A recent taxonomic revision of the genus Macrochelys resulted in the recognition of 3 extant species (M. apalachicolae, the Apalachicola alligator snapping turtle; M. suwanniensis, the Suwannee alligator snapping turtle, and M. temminckii, the alligator snapping turtle; Thomas et al. 2014). This taxonomic arrangement recognizes the 3 genetic lineages of M. temminckii previously identified by Roman et al. (1999) and Echelle et al. (2010). The majority of the published literature on these turtles is based on populations west of the Florida panhandle (i.e., M. temminckii; Ernst and Lovich 2009). Comparatively little is known about the biology of M. apalachicolae and M. suwanniensis.

Macrochelys suwanniensis is endemic to the Suwannee River drainage in northern Florida and southeastern Georgia (Thomas et al. 2014). Available literature is limited to regional status surveys (Moler 1996; Jensen and Birkhead 2003), analysis of blood chemistry (Chaffin et al. 2008), a description of blood parasites (Telford et al. 2009), an observation of basking behavior (Thomas 2009), and a report of juvenile growth rate (Johnston et al. 2012b). Macrochelys suwanniensis has been protected from commercial harvest since 1972 in Florida (Ewert et al. 2006) and 1992 in Georgia (Jensen and Birkhead 2003). Little commercial harvest occurred in Florida prior to 1972 (Moler 1996). A possession limit of 1 turtle/person was allowed in Florida until July 2009 when all take or possession became prohibited (Florida Wildlife Code, Rule 68A-25.002; Moler 1996; Lau 2009). The unknown level of historical and personal harvest in the Suwannee River drainage and the lack of ecological information led us to obtain information to inform future conservation decisions.

We conducted a study of M. suwanniensis over 8 yrs as part of a broad investigation of the turtle assemblage inhabiting the Santa Fe River (SFR), the major Florida tributary of the Suwannee River (Johnston et al. 2011, 2012a). We focused on this species because of the previously identified genetic divergence of this population (Roman et al. 1999; Echelle et al. 2010) and the deficit of basic ecological knowledge relating to this lineage. In this paper, we describe spatial distribution, size structure, sex ratio, and body sizes of M. suwanniensis in the SFR.

METHODS

Study Site

The SFR originates in swamps near Lake Santa Fe and Lake Alto in northern Florida and is classified as a blackwater stream (Florida Natural Areas Inventory 2010), but it is a heterogeneous system that becomes increasingly spring-fed as it flows along its ~ 115-km westward course to the Suwannee River (Fig. 1; Scott et al. 2004; Nico et al. 2012). Approximately 60 km downstream from its origin (elevation =  42 m), the upper SFR disappears underground into a swallet known as the SFR Sink. The subterranean river then re-emerges ~ 5 km away at a site known as the SFR Rise (elevation  =  10 m). From this point, the lower SFR flows ~ 50 km to its confluence with the Suwannee River (elevation  =  3 m), receiving substantial input from artesian springs in its final 37 km (Scott et al. 2004).

View Figure

• Download as Powerpoint
• Download Figure

The surface area between the upper and lower SFR is a natural limestone land bridge under which the subterranean river flows (Fig. 1). In this area, mesic flatwoods and upland pine forest are the primary habitats, but erosion of surface limestone has created numerous sinkhole lakes (Florida Department of Environmental Protection 2003). These lakes have steeply sloped shorelines, are 40–45 m deep, and are confluent with the subterranean river. At their surfaces, they are isolated from each other, as well as from the upper and lower SFR, except during extreme flood events.

In the upper SFR, tannin-stained water inhibits growth of submerged aquatic macrophytes. Vegetation in this part of the river is therefore limited to patches of emergent and floating plants such as spatterdock (Nuphar advena), duckweed (Lemna sp.), water spangles (Salvinia minima), and water hyacinth (Eichhornia crassipes). Coarse woody debris (partially submerged fallen trees and completely submerged logs) is abundant. Water temperatures fluctuate seasonally (5.0°–31.3°C; Suwannee River Water Management District 2012). The river varies in width (2–40 m) and depth (< 1–3 m) during normal base flow, and water flow is primarily determined by rainfall and surface runoff. During periods of high rainfall, water levels may rise by as much as 6 m (Suwannee River Water Management District 2012). At its most recent extremes in February 1998, September 2004, and June–July 2012, surface overflow from the flooded upper SFR joined the lower SFR (D. Kendrick, pers. comm., June 2005; R. Owen, pers. comm., July 2012). In contrast, the upper SFR may stop flowing and even dry up in long (> 1-km) reaches during extreme droughts such as occurred during June–July 2007 and April–May 2012 (G. Johnston, pers. obs.).

