1 Introduction

Governments around the world are heavily investing in building new transit infrastructure or upgrading or expanding existing ones [1, 2]. This is in order to draw higher levels of transit ridership, while decreasing the attractiveness of non-sustainable modes of transport (such as private car use). Achieving these goals would help societies in reaching their emerging climate goals by reducing the emissions from the transport sector. The construction of these new transit projects does not happen overnight and can take from a few weeks to several months or even years, which can lead to extended long-term disruptions in the transit network. These long-term disruptions, or transitional periods, can have undesirable impacts on people’s travel behavior, and they could also increase traffic congestion as well as decrease air quality in the short and long terms. The impacts of these transitional periods, in general, seem thematically and geographically dispersed. Similarly, a consolidated understanding of the impacts of long-term disruptions due to other causes, such as labor strikes and transit system failures, on travelers’ behavior seems to be missing from the literature. Albeit different in their origins and characteristics, both transitional periods and long-term disruptions share several attributes, especially in terms of their effects on travel behavior as well as travelers’ needs and perceptions. They both impact travelers for an extended period, which can lead to them developing new habits and attitudes to adapt to such changes. Therefore, both transitional periods and long-term disruptions that resulted from transit system construction and repairs, infrastructure failure, and labor strikes were included in this study.

In this study, the term “transitional periods” refers to any planned changes in the transit system that alter the service’s structure and quality and require an extended period of disruption of regular service operations to implement, such as the construction of new transit infrastructure or the substantial upgrade of such infrastructure. After these periods, users expect to have improved transit service quality or options (Fig. 1), which may have an impact on their travel behavior and perception. It should be noted that not all infrastructure-related disruptions should be considered transitional periods, as some projects may not improve service afterwards. For example, maintenance-related projects can lead to similar levels of service.

Fig. 1
figure 1

Conceptual framework

Full size image

The term “long-term disruptions” was used to refer to any long-term transit system disruptions in which transit returns to its initial service configuration after the disruption, with no to minor changes to the service. For both types of disruptions, transit agencies and cities implement a wide array of mitigation strategies. Therefore, this study aims at achieving the following three goals: comprehending the current state of knowledge in the academic literature regarding the impacts of long-term transit system disruptions and transitional periods on travel behavior, travelers’ perceptions and travel needs; understanding the applied mitigation strategies and technologies used to address any undesirable impacts of these disruptions; and synthesizing the findings to identify areas of overlap between studies and prominent gaps in the current state of knowledge. Exploring these issues together informs transit planners and practitioners of lessons learned across studies regarding similarities and differences in terms of impacts, thereby guiding their future practice.

2 Research context

A considerable number of academic studies explored the factors affecting travel behavior and ridership at the city, route, and stop level during regular operational periods of the transit service. Several studies provided a systematic literature review of these factors’ impacts on travel behavior and ridership [3, 4]. Nevertheless, there are numerous types of disruptions that can affect the normal operations of the transit network. There are also various classifications for these disruptions. Some studies differentiated between them in terms of whether they are planned or unplanned [5,6,7], while another categorization can be in terms of duration (long-term or short-term disruption) as was discussed by Kattan, de Barros [8].

Disruptions can also be divided based on the transport mode or system they affect (like rail transit or road disruptions) or in terms of magnitude—whether these disruptions resulted in closures of the affected transit stations or only caused the redirection of the stations’ lines, for example [9]. Furthermore, Zhu and Levinson [10] categorized transport network disruptions based on their causes; this included transit strikes, bridge closures, earthquakes and special events. A considerable number of researchers investigated the impacts of short-term transit system disruptions that last from a few minutes to hours on travel behavior and transit users’ perceptions [11]. For example, Saxena, Hossein Rashidi [12] compared how travelers weigh trip attributes differently in the case of either canceled or delayed transit service when choosing a mode of transport. Other studies focused on understanding the impacts of long-term disruptions of the transport network [6, 8, 10, 13,14,15,16,17].

In regard to review papers that focused on transport network disruptions, Shalaby, Li [18] conducted a systematic review to identify and analyze journal articles that focused on management strategies for short-term rail transit disruptions and modeling approaches. Zhang, Lo [19] provided a similar review of the academic literature concerning metro system disruption management and substitute bus service, whilst Zhu and Levinson [10] discussed theoretical and empirical studies that focus on traffic and behavioral impacts of transport network disruptions. To the best of the authors’ knowledge, none of the previous research efforts provided an in-depth systematic review of the contemporary academic literature concerning the impacts of long-term transit system disruptions and transitional periods on travel behavior, travelers’ perceptions and travel needs. To address this gap, this paper focuses on developing a comprehensive systematic review of the literature regarding the topic.

