Background

For the last decades, human activities have resulted in a massive worldwide erosion of biodiversity [1]. Loss of natural habitats due to landscape urbanization and fragmentation stands among the reasons for these extinctions [2]. Landscape fragmentation splits natural habitats into multiple isolated patches [3, 4]. Breaking apart of habitats per se has immediate and time-delayed effects on biodiversity [5], independently of habitat loss [6]. In the short term, fragmentation has negative consequences for habitat selection, abundance and species diversity [79]. In the long term, fragmentation limits or disrupts migration and dispersal of juveniles and adults, which can substantially impair metapopulation dynamics [10, 11]. Individual species are then exposed to various stochastic threats, leading in some cases to extinction cascades [12].

Both urbanization and the development of linear transportation infrastructures (LTIs) are causes of fragmentation [13, 14]. LTIs lead to a net disruption of the natural habitats that they cross, splitting them into several distinct patches. Most of the effects of fragmentation on biodiversity detailed above are true for LTIs and have been documented [15]. In particular, LTIs result in an indirect decrease of abundance and species diversity due to habitat fragmentation [16, 17]. They also cause direct animal mortality due to vehicle collisions, electrocutions and drownings of individuals attempting to cross the infrastructures [1820]. Over several generations, LTIs were also shown to lead to a genetic isolation of populations [2123]. The specific case of fragmentation due to the combination of several types of LTIs, such as a road crossing a waterway, remains to be studied.

In the last decades, scientists have paid a lot of attention to the potential of blue-green infrastructures, i.e. networks of ecological land and aquatic continuities, for decreasing habitat fragmentation in the short term [24]. Moreover, in the long term, maintaining a network of ecological corridors could mitigate the effects of global warming through enhanced dispersal of species to newly suitable areas [25]. In the context of biodiversity loss due to habitat fragmentation, the potential of anthropogenic areas for conserving nature deserves to be considered [26].

Up to now, studies about habitat fragmentation have considered LTIs transversally, i.e. they have focused on biodiversity dispersal flows perpendicular to LTIs. Yet, it is unclear whether LTI verges considered longitudinally, i.e. focusing on biodiversity dispersal flows parallel to LTIs, have potential for conserving wildlife. LTIs are generally made up of a transportation lane (road, railway, pipeline, power line, river or canal) and of verges (road and railway embankments, strips of grass under power lines or above buried pipelines, or waterway banks, etc.). A verge is a strip along, above or below the carriageway, inside the LTI boundaries, not directly used for transportation and belonging to the LTI manager. In most cases, verges are covered with plants and may constitute semi-natural habitats. It is thus interesting to assess if, despite their fragmenting effect, LTI verges could contribute to a network of blue-green infrastructures and thus to biodiversity conservation.

At present, few studies have considered LTI verges longitudinally and the studies that have investigated the potential of habitat or corridor of LTI verges provided contrasted results. For instance, according to Bolger et al. [27], revegetated highway rights-of-way could serve as ecological corridors for Californian native rodents and fragmentation-tolerant bird species. On the contrary, Benítez-López et al. [28] showed that mammal and bird population densities, with the exception of raptors, decline with their proximity to infrastructures. Moreover, antagonisms may exist between the potential positive role of habitat or corridor of LTI verges for biodiversity and the risk of animal collision with moving vehicles [29]. A verge management practice can be beneficial to some species and detrimental to others [30, 31]. Besides, some verge management practices positive for biodiversity may be impossible to operate for safety reasons (maintenance of a low-vegetation along roads to avoid collisions, mowing of power line verges to prevent wildfires, etc.) [32].

As there is no consensus in the scientific community regarding this topic, a systematic review taking into account all studies and synthesizing their results seem necessary. Such a review will be helpful for making recommendations to LTI managers by identifying the conditions to which LTI verges could play a role of habitat and/or corridor for biodiversity. Other sorts of literature reviews have been published regarding the role of one specific type of LTI for biodiversity [33, 34]. However, no systematic review exists regarding the potential of all types of LTIs to provide habitats and/or corridors for biodiversity.

