Alternative silvicultural practices with variable retention to improve understory plant diversity conservation in southern Patagonian forests

Understory plants could can act as indicators of temperate forest sustainability, health and conservation status due to their importance in ecosystem function. Harvesting impacts on understory plant diversity depends on their intensity. Variable retention has been proposed to mitigate the harmful effects of timber harvesting, but its effectiveness remains unknown in southern Patagonian Nothofagus pumilio forests. The objectives of this study were to: (i) define a baseline of understory plant diversity in old-growth forests along a site quality gradient and under canopy gaps; (ii) evaluate stands with three different variable retention treatments compared to old-growth forests; and (iii) assess temporal changes during 4 years after harvesting (YAH). A 61 ha N. pumilio forest was selected. Understory plant (Dicotyledonae, Monocotyledonae and Pteridophyta) richness, cover (including woody debris and bare forest floor) and aboveground dry biomass were characterized in summer for 5 years. Before harvesting, baseline samples were conducted along a site quality gradient and outside/inside canopy gaps. Analyzed treatments include a control of old-growth forest (OGF) and three different harvesting treatments with variable retention: (i) dispersed retention (DR) of 30 m² ha⁻¹ (20-30% retention); (ii) aggregated retention (AR) with one aggregate per hectare and clear-cuts (28% retention); and (iii) combined dispersed and aggregated retention (DAR) with one aggregate per hectare and dispersed retention of 10-15 m² ha⁻¹ (40-50% retention). Data analyses included parametric and permutational ANOVAs, multivariate classification and ordinations.Before harvesting, 31 plant species were found, where richness, cover and biomass were directly related to site quality. The presence of canopy gaps did not have a significant impact on the measured variables. After harvesting, 20 new species appeared from adjacent associated environments (two from N. antarctica forests and 18 from grasslands and peatlands). At the stand level, understory values were higher in AR > DR > DAR > OGF. Most (81-95%) plant richness at baseline conditions was conserved in all treatments, where inside the aggregates understory remained similar to OGF. Combination of aggregated and dispersed retention (DAR) better limited exotic species introduction and protected sensitive species, improving conservation in harvested stands. Changes in understory variables were observed after the first YAH in all treatments; greater changes were observed in the harvested areas than in aggregates. Changes stabilized at the fourth YAH. As a conclusion, the location of retention aggregates should be selected to preserve species understory diversity of more speciose and diverse habitats or particularly uncommon stands. Implementation of different kinds (patterns and levels) of retention for improvement of biodiversity conservation in harvested forests should be included in timber and forest management planning.     

Above- and belowground nutrients storage and biomass accumulation in marginal Nothofagus antarctica forests in Southern Patagonia

The above- and belowground biomass and nutrient content (N, P, K, Ca, S and Mg) of pure deciduous Nothofagus antarctica (Forster f.) Oersted stands grown in a marginal site and aged from 8 to 180 years were measured in Southern Patagonia. The total biomass accumulated ranged from 60.8 to 70.8 Mg ha⁻¹ for regeneration and final growth stand, respectively. The proportions of belowground components were 51.6, 47.2, 43.9 and 46.7% for regeneration, initial growth, final growth and mature stand, respectively. Also, crown classes affected the biomass accumulation where dominant trees had 38.4 Mg ha⁻¹ and suppressed trees 2.6 Mg ha⁻¹ to the stand biomass in mature stand. Nutrient concentrations varied according to tree component, crown class and stand age. Total nutrient concentration graded in the fallowing order: leaves > bark > middle roots > small branches > fine roots > sapwood > coarse roots > heartwood. While N and K concentrations increased with age in leaves and fine roots, concentration of Ca increased with stand age in all components. Dominant trees had higher N, K and Ca concentrations in leaves, and higher P, K and S concentrations in roots, compared with suppressed trees. Although the stands had similar biomass at different ages, there were important differences in nutrient accumulation per hectare from 979.8 kg ha⁻¹ at the initial growth phase to 665.5 kg ha⁻¹ at mature stands. Nutrient storage for mature and final growth stands was in the order Ca > N > K > P > Mg > S, and for regeneration stand was Ca > N > K > Mg > P > S. Belowground biomass represented an important budget of all nutrients. At early ages, N, K, S, Ca and Mg were about 50% in the belowground components. However, P was 60% in belowground biomass and then increased to 70% in mature stands. These data can assist to quantify the impact of different silviculture practices which should aim to leave material (mainly leaves, small branches and bark) on the site to ameliorate nutrient removal and to avoid a decline of long-term yields.     