Habitat in the first 13 km of the lower SFR is similar to that of the upper SFR, except that the river is consistently wider (20–30 m) and less thermally variable (11°–31°C; Suwannee River Water Management District 2012). The habitat downstream in the terminal 37 km of the lower SFR is substantially different because of input of clear, thermally stable (22.0°–23.0°C), mineral-rich water from ≥ 45 artesian springs (7 first magnitude, 16 second magnitude, 22 third magnitude; Scott et al. 2004). In this part of the river, greater water clarity allows growth of large patches of submerged aquatic vegetation such as tapegrass (Vallisneria americana), Indian swampweed (Hygrophila polysperma), and hydrilla (Hydrilla verticillata). Scattered patches of E. crassipes are also present. Water temperatures since 1989 have varied seasonally between 15°C and 29°C (Suwannee River Water Management District 2012). Midchannel water depth ranges from 0.5 to 5.0 m during normal base flow. In some reaches, the channel bottom is characterized by submarine limestone ledges (4–5 m deep) extending ≥ 50 m in length. Water level is determined more by groundwater sources than by rainfall runoff. Variation in rainfall has profound effects on water clarity. During periods of high rainfall, the entire lower SFR becomes dark because of large volumes of tannin-stained water flowing downstream from the upper SFR. In contrast, during periods of low rainfall, the lower SFR may be clear because most river water originates from groundwater sources.

Sampling

We sampled turtles throughout the SFR downstream from its confluence with the New River (29.9231°N, 82.4187°W, WGS84), including most spring runs feeding the lower SFR and 5 sinkhole lakes in the land bridge area separating the upper and lower SFR, year-round between June 2004 and October 2011. We used single-funnel nylon hoop-traps baited with fresh cut fish. We used 3 different trap sizes (76-cm diameter, 2.5-cm mesh; 91-cm diameter, 6.4-cm mesh; 122-cm diameter, 6.4-cm mesh) to facilitate placement of traps in sites of varied water depths and to increase our capacity to detect smaller size classes. We placed traps immediately upstream of snags, submerged logs, undercut banks, and outer bends of the river, with a minimum of 50 m between traps. During each trap session, we set 8–20 traps (1 2.5-cm mesh trap for every 3 6.4-cm mesh traps) during late afternoon and checked them the following morning. Each trap set overnight constituted 1 trap-night (TN). Trap captures were supplemented by opportunistic hand-captures.

Each captured turtle was measured for straight midline carapace length (CL), curved carapace length (CCL), maximum carapace width (CW), maximum head width (HW), plastron length (PL), and precloacal tail length (PTL) to the nearest 1 mm using aluminum tree calipers (Haglöf®, Långsele, Sweden; CL, CW, HW, PL) or a flexible nylon measuring tape (CCL, PTL). Turtles < 5 kg were weighed to the nearest 1 g using a portable digital scale (Ohaus®, Pine Brook, NJ). Turtles between 5 kg and 50 kg were weighed to the nearest 10 g using spring scales (Pesola®, Baar, Switzerland). Turtles > 50 kg were weighed to the nearest 500 g using a heavy-duty mechanical hanging scale (Rubbermaid®, Huntersville, NC). We marked each turtle individually by drilling holes in the marginal scutes and peripheral bones using a standard numbering system (Cagle 1939). All turtles > 130 mm CL were also marked by injecting a passive integrated transponder (Biomark®, Boise, ID) into the ventrolateral tail muscle (Trauth et al. 1998). We considered individuals with a CL > 330 mm and a PTL < 115 mm as adult females and individuals with a CL > 370 mm and a PTL > 115 mm as adult males (Dobie 1971). We considered all other individuals immature.

