Vol.:(0123456789) Oecologia (2024) 206:61–72 https://doi.org/10.1007/s00442-024-05613-z ORIGINAL RESEARCH Correlation between mammal track abundance and Forest Landscape Integrity Index validates actual forest ecological integrity Francesca Malcangi1   · Andreas Lindén2 · Janne Sundell1 · John Loehr1 Received: 9 November 2023 / Accepted: 21 August 2024 / Published online: 4 September 2024 © The Author(s) 2024 Abstract Human disturbance compromises the ecological integrity of forests, negatively affecting associated species. Assessing the impact of forest integrity on biodiversity is complex due to the interplay of various human activities, ecological factors, and their interactions. Current large-scale indices assess forest integrity but often lack a direct connection to the biotic environment. We tested the effectiveness of the global Forest Landscape Integrity Index (FLII) in evaluating aspects of anthropogenic forest degradation on the biotic community. We analyzed the relationship between changes in the ecological integrity of Finnish forests and variations in mammal species abundance, using the number of tracks from 17 different species collected during the winter seasons between 2016 and 2020 in south-central Finland. Beyond the FLII, we analyzed forest and canopy cover to enhance the accuracy of habitat preference assessments. We found that the FLII captures the varying degrees of forest integrity, as reflected by the correlation between the abundance of winter tracks and the FLII for most mammals. Species that were positively associated with forest integrity were all native to the boreal forest, while mammals that adapt well to human-disturbed environments including two invasive species were more common in lower FLII forests. Significant differences in habitat preferences were also observed in relation to forest and canopy cover, revealing additional nuances that the FLII alone did not capture. This study demonstrates that the FLII, when combined with a comprehensive dataset and supplemented with region-specific factors, can assess species' adaptability to human-modified forests, aiding in the development of conservation strategies. Keywords  Ecological integrity · Habitat loss · Boreal forest · Mammals · Native species Introduction The degradation of natural habitats due to human activi- ties is the major driver of biodiversity loss globally (Storch et al. 2022). In forests worldwide, tree cutting for logging and other land uses, such as agriculture, livestock areas, and urbanization, are primary causes of disturbance (Seidl and Turner 2022). These activities alter the ecological integ- rity of forests, defined as the degree to which an ecosystem remains free from anthropogenic changes in its structure, composition, and function, leading to negative consequences for many associated species (Parrish et al. 2003; Grantham et al. 2020). Forest disturbances, both natural and human-induced, have significant ecological consequences. In boreal forests, for instance, natural events, such as fires and insect out- breaks, can cause substantial impacts. In addition to natural disturbances human activities, particularly those that are intense and prolonged, add further pressure on ecosystems (Rockström et al. 2009). These disturbances can lead to changes in community composition (Hagen et al. 2012; Ven- ier et al. 2014) and ecosystem functionality (Cardinale et al. 2006; Folke et al. 2004), ultimately resulting in the loss of essential ecosystem services with significant socioeconomic impacts (Cardinale et al. 2012). In Finland, despite having the highest forest cover in Europe—over 75% according to the Ministry of Agricul- ture and Forestry—42% of all species in Finland live in these forests, and 30% of the threatened species use them Communicated by Indrikis Krams. * Francesca Malcangi francesca.malcangi@helsinki.fi 1 Lammi Biological Station, University of Helsinki, Pääjärventie 320, 16900 Lammi, Finland 2 Natural Resources Institute Finland (Luke), 00790 Helsinki, Finland http://orcid.org/0000-0002-8412-1867 http://crossmark.crossref.org/dialog/?doi=10.1007/s00442-024-05613-z&domain=pdf 62 Oecologia (2024) 206:61–72 as their primary habitat. The Red List 2019 indicates that of the 58 mammal species assessed, around 21% are classified as threatened (Hyvärinen et al. 2019). Critical factors leading to species endangerment or extinction include changes in forest habitats due to forest management activities, the reduction of old-growth forests, large trees, and the decreasing amounts of dead and decay- ing wood (Jonsson et al. 2016; Mönkkönen et al. 2022). The challenge of achieving sustainable forest use in Finland arises from the intensive industrial exploitation of boreal forests, primarily through systematic silvicultural practices. This excessive use severely damages ecologically essential forest structures and processes, reducing habitat quality and the ecological capacity to support various native species vulnerable to environmental disturbances, thereby altering the composition of forest-dependent species communities (Semenchuk et al. 2022). Given these impacts, it is crucial to consider the potential effects of landscape changes on species abundance and dis- tribution, and to implement measures to mitigate any nega- tive consequences. However, making informed decisions on biodiversity is challenging for policymakers due to the lack of clear, measurable targets and the complexity of ecosys- tems, which complicates the assessment for the public and stakeholders (Oettel et al. 2021). Several studies have highlighted the need for indices that integrate both abiotic and biotic data to more accurately reflect ecological integrity (Weigel et al. 2002; Zhao et al. 2019; Zelený et al. 2021). Despite this, assessing forest integrity and degradation at a global scale is complex and requires reliable proxies, which often do not have a direct connection to the specific biotic environment (Pan et al. 2021; Jia et al. 2023). The first index to assess the ecological integrity of forests on a global scale is the Forest Landscape Integrity Index (FLII). It was developed to provide a comprehensive meas- ure of forest ecological integrity by combining observed reductions in forest connectivity with proxy data on human activity intensity known to cause forest degradation (Gran- tham et al. 2020). The FLII has revealed important insights into the state of the world's forests, showing, for example, that boreal forests in northern Russia, Canada, and Alaska have large areas of high integrity, whereas regions like Fen- noscandia, including Finland, show low overall ecological integrity. In Finland, the national average FLII value is 5.08 out of a maximum of 10, with particularly low scores in southern Finland. This low ecological integrity can be cor- related with a higher concentration of threatened and red- listed species in these areas, highlighting the strong pres- sure exerted on their habitats by various land-use forms (Hyvärinen et al. 2019). In this context, understanding whether the FLII represents the ecological condition of the forest, such as the diversity and abundance of the species that inhabit it, can provide valuable insights into the relationship between forest integ- rity and species distribution, aiding in the development of