Raptors benefit from biosolids applications on rangelands

,


INTRODUCTION
Grasslands around the world have been degraded and lost because of land cover changes from urban encroachment, disrupted fire regimes, and expansion of agriculture (Gayton 2004, Vukovich and Ritchison 2008, Krzic et al. 2014, Newman et al. 2014. In western North America, more than 90% of semi-arid grasslands have been degraded because of human causes (Sullivan and Sullivan 2018). In British Columbia (BC), Canada, grasslands have become degraded since cattle began grazing on the landscape starting in the mid-1800s, with reduced productivity in overgrazed areas (Fleischner 1994, Krzic et al. 2013). These losses of grassland have contributed to major declines in many bird species, including raptors (McClure et al. 2018, Rosenberg et al. 2019). These population declines suggest that habitat restoration will be an important tool for conservation of avian predators.
Some ranchers use biosolids, the solid and pathogen-treated remains from wastewater treatment facilities (Newman et al. 2014, Elkhatib andMoharem 2015), to increase plant growth and yield (Wallace et al. 2009, Larney and Angers 2012, Avery et al. 2018. Even one application of biosolids can have multi-year impacts on soil quality and vegetation cover (Avery et al. 2019). Although studies have addressed how biosolids affect soil, vegetation, and cattle (Meyer et al. 2004, Wallace et al. 2009, Borden and Black 2011, Lu et al. 2012, there are few studies on the impacts of biosolids on wildlife (Washburn and Begier 2011). Both herbivores and the predators that feed on them are likely to be affected by these bottom-up nutrient additions and the strong vegetative changes that result. Further, avian predators have different preferences for hunting over sparsely versus thickly vegetated pasture (Vukovich andRitchison 2008, Massey et al. 2009), so biosolids may also affect whether a hunting bird finds a given area attractive in terms of plant cover.
Given that some migratory raptors rely on high populations of small mammals, it is important to determine if biosolids can improve prey populations and support birds of prey that forage in grasslands. American Kestrels (Falco sparverius) are resident generalists after migration (Poulin et al. 2001), consuming whatever prey are locally abundant (Rojas andStappung 2004, Smith et al. 2017). American Kestrels commonly switch between eating small mammals and insects, primarily grasshoppers (Orthoptera) and beetles (Coleoptera; Sherrod 1978, Smallwood and Bird 2020. American Kestrels are weakly philopatric, nesting in different cavities or boxes within the same region in multiple years Peterson 2009, Smallwood andBird 2020); shifts in location are tied to nesting opportunities rather than diet (Steenhof and Peterson 2009).
In contrast, Short-eared Owls (Asio flammeus), Long-eared Owls (A. otus), and Northern Harriers (Circus hudsonius) are migratory to nomadic specialists that focus on voles (Korpimäki 1985, Poulin et al. 2001, Smith et al. 2011, Booms et al. 2014, Marks et al. 2020, Schimpf et al. 2020. Because some vole populations are cyclic (Oli and Dobson 1999, Krebs et al. 2002, owls and harriers move each year to areas where vole populations are high (Hooper and Nyhof 1986, Korpimaki and Norrdahl 1991, Poulin et al. 2001, Houston 2005, Booms et al. 2014). However, grassland degradation has caused widespread interruptions in vole cycles and reduced vole populations Sullivan 2010, Booms et al. 2014), thus reducing areas that can support these raptors. Short-eared Owls are provincially blue-listed in BC (a listing that indicates conservation concern) and federally listed as Special Concern under the Species at Risk Act; both listings enable habitat protection.
Here, we examine temporal plasticity from 2016 to 2020 in diets and habitat use for these raptors on the OK Ranch, a large cattle ranch in central British Columbia, Canada, where biosolids were used as a remediation tool. The vole population on the Ranch peaked in 2017, then crashed in 2018. On-going biosolids applications meant the area of the ranch with biosolids application doubled during this study period, thus also raising the question whether the enhanced productivity of these grasslands would support the nomadic owls in years when voles were not at cyclic peaks.
