Effects of even-aged timber harvest on stream salamanders: Support for the evacuation hypothesis

Habitats worldwide are increasingly threatened by degradation and conversion. Critical to the process of habitat loss is the organismal response, which can have effects on immediate conservation measures or future restoration. Among the most threatened and underappreciated habitats are headwater streams, which are small but abundant features of montane forests. These habitats comprise a significant proportion of the total stream length, can harbor remarkable biodiversity, and are critical for numerous ecosystem processes. One of the most abundant organisms in montane headwater ecosystems are salamanders, and therefore what happens to salamanders when the forest habitats surrounding headwater streams are altered? Three main hypotheses exist: (1) mortality hypothesis; (2) retreat hypothesis; and (3) evacuation hypothesis. To examine these hypotheses we evaluated the impacts of even-aged riparian timber harvest on stream-breeding salamanders. Riparian forests along headwater streams were logged, leaving riparian buffers of 0 m, 9 m, and 30 m. Responses to each riparian alteration were measured in terms of salamander terrestrial habitat use and growth in the riparian habitat, as well as changes in population density within headwater streams. Adult and juvenile salamander densities measured in headwater streams were significantly greater in logged riparian treatments than in unaltered riparian treatments. In addition, salamanders significantly reduced their terrestrial habitat use following riparian logging with both the average distance from the stream and the relative abundance of salamanders decreasing. It is unlikely that salamanders will persist in highly modified riparian habitats, as we measured significantly reduced body conditions over short periods of time at these sites. We present corroborative evidence that salamanders evacuate the riparian habitat following intensive riparian logging, emigrating to adjacent headwater streams. Our results underscore the sensitivity of stream salamanders to riparian habitat alteration as well as the importance of riparian buffers in preserving amphibian assemblages.     

Factors influencing stream occupancy and detection probability parameters of stream-associated amphibians in commercial forests of Oregon and Washington, USA

Potential immediate and prolonged impact of timber harvests on stream-associated amphibians (SAA; torrent (Rhyacotriton spp.) and giant salamanders (Dicamptodon spp.) and coastal tailed frogs (Ascaphus truei)) in Oregon and Washington, USA is a management concern, in part because of widespread commercial management of forests across their geographic distributions. Relationships between SAA occupancy and detection probabilities, environmental variables, and management effects were examined in 141 randomly selected perennial streams in commercial forests west of the Cascade Crest in Oregon and Washington from July to October, 2006. Giant salamander occupancy varied by stream substrate type (i.e., consolidated vs. unconsolidated geologies) and was positively associated with stand age. Torrent salamander occupancy varied by species, was positively associated with channel gradient, and was reduced in the youngest and oldest sampled stands. Tailed frog occupancy was negatively associated with the presence of crayfish, negatively associated with low and high bankfull widths, and positively associated with stand age, although stands less than 5 years old had occupancy rates >50% in the absence of crayfish. Mean detection rates were less than 1 for all three genera (0.85, 0.66, and 0.63 for Dicamptodon, Rhyacotriton, and Ascaphus, respectively), which indicates that conclusions from previous research about relationships between SAA occupancy, environmental variables, and forest practices have an unknown element of uncertainty, particularly if detection rates co-vary with habitat change. We believe that our study is the first to present unbiased estimates of occupancy and detection parameters for SAA on forested landscapes in Oregon and Washington.     

Efficacy of riparian buffers in mitigating local population declines and the effects of even-aged timber harvest on larval salamanders

Headwater streams are an important and prevalent feature of the eastern North American landscape. These streams provide a wealth of ecosystem services and support tremendous biological diversity, which is predominated by salamanders in the Appalachian region. Salamanders are ubiquitous throughout the region, contributing a significant biomass that supports ecological and ecosystem processes. One of the greatest threats to salamanders is loss of headwater-riparian habitat through timber harvest. In this study, we measured larval salamander abundance at five headwater streams with different riparian buffer widths retained following logging. By sampling larval salamanders using leaf litter bags, we assessed the impacts of even-aged timber harvest on aquatic larval salamander abundances, where it was found that larvae are negatively impacted by increased stream sedimentation and a decrease in riparian buffer width. We found that retention of a 9-m buffer was effectively no different than complete removal of all riparian forest, and as such, current regulations to protect headwater streams are ineffectual. Furthermore, no significant differences were observed between the 30 m buffer treatment and uncut control treatments suggesting that a 30 m or larger riparian buffer may assuage the in-stream effects of riparian timber harvest. Management guidelines for Appalachian forests should be revised to accommodate the biology of plethodontid salamanders.     