3 Methodology

A comprehensive systematic review of the academic literature was carried out. Systematic literature reviews are a powerful approach to identify and analyze all relevant research on a given topic within certain parameters. The research process followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines for systematic reviews [20]. It was initiated by conducting a scan of a few relevant articles to establish the keyword search syntax that would then be used to conduct an article search on three notable research databases. These databases were the Transport Research International Documentation (TRID), the Web of Science Core Collection and Scopus. TRID is a comprehensive database that includes more than one million records of transport research [21]. Following this initial review, four different combinations of keyword search syntaxes were generated based, mainly, on the different categories of long-term transit system disruptions within the scope. More specifically, the four themes of the search queries were as follows: construction and maintenance, general disruptions and closures, labor disputes and strikes, and a more general search related to transport system upgrades or improvements (Table 1). It should be noted that the database search queries pertaining to the fourth keyword category were developed and carried out at a later stage to further expand the limited pool of relevant articles found.

Table 1 Keywords of database search queries for each category
Full size table

Using EndNote, research records for all the documents, which included the abstracts, were organized and some duplicate records were removed. Afterwards, the entire list of records was uploaded to the Rayyan web application [22] where the remaining duplicates were removed and the record screening process was initialized. Rayyan provided a platform for a collaborative work environment for both authors to screen articles, add notes, label articles, and conduct keyword searches. The screening was mainly done by the first author, while checking the undecided articles was done by both authors. To establish clear guidelines to elucidate the process of determining which documents are relevant, inclusion and exclusion criteria were formulated and subsequently followed to filter out the out-of-scope documents (Table 2).

Table 2 Inclusion and exclusion criteria for article selection process
Full size table

The criteria that guided the selection process included that all papers must be in English and must be full peer-reviewed journal articles published no earlier than in 2011. They must also be mainly concerned with the effects of long-term transit system disruptions. There is no clear distinction between short- and long-term disruptions in the literature. Some researchers generally classified short-term disruptions by lasting up to two or three days [23]. Therefore, and to increase the number of included studies in this literature review, we considered the threshold of 2 days or more for identifying long-term transit service disruption. Moreover, research articles investigating disruptions due to highway construction or road maintenance works or that study long-term disruptions of other transport modes, such as air travel and ferries, were beyond the scope of this review and, as a result, were excluded. Additionally, efforts that investigated the effects of natural disasters, epidemics, and pandemics such as the COVID-19 pandemic were removed from the analysis, as these types of events are not only impacting the transit system, but also impacting all transportation modes and people’s willingness to make trips to different destinations.

Based on the criteria presented, the number of records that had their titles and/or abstracts manually reviewed by the authors and were subsequently excluded was 6,588, while the number of records deemed preliminarily relevant and were then pursued for full text retrieval were 47 articles. The review of the articles was based on reviewing the papers’ titles first, if the paper title is not clear or suspected of inclusion, the abstract was then reviewed. Afterwards, a swift review of the full text was carried out for these 47 articles. As a result, 30 articles were excluded, and 17 were included (end of Phase 1). Phase 2, the final phase of this process, comprised scanning the reference lists of the 17 acquired articles for any new articles, while applying the same inclusion and exclusion criteria. This step led to the procurement of 5 additional articles that required a full-text review. Only 2 were deemed relevant and were included in the finalized list of articles for the systematic review (19 in total). This process is illustrated in Fig. 2, which was adapted from Page, McKenzie [20]. Articles that made it to the finalized list were categorized based on the type of disruption that they each discuss. Subsequent to this categorization, a full review of the 19 articles was performed, and the relevant information regarding the different aspects of each paper, such as the issues addressed; the data used; the utilized research methods; and the key findings, was extracted and organized. This was followed by conducting a qualitative analysis of the papers by analyzing the information on individual, categorical, and general scales as well as synthesizing and comparing the findings and other significant features of the articles.

Fig. 2
figure 2

Flow diagram of the systematic review process conducted

Full size image

4 Results

In total, 19 articles were found to focus on long-term transit system disruptions and transitional periods. Even though this may seem like an inconsiderable number of articles for a systematic review, other systematic reviews in the field had similar numbers [24,25,26]. Of these 19, three focused generally on long-term disruptions. Seven papers discussed construction-related transit disruptions, and nine papers explored the effects of transit labor disputes and strikes. Appendices 1, 2, and 3 depict the results of this systematic review. In the appendices, studies that have modeled the impact of a mitigation strategy or controlled for it in the model, or  have critically discussed the effects of a mitigation strategy were highlighted.

4.1 General disruptions studies

As can be seen in Appendix 1, only three articles focused on general long-term transit service disruptions or included sections that focused on them. One paper used semi-structured interviews to understand the factors influencing the mode shift to car among transit users in the case of a transit service disruption. In this study, long-term disruptions referred to the hypothetical absence of transit for 10 years [27]. Pnevmatikou, Karlaftis [17] investigated the impact of a 5-month partial closure of a metro line in Athens, Greece. Lastly, Yap, Nijënstein [28] analyzed the impacts of four tram line disruptions in The Hague, the Netherlands. These analyzed disruptions lasted 5 and 20 days.

