Spatial and temporal scales of coral reef fish ecological research and management: a systematic map protocol

Coral reefs are rapidly changing in response to local and global stressors. Research to better understand and inform the management of these stressors is burgeoning. However, in situ studies of coral reef ecology are constrained by complex logistics and limited resources. Many reef studies are also hampered by the scale-dependent nature of ecological patterns, and inferences made on causal relationships within coral reef systems are limited by the scales of observation. This is because most socio-ecological studies are conducted at scales relevant to the phenomenon of interest. However, management often occurs across a significantly broader, often geopolitical, range of scales. While there is a critical need for incisive coral reef management actions at relevant spatial and temporal scales, it remains unclear to what extent the scales of empirical study overlap with the scales at which management inferences and recommendations are made. This systematic map protocol will evaluate this potential scale mismatch with the goal of raising awareness about the significance of effectively addressing and reporting the scales at which researchers collect data and make assumptions. We will use the Collaboration for Environmental Evidence (CEE) systematic mapping guidelines to identify relevant studies using a framework-based synthesis to summarise the spatial and temporal scales of coral reef fish ecology research and the scales at which management inferences or recommendations are made. Using tested predefined terms, we will search for relevant published academic and grey literature, including bibliographic databases, web-based search engines, and organisational websites. Inclusion criteria for the evidence map are empirical studies that focus on coral reef fish ecological organisation and processes, those informing management interventions and policy decisions, and management documents that cite coral reef research for management decision-making. Study results will be displayed graphically using data matrices and heat maps. This is the first attempt to systematically assess and compare the scales of socio-ecological research conducted on coral reef systems with their management.

