Experimental Assessment of the Impact of Agricultural Machinery on Wood Turtles (Glyptemys insculpta)
Abstract
The wood turtle (Glyptemys insculpta) is a threatened species that uses agricultural fields, in part, to nest, bask, and feed. This use of anthropogenic habitats has resulted in the injury and death of individuals during crop harvest. Turtles have been exposed to hay harvests for over a century but the increased use of rotary disc mowers may have increased mortality rates to unsustainable levels. Mitigation measures have focused on raising blade height to minimize contact with turtles but these measures are based on the dimensions of the largest turtles and the effects of agricultural machinery on subadults, juveniles, and hatchlings have not been quantified. We sought to quantify how different agricultural machinery affects wood turtles at multiple life stages through experimental trials using 2 mower types. Cantaloupe (Cucumis melo) halves were used as turtle proxies to avoid actual harm to turtles. We found that rotary disc mowers resulted in a significantly greater injury to adult (i.e., female, male, and subadult) and young (i.e., juvenile and hatchling age class) proxies at low grass heights (10–20 cm) compared with sickle cutter bar mowers. Young proxies were found to be essentially unaffected by the sickle mower at all grass heights, but they were highly affected by the rotary disc mower at grass heights from 10 to 18 cm. If blade heights were to be raised to reduce the risk to wood turtles, a minimum blade height of 17 cm would be needed to obtain > 50% of adults remaining undamaged during harvest and a minimum height of 15 cm would be needed for young turtles.
The wood turtle (Glyptemys insculpta) is a terrestrial freshwater species that is long-lived (> 40 yrs) and reaches sexual maturity at 12–14 yrs old in northern populations (Ernst 2001; Walde et al. 2003; Marchand et al. 2018). Female wood turtles nest once per year and the likelihood of hatchlings reaching sexual maturity is low, especially when they face multiple threats (Powell 1967; Harding and Bloomer 1979; Schneider et al. 2018). Throughout their range, the wood turtle is considered at risk (Committee on the Status of Endangered Wildlife in Canada [COSEWIC] 2007; van Dijk and Harding 2011; US Fish and Wildlife Service 2015). This status is primarily due to threats such as poaching, road mortality, residential development, destruction of nests by all-terrain vehicles, increased predation, and agricultural machinery (e.g., Harding and Bloomer 1979; Garber and Burger 1995; Daigle and Jutras 2005; Saumure et al. 2007; van Dijk and Harding 2011). Agricultural machinery has become a substantial threat because wood turtles use agricultural fields during the summer months to bask, feed, and nest (e.g., Compton et al. 2002; Saumure et al. 2007; Tingley et al. 2009; McCoard et al. 2018). Wood turtles can be hit by agricultural machinery during crop harvest, causing injury or death; for example, Saumure and Bider (1998) found that carapace injuries were twice as common in agricultural landscapes as in forested landscapes. In Québec, Saumure et al. (2007) recorded that 20% of their study population died as a result of agricultural machinery within 2 yrs and 27% were injured. They noted that adult survivorship declined by 10%–13% and juveniles by 18% in a 2-yr period. In Nova Scotia, Tingley et al. (2009) recorded 5 turtle mortalities during the first harvest and that 14 of 27 turtles tagged had been within the field at least once throughout the active harvest season. Low adult survivorship is a common concern for turtle conservation because they are strongly K-selected species with little potential to recover populations quickly (e.g., Brooks et al. 1991; Congdon et al. 1993, 1994; Keevil et al. 2018). For example, in Ontario, river otters (Lontra canadensis) killed adult snapping turtles (Chelydra serpentina), decreasing the known population from 47 to 16 turtles within 3 yrs (Brooks et al. 1991) and 23 yrs later the population size had not recovered (Keevil et al. 2018). Thus, conservation actions that increase juvenile recruitment may be needed to achieve viability (Congdon et al. 1993; Mullin 2019).
Hay harvest has occurred in parts of wood turtle range for hundreds of years and likely has affected individuals and populations, but the threat is now considered more significant with continually improving machinery (Saumure et al. 2007). Farmers aim to maximize yield and minimize cost and, within the past 50 yrs, there has been a shift from using sickle cutter bar mowers to rotary disc mowers (Rider et al. 1993; Humbert et al. 2009). Rotary disc mowers are preferred over sickle bars because they can cut on uneven ground, run at a faster speed, and are more durable (Pogue et al. 1996) because the blade is tilted on an angle with no blade guards. In comparison, sickle bar blades cut parallel to the ground at lower speeds (Miller and Rotz 1995). Rotary disc mowers can cause a higher injury rate on adult turtles compared with sickle cutter bar mowers (Erb and Jones 2011). This is likely because the multiple, small rotary discs cut downward into each depression, whereas the rigid sickle blades miss some turtles.
