As stated by Ernst et al. (1994), “Turtle populations have been declining at an alarming rate in North America and if this trend continues, all North American turtle species will be threatened with extinction in this century.” River turtles have the highest proportion of threatened species when compared with other vertebrate groups, and smooth softshell turtles have been reported as declining worldwide (Johnson 2000; Bodie 2001). The decline of river turtles is often related to human activities such as habitat alteration/degradation (e.g., river channelization) and harvesting (Johnson 2000; Moll and Moll 2000; Trauth et al. 2004; Riedle et al. 2005).

Many large rivers in the United States have been modified for navigation and/or flood control (Koebel 1995), including the Mississippi River. Reported alterations to the river system include the reduction of more lentic habitats, such as oxbows, the loss of many deep holes associated with snags in unchannelized river systems, and the disjunction of low-velocity water associated with side channels from the main river channel by closing structures and wing dikes (Beckett et al. 1983; Barko and Herzog 2003). Many river turtles are adapted to inhabit rivers with variable depths, widths, and water velocities (Moll and Moll 2004), yet modifications often homogenize these conditions. The impact of river modifications on many aquatic organisms is still largely unknown (Plummer 1977a; Koebel 1995), and smooth softshell turtles have been reduced or eliminated from the Illinois River because of channelization (Moll 1980). Further, the ecology of many riverine turtles inhabiting large river systems of the United States is poorly understood and has received little investigation, in part, because of the difficulty of sampling large river systems (Plummer 1977a; Moll and Moll 2000). Most studies conducted on river turtles have been conducted within a small geographic area over a short temporal scale (see Anderson et al. 2002) or have focused on a single life history trait such as reproduction (see Plummer 1977b; Nagle et al. 2003). As a result, little information on the status, ecology, or long-term trends of many species is available (Plummer 1977a; Moll and Moll 2000).

In 1986, the U.S. Congress created the Environmental Management Program (EMP) under the Water Resources Development Act (Public Law 99–662) in response to concerns over long-term ecosystem sustainability of the upper Mississippi River (UMR) for both biological communities and human interests (Lubinski 1999). The EMP includes a biological monitoring program, known as the Long Term Resource Monitoring Program (LTRMP; Jackson et al. 1981), which is one of the largest systemic monitoring programs in North America. The main focus of this program is to identify and understand long-term trends in fishes, vegetation, benthic macroinvertebrates, and water quality within the UMR. In the 1990s, LTRMP biologists began collecting turtle by-catch data, concurrently with fisheries data, to better understand the role of these vertebrates in the unimpounded UMR (Moll 1993).

The 2 species of softshell turtles inhabiting the Middle Mississippi River (MMR) are the midland smooth softshell (Apalone mutica) and spiny softshell (Apalone spinifera). The smooth softshell is an inhabitant of the Ohio, Mississippi, and Missouri Rivers (mainstream and tributary) and ranges from Texas, east to Florida, and north to Minnesota (Conant and Collins 1998). This species is a riverine turtle, inhabiting waters with moderate to fast velocities and has more specialized habitat requirements than other softshell turtles (Williams and Christiansen 1981; Conant and Collins 1998; Nagle et al. 2003). Smooth softshell turtles have been reported to inhabit areas composed of sand substrates, and they feed on insects, crayfish, frogs, snails, and small fishes (Johnson 2000; Minton 2001; Trauth et al. 2004). The species reaches a maximum carapace length (CL) of 356 mm, with females being the largest (Johnson 2000), and reaches sexual maturity near age 6–7 years (female CL = 170–220 mm; male CL = 110–126 mm; Johnson 2000).

The spiny softshell turtle ranges from Louisiana, north to Minnesota, west to Wyoming, and east to New York (Conant and Collins 1998; Minton 2001). Although the spiny softshell is often characterized as a river turtle, it also has been collected from backwater and offshore areas (e.g., ditches, streams, and gravel pits) with mud and sand bars (Johnson 2000; Minton 2001; Trauth et al. 2004). This species feeds on insects, crayfish, and plant material (Minton 2001). Spiny softshell turtles have a maximum CL of 466 mm (Minton 2001); females reach sexual maturity at a CL of 180–200 mm, males at CL of 90–100 mm (Johnson 2000).

