(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .
Long-term effect of forest harvesting on boreal species assemblages under climate change [1]
['Ilhem Bouderbala', 'Département De Physique', 'De Génie Physique Et D Optique', 'Université Laval', 'Québec', 'Qc', 'Département De Biologie', 'Centre Interdisciplinaire En Modélisation Mathématique De L Université Laval', 'Guillemette Labadie', 'Jean-Michel Béland']
Date: 2023-04
Logging is the main human disturbance impacting biodiversity in forest ecosystems. However, the impact of forest harvesting on biodiversity is modulated by abiotic conditions through complex relationships that remain poorly documented. Therefore, the interplay between forest management and climate change can no longer be ignored. Our aim was to study the expected long-term variations in the assemblage of bird and beetle communities following modifications in forest management under different climate change scenarios. We developed species distribution models to predict the occurrence of 88 species of birds and beetles in eastern Canadian boreal forests over the next century. We simulated three climate scenarios (baseline, RCP4.5 and RCP8.5) under which we varied the level of harvesting. We also analyzed the regional assemblage dissimilarity by decomposing it into balanced variations in species occupancy and occupancy gradient. We predict that forest harvesting will alter the diversity by increasing assemblage dissimilarity under all the studied climate scenarios, mainly due to species turnover. Species turnover intensity was greater for ground-dwelling beetles, probably because they have lower dispersal capacity than flying beetles or birds. A good dispersal capacity allows species to travel more easily between ecosystems across the landscape when they search for suitable habitats after a disturbance. Regionally, an overall increase in the probability of occupancy is projected for bird species, whereas a decrease is predicted for beetles, a variation that could reflect differences in ecological traits between taxa. Our results further predict a decrease in the number of species that increase their occupancy after harvest under the most severe climatic scenario for both taxa. We anticipate that under severe climate change, increasing forest disturbance will be detrimental to beetles associated with old forests but also with young forests after disturbances.
Despite the number of studies made to better understand the cumulative effects of the different types of disturbances on biodiversity [ 1 ], there are considerable uncertainties over the magnitude of changes in future animal communities [ 27 ]. The cumulative effect of anthropogenic disturbances could induce complex effects on biodiversity through the adjustment in species’ ecological traits, which may have cascading repercussions from individual changes to community-level responses [ 6 ]. However, most literature evoking the modelling of biodiversity changes considered the assumption of no interactions among the various drivers [ 1 ]. In this work, we investigate how the interaction between forest harvesting and climate change would impact biodiversity. Specifically, we examine future implications of forest harvesting levels on the variation of bird and beetle assemblages in Québec’s boreal forest under different climate change scenarios. We aim first to quantify the extent of assemblages’ alteration by increasing harvest levels under different climate change scenarios and second to determine the main drivers. Moreover, we study how species associated with different types of habitat classes will be affected by the combined effects of forest harvesting and wildfire.
Although modelling biodiversity changes is becoming essential to predict the effects of human activities on the global environment, many challenges remain because there are many related issues that need to be integrated [ 24 ]. There are urgent needs to identify the most appropriate forest harvesting scenarios to achieve management objectives, such as species recovery [ 25 ] or habitat restoration [ 26 ]. This would help identify more targeted conservation actions under climate change such as vegetation restoration.
Climate change is the second most important driver of change in biodiversity after land use, for the terrestrial ecosystems [ 1 , 13 ]. Climate change may impact forest landscapes indirectly by triggering natural disturbances such as wildfire, which strongly alter forest age structure and composition [ 14 – 16 ], and to which are associated a unique biodiversity. Climate change may also directly impact tree growth and alter competitive abilities of boreal tree species [ 16 ]. These alterations are predicted to be detrimental to biodiversity especially those associated with older mixedwood and coniferous forests [ 17 , 18 ]. It may also impact species directly by changing their occurrence or abundance, when climatic conditions go out of their tolerance limits, due to modifications in temperature and precipitation conditions [ 19 , 20 ], which might reshape community composition by species turnover [ 20 ]. For example, the displacement of species to higher elevations is one of the predicted consequences of climate change to overcome the potential increase in temperature [ 21 ]. However, climate change could have implications on the abundance of species with low dispersal capacity that are unable to follow the climate conditions to which they are currently adapted [ 21 , 22 ] and could lead to an increase in extinction risks [ 23 ].
