Introduction
Many natural and semi-natural ecosystems have been destroyed, degraded or fragmented by human activities, to a point that, without dedicated restoration, some critical ecosystem functions are severely compromised in many remaining patches (Banks-Leite et al. Reference Banks-Leite, Ewers, Folkard-Tapp and Fraser2020). Forests, which harbour many species sensitive to isolation and fragmentation, are the frequent object of restoration initiatives (Hutto et al. Reference Hutto, Flesch and Fylling2014; Mansourian et al. Reference Mansourian, Kleymann, Passardi, Winter, Derkyi and Diederichsen2022). A particular case is represented by the increasing attention devoted to urban and peri-urban natural or semi-natural environments (McDonnell and MacGregor-Fors Reference McDonnell and MacGregor-Fors2016): many interventions in different continents are aimed at restoring portions of forest habitats in or near cities and towns. This trend is due to multiple reasons, including the increased awareness of the importance of urban and suburban forest for crucial ecosystem services (Dobbs et al. Reference Dobbs, Escobedo and Zipperer2011; Endreny et al. Reference Endreny, Santagata, Perna, De, Rallo and Ulgiati2017; Prigioniero et al. Reference Prigioniero, Paura, Zuzolo, Tartaglia, Postiglione and Scarano2022), such as climate and flood regulation, and recreational, aesthetic, and spiritual values (Brown et al. Reference Brown, Schebella and Weber2014; Dobbs et al. Reference Dobbs, Escobedo and Zipperer2011; Endreny et al. Reference Endreny, Santagata, Perna, De, Rallo and Ulgiati2017; Jenerette et al. Reference Jenerette, Harlan, Brazel, Jones, Larsen and Stefanov2007). Also conservation strategies aimed at promoting biodiversity in forests of urban and peri-urban sites increasingly include restoration activities (Wallace and Clarkson Reference Wallace and Clarkson2019); such habitats are indeed key to preserving taxonomically diverse communities (Dale Reference Dale2018; Morelli et al. Reference Morelli, Benedetti, Ibáñez-Álamo, Tryjanowski, Jokimäki and Kaisanlahti-Jokimäki2021), even if they are often affected by higher levels of disturbance and lower habitat and micro-habitat availability (e.g. Nieoczym et al. Reference Nieoczym, Polak and Wiącek2022), and can be subject to a high degree of degradation, fragmentation, and isolation (Gounaridis et al. Reference Gounaridis, Newell and Goodspeed2020).
Birds are frequently selected as indicators for the effects of environmental changes, including that of ecosystem restoration (Ortega-Alvarez and Lindig-Cisneros Reference Ortega-Alvarez and Lindig-Cisneros2012) and changes of habitat cover in urban and peri-urban sites (Fraissinet et al. Reference Fraissinet, Ancillotto, Migliozzi, Capasso, Bosso and Chamberlain2022), thanks to their ecological sensitivity, indicator value, occurrence at various trophic levels, links with habitats, and quick reactions to environmental changes, as well as to the relatively easy survey methods, which allow the gathering of abundant and reliable data with relatively limited efforts (Sutherland Reference Sutherland2006). Several studies have covered bird communities of restored sites (e.g. Brambilla Reference Brambilla2020; Versluijs et al. Reference Versluijs, Eggers, Hjältén, Löfroth and Roberge2017), as well as the contribution of birds to habitat restoration (Ortega-Alvarez and Lindig-Cisneros Reference Ortega-Alvarez and Lindig-Cisneros2012). Particular emphasis has been placed on degraded sites (e.g. exhausted mining sites), which became objects of restoration (Sampaio et al. Reference Sampaio, Pereira, Nunes, Clemente, Salgueiro and Silva2021), whereas a number of studies have dealt with urban forest restoration (Noe et al. Reference Noe, Innes, Barnes, Joshi and Clarkson2022). Examples related to peri-urban forest restoration are rarer, even if ecological restoration of urban and suburban landscape is gaining momentum (McDonnell and MacGregor-Fors Reference McDonnell and MacGregor-Fors2016; Standards Reference Group SERA 2017), and could be further boosted in the EU by the Nature Restoration Law.