We hypothesized that biosolids application would support a bottom-up increase in prey (small mammals, Orthoptera) and these avian predators. We also predicted the vole specialists would have higher densities in 2017, during the vole peak, but be rare or absent when vole populations were low, then start to rebound in 2020 as vole populations increased. We predicted American Kestrel populations would remain similar across years, and that kestrels, harriers, and owls would use biosolids-amended pastures more than untreated areas, as biosolids bolster the insect and small mammal prey base at the OK Ranch , Gaudreault et al. 2019, Ormrod et al. 2021. Owl and harrier diets were expected to be dominated by voles in all years, with kestrel diets being variable.

METHODS
We worked on the OK Ranch, a privately-owned cattle ranch near Jesmond, British Columbia (51°33′N, 122°0′W), from 2016 to 2020. Our research focused on 45.3 km² of rangeland that consisted mostly of grasslands dominated by bluebunch wheatgrass (Pseudoroegneria spicata), needle-and-thread grass (Hesperostipa comate), junegrass (Koeleria macrantha), and Nevada bluegrass (Poa secunda juncifolia). Within these grasslands, there were also small wooded stands (primarily Douglas fir, Pseudotsuga menziesii and aspen, Populus tremuloides) that have a history of logging, and some small wetlands and ponds. Cattle grazing occurred throughout the ranch.
The average yearly temperature at the OK Ranch from 2014 to 2020 was 3.1 °C, with January and July averages being -6.7 °C and 14.3 °C, respectively (Environment and Climate Change Canada 2020). The average snowfall per year at the OK Ranch from 2014to 2020 was 108.9 cm with snow falling an average of 8.4 days per month (World Weather Online 2020). The average elevation is 1100 m and the annual precipitation is 401 mm per year (Avery et al. 2019).
From 2014 to 2020, SYLVIS Environmental Ltd. applied biosolids regularly to the OK Ranch. We used GIS data layers provided by SYLVIS to quantify how much of the study area had been treated with biosolids each year. Between 2014 and 2018 biosolids were spread on 22.78 km² of the 45.3 km² study area. In 2016, 24.5% of the pastures had had biosolids applied, in 2017, 34.7%, and 2018, 50.3%. In 2019, the reapplication of biosolids began in pastures that had biosolids applications in 2014 and 2015, and new sites also had biosolids applied, with 52.0% of pastures spread by the end of the season. In 2020, biosolids were reapplied to some pastures initially spread from 2014 to 2017, but no new areas were spread. The OK Ranch also contains areas of publicly owned Crown Land interspersed throughout and around it; these areas were not spread with biosolids. Pastures applied with biosolids had higher grass and vegetation cover than did sites that had no biosolids (Meineke et al., unpublished manuscript).

Prey abundance
In 2017, there was a natural peak in vole populations at the OK Ranch (Ormrod et al. 2021), followed by a crash overwinter such that densities were very low in 2018. In 2019, small mammals were live-trapped on four trapping occasions between May and August on nine sites each with 50 traps spaced 15 m apart, baited with peanut butter, oats, and apple or corn for moisture (Meineke 2020;Meineke et al., unpublished manuscript). Because of pandemic restrictions in travel in 2020, we trapped eight sites in August only (one site was inaccessible because of fire risk); voles are usually at their highest density in late summer, so this timing was chosen to focus on this annual peak in abundance. Here, we summarize Lincoln-Peterson density estimates from eartagged animals; detailed methods and results are in Meineke (2020;Meineke et al., unpublished manuscript). Because meadow voles (Microtus pennsylvanicus) and montane voles (M. montanus) are ecologically similar but difficult to differentiate in hand (Pearson et al. 2001, Hales 2011), we grouped them as "vole." Long-tailed voles (Microtus longicaudus) were sometimes consumed by raptors, but we did not catch any during our live-trapping.
