Optimizing vole trapping strategies in annual and perennial cropping systems

Preliminary Work

We received final approval (IACUC-2021-0187) to work with voles and detection dogs on August 31, 2021; the approval expires August 30, 2024. We started recruiting dog and handler teams (canine teams) in May 2021, and by late August we had recruited seven teams for our first year of vole detection training. In July and August 2021, we developed vole detection and trapping protocols and determined plot size by incorporating what we learned by timing ourselves doing these tasks in dairy pasture, vegetable farm, and hazelnut orchard settings. We also forced smoke down vole holes using vegetable oil in an insecticide fogger to assess tunnel connectivity.

Dog Training

In September and October 2021, we constructed a vole detection training arena at OSU’s Fairplay Farm in Corvallis using a trench design adapted from Earthdog training tunnels and with discrete holes drilled in the plywood covering the network. We purchased twelve PVC Barn Hunt rat tubes to contain live voles during nose work training; the design adapted was from the 2019 Barn Hunt Association rule book. Nose work training workshops with seven nose work dog-and-handler teams and our professional dog trainer started in November 2021. Training started with a lie-down alert to an object with no odor.  We transferred this lie-down alert to voles in tubes and empty tubes above the ground. Next, we placed the tubes with and without voles in the covered trenches and discrete holes in the training arena and reinforced the lie-down alert.

The voles used for nose work training of the dogs were captured in late October and early November. We used Sherman folding traps to catch live voles at a dairy in Corvallis. Six captured voles were subsequently housed at OSU’s Research Animal Isolation Laboratory (RAIL) for use during canine detection training. Two voles were found dead in the Sherman traps.  Dr. J. Sargent, DVM, Director of the OSU Laboratory Animal Resources Center and OSU attending veterinarian, conducted necropsies and testing that identified leptospirosis infection in the dead voles. Pooled urine samples from the six live voles housed at RAIL also tested positive for leptospirosis, so our training with live voles was discontinued pending revision of our vole handling procedures. After consulting with OSU physicians and veterinarians, the voles were labeled as biohazards and new standard operating procedures (SOPs) for vole handling were developed and implemented for all aspects of housing, transportation, handling, and exposure to dogs and dog handlers. The SOPs included education of dog handlers and strong encouragement to complete leptospirosis vaccination for those dogs not already vaccinated. Dr. Sargent attended one of our nose work sessions to observe our vole handling and dog interaction procedures.

One of the animal caretakers at RAIL became ill with a presumed but unconfirmed case of leptospirosis, at which point Dr. Sargent and the OSU medical advisors terminated our work with live voles, which were subsequently euthanized with isoflurane on January 19, 2022. Blood samples from each vole showed that three were positive for leptospirosis. Based on consultations with canine detection experts around the U.S. we adapted our nose work training using three new approaches that avoided live vole handling: 1) we transferred vole scent to cotton swabs, 2) we freeze-dried voles, and 3) we trained dogs in suspected vole hot spots.

We transferred vole scent to cotton swabs in two ways: 1) swabs placed in metal strainers were hung in the cages for about 72 hours before euthanizing the voles, and 2) at the time of death, the voles were rubbed with cotton swabs. Swabs used for these two types of scent transfer were stored separately for potential use later in scent training. After the voles were euthanized, they were immediately placed in coolers on dry ice to prevent formation of decomposition compounds. Within two hours they were placed in a -80°F freezer to await preservation by freeze drying on a unit equipped with a HEPA filter on the exhaust, per OSU’s Biological Safety Officer. On January 25, we transferred the voles from the freezer to the freeze dryer. On January 27 after approximately 48 hours the freeze dryer automatically shut off, and the voles were removed and stored in amber glass jars with desiccant for use in scent training.

Cotton swabs that had been hung from vole cages elicited a slightly greater response from the dogs. We placed the freeze-dried voles in shallow mason jars with perforated lids and then placed those in shallow holes in a field at Fairplay Farm, which also had active vole colonies. The dogs did not give a reliable alert to the freeze-dried voles, but some dogs alerted on apparently active voles holes. Therefore, we decided to conduct canine training workshops at suspected vole hotspots. First, we returned to the Corvallis dairy where we had caught the live voles for training. Although a few dogs alerted on some holes, we didn’t catch any voles. Then we moved to a Corvallis-area sheep pasture that reported high vole damage. More of the dogs alerted on voles there, and we caught 9 voles and 3 shrews at the first field training at that location and 3 voles at the second. Finally, we moved to a collaborating dairy near Monmouth with heavy vole pressure and caught 15 voles. At each location, we recorded the number of holes identified as active (“sets”) by the canine teams, and we used the number of voles caught per set to select four dog and handler teams for the 2022 field work; three teams (dogs Daphne, Sierra and Bijou) were available for the trials.

The approach of training dogs in vole hotspots appeared to be the most successful; however we noticed during the early winter that apparent vole abundance (via sign and sightings) had declined at some locations, suggesting that the population may have dropped after two-to-three successive years of extremely high vole populations and crop damage. This fits the "boom and bust" irruptive pattern that vole populations are known for. The multiple delays due to the complications described above caused us to be out of sync with our system's prime population-reduction season (January–February), but we conducted field trials in each cropping system to further refine our methodology.

