Improved ecological insights commission new conservation targets for a crepuscular bird species

Knowledge on species’ ecological requirements is the key to an effective evidence‐based conservation. An adaptive approach can refine management recommendations following the outcomes of previous actions or as improved scientific knowledge about a species’ ecology becomes available. Following severe population declines the majority of the Swiss European Nightjar (Caprimulgus europaeus) population is now restricted to a small core area in the canton of Valais, despite the local application of evidence‐based restoration measures to the breeding habitats. The habitat use of this crepuscular bird species is presumed to be restricted to semi‐open oak‐pine scrublands and rocky steppe. By deploying miniature GPS loggers we assessed the spatial and habitat selection of 42 individuals (25 individuals in 2018 and 30 individuals in 2019; comprising 13 redeployments between years) in five study sites in the canton Valais. We highlight the importance of using fine‐scaled movement data to obtain insight into complex multi‐scale habitat requirements of a species. Nightjars used multiple habitats, indicating the importance of complementary resources to breed (e.g. open forest and shrub) and to forage (e.g. semi‐extensive grasslands and vineyards). The connectivity between these resources was influenced by habitat configuration and composition. Given these new insights for future conservation strategies, our results also suggest that national‐level land use changes, mainly due to agricultural intensification processes, have contributed to the long‐term population declines in Switzerland. We therefore consider our result in the context of knowledge gaps for species that exploit complementary habitats and the potential shortcomings for conservation planning on discrete species showing complex ecological requirements.


Introduction
Knowledge about species' ecological requirements holds a pivotal role in evidence-based species conservation during the formulation of conservation strategies, the monitoring of applied restoration actions and the empirical evaluation and reporting of the outcomes (Schwartz et al., 2017;de Bie, Addison & Cook, 2018). The availability of accurate knowledge on species' requirements differs greatly between taxonomical groups and is often lacking for species with a hidden lifestyle, such as crepuscular or nocturnal species (Gaston et al., 2008). The absence of accurate knowledge is particularly problematic when urgent conservation actions are required to maximize the benefits for rapidly declining species.
Studying habitat use (Johnson, 1980;Morris, 2003) is elementary to comprehend the ecological requirements of a species and spatio-temporal resource availability is a wellrecognized determinant of species' distribution (Weber et al., 2017). Habitat use should, herein, be considered as a multiscale process (Fattebert et al., 2018;Scherrer, Christe & Guisan, 2019). At a local scale, habitats and their associated resources might, for example, provide opportunities for a species to either breed, forage or rest (Wakefield, Phillips & Matthiopoulos, 2014). However, at a wider landscape shale their importance for a species, that is, the connection and accessibility of complementary resources, can be determined by the composition and configuration of the local landscape (Ripperger et al., 2015;Collins & Fahrig, 2017) in combination with the variation in individual space use of a species (Jones et al., 2018).
Detailed animal movement data, collected from GPS logging, can be collected in large quantities and provide unbiased knowledge about species' habitat use at the finest possible spatial scale, which can subsequently be extrapolated to investigate habitat use at larger spatial scales (Kay et al., 2017;Fattebert et al., 2018). Such data, collected during the breeding season, are particularly valuable to understand the importance of complementary resources for breeding and foraging (Schradin et al., 2010). It helps to reveal how the distribution of resources in the available landscape affects foraging behaviour, territory settlement and breeding phenology (Ropert-Coudert et al., 2004;Hinsley et al., 2008;Catry et al., 2013) and whether conservation practices deliver the resource requirements of the species (Law & Dickman, 1998;Sutherland et al., 2004;Lemieux et al., 2018).
There is an urgent need to review the ecological requirements of endangered species, in order to assess the accurateness of current management practices and to introduce more effective and relevant conservation measures into future conservation planning. In this perspective, the European Nightjar (Caprimulgus europaeus, hereafter referred to as nightjar) can be seen as a model system for species with a hidden lifestyle that require urgent conservation actions on different spatio-temporal scales (Evens et al., 2017a(Evens et al., , 2018a. During the breeding season, it has been demonstrated that these crepuscular, insectivorous birds require complementary habitats to breed and to forage (Alexander & Cresswell, 1990;Evens et al., 2017aEvens et al., , 2018a. The ongoing decline of this species in several European breeding populations (IUCN, 2020;Jacob et al., 2010;Knaus et al., 2018) may indicate that local conservation measures are most likely not applied to all required habitats and, given the high mobility of the species, not at the proper spatial scale.
In this study, we investigate the multi-scale habitat use of nightjars using fine-scaled space use data in an inner-alpine context in Switzerland. Nightjars are an endangered bird species in Switzerland (Keller et al., 2010). The studied population has undergone a rapid decline in the last decades (Knaus et al., 2018), despite the local application of evidence-based restoration measures to create new breeding habitats (Sierro, 2013(Sierro, , 2016. By following a multi-scale approach, we characterized foraging behaviour and assessed local habitat use (i.e. within home range habitat use) to define foraging habitats. Subsequently, we considered habitat selection in a wider landscape context (i.e. home range placement) to investigate whether the conservation of secondary key habitats, such as meadows and extensively cultivated grasslands, has been omitted. Finally, we extrapolated nightjars' habitat selection from the wider landscape to the national level (i.e. landscape scale) and investigated whether the disappearance of the species from many atlas grid cells (Knaus et al., 2018) was related to land cover changes. We consider our result in the context of knowledge gaps for species that exploit complementary habitats and the potential shortcomings for conservation planning on discrete species showing complex ecological requirements.