4.1.1 Methods and data

Nguyen-Phuoc, Currie [27] used a discourse analysis approach to analyze data collected from semi-structured interview responses of 30 transit users in Melbourne. Most of the participants were from an academic institution (i.e., Monash University). Interviews were coupled with brief questionnaires to collect participants’ socioeconomic information (age, income, occupation, car ownership and driving license) and current travel behavior (i.e., last transit trip). In contrast, Pnevmatikou, Karlaftis [17] adopted nested logit (NL) and multinomial logit (MNL) models to analyze revealed preference (RP) and stated preference (SP) data. The RP survey (1038 responses) examined the impact of a 5-month metro line closure on transport mode choice, which was conducted immediately after the line’s reopening. The SP survey of transit users was web-based and was collected to understand stated preferences towards travel patterns during hypothetical metro closures. In contrast to the previous two studies, Yap, Nijënstein [28] utilized smart card data obtained from an automated fare collection (AFC) system during disruptions to compare between predicted and realized transit ridership.

4.1.2 Studies key findings

Nguyen-Phuoc, Currie [27] found that in the long term, only context-specific factors (travel distance, travel time, travel cost, trip destination and flexibility of alternative mode) have an influence on transport mode shift. The postulated reasons were that participants did not perceive any alteration to the individual-specific factors in the future and that the authors focused on the unavailability of transit for an extremely long period (i.e., next ten years). The study indicated that the prolonged absence of transit services is expected to have an impact on land use and individuals in terms of changing their residential or workplace locations, or both. Participants did not consider trip cancellations during the long-term unavailability of the transit service.

Pnevmatikou, Karlaftis [17] showed that a metro user’s income was an important element in their decision-making process regarding whether to shift to buses or cars during metro service disruptions. Low-income metro users, regardless of car ownership, prefer using buses during metro disruptions. Additionally, during the disruptions, using a car for travel was negatively correlated with having a flexible work schedule. Yap, Nijënstein [28] indicated that in-vehicle time in the shuttle bus service (i.e., rail replacement buses) was perceived about 1.1 times more negatively compared to the in-vehicle time in the initial tram line, while waiting times for the shuttle bus service were perceived as approximately 1.3 times higher compared to the waiting times for the regular bus and tram services. The paper also indicated that if the prediction model does not account for vehicle capacity, integrating the positive effect of higher bus frequency would only overestimate the level of service provided by the shuttle buses during disruptions.

4.1.3 Section summary

Very few studies focused on long-term transit service disruptions or included sections that centered around long-term disruptions. One of them used a qualitative approach to develop a conceptual model of mode shift to car among transit users. The other two articles used specific case studies of partial closures of the metro and tram system in Athens and The Hague, respectively. There were no studies exploring system-wide (or a large portion of the system’s) long-term disruptions, nor were there studies that explored long-term disruptions within the North American context. All three studies focused primarily on current transit users’ travel behaviors. Nevertheless, other travelers may respond differently to additional costs imposed by increased traffic congestion. Additionally, since these studies explored only two cases of disruptions, future work is needed to include a wider array of disruptions because transit users’ responses will depend on: available travel options; duration, type, and degree of disruptions; and the used mitigation strategies’ effectiveness. The two quantitative studies investigated the impacts of providing shuttle buses and using an existing parallel transit service to mitigate the impacts of tram and metro closures, respectively. However, the effectiveness of other mitigation strategies is yet to be explored.

4.2 Construction studies

As seen in Appendix 2, seven studies explored the impacts of construction-related disruptions. Most of them focused on the impacts of heavy rail systems’ (e.g., metro and rail systems) construction and maintenance, while only two discussed light rail transit (LRT) systems [8, 9]. Construction studies are related to the idea of transitional periods, in which people are expected to have improved transit service quality (or options) after these periods. Of the seven papers, 3 focused on the impacts of construction-related transit disruptions on bike-sharing systems usage. The four others investigated the impacts of construction on: travel behavior changes and travelers’ responses to enroute real-time information disseminated through variable message signs (VMS) [8, 13], local air quality [29], and bus performance [14]. Additionally, only five articles critically discussed or analyzed the impacts of mitigation strategies.

