Failed landings after laying hen flight in a commercial aviary over two flock cycles 1

Failed landings after laying hen flight in a commercial aviary over two flock cycles 1 D. L. M. Campbell, S. L. Goodwin, M. M. Makagon, J. C. Swanson, and J. M. Siegford,2 Animal Behavior and Welfare Group, Department of Animal Science, Michigan State University, East Lansing; and Department of Animal Science, University of California, Davis ABSTRACT Many egg producers are adopting alternative housing systems such as aviaries that provide hens a tiered cage and a litter-covered open floor area. This larger, more complex environment permits expression of behaviors not seen in space-limited cages, such as flight. Flight is an exercise important for strengthening bones; but domestic hens might display imperfect flight landings due to poor flight control. To assess the potential implications of open space, we evaluated the landing success of Lohmann white laying hens in a commercial aviary. Video recordings of hens were taken from 4 aviary sections at peak lay, mid lay and end lay across two flock cycles. Observations were made in each focal section of all flights throughout the day noting flight origin and landing location (outer perch or litter) and landing success or failure. In Flock 1, 9.1% of all flights INTRODUCTION In many parts of the world, including within North America, the laying hen industry is gradually phasing out conventional cages in favor of alternative housing that is designed to address consumer concerns for improved hen welfare. One alternative system being adopted is the aviary, consisting of a tiered structure (sometimes enclosed) with perches, nestboxes, feeders and waterers, and a floor litter area. This open space and presence of vertical structures permits behaviors that are not seen in conventional or even enriched cages, such as jumping or flight. Studies are needed to further our understanding about how hens utilize the resources of this system to best implement them or prepare hens to use them. There are few studies that have looked at hen flight in aviary systems (Colson et al., 2008). However, there have been studies conducted in com- C 2015 Poultry Science Association Inc. Received January 14, 2015. Accepted August 3, 2015. 1 Research support provided in part by a grant from the Coalition for a Sustainable Egg Supply (Kansas City, MO). 2 Corresponding author: siegford@msu.edu failed and 21% failed in Flock 2. The number of flights decreased across the laying cycle for both flocks. Proportionally more failed landings were observed in the double row sections in Flock 2. Collisions with other hens were more common than slipping on the ground or colliding with aviary structures across sections and flocks. More hens slipped on the ground and collided with physical structures at peak lay for Flock 2 than at other time points. More collisions with other hens were seen at mid and end lay than at peak lay for Flock 2. Landings ending on perches failed more often than landings on litter. These results indicate potential for flight-related hen injuries in aviary systems resulting from failed landings, which may have implications for hen welfare and optimal system design and management. Key words: behavior, welfare, aviary, flight, laying hen 2016 Poultry Science 95:188 197 http://dx.doi.org/10.3382/ps/pev270 mercial aviaries or similar litter-based systems that have observed both hen occupancy of the litter area across time (Carmichael et al., 1999; Channing et al., 2001; Odén et al., 2002) and behaviors displayed by hens when using this resource such as foraging and dust bathing (e.g., Channing et al., 2001). Furthermore, within the commercial aviary used for this research, we found that hens used the litter area throughout the day with periods of peak occupancy, typically during afternoons when hens dust bathe in large groups, and hens exhibited group piling behavior (Campbell et al., 2015a). In aviaries, hens travel to the floor to access litter, travel to perches, and to the aviary cage, for which one mode of travel is flight, and few studies have examined the occurrence of this behavior in aviary open litter areas (Colson et al., 2008). Flight can be important for hens as it is the expression of a natural behavior and the movement exerts mechanical loading on bones (of the wings in particular), leading to better bone development for improved welfare (Newman and Leeson, 1998; Leyendecker et al., 2005; Regmi et al., 2014). But domestic hens have increased wing loading (i.e., more weight per wing area) in comparison to their jungle fowl ancestors, leading to a reduced ability to control lift when flying (Moinard 188

FAILED LANDINGS AFTER LAYING HEN FLIGHT IN AVIARIES 189 et al., 2004a). Laboratory studies have demonstrated that hens are less successful at navigating between perches of increasing distances apart (Scott and Parker, 1994) with a lack of control and clumsy or missed landings being witnessed (Moinard et al., 2004a). Hens also experience more difficulty and higher energy expenditure when trying to take off from and land on perches when obstructions are present (Moinard et al., 2005). This suggests that within an aviary, both the distance available for open litter flight and the presence of other hens might lead to imperfect flights and landings but as yet, this has not been described in a commercial system. Furthermore, comparisons of laying hens from varying housing types show birds in aviaries and other non-cage litter systems have elevated rates of keel fractures (Käppeli et al., 2011; Wilkins et al., 2011). It is possible that flight collisions with aviary structures or