Materials and methods
We tracked the spatial use of nightjars from May to August in 2018 and 2019 at five sites in the canton of Valais (Switzerland), between Sierre (46°N, 7°E) and Visp (46°N, 8°E) in the upper Rhône valley. The sites are 4-16 km apart ( Fig. 1) and hold approximately 70% of the Swiss nightjar population (Knaus et al., 2018). The study sites are located on south-exposed slopes between 600 and 1800 m above sea level and are characterized by a continental climate with hot summers and a low amount of annual precipitation (550-700 mm/year) (Winiger et al., 2018). These breeding habitats (Fig. 2) are characterized by semi-open landscapes, mainly dominated by oak-pine scrublands, rocky steppe and open dry forests (Sierro et al., 2001). Within the same altitudinal band, the landscape is interspersed with settlements and grasslands of different degrees of intensification, such as meadows, steppe and extensively grazed pastures (Theux, 2019). On higher elevations, dense pine forests and alpine meadows are the predominant land cover types. On lower elevations, intensively managed vineyards are the predominant agricultural land cover type. Most human activities are concentrated on the lowest elevations, in the plain of the valley (Sierro & Erhardt, 2019).
We captured nightjars using ultra-fine mist nets (Ecotone, 12 × 3 m) and tape lures (Evens et al., 2017a). All birds were captured within their presumed breeding territories and marked with a unique alphanumeric ring from the Swiss Ornithological Institute. We identified sex [M/F] and age [≥1CY, 2CY, ≥2CYs] based on individual's plumage, but did not discriminate between paired or unpaired individuals because the reproductive state of tracked individuals was often unclear. We fitted a VHF/GPS logger, comprising a 0.7 g Biotrack Ltd. radio tag and a 1.8 g Pathtrack Ltd. nanoFix GPS logger, to the tail using a simple drop-off mechanism (Evens et al., 2018). Tags weighed less than 3.8% of the mean weight of tagged birds (65.3 AE 5.1 g, range: [57-78.1 g], n = 46 weight measurements; for a list of the birds see Supporting Information). We programmed GPS loggers to start fixing positions (3 min intervals) from sunset until sunrise (i.e. 9 PM until 5 AM).

Land cover data
We used two types of land cover data: (i) high resolution (10 × 10 m) remote sensing data and (ii) a lower resolution digital land cover map (100 × 100 m). Both maps were used to study (i) within home range habitat use and (ii) home range placement and national habitat use respectively.
The digital habitat layers containing information on land cover on a 100 × 100 m grid, available for 3 time periods (1979-1985, 1992-1997, 2004-2009Swiss Federal Statistical Office, 2019). We reclassified 72 available habitat types into 12 relevant habitat types (Sierro et al., 2001;Evens et al., 2017a, Evens et al., 2018aSupporting Information). To investigate changes in land cover on a national scale, we  used habitat layers for the periods 1979-1985, 1992-1997 and 2004-2009. To investigate home range placement, we used the most recent land cover map (period 2004-2009; hereafter referred to as structural habitat map). From the structural habitat map, we then created a functional landscape map by grouping 12 habitat types into three functional categories: (i) breeding or roosting habitat, (ii) foraging habitat and (iii) other habitat types (Evens et al., 2018).