In 2023, dog Reacher replaced Bijou for the research trial. Reacher and his handler are a trained conservation detection team, therefore they bring a different training and working approach than competition nose work teams. Reacher was introduced to vole scent by following Daphne and Sierra during vole searches and being awarded for alerts that were consistent with Daphne and Sierra’s alerts. Reacher’s handler supplemented this training by transferring vole scent from the freeze-dried voles to the interior of foam tubes typically used to insulate pipes. The foam tubes were placed in cinder blocks for introductory scent discrimination training using an approach developed by Working Dogs for Conservation. When Reacher was alerting reliably on vole scent in this setting, his handler used the foam tubes as hides for field training. Blank decoys (foam tubes with no vole scent) and hides with vole scent were placed in small holes in the ground, mimicking a natural vole hole with fresh vole odor. Reacher was rewarded when he alerted by sitting next to the hides with vole scent. This training continued until Reacher was reliably alerting on hides with vole scent but not decoys, and his handler was confident that they were ready to start field work.

In 2024 we received a no-cost extension until December 31, 2024. We continued the field research at one location with larger plot sizes (1/2 acre) searched solely by the professional canine team.

Trapping and crop damage

Our first on-farm trapping season started on March 14, 2022 and ended on March 25, 2022. Five farm locations were surveyed for vole activity with a pair of treatment plots (canine assisted and human searching) on each farm for a total of ten quarter-acre plots. Plot corners were marked with wood stakes. For each treatment pair of plots, one plot was surveyed by a canine-handler team and the other plot was searched by a person alone. The same person served as the human-alone searcher for all plots. During canine searches, 300-foot long tape measures or rope were laid around the perimeter of the plot to help dog handlers keep their dogs within the plots. Holes identified as occupied during canine-assisted and unassisted-human searching were marked with flags. Trap placement began within 30 minutes of the end of the survey. Two to four traps (a “trap set”) were set around most of the holes identified. The number of traps varied according to the number of runways at each hole, with a few holes getting five to eight traps. Burlap squares approximately 6”x6” were used to plug holes (presumed alternative burrow entrances) within a 1 m diameter of each hole with trap sets, based on preliminary observations with forced smoke. Milk carton roofs were installed above trap sets to reduce non-target catches and false triggers from wind, rain, or larger animals. Trap sets were left overnight and then checked and removed approximately 24 hours after initial trap setting. Preliminary trapping during the dog training phase indicated that few voles were caught in traps reset for a second day and more non-target species, especially shrews, were caught.

During the 2022 trials, it became apparent that different cropping systems had pros and cons for the field trials. Regular tillage in vegetable production fields disrupts vole tunnels; after tillage, voles gradually move into the production fields and can cause substantial damage. This disturbance and subsequent vole migration confounds our main research question. Operators of conventional hazelnut orchards spray hazelnut suckers, use rodenticides and sometimes apply residual herbicides to maintain a bare orchard floor. These practices posed safety concerns for our search teams and can impact vole populations. One vegetable farm and the hazelnut orchard had relatively high populations of deer mice (Peromyscus maniculatus) during the 2022 trapping season, this is a non-target species in this project.

During the summer of 2022 we scouted several organic or low input pasture, hay and hazelnut sites and found sporadic evidence of recent vole activity. Pasture and hay fields appeared to have the most consistent signs, and farmer reports of recent vole activity at the sites we visited. For these reasons (vole disturbance from tillage, high incidence of non-target species at some sites, pesticide safety concerns, and signs and reports of recent vole activity) we decided to conduct the 2023 field trials in pastures or hay fields that had not recently been reseeded. Regional vole populations crashed prior to the 2023 season. We surveyed several sites from July to December 2022 and identified sites that still had apparently high levels of vole activity. Heavy rain and ponding at some sites further limited our options, and in December 2022 we selected three sites for the 2023 field trials. We set up two pairs of three plots (one canine assisted, one human searching, one untrapped control) on each of three dairy farms.

Each site was trapped approximate every 14 days from early January until the first juvenile voles were trapped in mid-March, using the same searching and trapping methods described above.

Forage damage assessment protocols include visual assessments of ground cover using the Daubenmire Method described in Coulloudon et al. 1999 (pp 55-63) and the number of vole holes in eight 50 x 50cm quadrats, and collecting dry matter samples from 50 x 25cm quadrats:  eight along the transects and two additional quadrats visually identified with severe vole. We conducted spring damage assessments of all treatment plots from March 28 to April 3 and again from June 20 to 23, 2023.  

We decided to conduct one more year of field work in early 2024 with larger ½ acre plots in the organic hay field with highest vole incidence in 2023. We set up three replicates of two treatments: canine assisted trapping (with the professional team) and an untrapped control. Traps were set approximately every two weeks—except one week when ice-storms caused delays—from January 3 to March 15, 2024, using the same searching and trapping methods as previously described. The sampling method was consistent, but rather than sampling along transects and at sites of severe damage, 10 quadrats were randomly placed in the plots. We also conducted a pre-trapping forage assessment. Vole damage assessments and forage samples were collected on December 14 (before trapping) and March 29 (after trapping). Our final damage assessment will be a few weeks after the hay is cut (anticipated for late May).