Managers require information on strategies to reduce the mortality of wood turtles during hay harvest. Many farmers cut their hay at a blade height of 5 cm. Consequently, it has been recommended to raise the blade height of the mower to 10 or 15 cm to reduce injury to wood turtles using agricultural fields (Saumure et al. 2007; Erb and Jones 2011; Wisconsin Department of Natural Resources 2015). These recommendations are based on the tallest carapace height of a wood turtle in a population. Erb and Jones (2011) tested the possible effects of different mower types (sicklebar mower, rotary mower, mulching head mower, and flail mower) on adult turtles with raised blade heights (10, 15, and 19 cm). They found that there was no injury to proxies using the sicklebar mower but the rotary and mulching head mower caused high levels of injury at a blade height of 10 cm. The rotary mower caused 50% mortality when the blade was set at 10 cm, but they also found that 38% were killed by tires. They noted that their sample size may have been too small to detect trends of injury due to the blade and the tires may have diluted the effect on the proxies. There is, however, no information as to whether different mower types also affect smaller, younger turtles. Smaller turtles (i.e., juveniles and hatchlings) could be severely injured by the blade or thrown as a result of the suction of the mower (Erb and Jones 2011). Juveniles and hatchlings use hay and corn fields during the summer and could be at equal risk of being injured during harvest (Tuttle and Carroll 2005; Castellano et al. 2008). If all age classes of wood turtles are at high risk during hay harvest, then future recovery of a population becomes increasingly unlikely. One unknown factor is whether vegetation can protect a turtle during a harvest.
The goal of our study was to refine existing conservation recommendations for wood turtle populations exposed to agricultural mowers. First, we investigated whether injury to wood turtles during haying operations was dependent on one or more of the following variables: size class, mower type, grass height, and location above or below thatch. Second, we sought to identify a blade height that could reduce the mortality of wood turtles of all age classes. Last, we provided management strategies that consider landowner limitations. Our study used two different mower types (i.e., sickle cutter bar and rotary disc mower) with blade heights ranging from 5 to 15 cm across varying simulated wood turtle body sizes to observe the probability of injury.
MATERIALS AND METHODS
We conducted our study in July and August of 2018 near Fredericton, New Brunswick (Canada), at 3 different study sites containing early successional hayfields. The fields were a mixture of timothy (Phleum pratense), bedstraw (Galium aparine), brome grass (Bromus spp.), and orchard grass (Dactylis glomerata), and are the common field type in much of the intervale landscapes in the Maritimes region. The fields were relatively flat with mounds (∼ 2.5 cm) every few meters. The fields are used as sheep pasture, for the production of hay, and/or maintained as early successional habitat.
We attempted to estimate possible injury of wood turtles as it relates to the following variables: 1) mower types, 2) blade heights, 3) vertical location of the turtle in the field (above or below thatch), and 4) varying subject sizes. Live wood turtles were not used to avoid harm to this imperiled species. We used cantaloupe (Cucumis melo) halves as a proxy for turtles because we could manipulate the size and weight to represent a wood turtle. In addition, the outer casing of unripe cantaloupes has a hard exterior, similar to turtle shells. To characterize the population, we separated the turtle proxies into 5 size classes: adult male, adult female, subadult, juvenile, and hatchling (Table 1). Adults were not classified by age because they reach sexual maturity at a certain size, not a certain age (Harding and Bloomer 1979). All other wood turtles were classified by age by counting the annuli on their scutes. Adults were defined as ≥ 16.4 cm straight carapace length, subadults as ∼ 7–11 yrs, juveniles as ∼ 3–6 yrs, and hatchlings as ≤ 2 yrs old (Farrell and Graham 1991; Mullin 2019). We separated each class based on the average size and weight recorded for a nearby wood turtle population (S.D.W., unpubl. data, 2020) and used a minimum of 5 replicates in each treatment (minimum n = 300; Table 1). To facilitate relocation of proxies after harvest, we applied food coloring (blue, red, yellow, green, and purple) to each size class. Two mower types were used to compare injury, with 3 different models. For the sickle cutter bar mower (henceforth, sickle mower), we used an International Model 1190. For the rotary disc mower (henceforth, disc mower), we used a New Idea 5209 and a Kubota DM1022. We used two tractor types: a Ford 4000 and a Case International 595. The rear tires of the Ford 4000 measured 37.8 cm in width and the total width of the tractor was 171.5 cm. The Case International 595 rear tires were 37.8 cm wide and the total width of the tractor was 190.5 cm.