Biodiversity is facing an unprecedented rate of change due to human activities [ 1 , 2 ]. For terrestrial ecosystems, land-use changes through forest harvesting, agriculture or urbanization are considered the main causes of biodiversity losses and reductions in ecosystem services [ 1 , 3 – 5 ]. Human exploitation of natural resources can induce animal and plant mortality, force species displacement, increase biotic interactions, modify their life-history traits, change their morphology and physiology, and even their polymorphic forms [ 6 ]. Moreover, the exploitation of natural habitats by humans results in habitat losses and biotope fragmentation, which constrains the size of animal and plant populations [ 7 ]. For instance, by returning stands to an early successional stage, forest harvesting homogenizes forest landscape, which results in an overrepresentation of early-succession forests at the expense of the older states [ 8 , 9 ]. Forest harvesting targets old-stands, which are economically more profitable but these stands are not homogeneous as they present structural diversity and can be distinguished into different groups [ 10 ]. Moreover, these stands are the habitat of a biodiversity indicator species, the Black-backed Woodpecker Picoides arcticus), associated with old-growth attributes. Thus, forest harvesting may threaten vertebrate species associated with old-growth forest attributes [ 11 ] but also organisms (insects, fungi, non-vascular plants) that have low-dispersal capacity and for show extinction risk increases rapidly when ecological continuity is broken [ 12 ].
Winning conditional to habitat or to relation to disturbance probability was calculated as (2) where is the indicator function and H i is the associated habitat for bird species i (H ∈ {G, EMSF, LSF}) or the relation to disturbance for beetle species (H ∈ {Fire, Harvest, Fire − Harvest, None}). In addition, we used the regional Bray-Curtis dissimilarity index (RBC Sim,Ref ) to measure the degree of dissimilarity in assemblage composition between the compared scenarios [ 55 , 56 ]. The RBC Sim,Ref between the scenarios Sim and Ref was defined as (3) We split the Bray-Curtis dissimilarity index into two additive components: where BC_grad Sim,Ref is the occupancy gradient and BC_bal Sim,Ref is the species balanced variation in occupancy. The latter was defined as where To quantify the main source of assemblage dissimilarity, we performed a decomposition of the regional Bray-Curtis dissimilarity, based on the probability of occurrence, into balanced variation in species occupancy and occupancy gradient, which are the generalizations of species turnover and the nestedness respectively in the context of continuous output [ 56 ] and occurrence probabilities in our case [ 20 ]. This decomposition of the dissimilarity measure computed based on scenarios helped us to characterize the mechanisms that drive variation in species composition. Hereafter the terms species turnover and nestedness will be used as equivalents of balanced variation in species occupancy and occupancy gradient respectively. We then computed the ratio (4) The case BC_ratio Sim,Ref > 0.5 then indicated that the community change was mostly caused by nestedness, whereas BC_ratio Sim,Ref < 0.5 indicated the dominance of species turnover in the assemblage dissimilarity. We used the package ‘betapart’ [ 57 ] for the assemblage analysis.
We calculated the percentage change in the regional occupancy probability (ROP) between the reference (Ref) and the simulated (Sim) scenarios [ 4 ] as follows: (1) where denotes the average of the ROP over all the species under the scenario s and is given by with and where N species , N pixel and P s,i,j respectively represent the number of species, the number of pixels in the study area, and the probability of occurrence of the species i for the scenario s at the cell j.
We suggested three full potential models, the first one contained only linear forms of the predictor variables. However, the second and third full models included also quadratic transformations of the covariates (see S2 Table ). The best models were selected for each full model based on a 10-fold cross-validation procedure by minimizing the AIC criterion (package ‘MuMIn’; [ 51 ]). For each species, we kept only the models with AUC ≥ 0.7 [ 52 , 53 ]. In addition, we analyzed the assemblage structure based on continuous occurrence probabilities instead of binarizing the outputs [ 46 , 54 ].
