Keel bone fractures induce a depressive-like state in laying hens

In commercial flocks of laying hens, keel bone fractures (KBFs) are prevalent and associated with behavioural indicators of pain. However, whether their impact is severe enough to induce a depressive-like state of chronic stress is unknown. As chronic stress downregulates adult hippocampal neurogenesis (AHN) in mammals and birds, we employ this measure as a neural biomarker of subjective welfare state. Radiographs obtained longitudinally from Lohmann Brown laying hens housed in a commercial multi-tier aviary were used to score the severity of naturally-occurring KBFs between the ages of 21–62 weeks. Individual birds’ transitions between aviary zones were also recorded. Focal hens with severe KBFs at 3–4 weeks prior to sampling (n = 15) had lower densities of immature doublecortin-positive (DCX+) multipolar and bipolar neurons in the hippocampal formation than focal hens with minimal fractures (n = 9). KBF severity scores at this time also negatively predicted DCX+ cell numbers on an individual level, while hens that acquired fractures earlier in their lives had fewer DCX+ neurons in the caudal hippocampal formation. Activity levels 3–4 weeks prior to sampling were not associated with AHN. KBFs thus lead to a negative affective state lasting at least 3–4 weeks, and management steps to reduce their occurrence are likely to have significant welfare benefits.

www.nature.com/scientificreports www.nature.com/scientificreports/ addition to spiral leg rings colour-coded for pen (Fieger AG, Untertuttwil, Switzerland), focal hens were marked with flexible legbands (Roxan Developments Ltd, Selkirk, United Kingdom) displaying individual identification numbers. Hens were left for three weeks to habituate to the new housing environment before mobility tracking commenced.
Each pen of the layer barn was equipped with a commercial aviary system (Bolegg Terrace, Vencomatic; Krieger AG, Ruswil, Switzerland). The system consisted of three tiers: (1) a lower tier (73 cm above ground), (2) a nest box tier with integrated group nests (153 cm above ground) and (3) a top tier (220 cm above ground), and was subject to minor modifications from the standard installed model. Specifically, the perch on the top tier was removed and one drinker line was moved from in front of the nest box to the top tier. Food and water were provided ad libitum through automatic feeding chains and nipple drinkers on the top and lower tier (with feed refreshed every two hours during light hours). Artificial light was provided from 02:00 h until 17:00 h, with a 5 minute dawn (02:00 h-02:05 h) and a 30 minute dusk (16:30 h-17:00 h) phase. Natural daylight was managed via curtains in front of the windows, which were open from 08:00 h until 16:00 h. Perches consisted of round metal rails (diameter: 3.2 cm, length: 230 cm) located on the top tier for roosting (six perches at 270 cm height, two perches at 300 cm height) and across the system to facilitate movement between tiers (three perches at 190 cm and 125 cm above ground on each side of the aviary, on top of the feeder on the lower tier). In total, 14 cm of perch space per hen was provided. The floor besides and underneath the aviary was covered with wood shavings (approx. 10 cm deep). Stocking density was 7.4 hens/m 2 of accessible floor space, which included the littered floor area and all mesh grid areas in the lower and top tiers. Pecking opportunities were provided in the form of autoclaved aerated concrete stones (Xella Porenbeton Schweiz AG, Zurich, Switzerland) and ad libitum mineralized pecking stones (FORS 228 Pickschale Geflügel; Kunz Kunath AG, Burgdorf, Switzerland). To increase opportunities for explorative behaviour (scratching, pecking), straw was supplied in racks placed in the litter area. The wintergarden provided a covered area (9.32 m 2 ) external to each pen, containing wood shavings and a dust bathing area filled with sand, and was accessible via popholes (15 cm above ground level) which opened automatically at 10:00 h and were closed manually between 16:00 and 16:30 h. Data collection. Data were collected at 11 time points during the production cycle. For one pen (20 focal birds), this occurred when the birds were 21,24,27,31,35,39,44,48,52,57, and 61 weeks of age. For practical reasons, data from the remaining two pens (40 focal birds) were collected the subsequent week, i.e., 22,25,28,32,36,40,45,49,53,58, and 62 weeks of age. Data on individual mobility were collected for six days per time point, and on the final day, hens were radiographed to detect fractures. individual mobility. Individual mobility was recorded using a custom-made infrared (IR) tracking system which has been previously described and validated 65 . Infrared emitters were installed on the vertical grid panels dividing pens and generated infrared beams encoded with specific signals for the five pre-defined zones: litter, lower tier, nest box, upper tier, and wintergarden. Infrared receivers were mounted on the legbands of focal hens and recorded the zone-specific signals produced by the IR emitters with a frequency of 1 Hz, along with the date and time of each zone change. The system therefore recorded vertical transitions made across the aviary, but did not track horizontal movement within zones. Receivers were covered by a small plastic container to protect from dust, moisture and faeces (outer diameter: 3.1 cm, height: 2 cm; Semadeni AG, Ostermundigen, Switzerland) throughout the experiment. Containers were replaced if they became opaque due to dirt or scratches. Equipment mounted on the hens weighed 9.4 g, well below the suggested limit of 5% of body mass 66 .