Available habitat maps
To delineate the area of available habitat and to derive the measures of landscape heterogeneity for each foraging flight, we cut out four circular maps (two from the structural and two from the functional habitat map) (sensu Evens et al., 2018a). We made the circular maps and determined their specific scale to avoid inclusion of large areas of unrepresentative habitat related to foraging activity (Evens et al., 2018a).
We defined extra-territorial flights and foraging flights sensu Evens et al. (2018a). Extra-territorial flights include the start in breeding habitat, an outbound flight, an inbound flight to the breeding habitat and arrival in the breeding habitat. An extraterritorial flight can be further defined as a foraging flight when an individual also perches between the outbound and inbound section. In this case, foraging activity is defined as an individual remaining stationary (i.e. flycatching when perching) in presumed foraging habitat. This approach did not account for foraging by hawking (i.e. foraging when flying) (Alexander & Cresswell, 1990) or foraging behaviour within breeding areas, which is not detectable using current GPS-tracking technologies. We used information on foraging activity to calculate Euclidean foraging distance, foraging duration and the timing of foraging events (Evens et al., 2018a).
The centre of each circular available habitat map was always placed at the start position of the corresponding foraging flight. For each habitat map, thus both the structural and functional habitat map, a circular surface was cut with a radius equal to the Euclidian foraging distance (hereafter referred to as 'foraging buffer') and another with a radius equal to the mean foraging distance (1311 m; calculated as the mean foraging distance of all foraging flights for all individuals; hereafter referred to as 'mean foraging buffer'). For each foraging flight, we then quantified four landscape characteristics: two measures of habitat composition (

Within home range habitat use
To investigate the effect of individual variables (age and sex) and landscape characteristics on foraging distance, we fitted linear mixed models to our data with individual nested within year (2018 or 2019) and study site as a random factor. After a backward selection procedure, individual variables (age and sex) were removed from the models as fixed effects because they did not significantly affect foraging distance. The selection procedure was initiated from two separate models to account for multicollinearity issues. More specific, we found a correlation between the following environmental variables: the size of foraging sites and Moran's I, habitat diversity and Moran's I, the amount of available foraging habitat and the size of foraging sites, and amount of available foraging habitat and habitat diversity. Finally, three models were fitted containing foraging distance (log-transformed) as outcome variable. The first model contained mean size of foraging habitats [calculated using the foraging buffer; km 2 ] and structural habitat diversity [calculated using the foraging buffer; Shannon Index] as fixed effects. The second model contained mean size of foraging habitats [calculated using the foraging buffer; km 2 ] and functional habitat diversity [calculated using the foraging buffer; Shannon Index] as fixed effects. The third model contained amount of foraging habitat [calculated using the mean foraging buffer; % cover] and Moran's I [calculated using the foraging buffer; classified as random, dispersed or clustered]) as fixed effects.
We also collected information on within home range habitat use by extracting data from the structural habitat map for each GPS position. We investigated the use of specific grassland types by extracting data from the fine-scaled grassland map for each GPS position falling in a grassland pixel. We fitted general linear mixed models to assess possible differences in i) grassland use between breeding and foraging sites and ii) grassland availability between occupied and historically managed sites (Table 2, Fig. 3).