We measured vegetation height before and after harvest to account for topography, tractor speed, and mower differences. Before each hay harvest, we laid cantaloupe halves of the same size class in a row within an area that a tractor could pass over in a single swath to allow for ease of injury assessment. We placed proxies above or below the thatch (undecomposed organic matter at the soil surface). After harvest, we assessed the injury of each proxy as undamaged or damaged; damage would include crushed or sliced sections of any single proxy. Assessment of the proxy's injuries was completed immediately after harvest of the field. We completed 2–3 replicates for each tractor type and blade height within uncut areas of the field. We did not separate mortality from injury on damaged proxies because turtles can sometimes survive with partially damaged shells (e.g., Saumure and Bider 1998; Saumure et al. 2007; Galois and Ouellet 2007; Bennett and Litzgus 2014).
Statistical Analysis
We used logistic regression to predict the probability of injury (0 = uninjured, 1 = injured) based on increasing grass heights, location of proxy, and size class. Proxies that were damaged by tires were removed from the analysis to obtain an isolated assessment of the damage by the blade. Grass height varied at each site postharvest, even though the blade height was the same. Grass heights varied, so we used this measurement (instead of blade height) to infer injury. We merged size classes that did not have a significant difference of injury between grass heights (p > 0.05), based on a Students' t-test. We compared models using Akaike's Information Criterion (AIC; Akaike 1974) to identify the best candidate model(s). We conducted all statistical analyses using the Program R Studio version 1.1.463 (R Core Team 2018).
RESULTS
We had a sample size of 398 proxies (cantaloupe halves) among 3 sites: 181 for the sickle mower and 217 for the disc mower. Blade heights of 5, 10, and 15 cm were used for the disc mower, which resulted in average postharvest grass heights ranging from 10 to 25 cm (range = 8–40 cm). We used grass height as a proxy for blade height because of the variation in grass height after harvest. Grass height was the height of grass left standing after the mower had harvested. Sickle mower had blade heights set to 5 and 10 cm, resulting in grass heights ranging from 10 to 15 cm. The average area that the tractor tires covered was 22% of each field. The logistic regression populated one top model for each mower type (Tables 2 and 3). For both models, grass height, placement of proxy (above or below thatch), and size class were all important in predicting injury. Female, male, and subadult classes had similarly low variability of injury among size classes (p > 0.05) and were merged to create an “adult” size class. Juvenile and hatchling were combined to create a “young” size class. Merged classes resulted in 4 treatments for each mower type: 1) adults above thatch, 2) adults below thatch, 3) young above thatch, and 4) young below thatch.
Disc mowers had a negative impact on the proxies. At the lowest cutting height (10 cm), 99.1% (95% confidence interval [CI] = 96.7%–99.7%) of adult proxies and 95.5% (95% CI, 88%–98.4%) of young proxies were damaged (Fig. 1). Proxies located above thatch had a higher damage rate on both adult and young proxies, compared with damage on proxies located below thatch (Fig. 1). The threshold for most undamaged proxies for adults was at a grass height of 23 cm for above-thatch and 19.7 cm for below-thatch trials. The threshold for young was at 18.5 cm for above-thatch and 15.1 cm for below-thatch trials. At the tallest grass height (25 cm), the younger size class had a 91% (95% CI = 80%–96.3%) probability of being uninjured by the blade, compared with the adults that had a 66% (95% CI = 49%–80%) probability.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1404.1
For the sickle mower trials, at the lowest grass height (10 cm), 58% (95% CI = 37%–77%) of adults and 21% (95% CI = 43%–98.7%) of young were damaged above thatch (Fig. 2). Adults below thatch had a 35% (95% CI = 21%–52%) probability of being damaged at 10 cm and young had a 10% (95% CI = 4%–22%) probability. When the grass height increased to 15 cm, damage to adults was reduced to 24% (95% CI = 11%–45%) and young to 6% above thatch (95% CI = 2%–17%; Fig. 2). The threshold for adults above thatch was at a grass height of 11.2 cm. All other treatments had more undamaged proxies than damaged for grass heights ≥ 10 cm.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1404.1
At a grass height of 15 cm, adult proxies above thatch had a 95% (95% CI = 87%–98.0%) probability of being damaged if cut by a disc mower, whereas for the sickle mower, it was only 24% (95% CI = 11%–45%). For young proxies, 78% (95% CI = 61%–89%) had a likelihood of being injured at a grass height of 15 cm for rotary but for sickle they had a damaged probability of 6% (95% CI = 2%–17%). The rotary blade trials revealed that a taller grass height is needed to achieve the threshold compared with sickle mower for all treatments. After harvest, 5% of hatchlings could not be located if a disc mower had been used, whereas for sickle mower, 19% of the hatchlings and one subadult could not be located.