Our objectives were to 1) expand the LTRMP monitoring efforts by analyzing turtle by-catch data collected in conjunction with annual LTRMP fisheries sampling, 2) examine softshell population characteristics, 3) assess associations between smooth and spiny softshell turtle abundance (i.e., number of individuals) and physical habitats, and 4) identify environmental variables correlated with variation in abundance.

METHODS

Sampling

Our study was conducted within the MMR between river kilometers (RK) 48 and 129, adjacent to Missouri and Illinois (Fig. 1). The MMR is located between the confluences of the Missouri (near St. Louis, Missouri) and Ohio Rivers (near Cairo, Illinois) and is unimpounded. Turtles were incidentally captured from 1996 to 2001 using fish sampling protocol developed by the LTRMP (Gutreuter et al. 1995). Fishing techniques included baited hoop netting (small and large), mini-fyke netting, fyke netting, daytime electrofishing, and gill netting (see Gutreuter et al. 1995, for gear descriptions and deployment methodology). Sample sites were determined before the sampling season by using a geographic information system to overlay a 50 m × 50 m grid on the study reach. Each sampling period, site locations were randomly chosen for each sampling technique within 5 aquatic areas, including the following: tributary, main channel border, wing dike, open side channel, and closed side channel (Fig. 2). Because tributary physical habitat was limited in our study area, these sites were fixed in the study. Main channel borders were defined as the zone between the margins of the main navigation channel and the nearest shoreline without wing dikes, whereas wing dikes were defined as main channel border with a wing dike as the main physical structure (Gutreuter et al. 1995). Tributary physical habitat was defined as the mouth of a floodplain stream and sampling was conducted 0.8–1.4 km upstream of the confluence with the Mississippi River (Gutreuter et al. 1995). Open side channels had both ends connecting to the main river channel, whereas closed side channels had only one end connecting with the main river channel during normal river elevation (Barko and Herzog 2003).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

RESULTS

Population Structure

Six types of fishing gear captured softshell turtles: daytime electrofishing (DE), fyke netting (F), large hoop netting (HL), small hoop netting (HS), gill netting (G), and mini-fyke netting (MF). In these samples we collected 150 smooth softshell (SM) and 56 spiny softshells (SP) (DE: 6 SM, 0 SP; F: 80 SM, 20 SP; HL: 6 SM, 14 SP; HS: 15 SM, 10 SP; G: 1 SM, 5 SP; MF: 21 SM, 2 SP). Because of unrecorded data, we only report the sex of 137 smooth softshell and 55 spiny softshell turtles. Female smooth and spiny softshells were more abundant than males and were larger for both species (Table 1). For smooth softshells, 85 adults (62% of captures) and 52 subadults (38% of captures) were captured (Table 1; Fig. 3a). For spiny softshell turtles, 48 were adult (87% of captures) and 7 subadult (13% of captures; Table 1; Fig. 3b).

Table 1. Population Structure of Smooth and Spiny Softshell Turtles Captured in the Middle Mississippi River from 1996 to 2001.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Environmental Variables and Abundance

Because of low catch rates from sampling gears, only samples containing one or more turtles (n = 117) were used in the statistical analyses. For spiny softshell turtles, the first 2 PCA axes had eigen values greater than one (1.77 and 1.09, respectively). However, only axis 1 entered and remained in the stepwise regression model. This axis explained 12.9% of the variation in spiny softshell abundance (F_1,31 = 4.57, p = 0.04). The eigen vectors that defined Axis 1 were Secchi transparency (0.59) and velocity (−0.52). Spiny softshell turtles were captured in waters with higher visibility and slower velocities (PC1 mean = 2.69E⁻¹⁶ ± 1.33 SD). For smooth softshell turtles, the first 3 PCA axes had eigen values greater than one (1.39, 1.22, and 1.01, respectively). However, only PCA Axis 2 entered and remained in the stepwise regression model. This axis explained 13% of the variation in smooth softshell abundance (F_1,60 = 9.0, p = 0.004). The eigen vectors that defined Axis 2 were velocity (0.70) and depth of gear deployment (0.58). Smooth softshell turtles were captured in deep waters with faster velocities (PC1 mean = 1.36E⁻¹⁶ ± 1.10 SD). Environmental variables measured at the sampling sites were significantly different between the softshell species, except for temperature and water velocity (Table 2).

