Patuxent Wildlife Refuge, Laurel, Maryland October 1999

Transcription

1 Patuxent Wildlife Refuge, Laurel, Maryland October 1999 Edited by: Christopher ~ Swarth Jug Bay Wetlands Sanctuary Lothian, Maryland Willem M. Roosenburg Department of Biological Sciences Ohio University Athens, Ohio Erik Kiviat Hudsonia Ltd., Bard College Annandale, New York Bibliomania! Salt Lake City, Utah 'l.."o~

2 Blanding's Turtle Response to Wetland and Upland Habitat Construction Erik Kiviaf.2 Gretchen Stevens 1 Krista L. Mungerl Laura T. Headyl Sven Hoeger3 Peter J. Petokas1 Robert Brauman3 Abstract: The Blandings Turtle is listed as Threatened in New York. Many of the wetlands used by an isolated population in Dutchess County have been altered or destroyed. We designed and monitored the construction of replacement habitats as part of a wetland mitigation project for a school expansion. The replacement project included wetlands, upland nesting areas, and a one-way barrier fence to discourage the turtles from entering the school compound. For three years post-construction, we compared use of constructed and nearby, pre-existing reference wetlands by adult Blandings turtles, and studied nesting success on the constructed nesting habitats. The turtles readily used constructed wetlands in late spring and summe1; but very little in early spring or winter. Turtles nesting on the constructed habitats appeared to select moister soils in a dry year and drier soils in a wet yea?; and moister soils produced larger hatchlings in one dry year. Hatchling production (as percentage of eggs) declined over' three years as vegetation cover increased. Although it is not possible to predict the long-term success of the restoration site based on three years of data, preliminary results suggest that habitats can be constructed for Blandings turtles, but uncertainties of suitability indicate that this should be done only where absolutely necessary. A1 a habitat restoration site in Dutchess County, New York, constructed wetlands and upland nesting areas were completed in May 1997 to replace habitats of the state-threatened Blanding's turtle (Emydoidea blandingit) ost to a school expansion (Kiviat et al. 2000). This paper reports on an additional year of data (1999) for the habitat restoration project. Vegetated wetland sods and underlying organic soil were salvaged and moved m to excavated basins to create 1.4 ha of deep-flooding, shrubby, groundwater-fed wetlands, each surrounded by dry, coarse-textured, sparsely vegetated, upland soils. A barrier fence with one-way "turtle curbs" was constructed to prevent movement of turtles from the restoration site into the school grounds, while allowing movement the other way. Dutchess County lies east of the Hudson River between New York City and Albany. The western two-thirds of the county support at least 12 small populations of Blanding's turtle in complexes of small wetlands near uplands that have gtavelly loam soils derived from glacial outwash. The vegetation of the wetlands used by the turtles includes a prominent shrub component, little canopy cover by trees, and little cover by grass-like plants (Kiviat 1997). There are also organic sediments and fluctuating water levels (0 m to 1.2 m deep). Environmental characteristics, from the landscape scale down to the microhabitat scale, appear to maximize warming in spring (Kiviat 1997). Study Area The environment around the restoration site is rural becoming suburban, and there are two major highways. The study area is between a public school and a state park on land that was farmed until approximately 1950 (Fig. 1). I Hudsonia Ltd., Bard Colkge, RD. Box 5000, Annandale, New York 12504, USA 2 . kiviat@bardedu 3 Creative Habitat, 253 Old Tarrytown Road; Whitt' Plains, New York 10603, USA C)~