conservation strategies for increasingly anthropized areas. The aim of this study is to assess whether the Forest Landscape Integrity Index (FLII) is a robust proxy for meas- uring forest integrity in relation to the biotic conditions, par- ticularly in the context of mammal abundance differences. This study also explores the potential of forest and canopy cover variations, which are not taken into account by FLII. This helps us to better investigate which habitat character- istics are important for the survival and thriving of different animal species in human-modified forests. By doing so, we can improve our understanding of the effective FLII’s appli- cation at fine-scale and offer valuable insights to address the challenge of evaluating the connection between forest integrity degradation and species status. Materials and methods Wildlife triangles The data on wild mammal abundance were based on the wildlife triangle scheme, which is a monitoring scheme of wildlife abundance organized by the Natural Resources Insti- tute Finland (Pellikka et al. 2005). The scheme is based on a network of triangle-shaped 4 × 4 × 4 km transects (totaling 12 km per triangle), with fixed locations covering the entire country. The spatial location of each triangle was available as the center point of the triangle. Mammal snow tracks have been surveyed each winter since 1989, at approximately 700 triangles located randomly in forested areas throughout Finland and the survey period is between January 15th and February 28th in southern and central Finland. All tracks of 24 mammal species crossing the transect are recorded. Counts are performed one or a few days after a snowfall or pre-count. During the pre-count, old tracks are marked to distinguish them from new tracks that accumu- late afterward. Considering the distance surveyed and the number of days since the last snowfall, a snow track index representing the number of snow tracks per 10 km per day is calculated. In addition to the tracks counted mainly in forested land- scapes, triangles in agricultural areas were also included in the analyses. These so-called “Field triangles” have sides of 2 km, totaling 6 km of tracking route, and are situated in mosaic-like landscapes where approximately half of the census lines traverse farmland, while the remainder passes through forests, yards, and built-up areas. Since their estab- lishment in 1999, field triangles have been regularly cen- sused, primarily located in Western and Southern Finland, where agricultural activity is most prevalent. 63Oecologia (2024) 206:61–72 Our study utilized snow track data collected from both wildlife and field triangles (Fig. 1), totaling 1063 triangle years distributed across 2016 to 2020 (247 triangles sampled in 2016, 217 in 2017, 230 in 2018, 182 in 2019, and 187 in 2020). Study species The analysis focused on 17 of the 24 mammal species sur- veyed in the winter wildlife triangle counts. The species were selected based on their relative abundance in the avail- able data, with those having fewer than 100 observations in total in the study area being excluded (The species and their relative abundance are reported in the Supplementary tables S1 and S2). We analyzed the following species that are native to the boreal forest: the mountain hare (Lepus timidus, L.), moose (Alces alces, L.), forest reindeer (Rangifer tarandus fennicus, L.), wolverine (Gulo gulo, L.), Eurasian lynx (Lynx lynx, L.), wolf (Canis lupus, L.), pine marten (Martes martes, L.), least weasel (Mustela nivalis nivalis,L.), stoat (Mustela erminea, L.), Eurasian otter (Lutra lutra), red squirrel (Sciurus vul- garis, L.), and red fox (Vulpes vulpes, L.). Although many of these species have distributions that include habitats other than the Eurasian taiga biome, they all are native to this environment. Some of them benefit from coexistence with humans in terms of increased density, reproduction, and/or survival. Among these, the red fox is considered a general- ist species (Wallén et al. 2018). The observed increase in fox populations can be attributed to human activities, which provide red foxes with more scavenging opportunities, habi- tats for preferred prey (e.g., voles), and higher predation success by increasing the amount of habitat edges (Jahren et al. 2020; Gorini et al. 2012; Güthlin et al. 2013; Vuorisalo et al. 2014). Red squirrel populations in Finland are more abundant in urban and rural areas than in forests. One factor that could explain this phenomenon is the increase in winter feeding sites in urban environments, especially during the harvest period of spruce cones, which are the main food of red squirrels (Jokimäki et al. 2017). Finally, the mustelids include the least weasel and stoat, which prefer farmland habitats, especially during the peak density phase of the Fig. 1   Track data collected from wildlife (in green) and field (in orange) triangles. At the bottom right of the figure is an example of a 2.5 km buffer around the center of a triangle with the tracks of different mammal species (in black) 64 Oecologia (2024) 206:61–72 Microtus vole on which they feed (Klemola et al. 1999). Other studies have found that fragmented agricultural land- scapes with residual edge habitats favor the cover and food requirements of these species (Magrini et al. 2009). Our analysis also included invasive species present in the study area: the brown hare (Lepus europaeus, P.) and roe deer (Capreolus capreolus, L.), which naturally dis- persed into Finland and have become abundant only in the last 50–150 years, showing a high ability to tolerate anthro- pogenic land use and habitat modifications (Levänen et al. 2019a, b; Linnell et al. 2020; Tasser et al. 2023). White- tailed deer (Odocoileus virginianus, L.), raccoon dog (Nyc- tereutes procyonoides, L.), and American mink (Neogale vison), which were introduced by humans and whose popu- lations were successful and increased rapidly in the new areas (Kauhala and Helle 1995; Bonesi and Palazon 2006; Poutanen et al. 2023). Study area We included data from the southern and central parts of Finland in our analyses (Fig. 1). The study area also encom- passes the wildlife districts and associations known as 'riistakeskus' and 'rhy,' where the survey triangles are geo- graphically grouped. The main reason we restricted our analysis to this study area is that all species are found there. However, for some species with a narrower distribution range—in particular the forest reindeer, lynx, wolverine, wolf, roe deer, white-tailed deer, brown hare and raccoon dog—we restricted our analysis to geographical areas with a relevant number of track observations, constructing a mini- mum convex polygon (MCP) or convex hull of all positive observations in the data for the years considered. By exclud- ing Northern Finland, it was also possible to separate the effects of latitude and forest integrity without encountering multicollinearity problems with these variables. Explanatory variables Initially, we tested whether FLII is a good proxy for forest ecological integrity by analyzing its association with the abundance of species tracks. Subsequently, in addition to FLII, we also analyzed forest cover and canopy cover to gain a more complete understanding of habitat suitability for mammal species. This is because, although the