Thinning the thickets: Foraging of hardy cattle, sheep and goats in green alder shrubs

Shrub encroachment is a global phenomenon affecting the structure and function of grasslands (Eldridge et al., 2011). In European Alps, mainly green alder shrubs Alnus viridis [Chaix] DC. expand into formerly open habitats subalpine zone (Brändli 2020; Tasser & Tappeiner, 2002). This shrub not native to most where it occurs today, but only emerged due human impact. post-glacial era, large areas Alps were covered by dense forests in which A. cannot prevail. Thus, was restricted naturally non-forested habitats, namely wet sites avalanche aisles. Bronze Age, early farmers settled (Gilck Poschlod, 2019). Woodcutting livestock grazing thinned finally created below natural tree line. Palaeobotanical pollen records alpine lake sediments demonstrate that appeared during this period, remarkably, simultaneously with typical grassland species (Schwörer 2014). indicates rapidly populated newly deforested, spreading beyond its initial habitats. Livestock thus enabled establishment shrubland, limited expansion pastures because high pressure prevented growth viridis. last century, social agricultural transformations favoured Depopulation mountainous non-agricultural employment have decreased labour availability sector (Muñoz-Ulecia 2021). Today, concentrated favourable, intensively managed locations, whereas less productive are underused (Tasser Consequently, like overgrow these extensively (Anthelme 2003; Zehnder 2020), among biodiverse on earth (Wilson 2012). process has been enforced shift species. Ruminants differ their preferred type forage. Goats, as opportunistic mixed feeders, eat woody plants, grazers sheep, even more notably cattle, supposed avoid them (Lu, 1988; Wood, 1987). During systems changed smallholder goat farming became unprofitable (Boyazoglu 2005). Accordingly, number goats Switzerland 80% since 1876 (Figure 1). They widely replaced cattle and, some extent, sheep. there fewer animals marginal pastures, remaining inclined Grazing assumed effectively mitigate (Wehn 2011; 1987), rarely practiced unprofitability. Clearing scrubs mechanically laborious, expensive often infeasible rough mountain terrain. Therefore, we tested whether effective alternatives mechanical clearance. Animals potentially impair two ways: (a) By visiting stands, damage branches seedlings thereby thin thickets via trampling. body weight, largest for cattle. However, may be too size enter densest parts stands visit steepest slopes. Sheep therefore better able penetrate thicket. Especially goats, whose wild ancestors prefer steep slopes providing hiding places (Zobel 2018), expected shrubland trample shrubs. (b) More efficiently, plant damaged debarking. If bark branch stripped all around, dies off irreversible interruption water nutrient transport. bite fodder using lips teeth, rip grass tongue. debark sheep plants than (Wood, farmer's observations unpublished, preliminary studies suggested different pattern certain hardy breeds, uniquely suited control. Although several analysed how ruminant deal (e.g. Iussig 2015; Rogosic 2006; Wehn lack scientific knowledge specific response We compared foraging movement behaviour positive control low-productive, breeds tolerate harsh environmental poor nutritive conditions pastures. To represent used Dexter 2a), particularly small (withers height cows: <110 cm), light (<350 kg weight) agile dual-purpose breed, cope well terrain (The Cattle Society, Hardy represented autochthonous breed Engadine 2b), originating from Valley, experiment conducted (Engadine association, selected postulations Bühlmann al. (2014) indicating preference Goats 2c) local, Pfauen (purebreds crosses chamois-coloured goat). over arch 3a). The middle area, Eastern Swiss (46.5812°N, 9.9029°E; 13.4 ha; elevation: 1,925–2,200 m a.s.l.; average slope: 55.5%; annual precipitation: 713 mm; temperature: 1.9°C), June August, entire period at elevation. study area 3b) partially including single Sorbus aucuparia L. (Raspé 2000). surrounded heterogeneous, mostly nutrient-poor grassland, dwarf-shrub heathland sparse forest, dominated Larix decidua Mill. Pinus cembra classified vegetation types 3c; details see Appendis S1) according Dietl (2002) mapped combining field observation high-resolution