The systematic review presented in this paper was initiated by French LTI managing companies and the French government, who wished to investigate the potential for LTIs to provide habitat or corridor for biodiversity. Indeed, in France, the concept of green infrastructures led in 2007 to the development, by the “Ministère de l’écologie, du développement durable et de l’énergie” (MEDDE), of a public policy project named “Trame Verte et Bleue”. Through this project, French administrative regions are currently identifying local ecological networks and developing action plans to preserve and restore these continuities. Moreover, this issue has now to be considered in local urban planning. Through different territorial scales, various stakeholders work on the issue of habitat fragmentation. As the LTI network is very dense in France, LTI managers can significantly contribute to ecological networks. For instance, the French road network, which is over a million kilometer long, is the longest (¼ of the European network) and one of the densest (1.77 km/km2) of Europe. As a comparison, Spain, which has an area close to the one of France, has a road density six times lower (0.32 km/km2). SNCF Réseau owns more than 50,000 km of railway lines, 30,000 of which are currently used, which constitutes the longest railway network of Europe. In France, the role of LTIs in habitat fragmentation is thus strong, compared to other countries. Meanwhile, as the LTI network is dense, the inherent area of verges is considerable. The total area of French road verges is estimated to 4500 km2, which is superior to the total area of 3450 km2 of the seven terrestrial national parks [35]. LTIs have thus a significant potential for contributing to green infrastructures. Aware of these issues, several French LTI managing companies gathered in an informal group, named “Club des Infrastructures Linéaires and Biodiversité” (CILB), to contribute to biodiversity conservation. Among the members of the CILB, six railway, power line, pipeline and waterway French companiesFootnote 1 decided to assess if their LTI verges could contribute to blue-green infrastructures and how to better manage these verges. The systematic review was assumed to be a relevant methodology for finding a scientific answer to this practical question from LTI managers. The French ministry of ecology (MEDDE), through its research incentive program relative to transportation ecology, named “Infrastructures de Transport Terrestre, Ecosystèmes et Paysage” (ITTECOP), undertook a call for tender for the systematic review, with the help of the CILB and the “Fondation pour la Recherche sur la Biodiversité” (FRB). The FRB, a French foundation supporting research in biodiversity, also trained the review team members in the implementation of systematic reviews and helped the review team with its methodological expertise. The “Muséum national d’Histoire naturelle” (MNHN), with the help of the “Institut national de recherche en sciences et technologies pour l’environnement et l’agriculture” (IRSTEA), was selected to conduct the present project.

Objective of the review

The objective of the review is to assess if LTI verges can be used as corridors for longitudinal dispersal movements and if they can provide a habitat for biodiversity. The review will also assess the effect of different management practices (mowing, grazing, etc.) on the potential role of habitat and/or corridor of LTI verges. Considering the constraints (such as safety) that managers are facing to maintain the function of transportation of LTIs, we will try to recommend practices that will contribute to biodiversity conservation. The review will proceed by first mapping the studies relevant to the broad review question followed by synthesis of subsets of studies relevant to more specific questions.

Mapping question

The broad mapping question will be: Can LTI verges constitute a habitat and/or a longitudinal corridor for biodiversity in temperate landscapes?

Synthesis questions

The mapping question can be split into the following topic groups: habitat/corridor and exposure to LTI verge/verge management intervention. This categorization leads to four topic group combinations. Examples of sub-questions for each combination are given below:

  • Habitat-Intervention: Does mowing increase, decrease or have no effect on LTI verge biodiversity?

  • Habitat-Exposure: Is the biodiversity of a LTI verge higher, smaller or equal to the biodiversity of meadow of the surrounding landscape?

  • Corridor- Intervention: Does pruning increase, decrease or have no effect on insect dispersal in a LTI verge?