FLII provides an assessment of anthropogenic disturbances in the forest, changes in forest structure, which can be an important indicator of species' forest preferences, forest cover and canopy cover are not directly extractable from the FLII score. In particular, the inclusion of these two variables makes it possible to distinguish areas that have the same level of anthropogenic disturbance (as indicated by the FLII), but differ significantly in their physical struc- ture and, consequently, in their suitability as habitat for different mammal species. Forest Landscape Integrity Index (FLII) The FLII is the first tool designed to infer the ecologi- cal integrity of forests for conservation, management, and restoration at global scale. It is constructed based on three main aspects: 1) observed pressures, 2) inferred pressures, and 3) changes in forest connectivity. The definition of forest in the FLII baseline dataset includes a 20% canopy cover threshold and all woody vegetation greater than 5 m in height, encompassing naturally regenerated forests, tree crops, planted forests, wooded agroforests, and urban tree cover. The observed pressures include direct human impacts, such as infrastructure, agriculture, and deforestation. These pressures are quantified using spatial datasets that map the presence and intensity of these activities. Each type of pressure is assigned a weight based on its esti- mated impact on forest integrity. The weighted values are then combined to form a composite measure of observed pressures for each forest pixel, ensuring that areas with higher levels of human activity are appropriately reflected in the index. For the anthropogenic pressures that were inferred, Gran- tham et al. (2020) accounted for indirect impacts, such as the creation of microclimates, species interactions at forest margins, hunting, firewood collection, pollution, livestock grazing, and fire spread. The impacts were modeled based on their spatial association with observed pressures, with intensity declining with distance. Changes in forest connectivity were determined by com- paring the current forest configuration with a potential con- figuration that assumes no extensive human alterations. This comparison indicates the degree of forest fragmentation. These values were spatially applied to a global forest extent map for early 2019 by subtracting the 2001–2018 annual tree cover loss from the Global Tree Cover 2000 product. The final base map for 2019 was generated from 300 m reso- lution pixels classified as forest or non-forest according to a majority rule. Finally, these elements were combined to generate an FLII score for each forest pixel, reflecting overall forest integrity. Higher scores indicate less human impact and better ecological integrity. The FLII was calculated for each forest pixel, but not for non-forest pixels. We downloaded the FLII map used in this study from the official website (https://​www.​fores​tinte​ grity.​com/) in tiff format for Europe with a WGS 1984/300 resolution per pixel and the data multiplied by 1000 to store them in full format. https://www.forestintegrity.com/ https://www.forestintegrity.com/ 65Oecologia (2024) 206:61–72 Canopy cover The FLII described earlier uses a minimum threshold of 20% canopy cover to define a forest; however, this thresh- old represents a simple binary classification to distinguish between forest and non-forest and does not consider more detailed information about the varying amounts of canopy cover. To consider this we used geographic data on canopy cover (%), estimated in 2019, defined as the percentage of forest floor covered by the vertical projection of tree crowns. We obtained these data from the multiple source inventory operating system, which uses field inventories, satellite images, numerical base maps, and contour maps (Tomppo et al., 1998; Ministry of Agriculture and Forestry, 2000). The maps are in raster format with a pixel size of 16 × 16 m and in the ETRS-TM35FIN coordinate system. Cover values were obtained by Korhonen et al. (2007) through a modeling technique, beta regression, specifically designed to model percentage variables such as canopy cover. For forest soils and poorly productive soils, estimation was done separately based on dominant species Scots pine (Pinus sylvestris L.), Norway spruce (Picea abies L.), and others. Deciduous tree canopy cover was calculated based on the proportion of deciduous trees in the plot (calculated from stem counts in seedling stands and basal area of immature stands). Forest cover To account for the proportion of forest that characterizes the study area versus the percentage of area characterized by other land-use categories, we also analyzed forest cover, described as the amount of forest that covers a particular area of land. In our case, we obtained forest cover by cal- culating from the canopy cover raster the percentage ratio between the number of pixels containing forest and the num- ber of pixels with no values for the canopy cover variable. This measure helps distinguish between areas with similar FLII scores but varying forest density and configuration. In particular, whereas forest cover provides a quantitative measure of the physical space occupied by forests, which can directly influence the habitat available for species, FLII adds a qualitative dimension by assessing the ecological integrity of these forests. Spatial analysis Geospatial analysis characterized the first stage of evaluating the relationship between the explanatory variables and the number of tracks recorded for each species in the wildlife triangles (Fig. 1). We calculated the average FLII and the percentage of other covariates within a buffer with a radius of 2500 m around the georeferenced center of each triangle. This approach ensured that we captured the relevant habitat characteristics influencing species presence. We selected tracks of one species at a time and accounted for the geo- graphic distribution of each individual species by including only triangles within the geographic area where tracks of a given species were found to be present. We used QGIS ver- sion 3.32 to perform this analysis. Statistical analysis We modeled the relationship between the snow track index (the response variable) for each species and the explana- tory variables described above using a generalized linear mixed model (GLMM) with a logarithmic link function and a negative binomial error distribution. The number of tracks observed in a triangle for the species under study was techni- cally the response variable. However, we modeled the snow track index (tracks/10 km/day) by including an offset vari- able—the natural logarithm of the survey effort (Lindén and Piha 2024). The survey effort was represented by the dis- tance surveyed, which is typically around 12 km (1.2 units of 10 km), although this distance can vary slightly due to practical surveying conditions. Additionally, we included the logarithm of days since the last snowfall as a covariate. This allowed us to account for changes in the number of tracks observed with the days since the last snowfall. We only con- sidered days since the last snowfall less than or equal to five to account for the nonlinearity between the number of tracks and days since the last snowfall. The ID of the wildlife triangle