We developed species distribution models (SDMs) [ 45 – 47 ] to estimate single-species occurrence probability based on extensive field surveys of 88 species (39 flying beetles, 15 ground-dwelling beetles and 34 bird species (see S1 File )). The methodology here is described in details in [ 20 ] as Habitat-Only-Based Models (HOBMs). We used generalized linear mixed models to predict species occurrence probability (package ‘lme4’; [ 48 ]). A random intercept was added to include any differences between sampling years based on three full potential models deferring only in the fixed effects of the models. We partitioned the modelling strategy into several steps starting from data pre-processing to the species occurrence probability projections. SDMs were estimated only for species recorded at least at 1% and 5% of sampling sites for birds and beetles, respectively. We used land cover variables extracted from LANDIS forest landscape simulations for the simulated scenarios (see S1 Table for predictor variables description). We also standardized the predictor variables just before any model calibration to facilitate the convergence of the models [ 49 ]. We removed highly correlated variables (|r| > 0.60, where r represented pairwise Pearson correlation coefficient) and kept the 5 most important potential predictor variables according to the Akaike information criterion (AIC) calculated on a univariate generalized linear model (GLM) with linear and quadratic terms for each variable [ 20 , 50 ].
A literature review was done on every beetle taxon to characterize two important functional traits: habitat preference and dispersal capacity. First, we identified species that respond positively to forest harvesting or burned forests, the remainder being considered as species associated with undisturbed natural forests. Then, in order to discriminate beetle taxa that mainly disperse by flight from those that mainly walk on the forest floor, we calculated abundance ratios from the data collected. For each beetle taxon, we pooled the total abundance (from 2004 to 2018) per flight interception trap and divided by the total abundance per pitfall trap. This ratio provides an index of beetle proficiency to fly or to walk on the ground. As the ratio provides continuous estimates, we kept only beetles with a ratio higher than 2 as flying beetles and beetles with a ratio lower than 3 as ground-dwelling beetles. Beetle species with ratios between 2 and 3 were removed from the study.
We considered three land cover classes to calculate species winning conditional probability (WCP), i.e., an increase in the probability of species occupancy from uncut to post-harvest in 2100, conditional to species habitat (see Eq 2 ). We calculated WCP for (1) generalist bird species; (2) species associated with early-to-mid succession forest (EMSF), which included the following land covers: wetland, deciduous, mixed-wood (young to mid-age), and coniferous habitat (young to mid-age); and (3) species associated with late succession forest (LSF), which included the following land covers: mixed-wood (mature to old), deciduous (mature to old) and coniferous habitat (mature to old). Stand age was classified as follows: young (stand age < 30 years), mid-age (stand aged between 30 and 60 years), mature (stand aged between 60 and 80 years), and old (stand age greater than 80 years). This classification was used only for birds because information regarding beetles’ associated land cover was unavailable at the moment of the study. Bird-associated land cover has been based on habitat associations from [ 43 , 44 ].
Land cover associated with sampling sites was determined from the Canadian National Forest Inventory referenced in the year 2001 [ 40 ], updated yearly for subsequent fires and cuts. The landscape was split into 11 land cover types. Specifically, we considered land covers with stand age greater than 50 years, i.e., closed-canopy conifer forest, open-canopy mature conifer forest, mixed forest, and open area (see Table 2 ), together with six disturbance classes, i.e, fire and forest harvesting subdivided by age classes: 0–10, 11–20 and 21–50 years [ 29 ]. The category of the other types of habitat was not included in the models to avoid collinearity. In the same region of study and forest types, [ 32 ] showed that ground-dwelling and flying beetles were influenced by landscape variables within radius around sample plots of 400 and 800 m respectively, while [ 4 ] used a 1 km radius for birds. Therefore, we used for ground-dwelling beetles 20 pixels (231 m resolution) centered on the focal pixel (i.e., 21 pixels in total) to calculate the frequencies of the land cover classes with a radius of 2.5 pixels (∼0.58 km). For flying beetles and birds, we used 37 pixels with a radius ∼0.81 km to calculate land cover frequencies. We also considered the distance to the nearest burned areas as well as the stand age as potential predictor variables.