4.2.1 Methods and data

Of the seven studies, five utilized statistical models to investigate construction-related transit disruptions’ effects on travel behavior and air quality. Bike-share system studies used ridership data from fixed docking stations [30] or from free-floating bike-sharing systems [9]. Based on bike-share data, these studies explored the impact of metro and LRT closures that lasted from 7 to 25 days in Washington, D.C., USA and Cologne, Germany. They used autoregressive Poisson log-level time series modeling [30], negative binomial regression [9] and linear regressions [31] to analyze ride-share data for periods before, during, and after disruptions, while controlling for a set of variables (weather conditions, season, day, time of day, etc.). One study [31] did not directly model changes in ridership due to disruptions, but rather used sensitivity analysis to explore the effects of implementing a new $2 single-trip fare (STF) on ridership, which was introduced concurrently with the SafeTrack program’s operations in Washington, D.C. Another study modeled bike-sharing usage from geographical and temporal perspectives [9].

On the other hand, other papers used participant survey data to understand changes in travel behavior using summary statistics. For example, Kattan, de Barros [8] used a revealed preference survey (430 responses) collected one year after the West LRT line’s construction in Calgary, Alberta, Canada, had started but before it ended (the construction project’s duration was ~ 3 years). Nevertheless, it did not model travel mode changes but focused on multinomial logit modeling to study travelers’ behavioral responses to VMS information. Zhu, Masud [13] used descriptive statistics to analyze panel survey data before (318 and 420 responses) and after (74 and 64 responses) two reductions and closures of service. Another study used fixed-effect modeling to understand the impacts of rail transit construction on the air quality index using air quality data from 28 cities in China [29]. Lastly, Shiqi, Zhengfeng [14] used fuzzy aggregation and summary statistics to evaluate the bus layout adjustment scheme from passenger and car driver perspectives and investigated passenger volumes at stops.

4.2.2 Studies key findings

Usage of bike-sharing systems generally increased during construction-related transit disruptions but at different levels. For example, Younes, Nasri [30] reported ridership increases on weekdays during disruption. Once the affected metro stations reopened, bike-share ridership returned to its pre-surge levels, suggesting a limited lasting effect of the studied disruptions. They also suggested the likelihood of travelers using bike-share as a first- and last-mile solution rather than as an alternative to transit. Similarly, Schimohr and Scheiner [9] reported a reversion to the original bike-share ridership levels once the disruption had ceased. On another note, Kaviti, Venigalla [31] indicated that implementing a new $2 single-trip fare increased the number of first-time bike-share riders by as much as 79% immediately after its introduction. The introduction of this new fare co-existed with metro service closures. Additionally, there was also a statistically significant increase in the daily ridership of registered members and casual users at docks near metro stations impacted by the metro service closures.

In regard to the other four studies, Kattan, de Barros [8] stated that the total demand for travel in areas affected by the construction did not decrease, neither were trip departure times rescheduled. Travel behavior changes were mainly route switching, followed by mode shifting, and then by destination changing. Zhu, Masud [13] indicated that transit users changed modes or destinations instead of departure time during the complete closure of metro stations. They also reported that ~ 20% of respondents did not return to using the metro even after the service’s full restoration. However, it is not known whether these mode changes are temporary or permanent. Additionally, they observed that wealthier riders are more likely to drive or switch to for-hire options (e.g., Uber and Lyft). On another note, Sun, Zhang [29] found that rail construction has a greater impact on improving air quality than urban road reconstruction, while Shiqi, Zhengfeng [14] suggested that factors attributed to transit service and traffic were degraded when bus routing schemes were implemented during disruptions.

4.2.3 Section summary

Few studies in the literature focus on long-term disruptions caused by transit construction or maintenance projects. Three of the seven studies focused exclusively on understanding the disruptions’ impacts on bike-share ridership rather than their effects on using active transportation modes and cycling behavior. Only two articles employed traveler surveys to gain a better picture of people’s travel behavior instead of their tendency to use one specific mode (bike-share) during such disruptions. Moreover, none modeled transport mode changes due to transit system construction projects, but they rather used descriptive statistics to provide information regarding, for example, route and mode changes, while indicating limited trip cancellations or changes in trip departure times.

Additionally, very few studies looked into long-term changes in travel behavior or home and work location decisions influenced by long-term disruptions as Nguyen-Phuoc, Currie [27] discussed. Most studies focused on immediate impacts and accounted for disruptions within limited time frames. Only five studies explored or modeled the impacts of mitigation strategies on travelers’ travel behaviors or perceptions (Appendix 2). Similarly, it was rare to find articles utilizing detailed mitigation strategy data (e.g., shuttle buses and travel information communications) and travelers’ data in combination with disruption data to explore the mitigation strategies’ effectiveness during transit construction projects. Furthermore, none of the studies were concerned with the relative impact of bus rapid transit (BRT) system construction projects; most of them focused on LRT and metro service construction.

4.3 Labor dispute and strike studies

Only nine papers focused on labor disputes and strikes (Appendix 3). Four of them discussed the impacts of transit operators’ strikes on air quality [32,33,34,35]. Three focused on transit service strikes’ effects on traffic conditions, while one article investigated the strikes’ impacts on the usage of bike-sharing systems. The remaining paper explored a strike’s effect on undergraduate students’ travel choices.