Most of the studies on the impact of forest restoration on birds are based on comparisons between restored and unrestored sites, or between restored/unrestored and reference (more natural) sites (Reeder and Wulker Reference Reeder and Wulker2017; Ruiz-Jaén and Aide Reference Ruiz-Jaén and Aide2005). In general, studies suggest that the longer the time after the restoration began, the higher the diversity and richness of the avian community (Noe et al. Reference Noe, Innes, Barnes, Joshi and Clarkson2022). Nevertheless, even after a very few years (e.g. 2–3 years) positive or negative effects can be detected (Hutto et al. Reference Hutto, Flesch and Fylling2014; Passell Reference Passell2000).
Full evaluations of any kind of interventions, including restoration effects (Conner et al. Reference Conner, Saunders, Bouwes and Jordan2016), on any kind of taxa or communities, including birds (Battisti and Marini Reference Battisti and Marini2018; Brambilla and Gatti Reference Brambilla and Gatti2022), ideally require a before/after-control/impact (BACI) experiment (Christie et al. Reference Christie, Amano, Martin, Shackelford, Simmons and Sutherland2019). BACI protocols have been implemented for forest restoration in natural and semi-natural environments (Hutto et al. Reference Hutto, Flesch and Fylling2014), whereas similar applications to urban or peri-urban contexts are virtually lacking (a search with “forest restoration” and “urban” and (“before/after” or “BACI”) in Scopus and Web of Science databases, performed on 23 February 2023, did not retrieve any relevant papers).
Many forest bird species are highly susceptible to forest reduction, isolation, and fragmentation (Clergeau and Burel Reference Clergeau and Burel1997). In peri-urban areas, urbanisation, sprawl, and other anthropogenic alterations are increasingly reducing, fragmenting, and isolating natural or semi-natural patches (Battisti et al. Reference Battisti, Barucci, Concettini, Dodaro and Marini2022), reducing forest functionality (Gounaridis et al. Reference Gounaridis, Newell and Goodspeed2020). Such dynamics need to be counteracted by strategies to improve habitat quality for forest birds and biodiversity in these contexts. Therefore, considering the relevance of forest restoration near urban areas on the one hand, and the value of birds as indicators for broader and more general patterns on the other hand, it is important to assess the effectiveness of peri-urban forest restoration on birds. Given the short response time shown by avian communities to many interventions, even evaluations performed over a short term could be relevant.
Birds may be affected by small-scale forest restoration after a very few years in different ways. Some species may be affected by structural habitat changes: those dwelling in young woodlands or in transitional habitats (such as areas recently encroached by shrubs and trees) may be favoured by interventions, while those tied to open habitats (such as grassland or arable land) may be negatively affected (Assandri et al. Reference Assandri, Bogliani, Pedrini and Brambilla2017). Forest species might be advantaged by a reduction of edge effects (Batáry and Báldi Reference Batáry and Báldi2004; Batáry et al. Reference Batáry, Fronczek, Normann, Scherber and Tscharntke2014), or by mitigation of the fragmentation of habitat patches through the creation of “stepping stones” (Saura et al. Reference Saura, Bodin and Fortin2014), rather than by the provision of high-quality habitats, which is unlikely over short time frames (Harmon and Pabst Reference Harmon and Pabst2015).
This study therefore explored the short-term effects on birds of small-scale restoration interventions within the largest Italian conurbation, targeted at increasing existing forest patches and/or reducing their isolation. We adopted a BACI protocol to relate observed changes in bird communities or abundance to the interventions implemented during the restoration initiative. To the best of our knowledge, this is the first time that such an assessment has been carried out for peri-urban forests. We anticipated that the restoration activities would provide immediate benefit to the local breeding bird community by increasing habitat availability for generalist and ecotonal species, and/or by reducing the edge effect around existing forest patches.