other hens might contribute to the high injury rates seen in these systems (Gregory and Wilkins, 1996). In this study, we observed hen flights and landings in open litter areas of 2 widths, across the laying cycle for two flocks of Lohmann White hens to provide an initial report of this behavior in a commercial setting. We documented the rate of failed landings, how the landing failed (e.g., slipping, collision with structural elements, or interference from other hens) and whether flight and landing failure rates differed depending on where the flights originated or ended. We predicted that a substantial number of flights would end in failed landings, flights would be less successful when hens were landing on a perch, and that longer flight distances would lead to more failed landings. We discuss possible implications of our results in reference to the high injury rate of aviary-housed hens and the sustainability of alternative systems. Aviary Housing MATERIALS AND METHODS The aviary system (aviary) used for this research housed 49,842 and 49,677 Lohmann White laying hens placed at 19 wk of age and 17 wk and depopulated at 77 or 78 wk, for Flock 1 and Flock 2, respectively. For further details on the rearing of the pullets, diet, and lighting see Jones et al. (2014). The house had belt manure removal under the cages every 3 to 4 d, but litter substrate was removed from the floor only following hen depopulation. The manure buildup throughout the flock cycle functioned as a litter substrate in addition to a thin(<1 cm) layer of wood shavings initially placed on the concrete floor prior to hens gaining floor-access. Data on litter coverage and depth were recorded at 3 points within each sampled section, directly underneath camera placement. Litter coverage was visually scored as follows: Score minimal : less than 25% of area is covered (no scores were assigned to this category so it was not used further); Score poor : 25 to 49.9% of area is covered; Score moderate : 50 to 74.9% of area is Figure 1. (a) Representation of the tiered aviary enclosures as seen from the end of a unit showing the outer perch, open litter, and underneath litter areas, the inner perches and ledges on each tier. (b) A schematic top-down diagram of the aviary house showing the tiered enclosure and open litter area, focal video-recorded sections, and single and double rows. The two striped-patterned sections in the single rows were extra sections used during peak lay of Flock 1 when no doublerow cameras were yet installed. The remaining 4 solid grey sections (2 single and 2 double rows) were used for all remaining recorded time points. covered; Score good : 75 to 99.9% of area is covered; Score full : 100% of area is covered. Within the aviary barn, there were 6 rows of threetiered cage structures (enclosures), all with litter access both in front of (40% of total litter area) and underneath the enclosures (60% of total litter area; Figure 1a, b). The tiered enclosures contained multiple perches, two water lines, and four feed troughs dispersed among the tiers and a row of nest boxes lined the back of the upper tier. One outer perch was situated above the litter area in front of the lowest tier where the birds exited the tiered enclosure to reach the floor (Figure 1a, b). Each row of tiered enclosures was subdivided by wire gates to create 10 sections down the length of a row, with each section 1,440 cm in length. In the two exterior single rows of tiered enclosures (i.e., when a single row of enclosures faced an outer wall of the house across an open litter area), 852

190 CAMPBELL ET AL. laying hens were placed in each section at the population of the house, and the open litter area was 120 cm wide. In the two interior double rows of tiered enclosures (i.e., when two rows of enclosures faced each other across a shared litter area) 1,704 hens were placed in each section at the population of the house, and the open litter area was 240 cm wide. In accordance with standard commercial practice, birds were not added to sections following mortalities that occurred throughout the lay cycle. Mortality was recorded as a cumulative percentage within the entire aviary house (peak lay Flock 1: 0.72%; peak lay Flock 2: 0.97%; mid lay Flock 1: 5.41%; mid lay Flock 2: 6.03%; end lay Flock 1: 11.29%; end lay Flock 2: 10.99%). (For additional details on available space per bird and provision of other resources, see Jones et al., 2014; Zhao et al., 2015). Experimental data were collected over two flock cycles: Flock 1 data collection spanned June 2011 to May 2012, and Flock 2 data collection spanned August 2012 to August 2013. System Management Doors on the lowest tier of the enclosures opened for the first time when hens reached 95% production (Flock 1 = 25 wk and Flock 2 = 24 wk), allowing birds access to the floor litter area. For the remainder of the flock cycle, doors opened each day after morning egg-laying (at approximately 11:00 across both flocks) and closed again each night 30 min prior to lights coming back on so as to keep access to the cage with feeders and drinkers available as long as possible for hens remaining out on the litter overnight. As Flock 1 was the first flock in this newly built system, there were several changes in management practices through Flock 1 that affected data recording and bird movement and that prevented us from directly comparing results from the two flocks. First, installation of video cameras was not complete prior to opening of the aviary enclosure doors allowing hens to access the open litter, thus, peak lay recording for Flock 1 occurred 2 weeks after hens initial exposure to litter. Second, the