4.3.1 Methods and data

Of the nine articles, six utilized statistical models to investigate transit labor disputes’ effects on people’s travel behavior and on air quality. Regarding air quality studies, they used pollutant levels as proxies to investigate people’s shifting from transit to using private vehicles. Using air quality monitoring stations data, most of them used case studies that lasted from 3 days [32] to 51 days [33, 34]. Additionally, two used a 2013 strike that occurred in Ottawa as their case study, and only one used a sample of more than one long-term strike for investigating their impacts on air quality [35]. The most commonly used data analysis methods included regression, difference-in-difference models [34, 35], and summary statistics [33].

On the other hand, three studies explored transit strikes’ effects on traffic conditions using freeways and highways’ loop detector data [36,37,38]. These studies used different indicators to understand changes in traffic conditions such as changes in average delay, traffic flow, mean speed, and travel time. The used methodological approaches include summary statistics and developing regression and generalized linear models [36, 37]. Only one study [39] used statistical modeling to isolate strikes’ impacts on bike-sharing system usage; it analyzed the impact of one strike that lasted 7 days in Philadelphia, Pennsylvania, USA, using interrupted time series models. Additionally, only one study [40] explored strikes’ impacts on transit users and non-users’ travel behavior using surveys and mobile phone app (called iEpi) data. Using four weeks of data, comprising two weeks during the strike and two weeks of normal operation after the strike ended, they developed descriptive statistics to understand changes in walking distance, trips frequency, and visited locations.

4.3.2 Studies key findings

Most of the air quality studies showed substantial increases in fine particulate matter (PM) concentrations during strikes, particularly in busy urban areas. Moreover, some studies indicated large increases in O3 and CO concentrations. These impacts were mainly attributed to travel behavioral changes; transit users shifted to using private cars. In contrast, two articles reported reductions in NO concentrations during strikes. NO is a gas that is mainly produced by diesel engines, which can be found in public transit buses. Interestingly, one article suggested that PM and O3 concentrations significantly decreased in the strike’s final 3 weeks. This was attributed to travel behavioral changes; travelers started using more environmentally friendly transport modes as they adapted to the transit service’s absence.

Regarding traffic studies, Anderson [37] reported an about 47% increase in highway delays during a 35-day strike in Los Angeles, California, USA. More delays were observed along freeways with parallel transit lines that are characterized by heavy ridership. Other researchers made the same spatial observation [38]. Furthermore, Spyropoulou [36], using a case study from Athens, Greece, of several strikes lasting between 1 and 5 days, reported similar results. Spyropoulou [36] stated that strikes increased congestions by increasing traffic flow (up to 30%), reducing mean speed (up to 27%) and increasing travel times (up to 25%).

Strikes also had significant impacts on using active transportation modes. Fuller, Luan [39] reported a 57% increase in bike-sharing system usage by members and non-members during the 7-day transit operators’ strike. However, bike-share usage quickly returned to previous trends directly after the strike. Additionally, some results suggest that non-members might have used bike-sharing for a slightly longer period. In contrast to previous studies that analyzed the usage of one mode during the strikes (i.e., car or bike-share system), Stanley, Bell [40] focused on the overall changes in the travel behavior of transit and non-transit users during a longer strike that lasted for more than 30 days. They indicated that transit users visited fewer places and walked more during the 30-day strike.

4.3.3 Section summary

Traffic studies showed that the impacts on the transport network were not equal; this is usually not captured by studies that used aggregate air quality data from fixed monitoring stations. Most of the studies examined changes in using one mode, namely cars or bike-sharing systems. However, only very few studies explored and modeled overall alterations in travel behavior. Moreover, most of the articles used passive data sources from traffic loop detectors, air quality stations, or a bike-sharing system at the aggregate level; this does not offer insights into individuals’ behavioral changes in terms of route choice, departure time, travel perception and overall experience. In fact, none of the studies relied on using social surveys to investigate the short- and long-term impacts of transit operators’ strikes on users. In other words, none of them explicitly focused on understanding changes in transit users’ perceptions and needs. Instead, studies tried to draw conclusions regarding transit users’ and non-transit users’ travel behavior.

Only one study utilized mobile-phone data to explore changes in travelers’ behaviors in more detail. This calls for more investigation into possibly using such tools in obtaining larger and more representative samples that can be combined with surveys to better understand different aspects of travelers’ decision-making behaviors during transit operators’ strikes. Furthermore, none of the reviewed studies explored the impacts of service strikes and information availability on travelers’ perceptions using data collected from social media platforms such as Twitter, for example. Table 3 shows an aggregated summary of the analyzed papers from the three categories (general causes, construction-related, and labor-related disruptions) and their different aspects including users’ perceptions and travel behaviors that were investigated.