Methods
Study system
In an area located just north of Milan (Lombardy, Italy), stretching towards the first pre-alpine hills and mountains (between 9.0414° and 9.1816°W and 45.5245° and 45.6947°N), small-scale peri-urban restoration interventions of forest habitats were realised in 2019 (Figure 1). The study area encompasses part of the provinces of Como, Monza e Brianza, and Milan (and marginally Varese), stretched along a N–S gradient. All the area is characterised by a temperate climate and is heavily modified by human activities; however, there is a general gradient of land-use intensity and alteration, increasing from north to south (i.e. towards Milan). Within this human-dominated landscape, some protected and agricultural areas still offer potential habitats to birds and other wildlife. Forest habitats are among the most valuable in the area in terms of both biodiversity and recreational opportunities. They are largely dominated by native broadleaved species, but non-native species are frequent or even dominant in the most disturbed sites. Several forest patches are rather small and/or fragmented, and many are also isolated from other patches (Figure 1). The high fragmentation and isolation level of many forest patches was the main reason for the restoration project.
The intervention plan, which did not target individual species or groups, was generally aimed at: (1) improving the status of existing forest patches by enlarging them and increasing the cover of native species by planting trees and shrubs at their margins, or secondarily by replacing exotic species with autochthonous ones; (2) increasing connectivity between forest patches and mitigating their isolation by establishing new patches between existing isolated patches, especially in the northern part of the area. The restoration activities involved the planting of native shrub and tree species, with the tallest planted individuals generally approaching or exceeding 4 m. Restorations were largely carried out between existing forest patches and cultivated fields, between forest and sport/recreational sites, or over uncultivated areas invaded by ruderal or invasive species. Sometimes native species were planted as tree-rows or hedgerows. In one site, tree and shrub plantation was associated with the creation of small ponds with riparian vegetation (targeted at amphibians), and in another site it was associated with the amelioration of a small wetland. All monitored interventions occurred over an extent of 0.1–1.2 ha; the amount of restored habitat encompassed by the 100-radius of point counts (see below) ranged between 0.1 and c.0.7 ha per site (see Figure 2 for some examples).
This study encompassed virtually all the interventions established within the project to improve the status, area, and/or connectivity of forest patches, adopting a BACI design. Given that restoration sites were either adjacent to existing forest patches (in the case of interventions increasing existing patches) or circumscribed by agricultural and/or urban land (in the case of restoration in areas between different forest patches), we scattered control plots so as to mirror the distribution of restoration sites, in all cases within 2 km of the latter. We surveyed similar numbers of control and treatment sites, which were grouped within five different areas, identified as critical for the restoration of ecological connectivity between forest tracts (largely included in protected areas) (Figure 1), and hence selected for intervention. Some control plots were therefore located in forest habitats, while others were in agricultural contexts, and finally some were located in suburban areas (See Supplementary material Table S1 for the dominant habitat found at each site).
Bird monitoring in relation to restoration efforts
Interventions took place in 2019 and were completed in the same (or, exceptionally, in the subsequent) year. Bird monitoring was carried out at 40 points (20 at intervention sites, with one point per restoration patch, and 20 control points, without any intervention within a 100-m radius around the point), in 2018 (before interventions), 2019 (during interventions), and 2021 (after interventions), to check for the effect of on-going interventions and then of the restored habitat, two years after the restoration action (Figure 2). Points were at least 200 m apart to avoid double counting of the same individuals. Point counts at intervention sites were located within or at the boundary of the restored patch, and the restored patches were entirely (or almost so, in one case) encompassed by the 100-m radius around the point count. Bird surveys were carried out by 10-minute point counts (Bibby et al. Reference Bibby, Burgess, Hill and Mustoe2000), because of the rather small but widespread areas to be surveyed, and the need to encompass all the area and all the species, with a particular focus on the relatively common ones (Sutherland Reference Sutherland2006).
Each survey site was surveyed twice in a year (with minimum variation due to logistical constraints; see Table S1): once in April, and a second time in late May–early June to cover the breeding season of the locally breeding species. Each observer surveyed the same exact points every year. Surveys were carried out in the early morning (from dawn to 9h00–11h00 depending on weather – stopping earlier under warm temperatures), avoiding windy or rainy days. All contacts with all species were recorded, discriminating between those which occurred within and those which occurred outside a 100-m distance from the sampling point (estimated by means of detailed maps – scale 1:1,000 – displaying the point and the 100-m radius over a high-resolution aerial orthophotograph of the site), as is commonly done in most avian studies based on point counts (Bibby et al. Reference Bibby, Burgess, Hill and Mustoe2000; Ceresa et al. Reference Ceresa, Kranebitter, Monrós, Rizzolli and Brambilla2021; Chamberlain et al. Reference Chamberlain, Brambilla, Caprio, Pedrini and Rolando2016; Sutherland Reference Sutherland2006), including restoration assessments (Brambilla Reference Brambilla2020). For the analyses, only the contacts that occurred within 100 m were considered. All the data are publicly available (Brambilla et al. Reference Brambilla, Foglini and Vitulano2023).