Home range placement
Using GPS positions, we calculated the home range of each individual in each year with a kernel density estimator (KDE) (fixed kernel), using Ranges 7 v0.77 (Anatrack Ltd.) (Aebischer, Robertson & Kenward, 1993). We calculated 50% (core area) and 95% (home range) kernels (Evens et al., 2017a). We used a fixed multiplier, between 0.3 and 2, that limited the number of multimodal home ranges or the inclusion of large unused areas (Fieberg, 2007;Supporting Information). To compare our results with earlier studies (Sierro et al., 2001) and to get a more general assessment of the space covered by individuals, we also calculated the 95% minimum convex polygon for each individual (Aebischer et al., 1993).
To quantify habitat availability for each individual, and to avoid the inclusion of large areas of unrepresentative habitats (Evens et al., 2017a(Evens et al., , 2018a, we created a habitat availability map (hereafter referred to as 'focal buffers') by cutting out a circular surface from the structural map (center = focal position/mean observation in breeding habitat, radius = furthest observation per individual; sensu Evens et al., 2018). The furthest observation per individual is not necessarily the same as an individual's longest foraging distance. To assess the availability of specific grassland types in proximity of nightjars' territories and historically managed sites (Sierro, 2016), we also created a 'mean buffer'. The centroid of the mean buffer was placed on individuals' focal position and on the centroid of historical managed sites. The radius of these buffers was equal to the mean of the distance to the furthest observation of all individuals (1905 m).
To determine regional habitat selection, we compared home range placement (50% and 95% kernels; i.e. habitat use) with the tailored habitat availability maps (focal buffer; i.e. habitat availability). Hence, we carried out a classical compositional analysis (Aebischer et al., 1993) using the Rpackage 'adehabitatHS' (Calenge, 2011). Paired t-tests determined the differences between habitats (habitats ranked, independent of availability, based on positive differences). Zero values were replaced with 10E-7 and 1000 iterations were chosen for data randomization (Evens et al., 2017a).

National dynamics
In order to assess whether land cover changes in recent decades could have contributed to the decline of the nightjar population in Switzerland, we combined information from the three reclassified land cover maps (1979-1985, 1992-1997-2009Swiss Federal Statistical Office [2019]) with three occupancy maps of the Swiss bird atlas. The occupancy maps contain information on the presence of breeding nightjars in Switzerland for 467 10 × 10 km squares and three atlas periods : 1972-1976(Schifferli et al., 1980), 1993-1996(Schmid et al., 1998  Spatio-temporal logistic Bernoulli models were fitted to estimate the effect of available habitat per 10 × 10 km square on nightjar presence. We used period-specific conditional autoregressive (CAR, Besag, York & Mollié, 1991) random effects, with a shared precision parameter, to capture compositional heterogeneity. The covariate effects of time and habitat coverage, which comprises 12 proportional habitat availability variables, were modelled univariately to avoid multicollinearity issues. The habitat availability variables were transformed through a folded exponential transformation (Piepho, 2003). An additional normally distributed unstructured normal random effect that was shared among the three periods accommodated additional overdispersion. This yielded 14 models, which were fitted using Markov chain Monte Carlo (MCMC) via the R (version 3.4.4) package R2OpenBUGS 3.2 (for full model specifications, see Supporting Information).

Animals subjects
The authors declare that all experiments have been performed according to the ethical guidelines of the Swiss veterinarian service under license number VS032018.

Results
We deployed 85 GPS loggers on 46 individuals and recovered 80 GPS loggers from 42 individuals. In total, 73 GPS loggers contained useful tracking data from 25 individuals in 2018 and from 30 individuals in 2019. Fifteen individuals were tracked more than once within a season and 13 individuals were tracked in both years. In order to account for possible differences in the habitat use between years, caused by habitat management, environmental or unknown factors, we will refer to a maximal sample size of 55 year-individuals in this manuscript, except in case of the linear mixed models where individual was nested within year (2018 or 2019). Altogether the GPS loggers contained 42816 observations on the spatial use and foraging behaviour of nightjars (808 AE 470 observations per individual per year, n = 55, [20-2049 observations]; mean AE standard deviation and [range]). We excluded eight individuals from home range analysis because of presumed nomadic behaviour and longdistance movements of unpaired males in search for a territory and two other individuals were removed from foragingdistance analysis because no foraging behaviour was detected (i.e. unpaired males remaining in a small territory) (Evens et al., 2018b).

