Introduction

‘Vulnerability assessment’ is a widely used term to describe approaches and techniques with objectives of understanding and management of risks or hazards. These assessments have been used in wide fields from engineering (Quiñones-bustos et al. 2021) to computing and cybersecurity (Shinde and Ardhapurkar 2016) in addition to the environmental sciences (Adger 2006; Ellison 2015; Sharma and Ravindranath 2019). Across these, details of the approaches and methods likely vary widely, along with definitions of the term itself.

In environmental sciences, vulnerability is defined with respect to climate change by the Fourth Assessment report by the Intergovernmental Panel on Climate Change (IPCC) as the degree to which geophysical, biological and socio-economic systems are susceptible to, and unable to cope with, adverse impacts of climate change (Schneider et al. 2007: 783). IPCC described vulnerability as consisting of three dimensions: sensitivity, exposure, and adaptive capacity (IPCC 2007, p.64). The later IPCC Fifth Assessment synthesis report considered vulnerability to consist of sensitivity and adaptive capacity, and exposure acting as an external hazard (IPCC 2014, p.54). Though the definitions vary, the definitions of individual dimensions are generally consistent.

Exposure refers to the external characteristics of a system which impact vulnerability (Adger 2006; Ellison 2015), presence of species or ecosystems in settings that could be adversely affected by external pressures (IPCC 2014). For example, exposure includes orientation to waves and tidal processes (Gornitz 1991). Contrastingly, sensitivity involves the internal characteristics of a system which impact vulnerability such as vegetation condition (Adger 2006; Ellison 2015), the degree to which a system or species is affected by climate variability or change (IPCC 2014). Adaptive capacity refers to the ability of a system to adapt to change, how much institutions, humans, and other organisms can adjust (IPCC 2014). Adaptive capacity can be controlled by a number of different variables from community involvement in environmental management to legislative protection (Adger 2006), and includes innate abilities of organisms to adapt such as migration inland with sea level rise (Ellison 2015).

During these assessments, a variety of different methods exist, from those that are complex and mathematical to those that are more qualitative or simpler in approach. In this study, vulnerability assessments carried out on saltmarshes, beaches and shorelines were reviewed. Similarities and differences among literature along with research gaps were analysed and identified.

Methods

In preparation for NVivo analysis, literature from 2010 to 2023 was sampled and collected from Google Scholar (Google 2023), using the advanced search function. Google scholar searches scholarly literature such as journal articles and books, but mostly excludes government and NGO reports. The start year of 2010 was to include studies influenced by the IPCC (2007) definitions, given that most studies take a couple of years from conception to publication. Table 1 shows key specific terms and selections utilised.

Table 1 Advanced search terms (words occurring anywhere in the article) used in Google Scholar to source relevant literature. "Salt marsh" as two words was searched as well as "saltmarsh". Relevance was then checked, and those that specifically performed a vulnerability assessment included
Full size table

Initial scoping found an enormous quantity (n = 8,260) of vulnerability research on mangrove systems, indicating sufficient for a standalone review. Publications on mangrove systems were excluded in this study, as other shoreline types were found to be more of a research gap.

All articles sourced were collated in Nvivo software. Nvivo is a versatile qualitative data analysis software widely used in literature review investigations (Bandara et al. 2015). Select functions within Nvivo were used in this study to facilitate the categorisation, organisation, and analysis of information and trends in vulnerability assessment literature (Table 2).

Table 2 Overview of Nvivo functions and their application in this review
Full size table

Only vulnerability assessments which focussed on ecosystem vulnerability were included, excluding applications to community or social vulnerability. Some research did mention of social vulnerability, such as Mahapatra et al. (2015) but the focus was on ecosystem vulnerability. Only English language articles were incorporated.