gates dividing the rows into sections were left open during peak lay video recording of Flock 1, allowing birds to travel between sections, affecting absolute numbers of hens per section. Following this incident, hens were redistributed evenly between sections prior to mid lay video recording, and section gates remained closed for the remainder of Flock 1 and for all of Flock 2. Finally, tiered enclosure doors providing litter access were left continuously open during mid lay video recording of Flock 1, meaning that birds could access open litter areas at any time, increasing the available hours during which flight could occur in open space. Continuous litter access did not occur during peak lay or end lay video recording for Flock 1 or at any point during Flock 2. Disturbance of the birds was minimized with no personnel entry into the aviary barn after morning egg collection was complete. Video Data Collection High-resolution digital video cameras (VF450: Clinton Electronics (Loves Park, IL) were ceiling-mounted in six of 40 sections within the aviary system (four sections of hens housed in single-rows and two sections of hens housed in double-rows; Figure 1b). At the time of recording for peak lay of Flock 1, cameras had been installed in the four single-row sections only (n = 4 recorded); however, for all other periods, data were collected from 2 single- and 2 double-row sections (n = 4 again). Three cameras were installed per section with each camera capturing one third of the open litter portion of the floor area, including the outer perch. Cameras were not able to see the litter area that extended under the tiered enclosures. Video recordings were made in each focal section at 3 significant time points in the laying hen production cycle for both flocks: peak lay (Flock 1: 27 wk of age - 95.93% production; Flock 2: 24 wk - 96.5% production), mid lay (Flock 1: 52 wk - 93.19% production; Flock 2: 55 wk - 89.33% production) and end lay (Flock 1: 77 wk - 82.86% production; Flock 2: 76 wk - 76.85% production). We attempted to record the birds at similar ages and points of production between flocks, but were limited in our ability to do so by the remote location of the commercial site and need to coordinate site access between multiple researchers. One full day of recording from lights on ( 06:00) until lights off ( 21:30) was made at each point. With the exception of Flock 1, mid lay period when enclosure doors remained open continuously (see System management ), behavioral observations of flight were made starting with aviary enclosure opening and continuing until lights went off, as this was when hens had access to open litter. During the Flock 1 mid lay period, when hens had continuous litter access, observations were made starting when lights came on and ending when lights went off. Though this created 5 h of additional data not present at other time points, this time of initial access to litter each day (irrespective of what time that occurs) was a period when hens were actively moving, and likely flying, in the system. Other behavioral data from this system showed the highest hen movement between the litter area and aviary enclosure and between areas on the litter when hens first had access to litter (Campbell et al., 2015b) regardless of whether this was immediately after lights on (mid lay Flock 1) or in late morning when aviary doors normally opened (all other time points for both flocks). In the present analysis, 100 of the 392 total flights observed in mid lay of Flock 1 were recorded during the first 5 h of observation. Video Analysis Video data for both flocks were decoded from 4 of the 6 recorded sections within the aviary house at the 3 production time points (peak lay, mid lay and end lay).

FAILED LANDINGS AFTER LAYING HEN FLIGHT IN AVIARIES 191 Table 1. The 3 classifications of failed landings and descriptions of behaviors observed corresponding to each classification. Failed classification Failed Ground Failed Hen Failed Structure Behavioral description Slipping on the ground when landing, legs are splayed Body colliding with other hen(s) upon landing Kicking other hen(s) with feet during landing approach Landing directly on top of other hen(s) Sliding collision with gate or wall Flying into gate or wall Flying into outer perch Attempting to land on outer perch but falling instead Due to the absence of double-row cameras at peak lay of Flock 1, data were collected from the 4 single-row sections only, but at all other time points, data were collected from 2 single-row and 2 double-row sections (Figure 1b). The video was watched continuously from aviary opening until lights off (except in Flock 1 at mid lay when observations began at lights on because aviary doors were already open), and all instances of flight within a section were documented. Flight was identified as elevation of a hen s body fully off the ground, involving flapping of the wings, and covering distance. This definition distinguished flights from wing-assisted jumps onto and off of perches. Flights were classified based on their origin and landing locations with four types of flights described: from the outer perch to the open litter ( perch to litter ), from the outer perch to the same outer perch or another outer perch ( perch to perch ), from the open litter to another location in the open litter ( litter to litter ) and from the open litter to the outer perch ( litter to perch ). Based on how the hen landed, a flight was further classified as successful : the hen landed on both feet and did not collide with any physical structure or other hen, or failed : the hen did not land successfully. Failed landings were