  • Corridor-Exposure: Is mammal dispersal in a LTI verge higher, smaller or equal to mammal dispersal in a hedgerow of the surrounding landscape?

Once the systematic map will be produced, these sub-questions might be addressed in a synthesis if they are covered by a sufficient number of studies.

Components of the mapping question

Table 1 Footnote 2 displays the components of the mapping question. The present study will not consider the transversal effects of LTIs on biodiversity, such as landscape fragmentation, which has already been demonstrated. It will focus on the longitudinal effects of LTI verges and LTI verge management on biodiversity. Yet, both potential positive (role of habitat/corridor, etc.) and negative (dispersal of invasive species, sink habitat effect, absence of role of habitat/corridor, etc.) longitudinal effects of LTI verges on biodiversity will be considered.

Table 1 Description of the PECO/PICO items of the primary question

Methods

Search terms

The search terms identified by the review team are displayed in Additional file 1.Footnote 3 A first search string, combining some of the search terms with Boolean operators, was tested on web of science. The search hits were compared to a list of 102 includable studies identified by subject expertsFootnote 4 , Footnote 5 (see Additional file 2) and the comprehensiveness was assessed. The search string was then modified, by removing some of the search terms and including new ones, until the highest comprehensiveness was reached. The search strings finally selected are detailed in Table 2.

Table 2 Search strings for two groups of LTIs and two strategies

A first scoping of search hits revealed that a global search string including all LTIs brought many irrelevant results linked to waterways. As a consequence, the search was split into a first string concerning all LTIs except waterways and a second string specific to waterways, which reduced the total number of search hits without decreasing the comprehensiveness.

For each of the non-waterway and waterway searches, two different search strings (strategies 1 and 2; Table 2) were developed in parallel, reaching similar high levels of comprehensiveness. As no argument justified to choose one rather than the other, both strings were retained, results were merged and duplicates were removed. Both search strings (strategies 1 and 2; Table 2) include LTI synonyms, verge synonyms and outcomes and the terms within each category are combined using the Boolean operator ‘OR’. However, while strategy 2 combines each of the three categories with the Boolean operator ‘AND’, strategy 1 combines the categories of verge synonyms and outcomes with the Boolean operator ‘OR’ and the category of LTI synonyms to the two other categories with the Boolean operator ‘AND’. Strategy 1 is based on the consideration that neither the list of verge synonyms nor the list of ecological outcomes is exhaustive. Thus, it combines both of them with the Boolean operator ‘OR’ hoping that articles about unlisted outcomes will be found by the use of a verge synonym and conversely. Strategy 2 separates verge and outcome synonyms in different strings but allows synonyms with a broader meaning so as to obtain a high comprehensiveness anyway.

The same search strings will be used on search engines and specialist websites as on databases. However, if search engines and specialist websites do not allow the same options (wildcards, quotation marks, etc.) as publication databases, search strings will be modified using the database help in order to obtain the search string most similar to the original one. Final search strings used for each database, search engine and specialist website will be recorded in an Appendix, together with search dates. Searches in databases will be undertaken using english terms only, while searches in search engines and specialist websites will be performed either with English or French terms. No time or document type restrictions will be applied.

Publication databases

The search will be conducted on the following publication databases:

  • Web of science core collection

  • Zoological record

These online databases will be used because they cover ecology and are accessible by the members of the research team. Although it would be interesting to include Scopus, no member of the research team has access to it.

Search engines

Internet searches will be performed using the following search engines:

Specialist websites

Websites of the specialist organizations listed below will be searched for links or references to relevant publications and data, including grey literature:

Other literature searches

To alert the research community to this systematic review and to ask for grey literature, national and international experts of transportation ecology will be contacted by e-mail, through the Ecodiff,Footnote 6 Transenviro, Wftlistserv and IENEFootnote 7 mailing lists and by posting a call on a social media (https://fr.linkedin.com/). Moreover, each member of the review team will use its professional network to get information on research related to the topic of the review and to find non peer-reviewed literature, including reports published in French and English. Organizations funding the study will also provide us with their unpublished reports. Finally, authors of unobtainable articles will be contacted by email to ask for their publications.