Sourced literature was categorised by coastal system using the Nvivo classification tool: beaches, saltmarsh and coastal wetlands, and shorelines (Table 2). During the reconnaissance process, a number of shoreline or coastal scale assessments were identified which analysed both saltmarsh and beach systems simultaneously. To avoid complexities in Nvivo analysis, such as counts of methods being inaccurate due to articles being in two system categories, these larger scale assessments were placed in a third “shoreline” category. Beaches and saltmarshes were selected as the systems of focus. Once the literature was categorised, matrix coding queries could be used to identify the distribution of methods across different systems.

The ‘text search’ tool was used to identify specific terms, such as articles which described or defined their definition of vulnerability. Finally, use of spatial analyses within the saltmarsh articles was assessed by searching for key terms identified by review of remote sensing in blue carbon systems (Pham et al. 2019). Relevance of usage of spatial terms was defined as those mentions which were:

  1. 1)

    referring to marsh spatial analysis, for example ‘extent’ or ‘aerial’ could both be used outside this context,

  2. 2)

    referring to methods or applications used within the article itself, such as the term Digital Shoreline Analysis System (DSAS), (e.g., Komi et al. 2022), or LiDAR actual usage (e.g. Webb et al. 2013 rather critiqued use of LiDAR), and

  3. 3)

    were in the main body of the text rather than only in the reference list or acknowledgements.

Results

Through screening, sampled articles were reduced to 31 (Table 2). Literature was primarily from the northern hemisphere, with certain regions being hotspots for research (Fig. 1) and was restricted to articles published in English language.

Fig. 1
figure 1

Distribution of study locations reviewed. Basemap: Google Satellite, 2021. Articles with a global focus (Spencer et al. 2016; Webb et al. 2013) were not included on the map

Full size image

An array of different methods were identified within the sampled vulnerability assessments. Thirteen different methods were found to have been used in the sampled articles (Fig. 2). One article (Myers et al. 2019) appeared to have no specific method, using both review and modelling approaches, hence was categorised ‘No named method’. Details of the sampled articles are provided in the supplementary material (Table S1).

Fig. 2
figure 2

Methods utilised in sampled articles (n = 31)

Full size image

Some articles incorporated multiple techniques, and in these cases, the main method was selected as the code (Fig. 2). For example, though Ekberg et al. (2017) utilised the SLAMM technique (see Fig. 2 for acronyms), their focus was on use of the vulnerability indice while SLAMM only contributed to composing this indice.

The literature selected included only those with actual site-based assessments as opposed to those which described methods without any practical application. For example, Webb et al. (2013) was included as although it detailed the surface elevation table method, it also incorporated a case study, effectively a small assessment. As part of the reconnaissance process, articles which did not incorporate a site-based assessment were coded to different groups within Nvivo such as ‘Papers contributing to theory’ and were not included in the sample analysed. The majority of the publications sampled were journal articles, though some grey literature was included. For example, published reports produced by county, local and municipal government associations were incorporated (Sherwood and Greening 2016; van Proosdij et al. 2018; Haag et al. 2016; Mount et al. 2010).

Results of the number of articles coded to each of the individual geomorphic systems were 10 for beaches, 10 for saltmarsh and coastal wetlands, and 11 for mixed shorelines (n = 31). Results of count from each method as determined using a ‘matrix coding query’ within Nvivo showed a range of results (Fig. 3).

Fig. 3
figure 3

Results of distribution of methods across the sampled articles on beach, saltmarsh and mixed shoreline systems. Figure 2 provides explanation of acronyms

Full size image

The distribution of methods across different systems as examined using the ‘matrix coding query’ within the Nvivo software showed many system-specific methods (Fig. 4).

Fig. 4
figure 4

Plot of distribution of methods across the sampled articles showing system differences

Full size image

Literature showed limited description of definition used for the term ‘vulnerability’ or ‘vulnerability assessment’. While the sampled literature mentioned “IPCC” or “Intergovernmental Panel on Climate Change” 100 times across 17 articles, there was only one mention of a “concept of vulnerability” or the phrase “vulnerability is defined”. However, where a definition was provided the IPCC’s 2007 definition of vulnerability occurred most, examples are Mussetta et al. (2017) and Sharma and Ravindranath (2019).

