The road to success and the fences to be crossed: Considering multiple infrastructure in landscape connectivity modelling (Dataset and script)
Cite this dataset
Botting, Isla et al. (2023). The road to success and the fences to be crossed: Considering multiple infrastructure in landscape connectivity modelling (Dataset and script) [Dataset]. Dryad. https://doi.org/10.5061/dryad.c59zw3rfg
Abstract
Linear infrastructure represent a barrier to movement for many species, reducing the connectivity of the landscapes in which they reside. Of all linear infrastructure, roads and fences are two of the most ubiquitous, and are understood to reduce landscape connectivity for wildlife. However, what is often neglected consideration is a holistic approach of modelling the effects of multiple types of linear infrastructure simultaneously. Few studies have examined this, typically assessing the impacts of a singular kind of infrastructure on landscape connectivity. Therefore, the aim of this study is to address the relative importance of considering multiple kinds of linear infrastructure in landscape connectivity modelling. We utilised presence data of red deer Cervus elaphus and wild boar Sus scrofa in Doñana Biosphere Reserve (Spain) to generate a sequential approach of scenarios of landscape connectivity; firstly only with environmental variables, secondly with roads as the sole infrastructure, thirdly with the addition of fences, and finally with the further addition of fences and wildlife road-crossing structures. We found that the connectivity of the landscape was greatly affected by the addition of fences and wildlife road-crossing structures in both species, with fences in particular causing considerable alterations to estimated movement pathways. Our finding impresses a need to consider multiple different types of linear infrastructure when modelling landscape connectivity to enable a more realistic view of wildlife movement and inform mitigation and conservation measures more accurately.
README: The road to success and the fences to be crossed: considering multiple infrastructure in landscape connectivity modelling (Dataset and script)
https://doi.org/10.5061/dryad.c59zw3rfg
Images showing presence probability for both study species, and R script implemented to build landscape connectivity models in CIRCUITSCAPE.
Description of the data
Landscape connectivity models were built using CIRCUITSCAPE software. This software relies on electrical circuit theory, creating multiple random movement pathways between focal nodes over a landscape permeability map. D’Amico et al. (2016) calculated the probability of occurrence of both species to the entire study area, using first environmental variables alone, and then environmental and road-related variables together. Here, we used the probability of occurrence maps from D’Amico et al. (2016) in CIRCUITSCAPE (i.e. we assumed that higher likelihoodof occurrence is related to higher landscape permeability). The first layers of presence probability from D’Amico et al. (2016) (without considering roads) fed thebaseline scenario of connectivity (i.e. environmental variables only), where movement was assumed tobe conditioned by habitat quality, without considering roads, fences or passages. We then created three other scenarios for each species, where in the first one we overlaid roads (hereafter road scenario), in the second one we overlaid both roads and fences (hereafter fence scenario), and a thirdscenario where we overlaid roads, fences and wildlife passages (hereafter mitigated fence scenario). The second layers of presence probability from D’Amico et al. (2016) (also considering roads) fed the road scenario, where movement was assumed to be conditioned by environmental variables and roads.
BottingetalDatasetRDenv = TIF image showing red deer presence probability in our study area considering only environmental variables.
BottingetalDatasetRDroads = TIF image showing red deer presence probability in our study area considering only environmental variables and road-related variables.
BottingetalDatasetWBenv = TIF image showing wild boar presence probability in our study area considering only environmental variables.
BottingetalDatasetWBroads = TIF image showing wild boar presence probability in our study area considering only environmental variables and road-related variables.
Sharing/Access information
Data was derived from the following sources: D'Amico M., Périquet S., Román J., Revilla E. (2016). Road avoidance responses determine the impact of heterogeneous road-networks at a regional scale. Journal of Applied Ecology 53(1): 181-190. DOI: 10.1111/1365-2664.12572.
Code/Software
BottingetalRScript = R script implemented to build landscape connectivity models in CIRCUITSCAPE.
Methods
Species occurrence information
Red deer and wild boar are ungulates commonly found in our study area. Presence–absence data used here was collected by D’Amico et al. (2016) using 40 randomly distributed transects, each 200 m long, perpendicular to and beginning from major paved or unpaved roads. The transects were positioned a minimal distance of 2 km apart and were divided into twenty 10 m-long segments. Presence of the focal species was determined by walking along the transect and georeferencing pellets found within a 1 m-wide buffer zone. For more detail, see methods in D’Amico et al. (2016).
Modelling landscape connectivity
Landscape connectivity models were built using CIRCUITSCAPE software (McRae et al. 2008). CIRCUITSCAPE relies on electrical circuit theory, creating multiple random movement pathways between focal nodes over a landscape permeability map. D’Amico et al. (2016) calculated the probability of occurrence of both species to the entire study area, using first environmental variables alone, and then environmental and road-related variables together. Here, we used the probability of occurrence maps from D’Amico et al. (2016) in CIRCUITSCAPE (i.e. we assumed that higher likelihood of occurrence is related to higher landscape permeability).
The first layers of presence probability from D’Amico et al. (2016) (without considering roads) fed the baseline scenario of connectivity (i.e. environmental variables only), where movement was assumed to be conditioned by habitat quality, without considering roads, fences or passages. We then created three other scenarios for each species, where in the first one we overlaid roads (hereafter road scenario), in the second one we overlaid both roads and fences (hereafter fence scenario), and a third scenario where we overlaid roads, fences and wildlife passages (hereafter mitigated fence scenario).