Statistical Analysis

Capture data were categorized according to 4 distinct areas: 1) upper SFR, 2) lower SFR, 3) spring runs feeding the lower SFR, and 4) sinkhole lakes in the land bridge between the upper and lower SFR (Fig. 1). To compare capture per unit effort (CPUE) among areas, we calculated the number of turtles captured per trap-night (TTN) for each trap session and then compared TTN per session among areas using a Kruskal–Wallis 1-way analysis of variance on ranks followed by multiple pairwise comparisons using the Holm–Šídák method (Holm 1979). We used a z-test to compare relative proportions of each demographic group (immature, adult female, adult male) captured in traps with those captured by hand and to compare relative proportions of immature individuals between the upper and lower SFR. We used a t-test to compare CL of trap-captured and hand-captured individuals from each demographic group. We tested whether the sex ratio of adults differed from 1:1 using a chi-square analysis. For each sex, we compared CL between upper and lower SFR with a Mann-Whitney rank sum test. To look for differences in relative mass between upper and lower SFR turtles, we performed a separate analysis of covariance (ANCOVA) for each sex using CL as the covariate and log₁₀-transformed all data. We used the Mann-Whitney rank sum test to compare female and male morphometric measurements and body masses. To quantify sexual size dimorphism in CL, CCL, CW, HW, PL, PTL, and mass, we calculated the sexual dimorphism index (SDI) advocated by Lovich and Gibbons (1992; mean size of larger sex divided by mean size of smaller sex, with the result arbitrarily defined as negative [+ 1] when males are the larger sex and positive [−1] in the converse case). To examine variation in relative CCL, CW, HW, PL, and mass between sexes, we performed an ANCOVA for each parameter using CL as the covariate. We transformed data to log₁₀ values for comparison of mass. Statistical significance was accepted at α ≤ 0.05. Means are followed by ± 1 standard deviation. All analyses were performed in SigmaPlot v12.3 or SAS v9.2.

RESULTS

We captured 109 individual M. suwanniensis throughout the SFR drainage. We captured 84 individuals in traps (48 in upper SFR, 35 in lower SFR, 4 in spring runs, and 1 in a sinkhole lake). All individuals captured in spring runs were also captured in the lower SFR. CPUE was significantly different throughout the SFR drainage (F_3,168  =  16.56, p < 0.001; upper SFR mean TTN per session  =  0.181, 42 sessions; lower SFR mean TTN per session  =  0.078, 72 sessions; spring runs mean TTN per session  =  0.006, 30 sessions; sinkhole lakes mean TTN per session  =  0.004, 28 sessions). In multiple pairwise comparisons of CPUE between areas, the only 2 areas that did not differ significantly from each other were sinkhole lakes and spring runs. Turtles in small mesh (2.5-cm) traps ranged between 51 and 575 mm CL (26 captures) and those in large mesh (6.4-cm) traps ranged between 129 and 623 mm CL (79 captures).

We captured an additional 25 individuals by hand while snorkeling in the lower SFR. Turtles captured by hand were not significantly different from turtles captured in traps in the lower SFR in the proportion of immature individuals (z  =  1.046, p  =  0.30), proportion of adult females (z  =  −0.267, p  =  0.79), proportion of adult males (z  =  0.917, p  =  0.36), CL of immature individuals (t  =  0.614, p  =  0.55), CL of adult females (t  =  0.810, p  =  0.43), or CL of adult males (t  =  −0.479, p  =  0.64).

Comparison Between Upper and Lower SFR

The proportion of immature turtles from the upper and lower SFR did not differ significantly (z  =  −0.021, p  =  0.98). The adult sex ratio was significantly female-biased (1M:2F) in the upper SFR (χ²  =  4.00, p  =  0.046) but did not differ significantly from 1:1 in the lower SFR (χ²  =  0.087, p  =  0.77). Adult females were significantly larger (CL) in the lower SFR (mean  =  435.8 ± 35.6 mm, range  =  336–492 mm, n  =  24) than in the upper SFR (mean  =  407.0 ± 34.7 mm, range  =  329–449 mm, n  = 24; U  =  146.5, p  =  0.004; Fig. 2). Adult females were proportionately heavier in the upper SFR (Fig. 3). Adult male CL did not differ significantly between the lower SFR (mean  =  541.0 ± 59.9 mm, range  =  447–623 mm, n  =  22) and the upper SFR (mean  =  523.4 ± 51.0 mm, range  =  443–598 mm, n  =  12; U  =  107.0, p  =  0.38), although all males > 600 mm CL (n  =  6) were captured in the lower SFR (Fig. 2). No difference was evident in relative mass between males from the lower and upper SFR (Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