and ecological interactions across spatial and temporal scales is considered one of the biggest challenges to effective management, and often mismatches occur between the scales of ecosystem observation, management decision-making, and the regulation of human impacts [4,5]. Given the variety of logistical and technological constraints that accompany data collection, our scales of observation often do not match the multiple scales at which drivers determine ecosystem dynamics [6][7][8]. Subsequently, inferences for management derived from this empirical data are often made beyond the scale of study [4,7,[9][10][11]. Although Socio-Ecological Systems (SES) management frameworks have improved the integration of social, natural, and governance systems [12][13][14][15], a multi-scale approach is crucial to better understand the interactions and feedbacks between them [16][17][18][19]. Coral reefs provide a useful lens to apply such an approach as a tightly coupled socio-ecological system [20], where ecological and social processes form a complex web of interactions across a range of spatial and temporal scales [21][22][23].
Mounting anthropogenic stressors have had adverse effects on the biodiversity and functioning of coral reef ecosystems [24][25][26] and their delivery of ecosystem services such as food security, culture, tourism, recreation, and coastal protection [27]. Chief among these stressors are mass coral bleaching events associated with warming ocean temperatures that result in broad-scale coral mortality [28][29][30][31], and restructuring of coral reef fish assemblages [32,33]. In some locations, these impacts are so severe they lead to ecosystem regime shifts away from hard calcifying corals to dominance by non-reefbuilding organisms [34,35]. In response to these threats, there has been a shift towards implementing broaderscale management tools such as networks of Marine Protected Areas (MPA) for integrated conservation and fisheries management [36][37][38]. Although most broadscale threats are beyond the direct influence of local managers, management decisions are based on the best available scientific evidence, and interventions must often be implemented within existing broader governance structures [39][40][41]. Additional challenges for managers arise from discrepancies in this evidence base, in particular related to the effectiveness of MPAs [1,42,43] and Herbivore Management Areas (HMAs) as tools to increase coral reef resiliency and prevent shifts [44][45][46][47][48]. Some of these inconsistencies may be due to the underlying complexities associated with coral reef ecosystem dynamics across scales and the failure to acknowledge observational scale limitations when making management recommendations.
Herbivore grazing and behavioural studies show clear links between fishes and algal populations at small (1-10 m 2 ) spatial scales [49,50]. However, the impact of herbivorous fishes on benthic community states and reef resilience to climate change across larger scales remains debated [42]. Despite this, some countries have proceeded in implementing HMAs as part of a resiliencebased management strategy that controls the fishing of herbivores [48,51,52]. To improve the interpretation of ecological observations for management decisions such as HMAs, it is imperative that studies acknowledge and address scaling issues by considering the geography and local bounds on the system.
The importance of scale is evident in both the ecological and social research domain. For example, coral reef benthic community composition, fish assemblage trophic structure and herbivore feeding behaviour and therefore functional impact varies across broad environmental gradients (e.g. temperature, irradiance, and within reef system wave exposure) [53][54][55][56]. Similarly, governance structures and anthropogenic pressures (e.g. fishing, land-based pollution) vary across contexts and scales, influencing the state of coral reef ecosystems [57,58]. They also influence the scale at which data collection is needed to evaluate these pressures and the appropriate management levers and governance levels for intervention [1,[59][60][61]. As our understanding grows on how local context-dependent variability in social-ecological systems interacts with larger scale processes, the need for scale-dependent nuance in management recommendations will become more pressing.
Emergent technologies that can be used to monitor the marine environment at larger scales and at a higher resolution [62,63], will undoubtably facilitate a better understanding of multi-and cross-scale variability in these ecosystems [64,65]. Integrating big data from new technologies with multi-scale social drivers and fine-scale ecological processes does have its associated challenges and significant practical hurdles [62,63]. Nonetheless, recent attempts to integrate new technological tools with traditional research methods have shown promising results [66,67], and should enable more effective and efficient monitoring if scaling issues are addressed as a critical requirement [6]. No matter how much multiscale data we may have access to in the future, in order to cross-calibrate, integrate and scale-up existing studies into this new big-data paradigm, there remains a key need to assess and take stock of the scale of the existing evidence base on coral reef socio-ecological research.
This paper presents a novel protocol to systematically map the scales of ecological and social data collection, and the scales at which researchers interpret data to make management inferences and recommendations. By doing so, we will highlight any mismatches in scale between the two with the goals of: (1) highlighting the shortfall and implication of addressing and reporting the scales at which researchers collect data and make assumptions, (2) ensuring existing ecological and social observations are discussed and used as an evidence base within the bounds of their spatial and temporal domains, (3) highlighting priority scales for future ecological and social evidence that managers and policy makers need to make more informed decisions to combat current threats to coral reefs. The systematic evidence map will be created by extracting and collating metadata from relevant studies, with a basic analysis of trends and patterns from which a full systematic review can be conducted at a later date.

Stakeholder engagement
The topic of this study was formulated by Project Team members, representing a broad range of research interests and expertise from the disciplines of ecology, geography and social science, some of whom are in regular contact with natural resource managers and decision makers. Given the scope of the project, it was not deemed necessary to consult with separate stakeholder groups during the development of the project objectives and methods.

Objective of the review
The primary research question of this systematic map is 'What are the spatial and temporal scales of coral reef fish ecological and social research and are these consistent with the scales at which researchers make management inferences and recommendations?' .
The systematic map will be built using the following predefined elements: Population: shallow and mesophotic tropical coral reef ecosystems (0-150 m depth).
Phenomenon of interest: spatial and temporal scales investigated and discussed within coral reef research studies.
Context: ecological, social, and socio-ecological research and management literature.
The focus of the systematic map is on all tropical coral reef-associated fish species, from shallow and mesophotic coral reef ecosystems. The systematic mapping process will involve assessing studies of these ecosystems that focus either primarily on ecological, social, or those that integrate ecological and social research. The evidence base will be categorised using a data coding framework (Additional file 1) designed to explore the following secondary questions: • Has there been a change in the spatial scales at which coral reef fish ecological and social research has been conducted over the last 10 years? • Is there a mismatch between the scales at which ecological data and social data are collected? • Is there a mismatch between the scale of ecological and social data collection and the scales at which management inferences and recommendations are made? • Is there a mismatch between the scale of the ecological and social indicators used to assess management effectiveness and the scale of the policy or management intervention? • Of the studies identified, are there spatial patterns in scale mismatches and do they vary by region or by the scale at which the studies are conducted (i.e. at island or regional scale).
To clarify the key terms to be used in this study, Table 1 lists the definitions of the terms used in this protocol document as relevant to the study objectives.