was included in the model as a random effect on the intercept to account for the non-independence due to the same triangle being sur- veyed in different years. We also considered spatial autocor- relation between triangles by adding the wildlife districts ("riistakeskus") and wildlife associations ("rhy"), where the triangles are geographically grouped, thus defining a nested random effect in our models. In the first model, called 'FLII model,' we analyzed FLII and the latitude of the triangle location as the main explan- atory variables (fixed effects). Latitude was included as a control variable to account for potential climatic and envi- ronmental gradients that may influence species abundance, ensuring that the effects of forest-related variables are not confounded by latitudinal variations. In the second model, called 'full model,' we included forest cover and canopy cover in addition to FLII and latitude. For all covariates, we applied centering, a technique in which the mean of the inde- pendent variables is subtracted from all values (Schielzeth 2010). This ensures that all independent variables have a mean of zero, helping the model to work efficiently. The models were fitted using the "glmmTMB" package (Brooks M.E. et al. 2017) in the R programming environment ver- sion 4.2.1 (R Core Team, 2022). In each model, those with 66 Oecologia (2024) 206:61–72 p-value < 0.05 and 95% confidence intervals of the estimated effects not including zero were considered significant. To evaluate the effectiveness of the Forest Landscape Integrity Index (FLII) in predicting mammal abundance, we calculated the coefficient of determination R2 using a method adapted from Nakagawa and Schielzeth (2013) for generalized linear mixed-effects models. This method allows us to calculate both marginal R2 (variance explained by fixed effects) and conditional R2 (variance explained by both fixed and random effects). The conditional R2 also includes distribution-specific variance and residual variance due to overdispersion. When calculating the marginal R2, we excluded offset and days since last snowfall, as these are effort variables used to account for the structure of the snow track index in our models. We also calculated the standardized coefficients of the fixed effects explanatory variables to illustrate their effect sizes, and predictive abilities in relation to each other (Schielzeth 2010). The standardized coefficients and their standard errors (SE) were calculated by multiplying the original estimates with the sample standard deviations of the explanatory variables. Hence, we standardize only with respect to the explanatory variables (Newman and Browner 1991), not all input variables, as is typically applied for lin- ear regressions. We assessed potential multicollinearity among the predic- tor variables using Variance Inflation Factor (VIF) analysis. In general, the VIF values were low for all variables, indi- cating low multicollinearity. Specifically, the VIF values for FLII ranged from 1.21 to 2.56, while those for forest cover ranged from 1.08 to 1.56. These values suggest low multicol- linearity among these predictors (Zuur et al. 2009). Additionally, residual diagnostics were checked using the ‘DHARMa’ package (Hartig and Lohse 2022). The ‘DHARMa’ package uses a simulation-based approach to create readily interpretable scaled (quantile) residuals for fitted (generalized) linear mixed models. The resulting residuals are standardized to values between 0 and 1 and can be interpreted as intuitively as residuals from a linear regression. Results We analyzed the relationship between the Forest Landscape Integrity Index (FLII) and snow tracks for 17 different spe- cies (Table 1, Fig. 2). We found that the FLII was positively associated with the abundance of seven of the studied spe- cies, including the Eurasian lynx, forest reindeer, moose, mountain hare, pine marten, wolf, and wolverine. A negative association was found for the brown hare, roe deer, red fox, and red squirrel. Observing the multiplicative effects of FLII on abundance, measured as standardized coefficients (beta), we found very large positive effects for the forest reindeer and wolf (beta = 1.3–1.4), and large negative effects for the brown hare and roe deer (beta = − 0.8 to − 1.2). Addition- ally, high positive effects were observed for the moose, pine marten, and wolverine (beta = 0.4–0.6), and moderately high positive effects for the Eurasian lynx and mountain hare (beta = 0.24–0.28) (Supplementary Table S3). We then tested the full model, which included forest cover and canopy cover as additional independent variables along with FLII (Table 2, Fig. 3). Our results suggest that forest cover was positively associated with the abundance of the same species that showed a positive correlation with FLII in the FLII model, and in addition for the white-tailed deer and European otter. In contrast, for other species, particularly the brown hare, raccoon dog, roe deer, red fox, and red squirrel, we observed an increase in track abundance with decreasing forest cover. Additionally, our analysis revealed that canopy cover was positively correlated with snow track abundance for the American mink, Eurasian lynx, least weasel, moun- tain hare, raccoon dog, red squirrel and stoat. Conversely, for the moose, red fox, and roe deer and wolf, we found a negative relationship. We calculated the marginal and conditional R2 values for both the FLII model (Table 1) and the full model, includ- ing forest cover and canopy cover as independent variables (Table 2). The results indicate that the full model explains more variance overall than the FLII model. Specifically, the marginal R2 values showed significant variation among different species. Higher marginal R2 values (> 0.3) model were found for the brown hare and white-tailed deer, indicat- ing that the fixed effects in the model explain a considerable portion of the variance in the snow track index for these species. In contrast, species like the forest reindeer and the wolf had very low marginal R2 values (< 0.01), indicating that the FLII model explains only a small fraction of the variance for these species. The inclusion of forest and canopy cover variables signifi- cantly improved the marginal R2 values for several species. This improvement was particularly pronounced, with an increase of at least 5%, for the brown hare, moose, mountain hare, and pine marten. Notably, for the moose and mountain hare, all the added variables were statistically significant. However, for certain species, such as the forest reindeer, the increase was marginal, indicating that these additional fac- tors do not substantially enhance the explanation of variance for these species. According to the effect sizes of the explanatory vari- ables (Supplementary Table S4), we observed that in the full model forest cover had a predominant effect, decreasing the explanatory power of FLII for most species. Although canopy cover also contributed to the full model, its impor- tance appeared less significant than forest cover. 