The effects of clearcuts and forest buffer size on post-breeding emigration of adult wood frogs (Lithobates sylvaticus)

The establishment of terrestrial buffer zones around vernal pools has been recommended to provide upland habitat for pool-breeding amphibians in areas where forestry practices occur adjacent to breeding sites. However, few studies have empirically tested the effectiveness of buffers. We assessed post-breeding emigration behavior (net emigration distance, rate of movement, proportionate use of available habitats) of radio-tagged adult wood frogs (Lithobates sylvaticus) at nine vernal pools with experimental forest buffer treatments in central Maine, USA. Buffer treatments were either 30-m (N = 3) or 100-m (N = 3) forest buffers surrounded by a 100-m wide clearcut; pools surrounded by uncut forest served as reference sites (N = 3). We tracked 33 individuals in 2004 and 2005, for an average of 41 days, as they emigrated from breeding pools. Recently clearcut habitat was permeable to emigrating adult wood frogs, particularly females. A higher proportion of frogs at 30-m buffer sites than at 100-m buffer sites traveled through the clearcuts to reach intact forest beyond, suggesting that 30-m buffers may not provide sufficient upland habitat to support adult wood frog populations. There was high variability in emigration behavior among frogs, regardless of buffer treatment, and males and females tended to exhibit differential responses to different buffer sizes and to clearcut habitat. Although wood frogs in this study utilized both 30-m and 100-m forest buffers, variability between sexes and density-dependent effects could render small buffers (e.g., 30 m or less) inadequate to support these populations.     

Understory vegetation in old-growth and second-growth Tsuga canadensis forests in western Massachusetts

We compared the understory communities (herbs, shrubs, and tree seedlings and saplings) of old-growth and second-growth eastern hemlock forests (Tsuga canadensis) in western Massachusetts, USA. Second-growth hemlock forests originated following clear-cut logging in the late 1800s and were 108–136 years old at the time of sampling. Old-growth hemlock forests contained total ground cover of herbaceous and shrub species that was approximately 4 times greater than in second-growth forests (4.02 ± 0.41%/m² versus 1.06 ± 0.47%/m²) and supported greater overall species richness and diversity. In addition, seedling and sapling densities were greater in old-growth stands compared to second-growth stands and the composition of these layers was positively correlated with overstory species composition (Mantel tests, r > 0.26, P < 0.05) highlighting the strong positive neighborhood effects in these systems. Ordination of study site understory species composition identified a strong gradient in community composition from second-growth to old-growth stands. Vector overlays of environmental and forest structural variables indicated that these gradients were related to differences in overstory tree density, nitrogen availability, and coarse woody debris characteristics among hemlock stands. These relationships suggest that differences in resource availability (e.g., light, moisture, and nutrients) and microhabitat heterogeneity between old-growth and second-growth stands were likely driving these compositional patterns. Interestingly, several common forest understory plants, including Aralia nudicaulis, Dryopteris intermedia, and Viburnum alnifolium, were significant indicator species for old-growth hemlock stands, highlighting the lasting legacy of past land use on the reestablishment and growth of these common species within second-growth areas. The return of old-growth understory conditions to these second-growth areas will largely be dependent on disturbance and self-thinning mediated changes in overstory structure, resource availability, and microhabitat heterogeneity.     

Enrichment planting as a silvicultural option in the eastern Amazon: Case study of Fazenda Cauaxi

Liana-dominated forest patches constitute 15–20% of old-growth forests in the Eastern Amazon but are generally excluded from management for timber production. Here we ask if liana-dominated patches may be brought into production by clearing lianas and conducting enrichment planting (EP) of native timber species. We present growth results from 8 years of such EP trials. Rapid growth and low mortality of all species in this study suggest that EP in cleared liana patches can contribute to timber stocks in second and third harvests of managed forests. The most vigorous individuals of Parkiagigantocarpa and Schizolobium amazonicum in each enrichment site grew more than 1 cm diameter per year (rates were initially >2 cm yr⁻¹), and attained dominant canopy positions and diameters equal to those of small canopy trees in the surrounding forest within 8 years of planting (mean dbh ∼18 cm and ∼20 cm, respectively, at year 8). Limited data on Ceiba pentandra plantings indicate a similar trajectory for this species (dbh ∼40 cm in 8 years). The most vigorous Swietenia macrophylla grew at least 1 cm per year in enrichment plots (mean dbh ∼10 cm in 8 years), but take longer to attain dominant positions. Tabebuia serratifolia may take much longer to reach the canopy than other species tested (rates <1 m yr⁻¹). We attribute the excellent performance to light availability; planting in intact soil with minimal compaction and abundant organic material; and low competition rates maintained by periodic thinning of competing vegetation.     