Methods
The systematic map protocol has been developed in accordance with the Collaboration for Environmental Evidence Guidelines and Standards for Evidence Synthesis [68] and the 'RepOrting standards for Systematic Evidence Syntheses' (ROSES) for systematic map protocols [69] (see Additional file 2). The PREDICTER tool (www. http://predi cter.org/) will be used to predict the number of days for each step of the process [70] to help to inform project scheduling and the Project Team members of project time commitments.

Term Definition
Management effectiveness The degree to which a policy or intervention meets its stated goal and/or meets the broader goal of improving ecological or social conditions Management intervention Direct actions (top-down or bottom-up), activities and/or policies designed to improve ecological and/or social conditions relating to coral reef ecosystems

Management inference
An assumption about the effectiveness of management interventions, or a recommendation for interventions, based on study findings

Searching for studies Search string
This search strategy details the steps that we will take to ensure that a comprehensive and unbiased search is conducted of the most relevant available knowledge. We will search multiple databases using keywords to search their title, abstract and keyword lists. Advanced search options will be utilised using phrases and Boolean operators and all options used will be recorded. The Project Team members compiled an extensive set of search terms relevant to different elements of the primary question and a scoping exercise was conducted using Web of Science Core Collections to modify the search terms (see Additional file 3). The asterisk was used as a wildcard to allow for singular or plural words to be identified in the same search. A list of alternative terms was established with the aid of a thesaurus and screening of relevant citations. We will use the following search string: (coral NOT temperate) AND (fish* OR detritivor* OR herbivor* OR browse* OR graze* OR scrape* OR excavat* OR planktivor* OR omnivor* OR corallivor* OR invertivor* OR piscivor* OR carnivor*) AND (ecosyst* OR *ecolog* OR spatial OR temporal OR assemblage* OR composition OR population* OR communit* OR structure OR function* OR process* OR social OR socio* OR econom* OR human OR pressure OR manage* OR anthropo* OR conservation OR sustainab* OR livelihood OR well-being OR service OR "food security" OR seascape* OR watershed OR protect* OR subsist* OR fisher* OR govern* OR change OR recover* OR network OR reserve OR marine OR interven*).
The searches will be limited to studies published from 2010 onwards in order to focus our study on the last decade of coral reef socio-ecological research. This particular time period has seen a noticeable increase in research that integrates social and ecological datasets to better understand socio-ecological systems [20] and studies that seek to understand the linkages between disturbances, ecological functioning, and ecosystem services [71].

Searching the literature
Searches for relevant published academic and grey literature will be conducted within (i) bibliographic databases, (ii) web-based search engines, and (iii) grey literature, including online publication databases and organisational websites and repositories. The literature will be searched in the English language with predefined search terms (see Additional file 4).

i) Bibliographic databases
The following academic databases will be searched to cover a range of time-periods, subjects, and geographical scope. Searches will be performed across 6 databases, as listed in Table 2, using the predefined search string (see Additional file 4):

