Forests harbor over half of all terrestrial biodiversity and generate ecosystem services essential to the humankind [1
]. Nevertheless, the request for goods and services from forest systems has increased constantly [2
]. Consequently, forest management is multifaceted [3
] and it targets the supply of a wide range of ecosystem services (ES) that include biodiversity conservation [4
]. Forest biodiversity—the diversity of species within a forest ecosystem—plays a central role in ecosystem functioning and in sustaining the provision of multiple ecosystem goods and services; that is, the benefits humans obtain from ecosystems [6
In this context, forest ecosystem productivity and its multifunctionality are intricately linked with underlying biodiversity [12
]. However, biodiversity has been steadily decreasing worldwide, thus calling for the implementation of conservation policies to mitigate this trend [15
]. Forest managers are faced with the need to tackle biodiversity conservation in the management of planted forests. Nevertheless, changes in ecological conditions are expected to impact the functioning of forest ecosystems and affect the provision of all ecosystem services [16
]. An accurate estimate of biodiversity will assist forest actors to design plans that in the long term increase the adaptability of forests to future environmental changes [18
Forest management planning needs to define and distribute over time and space silvicultural practices in order to increase the supply of market goods and services while mitigating negative impacts of climate drivers and disturbance regimes on biodiversity [19
]. This may encompass the modification of thinning regimes, rotation lengths, and species composition as well of the structure of forest stands [20
]. Brunette et al. [23
] reported that the rise in tree species diversity is an appropriate option for preserving ecosystem functioning under climate change, while Seibold et al. [24
] showed that boosting the share of broadleaf species in the landscape can benefit forest biodiversity. Other authors have recommended the design of landscapes where specific areas are left uncut or retained to provide wildlife habitat (e.g., green-tree retention, GRT) [25
] as a relevant biodiversity-oriented management response to forest biodiversity decline in managed landscapes [29
]. More recently this approach has been popularized as variable retention (VR). It may encompass further the allocation of areas to multi-aged stands [31
], the increase of old-growth areas and its distribution to achieve greater continuity and structural complexity [29
]. Shea et al. [33
] showed how this strategy is effective in conserving biodiversity at landscape-level, especially when there are moderate levels of fragmentation [34
]. Augustynczik et al. [36
] and Ezquerro et al. [20
] demonstrated the application of this approach to address biodiversity conservation in landscape-level management. Notwithstanding, practical applications of these adaptation practices are still scarce and/or recent [37
], hampering the development of reliable information to assist forest management planning and stakeholders’ decision-making processes.
Nevertheless, the definition of biodiversity indicators is influential to explicit conservation objectives in forest management planning and to help assess the impact of alternative landscape mosaics on biodiversity. The literature reports, since the late eighties, the use of a wide range of indicators [20
]. For example, some authors report the use of the amount of deadwood [39
]. Kouki et al. [34
] and Augustynczik and Yousefpour [38
] highlighted the importance of increasing the amount of deadwood in European forests to improve the habitat of threatened species. In general, the selection of biodiversity indicators for use in management planning must consider its practicality, cost-effectiveness and environmentally meaningfulness [44
]. Moreover, the selection must consider the availability of models for projecting the indicator value over the management planning temporal horizon.