Statistical analyses
We first carried out some exploration to see whether possible variations in bird species richness and abundance of forest species were potentially linked to restoration interventions, or to year effects. Then, we assessed the effect of restoration and year on avian communities; given that no tangible environmental changes apart from restoration occurred at the survey sites, we did not include any other predictor. The dependent variables used were: (1) the number of species at each site; (2) the number of species at each site after the exclusion of species occurring only as migrants (i.e. non-breeding, based on our knowledge of the local avian communities and on Lardelli et al. Reference Lardelli, Bogliani, Brichetti, Caprio, Celada and Conca2022) within the study area, raptors, aerial foragers (i.e. bee-eaters, swifts, swallows, and martins) and introduced (non-self-sustaining) species; (3) the number of forest species, considering only the non-raptorial species regularly and predominantly breeding in forest habitats in the study area: Marsh Tit Poecile palustris, Coal Tit Parus ater, Crested Tit Lophophanes cristatus, Eurasian Jay Garrulus glandarius, Eurasian Nuthatch Sitta europaea, Black Woodpecker Dryocopus martius, and Short-toed Treecreeper Certhia brachidactyla; (4) the abundance of the above listed forest species; (5) abundance (maximum number of individuals per year per point) of each species found in at least 15 surveys over the three years (rarer species were not considered for the latter analysis, as the sample size was too low for species-specific assessments). Individuals observed only flying over the site, without any kind of interactions with it, were discarded from all the analyses apart from the first one (Assandri et al. Reference Assandri, Bogliani, Pedrini and Brambilla2019).
Generalised linear mixed models (GLMMs) were employed to check for the effect of restoration. A Poisson model was used given that the dependent variables were counts (species richness or abundance of single species). Site identity was used as a random (grouping) factor, while year and intervention phase were entered as categorical predictors. A variable “phase” was used to describe the restoration status of each survey/point: it was set to (1) “no intervention” for all sites in the first year, and then in subsequent years for control sites (not affected by restoration); (2) “on-going interventions” for restoration sites in 2019; (3) “after interventions” for restoration sites in 2021, when all restoration activities had been concluded. Therefore, phase classified sites according to their relationship with restoration: no intervention at all for control sites and for all sites before restoration occurred (baseline), on-going restoration activities, and concluded restoration activities. Models were fitted through a Bayesian approach, using the package “brms” (Bürkner Reference Bürkner2017, Reference Bürkner2018) in R (R Development Core Team 2020). An actual effect of interventions was assumed when the estimate (mean of the posterior distribution) plus the 95% credible intervals (two-sided) based on quantiles did not encompass zero (Bürkner Reference Bürkner2017). In Table S2, we report for each variable the mean value (estimate) and standard deviation (SD) of the posterior distribution and its 95% credible intervals, Rhat (potential scale reduction factor on split chains), and bulk and tail effective sample size. Rhat was equal to 1.00 for all variables in the final models, indicating convergence.
Results
A graphical visualisation of changes in species richness and in the abundance of forest species suggested the occurrence of effects of restoration, with larger variations occurring with restoration interventions than year (Figure S1). The analyses revealed a positive effect of achieved restoration (estimate ± SD of the posterior distribution: 0.25 ± 0.11) on the species richness of breeding species (i.e. breeding, non-raptorial, and non-aerial foraging taxa), whereas the effect was positive but not supported for the other community variables (Table S2). Year had no supported effects on community variables.