Within home range habitat use
We collected 34310 observations while birds were present in the known breeding habitats (for an overview see Table 1) We identified 452 flights leaving the breeding habitats (for an overview see Table 1), containing 312 foraging flights. The mean foraging distance is 1.3 AE 0.7 km ([0.2-4.3 km], n = 312 flights of 45 individuals). We did not find any differences in foraging distance between sexes and ages ( Table 2) (Table 1).
Concerning the use of grasslands, we found a significant difference between the types of grasslands used in breeding and foraging areas (estimate = −0.222, SD = 0.025, z-value = 55, Pr(>|z|) = <0.0001; Supporting Information). The difference between the two seems to arise due to a higher proportion of extensive and semi-extensive grasslands in the breeding areas and presence of semi-intensive grasslands in foraging areas (Table 2, Fig. 3). When comparing the The estimates are based on the amount of GPS observations made in breeding areas and while birds are presumed to be foraging (perched) for each habitat type (Supporting Information). availability of grassland types between occupied and historically managed sites, we found no clear difference between both groups (estimate = −0.001, SD = 0.025, t-value = −0.053, Pr(>|z|) = 0.958; Table 2), both breeding grounds sharing similar extensive grasslands habitat types.

Discussion
This study highlights the importance of using fine-scaled movement data to obtain insight into complex multi-scale habitat requirements of a crepuscular species. Nightjars used multiple habitat types indicating the importance of complementary resources to breed and to forage. Requirements to breeding habitats of the studied population are in line with previous findings (Sierro et al., 2001;Winiger et al., 2018), while (i) the preference for specific grassland types, alpine meadows and vineyards to forage and (ii) the requirement of a larger space, on average six times higher per individual compared with previous estimates (Sierro et al., 2001), shed new light on future conservation strategies. Furthermore, our results suggest that the land use changes, especially in potential foraging habitat, have contributed to the long-term population declines in Switzerland.

Multiple habitat use
Breeding habitat, in our study, mainly comprises open, lownutrient habitats such as oak-pine scrubland and pine forests. Here nightjars laid eggs on bare ground close to open vegetation, such as shrub or low-growing oak trees (personal observations). These findings are in line with earlier studies in Switzerland (Sierro et al., 2001;Winiger et al., 2018) and elsewhere in Europe where nightjars have been observed to breed in various types of semi-open, low-nutrient habitats (Wichmann, 2004;Conway et al., 2007;Evens et al., 2017a). It has been suggested that nightjars would also stay within their breeding habitats to forage owing to sufficient food supplies (Sierro et al., 2001;Sharps et al., 2015a) or suitable micro-habitat structures to forage (Sierro et al., 2001;Wichmann, 2004). The lacking evidence of multiple habitat requirements, for the Swiss population, resulted in conservation measures mostly focusing on the management and restoration of lost breeding habitats (Sierro, 2013(Sierro, , 2016. However, almost all tracked nightjars (visual inspection of tracking data) in our study have been recorded foraging in extensively cultivated agricultural grasslands, alpine meadows or in vineyards.
The use of extensively cultivated grasslands for foraging seems to be common among nightjars (Alexander & Cresswell, 1990;Evens et al., 2018a). In our study, both extensive and semi-extensive grasslands, and not the natural xerophilic vegetation, are the predominant grassland types used by nightjars in foraging sites (Fig. 3A). The former grasslands are managed with a low to intermediate level of fertilizers; and it may be that these grasslands host a higher biomass of invertebrate fauna than natural grasslands, despite the presence of rare fauna and flora in the latter (Andrey et al. 2016).
Due to the high availability of vineyards, the compositional analysis probably did not identify nightjars' preference for this habitat type. From GPS-based foraging data, however, we observe that some individuals almost exclusively forage in vineyards. From field observations we expect that nightjars most likely select specific vineyards with a biodiversity-friendly management and a diverse ground vegetation that is known to promote invertebrate abundance (Bosco, Arlettaz, Jacot 2019a, Bosco et al., 2019b. That nightjars selected intensively managed vineyards would not be expected but can be explained by ongoing changes in management techniques. While still roughly 80% of vineyards in Valais are intensively managed and use herbicides to kill the ground vegetation, a growing minority is applying a more biodiversity-friendly management. Such parcels show a permanent ground vegetation that has a positive effect on insectivorous birds and their prey (Bosco et al., 2019a). In addition, semi-natural structures such as bushes and hedges on or around vineyard plots are known to be important structures for perching and foraging (Guyot et al., 2017).
It is important to stress that it remains challenging to assess the actual importance small landscape elements or remnants of specific habitat types when such fragments are misclassified at larger spatial scales and important environmental information is lost (English et al., 2017). For this reason, we assume that the actual importance of extensively cultivated agricultural grasslands and biologically managed vineyards is not identified by the compositional analysis in our study, due to their fragmentation within the surrounding intensively cultivated landscape.

