) and are not true estimates of abun- differences in hydraulic head and possible tidal dance (Edwards et al. 2007, Fonnesbeck et al. 2009). pumping (Ganju 2011). Discharge calculations for The synoptic survey data allowed us to compare the the FU-1 weir at POI were obtained from DBHY- number of reported carcasses to the number of live DRO (SFWMD website). Discharge at BCNP was manatees seen in TTI and WCE. estimated by USGS at bimonthly intervals using Stith et al.: Passive thermal refugia for manatees 291

velocity profile measurements and a cross-section RESULTS estimate. Discharge data were plotted for POI and BCNP in vertical panels with temperature and PTR temperature inversions and stratification salinity data. Vertical profiles of salinity and temperature were Winter 2008−2009 measured during a peak cold period on January 8, 2010 along a north−south gradient from just up - Bottom temperatures at POI remained above 20°C stream of the FU-1 weir to the southern end of the and temperature inversions were strong for the first canal at the Faka Union Bay (6 sampling locations, half of the winter, then became cold and fluctuated in see Appendix, Fig. A1). Three sites were within side near-synchrony with the surface as the halocline dis- canals in deeper holes used by manatees. These data appeared in the second half of the winter (Fig. 3). allowed us to characterize the depth of the fresh - During this period with no salinity stratification, the water layer and the spatial extent of the salinity strat- bottom temperature still cooled more slowly than the ification in the canal system. surface, showing a thermal lag of 1 or 2 d. Freshwater Following the approach of Stith et al. (2011), we discharge was highest at the start of winter, and calculated the density of water at the surface and declined steadily to 0 by mid-February. This associa- bottom for POI (Fofonoff & Millard 1983). We used tion between loss of discharge and loss of tempera- data from both winters to portray a wide range of ture-inverted haloclines is similar to 2006−2007, as values under stratified and unstratified conditions. reported by Stith et al. (2011; Fig. 3). Brief periods For each pair of bottom−surface measurements, we can be seen during both winters when the bottom plotted the vertical difference in temperature and was warming while the surface was cooling (example salinity, for 3 surface temperature thresholds re - in Fig. A2), but only when a halocline was present. presenting increasingly stressful conditions for As with all other years reported here and in Stith et manatees (<20, ≤18, ≤16°C). We also calculated the al. (2011), BCNP bottom temperatures were substan- density differences between the bottom and surface tially warmer and more stable than POI. Discharge and overlaid a contoured surface of these density was much lower in 2008−2009 at BCNP than in differences on a plot of the corresponding vertical 2009−2010, with bi-monthly readings falling close to salinity and temperature differences. The plot 0 by early December (Fig. 4). Accordingly, 2008− emphasizes extreme cold temperatures (≤16°C) not 2009 was the only winter when the halocline disap- found in previous years (Stith et al. 2011), with peared at BCNP (end of February 2009), and during density contours to show whether warmer, more this period the bottom temperatures fluctuated much saline bottom water remained denser than ex - more strongly than any other year (Fig. 4). Surface tremely cold surface water, and whether larger waters were rapidly warming at that time, so the bot- density differences were associated with stronger tom water at BCNP did not dip below 20°C. salinity stratification. Correlation analyses were conducted for data at POI as in Stith et al. (2011) to examine relationships Winter 2009−2010 among several factors: (1) temperature difference (bottom minus surface), (2) salinity difference (bot- Bottom temperatures remained above 20°C and tom minus surface), and (3) freshwater discharge. We strong temperature inversions persisted at POI examined the relationship between freshwater dis- throughout the severe winter of 2009−2010 (Fig. 5) charge and temperature inversion to test the hypoth- with the exception of some brief bottom temperature esis that increasing flow rates and decreasing surface dips to 18 to 19°C for a 1 wk period in mid-January. water temperatures would result in smaller tempera- The temperature of the mud layer at POI remained ture inversions. For this analysis we binned the sur- above 22°C for the entire winter. During several cool- face temperatures into 4 categories (>20°C, 17.99 to ing and warming cycles, a de-coupling of the bottom 20°C, 16 to 18°C, and <16°C). We also examined the and surface temperatures was evident at POI, with relationship between an indicator of head difference, the bottom layer warming while the surface was measured as groundwater level minus tide stage, cooling (Fig. 6). These bottom-warming, surface- versus vertical temperature difference (bottom minus cooling trends lasted multiple days, making it un - surface) to test the hypothesis that larger head dif - likely that the semi-diurnal tidal movement of water ferences were associated with larger temperature bodies with different temperatures could explain the inversions. ob served patterns. The bottom-warming, surface- 292 Mar Ecol Prog Ser 462: 287–301, 2012

) 12 a much colder than the bottom, unless –1 9 there was no salinity stratification. A s 3 6 major difference was the large num- 3 ber of hourly readings that fell below 16°C at the surface in 2009−2010