In this context, indicators that may be easily measured and for which projection models are available appear to be the most practical. The literature reports the need to identify indicators that reflect tree species composition and vertical structural diversity (e.g., diameter heterogeneity, tree height, basal area, stand age, and canopy cover), including the shrub layers and those that are environmentally significant and suitable for different types of managed forests [39
]. Understory vegetation comprises one of the most significant elements of biodiversity within plantations and is often the sole best predictor of animal diversity (e.g., [53
Effectively addressing biodiversity concerns in forest management planning requires further the development of methods that may provide information about the impacts of alternative landscape mosaics on the values of biodiversity indicator, across a wide diversity of forest types and biogeographical–socio-economic conditions and other relevant forest ES. This is influential to compare planning strategies and to propose efficient solutions. The literature reports several alternatives and silvicultural options to integrate biodiversity and wood production in forest management [26
]. The literature reports several methods to optimize wildlife habitat and biodiversity conservation management. Thompson et al. [55
] pioneered the use of linear programming (LP) to integrate wildlife and timber objectives while Hof and Joyce [56
] and Hof et al. [57
] built mixed-integer and integer programming models to facilitate this integration. Other authors reported the use of heuristics to address wildlife habitat concerns [58
]. Mathematical programming approaches have been developed further to address reserve selection problems [59
]. More recently, Marto et al. [62
] used a combination of Pareto frontier approaches [63
] and multiple attribute techniques [64
] to address biodiversity objectives in a multiple-objective decision-making framework. The reader is referred to Ezquerro et al. [65
] for a comprehensive review of methods to integrate biodiversity in forest management planning. However, a comprehensive understanding of how silvicultural treatments affect biodiversity indicators and how these effects interrelate with climate change and other ES outcomes in managed forests is still limited. Moreover, despite the large body of literature dealing with the ecological aspects of biodiversity-oriented models, the estimation of practical and quantitative indicators for application in the framework of landscape-level management planning Mediterranean managed forests remain unexplored.
We address this gap by developing an approach that integrates a biodiversity-oriented management indicator that reflect the interaction between tree species composition, stand age, and understory coverage, under scenarios of divergent climate conditions, and linear programming-based (LP) optimization techniques. We use indicators addressing biodiversity at the stand and forest management unit scales. Forest management at the landscape scale addressing, for example, issues related to habitat fragmentation and connectivity are not addressed by the present study. After describing the approach, findings are discussed for an application to a large-scale problem in Northwestern Portugal (Vale do Sousa) encompassing 14,765 ha classified into 1373 stands, for a 90-year planning horizon across two local-climate scenarios. We examine the potential of the approach to help assess the quantitative impact of both stand and landscape-level forest management models on biodiversity. Further, we analyze the insights provided by the approach on how climate change may impact plans that target the optimization of landscape-level biodiversity values.
In this research, a landscape-level LP-RMC was extended to integrate a biodiversity-oriented forest management indicator and two local climate scenarios, i.e., business-as-usual (BAU) and REF (with a RCP 8.5) to measure the quantitative impact of forest management models on the potential biodiversity landscape provision. The proposed methodology explores the instrumental interaction between forest management models and biodiversity levels at the stand and landscape scales in the face of climate change, using tree species composition and understory vegetation as biodiversity proxies. Likewise, the application of a simulation-optimization framework to guide forest actors in considering optimal practices to safeguard future biodiversity; seeking to benefit native forest species, while other conservation commitments are reached in long-term forest management. For that purpose, several forest management models (four current and four alternative FMMs), through different periodicity of fuel treatments and effects of local-climate change on biodiversity indicators and related proxies were tested.