Landscape complementation
In order to reach foraging habitats, nightjars flew considerable distances (mean foraging distance: 1.3 AE 0.7 km), even into the urbanized valley. The recorded foraging distances are in line with other telemetry (2.6 AE 1.1 km in Evens et al. [2017a], 3.1 AE 1.2 km in Alexander and Cresswell, 1990) and GPS-tracking studies (range 0.6-3.3 km in Evens et al., 2018aEvens et al., ,2018b performed elsewhere in Europe, but foraging distances in our study are higher compared to an earlier telemetry-based study (0.7 AE 0.5 km in Sharps et al. [2015]). Nightjars are able to connect complementary habitats, even when these are separated by several kilometres (Alexander & Cresswell, 1990;Camacho et al., 2014;Evens et al., 2018a). Herein, landscape heterogeneity influences the connectivity between complementary resources. In case of habitat composition, shorter foraging distances are observed when structural habitat diversity is lower and functional habitat diversity is higher. In our particular case, these findings seem to suggest that high structural habitat diversity was mainly explained by high diversity of unsuitable and breeding/roosting habitat types, whereas low functional habitat diversity most likely is determined by the overabundance of one of these functional habitats. This could imply that nightjars are more reluctant to traverse a high amounts of (unsuitable) habitat in order to reach foraging sites. In line with these findings, we observe that foraging distance is also shorter in randomly distributed landscapes where breeding and foraging habitats probably can be found on a much smaller spatial scale.
Although our study has been performed in a different environment compared to Western-European studies on the same species (Alexander & Cresswell, 1990;Evens et al., 2018a), our results confirm that nightjars require complementary habitats to breed and to forage (Alexander & Cresswell, 1990;Evens et al., 2018a), something which has also been demonstrated for other species of Nightjars (e.g. Caprimulgus ruficollis in Camacho et al., 2014). These findings further support the belief that complementary habitat requirements are common among nightjars. Because habitat composition and configuration influence nightjars' foraging distance in our study, it is clear that landscape heterogeneity affects landscape complementation (sensu Dunning, Danielson & Pulliam, 1992) for nightjars. The similar observations in the afore-mentioned studies suggest that ongoing landscape homogenization and fragmentation of foraging habitats, and in particular those further from breeding habitats, may increase travel distance between non-suitable resources and thus affect nightjar populations. Landscape complementation has a strong influence on the viability of a wide range of animal communities (e.g. Chadés et al., 2015;Ripperger et al., 2015;Collins & Fahrig, 2017), contributing to species occurrence over time (Ikin et al., 2018), yet, factors beyond landscape heterogeneity might also be driving rapid declines of the aerial insectivore guild (English et al., 2017;Grubisic et al., 2018;Wagner, 2020). At a wider landscape scale (e.g. home range placement), animals might select areas where complementary resources are close to minimize movement costs (i.e. high landscape complementation). Yet at a local scale (e.g. within home range habitat use), resource depletion may lead to habitat patches that are ultimately avoided by these animals (Valls-Fox et al., 2018). In our study area, we have no proof of resource depletion in foraging sites since many individuals have been observed to repeatedly forage in the same spot together during the entire study period (personal observations). Urbanization drives global environmental change causing, among many thing, severe declines of aerial insects (Kronfeld-Schor et al., 2013;Owens & Lewis, 2018;Piano et al., 2020). Unfortunately it is still unclear how nightjars respond to changes in food availability in breeding (Sierro et al., 2001;Sharps et al., 2015b) or foraging habitats (Evens et al., 2018a) and whether alterations in micro-habitat structures (Sierro et al., 2001;Wichmann, 2004;Camacho, 2014) or perhaps even anthropogenically mediated food availability (personal observations; Jackson, 2003;Sierro & Erhardt, 2019) might play a role in the habitat and spatial use of nightjars.