Search engines
The academic search engine Google Scholar will be used to search for relevant literature not identified by the bibliographic database search results. Search terms will be simplified by modifying the original search string (see Additional file 4: Table S2) and limiting the search to studies published from 2010 onwards. The citations for the first 200 results will be extracted as citations and added to the bibliographic databases records before removal of duplicates [72]. Documents will be downloaded using the free software 'Publish or Perish' (http:// www.harzi ng.com/resou rces/publi sh-or-peris h), and the free software Zotero (http://www.zoter o.org) will be used to download 20 full studies at a time [73].
ii) Searches for grey literature The search for relevant grey literature will include practitioner-generated research such as unpublished monitoring reports from government agencies, consultancies, thesis repositories, and organizational websites and databases. Specifically, the following websites and sources will be accessed: Searches will also be conducted across 29 relevant organisational websites, using simple search terms adapted from the search strategy. The websites were chosen to represent a diversity of information across all ocean basins. For each website, the first 100 results will be screened in situ, and information from each search will be recorded and described in the final systematic map publication. Information will include website name, date accessed, search string, number of relevant studies identified at full text. Relevant full texts will be recorded for inclusion in the systematic map database. The following organisational websites will be searched for relevant literature:  Table S3). The authors selected the list to represent a range of ecological and social studies on coral reef ecosystems, covering a range of authors, journals, research projects, and geographies relevant to the scope of the question. The search was finalised once all the studies in the benchmark list were found. During the final scoping exercise, 6801 studies were returned using the search string's final iteration, which matched all 20 benchmark studies. During each database search, alert services will be established to capture future updates by identifying and including relevant new studies published during the project. Amendments to the search string and search strategy during the review process will be reported in the final systematic map publication. The term 'coral' is not always used in the title or abstract of social studies related to small-scale coral reef fisheries and marine conservation's human dimensions. Subsequently, a separate list of social benchmark studies was developed to account for this important body of work. A 'benchmark' list of 10 key social study studies were selected to represent this body of work, covering a range of authors, journals, and research topics relevant to the scope of the question (see Additional file 3: Table S4). These studies will be used with non-search related methods to identify relevant studies to be included in this study. This will be conducted by using citation chasing, along with checking relevant bibliographies. We also aim to put out calls to the coral reef science community for relevant studies.

Assembling a library of search results
Studies from all bibliographic database search results will be combined with those from Google Scholar searches into the Zotero reference management software. This software was chosen because the singleclick capture function used to download pdf versions of studies is compatible with more databases and websites than other similar software.
The CADIMA online review management tool (www. cadim a.info) will be used to create a project workspace which will assist the Project Team in organising and managing the sources of evidence and the screening process. The tool is permanently hosted and maintained by Julius Kühn-Institut (JKI) and allows for upload of relevant titles and abstracts for candidate studies identified through the search strategy. CADIMA helps to structure and document the literature search by associating a search string with a search engine, whilst the individual search results can be uploaded to CADIMA as RIS files [74]. The first key steps involve checking if all abstracts are entered for each retrieved reference and removing duplicates from the generated merged reference list.

Screening process
The full list of studies will be screened for relevance using the eligibility criteria (listed below) by initially reviewing each study title and abstract text at the same time. Studies that meet the inclusion criteria will then be reviewed at the full text stage. Additionally, any studies with uncertainty about the relevance based on the title and abstract will be included and reviewed at the full text stage.
To ensure consistency between reviewers at each stage of the screening, the CADIMA software provides an automated calculation of a kappa-statistic to test inter-reviewer agreement when applying the defined criteria [75]. The observed agreement is placed in relation to the one being expected by chance using a value between 0 and 1 (where 0 = a weak agreement and 1 = a perfect agreement). When a reviewer is uncertain about including a study, it will be marked for a second opinion and will be screened by a second reviewer. Where more than one reviewer independently assesses studies and inconsistencies between reviewers occur, they will be automatically identified by CADIMA and the respective reviewers asked to solve those conflicts. If necessary, the Project Team will meet to discuss inconsistencies to help improve the understanding of the inclusion criteria.
The estimated proportion of studies that will be randomly screened and checked for consistency by two reviewers will be 10% at each of the title, abstract, and full text stages. Reviewers who have authored studies to be considered within the review will be prevented from unduly influencing inclusion decisions by delegating tasks related to these studies to other Project Team members.