We simulated forest succession across landscapes using LANDIS-II [ 35 ], a forest landscape model that captures forest succession across landscapes as an emergent property of both stand and landscape-scale processes [ 36 ]. Specifically, we used the LANDIS-II Biomass Succession extension v5.0, Base Fire v4.0, BDA v4.0, together with the Biomass Harvest v5.0 [ 35 ] to simulate forest succession, fire [ 37 ] and spruce budworm outbreak disturbances as well as harvesting in each 250-m cell at a 10-year time step. Outbreaks of SBW were simulated using the LANDIS-II Biological Disturbance Agent (BDA) extension v3.0 [ 38 , 39 ]. The Biomass Succession extension simulates modifications in cohort aboveground biomass (AGB) over time by taking into consideration cohort age, life-history traits, and land type responses of individual tree species. A random forest model was used to predict crown closure covariate by using Canadian National Forest Inventory (NFI) forest attribute maps [ 40 ]. These maps are a k-Nearest Neighbours interpolation of the NFI photoplot data acquired in 2001 and are depicting over 130 forest attributes including species-specific biomass, stand age and crown closure at a 250-m resolution (see [ 40 ]). We, therefore, build a random forest model predicting cell-level crown closure in NFI products from NFI species-specific biomass as well as stand age. This model had a very high goodness of fit (R 2 = 0.86). The model was then applied on LANDIS-II outputs to predict crown closure all along the simulation, for each cell by using simulated species-specific biomass and stand age [ 41 ]. Using species group and predicted crown closure, we created five land cover classes from the Earth Observation for Sustainable Development of Forests (EOSD) Land Cover Classification Legend ([ 42 ]; see Table 2 ). See [ 41 ] for more details regarding the spatially-explicit forest simulation model with LANDIS-II.
Three harvest levels were considered: no harvest activity (NoHarvest), 8% of the areas harvested over 10 years (Harvest)—the typical harvest rate for the study area—and half this level (i.e., 4%, Harvest 0.5 ). The harvest levels were combined with three climate scenarios: baseline level, radiative forcing associated with the Representative Concentration Pathway scenario (RCP) 4.5, which involved a 3°C increase in mean temperature from 2000 to 2100; and RCP 8.5 associated with a 7.5°C increase over the same period. We compared the response of species occurrence and occupancy between different harvest intensities under three climate scenarios, which explains why we had three reference scenarios ( Table 1 ). Also, we took into consideration in our simulations the northern limit of territorial forest attributes of harvest activities. Beyond this limit, no forest management was authorized by the Ministry of Forests, Wildlife, and Parks [ 34 ].
We used presence-absence data collected between 2004 and 2018 to model the species distribution of two taxa: birds and beetles. For birds, we used the data from the Second Atlas of Breeding Birds of Québec [ 30 ], based on species occurrence detected using unlimited distance 5-minute point counts [ 31 ] collected during the breeding season, from late May to mid-July. For beetles, we merged different databases collected in 2004, 2005, 2007, and 2011 [ 4 , 32 , 33 ]. We added 54 sites sampled in 2018 in the northern part of the territory, along the northeast principal road going to Labrador. Insect sampling involved one multidirectional flight-interception trap per site to capture the flying beetles, and four meshed pitfall traps per site to sample beetles that move on the soil surface [ 4 , 32 ]. Data from the sites heavily impacted by the SBW outbreak [ 34 ] were removed to focus only on the effect of the wildfires. We used for this purpose a cumulative index of defoliation severity [ 29 ] (we removed 4% and 7% of sampling sites for beetle and bird databases, respectively).
The study area was located in the Côte-Nord region of Québec, Canada (48°N to 53°N, 65°W to 71°W) within an area of 114,118 km 2 ( Fig 1 ). This region is part of the eastern spruce-moss subdomain and is dominated by black spruce (Picea mariana [Mill.] BSP) with balsam fir (Abies balsamea [L.] Miller). Forest harvesting had been the main source of anthropogenic disturbances since the late 1990s [ 28 ]. Moreover, This area is mainly affected by two major natural disturbances: spruce budworm outbreak (SBW) (Choristoneura fumiferana) and frequent wildfires [ 9 , 29 ].
(a) Regional Bray-Curtis dissimilarity ratio (BC_ratio) with the percentage of change in the regional occupancy probability ( ). At each climate change scenario (Baseline, RCP 4.5, RCP 8.5), we computed the change in the ROP ( Eq 1 ) and Bray-Curtis ratio ( Eq 4 ) between Harvest 0.5 and Harvest vs. NoHarvest in 2100. (b) Percentage of change in the regional occupancy probability between Harvest and Noharvest under the baseline and RCP 8.5 for bird species. Abbreviations : late succession forest (LSF), early-to-mid succession forest (EMSF), generalist species (G). (c-d) Percentage of change in the regional occupancy probability between Harvest and Noharvest under the baseline and RCP 8.5 for flying and ground-dwelling beetles species with relation to a disturbance.