In the last thirty years, the integration of biodiversity into forest management has been classified and evaluated in accordance with different attributes such as model components, forest management elements, or biodiversity indicators. However, landscapes are rarely managed to provide levels of biodiversity, in part because the indicators remain complex, and new key-related variables need to be considered to drive management decisions [14
]. Given the long rotations used in several forest systems, it is essential to have a set of indicators that encompass such long-term time scales and which values can be assessed periodically, after a particular management intervention, but also at the end of the planning horizon. In that sense, the biodiversity indicator used in the present study aims to fulfil such purpose, as it is a function of species composition (e.g., chestnut, cork oak, eucalypt, pine, pure pedunculate oak, and riparian trees), the stand age, and understory biomass accumulation, and there is a linear trade-off between area of each FMM and the biodiversity indicator. Several researchers identified that the understory vegetation is a practical indicator and a major driver of many forest processes such as forest productivity, litter decomposition and light interception [47
]. In fact, our research CSA area provides relevant outputs addressing the understory shrub conditions, thus allowing differentiation between forest characterization/attributes of higher or lower importance of biodiversity. The newly criteria used to classify the biodiversity proxies in Vale de Sousa have proved to be a practical indicator for assessing forest biodiversity provisions under forest management plans in Europe [94
]. It is important to remark that 21 studies reviewed by Thompson et al. [95
] reported that in 76% of the cases, vertebrate and arthropod diversity are positively correlated with plant diversity (measured as tree species and understory richness), establishing a direct relationship between increased biodiversity and forest productivity. Further, forest stand structural variables derived from inventories can help enhance management plans to place European forests on the path to an uncertain future [94
As far as we are concerned, indicators such as the structural heterogeneity of both tree species and understory may reflect better the biodiversity value than the amount of deadwood or the number of large trees [96
]. In Portugal, the richness and diversity of tree species are poor because most are single or two species stands. Even conservation valued cork oak woodlands are frequently mono—(cork oak only) or dual-specific (e.g., cork and holm oak) [46
]. whilst in central Europe or Northern forested landscapes, we are dealing with a few different tree species, here biodiversity provision is not in the diversity of tree species but on the shrubby understory. Thus, shrub species composition considered in forest management planning is particularly important in Portugal since Mediterranean basin is one of the "hot spots" of biodiversity worldwide [97
]. In this regard, increasing the availability of specific forest structures (e.g., dead wood and large trees) are not a strategy for achieving biodiversity goal across Portugal. Indeed, what make Portugal ecosystems interesting for biodiversity conservation is the mixture of shrub and grassland species underneath tree cover. Consequently, the relevance of pine, eucalypt, chestnut, and oak stands for conservation in these forested areas relies on the diversity and amount of shrub cover as an important habitat type for wildlife. This may impact the assessment of trade-offs between biodiversity and other ES. Nevertheless, these differences in outcome highlight the potential of this approach to be actively applied within CSAs to dictate the future biodiversity of forest production areas. Such an assumption, however, needs to be assessed in the future with increased wildfire hazard in shrub encroached stands, which may ultimately lead to loss of conservation value.
Our results are placed in the context of forest management and practically revealed no significant reductions in outcomes of biodiversity indicators with the rise in wood production. Further, we aimed to sustain the supply of timber at levels suitable for each climate scenario. Thereby, well-balanced results arose when diversity was actively promoted with new FMMs as part of a sustainable management concept. Liang et al. [13
] stated that there is a predominant positive biodiversity-productivity relationship in global forests. Regarding this component, our results suggested that there was not significant trade-off between biodiversity and sustainable wood production. These results agree with a study for then forest landscapes across Europe [96
] were they stated that almost no reduction in biodiversity indicators are associated with an increase in sustainable wood production. Indeed, in the recent years, potential synergies have been discussed in the connection between a landscape biodiversity conservation and wood production [21
The cFMMs have typical characteristics associated with lower biodiversity scores than the aFMMs in the same period. In fact, the cFMMs 1, 2, and 4 with eucalypt species had a reduced biodiversity score over the entire 90-year horizon. This is also in agreement with the finding in the work by Proença et al. [100
] and Goded et al. [101
] from Spain, where biodiversity in eucalyptus stands is compared to native stands, and the results showed the lowest values was precisely for eucalyptus. Indeed, diversity and composition of understory vegetation tended to be higher in native forests and shrublands, and lowest in eucalypt plantations in several stages (young, intermediate and mature) [102
]. Eucalyptus plantations display extremely low biological and aesthetic diversity; although, they seem to be admirably adapted to VA_CSA conditions. It is well associate with climax species and allow the development of diverse shrub layer. The presence of eucalypt may be instrumental to landscape heterogeneity and generate financial resources to support set-aside conservation areas, thus contributing to landscape-level biodiversity. Basically, the management activities of the cFMMs landscape-solution for our CSA, results in less biodiversity score due to essentially two factors: shorter rotations from the introduced eucalyptus species resulting in a younger forests and more compressed age class structure, and more coniferous trees over lager areas. Nevertheless, the combined effect of market, technical and human capacities play a major role in keeping the current almost total dominance of the production of the eucalyptus and maritime pine systems [88
]. Still, due to the recent changes on the National Forest Policy, there are now stronger planting restrictions on eucalyptus, and thus the forest owners are looking for native alternative species for timber production, resin and cork production, and wildfire risk reduction.