A total of 28 species were surveyed in at least 15 census events and were thus selected for the species-specific analyses. Models showed some converge issues for two species, namely Common Swift Apus apus and Eurasian Jay Garrulus glandarius, which were then left out of the analyses. Out of the remaining 26 species, models showed an effect (i.e. estimates plus credible intervals did not encompass zero) of restoration interventions for five species. The effect of on-going interventions (summarised in Fig. 3) was negative (estimate ± SD: -0.92 ± 0.46) for Barn Swallow Hirundo rustica and positive for Common Blackbird Turdus merula (estimate ± SD: 0.59 ± 0.23) and Hooded Crow Corvus corone cornix (estimate ± SD: 0.73 ± 0.24). The after-intervention restoration effect was positive for Common Blackbird (estimate ± SD: 0.57 ± 0.23) and Marsh Tit (estimate ± SD: 2.31 ± 1.32) and negative for Feral Pigeon Columba livia var. domestica (estimate ± SD: -1.12 ± 0.22). For all the other effects, see Table S2.
Discussion
Bird communities in peri-urban areas have been the object of many studies addressing planning and conservation issues (e.g. Hamza et al. Reference Hamza, Hanane, Almalki and Chokri2023; Mason et al. Reference Mason, Moorman, Hess and Sinclair2007), including within the study region (Bani et al. Reference Bani, Massimino, Bottoni and Massa2006; Padoa-Schioppa et al. Reference Padoa-Schioppa, Baietto, Massa and Bottoni2006; Saporetti Reference Saporetti2022). However, the effects of forest restoration on avian communities in the peri-urban context have been poorly considered. In the conurbation stretching northwards from Milan, our monitoring programme based on a BACI protocol allowed an evaluation of the “true” immediate effect of peri-urban forest restoration on the local avian communities. Results showed how the local avian community quickly reacted to a relatively limited restoration programme, only two years after interventions, with statistically supported positive effects of restoration on breeding species richness and on the abundance of some individual species, including a forest bird, i.e. Marsh Tit. Restoration outcomes included both transitory impacts of on-going interventions on a few species, and effects on communities and individual species. Those findings represent the first evidence based on a BACI approach of the effectiveness of forest restoration in peri-urban areas for biodiversity conservation using breeding birds as a model.
These patterns confirm the extremely high sensitivity of birds to habitat changes, and highlight the potential importance of forest restoration programmes in peri-urban areas. Such restoration efforts may thus contribute to biodiversity conservation, and, at the same time, to key ecosystem services. By improving conditions for bird communities, these interventions targeted at forest restoration in areas close to towns may be important not only for landscape-based (Dobbs et al. Reference Dobbs, Escobedo and Zipperer2011), but also for bird-related ecosystem services (Gaston Reference Gaston2022), including the promotion of mental health (Methorst et al. Reference Methorst, Bonn, Marselle, Böhning-Gaese and Rehdanz2021) or recreational opportunities (Brambilla and Ronchi Reference Brambilla and Ronchi2020).
Our study clearly shows some limitations. The main one is represented by the short period encompassed by the study, which does not allow for the evaluation of long-term effects of the restoration interventions (see further comments below). A second limitation is related to the fact that slightly different interventions had been implemented in different sites. Although all the interventions were targeted at improving the condition and/or the connectivity of forest patches, the use of different plant species, different restoration area, and/or the spatial arrangement of planted trees and shrubs might prove to be more or less beneficial to forest birds. The sample size was not large enough to explore the potential effect of individual restoration interventions, which should ideally be investigated over larger areas and sample sites.
Species-specific trajectories
Some generalist species were favoured by early-stage habitats resulting from restoration activities. Common Blackbird and Hooded Crow were positively associated with on-going interventions. Blackbirds might have been advantaged by the availability of bare, soft ground, particularly suitable for hunting earthworms (Perkins et al. Reference Perkins, Whittingham, Bradbury, Wilson, Morris and Barnett2000), while Hooded Crows could have benefitted from stressed conditions, with the on-going works increasing the availability of highly detectable prey (exposed because of tillage, clearing, etc.; cf. Atkinson et al. Reference Atkinson, Buckingham and Morris2004). On the other hand, Barn Swallow was negatively affected by on-going interventions, where disturbance due to works might have impacted foraging conditions for the species by direct (e.g. noise, high human presence) or indirect (e.g. reduction of flying insects) effects.