Table 2. Mean (± SD), Range (in parentheses), and t-Test Results of Values for Environmental Variables at Sample Sites Where Smooth and Spiny Softshell Turtles Were Captured in the Middle Mississippi River from 1996–2001.

View Fullscreen

DISCUSSION

Our findings suggest that smooth and spiny softshell turtles inhabit different physical habitats of the MMR. The greatest abundance of smooth softshells was collected from open side channels and main channel borders. Both of these habitats were once common in the MMR, are associated with faster water velocities, and have been reported to be used by large river dependants and specialists such as Ohio shrimp (Macrobrachium ohione), goldeye (Hiodon alosoides), and river shiner (Notropis blennius; Barko and Herzog 2003; Barko and Hrabik 2004; Barko et al. 2004a). Substrates in these habitats are often sandy, which is a preferred substrate for smooth softshell turtles (Minton 2001).

Conversely, spiny softshells had the highest abundance in tributaries and closed side channels and the lowest abundance in open side channels and at main channel border wing dikes. We sampled 2 large tributaries that are present in our reach of the MMR, the Big Muddy River (RK 121.8), and the Diversion channel (RK 78.5). Substrates in the tributaries and closed side channels are often composed of mud and sand, which is often selected substrate by the spiny softshell turtle (Graham and Graham 1997; Minton 2001).

Plummer (1977a) and Minton (2001) reported that smooth softshell turtles inhabit clearer, faster waters when compared with other softshell species. Our findings in the MMR were similar for velocity. However, we found that spiny softshells were associated with waters having higher visibility (e.g., Secchi transparency), while smooth softshell turtles were associated with deeper waters. Barko and Herzog (2003) reported that open side channels (habitat of smooth softshells) have faster water velocities and lower visibility (Secchi transparency) than closed side channels (nonflood stages) in the MMR.

Our findings suggest that smooth softshell turtles have successful reproduction in this river reach and the percentage of subadults in the population is similar to the percentage (37%) reported by Plummer (1977a) from the Kansas River. Conversely, spiny softshell turtles appear to have limited reproductive success in this reach based on the high number of adults and relatively small number of subadult individuals captured during our study. For both species, the populations appear to be female biased. The sex ratio for smooth softshells was 1M:1.6F, whereas the sex ratio for spiny softshells was 1M:2.1F. Graham and Graham (1997) reported the sex ratio of spiny softshell turtles captured in Vermont was near 1:1. Reasons for skewed reproductive classes and sex ratios among and within the species could be because of sexing errors, trapping techniques (Gibbons 1983), and/or biological factors (Swannack and Rose 2003). Plummer (1997a) reported capturing more males than females in a Kansas River softshell turtle population, a result opposite ours. Ream and Ream (1966), Gibbons (1983), and Swannack and Rose (2003) reported sex ratios were male biased when baited hoop nets were deployed. Our study used baited hoop nets, in addition to unbaited mini-fyke nets, fyke nets, and gill nets; hence, we do not believe the female biased sex ratios are a result of trapping methods deployed. However, it could be because of gear placement. Most gears were set in deep water with steep banks, preferred female softshell turtle habitat (Plummer 1977a). In addition, female smooth softshell turtles are more mobile and likely had more chances to come into contact with the nets (Plummer 1977a; Galois et al. 2002). Male softshell turtles and hatchlings inhabit shallow waters associated with sandbars (Plummer 1977a). There are not many of these areas in our river reach because of channelization, and this could be affecting nesting (Bodie 2001), male, and hatching softshell survivorship and provides an alternate hypothesis to the perceived female-biased sex ratio and the low number of subadult spiny softshell turtles captured in our study. Fitch and Plummer (1975) reported that nearly all smooth softshell nesting, feeding, and mating occurred on sandbars.

