•Linear feature decommissioning is a top priority for boreal caribou conservation.•We used camera traps to link large mammals use of forest roads to road attributes.•Use of forest roads by moose and ...bears was positively linked to forage availability.•Use of forest roads by wolves was related to habitat conditions favoring movements.•Our results could inform road decommissioning efforts to the benefit of caribou.
Decades of expansion of industrial resource extraction in boreal forests have resulted in the legacy of thousands of kilometers of linear features (seismic lines, forest roads) that have fragmented several wildlife habitats. The decommissioning of anthropogenic linear features and the restoration of suitable habitat are top priorities for the recovery of several species at risk, among which, the threatened populations of boreal caribou (Rangifer tarandus caribou). However, the decommissioning of linear features found in caribou range is expensive, and determining which characteristics make them more beneficial to caribou predators and competitors could assist in prioritizing those that may be most critical for boreal caribou habitat restoration. We thus aimed to determine how fine-scale forest road characteristics influence their use by gray wolf (Canis lupus), black bear (Ursus americanus), moose (Alces americanus) and caribou. We used camera traps and generalized linear mixed models to test the effect of road-scale characteristics on the use of forest roads by wolves, bears and moose while also considering larger-scale covariates. Wolves had a greater probability of using roads that were surrounded by wetlands and had a low lateral cover density. For bears, the intensity of use was lower on 20+ year-old roads when compared to 0–10-year-old roads, and higher on roads surrounded by coniferous stands. Moose intensity of use was higher on 11–20-year-old roads and lower on 30+ year-old roads, and decreased on roads surrounded by clearcuts and with a lower number of deciduous stems growing on them. We could not test for caribou use as we did not capture enough events. Nevertheless, by showing which forest roads are more used by caribou predators (wolves and bears) and its apparent competitor (moose), our study highlights the importance of considering both road-scale characteristics and the landscape context in which roads are built to prioritize the most detrimental roads to caribou conservation and guide efficient restoration efforts of its habitat.
•We extracted unpaved roads from RapidEye imagery with 87 % precision and 89 % recall.•Post-processing improved initial predictions produced by the neural network.•Results changed the public road ...database by 20 % through additions and removals.
Accurate and current road network data is fundamental to land management and emergency response, yet challenging to produce for unpaved roads in rural and forested regions using traditional cartographic approaches. Automatic extraction of roads from satellite imagery using deep learning is a promising alternative gaining increasing attention, however most efforts have focused on urban paved roads and used very high spatial resolution imagery, which is less frequently available for rural regions. Additionally, road extraction routines still struggle to produce a fully-connected, vectorized road network. In this study covering a large forested area in Western Canada, we developed and evaluated a routine to automatically extract unpaved road pixels using a convolutional neural network (CNN), and then used the CNN outputs to update a pre-existing government road network and evaluate if and how it would change. To cover the large spatial extent mapped in this study, we trained the routine using moderately high-resolution satellite imagery from the RapidEye constellation and a ground-truth dataset collected with smartphones by organizations already operating and driving in the region. Performance of the road extraction was comparable to results achieved by others using very high-resolution imagery; recall accuracy was 89–97%, and precision was 85–91%. Using our approach to update the pre-existing road network would result in both removals and additions to the network, totalling over 1250 km, or about 20 % of the roads previously in the network. We discuss how road density estimates in the study area would change using this updated network, and situate these changes within the context of ongoing efforts to conserve grizzly bears, which are listed as a Threatened species in the region. This study demonstrates the potential of remote sensing to maintain current and accurate rural road networks in dynamic forest landscapes where new road construction is prevalent, yet roads are also frequently de-activated, reclaimed or otherwise not maintained.
•The relative importance of food- and nesting resource for two bee species was studied.•Which resource that influenced population size depended on the bee's specialization.•Dead wood retention ...favored the cavity-nesting bee with specific demands on its nest.•The floral resource along forest roadsides had great impact on the food generalist.