Spatial pattern in herb diversity and abundance of second growth mixed deciduous-evergreen forest of southern New England,USA

This study was designed to answer questions about the patterns of understory diversity in managed forests of southern New England, and the factors that appear associated with those patterns. At the landscape-level, we used plot data to answer questions regarding the spatial distribution of forest understory plant species. Data from a combination of fixed area (understory vegetation) and variable radius (overstory trees) plot methods are combined with site variables for the analysis. Univariate and multivariate statistical methods are used to test for understory diversity relationships with overstory cover types and topography separately, and in combination. Analyses also test for relationships between specific understory species and cover types. In general the understory flora is dominated by four common clonal species that occur across the range of forest cover types: wild sarsaparilla (Aralia nudicaulis L.), Canada mayflower (Maianthemum candense Desf.), star flower (Trientalis borealis Raf.), and partridgeberry (Mitchella repens L.). Results also show that over story composition and structure can be used to assess understory species richness. Species richness follows a general trend among cover types of: hardwood ≥ regenerating forest, hardwood–pine, and pine ≥ mixed ≥ hardwood–hemlock > hemlock. Eastern hemlock (Tsuga canadensis L. Carriere) and mountain laurel (Kalmia latifolia L.) (which decreased in dominance from ridge to valley) both showed negative trends with understory species richness. Topographic position also appears associated with understory floristic patterns (particularly for the hardwood cover type), both in terms of species richness and compositional diversity which both increased from ridge, to midslope, to valley. However, overstory composition (covertype) appears to have a higher order influence on vegetation and mediates the role of topography. The results from this study provide foresters with a better understanding for maintaining floristic diversity and composition of the understory in managed forests.     

Snag longevity of Douglas-fir,western hemlock,and western redcedar from permanent sample plots in coastal British Columbia

Snags are important both as structural components and as animal habitat in forests, but abundance is often low and their dynamics poorly understood in young, managed stands. Using a large data set of 19,622 snags from permanent plots in second-growth forests of coastal British Columbia, we modeled snag longevity (time from tree mortality to snag fall) for three species: Douglas-fir (Pseudotsuga menziesii), western hemlock (Tsuga heterophylla), and western redcedar (Thuja plicata). Snag longevity was strongly related to species and snag size (diameter): the median snag longevity was 16 years for Douglas-fir, 11 years for hemlock and 5 years for redcedar. Western redcedar was predominantly in the subcanopy and its rapid fall rate was related to the small size of its snags. In addition to diameter, other attributes (height to diameter ratio, height, and live crown ratio before death) contributed significantly to models for one or two of the species. However, site level variables did not contribute significantly to any of the models. Snags greater than 50 cm diameter, especially Douglas-fir snags, have the potential for persistence well beyond 20 years in these second-growth forests, and could be important for wildlife.     

Influence of coarse woody debris on herpetofaunal communities in upland pine stands of the southeastern Coastal Plain

Coarse woody debris (CWD) is thought to benefit herpetofauna in a variety of ways including serving as feeding sites, providing a moist environment, and providing protection from temperature extremes. We investigated the importance of CWD to amphibian and reptile communities in managed upland pine stands in the southeastern United States Coastal Plain during years 6 and 7 of a long-term study. Using a randomized complete block design, 1 of the following treatments was assigned to 9.3-ha plots: removal (n = 3; all downed CWD ≥10 cm in diameter and ≥60 cm long removed), downed addition (n = 3; five-fold increase in volume of down CWD), snag (n = 3; 10-fold increase in volume of standing dead CWD), and control (n = 3; unmanipulated). Herpetofauna were captured seasonally using drift-fence pitfall trapping arrays within treatment plots. We compared abundance, diversity, and richness of anurans, salamanders, lizards, and snakes using analysis of covariance with topographic variables (slope, elevation, aspect, and distance to nearest stream) included as covariates. We captured 355 amphibians and 668 reptiles seasonally from January 2007 to August 2008. Abundance, species richness, and species diversity were similar among treatments for anurans, salamanders, and lizards. Snake abundance, species richness, and diversity were higher in removal than downed addition plots. Anuran abundance increased as distance to nearest stream decreased. The majority of species captured during this study are adept at burrowing into the sandy soils of the region. Lack of reliance on CWD may be the result of herpetofaunal adaptation to the longleaf pine (Pinus palustris) ecosystem that historically dominated the upland areas of the study area.     