divided into three categories ( Ground, Hen or Structure ) based on several different behavioral situations outlined in Table 1. Ethics All research was approved by the Michigan State University Institutional Animal Care and Use Committee prior to the start of data collection. Data Analyses Data from each flock are presented separately due to the substantial differences in management between flocks. Within each flock, data from 3 time points are presented: peak lay, mid lay and end lay, with a section as the experimental unit. Due to observations being conducted on 4 sections only at each time point within a flock, we did not have enough power to conduct meaningful statistical analysis of our data, but instead present descriptive statistics (means ± SE). Total number of flights and the percentage of failed landings within each section were used to illustrate overall differences in flight occurrences and failed landings between single versus double rows (each flock separately), to compare differences between time points for the types of failed landings and to compare differences within each time point depending on the origin and landing location. Chi-square tests were used to evaluate differences in litter coverage (by pooling the measures made under each of the three cameras/section) at each time point separately for Flock 1 and Flock 2. A paired t-test and repeated measures ANOVA were used to determine differences in litter depth at the two measured time points (mid lay and end lay) of Flock 1 and at all time points of Flock 2, respectively. All data analyses, distributions, and graphs were completed using JMP 11.1.1 (SAS Institute Inc., Cary, NC with α set at 0.05. RESULTS Behavioral observations yielded a total of 1,588 flights recorded from Flock 1 with 9.1% failed landings overall, and a total of 4,695 flights were recorded from Flock 2 with 21% failed landings overall. For both flocks, the total number of observed flights decreased as the hens aged with the greatest decrease occurring between peak and mid lay (Figure 2). A higher percentage of flights ended in failed landings at end lay compared to peak lay and mid lay of Flock 1 (peak lay: mean = 7.62, ± SE = 0.69; mid lay: 7.84 ± 1.92; end lay: 13.50 ± 0.84; Figure 2). In Flock 2, similar percentages of failed flights were seen across all time points (peak lay: 15.99 ± 7.04; mid lay: 24.58 ± 4.04; end lay: 20.33 ± 5.28; Figure 2). We pooled data from the 2 time points of Flock 1 where both single and double rows were observed (mid lay and end lay) and from all time points of Flock 2 (peak lay, mid lay and end lay) and found similar total percentages of failed landings occurring between hens housed in the single and double row types for Flock 1 but proportionally more failed landings were observed in the double rows than in single rows of Flock 2(Figure3). We pooled all flight data from both single and double row types to show that regardless of flock, time point, or section type the majority of flight failures were due to collisions with other hens (Figure 4). Failed flights due

192 CAMPBELL ET AL. Figure 2. The total number of observed flights, combined across all double and single row sections that had successful or failed landings for hens at peak lay, mid lay and end lay in Flock 1 and Flock 2. NB: During peak lay of Flock 1, only single row sections were observed. to slipping on the ground were not observed in Flock 1. Percentages of failed landings due to collisions with hens and physical structures appeared similar over time in Flock 1. However, there were higher percentages of failed landings resulting from both ground slippage and structure collisions at peak lay of Flock 2 than at other times (Figure 4). Additionally, more collisions with hens were seen at mid lay and end lay compared to peak lay in Flock 2 (Figure 4). When flights ended on perches, more failed landings were generally seen than when flights ended on litter (Figure 5). The perch to perch flight type was only seen at peak lay of both flocks and only one instance was recorded from Flock 1. In Flock 1 at peak lay the percentages of flights ending with failed landings were similar across the different flight types, but at mid lay and end lay of Flock 1, more litter to perch flights ended in failed landings compared to the other flight types (Figure 5). In Flock 2 at peak lay, higher percentages of perch to perch and litter to perch flights ended in failed landings compared to flights ending on litter (Figure 5). The percentages of flights ending in failed landings depending on flight origin and landing location were similar at mid lay and end lay for Flock 2(Figure5). The smallest proportion of the open floor area was covered by litter substrate at peak lay in both Flock 1(χ 2 (6, n = 36) = 36, P <.0001) and Flock 2 (χ 2 (2, n = 36) = 28.99, P <.0001), and by mid lay in both flocks, the open floor area was fully covered (Table 2). Litter depth increased over time for both Flock 1 (t(11) = 5.91, P < 0.001) and Flock 2 (F 2, 33 = 74.93, P < 0.001; Table 2). DISCUSSION Our observations of laying hens in a commercial aviary system showed that birds did use the open litter area for flight, but that 9 to 21% of flights did not end with successful landings. Hens continued to fly throughout the lay cycle, and though the number of flights declined over time, the greatest difference was between peak and mid lay, with similar numbers of flights between mid lay and end of lay. There was a higher proportion of failed landings at the end of the lay cycle during Flock 1, but these differences were not observed during Flock 2. More flights were observed in Flock 2 (4,692) than in Flock 1 (1,588) and, in Flock 2, a higher proportion of failed landings also occurred (21% compared to 9.1% for Flock 1). The number of flights observed at peak lay for Flock 2 (3,389) was 4 times as many flights as was observed during peak lay in Flock 1 (833). However, we are unable to parse out whether this effect was the result of different timing of sampling during peak lay between the two flocks or the result of having data only from single-row sections in Flock 1, which contained half as many hens and half as much open litter area as double-row sections. Most likely the difference in number of flights observed between the two