Flow (m 0 –3 (Fig. 7), compared to virtually none in b all previous years. These coldest points 30 were associated with the largest tem- 20 perature inversions ob served during Ground all years, but still fell within the most Salinity 10 Bottom stable region where density differ- Surface ences were largest. The near vertical c orientation of the density contours 2 indicates that salinity had a stronger 1 influence on density than temperature 0 Ground for the observed conditions, allowing

Tidal stage M tide –1 Tide the maintenance of these large tem- 24 perature inversions. 22

(°C) 18 Ground 14 Bottom Surface water discharge, salinity Surface Temperature 10 d stratification, 15 22 29 06 13 20 27 03 10 17 24 31 07 14 21 28 07 14 21 28 and temperature inversion Dec 08 Jan 09 Feb 09 Mar 09 The salinity difference between the Day of the month bottom and surface at POI for the 2008−2009 and 2009−2010 winters Fig. 3. Port of the Islands (POI) (a) POI daily mean discharge, (b) POI surface, bottom, and SGT5W3 groundwater salinity, (c) SGT5W3 groundwater level was strongly correlated with upstream and POI tide stage (15 min interval, Tide; and daily mean, M tide), and (d) freshwater discharge over the POI SGT5W3 groundwater, POI surface and bottom temperatures at 15 min weir (Fig. 8; Pearson’s coefficient = intervals for winter 2008−2009 0.699, df = 230, p < 0.0001) and was very similar to previous years (Fig. 10 cooling trends indicate a source of bottom heat sepa- of Stith et al. 2011). The relationship between dis- rate from surficial interactions. charge and temperature inversion for different Freshwater discharge was continuous during win- ranges of surface temperatures (Fig. 9) indicated that ter 2009−2010, fluctuating around 6 m3 s−1 for most of temperature inversions increased as flow increased the winter (Fig. 5). The continuous discharge and and as surface temperatures decreased. However, a persistent halocline can be attributed to above aver- limit must exist where a sufficiently large, long- age winter precipitation (NCDC 2010). BCNP bottom duration flow rate at a sufficiently low enough tem- temperatures were substantially warmer and more perature would cause the bottom to cool to detrimen- stable than those at POI, remaining above 24°C tally low temperatures (Decker et al. 2012). This limit nearly the entire winter (Fig. 4). Positive freshwater was not reached in the observed data and appears to discharge at BCNP was not as large as at POI, be unlikely, given that the largest flow rates are asso- remaining below 2 m3 s−1 during all but 1 bi-monthly ciated with the warmer wet season. reading (Fig. 4).

Salinity profile Density relationships The vertical profile data at different locations Density relationships in 2008−2009 and 2009−2010 within POI on January 8, 2010 showed a pro- were similar to the previous 2 winters reported in nounced halocline across most of the upper canal Stith et al. (2011), showing greater densities in the system. A fresh or brackish water layer at the sur- bottom versus surface even when the surface was face ranged from 0.33 to 1 m in depth, with the Stith et al.: Passive thermal refugia for manatees 293

3 a temperature difference between the 2.5 bottom and surface layer for winter )

–1 2 2009−2010 (Fig. 10, surface < 16°C; s 3 1.5 Pearson’s coefficient = 0.46, df = 1523, 1 p < 0.0001). Stable isotope analysis conducted 0.5 Flow (m Flow for another study indicated that the 0 bottom water at POI has tidal water –0.5 from the Gulf as its primary source b (USGS unpubl. data), suggesting that 30 the groundwater is recirculated seawater that has intruded into the surfi- 20 cial aquifer around the canal. Salinity 10

0 Live and dead manatees 28 c 24 Aerial survey counts of live manatees in the TTI region were more than 20 twice that of the WCE region (Table 1; 16 166 and 75 ind., respectively). Sev- Bottom enty-five percent of the total count for 12 Surface

Temperature (°C) TTI occurred at the 2 PTR. The largest manatee aggregation in the WCE Nov 08 Dec 08 Jan 09 Feb 09 Mar 09 Nov 09 Dec 09 Jan 10 Feb 10 Mar 10 Apr 10 region was at Mud Bay (45 ind.), Month accounting for 60.0% of the total Fig. 4. Big Cypress National Park (a) daily mean discharge, (b) surface and count in WCE. Because of heteroge- bottom salinities, and (c) surface and bottom temperatures at 15 min intervals neous detection rates and other issues, for winter 2008−2009 and winter 2009−2010 these survey results represent minimum counts. greater depth of the surface layer occurring nearer Counts of dead manatees, however, were many to the source of freshwater (Appendix, Fig. A3). The fewer in TTI compared to WCE (Table 1; 11 and 2 sites in the main canal showed higher salinities in 76 ind., respectively). Between January 13 and 28, the upper layer (15 to 18) than in the side canals. 2010, a total of 5 carcasses were found at POI and the The southern entrance to the canal (5 km down- Faka Union Canal connecting it to the Gulf, with 6 stream of POI) had no stratification and showed uni- additional carcasses recovered throughout the rest of formly high salinities that were equivalent to Faka the TTI region. Cause of death was determined to be Union Bay. cold related for all but 2 of the carcasses from POI and TTI. Because of the remoteness and inaccessi bility of many of the carcasses in WCE, 65 of the 75 were veri- Tidal pumping and groundwater discharge fied but because a full necropsy was not performed, cause of death could not be determined. Cold-related During several major cold periods, bottom temper- causes were identified for the remaining 10 carcasses atures showed warming trends even while surface and strongly implicated in the 65 others due to the ex- temperatures were rapidly declining and colder than ceptionally high number of carcasses found associ- the bottom (Figs. 6 & A2). These bottom-warming, ated with the brief, unprecedented cold period. surface-cooling trends occurred during downward Carcass counts for the remainder of the winter trending or extreme low tides that are usually associ- (January 29 to March 19) showed a similar pattern, ated with the passage of strong cold fronts, suggest- with an additional 26 carcasses with undetermined ing that tidal pumping was advecting warm ground- cause of death reported in WCE, and 8 cold-related water into the bottom canal water. A significant carcasses (7 cold-related, 1 with undetermined cause correlation was found between the hydraulic gradi- of death) found in the TTI region (4 at POI or the ent (groundwater stage minus tidal stage) and the Faka Union Canal). 294 Mar Ecol Prog Ser 462: 287–301, 2012