From 2016 to 2020, we counted grasshoppers in 12-16 pastures from 12 July to 2 August (sampling dates varied with field access related to regional wildfires). Our 2016 data are from Gaudreault et al. (2019) and we mostly resampled these sites in 2017-2020, but we adjusted some sites because of additional spreading of biosolids that converted control sites. At each site, we laid out 16 circular hoops (each 0.25 m²) in a 4 x 4 grid with hoops 4 m apart. We deployed hoops, let them sit for at least 1 hour so that grasshoppers would re-settle after the disturbance, then we counted grasshoppers within each hoop.

Raptor observations and habitat use
During 2016-2019, we were in the field for ~4 months from late April to mid or late August, although in 2017 we left for several weeks in August because of wildfire risk. In 2020, because of pandemic travel restrictions, we had three short field visits, of nine days in late May, six in late July, and 10 in late August. Throughout each field season, the presence of raptors at the OK Ranch was monitored through the use of "Seen Sheet" data, visual observations of species of interest recorded while we drove around the OK Ranch (Hochachka et al. 2000, Aumann 2001, Poulin et al. 2001, Ormrod et al. 2021. We used roads, fences, and other obvious features to divide the ranch into 75 pasture or woodland sections, and we classified each section as either biosolids-amended or untreated. We re-evaluated classifications each year using annual GIS spatial layers provided by SYLVIS (Fig. 1). When species of interest were observed, the time, location, and number of individuals were recorded, but we did not record behavior systematically. For effort, we recorded start and end time, and start and end km each day. Roads on the OK Ranch are primarily single-track dirt or gravel roads that are traveled between 10-30 kph. Multiple sections could be seen simultaneously from the roads, so the section the focal species was first sighted in was recorded as the location for that detection. Observations occurred between 0500 and 2200.

Dietary analyses
To address kestrel and owl diets, regurgitated pellets were collected and dissected. We used pellets rather than cameras at nests so that we could sample diets across the summer and for better resolution to species than is enabled from photographic records (Ormrod et al. 2021).
Pellets were collected by walking fence lines at the OK ranch and collecting pellets at the base of fence posts. We surveyed the samẽ 11.8 km of fenceline each month. We also spent 2-5 hours in spring and again in summer in each of the woodland patches, looking for owl pellets and nests. These searches either resulted in finding no or very few pellets, or in locating a nest or roost site with many pellets under one or several trees. Pellets from nest sites were collected after chicks had fledged. Pellets were identified to the raptor or owl species based both on the location of collection, such as being found near known nest sites, and by the use of morphometric measurements including length, mass, and shape (Holt et al. 1987, Ormrod et al. 2021.
When skulls were present, small mammal remains were identified to species using features of the dentition and incisive foramina (Nagorsen 2002, CalPhotos 2020. Skulls missing key features were identified to genus or family. We counted the minimum number of small mammal individuals per pellets using the number of jaws and pelvis bones found in each pellet. When nondiagnostic fur or bones were found in pellets, we identified remains as "small mammal." Grasshopper mandibles were identified to species by using a reference collection from the ranch (Gaudreault et al. 2019). The number of grasshoppers per pellet was calculated from counts of individual mandibles of each species, with each pair counted as a single individual, or the higher number of multiple left or right mandibles counted to give the minimum number of prey. Any unidentifiable grasshopper remains found in pellets that lacked mandibles were counted as one individual. Beetle remains found in pellets were identified as Meloe spp. based on field observations but were not quantified to individuals. Grass and seeds were found in several pellets each year; however, these remains were minor and were likely incidental ingestion or secondary consumption and were not included in our analyses.
Bird remains in pellets mostly consisted of feathers and bones; multiple individuals were identified if several beaks were present. We identified birds as Passeriformes or Galliformes based on structures of the feathers as viewed under a compound microscope (Dove and Koch 2011). We counted birds as part of the diet when feathers made up more than 20% of the pellet, to avoid counting accidental ingestion from preening. The Vesper Sparrow (Pooecetes gramineus) was used as the mass for passerine remains because they were the most common grassland nesting bird in the study area  and are known prey for Short-eared Owls and American Kestrels (Tomkins 1936, Sherrod 1978).