Table 3 Summary of the reviewed studies
Full size table

4.4 Mitigation strategies

Several mitigation strategies were discussed in the reviewed literature. They generally fall under three broad categories: backup transport services, policy-based measures, and impact assessments (Table 4). In the table, articles are also sorted by the disruption type that they are associated with. Of the 19 papers chosen for this review, 10 articles discussed disruption mitigation strategies in some form.

Table 4 Categorization of mitigation strategies in academic articles
Full size table

Backup transport services are mitigative actions where some alternative form(s) of transport is provided during disruptions to regular transit services. They can be considered to be a form of policy-based measures; however, a distinction has been made between the two since not all of the policy-based measures discussed in the articles were backup transport services. Several articles discussed the level of backup transport services and their impact. For example, Kattan, de Barros [8] discussed the benefits of implementing a temporary BRT service, which followed an alignment similar to that of the LRT under construction, as a proactive mitigation measure that encouraged travelers to shift to using transit. Yap, Nijënstein [28] found that people overestimated the in-vehicle and waiting times associated with using bridging buses compared to their in-vehicle and waiting times using the initial tram service. On another note, Schimohr and Scheiner [9] indicated that travelers’ proximity to stations with substitute or redirected lines was associated with a decrease in the number of bike-sharing trips, suggesting that people used transit at these locations more than they used the bike-share system during service disruption. Nevertheless, most of the studies that discuss this point agree that bus bridging involves a higher level of inconvenience for users, thereby encouraging them to change modes or destinations.

The second category is policy-based measures. This strategy includes providing and improving communications and the dissemination of information to travelers through delivering consistent updates regarding the disrupted transit services, traffic conditions, available alternative modes of transport, etc. It could also include reduced fares, for using transit or a form of active transportation for example, to promote using sustainable transport modes during the disruption to alleviate the hike in traffic congestion. The reviewed articles discussed or analyzed several policy-related amendments or measures. They include introducing a single-trip fare option to encourage riders to use the bike-sharing system and using improved communications methods to disseminate information; those approaches were studied by Kaviti, Venigalla [31] and Kattan, de Barros [8], respectively. In addition, Anderson [37] reported that an additional transit service was contracted (i.e., the Red Line Special bus service) to duplicate part of a closed metro route; this could also be considered a backup transport service. Similarly, Moylan, Foti [38] indicated that during the Bay Area Rapid Transit (BART) service shutdown, a local bus agency (i.e., AC Transit) increased frequencies on Transbay bus service routes. However, the benefits of policies like contracting new services or increasing transit services offered by other agencies were not explicitly measured in previous efforts. Regarding the third category, only one paper [14] focused on evaluating a metro construction project’s disruptive effects on bus performance in the city of Ningbo, China.

5 Discussions and policy implications

The results of this study demonstrate that there is generally a lack of academic research concerning long-term transit service disruptions and transitional periods. Nevertheless, the majority of the identified academic papers are relatively recent and were published during the past five years. This may suggest that this key topic has been gaining more traction in recent years. This showcases this topic’s relevance and the possibility of having more efforts in this area soon. This is in alignment with the increase in worldwide governmental funding opportunities to develop new transit infrastructure to foster economic growth and face climate change. For example, Canada’s federal government revealed a new sizeable funding of $14.9 B for new public transit infrastructure in February 2020 [41]. Similar efforts dedicated to providing more funding for building and upgrading public transport can be found in Europe, the US and China [44,45,46]

According to the number of identified documents, there is a wide agreement and overlap in the reviewed literature regarding the negative impacts such disruptions and transitional periods have on travelers and also regarding the importance of using a range of mitigation strategies. The most common impacts are mode changes. Very few studies indicated other types of changes such as route changes, trip departure time changes, and destination changes. Some evidence, which is rather limited, shows that transit users did return to their previous travel behavior after the end of long-term service disruptions. Nevertheless, it is not clear if these changes are temporary or permanent. Other studies indicated that bike-share systems ridership increased during disruptions. However, these ridership levels returned to their pre-disruption levels after the reopening of the transit service, suggesting a limited lasting effect of long-term disruptions on people’s mode choice to continue using the bike-sharing systems. Providing new backup transit services and rerouting and enhancing parallel services were the most common mitigation approaches widely discussed in the literature to deal with long-term transit disruptions. Previous efforts showed good use of passive data sources from air quality monitoring stations, highway loop detectors, bike-share system counters, and automated fare collection systems to establish evidence of the negative effects of such periods.

Despite these efforts in the literature, travelers’ perceptions and needs during these periods are minimally addressed or analyzed. Additionally, it was rare to find studies that explicitly incorporated or controlled for the expected impacts of the transit projects after their completion. In other words, transitional periods may have more positive outcomes on transit users’ perceptions and travel behaviors after the project is finalized, due to enhanced service quality for instance, compared to other long-term disruptions. Such effects were underexplored in the literature.

