Hiromitsu Samejima

Bio: Hiromitsu Samejima is an academic researcher from Kyoto University. The author has contributed to research in topics: Biodiversity & Logging. The author has an hindex of 10, co-authored 51 publications receiving 1068 citations.

The importance of correcting for sampling bias in MaxEnt species distribution models

Abstract: Aim Advancement in ecological methods predicting species distributions is a crucial precondition for deriving sound management actions. Maximum entropy (MaxEnt) models are a popular tool to predict species distributions, as they are considered able to cope well with sparse, irregularly sampled data and minor location errors. Although a fundamental assumption of MaxEnt is that the entire area of interest has been systematically sampled, in practice, MaxEnt models are usually built from occurrence records that are spatially biased towards better-surveyed areas. Two common, yet not compared, strategies to cope with uneven sampling effort are spatial filtering of occurrence data and background manipulation using environmental data with the same spatial bias as occurrence data. We tested these strategies using simulated data and a recently collated dataset on Malay civet Viverra tangalunga in Borneo. Location Borneo, Southeast Asia. Methods We collated 504 occurrence records of Malay civets from Borneo of which 291 records were from 2001 to 2011 and used them in the MaxEnt analysis (baseline scenario) together with 25 environmental input variables. We simulated datasets for two virtual species (similar to a range-restricted highland and a lowland species) using the same number of records for model building. As occurrence records were biased towards north-eastern Borneo, we investigated the efficacy of spatial filtering versus background manipulation to reduce overprediction or underprediction in specific areas. Results Spatial filtering minimized omission errors (false negatives) and commission errors (false positives). We recommend that when sample size is insufficient to allow spatial filtering, manipulation of the background dataset is preferable to not correcting for sampling bias, although predictions were comparatively weak and commission errors increased. Main Conclusions We conclude that a substantial improvement in the quality of model predictions can be achieved if uneven sampling effort is taken into account, thereby improving the efficacy of species conservation planning.

Estimating animal density without individual recognition using information derivable exclusively from camera traps

Abstract: Efficient and reliable methods for estimating animal density are essential to wildlife conservation and management. Camera trapping is an increasingly popular tool in this area of wildlife research, with further potential arising from technological improvements, such as video-recording functions that allow for behavioural observation of animals. This information may be useful in the estimation of animal density, even without individual recognition. Although several models applicable to species lacking individual markings (i.e. unmarked populations) have been developed, a methodology incorporating behavioural information from videos has not yet been established. We developed a likelihood-based model: the random encounter and staying time (REST) model. It is an extension of the random encounter model by Rowcliffe et al. (J Appl Ecol 45:1228, 2008). The REST model describes the relationship among staying time, trapping rate, and density, which is estimable using a frequentist or Bayesian approach. We tested the reliability and feasibility of the REST model using Monte Carlo simulations. We also applied the approach in the African rainforest and compared the results with those of a line-transect survey. The simulations showed that the REST model provided unbiased estimates of animal density. Even when animal movement speeds varied among individuals, and when animals travelled in pairs, the model provided unbiased density estimates. However, the REST model was vulnerable to unsynchronized activity patterns among individuals. Moreover, it is necessary to use a camera model with a fast and reliable infrared sensor and to set the camera trap's parameters appropriately (i.e. video length, delay period). The field survey showed that the staying time of two ungulate species in the African rainforest exhibited good fit with a temporal parametric distribution, and the REST model provided density estimates consistent with those of a line-transect survey. Synthesis and applications. The random encounter and staying time model provides better efficiency and higher feasibility than the random encounter model in estimating animal density without individual recognition. Careful application of the random encounter and staying time model provides the potential to estimate density of many ground-dwelling vertebrates lacking individually recognizable markings, and thus should be an effective method for population monitoring.

Co-Benefits of Sustainable Forest Management in Biodiversity Conservation and Carbon Sequestration

Abstract: Background Sustainable forest management (SFM), which has been recently introduced to tropical natural production forests, is beneficial in maintaining timber resources, but information about the co-benefits for biodiversity conservation and carbon sequestration is currently lacking. Methodology/Principal Findings We estimated the diversity of medium to large-bodied forest-dwelling vertebrates using a heat-sensor camera trapping system and the amount of above-ground, fine-roots, and soil organic carbon by a combination of ground surveys and aerial-imagery interpretations. This research was undertaken both in SFM applied as well as conventionally logged production forests in Sabah, Malaysian Borneo. Our carbon estimation revealed that the application of SFM resulted in a net gain of 54 Mg C ha-1 on a landscape scale. Overall vertebrate diversity was greater in the SFM applied forest than in the conventionally logged forest. Specifically, several vertebrate species (6 out of recorded 36 species) showed higher frequency in the SFM applied forest than in the conventionally logged forest. Conclusions/Significance The application of SFM to degraded natural production forests could result in greater diversity and abundance of vertebrate species as well as increasing carbon storage in the tropical rain forest ecosystems.