Data analysis
To analyze seen sheet data, we present "detections," the number of times we saw one or more individuals per species, and "counts," the sum of individuals seen during all detections. For example, if we saw two American Kestrels together and then another kestrel later, that would result in two detections and a count of three. To examine how the raptors used the pastures amended with biosolids, we used chi-square tests to compare bird observations in areas with and without biosolids applications against the proportion of the study area with biosolids-amended pastures; this analysis was done separately for each year because of differences in the amount of the ranch that had biosolidsamended pastures. We also calculated Manly standardized selectivity ratios (Manly's α) using the "adehabitatHS" package version 0.3.15 (Calenge 2006) in RStudio version 1.2.1335 (R Core Team 2019) to measure raptor selection of biosolidsamended and untreated pastures (Manly et al. 2002). Alpha values over 0.5 (1/number of habitat types available) indicated selection for that pasture type.
To analyze diets, we focused on the relative frequency of occurrence (n prey type/total prey consumed). The relative frequency of occurrence best reflects the prey killed, whereas the absolute frequency of occurrence (n prey type/n pellets) addresses how often each prey type occurs per pellet, and biomass accounts for difference in prey masses and better reflects the calories provided from each prey type. Biomass estimates and absolute frequencies are provided in the Appendix 1.
To relate prey selection to small mammal density in 2019 and 2020, we calculated the density of small mammals on biosolidsamended and untreated pastures to determine the ratio of deermice (Peromyscus maniculatus) to voles. We used a chi-square test to compare the prey available against the remains in pellets to see if owls and raptors ate mice and voles in proportion to their availability. We also used Manly's standardized selectivity ratios (Manly et al. 2002) to measure whether voles were selected over deermice, with alpha values over 0.5 indicating selection for voles.

RESULTS
Deermice were more abundant than voles across the ranch in both 2019 and 2020 (Meineke et al., unpublished manuscript). Deermice densities in August on pastures with biosolids ranged from 0.9 to 34.7 deermice/ha, whereas August densities on untreated pastures ranged from 1.8 to 80.7 deermice/ha; deermice were more abundant on control sites. We did not capture enough voles in either year to calculate credible density estimates. In 2019, the rodent prey population at the OK Ranch was 93% deermouse and 7% vole (Meineke 2020), and in 2020 it was 87% deermice and 13% vole (Meineke et al., unpublished manuscript). Grasshopper densities were highly variable across the years, with densities on control sites ranging from 0.09 per 0.25 m² in 2017 to 0.56 in 2016 (Table 1). In all years except 2017, grasshoppers were more abundant on sites that had had biosolids applied.
From 2016 to 2020, we detected American Kestrels 1404 times, Short-eared Owls 105 times, Long-eared Owls twice, and Northern Harriers 548 times (Table 2). American Kestrels were detected much more often in 2016 than in other years, and were often seen in groups. Kestrels were seen more than twice as often in 2017 than in 2018 or 2019. In 2020, we also had high kestrel sightings, despite being in the field for only ~3 weeks. Northern Harriers were seen least often in 2018. Most harriers were alone at the time of sighting in all years.  Ormrod et al. (2021). . § We did not locate American Kestrel nests in 2018-2020; the estimates are based on field observations of family groups in late summer. | In August 2020, we camped close to a kestrel family group. To make seen sheet records comparable to other years, we omitted 22 detections of 51 kestrels that we saw within 5 minutes of arriving to or leaving from camp. ¶ The nest in 2017 was not located on the main study area, but we strongly suspect those birds foraged on the study area. The 2019 and 2020 nests were in the same location on the study area, in a dense willow thicket.
Short-eared Owls were abundant in 2017; we saw owls 95 times and had three confirmed and one probable nest location (Ormrod et al. 2021). Short-eared Owls were also present in 2016 (3 sightings), 2018 (0 sightings, but fresh pellets were collected), 2019 (3 sightings and one nest), and 2020 (4 sightings). Long-eared Owls were present in 2017 (Ormrod et al. 2021) and 2020, with two nests in 2017, one in 2020, and one sighting and multiple pellets in each of these two years. We did not detect Long-eared Owls in 2016, 2018, or 2019, despite dedicated effort looking for nests and pellets.