At each of the 11 study time points, hens were caught and equipped with IR receivers on the day preceding mobility data collection (day 1). This allowed time for both habituation to the additional weight of the receiver and re-establishment of normal mobility behaviour after handling, given this experience has been shown to impact tonic immobility behaviour occurring directly afterwards 67 . Monitoring devices have been found to reduce exploration by adult hens on the day of fitting, but to have negligible effects on behaviour from two days onwards 68 . Mobility data were therefore collected from day 2 to day 7, before receivers were removed and data were downloaded as CSV files on day 8.
Keel bone assessment/fracture severity scoring. At each time point, after removing the IR receiver on day 8, hens were radiographed to detect keel bone fractures using a mobile X-ray unit (GIERTH HF 200 ML; x-ray tube Toshiba D-124 with maximal acceleration voltage of 100 kV; x-ray plate Canon CXDI-50G; software Canon CXDI Control Software NE; distance: 80 cm, voltage: 46 kV/2.4 mAs). Hens were firmly held by both legs, carefully turned upside down and fixated in padded metal shackles to induce immobility 69 . The procedure took approximately 10-20 seconds per radiograph and occurred 11 times (T1-T11) throughout each hen's lifetime. In a previous study, KBF severity at 61 weeks did not differ between repeatedly-radiographed focal hens and non-focal hens radiographed only at this time, indicating that multiple radiographs had no negative impact on keel integrity 70 . Though focal LSL hens produced fewer eggs, it is not clear if this related to stress of the radiograph procedure or of being the minority hybrid, whilst egg production in focal LB hens (used in the present study) was not affected 70 .
Radiographs were imported to the PACS (Picture Archiving and Communication System; IMPAX EE, Agfa Healthcare, Bonn, Germany) of the Department of Clinical Radiology (Vetsuisse Faculty, University of Bern) as DICOM files. For subsequent analysis, radiographs were downloaded from the PACS as JPEG files.
Radiographs were analyzed according to aggregate fracture severity and the presence of a visible fracture gap. The observer was blind to both the age and identity of the hen. Aggregate fracture severity was assessed using a tagged visual analogue scale ranging from "no fracture" to "extremely severe", resulting in a continuous variable ranging from 0.0 to 10.0. The system and its validation is described in detail by Rufener et al. 71 . Eleven days after final radiographs were conducted, preliminary analysis of processed radiographs up to the penultimate time (2020) 10:3007 | https://doi.org/10.1038/s41598-020-59940-1 www.nature.com/scientificreports www.nature.com/scientificreports/ point (T10) was used to select 12 birds with minimal KBFs and 12 birds with severe KBFs for brain sampling. This sample size previously afforded sufficient power to detect an effect of chronic stress on the same AHN outcome measure in laying hens 50 . It was not possible to select a control group of hens with no fractures, as although these were very minor in some cases, birds entirely free of KBFs did not exist by the end of the study. The higher proportion of KBFs detected here compared to previous studies may relate to the sensitivity of the relatively novel radiography technique, compared to less reliable methods such as palpation 71 . tissue collection & processing. Following a delay of 3.5 (pen 1) or 4.5 (pens 2 & 3) weeks since the final x-ray, focal hens (n = 24) were killed via an intravenous injection of pentobarbital (Esconarkon, 0.3 ml/hen). Immediately thereafter, brains were removed from the skull, placed into 0.1 M PBS in a Petri dish and divided along the longitudinal fissure with a scalpel. The hippocampus from one hemisphere was allocated for molecular biology, with results to be reported elsewhere. The remaining hemisphere from each brain was immersion fixed for 44-48 h in 4% paraformaldehyde in 0.5 M Phosphate Buffered Saline (PFA -PBS) at 4 °C in preparation for immunohistochemistry. To balance possible lateralisation of AHN and its responsivity, tissue collected for each purpose was alternated between the left and right hemisphere within the minimal and severe KBF groups. Samples were then cryoprotected in a solution of 30% sucrose in 0.5 M PBS, before being embedded in OCT (4583, Electron Microscopy Sciences -USA). Coronal sections (50 μm) were cut on a cryostat (HM 550, Microm -Germany) and stored in cryoprotectant solution (30% glycerol, 30% ethylene glycol, 0.1M PBS) at −20 °C. Serial sections taken at 400μm intervals were processed for immunohistochemistry.