We observed a change in the regional occupancy between the scenarios of forest harvesting (Harvest 0.5 and Harvest vs. NoHarvest) under all three climate change scenarios. More precisely, we observed an increase in regional occupancy in the bird community from nonharvest to harvest (passage from 13% to 15% in the regional occupancy from nonharvest to harvest scenarios under RCP 4.5). The opposite behaviour was observed in the beetle community (for instance a decrease in regional occupancy from 31% to 27% was observed for flying beetles by comparing the same scenarios). These changes in the overall occupancy of birds, and the decrease for beetles from NoHarvest to Harvest cases was visible over 114,118 km 2 of the study area ( Fig 4 ) and also regionally ( Fig 5a ). Post-harvesting changes in species assemblages were mainly caused by species turnover (BC_ratio Sim,Ref < 0.5) for the two taxa with a strongest turnover for ground-dwelling than flying beetles. Under each climate scenario, the occupancy of bird species associated with young habitats together with generalist species increased significantly contrary to species associated with mature and old forests, when we compared harvested to unharvested scenarios ( Fig 5b ). Under baseline and RCP 4.5, 50% of bird species associated with late succession forest increased in occupancy from no harvested to harvested scenarios ( Fig 6a ), whereas 82% of species associated with early to mid succession forest increased in occupancy (see S1 Fig for the variation at the species scale). By comparing RCP 8.5 to RCP 4.5, the proportion of winner species (from non-harvested to harvested landscapes) that were associated with early to mid-succession forests decreased from 82% to 73%. Under the same conditions, species associated with late succession forests increased from 50% to 63% ( Fig 6a ). When we compared non-harvested to harvested scenarios for flying beetles, species that were associated with burned stands had the weakest increase in occupancy compared to species associated with harvested stands or closed forests ( Fig 5c ). The percentage of winning flying beetles associated with closed forests decreased from 53% to 47% from baseline to RCP 8.5 under a forest harvest scenario ( Fig 6b ). Winning species of flying beetles associated with harvested stands even decreased from 42% to 33% under the same conditions. Among flying beetles associated with burned forests, the percentage of winning species went from 40% to 0%. Regarding ground-dwelling beetles, the percentage of winning species associated with closed forests decreased from 64% to 57% under the harvest scenario under climate change (baseline to RCP 8.5) ( Fig 6c ).
At each climate change scenario (Baseline, RCP 4.5, RCP 8.5), we computed the dissimilarity between Harvest 0.5 and Harvest vs. NoHarvest in 2100. Under Baseline, we compared respectively Baseline-NoHarvest to Baseline-Harvest 0.5 and Baseline-NoHarvest to Baseline-Harvest. Under RCP 4.5 we compared respectively RCP 4.5-NoHarvest to RCP 4.5-Harvest 0.5 and RCP 4.5-NoHarvest to RCP 4.5-Harvest. Finally, under RCP 8.5 we compared respectively RCP 8.5-NoHarvest to RCP 8.5-Harvest 0.5 and RCP 8.5-NoHarvest to RCP 8.5-Harvest. The abbreviation “None” represents the case of species associated to closed forests.
Species assemblage dissimilarity was higher in NoHarvest-Harvest than in NoHarvest-Harvest 0.5 , under the three climate scenarios in 2100 ( Fig 3 ). Under RCP 4.5, Bray-Curtis dissimilarity increased from 0.08 to 0.12 for birds and 0.06 to 0.09 for flying beetles and from 0.08 to 0.11 for ground-dwelling beetles when comparing NoHarvest-Harvest 0.5 to NoHarvest-Harvest scenarios. Compared to birds and ground-dwelling beetles, flying beetles had lower dissimilarity in assemblages regardless of climate scenarios ( Fig 3 ). However, the dissimilarity in species assemblage over harvest gradient decreased with climate change (see the difference in Bray-Curtis dissimilarity between baseline and RCP 8.5 in Fig 3 ). This result was also observed in the structure of land cover under RCP8.5 compared to RCP4.5 and baseline by varying harvest intensity (see S4 Fig ).