Eligibility criteria
The following inclusion criteria will be used to screen the relevance of studies to determine whether they will be assessed at the meta-data extraction stage: Population: studies that include tropical and subtropical shallow (between 0-30 m depth), and mesophotic (between 30-150 m depth) coral reef ecosystems.
Subject: studies focusing on tropical coral reef fish ecology and management.
Phenomenon of interest: spatial and temporal scales investigated and discussed within research studies.
Context: ecological, social, and socio-ecological research and management literature.
Eligible types of study design: primary research studies conducted by the authors that meet the following criteria will be included: • Studies that have observed or measured coral reefassociated fish, live (surveyed in situ using ecological methods), or dead (including catch landings, market surveys and archaeological fossil-remains). • Purely social studies that do not encounter fish, but include surveys with fishers or community leaders about their fishing behaviour, changes in fish catch or consumption over time. • Any quantitative or mixed method (quantitative and qualitative) study, including empirically tested data-driven theoretical models.
The following studies will be excluded: • Purely qualitative research, theoretical, laboratorybased, or modelling studies. • Literature reviews which do not describe methods used for search, data collection, and synthesis. • Editorials and commentaries, meta-analyses, systematic reviews, systematic maps, evidence gap maps. • Literature published before 2010. • Literature whose full texts exist in a language other than English.
A full list of studies excluded at full text stage will be provided in the final systematic map publication as an additional file including reasons for exclusion. Studies that cannot be located or accessed will also be reported.

Study validity assessment
Due to the likely number of studies and wide breadth of study designs, critical appraisal and assessment of the quality of studies will not be undertaken.