FAILED LANDINGS AFTER LAYING HEN FLIGHT IN AVIARIES 193 Figure 3. The mean (± SE) percentage of all flights with failed landings presented based on whether the failures occurred within single or double row sections. Within Flock 1, data were pooled across the 2 sampling time points where both single and double rows were observed (mid lay and end lay), but pooled across all 3 time points in Flock 2 (peak lay, mid lay and end lay). Table 2. The mean (± SE) litter depth (cm) measurements and % of the sampled litter areas in focal sections that were assigned the different floor litter coverage scores ( poor was defined as < 25 to 49.9% of the area covered by litter, full was defined as 100% of the open litter area covered by litter. Measurements and scores are presented for the 3 time points of both flocks, but no litter depth measurements were taken for peak lay of Flock 1. Litter coverage scores Flock Time point Depth (cm) Full Good Moderate Poor One Peak lay N/A 83.34% 8.33% 8.33% One Mid lay 3.32 ± 0.36 100% One End lay 5.33 ± 0.26 100% Two Peak lay 0.52 ± 0.25 100% Two Mid lay 1.27 ± 0.18 100% Two End lay 3.60 ± 0.26 83.33% 16.67% flocks is a combination of these factors; however, this difference in absolute numbers of flights did not appear to affect other patterns in the data. Although we collected data at similar stages of production for both flocks, video recording at peak lay for Flock 1 took place 2 wk after hens were first allowed out into the open litter area; while for Flock 2 we were able to record on the day of aviary opening. This difference in timing could be responsible for the high flight numbers observed at peak lay of Flock 2 relative to other flight numbers for several reasons. First, all hens were reared as pullets in aviaries with litter access that allowed flight before being placed into closed tiered enclosures for 8 to 10 wk. When litter access was granted, these hens may have shown a behavioral rebound and flown more frequently immediately after aviary opening than they would after acclimation (Nicol, 1987). This rebound would have been observed in Flock 2, as we recorded on the day of aviary opening, but not in Flock 1 when recording occurred 2 wk later. It is also possible that as the hens experienced bone breaks over time (regardless of reason), flight may have become more physically difficult or painful (Nasr et al., 2012; Richards et al., 2012). By peak lay ( 25 to 27 wk) hens in aviary systems are already reported to have keel fractures or abnormalities (Fleming et al., 2004). It may be that a difference of a few weeks in age could have led to differences in keel damage that reduced flight by

194 CAMPBELL ET AL. Figure 4. The mean (± SE) percentage of all failed landings (totaling 100%) presented based on whether failures were attributed to slippage on the ground, collision with other hens or collision with the aviary structure for peak lay, mid lay, and end lay in both flocks. Within each flock, data from single and double row sections were pooled together at each time point. Figure 5. The mean (± SE) percentages of each flight type (based on flight origin and ending) that ended in failed landings, presented for peak lay, mid lay and end lay in both flocks. Within each flock, data from single and double row sections were pooled together at each time point. PL = perch to litter, PP = perch to perch, LL = litter to litter, and LP = litter to perch. Only 1 instance of PP was recorded for Flock 1 (peak lay) with all other observations of PP seen in peak lay of Flock 2. The numbers above each bar indicate the total number of flights that were observed for each category of flights.

FAILED LANDINGS AFTER LAYING HEN FLIGHT IN AVIARIES 195 hens in Flock 1 at peak lay (27 wk versus 24 wk of age for Flock 2); although keel palpation data from Flock 2 (only) in this study showed that 15% of sampled pullets had keel fractures at placement (Blatchford and Mench, 2014). Thus, future research could examine hen flight both immediately after aviary opening and a few weeks later to determine if there is a large drop in flights in the same group of birds within a short period of time. These observations coupled with keel palpations might determine if fractures play a role in flight reduction or if recordings at aviary opening are just capturing a rebound effect of birds accessing litter. Pertaining to the reduction of flights from peak lay to mid lay, it is feasible that if hens sustained injuries during the early months of the production cycle, possibly beginning even in the first few days of litter access, they may not have healed properly and, thus, reduced the ability of hens to flap their wings properly to generate enough lift to fly (Sandilands et al., 2009). If these fractures occurred early in the lay cycle this may account for the small difference in flight numbers seen between mid lay and end lay periods. In fact, the presence of old, healed, keel fractures at the end of lay, which are more common in non-caged hens than in caged layers is