Changing climate from the baseline to RCP 8.5 scenarios in 2100 increased the proportion of burned stands at each harvest scenario. For example, the proportion of burned stands was 17.2% under the baseline-Harvest scenario, 24.7% under RCP 4.5-Harvest scenario, and 31.6% under RCP 8.5-Harvest scenario ( Table 3 ). Furthermore, the increase either in wildfire or in harvest levels decreased the proportion of coniferous forests (dense, open and mixed wood). Moreover, climate change as well as forest harvesting decreased significantly stand age. For instance, under RCP 4.5 the mean stand age decreased from 90 to 46 years from NoHarvest to Harvest scenarios. Also, the mean stand age under Harvest 0.5 went from 67 to 49 years from baseline to RCP 8.5 ( Table 3 ).
Of 61, 105 and 42 of candidate species of birds, flying beetles and ground-dwelling beetles, 34, 39 and 15 were selected for the projection with AUC ≥ 0.7 ( Fig 2a–2c ) respectively. Stand age was the most frequently selected predictor variable for the two taxa ( Fig 2d ). Moreover, ground-dwelling beetles were more closely associated with old undisturbed forests than flying beetles. However, the response of flying beetles was much greater to the distance from burned forests than ground-dwelling-beetles, which showed a weaker response that decreased with time after wildfire ( Fig 2d ).
Discussion
Forest harvesting can impact significantly forest biota and may have lasting effects on biodiversity [58]. We aimed to quantify how the interaction between forest harvesting and climate change would impact biodiversity. We analyzed how species assemblages can be expected to change under different forest management scenarios and climate conditions. Our models suggest that forest harvesting will cause pronounced changes in species assemblages for birds and beetles by 2100 under all climate change scenarios. Our result is supported by many empirical studies [4, 59] evoking the noticeable effect of harvesting on the assemblage of common species and predicted that the effect could be stronger by including rare species [4]. Moreover, [17] projects a drastic reduction in the habitat of the Black-backed Woodpecker (Picoides arcticus), a focal species associated with deadwood and representative of old-growth forest biodiversity in eastern Canada [60]. The simulated change in assemblage dissimilarity was mainly dominated by species turnover. This implies that the change in the assemblage composition from NoHarvest to Harvest scenarios was dominated by an increase in occupancy for a subset of species that was balanced mostly by a decrease in occupancy for the other subset of species.
It is important to underline that the effects of forest harvesting were studied under each climate scenario. So, the difference between the less disturbed scenario (baseline-NoHarvest, i.e. 15% of disturbance) and the most disturbed scenario (RCP8.5-Harvest, i.e. 55% of disturbance) on species occurrence could be different from our results based on comparing different harvest levels within the same climate scenario. Our results predict an increase in occupancy in the majority of bird species associated with Early-to-Mid succession forests contrary to species associated with Late succession forests. This result is coherent with previous studies [4, 61]. However, under the extreme climate scenario, there is one more late-succession species increasing in occupancy when we compared harvested to unharvested scenarios, while this is the opposite for early-to-mid succession species which decreased by one species. This difference could be explained by the proportion of burned stands on the harvestable stands under each climate scenario. Indeed, we observed a decrease in the harvested stands with climate change contrary to the burned stands. Those interactions between forest harvesting and future climate change could impact the occupancy of species that depend on disturbed stands. For example, on the one hand, we observed that Yellow Warbler (YEWA, Setophaga petechia), a bird species associated with early-to-mid succession forest and negatively affected by both harvested forest and stand age regressed under RCP 8.5 contrary to the baseline climate scenario. On the other hand, Swainson’s Thrush (SWTH, Catharus ustulatus) that was associated with late succession forest and negatively affected by burned forests increased under RCP 8.5 (see S1 Fig). The increase in wildfires under the extreme climate scenario could affect negatively species that prefer harvested stands if forest managers increase salvage logging. It has been shown that bird trait assemblages differ between burned, post-fire salvage, and traditionally logged habitats in North American forests [62].