12 a parsimoniously by the availability of )

–1 9 warm water we measured at POI and s 3 6 BCNP during this record-setting cold, 3 suggesting that these PTR played an important role as warm-water sites for 0 Flow (m manatees in this region. –3 b The findings that PTR provided warm water even during severe cold 30 Ground are significant because uncertainty 20 Bottom exists about the ability of PTR to func- Surface Salinity 10 tion adequately as warm-water refugia (Laist & Reynolds 2005b). PTR are generally thought to provide marginal 2 or unreliable warm water compared to 1 the 2 major sources of warm water used by most manatees: springs and 0 Ground industrial thermal effluent (primarily

Tidal stage M tide –1 power plants) (Laist & Reynolds c Tide 24 2005a,b). Possible loss of major warm- water sites, due to power plant shut- 22 Ground downs or reductions of spring flow, 18 Mud has been identified as a significant 14 Bottom Surface threat to the long-term viability of the 10 d

Temperature (°C) manatee population, second only to 28 05 12 19 26 02 09 16 23 30 06 13 20 27 06 13 20 27 watercraft mortality (Runge et al. Dec 09 Jan 10 Feb 10 Mar 10 2007). PTR could become more attrac- Day of the month tive or important to manatees with the loss of warm-water sites. A complicat- Fig. 5. Port of the Islands (POI) (a) POI daily mean discharge (b) POI surface, bottom and SGT5W3 groundwater salinity, (c) POI tide stage (15 min interval, ing factor is that various larger scale tide; and daily mean, M tide) and SGT5W3 groundwater level, and (d) temp- changes, such as water-management eratures for POI shallow mud, surface, bottom, and SGT5W3 groundwater at policies or climate change, could alter 15 min intervals for winter 2009−2010 the availability of freshwater at PTR and perhaps the temperature regimes. DISCUSSION Thus, understanding the mechanisms responsible for providing warm water at PTR has been identified as Despite the record-setting cold temperatures during important to the management and recovery of mana- the severe winter of 2009−2010, the PTR at POI and tees (USFWS 2001, FFWCC 2007). BCNP maintained bottom water near or above 20°C, Multiple mechanisms appear to operate at POI and providing temperatures sufficiently warm for most BCNP to create and maintain warm water, and simi- manatees to avoid severe cold stress or hypo thermia lar mechanisms may operate at other PTR in Florida. (Bossart et al. 2003, Laist & Reynolds 2005a). The We suggest 3 primary mechanisms can be used to large counts of live manatees at these 2 PTR and the characterize PTR: (1) thermal inertia, (2) salinity relatively small carcass counts in the associated TTI stratification, and (3) groundwater discharge. All 3 of region contrasted sharply with WCE, where a large these mechanisms may operate at a given site. Salin- number of manatees were found dead, most likely ity stratification and groundwater discharge may because they did not have access to PTR that func- be absent or operate intermittently or continuously tioned as well as POI or BCNP as warm-water re - at a given PTR, resulting in different warm-water fugia. Although detection rates, survey effort, and regimes. Thus, identifying the dominant mechanisms other factors undoubtedly differed between regions, may be important to understanding the conditions such biases seem unlikely to account for the large dif- under which a site might be unreliable or manage- ferences in the number of dead and live manatees de- able as a winter refuge. tected in the 2 regions. The small carcass count and Thermal inertia is a mechanism that is present at large live count in the TTI region can be ex plained many PTR simply because they have greater depths Stith et al.: Passive thermal refugia for manatees 295

tia effect can be seen at POI in the second half of winter 2008−2009 (Fig. 3), 30 when salinity stratification was absent. Ground 20 Bottom Warmer minimum bottom tempera- Surface

