Ann. Zool. Fennici 51: 349–370 ISSN 0003-455X (print), ISSN 1797-2450 (online) Helsinki 29 August 2014 © Finnish Zoological and Botanical Publishing Board 2014 Evaluating habitat suitability for the middle spotted woodpecker using a predictive modelling approach Krystyna Stachura-Skierczyńska & Ziemowit Kosiński* Department of Avian Biology and Ecology, Faculty of Biology, Adam Mickiewicz University, Umultowska 89, PL-61-614 Poznań, Poland (*corresponding author’s e-mail: [email protected]) Received 3 June 2013, final version received 1 Nov. 2013, accepted 28 Nov. 2013 Stachura-Skierczyńska, K. & Kosiński, Z. 2014: Evaluating habitat suitability for the middle spotted woodpecker using a predictive modelling approach. — Ann. Zool. Fennici 51: 349–370. This paper explores the influence of forest structural parameters on the abundance and distribution of potential habitats for the middle spotted woodpecker (Dendrocopos medius) in three different forest landscapes in Poland. We applied predictive habitat suitability models (MaxEnt) based on forest inventory data to identify key environ- mental variables that affect the occurrence of the species under varying habitat condi- tions and the spatial configuration of suitable habitats. All models had good discrimi- native ability as indicated by high AUC values (> 0.75). Our results show that the spe- cies exhibited a certain degree of flexibility in habitat use, utilizing other habitats than mature oak stands commonly associated with its occurrence. In areas where old oak- dominated stands are rare, alder bogs and species-rich deciduous forests containing other rough-barked tree species are important habitats. Habitat suitability models show that, besides tree species and age, an uneven stand structure was a significant predictor of the occurrence of middle spotted woodpecker. The total area of suitable habitats in the studied forests varied from 9% to 60%. Predictive habitat models identified several concentrations of suitable habitats (clusters) with the average distances between them ranging from 3.2 to 5.0 km. Although these distances lie within the species’ dispersal ability, the migration of individuals between these sites might be difficult due to the necessity of travelling long distances through unsuitable forest types. Introduction recent years, predictive modelling of species distributions has become an important tool for Identifying the habitat structure and composi- supporting conservation planning, i.e., for the tion that fulfils critical requirements for threat- selection of representative habitat networks to ened or indicator species is an important step meet species conservation goals (e.g. Rodriguez in developing promising conservation strategies. et al. 2007, Suárez-Seoane et al. 2008, Stachura- In particular, if habitats are subjected to human Skierczyńska et al. 2009, Fernández & Gur- exploitation, responsible management decisions rutxaga 2010, Mateo et al. 2013). must be based on reliable, quantitative habitat The scope of most studies on forest-dwelling models (Angelstam et al. 2003, 2004, Angelstam birds, such as woodpeckers (family Picidae), & Donz-Breuss 2004, Roberge et al. 2008). In is limited due to the cost and time required 350 Stachura-Skierczyńska & Kosiński • Ann. ZOOL. FEnnici Vol. 51 to execute the study. Environmental variables in the most recent decades. This decline has used to explain the habitat selection of birds resulted from the interaction of biotic and abiotic are frequently derived from small sample plots factors and has a mass character with periods of (e.g. Pasinelli 2000, Kosiński & Winiecki higher and smaller disease intensity (Thomas 2004, Müller et al. 2009, Delahaye et al. 2010, et al. 2002, Sonesson & Drobyshev 2010). The Rehnus et al. 2011). It is accepted, as is apparent appearance of oak decline may affect the sur- from the management or conservation implica- vival of middle spotted woodpecker in the future. tions developed by various authors, that results Therefore, identification of factors affecting the obtained from a limited area can be used effec- persistence of the middle spotted woodpecker tively for habitat management at a larger scale, in different types of forests is essential to main- e.g. on a landscape or regional scale. However, taining viable populations across the species habitat variation occurs across multiple scales, range. Moreover, identification of key habitats and detected patterns should not be generalized for middle spotted woodpeckers and their pro- beyond the extent of a given study (Wiens 1989). tection may also benefit non-vertebrate species Since it is not easy to obtain reliable extrapola- with similar habitat requirements, e.g. saproxylic tions of data from small to larger