A ‘text search query’ for the term DSAS indicated limited spatial change analysis in the sampled saltmarsh vulnerability assessments (Table 3). Results further showed limited use of other types of spatial change analysis specifically in saltmarsh assessments (Table 4).

Table 3 Results from text search query, ‘DSAS’ across vulnerability assessment articles (n = 31)
Full size table
Table 4 Use of spatial change analysis in saltmarsh and coastal wetland articles
Full size table

Mount et al. (2010) was excluded from spatial analysis results in Table 4 as such terms were only found in sections of the report that were irrelevant to the actual vulnerability assessment.

Discussion

Past reviews of vulnerability assessments have not focussed on saltmarsh systems, rather have been applied to broad coastline and beach assessments (Sudha Rani et al. 2015; Abuodha and Woodroffe 2006). Results showed literature (Table S1) was focussed on worldwide coastal systems, though largely temperate mid-latitudes (Fig. 1). Even though this study was limited to English-language studies, the region surrounding the Mediterranean (roughly between Spain and Lebanon) was a hotspot for research (shoreline and beaches specifically), about equal to North America. Very few (16%) articles were focussed on the southern hemisphere.

The analysis tools in Nvivo provided an effective method to identify patterns and trends within the sampled literature. Of the thirteen different methods identified (Fig. 3), the indices approach was by far the most utilised, characterised by use of a numeric index calculated from data like tidal range, mean wave height, relative sea level trends, coastal slope, past spatial change and coastal geomorphic type (Pendleton et al. 2010). Results were ranked from very low to very high vulnerability across 4 or 5 categories. Examples of different indexes include the coastal vulnerability index (Pendleton et al. 2010; Palmer et al. 2011; Ghoussein et al. 2018; Mafi-Gholami et al. 2019), beach vulnerability index (Alexandrakis and Poulos 2014) and dune vulnerability index (Pennetta et al. 2018). The multi-variable approach (Fig. 3) was very similar in combining results from a selection of variables into mostly three ranks of vulnerability, which could then be applied by other researchers or managers to other sites.

Many of the methods identified were only used in one study (Fig. 4), indicative of a prevalence of unique approaches within vulnerability research as opposed to frequently used techniques, particularly within saltmarsh focussed studies. An exception was the surface elevation table (SET) method, the most commonly used vulnerability technique for saltmarsh systems here (Fig. 4), allowing a technical analysis of different components of wetland surface elevation change. This method however, as is noted in the seminal article on the technique (Webb et al. 2013), should always be complemented by assessment of other variables in order to obtain a multi-dimensional understanding of marsh vulnerability.

There was otherwise little sharing of methods between the different systems: beaches, shorelines, and saltmarsh/coastal wetlands (Fig. 4). This indicates that methods used in vulnerability assessments are specific to systems. For example, techniques such as SLAMM or the SET method would not be used in systems outside of wetlands/saltmarshes. However, the multi-variable analysis method was used for each of the three systems, though the variables used differed. Beach vulnerability assessment included variables of riverine sediment input, long-shore and cross shore sediment transport, and an aeolian transport indicator (Alexandrakis and Poulos 2014). Dune vulnerability index (Pennetta et al. 2018) included anthropogenic influences and vegetation condition as well as marine and aeolian influences on different dune types. Multi-variable analysis was categorised as separate from the indices approach (Fig. 4) as the indices approach tended to involve repeatability whereas multi-variable methods included more site or system-based tailoring in the selection of individual variables.

Across the thirty-one articles sampled, very few defined or explained vulnerability and only six incorporated any form of discussion of the concept. Cui et al. (2015) and Mafi-Gholami et al. (2019) both described their chosen definition for vulnerability, aligning with the IPCC 2007 definition. Ghoussein et al. (2018), Palmer et al. (2011), Bosom and Jiménez (2011), and Sousa et al. (2013) all described varying definitions, some of which incorporated aspects of the IPCC definitions combined with their own additions. The remaining twenty-five articles appeared not to use any definition, reflecting an issue in vulnerability literature in that comparisons are difficult to make if definitions of the concept are unclear, and in some cases contradictory. Interrelations between research are necessary to understand the wider vulnerability of systems and regions, and would enable inter-relation of research.