3 CONSERVAllON AND EcOLOGY OF TURTLES b State Park Sial. Park ~ Pre-existing WeUands ~ Constructed WeUands 00 Constructed Nesting Areas ~. Nesting Berms 0 Turlie Barrier Drought Refuge Pool / SE.. State Park Athletic Fields I New A1hlelic Fields Expended School Building Local ROOd New:";. J ~ if Parking :::=~==::::::~:::::::- Fig. 1. Map of study area before (a) and after (b) habitat construction. AI, A2. B = constructed wetlands; CS = Corner Swamp; NCP = North Campus Pond; SE = Southeast Swamp; D = donor wetland. Wavy line on right edge of study area denotes woodland border. (Drawn by Kathleen A. Schmidt.) There are approximately 6 ha of pre-existing wetlands suitable for use by Blanding's turtles and 1.4 ha of constructed wetlands built in 1996 and A "donor" wetland (0.7 ha), formerly used by juvenile and nesting female Blanding's turtles, provided much of the soils and vegetation for the constructed wetlands and was filled after translocating these materials. All the wetlands have organic sediments. The two pre-existing wetlands most used by the Blanding's turtles are: 1. "Corner Swamp," 4.0 ha, with a central red maple (Acer rubrum) swamp and peripheral areas with buttonbush (Cephalanthus occidentalis), purple loosestrife (Lythrum salicaria), and a deeper moat; and 2. "Southeast Swamp," 0.6 ha, dominated by purple loosestrife, tussock sedge (Carex stricta), and shrubs of several species including buttonbush. Since construction, purple loosestrife and tall shrubs interspersed with shallow pools have dominated the constructed wetlands. Constructed nesting areas (Fig. 1) were built from the local gravelly glacial outwash soils (Hoosic gravelly loam); there are also four constructed nesting berms about 1 m high with a 20 cm deep layer of sand on top. Portions of the nesting areas have become covered with tall forbs such as white sweet-clover (Melilotus alba) and srar-thistle (Centaurea maculosa). A fence was built between the constructed habitats and the school areas to discourage turtles from entering parking lots and roads. The fence is chain-link with added aluminum sheeting at the base that is sunk 30 cm into the soil. Every 30 m along the fence there is a "gate" which has an earthen ramp on the school side and a 25 cm high curb on the restoration side. Turtles can pass through the gates from school side to restoration side but not the reverse (Kiviat et al. 2000). Spring and summer 1996 were very wet, and most of the pre-existing wetlands retained water all year. In 1997, there was a severe drought, wetlands drew down rapidly in late spring, and by August water sufficient for adult Blanding's turtle use (> 25 cm deep) remained only in constructed wetland AI, the dredged pool in North Campus -Pond (Fig. 1), and in artificial ponds in the park north of Corner Swamp. Spring and summer 1998 were much wetter

4 OF THE Mro-An-ANTIC REGION than in 1997, and water levels in all wetlands remained high into July; water levels fell rapidly in late July and early August, but all wetlands retained sufficient water for adult turtles through August. Spring and summer 1999 were very dry. Methods Wetland Vegetation-We sampled wetland vegetation on stratified random 3 x 3 m plots in and on the banks of the constructed wetlands in August or September Results presented here for the constructed wetlands refer to "fringe" plots (n = 9, mostly outside the wetland boundary, not among translocated sods) and "sod" plots (n = 12, among translocated sods within the wetland boundary). We also sampled vegetation on plots (n = 41) in the three reference wetlands (Corner Swamp, Southeast Swamp, North Campus Pond) most used by adult Blanding's turtles, in August and September In May and July-September 1999 we collected preliminary data on "turtle-centered" plots; these circular plots were 2 m radius and centered on a Blanding's turtle as located by radiotelemetry and visually. Turtles were approached carefully. We believe these locations are not biased because adult Blanding's turtles that are in the water when approached do not normally move away from the observer. Turtle-centered plots within reference and constructed wetlands (n = 43) were compared to random plots within reference and constructed wetlands (n = 62). All vascular plants were identified and cover percentages were estimated visually for each plant species in the various types of plots studied Ṡoilr at Nesting Sites-We collected a single soil sample at approximately 18 cm depth and 45 cm away from each Blanding's turtle nest on 10 July in 1997 (n = 11), 1998 (n = 7), and 1999 (n = 7). Soil samples were analyzed for field moisture content, soil organic matter (SaM) (1997 and 1999 only), and texture in four classes (gravel, coarse sand, fine sand, and silt plus clay) by weight. We also recorded distances from each nest to the turtle fence and the nearest wetland boundary, and estimated total cover by herbs in 1 m radius plots, and herbs, shrubs, and ttees in 10m radius plots centered on the nests. In 1998 and 1999 we collected similar data on random