Study screening and inclusion criteria

Scientific articles collected in databases will be assessed for inclusion at three successive levels: first on titles, second on abstracts and third on full-texts. At each stage, in case of uncertainty, articles will be retained for assessment at the following stage. Article eligibility, at the title screening stage, will be based on the list of selection criteria detailed in Table 3. These criteria encompass both the subject (ecology and related disciplines) and the population (all species of the temperate climatic zoneFootnote 8) of the study. During title screening, studies will be classified into one of the following groups: vertebrates, invertebrates or flora/fungi. Next, each article found to be potentially relevant on the basis of its title will be judged for inclusion on the basis of its abstract. Article assessment on abstracts will rely on the same criteria as for the title stage. Moreover, criteria regarding the exposure/intervention, the comparator, the outcomes or the study type will be added, as detailed in Table 4. Articles will then be assessed on full-texts. Since grey literature does not comply with scientific publishing standards, its assessment will be performed directly on full-texts. Article assessment on full-texts will be based on the same criteria as for the title and abstract stages. Moreover, some inclusion/exclusion criteria specific to the full-text stage, such as the study language, the climate of the study zone or the type of study design, may be added as the review proceeds. A list of studies rejected on the basis of full-text assessment will be provided in an additional file, together with the reasons for exclusion.

Table 3 List of exclusion/inclusion criteria at the stage of title screening
Table 4 List of inclusion criteria at the stage of abstract screening

Before the onset of screening, the review team members taking part in the assessment process will test the consistency of their inclusion/exclusion decisions. For each of the sets of articles about waterways and other LTIs and for each of the assessment stages on title and abstract, a sample of articles will be randomly selected and studies will be screened by each of the reviewers independently of each other. As more than two raters will take part in the study inclusion assessment, a Randolph’s Kappa coefficient will be computed with the R statistical software. A coefficient of 0.6 will be set as the minimal acceptable level of estimated agreement between raters. If the coefficient is lower than 0.6, disagreements will be discussed by raters until common selection criteria are chosen and the operation will be repeated until reaching a coefficient superior to 0.6.

Potential effect modifiers and reasons for heterogeneity

In order to identify potential effect modifiers and reasons for heterogeneity of the results of included studies, eight external experts in landscape connectivity and transportation ecology, and seven scientists of the review team, were gathered and consulted during a one-day workshop. During the workshop, the golden protocol of an ideal study answering our primary question with unlimited resources was discussed. Then, potential effect modifiers and reasons for heterogeneity of our included studies were debated, considering that resources for conducting a study are actually limited.

The following potential effect modifiers were foreseen

  • Study geographic location

  • Site characteristics (type of LTI, type and width of verge, presence of fences, surrounding landscape and history of site disturbances)

  • Timing of the study (study duration, seasonality, duration between exposure/intervention and data sampling, etc.)

  • Biological group studied

  • Verge management practices (mowing, grazing, vegetation burning, pesticide use, etc.)

  • Comparator type (spatial/temporal, etc.)

  • Sampling method (sample size, randomization of sample selection, number of replicates, etc.)

Examples of how geographic location, site characteristics and timing can affect results of a study are respectively given below. Species diversity of a LTI verge will likely be higher in a Mediterranean region than in a northern region. Similarly, the site disturbance history, such as the flooding frequency, will probably affect the floristic composition of a waterway bank. Whether the abundance of a species is measured in spring or winter may affect results. As the present list is not exhaustive, a final list of effect modifiers and reasons for heterogeneity will be established as the review proceeds.