Alternative FMMs are encouraged, especially in addressing climate change and potential future conditions in forest productivity. The optimal-solutions depict the dominance of native species in the new FMMs for our case study area. Eucalyptus stands exchanges by maritime pine and oak forests areas contributed to the biodiversity increase along the planning horizon. Maritime pine stands plantations (aFMM5) retain a significant recreation interest, as they are widely used for summer picnics. In the past, pedunculated oaks dominated the landscapes in the northern Portugal. aFMM6 and aFMM7 relates to native species of Fagaceae where, from a biodiversity perspective, there is a larger proportion of broadleaves. aFMM3 has a reasonable biodiversity score values, by the richness associated vegetation of Castanea sativa species. Although this species is associated to high values of biodiversity, the chosen prescriptions have shorter years of rotation than the Q. robur and Q. suber species (aFMM6 and aFMM7, respectively). aFMM8 (riparian system) have the highest biodiversity score values over the 90-years of planning period. aFMM8 will contribute to biodiversity by providing habitats for specific flora and fauna, both in the tree itself and in the flooded root system.
In this respect, the aFMMs demonstrate higher values than de cFMMs in terms of species richness, shrubby understory diversity, and stand structure. This is also associated with the periodicity of fuel treatments on the selected prescriptions of aFMM and cFMM and the corresponding impact on the accumulation of fuel understory. The set of aFMM fits properly with the general goal to manage forests for increased resistance and resilience. Horl et al. [103
] reviewed the performance of adaptive forest management strategies and identified changes in species composition as one of the most resilience-oriented strategies recommended among adaptation scenarios in relation to multiple goods and services, including wood production and biodiversity. In fact, it is a major concept for facing an uncertain future in forestry [104
The local-climate scenarios applied did not cause necessarily different outputs at the case study level. The classification of biodiversity scores suggests that the increase of temperature and precipitation from the reference local-climate scenario, compared with the BAU scenario, has a limited impact on the provision of the overall biodiversity. While the frame scenarios did not make a big difference, the outcomes of the silvicultural treatment showed interesting patterns. Previous research in Vale de Sousa with a focus on simultaneous maximization of multiple objectives has shown that biodiversity conservation values are lower (average values 1.52) when only current FMMs are assigned to all suitable management units, mainly explained by the proportion of eucalyptus [62
]. Our optimal forest management solution accomplishes an increase of 2.52 for maximizing biodiversity concerns with the share of native species as alternative FMMs across the current landscape (Biodiversity score = 3.82), and a slight increase of 1.78 and 1.72, when it reaches forest actor interests in wood provisioning and wildfires reduction, respectively.
Some limitations might be underlined. For example, the understory biodiversity relates mostly with shrub biomass, assuming that shrub cover is essential for wildlife, but do not reflect plant diversity. This can be improved in future assessments. There is also the degree of uncertainty that is linked to the results—mainly due to the use of empirical growth and yield models, disregarding the inclusion of physiological data, but also due to the derived assumptions of expected climate change effects on forest growth under the REF scenario. Although our database did not allow to simulate the effects of large tree removal on wood production and biodiversity it is know that large trees are frequently used for bird nesting namely threatened bird raptors such as the red kite (Milvus milvus
) which in Iberian Peninsula may nest is pine plantations similar to those in our study area [105
The strong relationship between the ecological and economic functions of forest services, under the impact of environmental factors, is crucial for the stakeholders decision-making processes. In this regard, future work directions screen a focus on the economic representation of ES values (e.g., net present value and soil expectation value) to allow a complete interpretation and efficient use of the described LP-RCM.