The academic literature on long-term disruptions and transitional periods is currently quite divorced from the practice. For example, there is a dearth of studies that seek to derive lessons from past and current practices to help advance the practice of using effective mitigation strategies in different contexts and for different purposes. Additionally, a considerable number of the articles focused solely on understanding the impacts of long-term service disruptions on the usage of one transport mode, such as bike-sharing system usage, or one element, such as air quality or traffic conditions, rather than drawing a complete picture of people’s decision-making process and changes in their travel behaviors and needs. It was also rare to find studies that used a statistical model to better understand travelers’ behaviors during and/or following different types of long-term disruptions and transitional periods.

Most of the studies focused on measuring the short-term impacts of transit service disruptions. This may be related to the fact that most of the analyzed disruptions in the academic literature lasted for a few days or weeks. Nevertheless, articles exploring both short- and long-term impacts of longer transit service disruptions and transitional periods that last for a few months or even years were very limited. In fact, only two studies focused on exploring the impacts of longer disruptions that lasted more than a few months. Using surveys, one study regarding disruptions in Athens explored the immediate impact of a 5-month disruption on travel behaviors, while another study from Calgary explored the 1-year impact of an LRT system construction project, which lasted for about 3 years. Therefore, it is challenging, based on the limited research available, to understand the changes in travel behavior during these prolonged periods and to understand whether these long-term disruptions have an extended or permanent impact on travelers’ behaviors. Potential changes that may not be considered in shorter disruptions include relocation and/or reductions in travel demand because of moving closer to work or school, changing jobs, joining a ride-sharing program, and increasing telecommuting. Some of the key policy recommendations of this research are listed below.

  • With the need for academic studies that focus on the short- and long-term impacts of different long-term disruptions and transitional periods, cities and transit agencies are encouraged to work with the academic community to test different sets of mitigation strategies in different contexts, scenarios, and at different scales. These studies should also include information about changes in travelers’ behaviors, perceptions, and well-being in order to evaluate the used mitigation strategies and their relative impacts, which would inform future policy making.

  • With the emergence of more academic studies, as well as non-academic reports, in this area, lessons from the literature and practice should be organized and used in more systematic ways to assist in developing a policy guide to help in managing these disruptions. This will aid in guiding future practices that should aim to maintain higher levels of the transit services’ attractiveness during such periods for both transit and non-transit users. The prospect of providing adequate active transport alternatives that would encourage people to shift to using active transportation modes should also be explored. This could potentially reduce the stress on the public transportation and road networks.

  • Using agreements with private bus operators, ride-hailing services, bike shops, advocacy groups, and bike-sharing and scooter-sharing companies, cities can help reduce the impact of such periods on travelers. As seen in the literature, offering bike-sharing services and making them cheaper or more accessible by offering more payment options during long-term transit service disruptions can work as a mitigation strategy. This could be coupled with looking beyond the physical availability of alternative modes by testing different pricing scenarios to provide transport alternative(s).

  • Research suggests that using greener transport options may be adopted more widely by travelers if adequate policies were in place during such disruptions. This would capitalize on the increased flexibility of travelers to try out new transport modes during such periods. This might help in increasing cities’ shares of active transportation if such mode changes could be sustained after the disruption and adopted permanently by travelers. Nevertheless, currently, there is limited evidence that this is the case.

  • People will not benefit or suffer increases in their monetary and non-monetary costs equally because of any long-term transit system disruptions or transitional periods. Therefore, transit agencies should assess the equity impacts of such extended time periods on different groups of travelers. This will be context-specific and will help in articulating more sensitive policies that match different groups of users’ needs.

  • Finally, it was reported in the literature that the provision of alternative public transport options with a high transit service level coupled with the efficient dissemination of pre-trip and enroute, real-time travel information (e.g., updates on traffic and on areas affected by the disruption) resulted in an increase in transit use during the disruption. Moreover, evidence of the importance of using social media, graphics, and short videos in communicating information during the COVID-19 pandemic was discussed in the literature [47]. Learning from these lessons, more understanding of the importance of efficient and timely dissemination of information plans using social media or other platforms is essential. This is to help cities in informing people about expected impacts and options, which can help in alleviating stress on transit segments.

6 Conclusions

This study aimed to explore the current state of knowledge concerning transit systems’ transitional periods and long-term disruptions and to understand the actively used disruption mitigation strategies and technologies that are implemented to address or alleviate any of their undesirable impacts on travelers. To achieve these goals, a comprehensive systematic review of the academic literature was conducted. In total, 19 peer-reviewed journal articles were identified and analyzed. This systematic review helps identify the major knowledge gaps in the literature. The results of this study demonstrate that there is generally a lack of academic research works concerning long-term transit service disruptions and transitional periods. In fact, travelers’ perceptions, travel behaviors and changing needs during these disruptive periods were minimally addressed or analyzed. Key conclusions and recommendations are discussed below.