immunohistochemistry. To quantify the number of surviving newborn cells generated through AHN, sections were stained for doublecortin (DCX), an endogenous protein marker of migratory immature neurons in the avian brain 63 . Exercise upregulates proliferation in the mammalian hippocampus, and a sustained effect of running on subsequent newborn cell survival has been reported to persist for five weeks after discontinuation of this exercise in mice 72 . Since DCX is expressed for approximately 4 weeks from the start of neuronal differentiation 63 , activity levels at the most recent time-point (~3-4 weeks prior) were those most likely to relate to survival of the stained cell population present at the time of death, and were thus analysed in relation to AHN. Staining was conducted over 6 batches, each containing tissue from 4 birds, balanced for KBF status. Free-floating tissue slices were washed in 0.1 M PBS to remove cryoprotectant (3 × 5 mins), then underwent 30 minutes endogenous peroxidase inhibition in 1% H 2 O 2 (Sigma-Aldrich, UK). Samples were again washed in PBS before 60 minutes incubation in blocking solution, containing 2% normal goat serum and 1% Bovine Serum Albumin [BSA] dissolved in 0.1 M PBS that contained 0.3% Triton X-100. After a quick rinse in distilled H 2 O, samples were incubated overnight in rabbit polyclonal to doublecortin primary antibody (Abcam Cat# ab18723, RRID:AB_732011) at concentration 1:1000 (4 °C). Following washes in PBS, samples were incubated for 120 minutes at room temperature in 1:500 biotinylated anti-rabbit IgG secondary antibody, made in goat (Vector Labs, BA-1000). Samples were washed in PBS before conjugate enzyme incubation in 1:250 Horse Radish Streptavidin (Vector Labs, SA-5004) for 60 minutes. Following washes in PBS and dH 2 O, 30 seconds chromogen incubation in 3,3′-Diaminobenzidine (DAB) was conducted by diluting SIGMAFAST tablets in pure water to final concentration of 0.35 mg/ml. Tissue was rinsed immediately in dH 2 O to stop the reaction. Following final washes in 0.1 M PBS, slices were mounted onto gelatine-subbed slides using a paintbrush in dH 2 O. Once dry, slides were soaked for 2 × 5 minutes in Histoclear before coverslipping using Eukitt ® (03989 FLUKA). Excess mounting medium was cleaned from the slides using a razor blade after drying.

Quantification of AHN.
For every animal, 4 to 6 hippocampal sections 800 μm apart were systematically analysed, starting with the rostral-most section bearing hippocampal tissue. This sampling spanned roughly 1/16 th of the rostral and 1/8 th to 1/4 th of the caudal HF, given the latter region curls around the back of the forebrain and is thus contained within fewer coronal slices. Hippocampal slices were analysed with an optical microscope (Leica DM6B-Z, Germany) equipped with a digital video camera (Leica DFC450 C, Germany) and motorized stage system (Leica AHM, Germany) to step through sections for systematic sampling. Quantification was performed by a single observer (EA), blind to the KBF status of the animals.
Image analysis was performed with Stereo Investigator software (version 2018.1.1, MBF Bioscience, USA). Hippocampal borders were outlined at 2.5X magnification (0.07 numerical aperture) on every analysed slice according to the chick stereotaxic atlas 73 . Because of the complex structure of the avian HF, we divided the whole structure in two major components: i) the rostral hippocampus (RH -interaural 5.68/0.50) and ii) the caudal hippocampus (CH -interaural 0.50/−0.50). Cell counting was performed at 100X magnification (0.65 numerical aperture). Stereological parameters were set to an optical fractionator grid of 120 × 120 µm for RH and 240 × 240 µm for CH, a counting frame of 50 × 50 µm for both regions and a mounted thickness of 20 µm.