The second layers of presence probability from D’Amico et al. (2016) (also considering roads) fed the road scenario, where movement was assumed to be conditioned by environmental variables and roads. For the fence scenario, we assumed that fences bordering main roads were impermeable to our study species, functioning as a barrier. Well managed fences, such as those paralleling roads, are highly impermeable to ungulates, whereas livestock fences can be more readily permeable (Burkholder et al. 2018, Laguna et al. 2022). In fact, throughout seven years of roadkill fieldwork in this region, we have recorded no roadkills of these species in fenced roads (D’Amico et al. 2015), suggesting that road fences do prevent access by ungulates. As for the livestock fences, scattered throughout the study area, we assumed that they represent a less restrictive movement filter. As there is no detailed information on the degree of barrier they represent, we assumed, based on personal observations, a conservative value in which they reduce the permeability value of the habitat they cross by 20%.
The mitigated fence scenario is similar to the fence scenario, except that we increase the permeability in places where wildlife passages exist, allowing movement through these areas. Overpasses were assumed to be completely permeable, but underpasses are likely to differ in their level of permeability. Nevřelová et al. (2022) defined levels of usability for red deer and wild boar to underpasses based on their OI, but otherwise there is limited evidence as to the level of underpass permeability to ungulates. Therefore, we assigned permeability values for road underpasses and red deer and wild boar based on their OI.
We utilised CIRCUITSCAPE using the wall-to-wall omnidirectional approach (Pelletier et al. 2014) for producing regional-scale maps of connectivity. Omnidirectional methods use the circuit theory algorithm to model the flow of electric current across a resistance grid from all directions, originating from the perimeter of the study area (Koen et al. 2014, Pelletier et al. 2014). The wall-to-wall models allow the flow of electrical current between thin, parallel source and ground strips placed on opposite sides of a buffered study region. Here, we used a buffer distance of 10 km, where the land use outside our study area was randomly assigned for each grid cell, to allow the diffusion of current before entering the study area. The area occupied by the ocean was ignored (NA value).
The flow of current is modelled using the ’advanced mode’ in CIRCUITSCAPE across the region from North to South, South to North, East to West and West to East. The resulting current maps in each of the four directions are then averaged together for a final map of current density, which we used as maps of expected use intensity. Each map of use intensity was scaled to range between 0 and 100, to allow comparisons across species and scenarios of use intensity.
Patterns of landscape connectivity were visually inspected considering the use intensity across the landscape and the different scenarios (i.e. considering the role of the roads, fences and passages). Additionally, in order to better visualise the results, we divided our study area into three distinct sub-areas: the North sub-area (i.e. north of the El Rocío lagoon, including both National and Natural Park), the South sub-area (i.e. the core of the National Park, south of the El Rocío lagoon and east of the paved road road A-483 El Rocío-Matalascañas), and the West sub-area (i.e. mostly including Natural Park, west of the paved road A-483 El Rocío-Matalascañas, Fig. 1). A two-way analysis of variance (ANOVA) was employed to determine whether there were significant differences in estimated use intensity among areas and among scenarios, for each focal species. Tukey’s honestly significant difference (HSD) test was applied to identify the specific groups that differed significantly from one another.
References
Burkholder, E. N., Jakes, A. F., Jones, P. F., Hebblewhite, M. and Bishop, C. J. 2018. To jump or not to jump: mule deer and white-tailed deer fence crossing decisions. – Wildl. Soc. Bull. 42: 420–429.
D’Amico, M., Périquet, S., Román, J. and Revilla, E. 2016. Road avoidance responses determine the impact of heterogeneous road networks at a regional scale. – J. Appl. Ecol. 53: 181–190.
Koen, E. L., Bowman, J., Sadowski, C. and Walpole, A. A. 2014. Landscape connectivity for wildlife: development and validation of multispecies linkage maps. – Methods Ecol. Evol. 5: 626–633.
Laguna, E., Barasona, J. A., Carpio, A. J., Vicente, J. and Acevedo, P. 2022. Permeability of artificial barriers (fences) for wild boar (Sus scrofa) in Mediterranean mixed landscapes. – Pest Manage. Sci. 78: 2277–2286.
McRae, B. H., Dickson, B. G., Keitt, T. H. and Shah, V. B. 2008. Using circuit theory to model connectivity in ecology, evolution, and conservation. – Ecology 89: 2712–2724.
Nevřelová, M., Lehotská, B. and Ružičková, J. 2022. Methodology of wildlife underpasses attractiveness assessment. – Ekológia (Bratislava) 41: 172–182.
Pelletier, D., Clark, M., Anderson, M. G., Rayfield, B., Wulder, M. A. and Cardille, J. A. 2014. Applying circuit theory for corridor expansion and management at regional scales: tiling, pinch points, and omnidirectional connectivity. – PLoS One 9: e84135.
Funding
HORIZON EUROPE European Innovation Council, Award: 101060429, European Union’s Horizon Europe Research and Innovation Programme
Government of the United Kingdom, Turing Scheme
Fundação para a Ciência e Tecnologia, Award: 2022.01773.CEECIND, CEECIND
European Commission, Award: LIFEWATCH-2019-09-CSIC-13
European Commission, Award: POPE2014-2020
Ministry of Economy, Industry and Competitiveness, Award: RYC2021-033192-I, Ramón y Cajal Program
Ministerio de Ciencia e Innovación, Award: IJC2019-039662-I, Juan de la Cierva Incorporación Program