areas, their beetles and moths associated with oaks (Müller applicability by conservationists and land man- & Goßner 2007, Ranius et al. 2009). The middle agers may be limited. Therefore, to gain knowl- spotted woodpecker is listed in Annex I of the edge that would be applicable to forest manage- European Community Birds Directive (Directive ment at the landscape or regional scale, studies 2009/147/UE). Therefore, it is the subject of spe- of the habitat requirements of species should be cial conservation measures concerning its habitat conducted at different spatial scales (e.g. Rolstad in order to ensure its survival and reproduction. et al. 2000, Robles et al. 2007b) and replicated Most studies regarding middle spotted wood- in several regions (Dunning 2002, Roberge et al. pecker habitat selection have been confined 2008). to small spatial scales, i.e., nest-site selection Due to the long-term exploitation of Euro- (Pasinelli 2000, Kosiński & Winiecki 2004, pean forests, majority of primeval woodpecker Kosiński et al. 2006) and the trees used for for- habitats have either disappeared or been replaced aging (Pasinelli & Hegelbach 1997). Studies at by semi-natural or artificial commercial stands larger, macrohabitat scales describe the habitat with lower tree species diversity and only a use or population with regard to the area, age and selection of the few most productive tree species type of forest (deciduous vs. coniferous), as well (Mikusiński & Angelstam 1997). Such human as to the size and density of potential nesting/for- activities have apparently degraded primary hab- aging trees (Kosiński & Winiecki 2005, Robles itats for middle spotted woodpeckers (Dendroco- et al. 2007b, Müller et al. 2009). Moreover, pos medius) (Petterson 1985a, 1985b, Pasinelli some studies focus on habitat use, demography 2003, Roberge & Angelstam 2006, Robles et and population persistence at the macrohabitat al. 2007a, Ciudad et al. 2009), a habitat spe- scale (Kossenko & Kaygorodova 2001, Robles cialist restricted to mature stands with rough- et al. 2007a, Robles et al. 2008, Ciudad et al. barked tree species, mainly oaks (Müller 1982, 2009, Robles & Ciudad 2012). Since habitat Jenni 1983, Török 1990, Pasinelli & Hegelbach suitability is correlated with population persist- 1997, Pasinelli 2000, Kosiński 2006, Müller ence (Cabeza et al. 2004, Rodriguez et al. 2007, et al. 2009). Other potential habitats are rarely Robles & Ciudad 2012), an understanding of the reported and include mature alder stands (Noah factors influencing habitat suitability for middle 2000, Weiß 2003) and very old pure beech for- spotted woodpeckers is necessary for identifica- ests (Winter et al. 2005). Oak forests have been tion and protection of optimal areas and for guid- favoured by humans for centuries as sources of ing forest management. brushwood, timber, and acorns for pigs (Kas- Environmental features important for the prowicz 2010). However, incidences of oak middle spotted woodpecker are most often those decline in central Europe have occurred repeat- that are recorded in classical forest inventories edly during the past three centuries as well as (Müller et al. 2009, Walczak et al. 2013). More- Ann. ZOOL. FEnnici Vol. 51 • Evaluating habitat suitability for the middle spotted woodpecker 351 over, some additional characteristics describing the complexity of within-stand structure can be simply obtained directly from basic data (Kurlavicius et al. 2004, Stachura-Skierczyńska et al. 2009). Some studies have reported the importance of decaying or cavity trees, which are not normally included in forest inventories; however, these factors have usually been found to be of lesser significance (Pasinelli 2000). Nev- ertheless, environmental parameters describing tree species composition, the vertical structure of stands, etc., are frequently omitted in studies of habitat selection of the middle spotted wood- pecker (however, see Pavlik 1994, Roberge et al. 2008, Delahaye et al. 2010, Walczak et al. 2013). In this study, we analyze environmental fac- Fig. 1. Distribution of study sites across Poland (KOF: tors affecting the occurrence of middle spotted Krotoszyn Oak Forest, FDR: Forest at the Drawa River, woodpecker in three forests sited within Special KF: Knyszyn Forest). Protection Areas, and differing in size, spatial structure, habitat conditions and history of use. Our goal was to develop a landscape-scale pre- petreae, locally Calamagrostio arundinaceae– dictive model of potential nesting habitats for Quercetum petraeae) in central Europe (Kaspro- middle spotted woodpeckers. In particular, our wicz 2010). Stands with dominant oak (mainly objectives were: (1) to identify key environmen- Pedunculate oak Quercus robur) comprise about tal variables that