orthophotos. grouped five herds: one herd seven adult she-goats yearlings; herds consisting ewes three lambs each; eight heifers each. defined 15 paddocks estimated forage topographic structures allowed fence installation. Three consecutively grazed rotations. six rotationally, other paddocks. Half (mean paddocks: 0.90, sheep: 0.52 ha) thus, (cattle: 0.63, 0.42 LU ha?1 year?1; 1 unit = 500 kg). half larger 1.45, 0.86 low (0.40 0.25 year?1 respectively). under (0.75 0.21 year?1). maintained higher or consume stocking rates. rate calculated based metabolic live weight (i.e. weight0.75) accounting turnover smaller animals. Because stocked same duration, directly corresponds density. Each paddock for, average, 12 days then moved another paddock. Thereby, alternatingly procedure repeated until had visited each once. twice. occupation given Appendix S2. Shrub-encroached pose challenge conducting controlled experiments they rocky. Since defining fencing out identical impossible environment, spaced inherent heterogeneity evenly across For instance, contained zones quality, slope, percentage cover (for further details, S3). track behaviour, equipped GPS loggers recording position every 10 s. After twice per paddock), quantified debarking location debarked (debarking occurrence) stand total location. Debarking intensity counted relative non-debarked Technical logging assessment Appendices S4 S5. animal approved veterinary office Grison (licence GR_2016_18). No permission needed fieldwork. analyses, observation. treatments rate, (c) (d) (e) (n 15). All calculations completed R 4.0.4 (R Core Team, recorded positions resting simple algorithm speed (see S6 details). Only data subsequent analysis normally caused feeding searching forage) resting. assigned recorded. Subsequently, presence while (RPA) proportion within normalized Values <1 indicate avoidance, indifference >1 Moreover, totalled 5 × grid cells computed Camargo's index (CI) space-use evenness (Pauler, Isselstein, Suter, 2020). tracked positions, RPA CI averaged paddock, weighted positions. Means post hoc Tukey tests 5% significance level package multcomp (Bretz 2016). explain spatial distribution space use debarking, variables resolution: distance pasture, Euclidean boundary between pasture set zero cells; slope swissALTI3D elevation model (swisstopo, Wabern, Switzerland; 2 m); difference digital surface obtained fixed-wing drone S7). Space aggregated averaging cell monitored individuals numbers ranged one, multiple linear regression beta likelihood spatially structured error term (Homburger 2015). fitted integrated nested Laplace approximation r-inla (Rue 2009). Model code output S8. also zeros bounded so zero-inflated glmmTMB (Brooks 2017). Logit links component conditional intensity) S9). unevenly 4), clustered foraging, indicated (CIforaging 0.45 CIresting 0.11, p < 0.001). spent time (41.2%) (34.5%) (38.9%; shown). differed although differences significant variability. least CIcattle 0.39; Figure 5a) avoided clearly RPAcattle 0.55; 5b). High increased (CIcattle_low 0.33; CIcattle_high 0.44) did significantly affect (RPAcattle_low 0.49; RPAcattle_high 0.60). slightly, CIsheep 0.52). At frequently (RPAsheep_low 0.67; RPAsheep_high 0.86) highly treatments. (CIsheep_low 0.45; sheep_high 0.59). when kept comparably (CIgoats 0.47). While similar RPAsheep 0.76; RPAgoats 0.80). highlighted parameters explaining use, especially steeper 6a), 6b) 6c), an area. flat short explored edge stands. Under deep inside strictly pressure. indifferent tended depth impact sheep's slightly affected rate. pressure, tall vegetation, contrasts observed behaviour. contrast deterred by, rather attracted negative, As edges however vegetation. activity. no registered 7a), understorey leaves (not quantified). 7% 7b), is, 244 7c). standard deviation (SD 7.4%). surprising: (185 paddock) shrubs, S. aucuparia. (S. aucuparia: 140 82.3%; SD 7.9%; viridis: 45 0.8%; 0.7%). equally distributed about 2% layer. any Upon closer examination, explained occurrence absence debarking; Table 1a) found; 1b). strip values zero), possible include model. Remarkably, (Table lower those (pgoats Taller greater reduced probability would (pveg.heigth 0.001; pdistance 0.001); words, locations contrast, (pgoats*veg.height distances (pgoats*distance 0.045). effect negative (pslope 0.02). 