Salinity tures and a delay in cooling occurred 10 compared to the surface temperatures. a The ob served thermal lags were brief, lasting 1 or 2 d, and bottom tempera- 2 tures still fell below 20°C during strong 1 cold fronts. A similar pattern was seen at the Matlacha, Florida, PTR, where 0

Tidal stage Ground Barton (2006) concluded that the –1 M tide Tide warmer temperature of the canals was b due to heat retention associated with 24 greater depths and re duced tidal flush - 22 ing. Thermal inertia may provide warm 18 Ground bottom-water at PTR during cold peri- Mud ods that are mild or short in duration. 14 Bottom However, during severe or prolonged c Surface Temperature (°C) 10 cold periods, the bottom temperatures 18 20 22 24 26 28 30 01 03 05 07 09 11 13 15 17 19 21 23 may become excessively cold if ther- Day of the month mal inertia is the only mechanism Fig. 6. Port of the Islands (POI), Dec–Jan 2010. (a) POI surface, bottom, and reducing heat loss. Under such condi- SGT5W3 groundwater salinity, (b) POI tide stage (15 min interval, tide; and tions, these PTR are likely to be unreli- daily mean, M tide) and SGT5W3 groundwater level, and (c) temperatures for POI shallow mud, surface, bottom, and SGT5W3 groundwater. (b) Green able as warm water refugia (Barton ellipses show low tide cycles associated with strong cold events. (c) Green 2006, Stith et al. 2011) and may func- ellipses show corresponding periods with bottom-water warming, surface- tion as temporary stopovers for migrat- water cooling ing or foraging manatees under mild winter conditions (Deutsch et al. 2003). compared to other water bodies used by manatees. Another heat-related mechanism that may be Typical artificial PTR consist of dredged canals or significant in some PTR is groundwater discharge. basins with in creased depths that increase thermal Springs have groundwater discharge that is charac- mass, have small surface dimensions relative to vol- terized as laminar flow through vents or seeps (Scott ume that reduces heat loss to the atmosphere, and are et al. 2004), whereas at PTR groundwater discharge, narrow with limited fetch that reduces wind-induced

0 0 1 0 5 0 convection (Stith et al. 2011). A 3-dimensional hydrol- −−10 1 5 0

10 1 2

1 20 ogy model of POI showed that increasing canal depth by 1 m substantially reduced the time duration of bottom temperatures falling below 20°C (Decker et al. ) C 1010 ° (

2012). This simulated thermal inertia effect occurred e in the absence of salinity stratification. A thermal iner- c n e r e f f i

d 5

Fig. 7. Vertical differences (bottom minus surface) at Port of e r the Islands for salinity versus temperature and correlations u t a when surface temperatures were below 16°C (large black r e

circles and thick black line), below 18°C (small black circles p

and thin black line), or below 20°C (small open circles and m 0 e

dashed line) for winter 2008−2009 and 2009− 2010. Points Temperature difference (°C) T with positive temperature differences show temperature inversions (bottom warmer than surface). Contours indicate vertical density differences (kg m−3; bottom to surface), with neutral density occurring at the zero contour and increas- 0 5 1010 1155 2200 2255 3030 ingly stable configurations (bottom denser than surface) occurring at points further to the right SalinitySalinity ddifferenceifference 296 Mar Ecol Prog Ser 462: 287–301, 2012

Top <16°C Top <20°C 25 15

10 15

10 5 Salinity difference 5 Temperature difference (°C)

0 0

0 2 4 6 8 10 12 –1 0 1 2 3 Mean daily discharge (m3 s–1) Head (groundwater − tide) Fig. 8. Correlation between salinity difference (bottom to Fig. 10. Port of the Islands (POI) temperature differences surface) versus freshwater discharge at Port of the Islands (bottom minus surface) versus index of hydrologic head for winter 2008−2009 and 2009−2010 (groundwater stage minus tidal stage) and correlations when surface temperatures were below 16°C (dark heavy 12 points and black correlation line) or 16–20°C (small grey cir- Top <16°C cles and gray dashed correlation line) for winter 2009−2010. Top 16−18°C Points with positive temperature differences indicate tem- 10 Top 18−20°C perature inversions (bottom warmer than surface). Points Top >20°C with positive head indicate groundwater levels higher than tidal stage 8