Plummer (1977b) reported that sexing errors in smooth softshell turtles were improbable because of cloacal position in subadults. Therefore, the female-biased sex ratio is likely not the result of sexing errors. Although the reason for the female-biased sex ratio is unknown, we speculate that it is the result of differential mortality among the sexes and age classes (Swannack and Rose 2003) and/or differential habitat use among the sexes (Plummer 1977a; Smith and Iverson 2002).

Temperate floodplain rivers are characterized by a spring flood pulse (Gutreuter et al. 1999). Because of modifications to the MMR, the exact timing of flood events varies annually, which likely influences the nesting of softshell turtles (i.e., May–June; Graham and Graham 1997; Johnson 2000). Both softshell turtles lay eggs on sandbars and sandbanks, (and spiny softshell turtles may also use gravel bars; Johnson 2000). These areas are often associated with island complexes and wing dikes in this river reach. Sandbars associated with island complexes in the MMR often have steep banks because of scouring, which may make them inaccessible to softshell turtles. During normal water gage, river flow is directed to the main channel by wing dike fields and away from offshore areas, allowing sand and gravel bars to be exposed. Wing-dike fields trap sediment, which allows sandbar creation and plant colonization from the adjacent shoreline (Smith and Stucky 1988). Sandbar exposure and length of exposure depends on hydrological events (e.g., drop in river gage; Bacon and Rotella 1998). Sandbars in the MMR are associated with island complexes and wing dikes (at low river gages). Floodpulses that caused river stage to exceed flood stage (Cape Girardeau gage ≥ 32 ft) occurred during the nesting months (e.g., May–June, smooth softshells; May–July, spiny softshells; Johnson 2000) during all years of our study except 1998 and 2000 (USGS). In 1998, river stages did not drop below flood stage until early July, and in 2000 river gage never exceeded flood stage. We hypothesize that smooth softshell turtles nest within island complexes, if available, because this species is a riverine specialist and evolved in systems with unpredictable hydrographs. These sandbars are less likely to be eliminated during flood spates and have lower predation risks when compared with sandbars associated with main channel borders. This could explain the higher percentage of subadult smooth softshell turtles captured, when compared with the number of subadult spiny softshell turtles captured during our study. Conversely, nests constructed in offshore areas are likely unsuccessful because of flood spates and mammalian predation (Renken and Smith 1993). We hypothesize that spiny softshell turtles nest in these areas because they are associated with slower water velocities and inhabit nonriverine habitats such as lakes and reservoirs (Johnson 2000). Studies need to be conducted in the MMR to verify nesting habitats, hatching success, and nest predation rates for these softshell species. At a minimum, we recommend focusing conservation efforts toward creating and maintaining shallow water sandbar habitat with varying amounts of solar radiation in the MMR that is accessible to softshell turtles throughout the nesting and hatching season (e.g., May–August) and less accessible to potential predators. Such efforts will also benefit other species requiring similar breeding sites, such as the Least Tern (Sterna antillarum; Smith and Renken 1991; Szell and Woodrey 2003), false map turtle (Graptemys pseudogeographica; Ernst et al. 1994), common map turtle (Graptemys geographica; Ernst et al. 1994), and pallid sturgeon (Scaphirhynchus albus; D. Herzog, ORWFS, unpubl. data).

In summary, smooth softshell turtles were the most abundant softshell turtles in the MMR. Softshell turtles are considered a game species in Missouri, yet have been on the decline in the Midwest (Johnson 2000). Barko et al. (2004b) reported that mortality rates for incidentally captured smooth and spiny softshell turtles using passive fishing techniques in the Mississippi River were 15% and 36%, respectively. The elimination of harvesting, development of turtle excluder devices or by-catch reduction devices for passive fishing gears, and the length of time passive nets can be safely set needs to be examined (Barko et al. 2004b). We suggest protection of aquatic areas inhabited by softshell turtles (e.g., side channels) and implementation of turtle-specific survey methodologies into the LTRMP and other large river monitoring programs to better understand population dynamics of these species in large river systems.