67Oecologia (2024) 206:61–72 Ta bl e  1   P ar am et er e sti m at es fr om sp ec ie s- sp ec ifi c G LM M s e xa m in in g fa ct or s a ffe ct in g sn ow tr ac k ab un da nc e in w ild lif e tri an gl e ce ns us es (2 01 6– 20 20 ) Th e ta bl e in cl ud es c oe ffi ci en ts fo r F LI I a nd la tit ud e, th ei r 9 5% c on fid en ce in te rv al s, an d si gn ifi ca nt v al ue s in b ol d. N es te d ra nd om e ffe ct v ar ia nc es a nd m od el d is pe rs io n pa ra m et er a re a ls o pr e- se nt ed , a lo ng w ith m ar gi na l a nd c on di tio na l R 2 v al ue s i nd ic at in g va ria nc e ex pl ai ne d by th e fix ed e ffe ct s a nd th e to ta l m od el , r es pe ct iv el y Si gn ifi ca nt c oe ffi ci en ts , w ith p -v al ue s < 0 .0 5 an d 95 % c on fid en ce in te rv al s n ot in cl ud in g ze ro , w er e hi gh lig ht ed in b ol d Sp ec ie s In te rc ep t FL II La tit ud e D ist ric ts v ar ia nc e A ss oc ia - tio ns v ar i- an ce Tr ia ng le s v ar ia nc e D is pe rs io n pa ra m et er R 2 m R 2 c Tr ia n- gl es sa m pl ed A m er ic an m in k −  2 .1 60 (−  2 .5 77 −  1 .7 43 ) 0. 12 1 (−  0 .0 15 0 .2 56 ) 0. 27 9 (0 .0 72 0 .4 85 ) 0. 20 5 0. 11 7 2. 15 1 0. 72 3 0. 07 4 0. 51 7 10 63 B ro w n ha re − 0 .3 94 (− 0 .9 96 0 .2 07 ) − 0 .7 00 (− 0 .8 39 − 0. 56 1) − 0 .6 41 (− 0 .9 74 − 0. 30 7) 1. 02 6 0. 58 9 3. 42 1 1. 47 5 0. 34 8 0. 89 6 87 1 Eu ra si an ly nx − 0 .6 97 (− 1 .1 83 − 0. 21 0) 0. 16 2 (0 .0 53 0 .2 71 ) − 0 .2 13 (− 0 .4 55 0 .0 28 ) 0. 61 1 0. 27 1 1. 27 0 0. 65 9 0. 01 8 0. 45 4 99 5 Eu ro pe an o tte r − 1 .8 35 (− 2 .1 39 − 1. 53 1) 0. 06 1 (− 0 .0 59 0 .1 82 ) − 0 .0 12 (− 0 .1 97 0 .1 73 ) 0. 02 0 0. 14 2 2. 21 9 1. 29 0. 02 0. 68 1 10 63 Fo re st re in de er − 3 .4 65 (− 5 .4 20 − 1. 51 1) 0. 82 0 (0 .2 66 1 .3 75 ) − 0 .1 39 (− 1 .1 78 0 .9 00 ) 3. 63 7 3. 81 2 1. 44 E- 07 0. 03 3 0. 00 2 0. 01 56 8 Le as t w ea se l − 0 .8 38 (− 1 .0 89 − 0. 58 6) − 0 .0 57 (−  0 .1 76 0 .0 62 ) − 0 .0 18 (− 0 .1 46 0 .1 10 ) 8. 66 E- 09 0. 04 1 2. 77 4 0. 66 5 0. 01 1 0. 48 3 10 63 M oo se 1. 77 6 (1 .5 49 2 .0 02 ) 0. 34 4 (0 .2 48 0 .4 39 ) − 0 .2 63 (− 0 .4 03 − 0. 12 2) 0. 08 5 0. 18 0 2. 34 8 1. 36 5 0. 05 1 0. 72 6 10 63 M ou nt ai n ha re 3. 06 2 (2 .7 91 3 .3 32 ) 0. 14 4 (0 .0 68 0 .2 21 ) 0. 05 4 (− 0 .0 97 0 .2 05 ) 0. 19 5 0. 09 1 1. 75 2 2. 31 1 0. 04 4 0. 8 10 63 Pi ne m ar te n 0. 08 2 (− 0 .2 29 0 .3 92 ) 0. 25 6 (0 .1 63 0 .3 50 ) − 0 .0 65 (− 0 .2 31 0 .1 02 ) 0. 20 7 0. 18 6 1. 72 9 1. 07 9 0. 04 1 0. 60 7 10 63 R ac co on d og − 1 .4 77 (− 1 .8 69 − 1. 08 5) 0. 01 8 (− 0 .1 00 0 .1 35 ) − 0 .9 52 (− 1 .1 49 − 0. 75 5) 0. 17 1 0. 34 1 1. 37 3 0. 40 6 0. 21 3 0. 37 8 10 05 Re d fo x 1. 82 1 (1 .5 17 2 .1 26 ) − 0 .0 79 (− 0 .1 54 − 0. 00 3) − 0 .1 33 (− 0 .2 96 0 .0 30 ) 0. 24 7 0. 31 9 1. 49 2 3. 01 3 0. 04 5 0. 84 7 10 63 Re d sq ui rr el 0. 96 9 (0 .8 17 1 .1 21 ) − 0 .0 98 (− 0 .1 83 − 0. 01 3) − 0 .2 21 (− 0 .3 16 − 0. 12 5) 8. 05 E- 09 0. 13 9 1. 81 6 1. 35 2 0. 08 2 0. 67 4 10 63 Ro e de er − 1 .7 59 (− 2 .5 64 − 0. 95 4) − 0 .5 01 (− 0 .6 62 − 0. 33 9) − 0 .6 31 (− 1 .0 29 − 0. 23 2) 1. 82 8 1. 01 3 4. 19 4 0. 76 8 0. 21 7 0. 79 8 94 4 St oa t − 0 .4 59 (− 0 .6 80 − 0. 23 7) 0. 00 4 (− 0 .1 05 0 .1 13 ) 0. 36 7 (0 .2 54 0 .4 81 ) 1. 40 E- 07 0. 00 7 1. 98 0 0. 67 5 0. 08 4 0. 44 9 10 63 W hi te -ta ile d de er − 0 .7 89 (− 1 .8 78 0 .3 01 ) 0. 07 2 (− 0 .1 01 0 .2 45 ) − 2 .2 60 (− 2 .8 36 − 1. 68 3) 3. 39 5 0. 83 9 3. 60 6 1. 13 2 0. 39 2 0. 90 9 68 3 W ol f − 4 .1 28 (− 5 .5 30 − 2. 72 7) 0. 78 6 (0 .4 28 1 .1 45 ) − 0 .2 71 (− 0 .8 01 0 .2 59 ) 0. 37 4 3. 63 8 0. 73 2 0. 06 1 0. 00 8 0. 02 5 86 1 W ol ve rin e − 3 .3 71 (− 4 .1 68 − 2. 57 3) 0. 23 8 (0 .0 48 0 .4 27 ) 0. 94 3 (0 .4 81 1 .4 05 ) 0. 77 2 1. 69 0 2. 20 0 0. 67 4 0. 22 8 0. 69 7 67 8 68 Oecologia (2024) 206:61–72 Discussion Our study demonstrated that the variation in track abun- dance among species reflects different levels of forest ecological integrity. The Forest Landscape Integrity Index (FLII) effectively captures these variations and provides valuable insights into how different mammal species respond to varying levels of human disturbance in Finn- ish forest habitats. Preference for forests with higher FLII scores—indicating less human disturbance—tended to be more common for native and more sensitive mammal spe- cies, such as the forest reindeer, wolf, moose, wolverine, pine marten, lynx, and mountain hare. Conversely, more disturbed forests with lower FLII scores exhibit higher abundances of adaptable species that thrive in human- altered environments, including the brown hare, roe deer, red fox, and red squirrel. This finding is further supported by the standardized coef- ficients presented in Supplementary Table S3, which shows large absolute effects of FLII on abundance in the models with FLII as the only forest-related variable. It seems that FLII alone serves as a strong indicator of forest integrity affecting species abundance. Notably, for two of the four species with negative associations with FLII, namely roe deer and brown hare, the FLII demonstrates a substantial effect size on track abundance. The brown hare, originally a steppe species from Central Asia and continental Europe, has expanded northward in Finland over the past three decades, partly due to its adaptability to various habitats, especially open fields (Thulin, 2023; Caravaggi et al. 2015). Similarly, the roe deer, recognized as the most widespread deer in Europe, adapts to diverse environments and relies on cultivated fields for foraging, especially in winter (Tinoco Torres et al. 2011). Given FLII’s effectiveness in predicting the abundance of both forest dwelling species, and species adaptable to low-integrity habitats, it could be utilized to establish thresh- old values for conservation efforts. For instance, FLII could help forecast the expansion of species that thrive in human- impacted environments, such as the brown hare and roe deer. However, it is important to note that incorporating for- est cover and canopy cover as additional explanatory vari- ables in the models, provides a more complete picture of the habitat preferences of some species, as evidenced by the substantial increases in their marginal R2. In particu- lar, when forest cover and canopy cover were included as additional variables in the full model, the effects of FLII became less pronounced for many species (Supplementary Table S4). This suggests that while FLII is a valuable sum- marizing indicator of ecological integrity, the direct physical attributes of the habitat, such as the amount of forest cover and canopy cover, play a more immediate and significant role in determining species abundance. Specifically in the study area, in fragmented forest landscapes, FLII may assign high scores to isolated forest patches naturally fragmented by undisturbed treeless peat- lands or waterbodies. However, such small forest patches might not be optimal habitats for many species requiring large areas of continuous forest cover, or due to poor con- nectivity with other suitable habitat patches. Similarly, peatlands that are often classified as high-integrity forest areas according to FLII have sparse tree cover but are not suitable habitats for species that prefer dense canopy cover. Dense cover generally provides shelter and concealment, Fig. 2   The error bar plot illustrates the FLII estimated coefficients values and the rela- tive 95% confidence intervals for each species obtained from the FLII model 69Oecologia (2024) 206:61–72 Ta bl e  2   P ar am et er e sti m at es fr om sp ec ie s- sp ec ifi c G LM M s e xa m in in g fa ct or s a ffe ct in g sn ow tr ac k ab un da nc e in w ild lif e tri an gl e ce ns us es (2 01 6– 20 20 ) Th e ta bl e in cl ud es c oe ffi ci en ts fo r F LI I, la tit ud e, fo re st co ve r, an d ca no py , t he ir 95 % c on fid en ce in te rv al s, an d si gn ifi ca nt v al ue s in b ol d. N es te d ra nd om e ffe ct v ar ia nc es a nd m od el d is pe rs io n pa ra m et er a re a ls o pr es en te d, a lo ng w ith m ar gi na l a nd c on di tio na l R 2 v al ue s i nd ic at in g va ria nc e ex pl ai ne d by th e fix ed e ffe ct s a nd th e to ta l m od el , r es pe ct iv el y Si gn ifi ca nt c oe ffi ci en ts , w ith p -v al ue s < 0 .0 5 an d 95 % c on fid en ce in te rv al s n ot in cl ud in g ze ro , w er e hi gh lig ht ed in b ol d Sp ec ie s In te rc ep t FL II Fo re st co ve r C an op y  co ve r La tit ud e D ist ric ts v ar i- an ce A ss oc ia - tio ns v ar i- an ce Tr ia ng le s v ar i- an ce D is pe rs io n pa ra m et er R 2 m R 2 c A m er ic an m in k − 2 .2 12 (− 2. 