There is also a considerable difference of injury rate between the two size classes for each mower type. When the disc mower blade was set to the lowest height (10 cm grass height), adult and young proxies above thatch had a relatively similar injury rate (adults 99.1%, young 95.6%), whereas for the sickle mower, the adult size classes were twice as likely to be damaged at the lowest grass height compared with young proxies.
DISCUSSION
We found that all wood turtle sizes are similarly affected by a disc mower if the blade is set to heights between 5 and 10 cm. However, raising the blade higher than 5 cm on a sickle mower can reduce injury significantly for all size classes. Location of the turtle, the equipment used, the blade height, and the size of the turtle can all influence the injury rate of individuals. Our results support other studies that suggest that disc mowers injure more turtles than sickle mowers (Liczner 1999; Humbert et al. 2009; Erb and Jones 2011). Proxies that we set above the thatch showed a higher injury rate than those set below the thatch, which was likely due to some protection for the proxies under vegetation. The disc mower caused a high injury rate at multiple blade heights; at the tallest grass height, > 30% of proxies were still damaged, whereas the sickle mower had < 50% undamaged proxies when the grass height was 11 cm. Erb and Jones (2011) conducted a similar experiment with adult proxies consisting of cabbage and found that sickle mowers did not injure any of their proxies at a blade height of 10 cm. For disc mowers, they found that 36% of their proxies were injured at a blade height of 15 cm. Our study found a higher rate of injury to the proxies, possibly due to a larger sample size or a difference in agricultural machinery and practices. We could not use blade height as an accurate indicator of injury because of the difference in grass heights postharvest. Grass height could be variable because of a combination of topography, blade wear, and tractor speed. These variables were not measured in Erb and Jones (2011) or our study; thus, we are uncertain which of these additional factors could be important for mitigation. Erb and Jones (2011) also noted that the number of “dead” turtles did not significantly decrease when the blade height increased from 10 to 15 cm with the disc mower. They speculated that this could be due to a small sample size or the results were altered by mortality induced by tires. We found similar results but had a larger sample size and tires did not influence our results. This implies that raising the blade height > 15 cm for the disc mower barely reduces the injury or mortality of wood turtles.
We found that the younger size class had varying degrees of injury between the two mower types. Young proxies had a relatively high injury rate (95%) at a grass height of 10 cm for disc mower, but for sickle mower they had a 21% probability of being injured. The number of hatchling proxies missing after harvest by disc mower was relatively small (4.5%), but for sickle mowers 20% of the hatchling samples were missing. A missing hatchling may indicate it was thrown too far away to detect, or more likely, was sliced into pieces too small to be found. It is also possible that the proxies were hard to detect because of their color and size. Hatchlings could be lifted by the suction of the mower and hit by the rotary blade, whereas the sickle mowers may simply displace the individuals, resulting in difficulty finding them. We excluded missing hatchlings from the injury predictions because we were uncertain whether the missing hatchlings were injured; if the missing hatchlings were added to the total damage on young proxies for sickle mower, the percent injured would be 39.5% at 10 cm and still fairly low relative to the disc mower. Although there is large variation in size between age classes, the disc mower caused a relatively high injury rate to all size classes, whereas the sickle mower has a relatively low impact on smaller turtles and even adults at low blade heights. We found that the number of injured young proxies decreased considerably more than the adults when the grass height was 15 cm for sickle mower. If blade heights are not raised for disc mowers, then all classes will have a high probability of being injured or killed. If the shells of young turtles are injured, it is likely that they would die because they cannot sustain the impact (Saumure et al. 2007). The difference in the injury between the two mower types of the size classes could be attributed to the mechanisms of the two mower types (Saumure et al. 2007). Disc mowers cut at an angle with no blade guards; therefore, the size of the turtle would likely not matter if the blade is set low enough.