Soricid response to coarse woody debris manipulations in Coastal Plain loblolly pine forests

We assessed shrew (soricids) response to coarse woody debris (CWD) manipulations in managed upland loblolly pine (Pinus taeda) stands in the upper Coastal Plain of South Carolina over multiple years and seasons. Using a completely randomized block design, we assigned one of the following treatments to 12, 9.3-ha plots: removal (n = 3; all CWD ≥ 10 cm in diameter and ≥60 cm long removed), downed (n = 3; 5-fold increase in volume of down CWD), snag (n = 3; 12-fold increase in standing dead CWD), and control (n = 3; unmanipulated). Therein, we sampled shrews during winter, spring, and summer seasons, 2003–2005, using drift-fence pitfall arrays. During 1680 drift-fence plot nights we captured 253 Blarina carolinensis, 154 Sorex longirostris, and 51 Cryptotis parva. Blarina carolinensis capture rate was greater in control than in snag treatments. Sorex longirostris capture rate was lower in removal than downed and control plots in 2005 whereas C. parva capture rate did not differ among treatments. Overall, the CWD input treatments failed to elicit the positive soricid response we had expected. Lack of a positive response by soricid populations to our downed treatments may be attributable to the early CWD decay stage within these plots or an indication that within fire-adapted pine-dominated systems of the Southeast, reliance on CWD is less than in other forest types.     

Optimizing dosage and preventing leaching of imidacloprid for management of hemlock woolly adelgid in forests

Hemlock woolly adelgid (Adelges tsugae Annand) threatens native hemlock (Tsuga canadensis and T. caroliniana) in the eastern United States. Imidacloprid used as a soil-applied systemic insecticide is effective against A. tsugae in natural forests and in landscapes. Non-target impacts are a special concern because hemlock is ecologically important; often growing next to streams which contain aquatic species that are sensitive to imidacloprid, some of which are protected or endangered fauna. Environmental risk can be mitigated by determining the minimum effective dosage in forests. Adelgid populations responded to imidacloprid dosage with approximately a linear relationship between the percent population reduction (probability scale) vs. log of dosage; 50% reduction in populations could be achieved with 0.15 g per 2.5 cm of trunk diameter at breast height (dbh), or 10% of the maximum labeled dosage. However, effectiveness was found to vary based on dbh; the dosage predicted to give ∼90% reductions in adelgid populations is given by the equation log(dosage) = 0.0153 × dbh − 1.074, where the dosage is g imidacloprid per 2.5 cm of trunk dbh, and dbh is measured as cm. For trees less than 82 cm dbh, these dosages are less than the typical 1.5 g imidacloprid per 2.5 cm dbh currently used. Formulation into tablets rather than as a wettable powder reduced the maximum imidacloprid concentration in soil solution at 5 cm depth from 80 to 1.6 ppm in columns constructed with the organic soil found beneath forest hemlocks. A combination of optimum dosing of trees and adoption of the tablet formulation when treating trees in sensitive habitats should minimize the risk of contaminating aquatic resources with imidacloprid.     