Alternative silvicultural practices with variable retention improve bird conservation in managed South Patagonian forests

Alternative silvicultural approaches to timber management, such as regeneration treatments with different degrees of stand retention, may mitigate negative effects of clear-cutting or shelterwood cuts in forested ecosystems, including changes in old-growth forest bird communities. The aims of this work were: (a) to compare bird species richness and densities among different silvicultural designs with variable retention (dispersed and/or aggregated) and unmanaged primary forests, and (b) to assess temporal changes at community and species levels before and after treatments. A baseline avian survey was conducted prior to harvesting to evaluate canopy gap presence and forest stand site quality influences. Subsequent to harvesting, data on bird species richness and density were collected by point-count sampling during the summer season for 5 consecutive years (4 treatments × 5 years × 6 sampling points × 5 counts). Bird species richness and density (15 species and 9.2 individuals ha⁻¹) did not change significantly with forest site quality of the stands and canopy gap presence in unmanaged forests. However, both variables were significantly modified in managed forests, increasing over time to 18 species and reaching to 39 individuals ha⁻¹. Inside the aggregated retention, bird communities were more similar to unmanaged primary forests than those observed within the dispersed retention or in clear-cuts. Opting for a regeneration method with dispersed and aggregated retention has great potential for managing birds in Nothofagus pumilio forests. This method retained enough vegetation structure in a stand to permit the establishment of early successional birds (at least in dispersed retention), and to maintain the bird species of old-growth forests which could persisted in the retention aggregates.     

Long term effects of fire on ectomycorrhizas and soil properties in Nothofagus pumilio forests in Argentina

The forests of Nothofagus pumilio have historically been affected by forest fires. The effects of fire on certain above and belowground, biotic and abiotic components of these ecosystems have been previously documented, albeit belowground components have received much less attention. It has been suggested that the effects observed in the short-term after a fire usually differ from the longer-term effects. The long-term effects of fire (i.e. >5 years after burning) on belowground components in Nothofagus forests are currently unknown. In the present study we evaluated the long-term effect of fire on ectomycorrhiza (ECM) colonization and morphotype composition in N. pumilio roots, as well as soil chemical properties in temperate forests in Patagonia. Sampling was conducted in three mature monospecific forests. In each, nearby burned and unburned sites were selected. The time since the occurrence of fires differed between areas (i.e. 6-10 years). Within each site, 3 transects of 40 m were established randomly along which 5 samples of roots and soil were collected in spring and autumn. The main results were: (1) in comparison with the unburned site, ECM colonization was lower in the burned site in the area with the shorter time length since fire occurrence and no effects in the other two areas were observed; (2) richness and diversity were not significantly affected by fire but there was a significant effect of season for both parameters, being higher in spring; (3) ECM dominance was significantly higher in the unburned than in the burned site in Tronador, while in Challhuaco the opposite was observed, mainly in autumn; (4) in general carbon, nitrogen and phosphorous decreased while pH increased in the burned sites; (5) ECM colonization positively correlated with NH⁴⁺ and phosphorus and negatively with pH but was not significantly correlated with organic matter or any other soil variable. Altogether the results suggest that effects of fire on ectomycorrhiza and soil properties in N. pumilio forests are probably related to the time elapsed since fire occurrence combined with site characteristics. In addition, the direct and indirect effects of fire in these forest systems may persist for more than 10 years.     

Carbon accumulation along a stand development sequence of Nothofagus antarctica forests across a gradient in site quality in Southern Patagonia