Although no granule cells are present in the avian HF, different types of DCX + cells have been previously described in avian brain literature 74 and can be divided into two groups according to neuronal morphology: (I) multipolar neurons and (II) bipolar (fusiform) cells. In line with Boseret et al. 74 , we assume that the fusiform neurons are younger and still migrating, while the multipolar neurons are more mature and settling. Multipolar cells were defined as medium-large sized cells, with a round or polygonal/angular cell body shape and process branching from it in three or more directions. Golgi analysis of the chick HF 75 suggests that this group includes a high proportion of multipolar projection neurons, but may also incorporate pyramidal and multipolar local circuit neurons. Bipolar/fusiform cells were defined as medium-small sized cells with elliptical or oval cell body shape and process branching from it in two or fewer directions. Cells of these two types lying inside the optical fractionator frame or bisected by its green lines were counted, according to the Optical Fractionator method.
Counts were exported to MS Excel from Stereo Investigator and used to manually calculate densities of each cell type. Specifically, the number of counted cells of each type was divided by the area of the counting frame (2020) 10:3007 | https://doi.org/10.1038/s41598-020-59940-1 www.nature.com/scientificreports www.nature.com/scientificreports/ (2500 µm), multiplied by both the number of counting sites sampled in that brain and the section thickness (50 µm), to produce a density per volume measure for the sampled tissue. Values were transformed to densities per cubic millimetre by multiplying by 10 9 . The rostral and caudal hippocampus were treated separately.
Statistical analysis. When analysing radiographs from the final time point (T11) after tissue collection, it was necessary to move 3 sampled birds from the minimal to the severe KBF group for the purpose of statistical analysis. Final sample sizes for between-group analyses were thus 9 minimal KBF and 15 severe KBF hens. Descriptive statistics are displayed as mean (M) ± standard deviation (SD), and all analyses were run in IBM SPSS Statistics (v24). To compare activity of the minimal and severe KBF groups at time point 1 (before most fractures occurred) and time point 11, a linear mixed model (LMM) was conducted for the mean aviary transitions made during the 6 days preceding each. Experimental pen was included as a random factor, with time point as a repeated fixed factor and final KBF severity group as a between-subject fixed factor. To assess the relationship between KBF severity and AHN, separate LMMs were conducted for raw densities of DCX-expressing multipolar and bipolar cells. In all models, staining batch and experimental pen were included as random factors, whilst HF subregion (rostral/caudal) was included as a repeated fixed factor. When exploring group differences, fracture status at the final time point (minimal/severe KBF) was included as a between-subject fixed factor, and in an interaction term with HF subregion. To assess within-individual relationships, KBF severity score at the final time point (T11) and the mean number of transitions made between aviary zones over the preceding 6 days were entered as covariates in separate models, each of which included their interaction with HF subregion. To determine which variable best predicted cell densities, a model including both covariates (and their interactions with subregion) was also run. Several analyses were conducted in order to explore the influence of the timescale of KBF acquisition on AHN at the end of life. Firstly, the time point at which a hen suffered its first KBF was included as a covariate in LMMs for the two cell types, again alongside staining batch and pen as random factors, HF subregion as a fixed factor and the interaction between time point and subregion. Secondly, to determine at which time point fracture severity best predicted DCX + cell densities, and if severity at any other times added explanatory power, LMMs including severity score at each time as a covariate were conducted in a stepwise forward manner. Lastly, the changes in KBF severity compared to the previous recorded score at each time point were included as covariates in LMMs, according to the same stepwise procedure. These analyses were conducted separately for DCX + multipolar and bipolar densities. All figures display cell densities normalised (z-scored) within their staining batch.

Results
Subsequent to readjusting sample groups based on KBF severity in radiographs taken at the final time point, hens in the minimal KBF group (n = 9) had a mean fracture severity score of 3.5 (±0.87 SD), with individual scores ranging from 1.9-4.8. In contrast, hens in the severe KBF group (n = 15) had a mean final fracture severity of 8.

Adult hippocampal neurogenesis & KBf severity group.