SFR Population

Our total sample consisted of 26 (24%) immature individuals, 48 (44%) adult females, and 35 (32%) adult males (Fig. 4; Table 1). The smallest individual (CL  =  51 mm, CCL  =  51 mm, CW  =  48 mm, HW  =  20 mm, PL  =  39 mm, PTL  =  11 mm, mass  = 36 g) was captured in a trap 28 October 2006 and exhibited 1 growth annulus on each plastral scute. The adult sex ratio did not differ significantly from 1:1 (χ²  =  2.04, p  =  0.15). Adult males were significantly larger than adult females for every measured morphological variable (Table 1). SDIs were −0.25 (CL), −0.25 (CCL), −0.23 (CW), −0.25 (HW), −0.20 (PL), −0.96 (PTL), and −0.98 (mass). Females and males did not differ significantly in relative CCL, CW, HW, PL, or mass.

View Figure

• Download as Powerpoint
• Download Figure

Table 1. Summary of morphometric measurements^a and mass of all Macrochelys suwanniensis captured in the Santa Fe River drainage, the main tributary of the Suwannee River, Florida (2004–2011). Data presented as mean ± 1 SD, minimum–maximum. The p-values from Mann-Whitney rank sum tests comparing females and males are all < 0.001.

View Fullscreen

DISCUSSION

Prior to our study, Suwannee alligator snapping turtles were known to occur throughout the SFR and associated spring runs (Ewert et al. 2006). To this we add the first documentation of occurrence in a sinkhole lake between the upper and lower SFR. However, M. suwanniensis appears to be relatively uncommon in sinkhole lakes. Lower abundance in sinkholes may be related to its dispersal ability. Dispersal via the subterranean river is unlikely given its role as a barrier to movement by several fishes (Hellier 1967) and overland dispersal is unlikely given the strongly aquatic habits of alligator snapping turtles (Harrel et al. 1996; Riedle et al. 2006; Howey and Dinkelacker 2009). Thus, dispersal to and from sinkholes may occur only during rare extreme flood events. The observation of 1 adult M. suwanniensis (unmarked, sex unknown) walking in shallow (< 20 cm) water flowing over the land bridge 1.5 km from the SFR Rise and > 100 m from the nearest sinkhole lake during a period of extreme flooding 17 July 2012 (R. Owen, pers. comm., July 2012) is consistent with this hypothesis. None of the marked turtles in our study was observed to have moved between the upper and lower sections of the SFR.

Our samples from the upper and lower SFR exhibited different sex ratios. The female-biased sex ratio in the upper SFR may be an artifact of small sample size or sampling bias, although similar sampling bias presumably occurred in the lower SFR. Alternatively, past harvest by humans in the upper SFR could have disproportionately reduced the number of individuals of the larger sex (males). Howey and Dinkelacker (2013) suggested that harvest may have been responsible for a similar female-biased sex ratio in a M. temminckii population in Arkansas. However, sex ratios that did not differ from 1:1 have been reported from previously harvested populations in Arkansas (Trauth et al. 1998), Georgia (Jensen and Birkhead 2003), Louisiana (Boundy and Kennedy 2006), and Oklahoma (Riedle et al. 2008), and female-biased sex ratios may occur naturally in some populations. In the upper SFR, the sex ratio may be related to the width of the river and social behavior of adult males. The upper SFR is narrow and space along this linear corridor may be limiting for adult males, which we have observed engaging in agonistic interactions outside of traps in the presence of bait. Large adult males were also observed displacing smaller males during a telemetry study in the Suwannee River (T. Thomas, unpubl. data, 2011). Folt and Godwin (2013) observed female-biased sex ratios in their survey of M. temminckii in south Alabama but were unable to attribute a cause. We are also uncertain about the cause of our observed sex ratios in the SFR.

Macrochelys suwanniensis from the upper and lower SFR also exhibited differences in body size. All males > 600 mm CL were captured in the lower SFR. Adult females were larger (CL) in the lower SFR but proportionately heavier in the upper SFR. Differences in CL could have been caused by past harvest of larger individuals in the upper SFR. This pattern is consistent with previous descriptions of commercially harvested M. temminckii populations (Shipman and Riedle 1994; Trauth et al. 1998; Riedle et al. 2008; Howey and Dinkelacker 2013). Alternatively, body-size differences could be a natural result of the different environments in the 2 river reaches. Perhaps the larger CL of adults in the lower SFR is a consequence of a greater food supply in the more productive, spring-influenced habitat compared with the less productive blackwater habitat of the upper SFR (Mattson et al. 1995; Whitney et al. 2004). The smaller CL and proportionately greater mass of females in the upper SFR may reflect a decrease in shell growth and reallocation of food resources toward reproduction. Future studies of diet, growth, and reproduction in the upper and lower SFR and other replicated blackwater and spring-fed habitats are needed to test our hypothesis about how habitat influences demographics of M. suwanniensis and other Macrochelys species.