Table 1. Trichechus manatus latirostris. Number of live and 6 dead manatees counted from 13 to 29 Jan 2010 at the major aggregation site of the Port of the Islands and within the Ten 4 Thousand Islands and western coastal Everglades (WCE) regions. Cause of death could not be determined for most carcasses in the SW Everglades 2 Location No. of No. of carcasses Temperature difference (°C) 0 live indi- (no. with unde- viduals termined cause of death) –2 Port of the Islands 135 5 (1) 0 2 4 6 8 10 12 Ten Thousand Islands Region 3 –1 (including Port of the Islands) 166 11 (2) Mean daily discharge (m s ) WCE 75 75 (65) Fig. 9. Correlation between freshwater inflow and temperature difference (bottom minus surface) when surface temperatures were below 16°C (large black circles and thick Head pressure in surficial aquifers is relatively small black line), 16–18°C (small black circles and thin black line), 18–20°C (large gray circles and thick dashed line) and due to the shallow elevational gradients found in above 20°C (small gray circles and thin dashed line), at Port Florida (Reich 2010). Nevertheless, groundwater dis- of the Islands for winter 2008–2009 and 2009–2010 charge can be a significant heat source in canals, rivers, and other water bodies, and heat measure- if present, is more diffuse and generally lacking ments are commonly used as a groundwater tracer observable point sources. Groundwater entering PTR (Anderson 2005, Constantz et al. 2008). If a PTR has a typically comes from unconfined surficial aquifers strong halocline that suppresses vertical mixing, rather than artesian aquifers (Laist & Reynolds 2005a). even a relatively small volume of warm groundwater Stith et al.: Passive thermal refugia for manatees 297

seepage could be held in a thin layer at the bottom not only persisted but became larger as surface that retains a significantly higher temperature. Shal- temperature decreased and flow increased (Fig. 9). low groundwater in south Florida is typically sub- This relationship might seem counterintuitive, since stantially warmer than 20°C even during winter in creased mixing might be expected as flow (Fig. 3), and in some areas discharge rates are suffi- increased, and greater heat loss and turnover might cient to maintain warm water temperatures even be ex pected as surface waters cooled and became during severe cold. For example, the freshwater L31 denser. However, the density of the saline bottom canal on the east side of the Everglades acted as a water remained greater than that of the colder, groundwater-driven PTR for nonnative freshwater fresher surface water (Fig. 7), resulting in a stable fish during the severe cold of January 2010 (Harvey density distribution that prevented the temperature et al. 2011) and remained above 17°C during the inversion from being disrupted by factors that might coldest period. Coastal canals in nearby Coral Gables cause vertical mixing, such as tidal action or wind. provide a well-known manatee PTR (Laist & Rey - The amount of freshwater inflow required to create nolds 2005a,b), as do some other canals in the exten- and maintain an effective temperature inverted halo- sive multi-county canal system, which receive signif- cline is likely site-specific. At POI, a surface layer of icant warm groundwater discharge from the shallow, freshwater roughly 1 m thick was associated with the porous Biscayne aquifer (Renken et al. 2005). pronounced temperature inversion during the severe At POI, a distinctive heat signature of groundwater cold of winter 2009−2010, forming a pycnocline or discharge can be seen during the extreme cold sharp boundary between layers with different salini- periods of 2009−2010, when bottom temperatures ties (Fig. A3). Pycnoclines may not be necessary to showed multi-day warming trends even while the maintain thermal inversions; partial stratification colder surface water rapidly declined in temperature with a gradual increase in salinity from surface to (Fig. 6). These heat signatures are associated with bottom may inhibit vertical mixing (Dyer 1997). POI periods of average downward trending or extreme appears to lose salinity stratification fairly rapidly as low tides, when canal water levels were low com- freshwater discharge declines, whereas at BCNP the pared to groundwater levels. The correlation analysis stratification appears to persist for several weeks or also showed that larger head differences created by longer even with much reduced freshwater inflow. low tide cycles were associated with larger tempera- BCNP was substantially warmer than POI during ture inversions. These are good indicators that tidal both winters, with bottom temperatures remaining pumping (Ganju 2011) was advecting warm ground- above 22°C. BCNP may remain warmer and retain water into the bottom water during low tidal periods stratification longer than POI at least in part because of winter 2009−2010, allowing bottom water to warm BCNP has less mixing due to deeper, narrower independently of the colder surface water. Similar canals, less flushing due to a more restricted tidal bottom-warming, surface-cooling trends can be seen connection and more uneven bathymetry, and higher during the first half of winter 2008−2009, but are groundwater influx associated with greater canal absent during the second half, when salinity stratifi- depth (Stith et al. 2011). cation was no longer present (Fig. 3). This suggests Other heat-related mechanisms undoubtedly oper- that salinity stratification was needed to retain ate at these PTR. Solar radiation is a dominant source groundwater heat that was advected into the bottom of shallow water heat (Ji 2008) and can rapidly warm layer. the entire PTR water column after the passage of cold Salinity stratification appears to be a key mecha- fronts. Solar gain may be enhanced by certain PTR nism maintaining warm water at POI, and freshwater characteristics, such as presence of windbreaks (e.g. inflow is needed to create salinity stratification trees, houses), or clear water, which enhances solar (Fig. 8; Stith et al. 2011). Loss of salinity stratification penetration. The warmer groundwater and substrate was associated with reduction and loss of upstream act as a large heat source that conveys heat upward freshwater discharge in winter 2008−2009. Numeri- via conduction, but conduction is generally minor cal experiments indicated that the observed thermal compared to advective mechanisms such as ground- inversions with warm bottom water temperatures water discharge. Heat may also be generated by would not exist for any significant period of time if benthic microorganisms, but relevant studies are salinity stratification was not represented (Decker et lacking or generally attribute a greater role for abi- al. 2012). Even though surface water temperatures in otic sources of heat (Pamatmat 1982, 2003). Addi- 2009−2010 were colder than previously recorded tional research is needed to clarify the modes and years (Stith et al. 2011), the temperature inversions rates of heat flux, but our results indicate that without 298 Mar Ecol Prog Ser 462: 287–301, 2012