Wild bees have separate food and nest sites, two essential resources that potentially could limit bee populations. Many solitary bee species nest in holes in deadwood. Female bees collect pollen and nectar, which is stored in the nests as a food supply for their offspring. It is not well understood how availability of either resource affects bee species with different life histories. This study aimed to demonstrate the relative importance of food and nesting resources on population size of two cavity-nesting wild bee species that differ in their requirements regarding food- and nesting resource. Standardized trap nests consisting of wooden poles with pre-drilled holes were deployed in 15 young boreal forest stands to monitor bee abundance. At each site the food resource (flowering plants) and nesting resource (holes in deadwood) for hole-nesting solitary bees were surveyed in the nearby surroundings. The food resource was differentiated into flowers occurring in young forest stands and flowers occurring along forest roadsides. Generalized and general linear models were used to predict nest abundance in the trap nests of two bee species. A total of 166 nests of a pollen-specialist species, Megachile lapponica, and 38 nests of a pollen-generalist species, Hylaeus annulatus were found in the traps. The nest abundance of M. lapponica across the sites was predicted only by this species’ specific food resource, fireweed, Chamerion angustifolium, whereas the nest abundance of H. annulatus was predicted by both food and nesting resources. In a simple linear regression, the density of suitable nesting holes for H. annulatus explained 38% of the variation in the number of nests. Corresponding values for the food resource density along sun exposed gravel roadsides and in young forest stands were 63% and 41% respectively. In a multiple regression, the three variables – nesting resource density, food resource density in young forest stands and food resource density along sun exposed roadsides – explained 86% of the variation in abundance of H. annulatus nests. Nesting and food resource densities for H. annulatus were not correlated with each other. Our results imply that creation and retention of standing dead wood are conservation measures that favor H. annulatus since the availability of nest holes in standing dead wood limited population sizes. To locate the high stumps near sun-exposed forest roads with a dense flora should increase the efficiency of this measure since H. annulatus was particularly favored by flower rich roadsides.
•This study analyzes the impact of various factors on forest road reconstruction.•We employed autoregressive distributed lag model for data analysis.•Timber harvesting and precipitation negatively ...affects forest road reconstruction.•This study used the CUSUM and CUSUMSQ tests to ensure parameter stability.
Forest roads are one of the most essential structures for the continuity of timber harvesting operations. To ensure continuity, forest roads must be open and stable during all seasons. The purpose of this study is to determine the relationships between the amount of forest road reconstruction (FRR) and the factors associated with it. These include average precipitation (AP) amount, timber harvesting (ındustrial wood) (TH) amount and forest road repair and maintenance (FRREM) amount. Also, study aims to test appropriate statistic models for forest road planning and management. In this study, an autoregressive distributed lag (ARDL) model, which is a time series analysis method, is used. Results from the ARDL model demonstrate that an increase in the amount of TH and AP in the long-run relationship negatively affected FRR amounts. In the short-run relationship, the TH and AP variables negatively affected the FRR variable, whereas the FRREM variable positively affected the FRR variable. Granger causality test results show that in the short run there was a one-way causality from FRREM to FRR. Also, for the Granger causality test in the long run, there was causality from other variables to FRREM. This demonstrates that timber harvesting amounts and changes in precipitation amounts due to climate change should be considered when determining road reconstruction activities. In addition, this work will enable forest road managers to make effective and accurate FRR budget planning decisions.
To predict future forest road collapses with the increasing influence of climate change, the relationship between the frequency of forest road collapse and rainfall intensity should be determined. ...Due to the long recurrence period between intense rainfalls, a long-term, broad-based record of forest road collapses is required. The Forest Road Register, which contains the records of the annual number of collapses on forest roads across Japan, is a good data source but does not identify the rainfall intensity that caused each collapse. Thus, we proposed an estimation method using Bayesian inference based on a general regression model to determine rainfall intensity and forest road collapse frequency using Forest Road Register data. To check the feasibility of the proposed method, we evaluated the model using either high- or low-resolution data for forest road collapse in Toyama Prefecture from 1998 to 2018. High resolution data included the date each collapse occurred; thus, the rainfall event that caused each collapse can be identified. Low resolution data included only the annual number of collapses of each road, just like the Forest Road Register does. The expected collapse frequencies were of the same order of magnitude in both models for rainfall events between 100 mm and 400 mm per 24 h. The proposed estimation method using low-resolution data sources is applicable to areas that have a similar or lower frequency of intense rainfall events than Toyama Prefecture. The applicability is also affected by road length registered as one road in the Forest Road Register.