Conservation
Our national approach suggests that, during the last four decades, the Swiss nightjar population has suffered from declines due to land cover changes in foraging habitats. Recently, other factors, such as light pollution, have been proposed to negatively influence the relationship between the nightjar breeding populations in Valais and the growing human population (Sierro & Erhardt, 2019). It is, however, still unclear which factors during the breeding (e.g. light pollution, food availability, land cover change, disturbance or predation) and the non-breeding period (e.g. land cover change or climate change) actually contributed to the decline of nightjars in Switzerland.
Missing the appropriate knowledge on nightjars' requirement for complementary habitats to breed and to forage, might have substantially contributed to the low efficiency of conservation measures taken during the last decades in our study population. This idea is further supported by our observations that suitable foraging habitats are still available close to the remaining breeding sites in Valais, but have disappeared from breeding sites where most historical conservation measures have taken place. Most of the managed breeding areas are small isolated clearings on southern slopes in areas where semi-extensive meadows have disappeared and have been replaced by vineyard and forest; or in the lowland on less dry and exposed areas close to intensive agricultural fields. As a result, the surroundings of the restored sites actually comprise monotone agricultural fields or intensive grasslands used for silage without small landscape elements. We expect that their suitability as foraging habitats is comparable with intensively managed grasslands (e.g. artificially sown meadows). Furthermore, we expect that, without small landscape elements to perch, the access to these grasslands is low since foraging site selection also reflects the ease of prey capture (Sierro et al., 2001) and predator avoidance (Camacho, 2014;Evens et al., 2017b).
Albeit the concept of landscape complementation is all but novel in conservation biology (Dunning et al., 1992), it cannot be considered as long as knowledge about species' ecological requirements is lacking. For long, the conservation action plan for the nightjar in Switzerland has focused on the optimization of forest structure used as breeding grounds (Sierro, 2016) while important habitats for this endangered species were disappearing. As a result, the process resulting in the abandonment of an important part of the Swiss nightjar territories is still slowly ongoing. Near occupied breeding areas, extensive and semi-natural grasslands are more and more scattered and isolated. These habitats are heavily affected by on-going land use changes and thus considered among the most threatened habitats in mountainous landscapes (Canals & Sebastià, 2000).
Areas with high complementarity between different resources support high population densities (e.g. Brotons et al., 2005;Haynes, Diekötter & Crist, 2007), also in case of nightjars . To promote landscape complementarity, conservationists should improve species' accessibility to complementary habitats (Pope, Fahrig & Gray, 2000) by considering both habitat configuration and composition (Haase et al., 2017). The similar findings between recent nightjar studies across Western Europe suggest that refining current conservation programs for nightjars in Europe could be done by (i) creating or restoring breeding grounds in proximity of suitable foraging habitats and (ii) focusing on creating/restoring foraging habitats in proximity to breeding habitats. Although higher amounts of habitat can support a higher diversity of macro-moth species (Merckx et al. 2019), the significance of small habitat patches for conservation purposes should not be neglected (Fahrig et al., 2019;Fahrig, 2020) to increase landscape complementation (Fahrig, 2017). This has also been demonstrated in our study by many nightjars foraging at the same time in remnant foraging patches near more intensively managed grasslands or vineyards (personal observations, Supporting Information). Particularly, the preservation and restoration of semi-extensive grassland habitats should be promoted in order to retain a high abundance, diversity and biomass of invertebrates. In the special case of vineyards, we also learnt from field observations that nightjars selected biologically managed vineyards. Since land-sharing policies provide higher ecological value than land-sparing policies (Fahrig, 2017), we suggest to support initiatives that want to create biologically managed vineyards. Moreover, in combination with unproductive areas between vineyards and small, extensively cultivated grasslands comprising small landscape elements such trees, bushes and hedges (Supporting Information), this may have a positive effect on habitat selection processes in nightjars as has been shown for a number of avian species (Guyot et al., 2017).