salinity stratification, heat can be rapidly lost from new warm-water sites have been identified in the the entire water column due to convective mixing Florida Manatee Management Plan (FFWCC 2007), (Stith et al. 2011). The combination of salinity stratifi- including minimizing reliance on technological solu- cation and groundwater discharge seems to provide tions, allowing for behavioral adjustments by mana- the warmest bottom-water temperatures. tees to adopt new sites and overcome strong site The favorable warm water characteristics of POI fidelity to old sites, and developing contingency and BCNP suggest that the creation or management plans to respond to temporary or permanent loss of of PTR for manatees could be considered as a man- warm-water habitat. agement alternative to other sources of warm water, Given the potential loss or reduction in carrying provided sufficient salinity stratification and/or capacity of some warm-water sites and the associ- groundwater discharge is present. Canal systems ated threat to population viability, identifying, pro- similar to POI or BCNP exist in other parts of the tecting, and enhancing warm-water sites has been a state, and winter aggregations of manatees have major goal identified for the recovery of manatees been seen at canal systems in Brevard, Miami-Dade, (USFWS 2001, FFWCC 2007). This threat may not be St. Lucie, Palm Beach, and Lee counties (Laist & reduced in the near future due to climate change, as Reynolds 2005a,b, Loomis 2010). Useful temperature greater variability and more extreme weather may measurements have been made at some of these sites occur (Kodra et al. 2011), and large-scale changes in (Loomis 2010, Barlas et al. 2011), but the role of salin- circulation patterns associated with loss of sea ice ity stratification, groundwater discharge, or tidal (Liu et al. 2012) or the Arctic Oscillation (Cohen et al. pumping remains unclear. Aside from the difficulty 2010) are expected to continue to produce sporadic and expense of collecting detailed data to character- cold-air outbreaks from the Artic, irrespective of any ize PTR, other problems arise once the site conditions global trends in mean temperature. Management of and mechanisms are understood. For example, fresh- the Florida manatee, as well as other temperature- water discharge might be unreliable or unavailable sensitive species, most likely will continue to require during winter, preventing the establishment of halo- consideration of warm-water availability issues in the cline-driven PTR. Even if freshwater were available, foreseeable future. Inventorying and monitoring the numerous competing demands for human use might state’s coastal canals, rivers, or small bays that have dictate against water release in winter, especially potential as halocline-driven and/or groundwater- during drought years, when winter water shortages driven PTR might help evaluate the potential of PTR are especially acute. Most coastal canals with fresh- from a statewide manatee perspective. Our results water discharge have salinity-blocking structures suggest that halocline-driven PTR can provide that control freshwater flows and levels to meet warm-water sites in extreme southwestern Florida conflicting goals, including prevention of saltwater and can support large numbers of manatees, but intrusion, flood protection, stormwater drainage, additional research is needed in other areas to deter- conservation of surface water for consumptive uses, mine how these findings might apply to understand- aquifer recharge, and maintenance or restoration of ing and managing warm-water habitats in other healthy aquatic ecosystems (SFWMD 2010). Never- parts of the state. theless, providing freshwater flow during the winter dry season may be beneficial or important to estuar- ine health (Fernald & Purdum 1998, Sklar & Browder Acknowledgements. Funding was provided by the following 1998, McVoy et al. 2011), thus PTR management may USGS programs: FISCHS Project (Future Impacts of Sea Level Rise on Coastal Habitats and Species), Ecosystems be compatible with some ecosystem restoration or Mapping, Greater Everglades Priority Ecosystems Science management goals. (PES), and Critical Ecosystem Studies Initiative (CESI). Other habitat needs and risk criteria have been Funding was also provided by the US Army Corps of Engi- identified for manatee use of canals (Ferrell et al. neers. We thank the following agencies for hydrology or 2004) that are relevant for PTR. These include (1) environmental data: South Florida Water Management Dis- trict, Everglades National Park, Big Cypress National Pre- manatee accessibility (absence of shallow, narrow serve, Ten Thousand Islands National Wildlife Refuge, and restrictions or excessively long distances), (2) avail- US Geological Survey. We thank L. Lefebvre for many help- ability of nearby food resources, (3) exposure to ful discussions and support related to this work. Discussions watercraft traffic, (4) presence of dangerous water with C. Deutsch, A. Spellman, S. Markley, B. Sobzak, and S. Barton were also helpful. Aerial survey and carcass count control structures, and (5) ability to establish and data were provided by the Florida Fish and Wildlife Conser- enforce protection zones at PTR to regulate human vation Commission. Use of trade or product names does not disturbance. Additional considerations for creating imply endorsement by the US Government. Stith et al.: Passive thermal refugia for manatees 299