The largest turtle in our study weighed 54.4 kg. The largest known wild-captured M. suwanniensis was a 57.1-kg male captured in the Suwannee River (T. Thomas, unpubl. data, 2012). Neither of these turtles weighed as much as the record wild-captured M. apalachicolae (143.3 kg) or M. temminckii (120.2 kg) reported by Pritchard (2006).

In addition to large body size, Macrochelys species exhibit a high degree of sexual size dimorphism among species in which males are the larger sex (Gibbons and Lovich 1990; Ceballos et al. 2013). Using mean CL as a measure of body size, the SDI reported for M. apalachicolae is −0.38 (Teare 2010), and values for M. temminckii range from −0.15 to −0.24 (Dobie 1971; Sloan et al. 1996; Boundy and Kennedy 2006; Bogosian 2010; Folt and Godwin 2013; Howey and Dinkelacker 2013). Thus, male Macrochelys are on average 15–38% larger than females. Our SDI value (−0.25) for M. suwanniensis falls within this range.

Comparisons of relative proportions of adults and juveniles in 5 populations of M. temminckii have yielded insights into the impacts of harvesting these long-lived turtles. The number of adults in a heavily harvested population in Louisiana was 1.1 times the number of juveniles (Boundy and Kennedy 2006). Adults in Arkansas populations (0.72:1–1.2:1; Trauth et al. 1998; Howey and Dinkelacker 2013) and in an Oklahoma population (0.42:1–0.97:1; Riedle et al. 2008; East et al. 2013) were approximately equal to or lower than the number of juveniles. Adults in a Missouri population significantly decreased between 1993–1994 (5.2:1) and 2009 (1.2:1; Lescher et al. 2013). Illegal harvest was listed among the causes of the decline. The ratio for the SFR population was 2.8:1 (upper 2.7:1, lower 3.3:1). Assuming that turtles captured in these studies were proportional to their actual abundance, then populations that had 3–5 times the number of adults may have been similar to natural, unharvested populations. Those with fewer or equal numbers of adults compared with juveniles may have been harvested. These comparisons support our view that M. suwanniensis in the SFR has not been as heavily harvested as other Macrochelys populations.

Several additional factors suggest a healthy M. suwanniensis population in the SFR drainage. The presence of juveniles of various sizes indicates ongoing recruitment and the occurrence of large adults of both sexes (especially males > 600 mm CL) suggests little widespread effect of past harvest. However, we suggest that M. suwanniensis in the SFR still merits conservation concern. All size classes are vulnerable to harvest or collection for illegal pet trade. In addition, blood-mercury levels in M. suwanniensis (0.603 ppm) from the SFR exceed the Florida health advisory level for consumption of fish (0.5 ppm; Chaffin et al. 2008). Nothing is known about how elevated blood mercury affects Macrochelys, but mercury in common snapping turtles (Chelydra serpentina) can be transferred maternally to eggs and negatively affects hatching success (Hopkins et al. 2013).

Although our study and limited previous studies provide insight into the ecology of M. suwanniensis in the SFR, critical information about growth, age-specific survivorship, reproduction, and population dynamics is still lacking. Because there are no available historical data on population structure or body sizes for this population, our study provides a baseline to which future studies can be compared. We must also acknowledge the geographic limitations of this study. The SFR comprises only a portion of the Suwannee River drainage, so without similar sampling in the Suwannee River and its other tributaries, conservation status of this species remains unclear. Previous sampling in the Suwannee River yielded 5 individuals (19 TN) in Florida (Moler 1996) and 0 individuals (53 TN) in Georgia (Jensen and Birkhead 2003). Hence, the results of a mark–recapture study presently being conducted in the Suwannee River by personnel from the Florida Fish and Wildlife Conservation Commission will be essential for assessing the overall status of M. suwanniensis.