•The first tests of two remote sensing methods in forest road surface deformations.•A handheld mobile laser scanning (TORCH) using the SLAM algorithm was introduced.•A PPK-integrated close-range ...terrestrial photogrammetry system was introduced.•Comparison of the methods by considering their advantages and disadvantages.
This study aimed to compare a handheld mobile laser scanning (HMLS), called TORCH that uses the SLAM algorithm, and a PPK-integrated close-range terrestrial photogrammetry (CRTP) to measure forest road surface deformation. The PPK-integrated CRTP includes a multiband GNSS-module and a camera mounted on a 5-m prism pole. 3D point-clouds were gathered/produced at three different dates with approximately 3-month intervals. And then road surface deformations were determined by applying the M3C2 algorithm. Each method was compared by considering some advantages and disadvantages. PPK-integrated CRTP, which could only be used in areas where the GPS signal is not blocked, provided highly denser 3D point clouds than HMLS. However, for the first period, the difference of mean deformation values between the two methods was not statistically significant, whereas it was statistically significant for the second period. Both methods can be suggested to use in forest road surface deformation yet considering their limitations.
Forest roads are composed of surface and subsurface layers. Determining the seasonal strength and associated moisture changes in these roads is required to understand their capacity. This study looks ...at changes in the subgrade layer. Laboratory and field studies were conducted to determine the changes in subgrade conditions by measuring moisture content and subgrade strength for fine-grained, non-plastic soils in Western Oregon. One field and one laboratory experiment were conducted. Drying in the field experiment occurred during the summer months in the absence of rainfall, while the laboratory study allowed drying for 48 h under controlled suction. In both cases, there was a slight reduction in the subgrade’s moisture content and no significant improvement in subgrade strength. These findings are supported by the soil physics theory that shows that limited water content reduction is expected for these fine-grained soils, as the numerous tiny soil pores can hold water at high capillary tensions. Adding rock layers adds an insulation effect for the subgrade, further reducing evaporation. Consequently, the moisture content remains high, and there will be little change in subgrade strength during the measurement periods.
This study presents a landslide susceptibility assessment for the Caspian forest using frequency ratio and index of entropy models within geographical information system. First, the landslide ...locations were identified in the study area from interpretation of aerial photographs and multiple field surveys. 72 cases (70 %) out of 103 detected landslides were randomly selected for modeling, and the remaining 31 (30 %) cases were used for the model validation. The landslide-conditioning factors, including slope degree, slope aspect, altitude, lithology, rainfall, distance to faults, distance to streams, plan curvature, topographic wetness index, stream power index, sediment transport index, normalized difference vegetation index (NDVI), forest plant community, crown density, and timber volume, were extracted from the spatial database. Using these factors, landslide susceptibility and weights of each factor were analyzed by frequency ratio and index of entropy models. Results showed that the high and very high susceptibility classes cover nearly 50 % of the study area. For verification, the receiver operating characteristic (ROC) curves were drawn and the areas under the curve (AUC) calculated. The verification results revealed that the index of entropy model (AUC = 75.59 %) is slightly better in prediction than frequency ratio model (AUC = 72.68 %). The interpretation of the susceptibility map indicated that NDVI, altitude, and rainfall play major roles in landslide occurrence and distribution in the study area. The landslide susceptibility maps produced from this study could assist planners and engineers for reorganizing and planning of future road construction and timber harvesting operations.