Study quality assessment

Following study inclusion assessment, the quality of included studies will be critically appraised. Based on assessments of their reliability and relevance, included studies will be categorized as having high, medium or low susceptibility to bias. Studies with a high susceptibility to bias will be excluded from the map and from the review. The exact criteria of study quality assessment will be developed as the review proceeds, but they will likely include study design type, coherence of the sampling design, duration between intervention and data sampling, sample size, randomization of sample selection, sampling replication, level of detail of the methodology, comparator relevance for our question, presence of measures of variation of outcomes and description of potential effect modifiers.

Several types of study designs can provide answers to our primary question. Comparison of the impact of different kinds of management on the use of verges as a habitat or a corridor by biodiversity can be made both temporally and spatially. Studies with a before/after (BA) design compare data collected at the same site prior to and following an intervention. Some studies with a BA design use data collected on a single sampling occasion after the intervention, while others use data collected on repeated sampling occasions after the intervention. On the contrary to studies with a BA design, studies with a comparator/intervention (CI) design compare data collected at the same time at different sites, some sites that were subject to a type of management and some sites that were left unmanaged. Finally, studies with a before/after/comparator/intervention (BACI) design combine the two approaches, making both a temporal and spatial comparison. Actually, in a BACI design, data are collected both in a control site and before and after intervention in the study site. These types of study designs have different levels of quality. BA designs with a sampling collection on the long term will likely be more valuable than BA designs with a sampling collection on the short term, since they may account for lasting effects of the intervention and seasonal variation. Moreover, BACI study designs will probably be more reliable than BA and CI designs. Similarly, studies that detail potential effect modifiers will probably be more valuable than studies that do not describe the local environment.

Detailed reasoning concerning critical appraisal will be displayed in a transparent manner. The table of study quality assessment will be included as an appendix. For rejected studies, a short explanation of the reason for exclusion will also be provided.

On top of the scientists of the review team, some external experts might be asked to contribute to the critical appraisal if the number of included articles is high. Each article will first be critically appraised by one reviewer. Uncertain cases will then be critically appraised by a second reviewer. Quality assessment conclusions of the two reviewers will then be compared, and where they differ, disagreements will be discussed until a consensus is reached.

Systematic map database

The studies included after critical appraisal will be mapped in a database. Table 5 displays the coding tool that will be used for the systematic map database. Key variables of interest and coding options within these variables were identified by the review team and subject experts. These key variables include the following type of information: bibliographic reference, geographical localization of the study sites, subject population, type of exposure/intervention, methodological design, setting/context and outcome measures. Studies may be coded with multiple keywords within each coding variable where appropriate. For instance, a study conducted in several regions will be coded with multiple keywords for the “Study region” variable.

Table 5 Coding tool for the systematic map

Each included article will first be coded by one reviewer. Uncertain cases will then be checked by a second reviewer. Potential disagreements will be discussed until a consensus is reached.

The systematic map will describe the evidence on the review topic. It will also identify knowledge gaps and potential specific synthesis questions. The database will be easily searchable and freely accessible.

Data extraction strategy

Once the systematic map will be produced, a synthesis might be realized on more specific questions. In this case, data will be extracted from included studies as follows. Outcome means, measures of variation (standard deviation, standard error, confidence intervals, etc.) and sample sizes will be extracted from tables and graphs, using image analysis software when necessary. If only raw data are provided, summary statistics will be calculated. If relevant data are difficult to extract accurately from graphs or if they are assumed not to be published, authors might be asked to provide primary data. Data on potential effect modifiers will also be extracted from the included articles.

Data synthesis and presentation

The mapping phase will provide descriptive statistics of the regions, taxa and LTIs studied. A narrative synthesis will describe the quality of the results of all the included studies, along with the findings of studies of sufficient quality. Tables will be produced to summarize these results with respect to each of the specific questions. Meta-analyses of effect sizes will be conducted if questions, designs and data formalizations of the included studies are sufficiently homogeneous and if the susceptibility to bias of these studies is low enough. The questions and modalities of the meta-analysis might only be specified once the papers have been read and assessed in order to properly weight the results and choose the metrics for net-effect calculations. Publication bias analysis will also be carried out where possible with the Egger test or the fail safe number.