LITERATURE CITED archical covariate model for detection, availability and abundance of Florida manatees at a warm water aggre- Ackerman BB (1995) Aerial surveys of manatees: a summary gation site. In: Thomson DL, Cooch EG, Conroy MJ (eds) and progress report. In: O’Shea TJ, Ackerman BB, Perci- Modeling demographic processes in marked popula- val HF (eds) Population biology of the Florida manatee tions. Environmental and Ecological Statistics Series, Vol (Trichechus manatus latirostris). Information and Tech- 3. Springer, New York, NY, p 563−578 nology Report 1, National Biological Service, US Dept FWRI (Florida Wildlife Research Institute) (2010) Cold-related Interior, Washington DC, p 13−33 mortality event winter 2009−2010. Accessed July 28, 2011. Anderson MP (2005) Heat as a ground water tracer. Ground http://myfwc.com/research/manatee/ rescue- mortality- Water 43: 951−968 response/mortality-statistics/cold-related-2009-2010/ Barlas ME, Deutsch CJ, de Wit M, Ward-Geiger LI (eds) Ganju NK (2011) A novel approach for direct estimation of (2011) Florida manatee cold-related unusual mortality fresh groundwater discharge to an estuary. Geophys Res event, January−April 2010. Final report to US Fish & Lett 38: L11402, doi:10.1029/2011GL047718 Wildlife Service. Florida Fish and Wildlife Conservation Harvey RG, Loftus WF, Rehage JS, Mazzotti FJ (2011) Commission, St. Petersburg, FL Effects of canals and levees on Everglades ecosystems: Barton SL (2006) The influence of habitat features on selec- circular. WEC304. IFAS Extension, University of Florida, tion and use of a winter refuge by manatees (Trichechus Gainesville, FL. manatus latirostris) in Charlotte Harbor, Florida. MS the- Irvine AB (1983) Manatee metabolism and its influence on sis, University of South Florida, Tampa, FL distribution in Florida. Biol Conserv 25:314−334 Bossart GD, Meisner RA, Rommel SA, Ghim SJ, Jenson AB Ji ZG (2008) Hydrodynamics and water quality: modeling (2003) Pathological features of the Florida manatee cold rivers, lakes, and estuaries. John Wiley & Sons, Hoboken, stress syndrome. Aquat Mamm 29: 9−17 NJ Cohen J, Foster J, Barlow M, Saito K, Jones J (2010) Winter Kodra E, Steinhaeuser K, Ganguly AR (2011) Persisting cold 2009−2010: a case study of an extreme Arctic Oscillation extremes under 21st-century warming scenarios. Geo- event. Geophys Res Lett 37:L17707, doi:10.1029/2010 GL phys Res Lett 38: L08705, doi:10.1029/2011GL047103 044256 Laist DW, Reynolds JE III (2005a) Influence of power plants Constantz JE, Niswonger RG, Stewart AE (2008) Analysis and other warm-water refuges on Florida manatees. Mar of temperature gradients to determine stream ex - Mamm Sci 21:739−765 changes with ground water. In: Rosenberry DO, Laist DW, Reynolds JE III (2005b) Florida manatees, warm- LaBaugh JW (eds) Field techniques for estimating water water refuges, and an uncertain future. Coast Manage fluxes between surface water and ground water. US 33: 279−295 Geological Survey, US Dept Interior, Reston, VA, p Liu J, Curry JA, Wang H, Song M, Horton RM (2012) Impact 117−128 of declining Arctic sea ice on winter snowfall. Proc Natl Decker J, Swain E, Stith B, Langtimm C (2012) Assessing Acad Sci USA 109: 4074−4079 factors affecting the thermal properties of a passive ther- Loomis CP (2010) Evaluation of thermal characteristics of mal refuge using three-dimensional hydrodynamic flow secondary warm-water sites for the Florida manatee. MS and transport modeling. J Waterw Port Coast Ocean Eng thesis, Duke University, Durham, NC (in press) McVoy CW, Said WP, Obeysekera J, VanArman JA, Deutsch CJ, Reid JP, Bonde RK, Easton DE, Kochman HI, Dreshel TW (2011) Landscapes and hydrology of the O’Shea TJ (2003) Seasonal movements, migratory beha - predrainage Everglades. University Press of Florida, vior, and site fidelity of West Indian manatees along the Gainesville, FL Atlantic coast of the United States. Wildl Monogr 151: NCDC (National Climatic Data Center) (2010) 2009/2010 1−77 cold season. Special reports. Accessed 29 July 2011. Dyer KR (1997) Estuaries: a physical introduction, 2nd edn. www1.ncdc.noaa.gov/pub/data/cmb/special-reports/ J. Wiley and Sons, New York, NY 2009-2010-cold-season/2009-2010_Cold_Season.pdf Edwards HH, Pollock KH, Reynolds JE III, Powell JA (2007) Pamatmat MM (1982) Heat production by sediment: ecolog- Estimation of detection probability in manatee aerial sur- ical significance. Science 215: 395−397 veys at a winter aggregation site. J Wildl Manag 71: Pamatmat MM (2003) Heat-flow measurements in aquatic 2052−2060 ecosystems. J Plankton Res 25: 461−464 Fernald EA, Purdum ED (eds) (1998) Water resources atlas Parkinson RW (1998) Decelerating Holocene sea-level rise of Florida. Florida State University Press, Tallahassee, and its influence on southwest Florida coastal evolution: FL a transgressive/regressive stratigraphy. J Sediment Pe - Ferrell D, Congdon R, Scotto L, Mortellaro S, Cook P, trol 59: 960−972 Mezich R (2004) Manatee suitability survey in the Ever- Reich CD (2010) Investigation of submarine groundwater glades region of the Central and Southern Florida discharge along the tidal reach of the Caloosahatchee project CERP Interagency Manatee Task Force. Vero River, southwest Florida. US Geological Survey Open- Beach, FL File Report 2009-1272. US Geological Survey, US Dept FFWCC (Florida Fish and Wildlife Conservation Commis- Interior, Reston, VA, p A1−A52 sion) (2007) Florida manatee management plan. Florida Renken RA, Dixon J, Koehmstedt J, Ishman S and others Fish and Wildlife Conservation Commission, Tallahas- (2005) Impact of anthropogenic development on coastal see, FL ground-water hydrology in southeastern Florida, 1900− Fofonoff P, Millard RC Jr (1983) Algorithms for computation 2000. Circular 1275. US Geological Survey, US Dept of fundamental properties of seawater. UNESCO Tech Interior, Reston, VA Pap Mar Sci, Vol 44. UNESCO, Paris Runge MC, Sanders-Reid CA, Langtimm CA, Fonnesbeck Fonnesbeck CJ, Edwards HH, Reynolds JE III (2009) A hier- CJ (2007) A quantitative threats analysis for the Florida 300 Mar Ecol Prog Ser 462: 287–301, 2012