American Kestrels strongly selected for pastures with biosolids every year from 2016-2020 (Fig. 2, Table 3). In 2016, 2019, and 2020, Northern Harriers preferred pastures that had had biosolids applied; we note that sightings were far fewer in 2018 (7 detections, vs. 88-226 in other years), so our ability to assess habitat preference was weaker in this year. When Short-eared Owls were abundant in 2017, they used pastures with biosolids significantly more than untreated pastures; they also nested in pastures with biosolids (Ormrod et al. 2021). In the other 4 years, we had 0-4 sightings per year, with 7 of 10 sightings on pastures with biosolids. The successful nest we observed in 2019 was on a biosolids-amended pasture. Long-eared Owls were absent in 3 of 5 years; the confirmed nests in 2017 and 2020 were in woodland adjacent to pastures with biosolids.
From 2017 to 2020, we collected and analyzed 729 pellets from kestrels, harriers, and owls (Fig. 3). In 2017, American Kestrel diets were evenly split between insects (50.6%) and voles (47.5%; . In 2018, American Kestrels ate grasshoppers and beetles as 98.7% of the individual prey in the diet (Table 4, Fig. 3). In 2019 and 2020, American Kestrels still primarily ate insects (82.1% and 71.4%, respectively), but the amount of small mammal prey increased, from 17.9% to 25.7%. In 2020, American Kestrels ate similar numbers of voles and deermice (χ² = 2.37, p = 0.19), but because voles were rarer, kestrels were actually selecting voles over deermice (Manly's α = 0.75).   . Birds were often sighted in groups, so we present detections (number of times that species was seen, regardless of group size), and counts (total number of birds observed). Between 2016 and 2019, 24%, 35%, 50%, 52% of the land had applications of biosolids, respectively, no new areas were spread with biosolids in 2020 so application remained at 52%. Chi-square testing was used to test whether bird observations were disproportionate in relation to sites with biosolids. We were unable to determine if these pastures had or had not been spread with biosolids at the time the observations took place. An additional 11 American Kestrel detections (count = 29) and 6 harrier detections (5 alone, one pair) occurred where birds were either on the edge between an untreated and biosolids area or moved quickly from one to the other. We have not included these records in our habitat use statistics. ‡ 2017 data can be found in  and Ormrod et al. (2021). § In this table, we included the kestrels observed near our campsite. Voles made up > 92% of individual prey in the Short-eared Owl diet in 2017 and 2018 (Table 5, Fig. 3; Ormrod et al. 2021). In contrast, in 2019 and 2020 voles made up 61.7% and 48.0% of the Short-eared Owl diet, respectively, while deermice were 27.7% and 44.0%. Short-eared Owls also ate a few birds in 2017 (2.6%) and 2019 (4.3%). Short-eared Owls ate voles disproportionally often compared to deermice (2019: χ2 = 266.37, p < 0.01, 2020: χ2 = 26.66, p < 0.01) and strongly selected voles over deermice (2019: Manly's α = 0.97, 2020, α = 0.87).

DISCUSSION
Biosolids applied to rangeland appeared to benefit migratory and nomadic birds of prey that use grasslands for foraging. All of the species preferred pastures amended with biosolids. Kestrels showed preference for biosolids in all years, whereas harriers  .7 2 2 5.7 † Melanoplus bivitattus or Melanoplus sanguinipes, these mandibles were too degraded to identify to species. ‡ A. longipes or A. simplex, but these mandibles were too degraded to identify to species. § Mandibles were identified as "grass-eating grasshopper spp." when mandibles of the following species were too degraded to identify: Arphia pseudonietana, Bruneri brunnea, or Pseudochorthippus curtipennis. preferred biosolids in 2016, 2019, and 2020. In 2017, when Shorteared Owl sightings were at their peak, they also used biosolidsamended pastures more than untreated pastures, and when we confirmed breeding (2017 and 2019) all the nests were in the biosolids-amended pastures. Long-eared Owls nested in forests adjacent to pastures with biosolids. This pattern of raptors preferring the biosolids-amended pastures was likely due to bottom-up changes in prey availability and vegetative cover.