Hens with severe KBFs 3-4 weeks before tissue sampling had fewer DCX + multipolar cells than counterparts with minimal fractures in the HF as a whole (F 1,14.5 = 22.56, p < 0.001), and there was no effect of HF subregion (F 1,20.2 = 2.93, p = 0.102). There was an interaction between subregion and fracture status (F 1,20.2 = 10.52, p = 0.004, Fig. 2a), whereby hens with minimal fractures had more DCX + multipolar neurons than hens with severe fractures in both the rostral (p = 0.011) and caudal (p < 0.001) HF, but the effect was stronger in the caudal HF. Hens with minimal fractures alone had a higher density of stained neurons in the caudal HF than the rostral region (p = 0.001), with no subregional difference in hens with severe fractures (p = 0.294).
Timescale of KBF acquisition. For hens that had developed severe KBFs by the final time point (T11), the median time point for acquisition of the first fracture was T3, compared to one time point later (T4) for hens with only minimal damage by the end of the study. Though time point at which a KBF was first acquired did not co-vary with AHN over the whole HF (F 1,16.2 = 1.87, p = 0.190), it interacted with subregion to predict DCX + multipolar cell density (F 1,20.0 = 17.63, p < 0.001). Specifically, hens that acquired KBFs earlier and thus had suffered them for longer appeared to have fewer new multipolar neurons at the caudal pole (Fig. 4a). To explore the integration window over which AHN assimilates changes in the experience of pain/stress, the same stepwise LMM procedure was conducted for the difference in KBF severity at each time point compared to  (Fig. 5a,b). Changes in severity between time points 4 & 5 (F 1,19.1 = 13.57, p = 0.002) and 3 & 4 (F 1,14.7 = 7.39, p = 0.016) similarly predicted DCX + bipolar cell densities, though severity difference between time points 9 & 10 did not (F 1,19.4 = 0.322, p = 0.577, Fig. 5c,d).
The predictive power of change in severity between these particular time points did not arise from their contribution to KBF severity at the final time point, as these variables were not correlated (T3

Discussion
implications for animal welfare. Hens with severe KBFs 3-4 weeks before tissue sampling had lower densities of DCX + multipolar and bipolar neurons in the HF when compared to birds with minimal KBFs, while the magnitude of keel bone injury an individual had sustained by this time also linearly predicted their number of surviving new-born cells. Such downregulated AHN occurs as part of a depression-like physiological state, arising from chronic exposure to stress across numerous paradigms in both mammalian 29,31,76 and avian 43,45,48,50 species. As previous work provided behavioural evidence of pain through a place preference for the location of analgesic administration in hens with KBFs 21 , altered HF morphology in terms of reduced AHN in the present study further suggests that the accumulation of fractures is sufficient to lead to long-term stress in affected birds. In mice, a model of chronic neuropathic pain severe enough to suppress AHN was also associated with anxiety-and depressive-like behaviour 32 . The observed neural and affective changes persisted beyond reversal of the injury, indicating that they were not an immediate result of nociceptive stimulation. Given that the hens in the www.nature.com/scientificreports www.nature.com/scientificreports/ present study showed a similar reduction in AHN following chronic pain, an accompanying depressive-like state may also have been induced. Moreover, the magnitude of keel bone damage present 3-4 weeks before end of life appears to predict the degree of suppression in neural differentiation exhibited, which might also reflect the level of stress experienced. Adult-born neurons in the ventral dentate gyrus of the mouse HF inhibit those mature neurons in the same region that respond preferentially to anxiogenic conditions 77 . Though currently unexplored, if adult-born cells in the chicken HF are similarly involved in inhibiting stress-responsive mature neurons in the caudal subregion, this linear relationship may relate to a cumulative loss of the stress resilience afforded by new neurons with increasing magnitude and/or duration of pain.