manatee (Trichechus manatus latirostris). U.S. Geologi- usgs. gov/pdf/ Stith _ et _al_ ENP_ Manatee _Administrative cal Survey Open-File Report 2007-1086, US Geological _ Report .pdf Survey, US Dept Interior, Reston, VA. Availalbe at www. Stith BM, Reid JP, Langtimm CA, Swain ED, Doyle TJ, Slone pwrc.usgs.gov/resshow/manatee/threats DH, Decker JD, Soderqvist LE (2011) Temperature Scott TM, Means GH, Meegan RP, Means RC and others inverted haloclines provide winter warm-water refugia (2004) Springs of Florida. Florida Geological Survey, for manatees in southwest Florida. Estuar Coast 34: Tallahassee, FL 106−119 Sklar FH, Browder JA (1998) Coastal environmental impacts US Army Corps of Engineers and South Florida Water Man- brought about by alterations to freshwater flow in the agement District (2004) Comprehensive Everglades res- Gulf of Mexico. Environ Manag 22: 547−562 toration plan, Picayune Strand restoration (formerly SFWMD (South Florida Water Management District) (2010) Southern Golden Gate Estates ecosystem restoration). Canals in south Florida: a technical support document. Final integrated project implementation report and envi- South Florida Water Management District, West Palm ronmental impact statement. Accessed June 2009. www. Beach, FL evergladesplan .org/ pm/ projects/ docs_30 _sgge _pir_final Stith BM, Slone DH, Reid JP (2006) Review and synthesis .aspx of manatee data in Everglades National Park. USGS USFWS (US Fish and Wildlife Service) (2001) Florida mana- Administrative Report. USGS Florida Integrated Sci- tee recovery plan, (Trichechus manatus latirostris), 3rd ence Center, Gainesville, FL. Available at fl. biology. edn. US Fish and Wildlife Service, Atlanta, GA

Appendix

Fig. A1. Color infrared image (water is black, vegetation is red) showing location of continuous monitoring station (BB- STAT) in Port of the Islands and additional sites where grab sample vertical salinity profiles were taken. Sites BBSTAT, BBNE, and BBSE were in deeper holes with heavier manatee use. Crosses show latitude–longitude graticule Stith et al.: Passive thermal refugia for manatees 301

Fig. A2. Port of the Islands (POI) (a) surface and bottom salinities, (b) SGT5W3 groundwater level and POI tide stage (15 min interval, tide; daily mean, M tide), (c) SGT5W3 groundwater, POI surface and bottom temperatures at 15 min intervals for November 18 to December 28, 2008. (b) Red ellipses show low tide cycles associated with strong cold events. (c) Red ellipses show corresponding periods with bottom-water warming, surface-water cooling

Fig. A3. Port of the Islands vertical salinity profiles for the afternoon of January 8, 2010 at sampling stations shown in Fig. A1

Editorial responsibility: Matthias Seaman, Submitted: August 8, 2011; Accepted: March 23, 2012 Oldendorf/Luhe, Germany Proofs received from author(s): August 7, 2012