The vole cycle also affected the raptor community. Northern Harriers were present in all years, but with lowest numbers during the vole crash. They preferred pastures that had had biosolids applied, which is consistent with their diet, as by far the majority of identifiable prey were voles. The nomadic Long-eared Owls were present in 2017, during a vole peak, and in 2020, as voles began to increase again. Short-eared Owls were most numerous during the vole population peak, but were also present in other years, thus indicating this biosolids-amended landscape was suitable for them even when voles were scarce. American Kestrels were abundant and nested at the Ranch each year, but ate considerably more voles when voles were abundant.
These results suggest that the Short-eared Owls and Northern Harriers tracked the vole cycle but nested at the ranch in low-vole years because of biosolids-induced increases in abundance of other prey species. Short-eared Owls are vole specialists in North America (Wiggins et al. 2020), with small mammals making up > 90% of the diet, and with vole species making up > 83% of these small mammals (Errington 1937, Clark 1975, Colvin and Spaulding 1983, Wiebe 1991, Holt 1993. During the vole population peak in 2017, Short-eared Owls were seen 95 times (Ormrod et al. 2021) versus a total of 10 observations across the other four years. Similarly, Poulin et al. (2001) observed Shorteared Owls 604 times during a peak vole year in Saskatchewan, when in other years they did not record more than two observations. Surprisingly, we documented successful breeding of Short-eared Owls in 2019, when they consumed a diverse diet: nearly 30% of the diet was deermice and shrews, and 10% consisted of birds and insects. These owls still selected voles, with voles making up 61.7% of their diet, despite the 2019 small mammal prey base consisting of 93% deermouse and only 7% vole. The diverse summer diet suggests that even though the number of voles in the diet was low, there were enough alternate prey on the biosolids-amended pastures to support breeding. Table 5. Diet of Short-eared Owls (Asio flammeus). In 2016, we had 2 pellets with 6 prey: 5 voles, and 1 Meloe spp. In 2017, we had 90 pellets with 194 prey (Ormrod et al. 2021); here, we present the relative frequency of occurrence (RFO) for comparison to other years. In 2018, we had 11 pellets with 34 prey, in 2019 we had 24 pellets with 47 prey, and in 2020 we had 11 pellets with 25 prey. Year 2017 was a high vole year, whereas 2018-2019 were low vole years. In 2019, the majority of pellets were collected from a single nest site. Relative frequency of occurrence is the percentage of a given prey type out of all prey recorded.  Table 6. Diet of Long-eared Owls (Asio otus). In 2017, we had 46 pellets with 72 prey (Ormrod et al. 2021); here, we present the relative frequency of occurrence (RFO) for comparison to 2020. In 2020, we had 324 pellets with 566 prey. Year 2017 was a high vole year, and 2020 had increasing vole densities after two low years. In 2020, the majority of pellets were collected from a single nest site. Relative frequency of occurrence is the percentage of a given prey type out of all prey recorded. Long-eared Owls were rarer than the Short-eared Owls. Similarly, Long-eared Owls in Saskatchewan were extremely rare except in years of vole population peaks (Houston 2005). In 2017, we located two Long-eared Owl nests (Ormrod et al. 2021) and in 2020, we located one. A second failed Long-eared Owl nest may have been present in 2020, based on the discovery of a large number of pellets and a deceased Long-eared Owl in a second location. All Long-eared Owl nests were located in forested areas where biosolids were not spread, but these forest patches were surrounded by high proportions of biosolids-amended pastures. It is possible that undetected Long-eared Owls used forest on the outskirts of our main study area, but from our woodland walking surveys we are confident owls did not nest in 2016, 2018, or 2019 in the main study area. These nesting owls primarily consumed voles.