Our results on beetles confirm that the dispersal capacity of taxa is an important functional trait to understand future extinction risks for this order of insects. As they are less mobile, ground-dwelling beetles are mainly associated with local conditions [32]. In our study, 93% of them were associated with mature and undisturbed forests (see S2 and S3 Figs). Nevertheless, the frequency of young harvested stands had some support in the models, suggesting that ground-dwelling beetle communities found in old boreal forest persist some years in recently harvested stands, as shown for carabids by [63] in western Canada. For instance, two common carabids, Platynus decentis and Pterostichus adstrictus, selected in our models were “winners” under each climate change scenario. Post-disturbance successions are usually characterized by a deciduous stage, which should favor P. decentis as it is usually found in deciduous and mixed forest [64]. Moreover, P. adstrictus is favored by wildfire [65] and is not affected by forest management [64]. Two species of Cryptophagus (Cryptophagidae), a mycetophagous family usually associated with mature and undisturbed forests, also showed a positive response to harvest intensification. They have low dispersal capacities and their persistence in harvested landscapes suggest that suitable conditions could be maintained as the area of study is covered by a matrix of old-growth forests growing in a humid climate that favor fungal growth. Certain ground-dwelling beetles also respond to recently burned stands. For instance, some carabids are known to be attracted to recently burned stands [66]. In a recent study, [67] suggested that pyrophily in insects have evolved to take advantage of unique post-fire environmental conditions. Indeed, heat-sterilized ovipositing substrates are used by carabids of the genus Sericoda to increase their reproductive output. These conditions are ephemeral and Sericoda disappears rapidly after fire. This could result from suboptimal reproductive conditions. Many other insects are known to respond in such a way to wildfire [68].
We showed that species turnover was the dominant source of dissimilarity in species assemblages for the two studied taxa but ground-dwelling beetles registered the highest rate. This might be due to their lower moving capacity. They cannot escape easily new conditions generated by disturbances. As seen earlier, they can persist for some years after harvesting, but the new habitat conditions end up inducing a higher species turnover. In the same boreal region, [69] showed that post-logging beetle assemblages during forest succession were mostly driven by interspecific interactions than by habitat attributes. It was probably a consequence of the presence of a pristine forest matrix in the surrounding landscape in which local communities differ as they have evolved together over a long time. Our results on ground dwelling beetle occupancy showed a reduction of the rove beetles (Staphylinidae), which are mainly associated with mature and undisturbed forests. They represent 83% of the ground-dwelling “losers” regardless of the climatic change scenario. As the rove beetles modelled were all predators and thus at the top of the food chain, interspecific interactions are critical, and they might be the first to disappear in such turnovers.
In comparison, beetle communities that fly to colonize burned trees after wildfire are much more structured according to habitat characteristics [70]. Among modelled flying beetles in our study, only 44% of species were associated with undisturbed forests. Most were saproxylic beetles associated with burned forests. The time window for successful colonization of burned trees could be narrow for some saproxylic beetle as phloeophagous species may simply benefit from the high-quality sub-cortical tissues of freshly burned trees, usually for 1–2 years only [71–75]. After tree felling, woody debris decay for several decades, allowing great persistence for late-succession beetles [75, 76]. The complexity of the different saproxylic beetle feeding guilds is expressed in the analysis of the change in occupancy for flying beetles. Even though 26% were associated with fire, the “winners” among them dropped from 40% under base scenario to 20% under the RCP8.5 scenario. For instance, in the literature, the six bark beetle species modelled were associated with fire, harvest or both disturbances, but were all “losers” under both climate change scenarios. [77] showed the preference of these beetles to colonize large diameter burned trees, showing the importance of this forest attribute for maintaining biodiversity. However, it is probably the first attribute to be negatively affected by harvesting and climate change. Moreover, wildfire [68, 70], salvage logging [78] and traditional logging [33] rapidly homogenize stands and biodiversity at the landscape level. On the long-term, this lowers landscape heterogeneity as well as biodiversity at the landscape level. As observed for ground-dwelling beetles, changes in occupancy are also most visible for predators in flying beetles. While the proportion of “losers” increases from 50 to 63% between climate change scenarios for predatory ground dwelling beetles, it increases from 44 to 77% for predatory flying beetles. The scenarios tested integrate burned forests interacting with increased harvesting and the resulting landscapes suggest that an erosion in the flying beetle food chain is likely to impact all trophic levels.
Our results were based only on the most common species, the response of rare specialists and threatened species might decrease the regional occupancy with both harvest activities and climate conditions [25]. Therefore, it will be interesting to analyze the response of the rare species to disturbance. One possible approach could be through the development of a probabilistic model which combines the occurrence probability of rare species with indicator species.
[END]
---
[1] Url:
https://journals.plos.org/climate/article?id=10.1371/journal.pclm.0000179
Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.
via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/