locations paired with nest sites. A random location was constrained within 30 m of the paired nest site and random locations were not selected in obviously unsuitable habitats, i.e., within wetland boundaries or within woods. Water levels were recorded weekly or semi-monthly during the ice-free season on staff gauges in several wetlands. Turtles-;-Trapping was conducted in 1983, 1985, 1986, 1993 and In we trapped Blanding's turtles (principally adults) with hoop nets for three to four weeks beginning 1 May. Turtles were measured (maximum straight-line carapace length) and weighed and numbered plastic tags were epoxied to the posterior ponion of the carapace. Juveniles 75 mm to 150 mm carapace length (CL) were not tagged but were individually marked by filing notches in marginal scutes; juveniles < 75 mm CL were not marked. Each year, we epoxied radio transmitters Gohnson Telemetry or Advanced Telemetry Systems) to the posterior carapace of about 10 to 15 adults, principally females. Locations of radio-tracked turtles were determined with Custom Electronics receivers and handheld Yagi antennas daily in May and June, twice per week in July through September, weekly in October, and semimonthly or monthly the rest of the year. During the nesting season, we radio-tracked and visually followed females until they nested. We immediately covered nests with inverted-box-shaped predator exclusion cages made of 12 mm square mesh hardware cloth with horizontal flanges buried approximately 18 cm below the soil surface. Beginning in early or mid-august we checked nests at least once each morning, measured hatchlings, and released them in densely vegetated shallow water wetland habitats near the nests. In late September we excavated the nests and counted undeveloped eggs, dead advanced embryos or hatchlings, and any remaining live hatchlings. Data Analysis-Statistical and graphical analyses were performed with Statistica versions 5.1 and 5.9 (StatSon 1995). We used Kruskal-Wallis analysis of variance by ranks to compare values among three groups, e.g., years ( ), and Mann-Whitney U tests to compare values between any two groups. For the Mann-Whitney test, we present the exact probability (P) where sample size is less than 25; for larger samples we present probability based on Z adjusted for ties (StatSon 1995). We used the Wilcoxon matched-pairs test to compare soil characteristics at nest sites and random locations. Results Wetland Vegetation-In 1999, vascular plant species richness was highest in constructed wetland plots (fringe + sod), intermediate in reference wetland plots, and lowest in turtle-centered wetland plots. Because turtlecentered plots were larger than random plots (12.6 m2 vs. 9 m2) and species richness is normally correlated with 95

5 plot size, we expected turtle-centered 12 plots to have higher rather than lower richness. Kruskal- Wallis H test results 10 indicated that two plant species had higher percent cover in turtle-centered ~ 8 plots than in random plots: buttonbush ~ (p = 0.017) and common duckweed ~ 6 (Lemna minor) (p = 0.018). Purple.2 loosestrife cover did not differ between ~ 4 random plots and turtle-centered plots Z (p = 0.743). Most of the turtle-centered 2 CONSERVATION AND ECOLOGY OF ThRTI.ES EJ :f!!~~.~~~). MeJe 0 Fem ale plots were sampled during summer and :",; these results may not be representative 0 of spring or of non-drought conditions )- of individuals Turtle observed Population-Numbers have been stable..... Sze dass mm I., (20 to 22) in , the four ears ~Ig..~. Size frequency. distribution of adult Blandmgs turtles for 28 d With. th y d mdividuals handled durmgthree field seasons ( ). Measurements f.. 0 mtensive stu y. I m e stu y 31. d..d al (16 fi al 12 are based on date of first capture. area, m IVI u s em es, males and 3 juveniles) were marked from The size class distribution of 28 adults measured during the peri9d of 1996 to 1999 (based on carapace measurements on the first date of capture) is shown in Fig. 2. Average size of adult males is larger than adult females. Nests monitored on the constructed habitats (AI, A2, and B) produced live hatchlings as follows: 104 hatchlings ftom 11 nests in 1997 (mean = 9.45), 53 hatchlings from seven nests in 1998 (mean = 7.57 for seven nests; mean = 8.83 excluding one nest drowned by high water), and 52 hatchlings from seven nests in 1999 (mean = 7.42). Animal populations normally contain "trap-shy" individuals that are not detected during a mark-recapture study; such individuals may be residents or transients in the study area. We are not able to assess the ll1lffibers of noncaptured individuals in the study population. It is also important to note that juvenile Blanding's turtles are almost universally difficult to detect, between the time the hatchlings enter the wetlands to the time individuals attain a CL of approximately 180 mm. We have captured five juveniles in the study area since 1981; the actual number of juveniles in any year is unknown and is likely to be large due to the high year-to-year survival rates of juveniles expected from the research of Congdon et al. (1993) in Michigan. Nest Sites-In 1996, a cold spring resulted in few Blanding's turtles nesting. Two nests occurred on the edges of the sandy horse racetrack that encircled the donor wetland which was subsequently removed. In 1997, 11 radiotracked females nested on the constructed habitat areas. We found seven nests in 1998 and seven nests in 1999 on the constructed habirat areas. Some of the radio-tracked females in these years nested during storms without being detected. During three nesting seasons ( ), the following proportions of radio-tracked females were gravid, respectively: 11 of 11, 10 of 11, and 11 of 11. Nest location prior to and following restoration is summarized in Table 1. Additional nests may have been located within the study area but went undetected. Nesting on the constructed habitats took place on sparsely vegetated soils of the banks and intervening areas between the constructed wetlands. Few nests were found on the nesting berms, probably because these soils were too dry. Females generally walked back and forth along the fence and mostly nested close to the fence (median distance from fence = 1.6 m, range 0-51 m, interquattile range m for three years combined). One nest in 1998 was found within the flood zone of wetland B and subsequently the eggs were drowned by unseasonably high water. We wanted to determine which characteristics of soils and vegetation were selected by nesting females on the constructed habitats. In July 1998, a wet year, soil moisture was lower at nest sites than at random locations (Wilcoxon p = 0.028) (Fig. 3). In July 1999, a drought year, soil moisture was higher at nest sites than at random locations (p = 0.028). The nesting season was relatively wet and dry, respectively, in these two post-construction years, although we did not sample soils until about 10 July. In 1998, soils at nest sites were lower in fine sand (p = 0.028) and silt + clay (p = 0.028) than at random locations, but gravel was not significantly different. Overall in 1998, soils at nest sites were two-thirds gravel by weight. No patterns were detected in SaM. Mean carapace length (CL) of live hatchlings was correlated with early July soil moisture at the nest in 1997 (Spearman p = 0.016) but not in 1998 or Grand mean CL (all nests for one year) was mm in 1997, for 1998, and for The 1999 grand mean was reduced by two clutches of very small hatchlings, and the cause of their small size was not identified. The clutch means for CL, however, were not significantly different between Q(.;

6 OF me MID-ATLANnC REGION Table 1. Location of nest sites. Site identity given with compass direction followed by wetland code (constructed sites = AI, A2, B, B2; reference wetland = CS; 0/5 = outside of restoration area). Missing values (*) indicate nest not observed in a particular year. Turtle Tag Number 58'; Year and Nest Location SW of Al E ofb2 EofB SW of Al NE of Al 809 racetrack NW of Al NW of Al 81 racetrack SE of Al EofB EofB SW of CS o/s OIS SE of Al near Al SW of Al E of A2 817 SW of Al SE of Al 818 SW of A2 N of A2 819 SE orb E of Al 820 SE of CS SE of CS 1997 and and between 1998 and 1999 (Mann-Whitney U tests, p > 0.05). Hatching production (i.e., number of live hatchlings divided by clutch size) appeared to decline during the three years post-construction (Kruskal-Wallis p = 0.03). However, data for subsequent years do not show a continued downward trend (unpubl. data). Turtle Activities--':'Figure 3 shows rdative use of the different constructed and reference wetlands by adult Blanding's turtles. Of the pre-existing wetlands, Corner Swamp and Southeast Swamp were the most highly used; North Campus Pond was used principally in summer and more in dry periods. The constructed