Previous research has indicated that it is common for hens to sustain multiple fractures to the keel at different points throughout the laying cycle, with a mean of 3.1 (±1.8 SD) KBFs per hen observed at the end of the laying period in a similar study 78 . Multiple fractures were also observed in the present study, but due to variability in dimensions, location, healing status, etc., quantified KBF severity scores collated all fractures present 71 (both existing & new), and thus tended to increase incrementally after damage was first sustained. Though it is not known for how long a KBF continues to be a source of pain after acquisition, hens that developed a conditioned place preference for the location of analgesic administration in a previous study had had KBFs for at least three weeks prior to the point of testing 21 , whilst hard callus formation takes four to six weeks to occur 14 . Radiographs in the present study were taken 4 to 5 weeks apart, leading to two possible interpretations of analyses relating the timescale of KBF severity to AHN. The first is based upon the assumption that, after this period of time, previous fractures have healed to the extent that they are no longer painful or stressful, and thus would only have exerted a suppressive effect on AHN during their recorded time point. Therefore, hens with more recent (still healing) fractures should be more affected, which is supported by the finding that severity of most recent KBFs (radiographed 3-4 weeks prior) were the best predictor of AHN. In this case, where hens with older fractures also show corresponding reductions in AHN, the window over which AHN integrates experience must span this length of time. Our results indicate that AHN at end of life is predicted by KBF severity from around time point 4 onwards, meaning AHN would thus integrate experience from around 34 weeks previously. www.nature.com/scientificreports www.nature.com/scientificreports/ The alternative interpretation is based on the assumption that healed breaks remain somewhat painful, and therefore stressful, over a hen's lifetime. Indeed, existing KBFs may often be repeatedly aggravated during normal hen behaviour, such as movement to food or water, flight and perching 15 . If this is the case, through integrating older and newer fractures, severity score at the final time point in itself reflects the cumulative pain or discomfort experienced. This means it is not possible to determine the time scale over which AHN integrates painful experience, as that occurring in the past persists until the end of life. As such, the contribution of recent versus longer-term pain or stress to the relationship between AHN and KBF severity measured at the latest point (3-4 weeks previously) cannot be distinguished. However, in the caudal HF subregion, a relationship existed between AHN and the duration of time since fractures were first incurred. Moreover, this variable predicted caudal AHN independently of severity at the final time point. Furthermore, changes in KBF severity score since the preceding time point occurring as early as ~37 weeks previously also predicted AHN over the whole HF, and were not correlated with final fracture severity. These relationships were not as strong as those with end of life KBF severity, suggesting that AHN may integrate temporally distant experience on a more subtle level, alongside that which is most recent or cumulative. The finding that the caudal HF is more sensitive to certain aspects of this integration are consistent with its hypothesised greater responsivity to chronic stress 60 . Before future studies definitively elucidate the time scale over which KBFs cause pain, it is not possible to know which assumption is correct, meaning the results from the present study cannot determine the window of time over which AHN integrates negative experience.
Though individual differences in activity levels masked differences between minimal and severe KBF hens in the absolute number of vertical aviary transitions made at the final time point, hens that had developed severe KBFs by this point made fewer transitions than they had at the first time point, before most individuals had suffered any damage. This indicates that birds with severe KBFs, as selected for the present sample, show a corresponding reduction in activity. The assumption that it is painful for them to move is consistent with reported analgesic-driven reductions in latency for hens with KBFs to descend from raised perches 18 . On the other hand, www.nature.com/scientificreports www.nature.com/scientificreports/ lower AHN in hens with severe fractures could not be explained by their recorded activity 3-4 weeks before tissue sampling, and thus cannot be attributed to a consequence of varying levels of activity or concurrent spatial cognitive processing at this time. Our findings thus support the inference that hens with KBFs experience a corresponding negative affective state beyond an acute sensory pain response. Since this is likely to influence overall quality of life for laying hens in commercial systems, management steps to reduce or delay acquisition of KBFs are likely to have a notable impact on welfare.
Hippocampal homologies. In the HF, hens with severe KBFs 3-4 weeks prior to tissue collection had lower densities of DCX + multipolar and bipolar neurons when compared to birds with minimal KBFs. These findings are consistent with the suppressive effect of induced chronic neuropathic pain on AHN in mice 32 and suggest a homologous response to prolonged pain in the avian HF. In line with the recent demonstration of reduced AHN in hens exposed to unpredictable chronic mild stress 50 , the current findings support the conclusion that the regulation of AHN in modern mammals and birds derives from a process established in their common ancestor over 300 million years ago 79 .
In rodents, some models of chronic stress have a ventral-specific influence on AHN, whereas others result in suppression across the whole dentate gyrus (see O'Leary and Cryan 29 for a review). Whether the effect is regionally-localised may depend on the nature of the stressor or paradigm employed. In the present study, KBFs were associated with suppressed AHN over the whole laying hen HF, rather than being restricted to the putatively more stress-sensitive caudal region 60 . In a mouse model, Dimitrov et al. 32 also observed a qualitatively similar reduction in AHN in the dorsal and ventral subregions of the HF following induced chronic pain. Therefore, whereas a sequence of mild, albeit unpredictable stressors (as in Gualtieri et al. 50 ) appears to produce a caudal-specific suppression of laying hen AHN, sources of chronic pain such as KBFs may conceivably present stressors of greater magnitude, which affect neurogenesis across more of the HF.