Above- and below-ground C pools were measured in pure even-aged stands of Nothofagusantarctica (Forster f.) Oersted at different ages (5–220 years), crown and site classes in the Patagonian region. Mean tissue C concentration varied from 46.3% in medium sized roots of dominant trees to 56.1% in rotten wood for trees grown in low quality sites. Total C concentration was in the order of: heartwood > rotten wood > sapwood > bark > small branches > coarse roots > leaves > medium roots > fine roots. Sigmoid functions were fitted for total C accumulation and C root/shoot ratio of individual trees against age. Total C accumulated by mature dominant trees was six times greater than suppressed trees in the same stands, and total C accumulated by mature dominant trees grown on the best site quality was doubled that of those on the lowest site quality. Crown classes and site quality also affected the moment of maximum C accumulation, e.g. dominant trees growing on the worse site quality sequestered 0.73 kg C tree⁻¹ year⁻¹ at 139 years compared to the best site where 1.44 kg C tree⁻¹ year⁻¹ at 116 years was sequestered. C root/shoot ratio decreased over time from a maximum value of 1.3–2.2 at 5 years to a steady-state asymptote of 0.3–0.7 beyond 60 years of age depending on site quality. Thus, root C accumulation was greater during the regeneration phase and for trees growing on the poorest sites. The equations developed for individual trees have been used to estimate stand C accumulation from forest inventory data. Total stand C content ranged from 128.0 to 350.9 Mg C ha⁻¹, where the soil C pool represented 52–73% of total ecosystem C depending on age and site quality. Proposed equations can be used for practical purposes such as estimating the impact of silvicultural practices (e.g. thinning or silvopastoral systems) on forest C storage or evaluating the development of both above- and below-ground C over the forest life cycle for different site qualities for accurate quantification of C pools at regional scale.     

Regeneration of Populus nine years after variable retention harvest in boreal mixedwood forests

Aspen and balsam poplar regeneration from root suckers were assessed in boreal mixedwood forests nine years after logging in a variable retention experiment (EMEND Project—Ecosystem Management Emulating Natural Disturbance) located north of Peace River, Alberta, Canada. Five levels of retention of mature trees (2%, 10%, 20%, 50% or 75% of the original basal area) were applied in stands dominated by aspen, white spruce or mixtures of the two species. Basal area of aspen (or that of aspen plus balsam poplar combined) prior to logging strongly influenced sucker density of aspen (or aspen + balsam poplar combined) and in some cases their growth. Nine years after harvest there was a decline in sucker density and volume ha⁻¹ with increasing retention levels of aspen (or both poplars combined); sucker density declined by 50% when only 20% of the original basal area was left in the stand. Retaining mature spruce trees in the stand had little influence on the number of suckers but did affect their total volume ha⁻¹. Thus, we suggest that by knowing stand aspen and balsam poplar density prior to logging and varying levels of retention of aspen and balsam poplar or conifers at harvest, the density of Populus regeneration can be predicted by managers, thereby allowing them to create a range of mixedwood conditions.     

Variable-retention harvests in the Pacific Northwest: A review of short-term findings from the DEMO study

In the Pacific Northwest (PNW) region of the contiguous United States, retention of live (green) trees in harvest units is an integral part of forest management practices on federal lands, yet the ecological benefits that result from various levels or patterns of retained trees remain speculative. The Demonstration of Ecosystem Management Options (DEMO) study was established to address these informational gaps. The experimental design consists of six treatments, each 13 ha in size, replicated at six locations (blocks) in western Washington and Oregon. Treatments represent strong contrasts in retention level (15–100% of original basal area) and pattern (trees dispersed vs. aggregated in 1-ha patches) in mature Douglas-fir (Pseudotsuga menziesii) forests. A wide variety of ecological responses and public perceptions of visual quality have been examined; this paper provides a comprehensive review of the short-term (1–7 years) results of these studies. Level of retention had a strong effect on many responses. At 15% retention, regardless of pattern, microclimate, ecological responses, and public perceptions of visual quality did not differ from those measured in the “clearcut” areas of aggregated treatments. In contrast to level of retention, pattern of retention had limited effect on most measures of biological response. Small changes within forest aggregates were balanced by large changes in adjacent harvested areas, thus on average, responses within aggregated treatments were comparable to those in dispersed treatments. Nevertheless, retaining trees in 1-ha aggregates provided several benefits over dispersed retention. Aggregates greatly reduced damage to and mortality of residual trees (particularly at lower levels of retention) and provided short-term refugia for forest organisms sensitive to disturbance or environmental stress (e.g., bryophytes and late-seral herbs). However, aggregates were susceptible to edge effects (e.g., elevated light and temperature), which may compromise their ability to serve as sources for recolonization of adjacent harvested areas. Collectively, our findings suggest that retention levels >15% are needed to effectively retain sensitive plants and animals, ameliorate harsh microclimatic conditions, and gain public acceptance of retention harvests in these forests. A combination of relatively large (≥1 ha) aggregates and dispersed trees at levels considerably greater than current minimum standards in the PNW may be the most effective strategy for sustaining a broad array of forest values in managed stands.     