Common Blackbird and Marsh Tit were positively affected by restoration (after the interventions had taken place). Conversely, in the case of Feral Pigeon, a negative effect was found. Feral Pigeons typically forage in peri-urban fields (Lardelli et al. Reference Lardelli, Bogliani, Brichetti, Caprio, Celada and Conca2022), and the conversion of open and cultivated habitats into shrub/tree mosaics might have reduced foraging opportunities. In this specific case, such an effect provides further broader benefits, considering that Feral Pigeon is often regarded as a pest species in urban and suburban areas (Giunchi et al. Reference Giunchi, Albores-Barajas, Baldaccini, Vanni, Soldatini, Larramendy and Soloneski2012). Negative impacts on open-habitat specialists could be expected to occur because of reforestation; however, interventions focused on small extents, took place largely close to existing forest patches and were implemented in areas where open habitats are generally scarce and/or degraded, and this probably limited the impact. Apart from the negative effect of on-going interventions on Barn Swallow, a typical open-habitat species (Roseo et al. Reference Roseo, Salvatori, Brambilla, Pedrini, Fedrigotti and Bertocchi2024), no other statistically supported negative effect of restoration efforts was found on open-habitat species (see Table S3).
Among the forest species occurring in the area, only Marsh Tit (a forest specialist) appeared to increase in relation to restoration interventions. The lack of effect for other forest species is consistent with the long time required by restored forests to properly develop and mature, achieving more suitable conditions for many other species. Forest can take centuries to reach maturity (Harmon and Pabst Reference Harmon and Pabst2015), and many species are largely or entirely confined to mature forests (Morris et al. Reference Morris, Porneluzi, Haslerig, Clawson and Faaborg2013): it is therefore likely that most forest specialists would benefit from restoration activities only after decades. On the other hand, the location of many restoration interventions within existing forest patches likely benefitted forest species even in the short term, and rapid responses to restoration have been shown by birds in other contexts (e.g. Hutto et al. Reference Hutto, Flesch and Fylling2014; Passell Reference Passell2000). In the specific case of Marsh Tit, which has been reported not to favour forest edges (either because of structure or predator/weather effects), enlarging the existing forest patches may have also expanded the suitable habitat by reducing such edge effects (Broughton et al. Reference Broughton, Hill, Freeman, Bellamy and Hinsley2012): consistent with this hypothesis, all points where Marsh Tits increased were characterised by interventions increasing existing forest patches, rather than creating new ones. This positive impact on Marsh Tit, a specialised species characterised by short dispersal distances and high sensitivity to habitat fragmentation (Broughton et al. Reference Broughton, Hill, Bellamy and Hinsley2010), suggests that the restoration interventions will likely exert positive impacts on forest biocenoses.
Conclusions
This study showed that even small-scale interventions in peri-urban areas may provide tangible benefits to breeding birds, one of the most sensitive and employed indicators of ecosystem restoration. Therefore, peri-urban restoration of forest habitats could potentially contribute not only to increasing the landscape potential for regulating ecosystem services and recreational purposes (e.g. Davies et al. Reference Davies, Doick, Handley, O’Brien and Wilson2017; Gómez-Baggethun and Barton Reference Gómez-Baggethun and Barton2013; Roeland et al. Reference Roeland, Moretti, Amorim, Branquinho, Fares and Morelli2019), but also for biodiversity conservation by increasing opportunities for avian communities, and for citizens’ mental health (Methorst et al. Reference Methorst, Bonn, Marselle, Böhning-Gaese and Rehdanz2021). Further studies should evaluate the long-term impact on bird abundance, and ideally also avian species’ fitness, ecology, and behaviour in restored habitats (Hale and Swearer Reference Hale and Swearer2017).
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0959270924000200.
Acknowledgements
We are particularly grateful to R. Falco and V. Bergero for their help during different phases of the project. M. Noseda, G. Fontana, and M. R. Gelso kindly provided bird pictures. We are grateful to an anonymous reviewer and the Editor, F. Casas, for helpful comments on a first draft of the manuscript. The study was partially co-funded by Fondazione Cariplo to Fondazione Lombardia per l’Ambiente and other project partners within the project “Dal Lura alle Groane e alle Brughiere, dal Seveso al Parco Nord: Fiumi e parchi in rete per erogare servizi ecosistemici alla città diffusa”.