American Kestrels had multiple nesting pairs on the ranch from 2016 to 2020. However, unlike our hypothesis that these dietary generalists would have stable populations (Rojas andStappung 2004, Santillán et al. 2009), American Kestrel populations varied across years. We think this decline was likely due to declines in suitable nest trees. Logging at the OK Ranch increased starting in winter 2017, removing about 70% of the small patches of forest in our study area by 2019. American Kestrels used biosolidsamended pastures more than untreated areas during all five years of study. This pattern was especially notable in 2016 when kestrels disproportionately selected for biosolids treatments when only 24% of the study area had had biosolids applied. American Kestrels consume grasshoppers throughout their range (Buers et al. 2019, Smallwood andBird 2020), and kestrels likely prefer the biosolids application areas because of this increase in grasshopper prey.
During the vole peak in 2017, American Kestrels' diet was split almost evenly between insects and voles ). The small vole populations in 2018 resulted in American Kestrels switching to eating more grasshoppers, with almost no small mammals in their diet. In 2019 and 2020, we obtained fewer pellets, but we focused on obtaining pellets from as many individuals as possible and they were collected from across the ranch. These pellets showed kestrels consumed both small mammals and grasshoppers in these years. This response is similar to American Kestrels in Argentina that changed their diets based on seasonal changes in prey availability, rather than selecting specific prey (Sarasola et al. 2003).  -

CONCLUSIONS
In British Columbia alone, some 38,000 dry tonnes of biosolids are produced annually, of which 21% are used in forestry, reclamation, or agricultural applications (BC Ministry of the Environment 2017). It is thus important to assess how these terrestrial applications of biosolids affect food webs. Collectively, our results indicate that there are bottom-up effects of biosolids through vegetation to prey to predators. Deermice were more abundant on untreated pastures, which reflects the fact these pastures had less plant cover and more bare soil (Meineke 2020;Meineke et al., unpublished manuscript), conditions the generalist deermice are able to use. Biosolids led to higher vegetative cover and were more suitable for voles than were the control sites. All four of these birds of prey clearly consumed more voles when voles were abundant, and they selected voles over mice. Grasshoppers were more abundant on treated pastures and were especially important components of the diet for kestrels.
Degradation of grassland habitats and declining raptor and owl populations are a focal conservation concern (Ims et al. 2008, Vucovich and Ritchison, 2008, Booms et al. 2014, Marks et al. 2020. In British Columbia, the Short-eared Owl is a species of special concern, while the Long-eared Owl has declined throughout its North American range (Kirk and Hyslop 1998, BC Conservation Data Centre 2020, Marks et al. 2020). Importantly, Short-eared Owls were present in this landscape during all years even when vole densities were low, and all four raptor species preferred pastures with biosolids. It therefore appears that grassland restoration via application of biosolids could be an important conservation tool for avian predators as well as for soils and plants.

Appendix 1. Supplemental dietary data.
We also calculated the Absolute Frequency of Occurrence and percent wet biomass of each prey type. Absolute frequency of occurrence shows the percent of pellets containing each prey type. Given that American Kestrels typically expel one pellet every 21 hours (Balgooyen 1971), Short-eared Owls expel about one pellet per meal (Duke et al. 1976), and Long-eared Owls expel about 1.4 pellets per day (Marti 1973), this metric can help to examine the daily diet of raptors and owls when pellets are collected frequently. The wet biomass of each prey type consumed reflects how much energy the raptor or owl gained from eating that prey item. We used prey biomasses as presented in  and Ormrod et al. (2021). c. Anabrus longipes or A. simplex, but these mandibles were too degraded to identify to species.
d. Mandibles were identified as "grass-eating grasshopper spp." when mandibles of the following species were too degraded to identify: Arphia pseudonietana, Bruneri brunnea, or Pseudochorthippus curtipennis. Table A1.3. Diet of the Short-eared Owl. Absolute frequency of occurrence (AFO) is the percentage of pellets containing each prey type, and the wet biomass of each prey type is shown as a percentage of the total diet. In 2017, we had 90 pellets with 194 prey, as published in Ormrod et al. (2021). In 2018, we had 11 pellets with 34 prey, in 2019 we had 24 pellets with 47 prey, and in 2020 we had 11 pellets with 25 prey.