Evidence of continued effects from timber harvesting on lotic amphibians in redwood forests of northwestern California

We compared species richness and relative abundance of stream-associated amphibians in late-seral redwood forests with those in mid-seral, second-growth forests to examine the continued (as opposed to immediate) effects of timber harvest on amphibian populations. Lacking pre-harvest data on amphibian abundances for streams in the second-growth stands, we assumed that nearby tributaries transecting late-seral stands with similar topography and flora harbored similar numbers of animals as second-growth stands prior to harvest. The study was conducted in two blocks (ca. 160 km apart) with three matched-pairs of streams per block. The mid-seral forests (treatment, n = 6) ranged from 37 to 60 years post-harvest; the late-seral forests (control, n = 6) consisted of unharvested stands. We conducted nocturnal visual encounter surveys to sample for amphibians in spring, summer, and fall, for 2 years, with three repeated visits per season. Environmental factors, including water temperature, air temperature, and fine sediment loads were also recorded. Results indicated that amphibian species richness and relative abundances of lotic amphibians were significantly greater in the late-seral forest streams compared with streams transecting mid-seral forests. Water and air temperatures were similar in both forest types, but streams in mid-seral forests had greater amounts of fine sediments compared with the streams in the late-seral forests.     

Ectomycorrhizal fungal communities of rehabilitated bauxite mines and adjacent,natural jarrah forest in Western Australia

Species richness and species composition of ectomycorrhizal (EM) fungi were compared among rehabilitated mine sites and unmined jarrah forest in southwest Western Australia. Species richness, measured in 50 m × 50 m plots, was high. In the wetter, western region, mean species richness per plot in 16-year-old rehabilitated mine sites (63.7 ± 2.5, n = 3) was similar to that of unmined jarrah forest (63.6 ± 9.6, n = 9). In the drier, eastern region, species richness in 12-year-old rehabilitated mine sites (40.3 ± 2.1, n = 3) approached that of nearby forest (52.4 ± 9.3, n = 9). Species composition was analysed by detrended correspondence analysis. Rehabilitated sites of similar age clustered together in the analysis and species composition was closer to the native jarrah forest in the older rehabilitated plots. In unmined forest, species composition of fungal communities in the wetter, western region was different from communities in the drier, eastern region.     

The effects of sudden oak death on foliar moisture content and crown fire potential in tanoak

The introduction of non-native pathogens can have profound effects on forest ecosystems resulting in loss of species, changes in species composition, and altered fuel structure. The introduction of Phytophthora ramorum, the pathogen recognized as causing Sudden Oak Death (SOD), leads to rapid decline and mortality of tanoak (Lithocarpus densiflorus) in forests of coastal California, USA. We tracked foliar moisture content (FMC) of uninfected tanoaks, SOD-infected tanoaks, SOD-killed (dead) tanoaks, and surface litter for 12 months. We found that FMC values differed significantly among the three categories of infection. FMC of uninfected tanoaks averaged 82.3% for the year whereas FMC of infected tanoaks had a lower average of 77.8% (ANOVA, P = 0.04). Dead trees had a significantly lower FMC, averaging 12.3% (ANOVA, P < 0.01) for the year. During fire season (June–September), dead tanoak FMC reached a low of 5.8%, with no significant difference between dead canopy fuels and surface litter (ANOVA, P = 0.44). Application of low FMC values to a crown ignition model results in extremely high canopy base height values to escape crown ignition. Remote estimation of dead FMC using 10-h timelag fuel moisture shows a strong correlation between remote automated weather station (RAWS) 10-h timelag fuel moisture data and the FMC of dead leaves (R² = 0.78, P < 0.01). Results from this study will help refine the decision support tools for fire managers in SOD-affected areas as well as conditions in other forests where diseases and insect epidemics have altered forest canopy fuels.     

Tree biomass estimation in regenerating areas of tropical dry vegetation in northeast Brazil

Allometric equations have been developed for various different vegetation types but have rarely been validated in the field and never for dry tropical forest such as caatinga. In three areas of semi-arid Brazil, with regenerating caatinga vegetation, we measured and weighed twelve hundred individuals of four tree species and used the data to validate equations previously determined in mature caatinga. They and several other equations developed for tropical vegetations overestimate the biomass (B) of trees from the regeneration areas by more than 20%, possibly because these trees have reduced crowns, with lower branch masses. We then determined new allometric equations for them, validating equations for one site against data of the others and pooling the data if they were cross-validated. The best equations were power ones, based on diameter at breast height (D), with little improvement by including height, crown area and/or wood density (Caesalpinia pyramidalis, B = 0.3129D^1.8838; Croton sonderianus, B = 0.4171D^1.5601; Mimosa ophthalmocentra, B = 0.4369D^1.8493; and Mimosa tenuiflora, B = 0.3344D^1.9648 and 0.4138D^1.7718).     