wetlands were used most in 1998, the wet year. A deep pool in wetland Al that hdd water throughout the droughts also was used often. The alternation of dry and wet years may have obscured any increase in use of the constructed wetlands over time. Use of "other" pre-existing wetlands, and occurrence on land, were prominent in the severe drought of During dry periods, the turtles remained in wetlands until water levds dropped below about 25 cm. Turtles then either burrowed into non-flooded wetland sediments, migrated to permanent water (a deep pool in wetland AI, the dredged pool in North Campus Pond, permanent ponds in the park, or a remnant pool in a small stream in the park), or estivated on land, e.g., in shrub thickets or next to down logs. We radio-tracked nine turtles in the winter of , 12 turtles in the winter of , and 12 turtles in the winter of Turtles were radio-located semimonthly or monthly. Most overwintering occurred in Corner Swamp and Southeast Swamp. Several turtles overwintered in a 30 m diameter spring fed pond following the drought of summer and fall In February 1998, there was a very warm spell, and the turtles left the pond and entered a small intermittent woodland pool 160 m west of Corner Swamp. About two weeks later these turtles moved back to Corner Swamp and Southeast Swamp, the commonly used wetlands at the park-school boundary. In November 1999, one female was in constructed wetland A2; this was the first indication of overwintering in a constructed wetland but radio contact with this turtle was lost in December In winter , one turtle overwintered in the same small woodland pool used by several turtles in February 1998, west of Comer Swamp. Springtime activity began in Fcbruary, March, or April depending on the weather each year. q7

7 CONSERVA1l0N AND ECOLOGY OF TuRTLES LAND cs SE NCP OTHER A1 A2 B LAND cs SE NCP OTHER A1 A2 B AND Fig. 3. Use of reference and constructed wetlands by adult Blanding's turtles prior to construction (1996) and following construction ( ). Bar height indicates frequency of turtles observed in a particular location. Asterisks (*) denote wetlands that had no standing water for at least half the month. Reference wetlands = CS, SE, NCp, constructed wetlands = AI, A2, and B; LAND = upland areas. Wetland abbreviations correspond to Figure 1. Discussion In the wetlands, turtles appeared to be selecting microhabitats with greater dominance by buttonbush, lower plant species richness, and more common duckweed. Low water levels in summer 1999 presumably influenced microhabitat selection and may have increased the apparent importance of plants associated with remaining pools. There did not seem to be selection for or against purple loosestrife. These patterns accord with qualitative observations at the study area and other Dutchess County Blanding's turtle sites. Female Blanding's turtles on the constructed habitats seemed to select moister soils for nesting, at least in a dry year, and this selection may favor the development of larger hatchlings. Laboratory studies of other freshwater turtles (e.g., Packard et al. 1980) indicate that moister incubation media produce larger hatchlings with presumably increased fitness. The avoidance of finer-textured soils on the constructed habitats suggests that nesting females are seeking gravelly spots that may transmit and retain solar heat or promote soil drainage. The three-year decline in hatchling production rates might be due to greatly increased cover by tall herbs. A three-year time series, however, is certainly too short to be conclusive. Nest site data such as ours may help in decisions about management of the constructed nesting areas at our study area and elsewhere. However, because of our small sample sizes, we believe further studies are needed. Few studies have tracked turtles through the winter or collected quantitative ~ata on soils and vegetation. Yearround radiotelemetry in our studies confirmed overwintering habitats and is a valuable adjunct to growing season studies. Likewise, study of the habitats of turtles and their nests is valuable. The resulting information allows better evaluation of the habitat construction project and will allow improvement of this or similar projects elsewhere. Three years post-construction is a short period in which to evaluate the responses of a very long-lived animal and its habitat to manipulation. The study population has used the constructed wetlands considerably during the nesting season and summer, but little in early spring and winter. These two periods probably are riskier for Blanding's turtles, so the turtles may stick more closely to their traditional "core" habitats (in our case, Corner Swamp and Southeast Swamp mainly). (Southeast Swamp became suitable Blanding's turtle habitat in 1986, suggesting that the constructed wetlands may eventually be used for overwintering.) The turtles have used the constructed nesting areas, probably because of their partial confinement by the fence. Nonetheless, hatchling production has been similar to production in 9R