thought to be the result of flight and landing accidents sustained throughout the lay cycle (Gregory et al., 1990), and this may also account for the higher proportion of failed landings seen at end lay in Flock 1. Keel palpations on a sample of hens from this system showed 45% of birds from Flock 2 had sustained fractures by the end of lay (Blatchford and Mench, 2014). Finally, it is possible that the reduction in flights was observed as the birds had less energy for any activities beyond production (Schütz and Jensen, 2001). We observed a higher number of landing failures in double-row sections within Flock 2. Experimental studies have shown that landing success of hens decreases significantly at distances greater than approximately 100 cm (Scott and Parker, 1994; Taylor et al., 2003; Moinard et al., 2004a, b). Each aviary section in the current study had an open litter area 1,440 cm in length, thus, hens could be landing poorly as a result of trying to fly distances beyond successful flight thresholds. However, the increased width in the double rows (240 cm vs 120 cm in single rows) could contribute to the higher number of failed landings seen in double-row sections in Flock 2, particularly as hens may have been motivated to fly from one tiered enclosure across the litter to the other. This suggestion may be substantiated by the fact that perch to perch flights were only observed at peak lay in Flock 2, and most of these flight types ended in failed landings. (Though, perch to perch flight could technically occur as hens moved between locations on the same outer perch, this was very rare.) This high perch to perch failure rate could have led to injury or pain or experiences that kept hens from continuing to attempt these flights. It would be of interest in future studies to assess the average distance traveled by hens flying in these systems to examine whether increased room leads to longer flights (or do they not typically fly beyond previously described successful flight thresholds). Collisions with other hens during landing, which were more frequent than collisions of other types, may have been partially the result of patchy distribution of hens on the floor. Channing et al. (2001) showed that irrespective of housing densities, laying hens tended to cluster in groups, not dispersing evenly across areas. In the commercial aviary studied here, hens were observed to form large piles or clusters on the open floor, and they also dust bathed in tight-knit groups (Campbell et al., 2015a). Flying hens would often land directly on top of these piles or dust bathing groups. The lack of space between hens on the floor, combined with movement of hens into landing areas targeted by hens in flight, as well as poor landing control due to high wing loading (Moinard et al., 2004a) could all contribute towards the hen collisions in this system. There were twice as many birds in these double row sections, and thus we observed an increased number of flights. However future studies could correlate the number of hen-to-hen collisions directly with hen occupancy of the litter area. This would enable determination of whether more hens on the litter in double row sections lead to higher collisions simply due to increased environmental chaos, even if actual percentage of litter occupied is similar between double and single row sections. At peak lay in Flock 2, more collisions into tiered enclosures were seen, as well as more slipping on the ground. These differences compared to later time points may have been due to both hen learning and environmental conditions. The pullets placed in this commercial system were aviary-reared with perches and floor access from 6 to 18 wk of age, suggesting they had opportunity to develop spatial navigation skills for more accurate flight (Gunnarsson et al., 2000; Colson et al., 2008). However, learning to navigate a new spatial environment would likely still be required following their move to the laying facility to avoid gates and supports between sections, walls, and upper cage tiers. Experimental research on chicks and hens respectively, found improved navigation of complex environments (Wichman et al., 2007) and improved accuracy of perchjumping over time (Scott et al., 1997). If pullets were reared in conventional cages prior to aviary housing during production, we might anticipate seeing an even higher prevalence of failed landings, particularly during initial access to litter (Gunnarsson et al., 2000). Differences in ground-slipping failures might be attributed to the environment via litter coverage of the floor. Floor coverage and litter depth increased as the flock cycle progressed, which provided a landing substrate with more friction compared to the initial smooth concrete with a scattering of wood shavings when hens were first let out. Hens have worse success landing on perches with smooth surfaces (Scholz et al., 2014), which likely translates to floor landings as well. This is supported by our findings that almost no collisions from slipping on the