Our study shows that recent changes in agricultural machinery, such as a move toward greater use of disc mowers over sickle mowers, is likely causing high rates of injury or mortality to all wood turtle age or size classes occupying fields during hay harvest. With no implementation of management strategies, wood turtle populations are likely to become extirpated in agricultural areas.
MANAGEMENT
Managers and researchers have suggested raising blade height by 5–15 cm to lessen injury and mortality. Raising blade height would benefit farmers in that there is little nutritional value in the stems and it would reduce wear on machinery (MacGregor and Elderkin 2003; Saumure et al. 2007; Tingley et al. 2009; Erb and Jones 2011; Wisconsin Department of Natural Resources 2015). However, based on our study, blade heights would need to be raised to a minimum of 17 cm to have ≥ 50% probability of an adult turtle remaining safe from the blade. A similar blade height would likely protect 80% of young turtles. Raising the blade to this height would result in decreased profit to landowners because a greater volume of hay will be left growing in the field. Depending on the type of hay, the relevant commodities markets, and time of the year, landowners could lose approximately US$35/ha (based on current markets) if the blade height was raised to 17 cm (Lantz 2007; Wood Turtle STRIDES n.d.). If landowners have large-scale operations, this could result in a high profit loss. As such, increasing blade height as a mitigation measure to protect wood turtles would likely need to be accompanied by financial compensation, which would allow farmers to purchase extra forage to make up for the loss of forgoing hay harvested at a lower height. Programs have been implemented in various jurisdictions to compensate farmers for implementing such management strategies for species at risk that use agricultural fields (Kleijn and Sutherland 2003; Government of Prince Edward Island 2018; Wood Turtle STRIDES n.d.).
Using a sickle mower instead of a disc mower would be a more suitable option because it would allow farmers to continue to cut hay at low blade heights while reducing the injury rate to adult and young wood turtles. However, the use of sickle mowers is in decline, and typically only small-scale farmers still have them in use. Within New Brunswick, approximately 31% of farmers still use sickle mowers (Noel et al. 2017).
If blade height is raised to reduce the injury to wood turtles, mower tires still pose a substantial threat. During harvest, tires can contact almost 20% of the field's surface area. Therefore, if a turtle does not move during harvest, then the turtle has a 1 in 5 chance of being struck by a tire. Erb and Jones (2011) found that 46% of mortality was due to tires alone and that the rear tires accounted for 37% of the tire width. The width of the tractor and tires used in our experiment were smaller, so the threat of the tractor tires depends on the tractor model and tires sizes used.
Researchers have also recommended delaying the hay harvest or implementing buffer strips to avoid the mower altogether (Semlitsch and Bodie 2003: Saumure et al. 2007; Tingley et al. 2009). Buffer strips would be ideal, but wood turtle populations vary in the length of time they use the fields, and female wood turtles will move up to a kilometer away from the river (Mullin 2019). Management strategies will be dependent on the equipment being used, type of hay production, willingness of landowners, and any knowledge on the current wood turtle population.
Not-for-profit organizations that want to increase wood turtle populations in landscapes under active agricultural can use these management strategies to help reduce mortality risk. Outreach to landowners and farmers who have suitable wood turtle habitat will be necessary to implement these strategies, and funding programs that can help compensate landowners for hay loss would likely increase public involvement. An assessment of each landowner's property would be helpful to assess which management option would be ideal for each landowner based on the amount of agricultural land, proximity to the main river, time of year harvested, and type of mower used. If jurisdiction can institute policy or regulation, using management options presented would benefit wood turtle populations.
Probability of adult and young wood turtle (Glyptemys insculpta) proxies being damaged during hay harvest using a disc mower in areas with different vegetation height. Cantaloupe halves of similar weight and dimensions as adult and young (hatchling to subadult; Table 1) wood turtles were used as proxies for wood turtles. Rotary blades were raised to 5, 10, and 15 cm heights. “Above” and “Below” are referring to the placement of the proxies in relation of the thatch (accumulation of undecomposed grass). Solid lines represent damaged turtles; dashed lines undamaged.
Probability of adult and young wood turtle (Glyptemys insculpta) proxies being damaged during hay harvest using a sickle mower in areas with different vegetation height. Cantaloupe halves of similar weight and dimensions as adult and young (hatchling to subadult; Table 1) wood turtles were used as proxies for wood turtles. “Above” and “Below” are referring to the placement of the proxies in relation of the thatch (accumulation of undecomposed grass). Sickle blades were raised to 5 and 10 cm heights. Solid lines represent damaged turtles; dashed lines undamaged.
Contributor Notes
Handling Editor: Peter V. Lindeman