59 2 − 1 .8 32 ) 0. 10 3 (− 0 .0 44 0. 25 0) 0. 01 2 (− 0 .0 04 0. 02 8) 0. 04 2 (0 .0 07 0. 07 7) 0. 35 9 (0 .1 59 0. 55 8) 0. 10 2 0. 02 7 2. 40 2 0. 76 5 0. 10 0 0. 55 6 B ro w n ha re − 0 .2 68 (− 0 .6 65 0. 12 9) − 0 .2 50 (− 0. 39 1 − 0 .1 08 ) − 0 .0 73 (− 0. 08 5 − 0 .0 62 ) − 0 .0 06 (− 0 .0 41 0. 03 0) − 0 .5 73 (− 0. 82 2 − 0 .3 24 ) 0. 35 3 0. 32 0 2. 81 0 1. 47 5 0. 44 2 0. 88 0 Eu ra si an L yn x − 0 .7 27 (− 1. 17 1 − 0 .2 84 ) 0. 10 7 (− 0 .0 09 0. 22 4) 0. 02 1 (0 .0 08 0. 03 4) 0. 04 1 (0 .0 12 0. 07 0) − 0 .1 70 (− 0 .4 06 0. 06 6) 0. 46 6 0. 29 0 1. 61 8 0. 66 8 0. 04 3 0. 46 1 Eu ro pe an o tte r − 1 .8 89 (− 2. 14 3 − 1 .6 36 ) − 0 .0 42 (− 0 .1 72 0. 08 9) 0. 02 7 (0 .0 14 0. 04 0) − 0 .0 13 (− 0 .0 38 0. 01 2) − 0 .0 89 (− 0 .2 27 0. 05 0) 1. 10 E- 10 3. 88 E- 07 2. 47 5 1. 41 7 0. 04 4 0. 72 3 Fo re st re in de er − 3 .4 35 (− 5. 42 3 − 1 .4 48 ) 0. 90 7 (0 .2 80 1. 53 4) − 0 .0 16 (− 0 .0 68 0. 03 5) − 0 .0 05 (− 0 .1 57 0. 14 8) − 0 .1 45 (− 1 .2 13 0. 92 4) 3. 46 7 3. 76 6 1. 27 E- 07 0. 03 3 0. 00 2 0. 01 0 Le as t w ea se l − 0 .8 48 (− 1. 09 5 − 0 .6 00 ) − 0 .0 44 (− 0 .1 74 0. 08 5) 0. 00 0 (− 0 .0 12 0. 01 2) 0. 03 5 (0 .0 09 0. 06 0) 0. 04 2 (− 0 .0 93 0. 17 8) 2. 89 E- 13 1. 82 E- 07 2. 83 1 0. 67 0 0. 01 8 0. 49 2 M oo se 1. 73 4 (1 .5 23 1. 94 4) 0. 09 9 (0 .0 03 0. 19 5) 0. 06 8 (0 .0 58 0. 07 9) − 0 .0 29 (− 0. 05 0 − 0 .0 09 ) − 0 .4 15 (− 0. 55 1 − 0 .2 79 ) 0. 07 1 0. 11 0 2. 04 4 1. 37 0 0. 20 2 0. 74 3 M ou nt ai n ha re 3. 06 8 (2 .9 03 3. 23 2) 0. 05 3 (− 0 .0 29 0. 13 5) 0. 02 8 (0 .0 20 0. 03 6) 0. 03 6 (0 .0 18 0. 05 4) 0. 11 8 (0 .0 05 0. 23 1) 0. 04 5 0. 05 4 1. 72 4 2. 31 3 0. 12 4 0. 80 0 Pi ne m ar te n 0. 02 3 (− 0 .2 08 0. 25 5) 0. 12 1 (0 .0 23 0. 21 9) 0. 05 0 (0 .0 38 0. 06 2) − 0 .0 01 (− 0 .0 24 0. 02 3) − 0 .1 63 (− 0. 30 5 − 0 .0 21 ) 0. 06 6 0. 15 0 1. 70 4 1. 09 2 0. 13 9 0. 63 0 R ac co on d og − 1 .4 74 (− 1. 86 9 − 1 .0 79 ) 0. 09 5 (− 0 .0 30 0. 21 9) − 0 .0 19 (− 0. 03 1 − 0 .0 08 ) 0. 04 5 (0 .0 12 0. 07 7) − 0 .8 53 (− 1. 05 5 − 0 .6 50 ) 0. 18 1 0. 34 4 1. 23 6 0. 39 9 0. 22 6 0. 37 7 Re d fo x 1. 83 0 (1 .5 57 2. 10 3) − 0 .0 26 (− 0 .1 07 0. 05 5) − 0 .0 16 (− 0. 02 3 − 0 .0 08 ) − 0 .0 39 (− 0. 05 8 − 0 .0 20 ) − 0 .1 89 (− 0. 34 3 − 0 .0 35 ) 0. 18 9 0. 26 4 1. 46 4 3. 01 1 0. 08 3 0. 84 4 Re d sq ui rr el 0. 96 5 (0 .8 13 1. 11 7) − 0 .0 42 (− 0 .1 34 0. 05 0) − 0 .0 11 (− 0. 02 0 − 0 .0 03 ) 0. 02 7 (0 .0 09 0. 04 6) − 0 .1 61 (− 0. 26 2 − 0 .0 60 ) 1. 56 E- 08 0. 14 1 1. 78 5 1. 35 1 0. 09 7 0. 67 6 Ro e de er − 1 .6 69 (− 2. 35 0 − 0 .9 88 ) − 0 .1 14 (− 0 .2 86 0. 05 8) − 0 .0 66 (− 0. 08 1 − 0 .0 52 ) 0. 01 5 (− 0 .0 30 0. 05 9) − 0 .5 21 (− 0. 87 5 − 0 .1 66 ) 1. 21 9 0. 75 4 3. 71 3 0. 76 2 0. 25 7 0. 77 4 St oa t − 0 .4 68 (− 0. 69 0 − 0 .2 46 ) 0. 05 9 (− 0 .0 59 0. 17 8) − 0 .0 08 (− 0 .0 19 0. 00 3) 0. 04 8 (0 .0 26 0. 07 1) 0. 46 8 (0 .3 44 0. 59 2) 2. 42 E- 09 0. 02 4 1. 92 7 0. 68 0 0. 09 8 0. 45 6 W hi te -ta ile d de er − 0 .7 92 (− 1 .8 97 0. 31 2) − 0 .0 37 (− 0 .2 31 0. 15 7) 0. 02 1 (0 .0 04 0. 03 7) − 0 .0 27 (− 0 .0 83 0. 03 0) − 2 .3 21 (− 2. 90 3 − 1 .7 39 ) 3. 49 7 0. 86 8 3. 53 5 1. 13 2 0. 39 4 0. 91 0 W ol f − 4 .1 74 (− 5. 44 8 − 2 .8 99 ) 0. 63 5 (0 .2 58 1. 01 3) 0. 04 6 (0 .0 02 0. 09 0) − 0 .0 97 (− 0. 17 7 − 0 .0 18 ) − 0 .4 92 (− 1 .0 42 0. 05 7) 0. 28 9 3. 82 7 0. 43 0 0. 05 9 0. 01 0 0. 02 5 W ol ve rin e − 3 .4 09 (− 4. 22 5 − 2 .5 94 ) 0. 16 3 (− 0 .0 37 0. 36 2) 0. 03 9 (0 .0 07 0. 07 1) − 0 .0 11 (− 0 .0 56 0. 03 4) 0. 87 7 (0 .3 91 1. 36 2) 0. 83 6 1. 60 4 2. 22 3 0. 67 7 0. 24 5 0. 70 5 70 Oecologia (2024) 206:61–72 facilitating escape from predators (Thompson et al. 2003). For example, pine martens (Hargis and McCullough 1984) and weasels (Mougeot et al. 2020) use stands with dense cover to avoid avian predators. Canopy cover could also be an important environmental element for protection from humans, emphasizing the need for more nuanced habitat assessments. Conclusions Our study supports the hypothesis that the Forest Landscape Integrity Index (FLII) provides a solid basis in measuring aspects of forest ecological integrity degradation on the biotitic community. However, integration with other envi- ronmental variables provides a more complete and accurate understanding of ecological dynamics. Therefore, we recommend that future research focus on adapting these global indices to local contexts to achieve more accurate and meaningful assessments of forest health and biodiversity. Such adaptations will allow the develop- ment of more targeted conservation strategies that take into account regional forest variations and species-specific needs. Supplementary Information  The online version contains supplemen- tary material available at https://​doi.​org/​10.​1007/​s00442-​024-​05613-z. Acknowledgements  We would like to thank Societas pro Fauna et Flora Fennica, Lammin biologisen aseman Ympäristötutkimuk- sen Säätiö, University of Helsinki funds, Ilkka Hanski rahasto, and Riistasäätiö provided FM with financial support. Author contribution statement  Initial conceptualization of the study was by John Loehr. Subsequent conceptualization, methodology, inves- tigation, and writing by all authors. Material preparation, data collec- tion, and analysis were performed by Francesca Malcangi, Andreas Lindén, and John Loehr. Fig. 3   Error bars plot illustrates the estimated coefficients values and the relative 95% confidence intervals of FLII, forest cover and canopy cover for each species obtained from the full model https://doi.org/10.1007/s00442-024-05613-z 71Oecologia (2024) 206:61–72 Funding  Open Access funding provided by University of Helsinki (including Helsinki University Central Hospital). Societas pro Fauna et Flora Fennica,-, Francesca Malcangi, Lammin biologisen aseman Ympäristötutkimuksen Säätiö,-, Francesca Malcangi,University of Hel- sinki funds,Ilkka Hanski rahasto,-, Francesca Malcangi,Riistasäätiö,-, Francesca Malcangi. Data Availability  Data on wildlife triangles that support the results of this study are available from the National Resources Institute Finland (LUKE), but restrictions from the Finnish government apply to the availability of these data, which were used under license for this study and are therefore not publicly available. However, the data are avail- able upon reasonable request and with the permission of the National Resources Institute Finland (LUKE). The data on canopy cover and forest cover are available from https://​kartta.