196 CAMPBELL ET AL. floor occurred in Flock 2 during mid lay and end lay when litter entirely covered the open area at a depth of 2 cm or greater. Further, hens in Flock 1 showed no ground-attributed failures at any time. Litter coverage in the open area was comparatively higher at peak lay for this first flock, which was a time point that showed many ground-attributed failures in Flock 2. Lastly, our observations showed differences in failure rates depending on flight landing location, with more failures for flights ending on the outer perch compared to litter landings. This result is consistent with laboratory studies documenting the difficulties hens have landing on perches when obstructed by inanimate objects or other hens (Moinard et al., 2005). In our commercial system, there were large numbers of hens perching on the outer perch or jumping onto or off it as they entered and exited the enclosure (Campbell et al., 2015b). These conditions created a dynamic landing environment and, thus, hens may have struggled more to land on the outer perch than in the open litter area. Failed landings within a commercial aviary are important for hen welfare as these collisions might cause bruising, breaks or fractures injuring both the flyer and other birds (Gregory and Wilkins, 1996). In particular, the keel bone is prone to damage, which is currently a widespread concern for laying hen welfare and production (Wilkins et al., 2004, 2011; Nasr et al., 2012; Nasr et al., 2013). Researchers examining hens at the end of lay have reported incidences of old keel bone fractures ranging from 49 to 74% in a variety of extensive housing systems within the European Union (Freire et al., 2003; Wilkins et al., 2004; Nicol et al., 2006), and hens in aviaries have significantly higher rates of keel fractures than those housed in floor litter pens (Käppeli et al., 2011). By the end of the production cycle, up to 95% of sampled hens of various commercial strains in non-cage systems sustained a keel fracture at some point (Wilkins et al., 2011). In agreement with these findings, hens within the commercial aviary in this study sustained more keel fractures (45%) than those in enriched colony or conventional cage housing (29% and 13%, respectively, Blatchford and Mench, 2014). However, these fractures may be due to multiple factors including flight, and research has not yet established a causative link between failed landings and keel breaks. Open space in alternative housing can provide freedom to express flight behavior, but many flights end with failed landings. As multiple variables influence whether hens fly and land well, we must manage and design aviary systems to ensure hen welfare is improved, not compromised. To implement these systems sustainably for production and hen welfare, we need to minimize the risk of injury related to flight, including failed landings. Simpler tiered enclosures might help reduce collisions with metal elements of the system (Wilkins et al., 2011). Shorter open space distances, available for flight but below the threshold distances at which failed landings have been shown to increase in laboratory studies, might also reduce the number of failed landings (Moinard et al., 2004a,b); although these distances would need to be confirmed between different strains and ages of hens. Providing litter to cover open floor areas (before litter naturally accumulates) during initial access to the floor may also reduce injuries occurring as a result of high flight numbers and slipping on the ground. We acknowledge we did not analyze data statistically nor have data on the day-to-day variation in occurrence of flights. However, the overall decrease in flight frequency with time for both flocks suggests consistency in hen behavior across the production cycle, in keeping with typically reported decreases in hen activity with age (Channing et al., 2001). Future research could examine the degree of day-to-day variation in flights and failed landings for comparison with changes resulting from hen age. Finally, more detailed description of hen flight behavior surrounding initial aviary openings is needed to determine whether hens do truly fly more immediately after gaining litter access than at other times and, if so, what leads to decreased flights and whether there are risk factors that can be managed. ACKNOWLEDGMENTS We would like to thank H. Albeer, K. Bradley, S. Dorey, M. Heydenburg, A. Hinson, E. Stefansky, and D. Voishich for assistance with video data collection and C. Daigle, K. Dunn, M. Erasmus, P. Regmi, L. Turner, and K. Wurtz for onsite data collection and camera installation (all from Michigan State University, East Lansing). We thank anonymous reviewers for their suggestions to improve this manuscript. REFERENCES Blatchford, R. A., and J. A. Mench. 2014. Welfare Quality Assessment of hen condition across housing systems. (abstract). In Proc. Poult. Sci. Annual Meeting, Jul 14 17, 2014, Corpus Christi, Texas, USA. Campbell, D. L. M., M. M. Makagon, J. C. Swanson, and J. M. Siegford. 2015a. Litter use by laying hens in a commercial aviary: dust bathing and piling. Poult Sci. In press. doi: 10.3382/ps/pev183. Campbell,D.L.M.,M.M.Makagon,J.C.Swanson,andJ.M. Siegford. 2015b. Laying hen movement in a commercial aviary: enclosure to floor and back again. Poult. Sci. In press. doi: 10.3382/ps/pev186. Carmichael, N. L., A. W. Walker, and B. O. Hughes. 1999. Laying hens in large flocks in a perchery system: influence of stocking density on location, use of resources and behaviour. Br. Poult. Sci. 40:165 176. Channing, C. E., B. O. Hughes, and A. W. Walker. 2001. Spatial distribution and behaviour of laying hens housed in an alternative system. Appl. Anim. Behav. Sci. 72:335 345. Colson, S., C. Arnould, and V. Michel. 2008. Influence of rearing conditions of pullets on space use and performance of hens placed in aviaries at the beginning of the laying period. Appl. Anim. Behav. Sci. 111:286 300. Fleming, R. H., H. A. McCormack, L. McTeir, and C. C. Whitehead. 2004. Incidence, pathology and prevention of keel bone deformities in the laying hen. Br. Poult. Sci. 45:320 330. Freire, R., L. J. Wilkins, F. Short, and C. J. Nicol. 2003. Behaviour and welfare of individual laying hens in a non-cage system. Br. Poult. Sci. 44:22 29.