Further to discussion of definition, articles also differed on their choice of external hazard (such as vulnerability to what). The majority of the literature was focussed on vulnerability to sea level rise related to climate change, demonstrated by the top ten terms found in the literature, three being ‘sea’, ‘level’, and ‘rise’. There were several exceptions identified such as Christie et al. (2018) which assessed vulnerability to storm surge flooding and Medellín et al. (2016) which examined vulnerability to wave related impact. Other exceptions are demonstrated in the supplementary material (Table S1). Conceptualisation of vulnerability (Romieu et al. 2010) differed significantly between research focussed on natural hazards and research focussed on climate change. This difference needs further assessment as the number of articles from a natural hazards background were limited in this study sample compared to those on climate change and related impacts, and many did not fully define concepts used.

Through the ‘word search’ queries conducted in Nvivo, it was found that there was a lack of spatial change analysis in the saltmarsh and coastal wetland assessments, specifically analysis of spatial extent (Tables 3 and 4). Some articles referred to erosion intermixed with shoreline retreat, and lacked clarity that erosion involves vertical change such as sediment loss (Webb et al. 2013) while spatial change is indicated by aerial photography or satellite imagery analysis over time. This identified gap is important as analysis of changes in spatial extent can provide long-term indications of progradation/retreat. Spatial change is important in demonstrating system sensitivity (Ellison 2015) and provides important information about how future sea level rise increases vulnerability. One wetland article in this study was found to analyse spatial extent over time by use of DSAS (van Proosdij et al. 2018; Table 3). DSAS is software produced by the United States Geological Survey to assess shorelines change over time through analysis of historical aerial imagery (Himmelstoss et al. 2021), and has been useful in shoreline and beach assessments (Table 3), as well as in other blue carbon ecosystems like mangroves (Crameri and Ellison 2022). DSAS has potential for application within vulnerability assessments, allowing incorporation of reliable information about site-specific sensitivity which would complement region-based information about relative sea level rise. The only drawback with use of DSAS is that it relies upon availability of historical aerial imagery for the study sites. Aside from one application of DSAS, other sampled saltmarsh literature utilised little spatial change analysis. Examples of the analyses conducted in the remaining literature included: digital elevation models, habitat change analysis and ground-truthed ecosystem mapping.

Extensions to Research

Review could be expanded, and extended to vulnerability assessment of coastal systems not published in English as this study reviewed, rather languages such as Chinese or Spanish language, to analyse similarities or differences. Furthermore, the inclusion of grey literature from government and non-governmental organisations on environmental protection could also be included, again to give wider scope to review of use of vulnerability assessment concepts. Such reports were excluded as Google scholar selects scholarly literature, and the website layout of these organisations would require manually searching and internal examination to identify relevant examples. However, review of such government and NGO literature may explore methods and approaches in use that could differ from those used in journal articles or other government reports. As this review was also limited to three systems (beaches, shorelines and saltmarsh), it would be useful to assess whether patterns identified in this study are also found in other systems. For example, early reconnaissance found the largest proportion of vulnerability research is focussed on mangrove systems, and a larger scope study could compare vulnerability assessment approaches in tropical to temperate systems, and analysis could investigate if mangrove-focussed literature is using varying definitions of vulnerability concepts.

Conclusions

Vulnerability assessments for saltmarshes, beaches and mixed shorelines were reviewed, and the following conclusions were identified:

  1. 1.

    The indices approach was the most common technique to assess vulnerability.

  2. 2.

    In saltmarsh systems, most articles used unique methods as opposed to repeats of previously published methods.

  3. 3.

    Most literature assessed vulnerability to climate change and sea level rise relative to natural hazards.

  4. 4.

    The volume of vulnerability assessment literature for mangroves was far greater than for saltmarsh or beach systems.