1b) 1a). Hence, strongly impacted animals' decisions once chosen Of evaluated both (pdistance 0.0098); Over centuries, important Europe 2005) efficiently expansion. benchmark consumption our study. terms strategy. feeders readily switch browse 1988). Their tolerance broad spectrum feed resources study: available (Space-use hypothesis) (Debarking hypothesis). exploited nutrients herbaceous sources found respectively. corroborates farmers' experiences former encroached ecosystems 2011 birch heathlands; 2006 Mediterranean maquis). complemented observations, demonstrating species, likely originate Capra shelter 2018). grow suggests supplement diet occasionally edges, largely subsist explore depths taller occurs. potential (re-)transform previously shown showed ideal antagonists, aucuparia, much abundant hypothesis must, therefore, rejected. Generally, select (Iussig 2015), selectively respect, behaved hypothesized primarily selecting Nevertheless, appropriate re-establishing viridis, though inappropriate if management aim re-establish forest abandoned case protection against avalanches hazards. prevents reforestation, out-competes saplings fast growth, shading late leaf shedding, cause humidity long snow fungal infestation (Huber Frehner, 2013). Few compete conditions: valuable pioneer succession (Emmer 1998), preferably roughage branches, technique—ripping tongue—does facilitate stripping bark. And indeed, never seem constrain shrubs—likely adaptation terrain—than statistically significant. completely, do destroy quickly seedlings, understorey, breaking trampling down young moving Thereby probably slowly. beef easier place market meat, could profitable alternative term. hypothesized, evenly, penetrated deeper Similar comparable suitable goats. confirmed Space-use counter-intuitive. presumed best destructive demonstrated first spread Furthermore, useful resettlement weakening competitive increasing survival light-demanding saplings. variability additional, yet unidentified, influence slope; soil humidity; proximity water; cent shrubs; respect parameters. Large variation reported Celaya (2007). Further investigations necessary identify additional drivers A generally selectivity, equalize increase ruminants Pauler, Berard, Pittarello 2017; Utsumi 2010). Indeed, (Grazing clearer stable niche fodder, opportunistic. concurs Sanon (2007) who available, browsing time, ruminating. branch. underlines roughage. implications: substantially raise welfare issues, underfed. Contrastingly, (at breed) broader will study, adapted rough, can slowly before (Meisser 2014; Svensk main trampling, heavier Highland destructive. Heavy, production-oriented however, recently 2020) altogether. assume production-oriented, conventional (Brand, 2000; Glasser remains clarified detected general unique breed. unexpected, beneficial local importance preserving rare breeds. assessed estimate replace antagonists. conclude adequate choice depends objective: just grazing, included here appropriate: consumed understorey. Even economically attractive (and sheep), such Alternatively, aims reducing restore grasslands, outperform slow thickets. Cattle, goal undemanding, Finally, towards without trees succession, did. It seems worth investigating long-term offer benefits. Our findings underline powerful option impacts sustain structure, outstanding biodiversity grasslands. authors thank municipality La Punt Chamues-ch, Ralph Fluor Jörg Pleisch site; breeders SwissDexters association; Erny, Hossli Schriber families loaning animals; Raphael Gago, Nikolai Eisenhut, Ramon Hofer, Pascal Jenni, Marina Roth, Cornel Stutz Eric Vogelsanger multifaceted assistance; anonymous reviewers helpful comments; Mercator Research Program ETH Zurich World Food System Center financial support. Open access funding provided Agroscope. conflict declare. J.B., M.K., A.L. M.K.S. conceived ideas designed methodology; T.Z., M.S. collected data; C.M.P. led writing manuscript. contributed critically drafts gave final approval publication. Data Zenodo repository https://doi.org/10.5281/zenodo.4751171 (Pauler 2022). S1-S10 Please note: publisher responsible content functionality supporting information supplied authors. Any queries (other missing content) should directed corresponding author article.