Data coding strategy
During the meta-data extraction process the free software SysRev (http://sysre v.com/) will be used to provide a platform for collaborative extraction of data from the selected studies. Metadata will be extracted from each relevant study that satisfies the inclusion criteria using a standardised coding and data-extraction form (see Additional file 1). The authors reviewed and refined the form by conducting a pilot meta-data extraction study with the 20 benchmark studies.
Using relevant studies, we will assess the research's spatial and temporal scale and the management recommendations and inferences. Scale usually refers to the spatial (space) and temporal (time) dimensions of a pattern or process [76]. In landscape ecology, a measured variable's spatial or temporal scale refers to the 'resolution' and 'extent' [77]. 'Resolution' refers to the precision used in measurement, where 'grain' is the finest spatial resolution, or the size or duration of a single observation [76,78]. The 'extent' of the study area fixes the outer boundary of the measured variable [77]. For example, benthic imagery collected in situ by divers has high resolution (cm), but limited extent (10 m 2 ), whereas satellite-derived temperature data (e.g. Pathfinder) has low resolution (4-5 km), but large extent (global).
In order to assess how well modern ecological research is achieving an all-inclusive and predictive understanding of ecosystems, Estes et al. [6] quantified the spatial and temporal domains of empirical ecological observations that were published between 2004 and 2014. To do this they defined two additional dimensions for measuring temporal scale; the 'interval' (time elapsed between successive temporal replicates) and 'duration' (time elapsed between first and last temporal replicates). The study evaluated how the actual scales of observation i.e. how much the measurement covers space and time, differ from the scales the observations attempted to represent. To do this they assessed the 'actual extent' , the summed area of spatial replicates, and the 'actual duration' the summed observational time of temporal replicates. Accordingly, we will estimate the following dimensions of coral fish reef ecological, social and socio-ecological research using the terminology listed in Table 3.
In this instance, we define 'replicate' as the lowest level of independent observations collected from which mean and variance is calculated, or the scale at which the authors describe as "independent replicates" within their statistical framework. Although 'sampling dispersion' is the spatial equivalent of 'observational interval' used for describing temporal scales, the Project Team decided not to incorporate it into this study given the complexities that coral reef ecosystems present when estimating areas of surveyable reef. These include spatially complex distributions of heterogenous habitat types, in addition to the complexities associated with depth and 3D habitat structure.
When assessing social studies, the term 'level' is used for describing social system organization which has no explicit spatial or temporal dimensions [79]. For the purpose of this study, the Project Team developed a list of categories representing a progression of increasing levels of social organisation that have been used by other studies and roughly relate to particular spatial scales [10,80,81]. The Project Team tested the suitability of the categories and amended as necessary using 4 of the benchmark studies listed in Additional file 3: Table S3 [ [82][83][84][85]. The final list of levels in order of increasing social organisation are: Individual; Household; Village; Community; User groups; Watershed; Sub-regional (island, district etc.); State governmental department (or similar); National governmental department; Regional (e.g. Caribbean, South Pacific etc.); and International body. The chosen level of social organization is relevant to our assessment of potential matches or mismatches and whether they are more common at certain levels. During the meta-data extraction phase there will be an opportunity to select multiple options if relevant to the particular study.
The following main categories of data will be extracted (see Additional File 1 for more detail): • General information on the study, including year initiated, duration, location, and type of data collected (ecological, social, socio-ecological), and if there are socio-ecological linkages included in the discussion. • Ecological study details, including depth, habitat type, survey techniques [63], and methods [86], response variables measured, habitat associations investigated, ecosystem processes [87], and biological characteristics [71] assessed in the study. • Social study details, including type of data sources used, type of data collected, methods used in study [88], and social system characteristics being assessed (including demographic factors, economics (e.g., markets and trade), technological factors, knowledge, attitudes and values, and institutions and governance systems [20]. • The spatial and temporal scales of the study, drivers and management interventions (spatial resolution including total extent and actual extent, and temporal interval including total duration and actual duration). • Socio-ecological system components, including: o Linkages between social factors and ecosystem conditions [20]. o Assessed and recommended management interventions addressed in study. o Management effectiveness indicators (social and ecological) used in study [89].
Before the final meta-data extraction is initiated, consistency checking will be conducted by all Project Team members. This will include at least 20 studies randomly assigned to multiple Project Team members. Any inconsistencies will be discussed, and the meta-data extraction coding list will be amended as required. If further inconsistencies arise during the review process, the Project Team members will meet to resolve the issues. As discussed in the screening criteria section, reviewers who have authored studies to be considered within the review will be prevented from unduly influencing inclusion decisions, by delegating tasks related to these studies to other Project Team members. If studies are missing information or details need clarifying during the review process, the corresponding authors will be contacted by email.

Study mapping and presentation
The CADIMA tool will be used to document the review process and provides the following information and data formats: • A flow diagram summarising the study selection process; • Reference lists for each database (xlsx) and the final reference list after duplicate removal (xlsx and RIS); • The outcomes of the consistency check and study selection across the different stages (title, abstract, and full text) including the reasons for exclusion (xlsx); • The filled meta-data extraction sheet (xlsx).
The project will follow a framework-based synthesis method to summarise the descriptive characteristics of the included studies according to the population, study designs, and outcomes, and will conduct additional analyses such as looking at number of studies published per year [90,91]. The meta-data extraction sheet will be formatted to produce summary data, figures, and tables to enable summarisation of key characteristics and trends. A data matrix will be developed to provide a graphical illustration of the distribution and frequencies of key data analysis results. Heat maps will be created to identify knowledge gaps of unrepresented or underrepresented subtopics that warrant further primary research. Knowledge clusters (well-represented subtopics that are amenable to full synthesis via systematic review) will also be identified and reported. The graphics will build upon work conducted by Estes et al. [6] on similar questions in ecology focused journals between 2004 and 2014.
The data and code will be uploaded to an online open access data repository such as GitHub. The CADIMA tool will be used to make synthesis results available to third parties by displaying the documents on the web site and enable external users to download them.