​luke.​fi/​opend​ata/​valin​ta-​ en.​html. The global raster on Forest Landscape Integrity Index (FLII) is available from https://​www.​fores​tinte​grity.​com/. Code availability  Code generated during the current study is available from the corresponding author on request. Declarations  Conflict of interest  Janne Sundell and John Loehr are handling editors of the journal. Ethics approval not applicable. Consent to participate not applicable. Consent for publication not applicable. Open Access  This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. References Bonesi L, Palazon S (2007) The American mink in Europe: status, impacts, and control. Biol Cons 134:470–483. https://​doi.​org/​ 10.​1016/j.​biocon.​2006.​09.​006 Brooks ME, Kristensen K, Benthem KJ van, others (2017) Modeling zero-inflated count data with glmmTMB Caravaggi A, Montgomery WI, Reid N (2015) Range expansion and comparative habitat use of insular, congeneric lagomorphs: invasive European hares Lepus europaeus and endemic Irish hares Lepus timidus hibernicus. Biol Invasions 17:687–698. https://​doi.​org/​10.​1007/​s10530-​014-​0759-1 Cardinale BJ, Srivastava DS, Duffy JEE et al (2006) Effects of bio- diversity on the functioning of trophic groups and ecosystems. Nature 443:989–992. https://​doi.​org/​10.​1038/​natur​e05202 Cardinale BJ, Duffy JE, Gonzalez A et al (2012) Biodiversity loss and its impact on humanity. Nature 486:59–67. https://​doi.​org/​ 10.​1038/​natur​e11148 Folke C, Carpenter S, Walker B et al (2004) Regime shifts, resil- ience, and biodiversity in ecosystem management. Annu Rev Ecol Evol Syst 35:557–581. https://​doi.​org/​10.​1146/​annur​ev.​ ecols​ys.​35.​021103.​105711 Gorini L, Linnell JDC, May R et al (2012) Habitat heterogeneity and mammalian predator–prey interactions. Mammal Rev 42:55–77. https://​doi.​org/​10.​1111/j.​1365-​2907.​2011.​00189.x Grantham HS, Duncan A, Evans TD et al (2020) Anthropogenic modification of forests means only 40% of remaining forests have high ecosystem integrity. Nat Commun 11:5978. https://​ doi.​org/​10.​1038/​s41467-​020-​19493-3 Güthlin D, Storch I, Küchenhoff H (2013) Landscape variables associated with relative abundance of generalist mesopreda- tors. Landscape Ecol 28:1687–1696. https://​doi.​org/​10.​1007/​ s10980-​013-​9911-z Hagen M, Kissling WD, Rasmussen C, others (2012) Biodiversity, Species Interactions and Ecological Networks in a Fragmented World. In: Advances in Ecological Research. Elsevier, pp 89–210 Hargis CD, McCullough DR (1984) Winter diet and habitat selection of marten in Yosemite national park. J Wildl Manag 48:140– 146. https://​doi.​org/​10.​2307/​38084​61 Hartig F, Lohse L (2022) DHARMa: Residual Diagnostics for Hier- archical (Multi-Level / Mixed) Regression Models Hyvärinen E, Aino A, Juslén K, et al (2019) Red List for Finnish Species. Ympäristöministeriö & Suomen ympäristökeskus Jahren T, Odden M, Linnell JDC, Panzacchi M (2020) The impact of human land use and landscape productivity on popula- tion dynamics of red fox in southeastern Norway. Mam- malian Research 65:503–516. https://​doi.​org/​10.​1007/​ s13364-​020-​00494-y Jia H, Luo P, Yang H et al (2023) Constructing an indices system for evaluating the ecological integrity of forests in western Sichuan, China based on structural equation modeling. Ecol Ind 146:109745. https://​doi.​org/​10.​1016/j.​ecoli​nd.​2022.​109745 Jokimäki J, Selonen V, Lehikoinen A, Kaisanlahti-Jokimäki M-L (2017) The role of urban habitats in the abundance of red squir- rels (Sciurus vulgaris L.) in Finland. Urban for Urban Greening 27:100–108. https://​doi.​org/​10.​1016/j.​ufug.​2017.​06.​021 Jonsson BG, Ekström M, Esseen P-A et al (2016) Dead wood avail- ability in managed Swedish forests–policy outcomes and impli- cations for biodiversity. For Ecol Manage 376:174–182. https://​ doi.​org/​10.​1016/j.​foreco.​2016.​06.​017 Kauhala K, Helle E (1995) Population ecology of the raccoon dog in Finland - a synthesis. Wildl Biol 1:3–9. https://​doi.​org/​10.​ 2981/​wlb.​1995.​004 Klemola T, Korpimäki E, Norrdahl K et al (1999) Mobility and habitat utilization of small mustelids in relation to cyclically fluctuating prey abundances. Ann Zool Fenn 36:75–82 Korhonen L, Korhonen K, Stenberg P et al (2007) Local models for forest canopy cover with beta regression. Silva Fennica 41:671– 685. https://​doi.​org/​10.​14214/​sf.​275 Levänen R, Pohjoismäki JLO, Kunnasranta M (2019a) Home Ranges of Semi-Urban Brown Hares (Lepus europaeus) and mountain hares (Lepus timidus) at Northern latitudes. Ann Zool Fenn 56:107–120. https://​doi.​org/​10.​5735/​086.​056.​0110 Levänen R, Pohjoismäki JLO, Kunnasranta M (2019b) Home Ranges of Semi-Urban Brown Hares (Lepus europaeus) and mountain hares (Lepus timidus) at Northern Latitudes. Anzf 56:107–120. https://​doi.​org/​10.​5735/​086.​056.​0110 Lindén, A. & Piha, M. 2024. Pienten ja keskikokoisten riistanisäk- käiden kannanmuutokset 1989–2024. Lumijälkilaskentojen tulosraportti. Luonnonvara- ja biotalouden tutkimus 40/2024. Luonnonvarakeskus. Helsinki. 20 s. Linnell JDC, Cretois B, Nilsen EB et al (2020) The challenges and opportunities of coexisting with wild ungulates in the human- dominated landscapes of Europe’s Anthropocene. Biol Cons 244:108500. https://​doi.​org/​10.​1016/j.​biocon.​2020.​108500 Magrini C, Manzo E, Zapponi L et  al (2009) Weasel (Mustela nivalis) spatial ranging behaviour and habitat selection in https://kartta.luke.fi/opendata/valinta-en.html https://kartta.luke.fi/opendata/valinta-en.html https://www.forestintegrity.com/ http://creativecommons.org/licenses/by/4.0/ https://doi.org/10.1016/j.biocon.2006.09.006 https://doi.org/10.1016/j.biocon.2006.09.006 https://doi.org/10.1007/s10530-014-0759-1 https://doi.org/10.1038/nature05202 https://doi.org/10.1038/nature11148 https://doi.org/10.1038/nature11148 https://doi.org/10.1146/annurev.ecolsys.35.021103.105711 https://doi.org/10.1146/annurev.ecolsys.35.021103.105711 https://doi.org/10.1111/j.1365-2907.2011.00189.x https://doi.org/10.1038/s41467-020-19493-3 https://doi.org/10.1038/s41467-020-19493-3 https://doi.org/10.1007/s10980-013-9911-z https://doi.org/10.1007/s10980-013-9911-z https://doi.org/10.2307/3808461 https://doi.org/10.1007/s13364-020-00494-y https://doi.org/10.1007/s13364-020-00494-y https://doi.org/10.1016/j.ecolind.2022.109745 https://doi.org/10.1016/j.ufug.2017.06.021 https://doi.org/10.1016/j.foreco.2016.06.017 https://doi.org/10.1016/j.foreco.2016.06.017 https://doi.org/10.2981/wlb.1995.004 https://doi.org/10.2981/wlb.1995.004 https://doi.org/10.14214/sf.275 https://doi.org/10.5735/086.056.0110 https://doi.org/10.5735/086.056.0110 https://doi.org/10.1016/j.biocon.2020.108500 72 Oecologia (2024) 206:61–72 agricultural landscape. Acta Theriol 54:137–146. https://​doi.​ org/​10.​1007/​BF031​93169 Mönkkönen M, Aakala T, Blattert C et al (2022) More wood but less biodiversity in forests in Finland: a historical evaluation. Finn- ish Soc for Sci 98:1–11. https://​doi.​org/​10.​14214/​msff.​120306 Mougeot F, Lambin X, Arroyo B, Luque-Larena J-J (2020) Body size and habitat use of the common weasel (Mustela nivalis vulgaris) in Mediterranean farmlands colonised by common voles (Microtus arvalis). Mamm Res 65:75–84. https://​doi.