That being said, for multipolar DCX + neurons, the interaction between individual KBF scores and HF subregion demonstrates that the suppressive effect of these fractures on AHN is stronger at the caudal pole. Moreover, the caudal HF appears more sensitive to temporal aspects of KBF acquisition, with the relationship between AHN and the duration of time since fractures were first incurred only existing in the caudal region.
Accounting for fracture severity scores also revealed an overall greater density of new multipolar neurons in the caudal HF than the rostral subregion, as found in our lab previously 50 . This regional gradient has been observed in certain other avian species, with the same pattern present in zebra finches 80 and two species of North American blackbird 81 . However, adult wild-caught chickadees instead have a higher level of AHN in the rostral HF subregion. Interestingly though, this gradient is eliminated after 6 weeks in captivity 45 . When comparing KBF groups in the present study, hens with minimal fractures alone had more new multipolar neurons in the caudal HF than the rostral region, whereas hens with severe KBFs exhibited no subregional difference. Though further exploration is required, it could be that sources of chronic stress, such as KBFs and captivity, have some form of homogenising effect on subregional AHN rates.
In general, the results obtained were qualitatively similarly for multipolar and bipolar/fusiform cells, assumed to be respectively more mature versus younger and still migrating 74 , though relationships with KBF severity were stronger for the multipolar cells. This may suggest that the chronic stress arising from severe KBFs particularly affects later neuronal cell survival. The relative impact of stress on various stages of AHN appears to differ between paradigms 82 , but psychosocial stress has similarly been reported to downregulate the longer-term survival, but not the proliferation, differentiation or immediate survival, of newly-generated cells in the rat HF 83,84 .
In contrast to fracture severity, the number of transitions hens made between aviary zones 3-4 weeks before tissue sampling was not associated with AHN when employed as a proxy for activity. Such a relationship might have been predicted based on rodent studies, wherein time spent voluntarily running in a wheel correlates positively with AHN 64,85 . However, flight exercise was previously found not to be associated with DCX-expression in starlings 86 . Infrared tracking in the present study also recorded only vertical activity within the aviary, meaning individual differences in activity within a single horizontal zone or tier were not accounted for. Additionally, while voluntary exercise robustly upregulates AHN in rodents, forced exercise does not have this stimulatory effect 35 . Some transitions recorded in the present study may have equated to forced movement, as hens evaded people entering the aviary or were displaced by flock-mates during competition for space or food, with injured hens perhaps particularly affected. Involuntary activity may as such have contributed to the lack of observed relationship. For these reasons, further exploration into the association between exercise and AHN in avian species may be required in future. It would also be worthwhile to consider more complex assessments of hen movement, that account for consistency and changes in patterns over multiple days and longer periods of a hens' life (as in Rufener, et al. 65 ).

conclusions
In commercial laying hens, severe KBFs present 3-4 weeks before sampling are associated with a reduced density of new-born neurons across the HF when compared to flock-mates with minimal KBFs. Additionally, KBF severity scores at this time negatively predict DCX + cell counts across individual hens, while in the caudal HF, hens that developed KBFs earlier tend to have fewer DCX + multipolar neurons. Further information regarding the duration that KBFs continue to be a source of pain or stress will be necessary in order to determine the time scale in which AHN over the whole HF integrates this painful experience. On the other hand, vertical movement within the aviary 3-4 weeks prior to sampling is not associated with AHN levels. Results suggest that, like the rodent hippocampus, the avian HF is sensitive to the experience of pain as a result of KBFs, on a time scale of at least 3-4 weeks. Downregulated AHN lends support to the notion that KBFs present a source of chronic stress, which is associated with induction of a depression-like state in mammals and thus likely detrimental to the affective experience of commercial laying hens. Management steps to reduce or delay acquisition of KBFs in commercial laying hen systems are thus likely to have a notable impact on animal welfare.

Data availability
The datasets generated during and analysed during the current study are available from the corresponding author on reasonable request.