Soil organic C as affected by silvicultural and exploitative interventions in Nothofagus pumilio forests of the Chilean Patagonia

This study evaluates the effect of silvicultural and exploitative interventions on soil organic carbon (SOC) in Chilean Lenga (Nothofagus pumilio (Poepp et Endl.) Krasser) forests in south Patagonia. We analyzed SOC and the organic soil horizons in five stands at different stages of development: intact native forest (NI); a 3-year-old shelterwood stand (S3); an 8-year-old shelterwood stand (S8); a 14-year-old stand that was initially treated with shelterwood and subsequently final cut (10 years after the first intervention) (S14), and a 25-year-old stand subject to a exploitative intervention (E25). The SOC under the forest stands, down to a depth of 50 cm (including the Oi horizon), was 60, 55, 71, 85, and 67 Mg ha⁻¹ for the NI, S3, S8, S14, and E25 forest stands, respectively. A significant decrease in SOC occurred 3 years after an intensive shelterwood cut (S3), particularly in the first 5 cm of the mineral soil. Slightly higher carbon contents were observed in the upper horizons of the mineral soil in both the S8 and S14 stands. Consequently, the applied shelterwood system appears to generate only short-term losses of SOC in the Lenga forest. Soil organic carbon increased over the medium term but decreased to the level observed in intact native forests over the long term. Regeneration, which influences stand microclimate (a factor in SOC storage) and provides an important source of organic soil material, was identified as one of the most important factors influencing SOC.     

Canopy microclimate response to pattern and density of thinning in a Sierra Nevada forest

Restoring Sierra Nevada mixed-conifer forests after a century of fire suppression has become an important management priority as fuel reduction thinning has been mandated by the Healthy Forests Restoration Act. However, in mechanically thinned stands there is little information on the effects of different patterns and densities of live-tree retention on forest canopy microclimate. This study compared gradients of air temperature and vapor pressure deficit (VPD) through the vertical forest profile among an overstory-thin, an understory-thin, an un-thinned control, and a riparian environment in a Sierra Nevada mixed-conifer forest. Temperature and humidity were recorded for a year by 60 data loggers arrayed in 12 trees at 5, 15, 25, 35, and 45 m above the forest floor. Both thinning treatments had significantly more extreme summer daily ranges of temperature and VPD than the control across heights. The overstory-thin resulted in the greatest maximum temperatures, VPDs, and VPD range among all sensors at 5 m, and significantly higher summer maximum temperatures and VPDs than the control in lower strata (≤15 m). The understory-thin also had significantly higher summer maximum temperatures than the control (≤15 m), but these too were significantly less than in the overstory-thin nearest the surface at 5 m. Understory thinning did not alter the mean or range of microclimate as much as overstory thinning. Riparian microclimate had significantly lower minimums and means, and greater daily ranges of temperatures and VPDs than the control. Results suggest that thinning canopy cover significantly increases the extremes and variability of understory microclimate compared to thinning from below and no-thin treatments.     

Sphagnum community change after partial harvest in black spruce boreal forests

In eastern Canada, boreal forests develop structural diversity in association with time since stand replacing fire. In some regions, this is associated with significant changes in the bryophyte community (Sphagnum moss invasion) and paludification (thick waterlogged forest floor development). The bryophyte community responds to opening of the canopy, and increasing moisture by replacement of slow growing species by faster growing Sphagnum spp. (e.g. magellanicum, fallax) that are dependent on constant hydration. Within a forest management context, partial harvest systems have been proposed as a strategy to maintain structural diversity, which is currently not accomplished with low retention systems. However, it is unknown whether these interventions will effectively accelerate community succession. The questions addressed in this study were: (1) is the composition of Sphagnum colonies in partially cut stands more similar to old-growth communities than in control, and low retention cut stands, (2) what aspects of harvest disturbance drive these changes, and (3) is the growth rate of Sphagnum capillifolium (an early successional shade tolerant species) different in partial versus low retention harvest systems? After harvest, Sphagnum patch size was reduced by 19.8% and 11.7% after low retention and partial harvest, respectively. While trends were not constant across three separate partial cut trials, the proportion of Sphagnum magellanicum, Sphagnum fallax and Sphagnum fuscum increased compared to controls and low retention 1–2 years after harvest. Models of percent Sphagnum cover indicated machinery track cover, percent cover of vascular plants, and patch depth were positive factors, while the influence of open canopy varied among species. Despite the inclusion of individual disturbance variables, the summary variable ‘treatment’ was significant in all models. Growth of S. capillifolium in partial cuts was intermediate to growth rates in control and low-retention cuts. Growth was positively influenced by slash cover and, contrary to the patch level, negatively influenced by track cover. These results indicate that partial harvest does represent an intermediate level of disturbance, as direct and indirect harvest effects were reduced, as was Sphagnum death. Change in composition 1 and 2 years after harvest indicates that partial harvests may effectively shift the bryophyte community towards an older community type and may thus be used to create landscape diversity. Long term trends and entire community compositions need to be assessed before this can be stated definitively. However, as paludified stands are less productive, the capacity of these partially harvested sites to produce merchantable timber is questioned.     