FAILED LANDINGS AFTER LAYING HEN FLIGHT IN AVIARIES 197 Gregory, N. G., L. J. Wilkins, S. D. Eleperuma, A. J. Ballantyne, and N. D. Overfield. 1990. Broken bones in domestic fowls: Effects of husbandry system and stunning method in end-of- lay hens. Br. Poult. Sci. 31:59 69. Gregory, N. G., and L. J. Wilkins. 1996. Effect of age on bone strength and the prevalence of broken bones in perchery laying hens. N. Z. Vet. J. 44:31 32. Gunnarsson, S., J. Yngvesson, L. J. Keeling, and B. Forkman. 2000. Rearing without early access to perches impairs the spatial skills of laying hens. Appl. Anim. Behav. Sci. 67:217 228. Jones, D. R., D. M. Karcher, and Z. Abdo. 2014. Effect of a commercial housing system on egg quality during extended storage. Poult. Sci. 93:1282 1288. Käppeli, S., S. G. Gebhardt-Henrich, E. Fröhlich, A. Pfulg, and M. H. Stoffel. 2011. Prevalence of keel bone deformities in Swiss laying hens. Br. Poult. Sci. 52:531 536. Leyendecker, M., H. Hamann, J. Hartung, J. Kamphues, U. Neumann, C. Sürie, and O. Distl. 2005. Keeping laying hens in furnished cages and an aviary housing system enhances their bone stability. Br. Poult. Sci. 46:536 544. Moinard, C., R. M. D. Rutherford, M. J. Haskell, C. McCorquodale, R. B. Jones, and P. R. Green. 2005. Effects of obstructed take-off and landing perches on the flight accuracy of laying hens. Appl. Anim. Behav. Sci. 93:81 95. Moinard, C., P. Statham, and P. R. Green. 2004a. Control of landing flight by laying hens: implications for the design of extensive housing systems. Br. Poult. Sci. 45:578 584. Moinard, C., P. Statham, M. J. Haskell, C. McCorquodale, R. B. Jones, and P. R. Green. 2004b. Accuracy of laying hens in jumping upwards and downwards between perches in different light environments. Appl. Anim. Behav. Sci. 85:77 92. Nasr, M. A. F., J. Murrell, L. J. Wilkins, and C. J. Nicol. 2012. The effect of keel fractures on egg-production parameters, mobility and behavior in individual laying hens. Anim. Welf. 21:127 135. Nasr, M. A. F., J. Murrell, and C. J. Nicol. 2013. The effect of keel fractures on egg production, feed and water consumption in individual laying hens. Br. Poult. Sci. 54:165 170. Newman, S., and S. Leeson. 1998. Effect of housing birds in cages or an aviary system on bone characteristics. Poult. Sci. 77:1492 1496. Nicol, C. J. 1987. Behavioural responses of laying hens following a period of spatial restriction. Anim. Behav. 35:1709 1719. Nicol, C. J., S. N. Brown, E. Glen, S. J. Pope, F. J. Short, P. D. Warriss, P. H. Zimmerman, and L. J. Wilkins. 2006. Effects of stocking density, flock size and management on the welfare of laying hens in single-tier aviaries. Br. Poult. Sci. 47:135 146. Odén, K., L. J. Keeling, and B. Algers. 2002. Behaviour of laying hens in two types of aviary systems on 25 commercial farms in Sweden. Br. Poult. Sci. 43:169 181. Regmi,P.,T.S.Deland,J.P.Steibel,C.I.Robison,R.C.Haut,M. W. Orth, and D. M. Karcher. 2014. Effect of rearing environment on bone growth of pullets. Poult. Sci. (Accepted). Richards, G. J., L. J. Wilkins, T. G. Knowles, F. Booth, M. J. Toscano, C. J. Nicol, and S. N. Brown. 2012. Pop hole use by hens with different keel fracture status monitored throughout the laying period. Vet. Rec. 170:494 498. Sandilands, D. V., C. Moinard, and N. H. Sparks. 2009. Providing laying hens with perches: fulfilling behavioural needs but causing injury? Br. Poultry Sci. 50:395 406. Scholz, B., J. B. Kjaer, and L. Schrader. 2014. Analysis of landing behavior of three layer lines on different perch designs. Br. Poult. Sci. 55:419 426. Schütz, K. E., and P. Jensen. 2001. Effects of resource allocation on behavioural strategies: a comparison of Red Junglefowl (Gallus gallus) and two domesticated breeds of poultry. Ethol. 107:753 765. Scott, G. B., N. R. Lambe, and D. Hitchcock. 1997. Ability of laying hens to negotiate horizontal perches at different heights, separated by different angles. Brit. Poultry Sci. 38: 48 54. Scott, G. B., and C. A. L. Parker. 1994. The ability of laying hens to negotiate between horizontal perches. Appl. Anim. Behav. Sci. 42:121 127. Taylor, P. E., G. B. Scott, and P. Rose. 2003. The ability of domestic hens to jump between horizontal perches: effects of light intensity and perch colour. Appl. Anim. Behav. Sci. 83:99 108. Wichman, A., M. Heikkilä, A. Valros, B. Fockman, and L.J. Keeling. 2007. Perching behaviour in chickens and its relation to spatial ability. Appl. Anim. Behav. Sci. 105:165 179. Wilkins, L. J., S. N. Brown, P. H. Zimmerman, C. Leeb, and C. J. Nicol. 2004. Investigation of palpation as a method for determining the prevalence of keel and furculum damage in laying hens. Vet. Rec. 155:547 549. Wilkins, L. J., J. L. McKinstry, N. C. Avery, T. G. Knowles, S. N. Brown, J. Tarlton, and C. J. Nicol. 2011. Influence of housing system and design on bone strength and keel bone fractures in laying hens. Vet. Rec. 169:414. doi:10.1136/vr.d4831. Zhao, Y., T. A. Shepherd, J. C. Swanson, J. A. Mench, D. M. Karcher, and H. Xin. 2015. Comparative evaluation of three egg production systems: Housing characteristics and management practices. Poult. Sci. 94:475 484.