Predicting the geographical distributions of the macaque hosts and mosquito vectors of Plasmodium knowlesi malaria in forested and non-forested areas

Abstract: Plasmodium knowlesi is a zoonotic pathogen, transmitted among macaques and to humans by anopheline mosquitoes. Information on P. knowlesi malaria is lacking in most regions so the first step to understand the geographical distribution of disease risk is to define the distributions of the reservoir and vector species. We used macaque and mosquito species presence data, background data that captured sampling bias in the presence data, a boosted regression tree model and environmental datasets, including annual data for land classes, to predict the distributions of each vector and host species. We then compared the predicted distribution of each species with cover of each land class. Fine-scale distribution maps were generated for three macaque host species (Macaca fascicularis, M. nemestrina and M. leonina) and two mosquito vector complexes (the Dirus Complex and the Leucosphyrus Complex). The Leucosphyrus Complex was predicted to occur in areas with disturbed, but not intact, forest cover (> 60 % tree cover) whereas the Dirus Complex was predicted to occur in areas with 10–100 % tree cover as well as vegetation mosaics and cropland. Of the macaque species, M. nemestrina was mainly predicted to occur in forested areas whereas M. fascicularis was predicted to occur in vegetation mosaics, cropland, wetland and urban areas in addition to forested areas. The predicted M. fascicularis distribution encompassed a wide range of habitats where humans are found. This is of most significance in the northern part of its range where members of the Dirus Complex are the main P. knowlesi vectors because these mosquitoes were also predicted to occur in a wider range of habitats. Our results support the hypothesis that conversion of intact forest into disturbed forest (for example plantations or timber concessions), or the creation of vegetation mosaics, will increase the probability that members of the Leucosphyrus Complex occur at these locations, as well as bringing humans into these areas. An explicit analysis of disease risk itself using infection data is required to explore this further. The species distributions generated here can now be included in future analyses of P. knowlesi infection risk.

Diversity of Bornean viverrids and other small carnivores in Deramakot Forest Reserve, Sabah, Malaysia

Abstract: We used camera-trapping and night spotlight surveys to investigate carnivores in a lowland tropical rainforest in Borneo. Here we report records of 14 small carnivore species from Deramakot, a commercial forest reserve, where a reduced impact selective logging system is practised. Some of the recorded species like the Otter Civet Cynogale bennettii or the Hairy-nosed Otter Lutra sumatrana have rarely or never been recorded with camera-traps in Borneo. The observed very high diversity of carnivores, especially of globally threatened wetland species, highlights the importance of this lowland forest complex and suggests that even commercially used forests may harbour a high diversity of carnivores.

spThin: an R package for spatial thinning of species occurrence records for use in ecological niche models

Abstract: Spatial thinning of species occurrence records can help address problems associated with spatial sampling biases. Ideally, thinning removes the fewest records necessary to substantially reduce the effects of sampling bias, while simultaneously retaining the greatest amount of useful information. Spatial thinning can be done manually; however, this is prohibitively time consuming for large datasets. Using a randomization approach, the ‘thin’ function in the spThin R package returns a dataset with the maximum number of records for a given thinning distance, when run for sufficient iterations. We here provide a worked example for the Caribbean spiny pocket mouse, where the results obtained match those of manual thinning.

The importance of correcting for sampling bias in MaxEnt species distribution models

Abstract: Aim Advancement in ecological methods predicting species distributions is a crucial precondition for deriving sound management actions. Maximum entropy (MaxEnt) models are a popular tool to predict species distributions, as they are considered able to cope well with sparse, irregularly sampled data and minor location errors. Although a fundamental assumption of MaxEnt is that the entire area of interest has been systematically sampled, in practice, MaxEnt models are usually built from occurrence records that are spatially biased towards better-surveyed areas. Two common, yet not compared, strategies to cope with uneven sampling effort are spatial filtering of occurrence data and background manipulation using environmental data with the same spatial bias as occurrence data. We tested these strategies using simulated data and a recently collated dataset on Malay civet Viverra tangalunga in Borneo. Location Borneo, Southeast Asia. Methods We collated 504 occurrence records of Malay civets from Borneo of which 291 records were from 2001 to 2011 and used them in the MaxEnt analysis (baseline scenario) together with 25 environmental input variables. We simulated datasets for two virtual species (similar to a range-restricted highland and a lowland species) using the same number of records for model building. As occurrence records were biased towards north-eastern Borneo, we investigated the efficacy of spatial filtering versus background manipulation to reduce overprediction or underprediction in specific areas. Results Spatial filtering minimized omission errors (false negatives) and commission errors (false positives). We recommend that when sample size is insufficient to allow spatial filtering, manipulation of the background dataset is preferable to not correcting for sampling bias, although predictions were comparatively weak and commission errors increased. Main Conclusions We conclude that a substantial improvement in the quality of model predictions can be achieved if uneven sampling effort is taken into account, thereby improving the efficacy of species conservation planning.

Mapping species distributions with MAXENT using a geographically biased sample of presence data: a performance assessment of methods for correcting sampling bias.

Abstract: MAXENT is now a common species distribution modeling (SDM) tool used by conservation practitioners for predicting the distribution of a species from a set of records and environmental predictors. However, datasets of species occurrence used to train the model are often biased in the geographical space because of unequal sampling effort across the study area. This bias may be a source of strong inaccuracy in the resulting model and could lead to incorrect predictions. Although a number of sampling bias correction methods have been proposed, there is no consensual guideline to account for it. We compared here the performance of five methods of bias correction on three datasets of species occurrence: one “virtual” derived from a land cover map, and two actual datasets for a turtle (Chrysemys picta) and a salamander (Plethodon cylindraceus). We subjected these datasets to four types of sampling biases corresponding to potential types of empirical biases. We applied five correction methods to the biased samples and compared the outputs of distribution models to unbiased datasets to assess the overall correction performance of each method. The results revealed that the ability of methods to correct the initial sampling bias varied greatly depending on bias type, bias intensity and species. However, the simple systematic sampling of records consistently ranked among the best performing across the range of conditions tested, whereas other methods performed more poorly in most cases. The strong effect of initial conditions on correction performance highlights the need for further research to develop a step-by-step guideline to account for sampling bias. However, this method seems to be the most efficient in correcting sampling bias and should be advised in most cases.