  • Given the range of different types of long-term transit system disruptions (e.g., construction, labor disputes and service failures); various disruption time frames (from a few days to several years); varying spatial coverage (from one line to system level); and different disrupted modes (e.g., bus, metro and tram), much more work can be done to explore the effects of such disruptions on people’s travel behavior and perceptions.

  • The possible impact of long-term disruptions on travelers in terms of changing modes, routes, trip departure times, frequency of trips, destinations, and work and home locations in addition to increasing telecommuting and trip sharing are widely recognized in the literature. However, studies rarely explored explicitly the factors affecting changes in travel behavior like shifting to different routes or changing the frequency of trips or the factors influencing relocation for transit and non-transit users using statistical models. This can be an important area for future work.

  • The relative importance and impact of disruption mitigation measures, while understanding how these measures could work together, within the context of long-term disruptions are rarely investigated in the literature. In fact, the current academic literature provides transit agencies with very limited information to assist them. Such knowledge can inform the processes of planning for long-term disruptions to implement more efficient and effective strategies.

  • Most of the studies were quantitative in nature and provided some relevant findings; however, qualitative studies can provide in-depth insights into the intersection between how, why, who, and what questions that are related to different types of disruptions. For example, it can help in understanding the importance of using different mitigation strategies for different groups of the population and their relation to different types of disruptions. Therefore, future research can focus on using qualitative approaches to elicit information not only from transit and non-transit users, but also from transit agencies and operators to understand their perspectives.

  • The reviewed literature generally used data from two main sources: from traveler surveys and from passive data sources that are obtained from air quality monitoring stations, bike-share systems, and highway loop detectors. Therefore, using emerging data sources such as cellular phone data and mobile app data can be explored for future studies to give a better understanding of the different long-term disruptions’ impacts. Social media data, farebox system data, and web surveys can be also incorporated in these studies.

  • Several studies used summary statistics, difference-in-difference approaches, or a dummy variable to isolate the impacts of long-term disruptions on different aspects (e.g., air quality, traffic conditions, and bike-share usage). However, these studies ignore that a long-term disruption entails an extended period of time, which can see different movement patterns within this period as indicated by Chandler and Shymko [34]. They stated that relying only on short-term results to draw conclusions regarding long-term impacts of long-term disruptions can lead to an overestimation of the negative effects. Therefore, future research could look into different patterns within such time frames.

  • Similarly, temporal changes in travel behavior after long-term disruptions or transitional periods end are rarely explored in the literature, particularly for longer disruptions that last for more than a few weeks. Some authors indicated that transit ridership can take several weeks to reach pre-disruption levels [37]. Therefore, exploring changes occurring over time to travelers and, more specifically, transit users’ travel behavior could be a viable future research endeavor.

  • Previous studies explored the impacts of long-term disruptions and transitional periods on bike-share usage; however, the academic literature is lacking in studies that investigate their impacts on using other active transportation modes, such as walking and cycling. Since these trips, particularly walking trips, are usually underreported in travel surveys, using sensors data from mobile phone apps can be beneficial to understand changes before, during, and after long-term disruptions for different groups of travelers.

  • It was found that very few articles explored changes in travelers’ perceptions, satisfaction, and needs due to transitional periods in comparison with long-term transit system disruptions triggered by other causes. Transitional periods, which usually lead to different outcomes in terms of improved service quality after ending compared to other long-term disruption periods, were not explicitly explored in the literature. Transitional periods’ ultimate positive outcomes on users’ perceptions and travel behaviors could be a viable focal point for future research efforts.

  • Incorporating social issues, such as equity concerns, and seeking to derive lessons to help understand the equity impacts of long-term disruptions and transitional periods on different groups of populations are yet to be accomplished. These groups can include Indigenous populations, visible minorities, and people with systemic barriers to using other transportation modes. Additionally, there is a lack of research to both understand and address the sociopolitical, institutional and community capacity dimensions, which is an important aspect during such periods.

Finally, this study aimed to derive lessons from the current academic literature on the effects of transit systems’ long-term disruptions and transitional periods. Exploration of this rather diverse research area will not only inform professionals but will also highlight important gaps in the current literature for researchers. Future research can expand the presented efforts and focus on reviewing transit agency reports and studies and conducting surveys and interviews with transit planners to record their experience and to better understand their perspective of the effects of transit systems’ long-term disruptions and transitional periods. This is to help cities and transit agencies to better anticipate and manage “change”. This will, in turn, help to facilitate and secure the development of their public transport networks while planning and being better prepared for long-term transit system disruptions, thereby aiding them in achieving their overarching sustainability goals.