8 OF THE Mm- A 'n-annc REGIO! other Blanding's turtle studies (e.g., Butler and Graham 1995 reported 87% overall egg hatcwing success). In the years covered by this report, hatchling success ranged from 79% in the year habitat construction was completed (1997) and decreased to 68% in a wet year when one nest was inundated by high water (1998) and 66% in a dry year (1999). In an extended study ( ) at the same site, hatchling success was 83-90% (unpublished data), comparing even more closely with what Butler and Graham (1995) reported for a population in Massachusetts. Many more years of study will be needed to determine in detail how constructed habitats develop and how habitat construction has affected the turtles. Overall, it appears that the translocation of wetland sods was effective in rapidly establishing a "mature" wetland tall-shrub community that is used by Blanding's turtles. However, because the long-term prognosis of the habitat restoration experiment is uncertain, wetland and nesting area construction should be used to increase habitat for Blanding's turtle rather than to mitigate the planned destruction of wetlands or nesting habitats. Acknowledgments Heidi Bock, Anne Marie Casper, Ashley Curtis, Amy Foster, Douglas Gaugler, Tanessa HartWig, Jason Martin, Stephanie Matteson, Karen Moore, Maribel Pregnall, Michael Rubbo, Samantha Searcy, Jim Siesfeld, John Sullivan, Stacey Thew, Kate Wallen, and many volunteers assisted us. Staff!;If the New York State Department of Environmental Conservation (DEC) (especially Larry Biegel, AI Breisch, Mike Kallaji, and Ted Kerpez), New York State Office of Parks, Recreation, and Historic Preservation (especially Edwina Belding, Jim Gell, Ron Janneck, Tom Lyons, and Larry Salvatore), and staff, consultants, and contractors of the Arlington Central School District were extremely hdpful. Our work was funded by Arlington Central School District, Guinness Water of Life, a Society for Ecological Restoration Project Facilitation Award, U.S. Environmental Protection Agency (via New York State Office of Parks, Recreation and Historic Preservation), and AmeriCorps. Prior Hudsonia studies of this Blanding's turtle system were funded by the New York State Office of Parks, Recreation and Historic Preservation, and the InstitUte of Ecosystem StUdies (1993 Summer Research Fellowship to Erik Kiviat). Field work was conducted under a DEC Endangered/Threatened Species Permit. Habitats were constructed by the Arlington Central School District under a New York State Freshwater Wetlands Permit. This is Bard College Field Station - Hudsonia Contribution 79. Literature Cited Butler, B.O. and T.E. Graham Early post-emergent behavior and habitat selection in hatchling Blanding's turtles, Emydoidea blandingii, in Massachusetts. Chelonian Conservation and Biology 1: Congdon, J.D., A.E. Dunham and R.C. van Loben Sels Delayed sexual maturity and demographics of Blanding's turtles (Emydoidea blandingii): Implications for conservation and management of long-lived organisms. Conservation Biology 7: Kiviat, E Blanding's turtle habitat requirements and implications for conservation in Dutchess County, New York. Pgs In: J. Van Abbema, ed. Proceedings: Conservation, Restoration, and Management of Tortoises and Turtles - an International Conference. New York Turtle and Tortoise Society, New York. Kiviat, E., G. Stevens, R. Brauman, S. Hoeger, P.J. Petokas and G.G. Hollands Restoration of wetland and upland habitat for Blanding's turtle. Chelonian Conservation and Biology 3: Packard, G.C., T.L. Taigen, M.J. Packard and T.J. Boardman Water relations of pliable shelled eggs of common snapping turtles (Chelydra serpentina). Canadian Journal of Zoology 58: StatSoft, Inc Statistica for Windows. StatSoft Inc., Tulsa, Oklahoma. 99