Bryophytes and decaying wood in Hepatica site-type boreo-nemoral Pinus sylvestris forests in Southern Estonia

The aim of the study was to establish the amount of decaying wood (logs and stumps) in various groups of Hepatica site-type pine forests of different age and management intensity and to analyse the composition of bryophytes in dependence of these factors. The average volume of CWD in old unmanaged forests was 47.5 m³/ha, which is rather well comparable with respective estimations from Fennoscandia. Reduced human impact contributes positively to the amount of CWD. Diversity of log diameter classes and decay stages is larger in old forests. Altogether 73 bryophyte species were recorded, 65 species on logs and 55 on stumps. Species richness on stumps was higher in managed forests than in unmanaged ones. At the same time, the species having high indicator value for man-cut stumps are very common species in boreal forests and grow on other substrata as well. Species composition and ecological conditions differed between stumps and logs. Logs are more humid microhabitats than stumps, therefore the occurrence of hepatics is more frequent on them. According to species composition on decaying wood the old unmanaged forests distinguished from others. As the differences of substratum characteristics were notable between old and young forests, the stand age described a considerable part of species variance on logs.     

Songbird response to experimental retention harvesting in red pine (Pinus resinosa) forests

Traditional harvesting practices frequently result in simplification of the structure and composition within managed forest stands in comparison to their natural counterparts. In particular, loss of heterogeneity within stands may pose a problem for maintaining biodiversity in perpetuity. In this study, we survey breeding bird diversity and abundance in response to different spatial harvesting patterns in mature red pine forests located on the Chippewa National Forest of northern Minnesota, USA. Treatments are designed to increase structural complexity over time and include three overstory manipulations (dispersed retention, aggregate retention with small gaps, and aggregate retention with large gaps), one understory manipulation (brush removal), and controls (no harvesting, and/or no brush removal). In 2003, the first breeding season following the harvest, we found little difference in bird community composition between control and treatment stands. In 2005, the third breeding season following harvest, avian abundance, richness, and diversity were all greater within treatments. Species associated with edge, shrub, and early successional habitats generally show positive response to treatments (e.g. Chestnut-sided Warbler [Dendroica pensylvanica], Mourning Warbler [Oporornis philadelphia], Chipping Sparrow [Spizella passerine]), as do some species associated with mature forest (e.g., Pine Warbler [Dendroica pinus], Rose-breasted Grosbeak [Pheucticus ludovicianus]). Ovenbirds (Seiurus aurocapilla) and Black-throated Green Warblers (Dendroica virens) were more abundant in control stands. There are, as of yet, no discernable differences in avian community composition among the three overstory treatments or between the single understory treatment and the understory control, but differences are expected as the treatments diversify due to understory development. While overstory retention harvests provide habitat for a diverse and abundant bird community, the temporal divergence in avian community composition that we observed between treatment and control stands reveals the importance of uncut, mature red pine forest as a component of a biodiverse landscape.     