Relative susceptibility of four species of African mahogany to the shoot borer Hypsipyla robusta (Lepidoptera: Pyralidae) in the moist semideciduous forest of Ghana

We examined the relative susceptibility of four mahogany species, Khaya ivorensis, Khaya anthotheca, Entandrophragma angolense, and E. utile, to Hypsipyla robusta attack. Seeds were obtained from one to three parent trees for each species. The research was conducted in the moist semideciduous forest zone in Ghana and used a randomized complete block design. Tree height and diameter and height to first branch were measured until 24 months after out-planting in the field. H. robusta damage was assessed by counting the numbers of shoots attacked, branches, and dead shoots. Khaya spp. grew better but experienced more attack than Entandrophragma spp. The relative susceptibility to H. robusta attack, from most to least, of the four species was: K. anthotheca > K. ivorensis > E. angolense > E. utile. At 24 months, the mean number of shoots attacked per tree ranged from 1.0 for an E. utile seed source to 3.6 on for a K. anthotheca seed source. At 15 months, K. anthotheca and K. ivorensis started branching at about 1.5 m, but height of clear trunk increased over time due to self-pruning. As K. anthotheca grew taller, the number of H. robusta attacks per tree declined. This suggested that selection of genotypes and species that are tolerant of H. robusta attack based on infestation of young plants may not be appropriate. Genetic factors more completely reflecting the response of different species and genotypes to H. robusta attack may manifest themselves at later growth stages.     

Determinants of yield in a non-timber forest product: Copaifera oleoresin in Amazonian extractive reserves

Developing sustainable extractive industries in otherwise intact tropical forest regions requires a sound understanding of the production potential of key resource populations. The oleoresin extracted from Copaifera trees is an economically important non-timber forest product harvested throughout the lowland Amazon basin. We studied oleoresin extraction from four species of Copaifera trees with known harvest histories within two contiguous extractive reserves in western Brazilian Amazonia. We conducted a large-scale experimental harvest of 179 previously unharvested Copaifera trees, in both seasonally flooded (várzea) and adjacent unflooded (terra firme) forests. The likelihood of trees yielding any oleoresin was principally determined by their species identity: C. multijuga was the only species to regularly yield oleoresin (70% of trees). Yield volumes varied both amongst species and forest types: C. multijuga (restricted to terra firme forest) had the highest mean yield of 505 ml, whilst C. guyanensis produced higher volumes of oleoresin in várzea (139 ml) than terra firme (15 ml) forest. Intraspecific differences were driven mainly by tree size. To assess extraction sustainability, we reharvested a sample of C. multijuga trees and compared the oleoresin production of 24 conspecific trees that had been initially harvested one year previously with that of 17 trees initially harvested three years previously. Reharvested trees produced just 35% of the oleoresin volume compared to that when originally drilled, but this response was not affected by the time interval between consecutive harvests. We demonstrate that, within a population of Copaifera, both morphological and environmental factors restrict total productivity; consideration of these factors should inform sustainable management practises. We additionally raise methodological considerations that may improve the comparability of studies.     

Roles of seed and pollen dispersal in natural regeneration of Castanopsis fargesii (Fagaceae): Implications for forest management

Natural regeneration is an important process to reverse the loss of forests. Understanding the process of natural regeneration is crucial for achieving sustainable forest management. In this study, we examined the effects of seed and pollen dispersal in naturally regenerated populations of Castanopsis fargesii. Genetic variation in six populations along two successional series (three successional stages in each series: early, pre-climax, and climax) was assayed using RAPD (random amplified polymorphic DNA) markers. High genetic variability was observed as measured with Shannon's information index. A majority of genetic variation was distributed within populations (Φ_st = 0.1271) and significant isolation by distance existed among populations. A contrasting pattern of genetic variation along these two series was observed, representing different scenarios of natural regeneration processes. The ratio of the number of migrants between the mature populations to the number of migrants from the mature to immature populations was estimated as 1.146 ± 0.099 to 1.981 ± 0.164, implying that comparable seed and pollen dispersal might exist at a fine spatial scale (∼2 km²). The results suggest the critical role of seed dispersal in shaping genetic composition and diversity in the second-growth forests. Barriers to seed dispersal from a variety of genetic sources could result in low genetic diversity in naturally regenerated populations. Management that facilitates the connectivity of newly regenerated stands with mature forests could be effective for natural regeneration given the predominant role of short-distance dispersal of seeds in this species.