​org/​ 10.​1007/​s13364-​019-​00465-y Nakagawa S, Schielzeth H (2013) A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol 4:133–142. https://​doi.​org/​10.​1111/j.​2041-​ 210x.​2012.​00261.x Newman TB, Browner WS (1991) In defense of standardized regres- sion coefficients. Epidemiology 2(5):383–386. https://​doi.​org/​10.​ 1097/​00001​648-​19910​9000-​00014.​PMID1​742391 Oettel J, Lapin K (2021) Linking forest management and biodiver- sity indicators to strengthen sustainable forest management in Europe. Ecol Ind 122:107275. https://​doi.​org/​10.​1016/j.​ecoli​nd.​ 2020.​107275 Pan Z, He J, Liu D et al (2021) Ecosystem health assessment based on ecological integrity and ecosystem services demand in the middle reaches of the Yangtze river economic belt. China Sci Total Envi- ron 774:144837. https://​doi.​org/​10.​1016/j.​scito​tenv.​2020.​144837 Parrish JD, Braun DP, Unnasch RS (2003) Are we conserving what we say we are? Measuring ecological integrity within protected Areas. Bioscience 53:851–860. https://​doi.​org/​10.​1641/​0006-​ 3568(2003)​053[0851:​AWCWWS]​2.0.​CO;2 Pellikka J, Hannu R, Linden H (2005) Monitoring wildlife richness: Finnish applications based on wildlife triangle censuses. Ann Zool Fenn 42:123–134 Poutanen J, Fuller AK, Pusenius J et al (2023) Density-habitat relation- ships of white-tailed deer (Odocoileus virginianus) in Finland. Ecol Evol 13:e9711. https://​doi.​org/​10.​1002/​ece3.​9711 Rockström J, Steffen W, Noone K et al (2009) A safe operating space for humanity. Nature 461:472–475. https://​doi.​org/​10.​1038/​46147​ 2a Schielzeth H (2010) Simple means to improve the interpretability of regression coefficients. Methods Ecol Evol 1:103–113. https://​doi.​ org/​10.​1111/j.​2041-​210X.​2010.​00012.x Seidl R, Turner MG (2022) Post-disturbance reorganization of forest ecosystems in a changing world. Proc Natl Acad Sci 119:e2202190119. https://​doi.​org/​10.​1073/​pnas.​22021​90119 Semenchuk P, Plutzar C, Kastner T et al (2022) Relative effects of land conversion and land-use intensity on terrestrial verte- brate diversity. Nat Commun 13:615. https://​doi.​org/​10.​1038/​ s41467-​022-​28245-4 Storch D, Šímová I, Smyčka J et al (2022) Biodiversity dynamics in the Anthropocene: how human activities change equilibria of species richness. Ecography. https://​doi.​org/​10.​1111/​ecog.​05778 Tasser E, Unterthurner B, Agreiter A et al (2023) Long-term game species dynamic as indicator for changing landscape quality. Sci Total Environ 874:162375. https://​doi.​org/​10.​1016/j.​scito​tenv.​ 2023.​162375 Thompson ID, Baker JA, Ter-Mikaelian M (2003) A review of the long-term effects of post-harvest silviculture on vertebrate wild- life and predictive models, with an emphasis on boreal forests in Ontario, Canada. For Ecol Manage 177:441–469. https://​doi.​org/​ 10.​1016/​S0378-​1127(02)​00453-X Thulin C-G (2003) The distribution of mountain hares Lepus timidus in Europe: a challenge from brown hares L. europaeus? Mammal Rev 33:29–42. https://​doi.​org/​10.​1046/j.​1365-​2907.​2003.​00008.x Tinoco T, Carvalho JC, Panzacchi M et al (2011) Comparative use of forest habitats by roe deer and moose in a human-modified land- scape in southeastern Norway during winter. Ecol Res 26:781– 789. https://​doi.​org/​10.​1007/​s11284-​011-​0837-0 Tomppo E, Olsson H, Ståhl G et al (2008) Combining national forest inventory field plots and remote sensing data for forest databases. Remote Sens Environ 112:1982–1999. https://​doi.​org/​10.​1016/j.​ rse.​2007.​03.​032 Venier LA, Thompson ID, Fleming R et al (2014) Effects of natural resource development on the terrestrial biodiversity of Canadian boreal forests. Environ Rev 22:457–490. https://​doi.​org/​10.​1139/​ er-​2013-​0075 Vuorisalo T, Talvitie K, Kauhala K et al (2014) Urban red foxes (Vulpes vulpes L.) in Finland: a historical perspective. Landsc Urban Plan 124:109–117. https://​doi.​org/​10.​1016/j.​landu​rbplan.​ 2013.​12.​002 Wallén J, Statham M, Agren E et al (2018) Multiple recolonization routes towards the north: population history of the Fennoscandian red fox (Vulpes vulpes). Biol J Lin Soc 124:621–632. https://​doi.​ org/​10.​1093/​bioli​nnean/​bly060 Weigel B, Henne L, Martinez Rivera L (2002) Macroinvertebrate-based Index of biotic integrity for protection of streams in west-central Mexico. Am Benthol Soc 21:686–700. https://​doi.​org/​10.​2307/​ 14684​39 Zelený J, Mercado-Bettín D, Müller F (2021) Towards the evaluation of regional ecosystem integrity using NDVI, brightness temperature and surface heterogeneity. Sci Total Environ 796:148994. https://​ doi.​org/​10.​1016/j.​scito​tenv.​2021.​148994 Zhao C, Shao N, Yang S et al (2019) Integrated assessment of ecosys- tem health using multiple indicator species. Ecol Eng 130:157– 168. https://​doi.​org/​10.​1016/j.​ecole​ng.​2019.​02.​016 Zuur AF, Ieno EN, Elphick CS (2010) A protocol for data exploration to avoid common statistical problems. Methods Ecol Evol 1:3–14. https://​doi.​org/​10.​1111/j.​2041-​210X.​2009.​00001.x https://doi.org/10.1007/BF03193169 https://doi.org/10.1007/BF03193169 https://doi.org/10.14214/msff.120306 https://doi.org/10.1007/s13364-019-00465-y https://doi.org/10.1007/s13364-019-00465-y https://doi.org/10.1111/j.2041-210x.2012.00261.x https://doi.org/10.1111/j.2041-210x.2012.00261.x https://doi.org/10.1097/00001648-199109000-00014.PMID1742391 https://doi.org/10.1097/00001648-199109000-00014.PMID1742391 https://doi.org/10.1016/j.ecolind.2020.107275 https://doi.org/10.1016/j.ecolind.2020.107275 https://doi.org/10.1016/j.scitotenv.2020.144837 https://doi.org/10.1641/0006-3568(2003)053[0851:AWCWWS]2.0.CO;2 https://doi.org/10.1641/0006-3568(2003)053[0851:AWCWWS]2.0.CO;2 https://doi.org/10.1002/ece3.9711 https://doi.org/10.1038/461472a https://doi.org/10.1038/461472a https://doi.org/10.1111/j.2041-210X.2010.00012.x https://doi.org/10.1111/j.2041-210X.2010.00012.x https://doi.org/10.1073/pnas.2202190119 https://doi.org/10.1038/s41467-022-28245-4 https://doi.org/10.1038/s41467-022-28245-4 https://doi.org/10.1111/ecog.05778 https://doi.org/10.1016/j.scitotenv.2023.162375 https://doi.org/10.1016/j.scitotenv.2023.162375 https://doi.org/10.1016/S0378-1127(02)00453-X https://doi.org/10.1016/S0378-1127(02)00453-X https://doi.org/10.1046/j.1365-2907.2003.00008.x https://doi.org/10.1007/s11284-011-0837-0 https://doi.org/10.1016/j.rse.2007.03.032 https://doi.org/10.1016/j.rse.2007.03.032 https://doi.org/10.1139/er-2013-0075 https://doi.org/10.1139/er-2013-0075 https://doi.org/10.1016/j.landurbplan.2013.12.002 https://doi.org/10.1016/j.landurbplan.2013.12.002 https://doi.org/10.1093/biolinnean/bly060 https://doi.org/10.1093/biolinnean/bly060 https://doi.org/10.2307/1468439 https://doi.org/10.2307/1468439 https://doi.org/10.1016/j.scitotenv.2021.148994 https://doi.org/10.1016/j.scitotenv.2021.148994 https://doi.org/10.1016/j.ecoleng.2019.02.016 https://doi.org/10.1111/j.2041-210X.2009.00001.x Correlation between mammal track abundance and Forest Landscape Integrity Index validates actual forest ecological integrity Abstract Introduction Materials and methods Wildlife triangles Study species Study area Explanatory variables Forest Landscape Integrity Index (FLII) Canopy cover Forest cover Spatial analysis Statistical analysis Results Discussion Conclusions Acknowledgements References