西藏天然林资源保护与可持续经营

探讨了西藏天然林资源的现状、特点、问题，提出了天然林资源的可持续经营措施。     

Response of terrestrial small mammals to varying amounts and patterns of green-tree retention in Pacific Northwest forests

To sustain native species in managed forests, landowners need silvicultural strategies that retain habitat elements often eliminated during traditional harvests such as clearcut logging. One alternative is green-tree or variable retention. We investigated the response of terrestrial small mammals to experimental harvests that retained large live trees in varying amounts (approximately 100, 75, 40, and 15% of original basal area) and patterns (aggregated versus dispersed) in mature coniferous forests of western Oregon and Washington. Treatments were applied in 36, 13-ha experimental units. We used pitfall traps to sample small mammals for 4 weeks each autumn during 2 years before and 2 years after treatments. We captured 21,351 individuals of 32 species. We analyzed effects of treatments on relative abundance of 12 species. As level of retention declined, we expected species associated with closed-canopy forests to decrease (Sorex trowbridgii, Neurotrichus gibbsii, Peromyscus keeni, Myodes [Clethrionomys] californicus, and M. gapperi); species associated with early successional habitats to increase (S. vagrans, P. maniculatus, Microtus longicaudus, and Microtus oregoni); and habitat generalists to show little response (S. monticolus, S. pacificus, and S. sonomae). As expected, M. californicus declined after harvest, and P. maniculatus and M. longicaudus increased. Sorex sonomae showed an unpredicted decrease. Other species did not show consistent changes. Responses of S. monticolus, S. sonomae, and M. gapperi varied among study areas. For M. gapperi, this variation was not explained by differences in habitat structure among areas. For all species, capture rates were similar in dispersed- and aggregated-retention units. Similarity in species composition between harvested sites and controls decreased with decreasing retention. Future sampling of these treatments is needed to assess long-term responses. Based on our initial results, green-tree retention strategies need to be sensitive to regional variation in environmental characteristics and small mammal community composition.     

Threshold effects of variable retention harvesting on understory plant communities in the boreal mixedwood forest

We studied the effects of six levels of dispersed green-tree retention (GTR) harvesting (clearcut (0%), 10%, 20%, 50%, and 75%, and unharvested reference (100%)) on understory plant communities in the 8th growing season post-harvest in the mixedwood boreal forest in northwestern Alberta. For the partial harvest treatments (10%, 20%, 50%, 75%) sample plots were located in the partially harvested (retention) strips as well as in the intervening machine corridors used by the harvesting equipment. The understory plant community was significantly influenced by the gradient of retention level. The cover of understory vegetation, especially graminoids, increased with increasing harvesting intensity for the retention strips and overall considering both plots types. Species richness was unaffected by retention level but did decrease as tree density increased. Lower levels of retention lead to increased abundance of early successional, shade-intolerant species. The results suggest a threshold in understory response to GTR harvesting between the 10% and 20% retention treatments. In terms of understory cover and composition, machine corridors within partially harvested forests resembled clearcuts. The results suggest that retaining more than 10% during GTR harvesting could have significant benefits in terms of maintaining understory plant communities more similar to unharvested reference forest.     

德国森林经营理念对慈利县林业发展的启示

通过对德国森林经营理念、近自然林业经营方法的介绍,分析了德国采用近自然林业理论经营管理取得的成效,结合慈利县林业经营工作实际,尝试将近自然林业这一新型理论融入慈利县林业发展进行了一些探讨.     

The decline of Austrocedrus chilensis forests in Patagonia, Argentina: soil features as predisposing factors

The forests of Austrocedrus chilensis in southern Argentina suffer mortality from “mal del ciprés”, whose causes remain unknown. The purpose of this work was to establish the relation of soil features with the occurrence of the disease. In Río Grande Valley, Chubut Province, Argentina, 14 areas with “mal del ciprés” were selected for study. The spatial pattern of the decline varied among the different areas and was classified as aggregated and disaggregated. In each area, symptomatic and asymptomatic plots were established and characterized by 11 edaphic and topographic variables. Three forest areas where the disease was totally absent were also included. Site features were related to the occurrence of the decline using principal component analysis and cluster analysis. Results indicated that soil properties related to poor internal drainage, such as the proximity to water streams, non-allophanized soils of fine textures, and redoximorphic features, act as predisposing factors to the development of “mal del ciprés”. Poor soil drainage was strongly associated not only with the occurrence of the disease, but also with its spatial pattern. Symptomatic and asymptomatic plots presented similar edaphic features in areas with a disaggregated distribution of the decline and were grouped together in the multivariate analysis. This result suggests that large areas with such a pattern are prone to develop the decline.